This work is also discussed in our archival publication:
M. S. Islam, E. Tuncer, and D. P. Neikirk, "Accurate Model for Schottky-Contacted Coplanar Waveguide Including Finite Epilayer Resistance Effects," Electronics Letters, vol. 30, pp. 712-713, 1994.
Accurate Quasi-Static Model for Schottky-Contacted Voltage-Controlled
Coplanar Waveguide Phase Shifters
M. Saiful Islam, Emre Tuncer and Dean P. Neikirk
Department of Electrical & Computer Engineering
The University of Texas at Austin
Austin, Texas 78712
Abstract
A new quasi-static model for Schottky-contacted coplanar waveguide (CPW)
on a semiconductor substrate is shown. Comparison between experimental measurements
for a CPW on a GaAs epilayer and calculations show excellent agreement.
The new model includes the effect of the finite resistance of the undepleted
epilayers under the CPW electrodes [1,2].
Introduction: Coplanar waveguide (CPW) on semiconducting substrates has
been extensively studied for possible use in phase shifting applications.
The most common structure consists of CPW electrodes Schottky-contacted
to a doped semiconducting layer on a semi-insulating substrate [3]. Two main techniques for the control
of the propagation constant in such CPWs have been used: optical control
[4] and voltage control [5-8]. For both cases, models of the behavior
of the CPW are usually based on slow-wave effects. Here we report a quasi-static
model for Schottky-contacted CPWs that is in excellent agreement with experimentally
measured data over a wide range of frequency and bias conditions. While
past models have emphasized the impact of depletion layer capacitance on
the slow wave effect, we show that significant sensitivity to bias voltage
is induced by resistance in the epi layer.
Model: A cross-sectional drawing of a Schottky-contacted CPW is shown in
Fig. 1. The CPW center conductor
is 2a wide, the gap between center conductor and ground plane is ,
and the ground plane is w wide. The depleted and undepleted layers under
the CPW electrodes contribute a distributed capacitance and resistance,
respectively, to the admittance per unit length Ytot for
the transmission line. Figure 2
shows the equivalent circuit for Ytot. The normal CPW
capacitance per unit length consists of air-side ( ) and substrate-side
( ) contributions, found using conformal mapping. The distributed
RC circuits formed by the CPW electrodes and the epi layer under them produce
Zgp and Zcntr. The finite resistance
of the undepleted epi layer has been ignored previously; however, it can
be of critical importance in capacitance-voltage measurements on doped epitaxial
layers [9,10]. The sheet resistances
shown in Fig. 1 are
(1)
where the superscript n is either gp, gap, or cntr, for the region under
the ground plane, in the gap between electrodes, or under the center conductor,
respectively. The epi layer is tepi thick, with conductivity
[[sigma]]epi, and the bias-dependent depletion layer is hn
thick. For simplicity, we have assumed a single, uniformly doped epitaxial
layer, on a high resistivity substrate with residual sheet resistance ,
in parallel with the epi layer. A simple depletion approximation is used
to find hn, given the bias condition and doping level in the
epi layer. For the geometry used here, we then find
(2)
where [[omega]] is the angular frequency, [[epsilon]] is the dielectric
constant of the epi layer, n is either gp or cntr, and wn is
the width of the region ( and ). If is small enough
is just the parallel plate capacitance across the depletion region.
However, if becomes large the series resistance prevents the capacitance
from contributing fully. Again referring to Fig. 2, Ytot
for the Schottky-contacted CPW is
(3)
where Rgap is the total gap region resistance, . Finally, the
complex propagation constant is given by , where the series impedance
per unit length Ztot is found using a new, highly accurate
quasi-static technique [11].
The propagation constant can be very sensitive to bias voltage when either
hgp or hcntr are close to the total epi layer thickness
tepi (i.e., the bias voltage has nearly depleted the epi layer)
since small changes in bias induce very large changes in RS (even though
h, and therefore the depletion layer capacitance, changes very little),
inducing large changes in Ytot. This also explains the
extreme sensitivity to optical illumination when biased near full depletion
[12], since very low levels of
illumination can induce large changes in RS.
