CHAPTER NINETEEN

The Electrooptic and Acoustooptic

Effects and Modulation of Light Beams
 19

The Electrooptic and

Acoustooptic Effects and Modulation
of Light Beams

In this chapter we shall explain how the distortion produced in a crys-

tal lattice by the application of an electric field or by the passage of a
sound wave affects the propagation of light through the crystal. These
effects – the electrooptic and acoustooptic effects are of considerable
practical importance as they can be used to amplitude and phase modu-
late light beams, shift their frequencies and alter the direction in which
they travel.

19.1 Introduction to the Electrooptic Effect

When an electric field is applied to a crystal, the ionic constituents move
to new locations determined by the field strength, the charge on the ions
and the restoring force. As we saw in Chapter 18 unequal restoring forces
along three mutually perpendicular axes in the crystal lead to anisotropy
in the optical properties of the medium. When an electric field is applied
to such a crystal, in general, it causes a change in the anisotropy. These
changes can be described in terms of the modification of the indicatrix
by the field – both in terms of the principle refractive indices of the
medium and in the orientation of the indicatrix. If these effects can be
described, to first order, as being linearly proportional to the applied
field then the crystal exhibits the linear electrooptic effect. We shall see
 The Linear Electrooptic Effect 625

that this results only if the crystal lattice lacks a center of symmetry. *
So, some cubic crystals can exhibit the linear electrooptic effect. If the
crystal possess a center of symmetry (or is even an isotropic material
such as a gas or liquid) a change in optical properties can result that
depends, to first order, on the square of the applied field. This is the
quadratic electrooptic effect. Both the linear and quadratic electrooptic
effects can be used effectively in various optical devices.

19.2 The Linear Electrooptic Effect

For any cartesian coordinate system the equation of the indicatrix has
the general form
( n12 )1 x2 +( n12 )2 y 2 +( n12 )2 z 3 +2 ( n12 )4 yz+2 ( n12 )5 xz+2 ( n12 )6 xy = 1
(19.1)
If x, y, z are chosen to be principal axes the cross terms in yz, xz and
xy disappear and, in general, the equation of a biaxial indicatrix (Eq.
(18.52)) results. For the linear electrooptic effect, an applied electric
field changes each of the coefficients ( n12 )i above in a way that is linearly
dependent on the Ex , Ey and Ez components of the field. This change
=
in these coefficients is described by the electrooptic tensor r through the
relationship
  X3
1
∆ = rij Ej (19.2)
n2 i j=1
where E1 = Ex , E2 = Ey , E3 = Ez . So for example:
 
1
∆ = r11 Ex + r12 Ey + r13 Ez (19.3)
n2 1
and
   
∆( n12 )1 r11 r12 r13
 ∆( n2 )2   r21 r22 r23  
1 
    Ex
 ∆( n2 )3   r31 r32 r33  
1
 =  Ey  (19.4)
 ∆( n12 )4   r41 r42 r43 
    E z
∆( n12 )5 r51 r52 r53
1
∆( n2 )6 r61 r62 r63
Typical magnitudes of the rij coefficients are 10−12 m/V . The form of
=
the matrix describing r depends on the symmetry of the crystal and is

* A crystal posses a center of symmetry if there are identical particles in the lattice
at vectors r and -r where r is a position vector measured from an arbitrary origin.
 626 The Electrooptic and Acoustooptic Effects
=
closely related to the symmetry of the piezoelectric∗ tensor d, which
relates polarization produced in a medium to stress σ ∗∗ The matrix
= =
describing r is the transpose of the matrix describing d. Nye has
=
tabulated in a convenient way the form of d. Table (19.1) is a slight
variant on the format given by Nye. ***
With an electric field applied the equation of the indicatrix changes
from Eq. (19.1) to
1 1 1 1 1 1
[ 2 + ∆( 2 )1 ]x2 + [ 2 + ∆( 2 )2 ]y 2 + [ 2 + ∆( 2 )3 ]z 2
n1 n n2 n n3 n
1 1 1 1
+ 2[ 2 + ∆( 2 )4 ]yz + 2[ 2 + ∆( 2 )5 ]xz· (19.5)
n4 n n5 n
1 1
+ 2[ 2 + ∆( 2 )6 ]xy = 1
n6 n
The use of the index i on ∆( n12 )i is an example of what is called con-
tracted notation: i takes the values 1,2,3,4,5, and 6 where 1 ≡ xx, 2 ≡
yy, 3 ≡ zz, 4 ≡ yz, 5 ≡ xz, 6 ≡ xy. Therefore the index i represents the

