We introduce a novel method of modeling PLZT phase modulators.
Traditionally, modeling has been based upon fitting the constant
quadratic electro-optic coefficient to empirical data. Our characterization
has shown that the electro-optic coefficient is not a constant
and that the electro-optic effect saturates at electric field
strengths that exist in standard surface electrode device configurations.
We have also found that the additional effects of light scattering
and depolarization, which depend on the strength of the applied
electric field, are significant factors for modeling device design
and optimization.
Keywords: PLZT, electro-optic effect, phase modulators,
finite element analysis, depolarization, scattering.
PLZT is an excellent material choice for use in spatial light
modulators (SLM) due to it's large electro-optic effect and low
absorption for thin wafers1. PLZT ceramics are used
as transverse electro-optic modulators where the electric field
is applied using interdigital surface electrodes (ISE). Such electro-optic
devices are modeled based upon the quadratic electro-optic effect2-6
as well as it's combination with the linear electro-optic effect7.
However these models do not accurately predict the performance
of an ISE device fabricated at UCSD8. In order to more
accurately model such devices we experimentally characterized
PLZT's electro-optic properties. Then we used finite element analysis
(FEA) to characterize the field distributions for ISE devices.
Finally, by combining the electric field values provided by FEA
with the experimental electro-optic data, we were able to predict
the performance of our ISE device. Although this methodology was
carried out with PLZT 9.0/65/35 material, it can be applied to
modeling electro-optic devices with arbitrary choice of material,
electrode structure and geometry.
In the following section we will review the basic theory of quadratic
electro-optic materials, as well as formulate the methodology
for characterization of PLZT electro-optic material. In section
3, we will apply FEA to model the electric field induced by an
electric potential difference between electrodes of ISE devices.
The results of section 2 are integrated into the FEA model to
determine the relationship between the change in relative phase
of two orthogonally polarized components of an incident beam and
the externally applied electric field. In section 4 we compare
this prediction with the actual values taken from the fabricated
ISE device. Conclusions and future directions are discussed in
section 5.
G. Haertling and C. Land initiated extensive studies characterizing PLZT ceramics1. They found that thin wafers with compositions containing greater than 8 at.% La had strong electro-optic properties with transmission values of close to 100%. At room temperature, PLZT is isotropic due to it's cubic crystallographic structure. When an external electric field is applied the PLZT material becomes polarized, demonstrating anisotropic optical characteristics. This same behavior is seen in crystals such as BaTiO3 (in it's cubic form) that exhibit primarily third order non-linear optical properties which in turn lead to quadratic electro-optic effects. Assuming PLZT follows this uniaxial crystal model, the optic axis will be determined by the direction of an externally applied electric field. The induced ordinary and extraordinary index of refraction is determined by
where, R12 and R11
are the quadratic electro-optic coefficients, n is the
refractive index of PLZT, and (E)
is the induced optical birefringence. For PLZT with 9.0 at.% La
the accepted value13 for R is approximately
3.81016 m2/V2. Haertling and
Land1 noticed that the induced optical birefringence
saturates with an externally applied electric field reaching a
maximum7 value of 1.110-2. More recently,
M. Title5 mentioned the saturation effect in modeling
embedded electrode PLZT modulators.
Figure 1.
(a) For ISE on PLZT the index ellipsoid follows the tangent of
the curved E-field lines. (b) PLZT wafer, 300 m thick and 2.01
mm wide, placed between gold coated copper plates. The plates
were inserted into a Teflon base to ensure good electrical isolation
and the cuvet was filled with mineral oil to prevent current arcing
due to exceeding the breakdown voltage of air.
A typical ISE device is constructed of stripe-shaped metal electrodes of width d and length L. An applied voltage across such electrodes creates curved lines of electric field within the PLZT (see Figure 1a). Since the optic axis follows the direction of the applied electric field, the axis orientation will vary as a function of position within the PLZT. Assuming the electrode length, the index of refraction parallel and perpendicular to the surface electrodes can be approximated5 by
Defining the relative change in index, we determine the relative phase between the parallel and perpendicular polarization components of an optical beam passing through the device by
where is the wavelength in vacuum and l is the thickness
of the electrooptic material.