Results and Conclusions: To verify the accuracy of this model, measurements
have been made on coplanar waveguide fabricated on doped GaAs layers grown
by molecular beam epitaxy on semi-insulating (SI) GaAs wafers. The epi layer
consisted of a 1.5 um thick n-type (4 x 1015 cm-3)
GaAs layer, a 0.1 um thick AlAs layer, and finally a 0.2 um thick GaAs undoped
buffer layer. Accounting for Fermi level pinning at the SI substrate interface,
the effective epi layer thickness used in eq. 1 is tepi = 1
um. In the gap region, Fermi level pinning also depletes 0.5 um from the
front surface, so hgap = 0.5 um. The center conductor
half-width a is 5 um, the gap is 7 um, the ground plane width
w is 500 um, and the CPW metal (silver) thickness is 1 um. Figures
3 (attenuation constant) and 4 (effective refractive index, neff = [[beta]]/[[beta]]o)
show a comparison between the experimental results (measured using an HP
8510B Network Analyzer) and our new model for the center conductor reverse
biased. Excellent agreement between the experimental results and the calculations
over the full frequency range (45 MHz - 40 GHz) is shown; excellent agreement
between measurement and calculation is also obtained when the ground planes
are reverse biased. Again, no fitting factors are used; only the dimensions
of the CPW and the characteristics of the semiconducting substrate are required
for the calculation.
Acknowledgments: This work was sponsored in part by the Joint Services Electronics
Program under grant number AFOSR 49620-92-C-0027 and the Advanced Research
Projects Agency ASEM program.
References
1. M. S. Islam, E. Tuncer, and D.
P. Neikirk, "Accurate Model for Schottky-Contacted Coplanar Waveguide
Including Finite Epilayer Resistance Effects," Electron. Lett.,
vol. 30, pp. 712-713, 1994.
2. S. Islam, "Modeling and Experimental Studies
of Schottky-Contacted Coplanar Waveguide Transmission Lines on Semiconductor
Substrates," 1994, PhD, The University of Texas at Austin.
3. H. Hasegawa and H. Okizaki, "M.I.S.
and Schottky slow-wave coplanar striplines on GaAs substrate," Electron.
Lett., vol. 13, pp. 663-664, 1977.
4. P. Cheung, D. P. Neikirk, and
T. Itoh, "Optically Controlled Coplanar Waveguide Phase Shifters,"
IEEE Trans. Microwave Theory Tech., vol. 38, pp. 586-595, 1990.
5. R. E. Niedert and C. M. Krowne,
"Voltage Variable Microwave Phase Shifter," Electron. Lett.,
vol. 21, pp. 636-638, 1985.
6. C. M. Krowne and R. E. Niedert, "Slow wave
monolithic variable phase shifter with operation into the millimeter wave
wavelength regime," Int. J. Infrared Millimeter Waves, vol.
7, pp. 715-728, 1986.
7. C. M. Krowne and E. J. Cukauskas, "GaAs Slow
Wave Phase Shifter Characteristics at Cryogenic Temperatures," IEEE
Trans. Electron Devices, vol. ED-34, pp. 124-128, 1987.
8. V. M. Hietala, Y. R. Kwon, and K. S. Champlin, "Broadband
Continuously-Variable Phase-Shifter Employing a Distributed Schottky Contact
on Silicon," Electron. Lett., vol. 23, pp. 675-677, 1987.
9. J. H. Wiley and G. L. Miller,
"Series Resistance Effects in Semiconductor CV Profiling," IEEE
Trans. Electron Devices, vol. ED-22, pp. 265-272, 1975.
10. J. D. Wiley, "C-V profiling of GaAs FET films,"
IEEE Trans. Electron Devices, vol. ED-25, pp. 1317-1324, 1978.
11. M. S. Islam, E. Tuncer, and
D. P. Neikirk, "Calculation of Conductor Loss in Coplanar Waveguide
using Conformal Mapping," Electron. Lett., vol. 29, pp. 1189-1191,
1993.
12. M. S. Islam, A. J. Tsao, and
D. P. Neikirk, "GaAs on Quartz Coplanar Waveguide Phase Shifter,"
IEEE Microwave and Guided Wave Lett., vol. 1, pp. 328-330, 1991.
Figure 1: Schottky-contacted
coplanar waveguide on an epitaxial layer. Depletion regions under the electrodes
produce distributed capacitance, while the undepleted epi layer produces
a distributed resistance.
Figure 2: Equivalent
circuit for the shunt admittance per unit length of the coplanar waveguide.
Figure 3: Measured
(solid lines) and modeled results (dashed lines) for attenuation constant
at 0 V, 2 V reverse bias on center conductor, and 7 V reverse bias on center
conductor; at 7 V the epilayer is fully depleted, so hcntr is
taken to be the full SI substrate thickness.
Fig. 4: Measured (solid
lines) and modeled results (dashed lines) for the effective index of refraction
[[beta]]/[[beta]]o (slow wave factor) at 0 V, 2 V reverse bias on center
conductor, and 7 V reverse bias on center conductor.