∗ Piezoelectricity is the phenomenon in which application of a force to a crystal

causes a voltage to develop between the faces of the crystal. The voltage appears
because the force has caused polarization – separation of positive and negative
changes within the medium.
∗∗ Stress is force per unit area and can take the form of compressive or shear stress.
For a more detailed discussion see J.F. Nye, Physical Properties of Crystals, Oxford
Unviersity Press, Oxford, 1957.
*** His piezoelectric coefficients dij describe polarization produced by stress in the
form
σ 
11
 σ22 
Px
! d11 d12 d13 d14 d15 d16  σ33 
! 
 
Py = d21 d22 d23 d24 d25 d26  σ23 
Pz d31 d32 d33 d34 d35 d36  σ13 
 
σ 
13
σ12
so, for example
Px = d11 σ11 + d12 σ22 + d13 σ33 + d14 σ23 + d15 σ13 + d16 σ12 σ12 .
σ12 is a shear stress – it is the stress produced in the y direction by a force
acting on the face of a volume perpendicular to x. To be consistent with our
=
notation for r , where factors of 2 appear in the cross terms in Eq. (19.1) we
need to write
Px = d11 σ11 + d12 σ22 + d13 σ33 + ( 12 d14 )2σ23 + ( 12 d15 )2σ13 + ( 12 d16 )2σ12
So all Nye’s dij coefficients where i 6= j are double the value they have in our
definition, where in symmetry terms dij ≡ rji .
 The Linear Electrooptic Effect 627

Table 19.1 Symmetry of the linear electrooptic tensor for the 32 crystal
classes.

zero element non-zero

element
equal non-zero
element
equal non-zero
elements of
opposite
sign
Centrosymmetric classes
All elements zero

specific quadratic term with which it is associated in the equation of the

indicatrix. We shall have cause to use this notation further. So, if our
original axes were principal axes the indicatrix becomes
The application of the electric field has introduced cross terms so
x, y, z are no longer appropriate principal axes – the indicatrix has been
changed in shape and rotated in space.
In a centrosymmetric crystal the change in shape of the indicatrix
produced by a field E must by the same as for field -E, as these 2 fields
are identical as far as the lattice is concerned. Therefore
  X3 X3
1
∆ = rij Ej = − rij Ej (19.7)
n2 i j=1 j=1
which can only be true if rij = 0. Consequently, there is no linear elec-
trooptic effect (or any piezoelectric effect) in a centrosymmetric crystal.
The linear electrooptic effect is closely related to the inverse piezo-
electric effect – where application of an electric field to a non-centro-
symmetric crystal leads to an actual change in crystal shape (Ref. Nye).
The change in crystal shape leads to strain and the shape and orienta-
tion of the indicatrix are altered. With this additional contribution to
the change in index coefficients we can write
  X 6 X3
1
∆ = p ik Sj + rij Ej (19.8)
n2 i j=1
k=1
j takes the values 1, 2, 3; i and k takes the values 1 → 6 (contracted
628 The Electrooptic and Acoustooptic Effects

notation). Sj is a component of the strain;* pij is the elastooptic tensor.