In the following we will experimentally determine the phase relationship
provided by Equation 5. By placing the PLZT wafer between large
parallel metal plates (see Figure 1b) a homogeneous electric field,
perpendicular to the plates (i.e. = 0), will exist within the
dielectric. The field strength is a function of the potential
difference between the plates (i.e., where
d is the distance between the plates). To measure the induced
phase retardation we illuminate the PLZT sample using a normally
incident HeNe laser beam. The incident beam is linearly polarized
at 45 with respect to the direction of the applied electric field,
providing two equal components parallel and perpendicular with
respect to the electrode structure. As the voltage, and thus the
electric field strength, is varied, there will be a change in
the relative phase between these two components.
By placing a crossed polarizer at the output of an ideal phase modulating device, the light intensity will vary as a function of the relative phase according9 to the relation
where is the relative phase, and a2 and b2
are the transmittances for the two orthogonal components of the
light. We measure the transmittance, T, through a crossed
polarizer, as well as the transmittances through vertically and
horizontally oriented polarizers, a2 and b2
respectively, as functions of the electric field. Then,
by solving Equation 6 for , we expect to find the relative phase
as a function of electric field. However, using this method, the
resultant phase is not continuous when a PLZT based device
is being investigated. This is due to depolarization effects observed
in PLZT phase modulators10.
By introducing a depolarizing term into the transmittance of the
orthogonal components in Equation 6, we obtain
where ca and cb are the fractions
of depolarized light corresponding to the incident vertically
and horizontally polarized components. Defining,
we get the relationship
where we also assume that = c.
Using the measured values for T, A2,
B2 and curve fit for c2, we
solve Equation 8 that provides continuous relative phase (see
Figure 2b).
Figure 2.
(a) Transmittance T through crossed polarizers set at 45
and through parallel polarizers set vertically and horizontally,
A2 and B2, as functions of
a horizontally applied electric field. (b) Relative phase as a
function of applied electric field. We also show the best fit
for a quadratic electro-optic coefficient as.
Observing the two curves A2 and B2
of Figure 2a we notice a dramatic intensity drop after the electric
field reaches approximately 7105 V/m. The main reason
for the attenuation is due to scattering effects. This also causes
a corresponding drop in the transmittance, T, through the
crossed polarizers. Furthermore, T varies sinusoidally
with the period varying as a function of electric field. The frequency
of the sine initially increases, whereas at fields above 1106
V/m the frequency decreases. Above the value of 2106
V/m the contrast ratio approaches 1:1, which is primarily due
to the depolarization effects. Due to the need for high contrast
ratios in phase modulators, scattering and depolarization effects
must be factored into the design of such devices.
Figure 2b shows the relative phase vs. electric field, where the
experimental data is determined by solving Equation 8. Below electric
fields of 2105 V/m there is practically no phase change
in the PLZT. Above values of 1106 V/m we observe that
the phase change begins to saturate. Our calculations show that
for PLZT surface electrode devices, there are regions where the
field strengths are on the order of 2106 V/m, and therefore,
the saturation effect must also be taken into consideration in
these devices.
Finite Element Analysis (FEA) is one of several methods3-6,11,12
available for calculating the electric field induced by metal
electrodes. We use Mentat, a commercial FEA program from Marc
Analysis, which provides an excellent tool for mesh generation,
field calculations and visualization (see Figure 3). We used FEA
to determine the electric field distribution in PLZT devices with
surface electrodes. A typical example of Mentat's output is shown
in Figure 3. For this particular ISE configuration, we observe
that 100 m from the surface the electric field strength drops
below the minima required for phase modulation. We also see that
near the edges of the electrodes the field goes beyond the maxima,
i.e. the phase modulation has become constant. Using FEA we are
able to calculate the electric field strength and direction, at
any point, for any configuration of electrodes.
To find an explicit relationship between the magnitude and the direction of the electric field and the relative phase we substitute Eqs. 1, 2 and 3 into Equation 5 and obtain
Using Taylor series expansion this can be approximated by ( 10)
where we define the birefringence as quadratic, but with the electro-optic coefficient also being a function of the electric field. This can be more simply stated as
where is the phase function from the experimental curve of Figure 2b. Applying the electric field results from FEA we find
where N is the number of finite elements that the light
ray passes through and li is the height of each
element.
Mentat FEA software calculates the x and y components of the electric
field (in this case we are using a two dimensional model) for
four integration points for each element. Taking the average over
each element we use the magnitude of the field (i.e. ),
the phase function and the orientation of the index ellipsoid
from Equation 4 to get the relative phase change for each element.
