You are on page 1of 6

9.4.

2 Electro-optic coupled wave modulators


The passive directional coupler allows a fixed proportion of the power in one waveguide
to be transferred to another guide. This is readily adapted to provide the controlled
switching of power from one waveguide to another by using the electro-optic effect
to control the phase-matching of the waves in the coupled guides. The basic modulator
structure is shown in Figure 9.5(a). The refractive indices of the 1wo guides are
altered differentially by applying a modulating field to the electrodes, with the actual
arrangement of electrodes and orientation of the crystal substrate determined by
what electro-optic material is used. What is required is that the field components
controlling the refractive index of the guides through the electro-optic effect are
oriented in opposite directions from each other for the two guides. This is achieved
with LiNbO3 when the crystal is oriented with the x. direction out of the plane of
the substrate and coupling is achieved through the Xy3 coefficient.

Gambar chapter 9

294 Chapter 9
If the two waveguides are identical and there is no modulating field present then
there will be perfect. phase-match between the waveguides. When an optical input is
supplied to guide a, and there is no input to guide b, there is progressive transfer of
energy from guide a to guide b as the input wave travels along the coupling region
until the distance traveled is z = xl2C, at which point all the energy has been transferred
to guide b. If the length of the coupling region is made equal to xl2C, the device then
acts as a switch in the same way as the passive switch does in Section 9.4.2.
As soon as a voltage is applied to the modulator electrodes, the electric field
below the electrodes produces an imbalance in the propagation velocities in the two
waveguides and phase-match is lost. For a small mismatch AB,the effect is to reduce
the power transferred by a small amount so that imperfect switching results. For a
large mismatch virtually no power is transferred out of the input guide and negligible
switching results. Thus if the length of rhe coupling region is set to L = r/2C, and a
rectangular wave is applied to the electrodes, a switching modulator results.
In principle, it is possible to obtain complete return of the input power to the
input guide by adjusting the magnitude of the applied modulating field. To see this we
return to (9.25a):

from which it can be seen that in the unmatched condition when /B * 0 the power coupled
into the second guide is 0 when

so that L can be chosen to give complete transfer of power to the second guide
output in the absence of a modulating voltage and AB can be chosen to give zero transfer
of power in the presence of a modulating voltage.
We define the switching crosstalk as the ratio of the power transferred to the
second guide in the presence of a modulating voltage, to the power transferred in the
absence of the modulating voltage. Crosstalk levels better than -30 db are achievable
in real devices.
The switching voltage is not very critical to obtaining good crosstalk performance,
but the technology of matching the physical lengrh of the coupling region to the
coupling coefficient C is critical and difficult. A simple modification of the basic
modulator geometryl'2 to that in Figure 9.5(b) removes this critical dependence on
processing technology. This form of modulator is known as the reverse AB modulator.
The electrodes are split into two sections so that in guide a AB changes sign from positive
to negative when the input wave reaches the halfway point along the coupling region,
and the wave traveling in guide b sees the opposite change in AB as it travels through
the coupling region.
To see how this modulator works we rerurn ro (9.24), (9.25) and (9.28) to
describe the coupling between guides.

When two guide structures are connected in tandem, as in the reverse /Bmodulator, the
overall system is described by

lfi []l =lzfr: fi,"l;?: :flIllli I (e.s6)

*6"ts 15(l)l describes the first half of the modulator and lG(2)l describes rhe second
half. It is convenient to rewrite these equations suppressing the phase terms in the G
matrices, by defining variables R(z) and S(z) as

We now define two alternative states € and @ to describe the fully unswitched and
fully switched states of the reverse /Bmodulator. The modulator is in the @ state when

A graphic representation of these states is provided by the switching diagrams


shown in Figure 9.6. It no modulating voltage is applied to the modulator and AF = 0,
there will be complete switching when L is equal to the characteristic length Ls = tr/2C,
as illustrated in Figure 9.6(a). When /p is non-zero because of the application of a
modulating voltage we get the pair of state diagrams for the @ and € states shown in
Figures 9.6(b) and 9.6(c). These are combined to form the switching diagrams of
Figures 9.6(d) and 9.6(e).
The € state diagram is derived from (9.60). There are two conditions that
satisfy this equation. Either Koo = 0 , which requires Ap = 0, and

which has the form shown in Figure 9.6(c).


Combining the two state diagrams together provides the switching diagram for
the modulator. It can be seen that switching to the two states now depends upon /B
as well as the length of the device. It can be seen from Figure 9.6(e) that when the
length of the device is not /," it is possible to switch between the two states by switching
AB from AB, to ABr. Since AB is proportional to the voltage Vmod itfollows that switching
modulation is accomplished by switching between two non-zero modulating voltages.
9.4.3 lnterferometric electro-optic modulators
An alternative method of providing modulation is by using a waveguide interferometer.
The basic waveguide interferometer is shown in Figure 9.7. The passive interferometer
is very simple. A single input is split between two waveguides;.,spaced so that
evanescent coupling does not take place. The two outputs are added together in a
single output guide where the two waves interfere, with the amount of interference
depending upon the difference in propagation time along ttre two limbs. This interference
may be intrinsic, due to a difference in the static propagation coefficient or a difference
in the lengths of the two guides; or it may be due to the two waveguides experiencing
a different environment. For example, we will see in Chapter l0 how a rotational
environment can change the interference and produce an output from an interferometer
that depends upon the rotation rate. In the context of modulation, the propagation,constant of the two
guides can be changed dif rentially by using the electro-optic

effect to change the guide refractive indices.