At high frequencies the inertia of the crystal prevents it straining macro-
scopically, so the first term on the right of Eq. (19.8) is zero. At high
frequencies the residual change in indices is called the Pockel’s effect.
At low frequencies the elastrooptic effects cannot be ignored. However,
since the deformation leading to strain is generally caused by the inverse
piezoelectric effect∗∗ – and is therefore also related to applied field - it is
possible to incorporate all changes in index into a single low-frequency
=
linear electrooptic tensor r dc and write
  X3
1
∆ = rdc Ej (19.9)
n2 i j=1 ij
We shall not deal explicitly with this point further, but it should be
noted that whenever a coefficient rij is written in what follows, this
coefficient may be significantly different in magnitude between low and
high frequencies. Table (19.2) gives some examples for important linear
electrooptic materials.
One additional effect is worth noting – the photoelastic effect. This is
the change in index coefficients produced directly by applied stress (no
applied electric field is involved). For this effect
1 X 6
∆( 2 )i = πik σk (19.10)
n
k=1
k runs from 1 → 6 (contracted notation). The πik are piezo-optical
coefficients whose typical magnitude is 10−12 m2 /newton; the σk are the
components of the stress (for further discussion see Nye).

19.3 Longitudinal Electrooptic Modulation

To illustrate how an applied electric field affects an electrooptic crys-
tal it is instructive to consider the specific important material KDP
(potassium dihydrogen phosphate, KH2 P O4 ). The related material

* Each strain component is a measure of the distortion of the crystal shape in a

particular axial direction or of the rotation of the structure about a particular
axial direction. For further details see Nye.
∗∗ If strain results solely from the inverse piezoelectric effect then

X
6
dc = r +
rij pik djk .
ij
k=1
 Longitudinal Electrooptic Modulation 629

Fig. 19.1.

KD∗ P(KD2 P O4 ), ADP (ammonium dihydrogen phosphate, (N H4 )

H2 P O4 ) and AD∗ P,(N H4 )D2 P O4 ) behave similarly. For these mate-
rials, symmetry 42m, the electrooptic tensor has the form
 
0 0 0
 0 0 0 
 
=  0 0 0 
r=   (19.11)
 r41 0 0 
 
0 r41 0
0 0 r63
In the presence of an applied field E = Ex î + Ey ŷ + Ez k̂ the indicatrix,
written in its original principal axes coordinates is
x2 + y 2 z2
+ + 2r41 Ex yz + 2r41 Ey xz + 2r63 Ez xy = 1 (19.12)
n20 n2e
For a field applied in the z direction only Eq. (19.12) becomes
x2 + y 2 z2
2 + 2 + 2r63 Ez xy = 1 (19.13)
n0 ne
To find the new principal axes we note that Eq. (19.13) is symmetric
in x and y, so the new x and y axes must be rotated by 45◦ from the
original axes, as shown in Fig. (19.1). The z axis is unchanged.
The old axes are related to the new according to
x = x0 cos 45◦ + y 0 sin 45◦
(19.14)
y = −x0 sin 45◦ + y 0 cos 45◦
Substitution in Eq. (19.13) gives
1 1 z2
( 2 − r63 Ez )x0 + ( 2 + r63 Ez )y 0 + 2 = 1
2 2
(19.15)
n0 n0 ne
The indicatrix now no longer has any cross terms so our choice of new
630 The Electrooptic and Acoustooptic Effects