Integrating the change in phase passing through a column of elements,
expressed by Equation 12, we find the change in phase for a light
ray passing through a line of elements (see the dotted line in
Figure 3). Looking at the series of columns across the electrode
gap gives us a phase profile for a plane wave passing through
the device.
Using the modeling procedures discussed in section 3 we determine
the calculated phase distribution for a simulated ISE device and
compare it to that found experimentally using a fabricated device.
The FEA modeling and the experimentally measured results are found
to be in good agreement for voltages of less than
and between electrode gaps of about 500 m (see Figure 4a). For
narrow gaps and higher voltages (see Figure 4b) there is a difference
between the modeling and experimental results. One possible explanation
would be that the electric field strength is weaker then the value
calculated by the FEA model due to screening effects. These screening
effects occur when free carriers (photo-induced or due to surface
states) create a space-charge distribution near the electrodes.
Our future work will entail investigating these phenomena.
Figure 4. Comparison of FEA model to experimental data for transmission through crossed polarizers. (a) Shows a relatively good fit for V and V2 (i.e. the first maxima and minima) for an electrode gap of 500 m. Whereas (b) shows a poor fit with a narrow electrode gap of 50 m.
Currently, in our model, we are compensating for this 'weakening'
effect by introducing a constant factor. Consequently, we are
able to accurately model the behavior of fabricated devices operated
at voltages less than (see Figure 5a).
Notice that the 'integrated phase' depends linearly on the applied
voltage in contrast to the quadratic behavior predicted by previous
models.
For surface electrode based devices, the two parameters that are
most important are the electrode width and the size of the gap
between electrodes. By varying the gap size, and holding the width
(160 m) and voltage (150 V) constant, in our FEA model we have
determined that as the gap decreases the phase shift increases
proportionately. This is expected based upon the linear relationship
of phase to electric field, since the electric field strength
scales proportionately with gap size. However, when the gap size
becomes smaller than 40 m there is very little increase in phase
modulation. This is due to electric field values being above the
'saturation' level. We also find that as the electrodes increase
in size there is a corresponding increase in phase shift, but
electrodes wider than 160 m were also of little benefit (see Figure
5b). The conclusion is that the optimal configuration for this
material is a 40 m gap between 160 m wide electrodes. It was experimentally
determined8 that for interdigital surface electrodes
on PLZT 8.8/65/35 the optimum configuration is 160 m electrode
widths with 40 m between electrode gaps. From this excellent agreement
we conclude that optimization of design parameters can be done
successfully using our FEA model.
Figure 5.
(a) Comparison of the FEA model with experimental
data for phase as a function of applied voltage for surface electrodes
160 m wide with a 40 m gap (after including a compensating constant
factor). (b) Holding the gap width (40 m) and voltage (150 V)
constant, the results of the FEA simulation show that as electrode
width approaches 160 m the increase in phase modulation slows
down. (c) The gradient of the phase front for various applied
voltages for etched electrodes 40 m deep. A relatively flat phase
profile across the entire gap can be realized around 50 Volts.
With our general modeling approach, by varying other parameters
within our FEA model we can design and evaluate many device configurations
without having to fabricate many devices. In the following we
will briefly discuss the optimization of two independent design
criteria, embedded electrode geometry and phase uniformity. One
of the limitations in using ISE devices is their low transmittance
due to a small gap to electrode ratio, i.e. small fill factor.
In contrast to surface electrodes, much larger fill factors can
be realized by using electrodes that are embedded into the surface.
To find the effect of using different etch depths we hold other
parameters constant and change the scale of the electrode structure.
Using a fill factor of 80% and assuming that a 5:1 aspect ratio
is possible in etching PLZT substrate, we observe that V decreases
steadily as the gap gets smaller (see Table 1), indicating that
the decrease in switching energy is proportional to the decrease
in gap size. Under the constraint of an 80% fill factor we observe
that the linear relationship no longer holds for a gap size below
40 m. For small geometries the electric fields need to be high
due to the short active modulation path length. Consequently,
we once again observe the effects of 'saturation' of the phase
modulation.
Another important characteristic of an electro-optic modulator
is the homogeneity of the modulated phase front. Our FEA modeling
allows us to calculate the phase distribution across the aperture
between the electrodes. A plane wave passing through the gap between
electrodes will attain a phase curvature across the aperture depending
on the field distribution. According to our model, for a 50 m
wide gap at 150 Volts, there is a phase difference of approximately
0.7 radians from center to edge. By simulating various electrode
geometries and applied voltages we find that by using electrodes
60 m wide, spaced 40 m apart and etched 40 m deep the wave front
has an almost perfectly flat phase profile at 53.6 Volts (see
Figure 5c). It is important to note here that we have been analyzing
one characteristic at a time. In order to find an optimum device
configuration, many coupled characteristics must be taken into
account and weighted according to specific device requirements.