The interferometric electro-optic modulator is illustrated in Figure 9.5(c), where
it is contrasted with the coupled wave modulator structure. A major advantage of the
interferometric modulator relative to the coupled wave modulator is that the length
of the device and the maintenance of identical propagation properties in the two
waveguides is not critical, since a DC bias can be applied to equalise the phase
difference of the two limbs in the absence of a modulation voltage. This avoids the
need for the more complex geometries such as the reverse /Bstructure discussed in
Section 9.4.2. On the debit side the interferometric modulator is more sensitive to
external environmental changes such as temperature gradients, and requires close
control of bias voltages.
Interferometric modulators using LiNbO3
been reported, and they have a considerable
optical computing.
9.4.4 Traveling wave modulators
LiNbO3 is the preferred material for coupled wave and interferometric modulators
because of its high electro-optic effect, where it outperforms GaAs and other III-V
compounds. However, there is another type of modulator where GaAs outperforms
LiNbO3. This is the waveguide form oi the bulk traveling wave modulators discussed
with bandwidths up to 40 GHz have
future in optical communication and

9.4.4 Traveling wave modulators


LiNbO3 is the preferred material for coupled wave and interferometric modulators
because of its high electro-optic effect, where it outperforms GaAs and other III-V
compounds. However, there is another type of modulator where GaAs outperforms
LiNbO3. This is the waveguide form oi the bulk traveling wave modulators discussed
with bandwidths up to 40 GHz have
future in optical communication and
Coupled wave and other waveguide devices 299
in Chapter 4. As we saw in that chapter, the modulation bandwidth for bulk devices
is limited by the time the optical wave takes to traverse the modulator electrodes. In
adapting the bulk geometry to a waveguide geometry, the modulation technique can
be modified to overcome transit time effects by launching a modulating wave along
electrodes attached to an optical waveguide, with the optic;l wave and thl modulating
wave both traveling in the same direction. In this way modulation bandwidth extendin!
into the microwave region can be achieved and bandwidths up ro 2AGHz have been
reported' If the optical wave and the microwave modulating wave travel at the same
velocity no transit time limitations are present. In practice the t*o waves travel with
different velocities so that there are residual transit time problems. The velocity of
the optical and microwave waves are much closer for GaAs structures than they are
for LiNbO3 structures and this is what gives GaAs the edge.
In bulk modulators the dependence of the velocity of ordinary and extraordinary
waves upon the appropriate electro-optic tensor components is exploited to changl
the polarization at the output. This is then converted to amplitude modulation by
using a crossed analyser. In the waveguide form it is the dependence of the velocity
of TE and TM modes upon the electro-optic coefficienrs thai is exploited to producl
a polarization change in the output. Again an external crossed analyser is required to
convert the polarization modulation into amplitude modulation, and this is a disadvanage
that this form of modulator has with respect to coupled wave and interferometric
modulators.
There are several different configurations for traveling wave modulators,3 two of
which are shown in Figure 9.8. The basic principle is the same for both structures,
but the modulating fields are oriented in different directions. For the simpler arrangement in Figure 9.8(a) the
optical waveguide consists of a thin film of intrinsic GaAs
sandwiched between n and p layers of GaAlAs all epitaxially deposited onto a GaAs
substrate. The dimensions of the guide are such that only a single TM and a single
TE mode can propagate. A ground strip and a microstrip of metal are deposited on
the bottom and top of the sandwich. The microstrip typically has a width oi the order
of 10 pm, and a length of the order of l0 mm, and constitutes a long pin strip diode.
The most convenient orientation of the substrate crystal structure is 001, so that
the electric modulating field is directed in the 001 direction, and there is a free
choice for propagation direction in the 001 plane. To obtain maximum electro-optic
interaction the same propagation direction is chosen as the one chosen when using
GaAs as a transverse bulk modulator (see Chapter 4). This propagation direction is
parallel to the T l0 axis and is labelled xi in Figure 9.8. From Section +.:.: ttre permittiviry
tensor in coordinates xi,xi,xi is:

It can be seen that for propagation in the xi


with its electric field in rhe xi direcrion, can
direction, the velocity of a TE mode,
be changed by Eg, whilst the velocity
of the TM mode cannot.
The modulating field E3 consists of a DC biasing component and a broadband

(a) Modulator with field directed along x3


arah keluar dari pesawat substrat dan kopling dicapai melalui koefisien Xy3

You might also like