principal axes was correct. The crystal has now become biaxial with
n20 n20
n2x0 = ; n 2
0 = ; n2z0 = n2e (19.16)
1 − n20 r63 Ez y
1 + n20 r63 Ez
If n20 r63 Ez << 1, which is true for any reasonable electric field, these
equations become
nx0 = n0 (1 + 12 n20 r63 Ez ); ny0 = n0 (1 − 12 n20 r63 Ez ); nz0 = ne (19.17)
If a wave propagates in the z direction through such a crystal the retar-
dation is
2πL 0 2πLn30 r63 Ez
∆φ = (nx − n0y ) = (19.18)
λ0 λ0
The crystal has become a waveplate whose retardation is linearly propor-
tional to the field. It is common practice to characterize such a crystal
by the voltage, Vπ , necessary to produce a retardation of π. Since this
applied voltage acts across a length L of crystal, Ez = Vπ /L, so from
(Eq. (19.18)
λ0
Vπ = (19.19)
2n30 r63
For a crystal of KDP which has r63 = −10.5 × 10−12 m/V, n0 = 1.51
and for an operating wavelength of 632.8 nm, Eq. (19.19) gives Vπ =
8752V . A crystal of KDP with this longitudinally applied voltage has
been transformed into a half-wave plate and will, therefore, transform
a linearly polarized input wave into a linearly polarized output beam
whose polarization direction has been rotated from its original direction.
If the input wave is polarized at 45◦ to the x0 and y 0 directions, as shown
if Fig. (19.2), that is polarized parallel to the original x or y directions of
the undisturbed indicatrix, then the half-wave plate rotates the plane of
linear polarization by 90◦ . When the crystal is operated in this way, as
the applied voltage increases from 0 to Vπ the output makes a transition
from linear polarization → elliptical polarization with the major axis
aligned along the original polarization direction → circular polarization
→ elliptical polarization with the minor axis aligned along the original
polarization direction → 90◦ rotated linear polarization as shown in
Fig. (19.3). We shall see shortly how this modulation of the polarization
state can be used to make an electrooptic amplitude modulator.
Longitudinal electrooptic modulators of this sort, in which the field is
applied along the light propagation direction, are somewhat inconvenient
insofar as the light must pass through the electrodes used to apply the
field: To minimize loss the electrodes are usually deposited as a fine
metallic grid on the crystal surfaces shown in Fig. (19.2).
 Transverse Electrooptic Modulation 631

Fig. 19.3.

Fig. 19.4.

19.4 Transverse Electrooptic Modulation

It is more convenient to apply the electric field transversely between elec-
trodes placed on side faces of the crystal as shown in Fig. (19.4). As an
example of how this is done we can consider the important electrooptic
material lithium niobate LiN bO3 .*
Its crystal symmetry is 3m so the form of its linear electrooptic tensor
is
 
0 −r22 r13
 0 r22 r13 
 
=  0 0 r33 
r=   (19.20)
 0 r51 0 
 
r51 0 0
−r22 0 0

* This material is also strongly piezoelectric and is used to fabricate acoustooptic

modulators.
632 The Electrooptic and Acoustooptic Effects

Fig. 19.5.

Without an applied electric field the crystal has uniaxial symmetry with
an indicatrix of the form
x2 y z2
2 + 2 + 2 =1 (19.21)
n0 n0 ne
with a field applied in the y principal axis direction the indicatrix be-
comes
1 1 z2
( 2 − r22 Ey )x2 + ( 2 + r22 Ey )y 2 + 2 + 2r51 Ey yz = 1 (19.22)
n0 n0 ne
Clearly, because of the introduction of the cross term in yz the original
principal areas are no longer appropriate. The effect of the field has
been to render the crystal biaxial and rotate the indicatrix in the yz
plane. Let us assume that the indicatrix has been rotated so that its
new principal axes y 0 , z 0 are at an angle θ to the original principal axes,
as shown in Fig. (19.5). The x direction remains a principal axis with
the field Ey applied, as no cross term involving x has been introduced
into the equation of the indicatrix.
We can write the old coordinates in terms of the new as
x = x0
y = y 0 cos θ − z 0 sin θ (19.23)
0 0
z = z cos θ + y sin θ.
If we substitute from Eq. (19.23) into Eq. (19.22) the condition that
x0 , y 0 , z 0 are the new principal axes is that the cross term in z 0 , y 0 must
vanish, that is
1 1
( 2 − 2 − r22 Ey ) sin θ cos θ + r51 Ey (2 cos2 θ − 1) = 0 (19.24)
ne n0
We can assume that r51 Ey is small enough that θ is a small angle and
 Transverse Electrooptic Modulation 633