We have used a uniform applied electric field within PLZT in order
to experimentally characterize the electro-optic response of the
material. This characterization has highlighted the fact that
scattering and depolarization effects need to be considered in
determining the phase function. Furthermore, electric field distributions
obtained using various electrode configurations have been calculated
using FEA. These resultant electric fields were integrated with
the phase function of the material to determine the characteristic
phase modulation of an ISE electro-optic device. The calculated
strength of the electric fields has shown us that 'saturation'
of phase modulation needs to be considered in device design. We
have also found that an electric field 'weakening' effect needs
to be factored into the model and in the future we will investigate
the cause of this phenomena. After compensating for these various
effects we are able to model the behavior of a device as a function
of a variety of parameters. We have shown that this model is useful
in optimizing individually such device characteristics as increased
transmittance and homogeneity of the phase front. Multiple characteristics
of such devices that are mutually coupled can also be optimized.
We wish to thank the invaluable help of Rebecca Bussjager at Rome
Labs, Eddie Rezler from Marc Analysis as well as Rong-Chung Tyan
and James Thomas at UCSD. This work is supported by Air Force
Rome Laboratory.
1. G.H. Haertling and C.E. Land, 'Hot-Pressed (Pb,La)(Zr,Ti)O3 Ferroelectric Ceramics for Electrooptic Applications', J. Am. Ceram. Soc., Vol. 54, No. 1, pp. 1-10, 1971.
2. R. Viennet, 'Driving Voltage Calculation for a Ferroelectric Display Device', J. de Math. et de Phys. Appliquées, Vol. 29, pp. 715-22, 1978.
3. E.E. Klotin'sh, et al., 'Geometrical Optics of an Electrically Controlled Phase Plate made of PLZT-10 Ferroelectric Ceramic', Avtometriya, No. 6, pp. 68-72, 1984.
4. K. Tanaka, et al., 'Analyses of PLZT Electrooptic Shutter and Shutter Array', Jap. J. Appl. Phys., Vol. 24, No. 2, pp. 177-82, February 1985.
5. M. Title and S.H. Lee, 'Modeling and characterization of embedded electrode performance in transverse electrooptic modulators', Appl. Opt., Vol. 29, No. 1, pp. 85-98, January 1990.
6. Q. Wang Song, P.J. Talbot and J.H. Maurice, 'PLZT based high-efficiency electro-optic grating for optical switching', J. Modern Optics, Vol. 41, No. 4, pp. 717-27, 1994.
7. B. Mansoorian, Design and Characterization of Flip-Chip Bonded Si/PLZT Smart Pixels (Ph.D. Thesis), Ch. 2, University of California, San Diego, 1994.
8. J. Thomas and Y. Fainman, 'Programmable diffractive optical elements using a multichannel lanthanum-modified lead zirconate titanate phase modulator', Opt. Lett., Vol. 20, No. 13, pp. 1510-12, July 1995.
9. A. Yariv and P. Yeh, Optical Waves in Crystals, Ch. 5, John Wiley and Sons, New York, 1984.
10. M. Ivey, 'Birefringent Light Scattering in PLZT Ceramics', IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, Vol. 38, No. 6, pp. 579-84, IEEE, New York, 1991.
11. H. Elgan, 'Excitation of Elastic Surface Waves by Spatial Harmonics of Interdigital Transducers', IEEE Transactions on Electron Devices, Vol. 16, No. 16, pp. 1014-17, IEEE, New York, 1969.
12. A. Y. Wu, T. C. Chen and H.Y. Chen, 'Model of electrooptic effects by green's function and summary representation: applications to bulk and thin film PLZT displays and spatial light modulators', Proceedings of Eighth IEEE International Symposium on Applications of Ferroelectrics, Eds. M. Liu, A. Safari, A. Kingon and G. Haertling, pp. 600-3, IEEE, New York, 1992.
13. G. Haertling, 'Electro-optic ceramics and devices', Electronic
Ceramics, Ed. L. Levinson, pp. 371-492, M. Dekker, New York,
1988.