write sin θ ' θ, cos θ ' 1. Eq. (19.24) then gives

−r51 Ey
θ= 1 (19.25)
( n2 − n12 − r22 Ey )
e 0

with r51 = 28 × 10−12 m/V, r22 = 3.4 × 10−12 m/V, ne = 2.21, n0 = 2.3
and a voltage of 1kV applied across a 1mm thick crystal.
−28 × 10−12 × 106
θ= 1 2 = 1.78 mrad = 0.1◦ (19.26)
( 2.21 ) − ( 2.3
1 2
) − 3.4 × 10−12 × 106
Indeed, θ is a small angle. Since in LiNbO3 n0 > ne , θ is a negative angle
so in reality the rotation of the axes is clockwise, not counter clockwise
as shown in Fig. (19.5). With respect to the new principal axes the
equation of the indicatrix is now
1 1 sin2 θ
( 2 − r22 Ey )x0 + [( 2 + r22 Ey ) cos2 θ + + r51 Ey sin 2θ]y 0
2 2

n0 n0 ne2

2
1 cos θ
− r51 Ey sin 2θ]z 0 = 1
2
+ [( 2 + r22 Ey ) sin2 θ + 2
n0 ne
(19.27)
and with cos θ ' 1, sin θ ' θ
1 1
( 2 − r22 Ey )x0 + ( 2
e

n0 n0
θ2 2
+ r22 Ey + 2r51 Ey θ + 2 )y 0
ne
1 1
+ (( 2 + r22 Ey )θ2 − 2r51 Ey θ + 2 )z 0 = 1
2
(19.28)
n0 ne
the new principal refractive indices satisfy the equations
1 1
2 = 2 − r22 Ey
nx0 n0
1 1 θ2
= + r22 E y + 2r51 E y θ + (19.29)
n2y0 n20 n20
1 1 1
2 = ( 2 + r22 Ey )θ2 − 2r51 Ey θ + 2
nz0 n0 ne
Since θ is very small we can actually neglect the rotation of the indicatrix
and write its equation as
x2 y2 z2
+ + =1 (19.30)
n2x n2y n2z
634 The Electrooptic and Acoustooptic Effects

where
n0
nx = = n0 (1 + 12 n20 r22 Ey )
(1 − n20 r22 Ey )1/2
n0
ny = = n0 (1 − 12 n20 r22 Ey ) (19.31)
(1 + n20 r22 Ey )1/2
nz = ne
For a wave travelling in the z direction the indices of refraction experi-
enced by wave polarized along x and y are nx and ny respectively. For
the transversely operated electrooptic modulator shown in Fig. (19.4)
the retardation is
2πL 4πLn30 r22 V
∆φ = (nx − ny ) = (19.32)
λ0 λ0 d
The half-wave voltage Vπ of the modulator is the voltage required to
make it act as a half-wave plate.
From Eq. (19.32), if φ = π
λ0 d
Vπ = (19.33)
4Ln30 r22
For example: with d = 5mm, L = 10mm, n0 = 2.3, r22 = 3.4×10−12m/V
at an operating wavelength of 530nm - Vπ = 1600V.
To illustrate further how a change in the orientation of the applied
field and the light propagation direction affects the way in which an
electrooptic crystal behaves, let us now consider what happens when
light propagates through a LiNbO3 crystal in the y direction, but with
the field applied along the z direction.
In this case the indicatrix becomes
1 1 1
( 2 + r13 Ez )x2 + ( 2 + r13 Ez )y 2 + ( 2 + r33 Ez )z 2 = 1 (19.34)
n0 n0 ne
The application of the field has left the crystal uniaxial, but the new
ordinary and extraordinary refractive indices satisfy
1 1
2 = ( 2 + r13 Ez )
n00 n0
(19.35)
1 1
= ( + r E
33 z )
n0e2 n2e
1 1
Since r13 Ez << n20
and r33 Ez << n2e for any reasonable applied field
strengths,
n0
n00 = = n0 (1 − 12 r13 n20 Ez )
(1 + r13 n20 Ez )1/2
ne (19.36)
n0e = = n e (1 − 1
2 r33 n 2
e Ez )
(1 + r33 n2e Ez )1/2
If a wave passes through such a crystal in the y direction it will, in
 High Frequency Waveguide Electrooptic Modulators 635

Fig. 19.6.

general, split into an O-wave polarized in the x direction, and an E-wave

polarized along the z direction. For the geometry shown in Fig. (19.6)
the retardation is

2πL 0
∆φ = (ne − n0 ) (19.37)
λ0
which gives
2πL
∆φ = (ne − n0 + 12 (r33 n3e − r13 n30 )Ez ) (19.38)
λ0
The half wave voltage of the crystal used in this way is
λ0 d
Vπ = (19.39)
L(r33 n33 − r13 n30 )
With d = 5mm, L = 10mm, r33 n3e − r13 n30 = 224 × 10−12 m/V, and
our operating wavelength of 530nm; Vπ = 1183V. Although this is a
slightly lower half wave voltage than in the equation configuration dis-
cussed previously, there is a potential drawback to using the crystal in
this way. Eq. (19.38) shows that the crystal operated in this orientation
acts as a waveplate – because the wave is not propagating along the op-
tic axis. The field-independent birefringent retardation 2πL(ne −n0 )/λ0
will be temperature sensitive and can lead to practical stability prob-
lems. However, in practice this difficulty can be overcome by building an
electrooptic device using pairs of crystals in series as shown in Fig. (19.7).
The orientation of the x and z axes in the 2 crystals are arranged to be
orthogonal so that the constant term in Eq. (19.38) cancels out. In order
that the field dependent retardation in Eq. (19.38) does not also cancel
out the field polarity in the second crystal must be reversed.
636 The Electrooptic and Acoustooptic Effects

Fig. 19.7.

19.5 High Frequency Waveguide Electrooptic Modulators

In a high frequency electrooptic modulator the electrical drive to the
modulator must be applied using some form of strysline configuration
so that the induced phase or polarization modulation propagates at, or
close to, the phase velocity of the light signal travelling through the
modulator. If this is not done, the light propagating through the crystal
sees an electric field of varying phase as it propagates and the effects of
the field on the phase or polarization state of the wave can camel out.
Fig. (19. ) shows such a travelling wave electrooptic modulator
(Suzanne Hiser, Fig. (1.2)). The electric field acts on the
waveguide region of the electrooptic crystal as shown in Fig. (19. )
(Suzanne Hiser, Fig. (2.4). In LiNbo3 , for example, the waveguiding
region is made by diffusing titaniun into the crystal. The buffer layer
is usually Al2 )3 or Si O2 , and prevents the propagating wave, in this
geometry a TM wave, from integrating with the metal electrodes. The
undiffused hammer is typically 7µ wide, the buffer layer 0.26µn thick,
with a 3µm thick Cr/Au electrode on the top. The electrode thickness
is chosen to be larger than the depth at the design frequency.
The mode difference between guiding region and substate is typically
0.01-0.02.
In the Z-cut, TM mode of operation, the electric field of the mode has
Ez , Ex components. The Ez component of the wave sees the change in
indix ∆( n12 )3 = r33 Ez , this is the only change in the radiation.
 High Frequency Waveguide Electrooptic Modulators 637

For LiNbo3
 
0 −3.4 8.6
 0 3.4 8.6 
 
 0 0 30.8 
 × 10−12 m/v
=
r=
 0 28 0 
 
28 0 0
−3.4 0 0
typical optical losses are ∼ |dB|cm at 633 nm and ∼ 0.3dB/cm at λ =
1.32µm
 
Ex
∆ ( 1/n2 ) = r  Ey 
=

Ez
For the Z− cut the applied field is in the Ez direction.
 
1
∆ = r33 Ez
n2 3
For the X cut, the field is also applied in the z direction, and the wave
propagates in the y direction Ez is the only electric field component
of the mode, which is a TE mode—it has |E|x and Hy magnetic field
components. With the X cut ∆ ( n12 )3 = r33 Ez is the only change in
the . Various kinds of waveguide structure can be fabricated
in LiNb03 , as shown in Fig. (19. ) (Suzanne Hiser, Figs. 2.1 and 2.2).
In the Mach-Zehnder interferometer configuration the waves propagating
in the two arms can experience different phase shifts so can emerge in
a cut of phase. The device can be used as an amplitude modulator or
switch. If the phase shift between the 2 arms is φ the output intensity
Iout will be added to the output intensity Iin according to
Iout 1
= (1 + cos φ) = cos2 φ/2
Iin 2
Variation of φ in such a device will not produce a linear modulator.
However, if the device is biased by π/2, for example by having one in-
terferometer arm slightly longer than the other (as shown in Fig. (19. ))
(Suzanne Hiser, Fig. (2.2)) then we can write φ = π/2 + ∆φ. In this
case
Iout 1
= (1 + cos(π/2 + ∆φ))
Iin 2
1
= (1 + cos π/2 cos ∆φ − sin π/2 sin ∆φ)
2
1 1
= − ∆φ
2 2
for small phase modulation ∆φ.
638 The Electrooptic and Acoustooptic Effects

This provides linear modulation of the intensity about a near value

Iout = 12 Iin . To achieve passive biasing of LiNbO3 the path difference
is ∆L = 152 nm, this provides φ = π/2. Typical devices are 10-20 mm
long with the guiding region 20µm apart.
Because it is not possible to match the velocity of the excitation field
and optical field over a long distance at high modulation fields, a phase-
reversed scheme can be used to extend the bandwidth. After each length
of the device over which the optical and drive fields get out of phase by
π the polarity of the drive field is reversed as shown in Fig. (19. ) (SH,
Fig. (2.7)). Such a device will in theory have a pass band response.
Broad frequency response can be obtained by aperiodically varying the
length between phase reversals of the drive field, as shown in Fig. (19. )
(SH, Fig. (2.11)).

19.6 Straight Electrode Modulator

In a electrooptic modulator of the straight type high
frequency response is limited by traveling wave mismastch effects. We
can represent the drive voltage at position z along this structure as
wm
V (z, t) = V0 sin(ωm t − nm z)
c0
where c0 is the velocity of light in vacuo, ωm is the drive frequency,
and nm is the effective refractive index for the drive field (such that the
phase velocity of the drive field is c0 /nm ).
The optical field propagates with phase velocity,
c0
c=
h
where n is the effective refractive index for the waveguide mode.
If the optical field enters the crystal at t = 0 at time t0 then the
voltage that is seen as it propagates can be formed from Eq. (19. ) by
writing t = zn /c0 which is the time it takes the optical wave to reach
l9ocation z
ωm z
V (z) = V0 sin( (n − nm ))
c0
The electrooptically induced change in the index n can be written as
∆n(z) = aV (z)
where a is a constant that depends on the specific geometry and crystal
being used.
The total phase shift experienced in propagating a length ` through the
 Straight Electrode Modulator 639

crystal is
Z `
ω∆n(z)
∆φ = dz
0 c0
Z
aV0 ω ` ωm z
= sin( (n − nm))dz
c0 0 c0
which gives
aV0 ω wm z c0
∆φ = − cos[ (n − nm)] |`
c0 c0 ωm (n − nm) 0
aV0 ωm `
=− [cos (n − nm) − 1]
(n − nm) c0
By making the substitution
θ = πωm /ωc
where
πc0
ωc =
(nm − n)`
the total phase shift can be written as
aV0 `
∆φ = ωc [cos θ − 1]
πc0
aV0 `ωm
= [cos2 θ/2 − sin2 θ/2 − 1]
2c0 (θ/2)
aV0 `ωm
= [−2 sin2 θ/2]
2c0 (θ/2)
aV0 ωm ` sin(θ/2)
=− [ ] sin θ/2
c0 θ/2
aV0 ωm 2
π` sin2 θ/2
∆φ = −
2ωcc0 (πωm /2ωc) θ/2
2
aπV0 ωm `
=− sin2 c(θ/2)
2c0 ωc
where*
2 2
aπV0 ωm `
∆φ = − (nm − n) sin2 c(θ/2)
2c0 πc0
The optimum length for the modulator satisfies
ωm `(n − nm )
= π, or ωm = ωc , θ = π
c0
For LiNbO3 nm = 4.2, n = 2.146, so for example, with ` = 10 mm
n − nm = 2.05, the cutoff frequency is 7.36 Hz.

* sin c(θ/2) = sin(θ/2)/ sin θ/2.