This work presents a novel design for a micromachined, capacitively
sensed hydrophone. The design consists of a fluid-filled chamber
constrained by two sets of membranes. The "input" membranes are
arrayed around the outside of the circular chamber. Incoming sound
generates a trapped cylindrical wave, creating mechanically amplified
motion of the 1 mm diameter central "sensing" membrane. The membrane
material is a LPCVD nitride/oxide/nitride triple-stack with respective
film thickness 0.1/0.65/0.1 micron. The chamber is filled with 200
cSt viscosity silicone oil. Fluid-filling eases design constraints
associated with submerging the sensor, especially with respect to
exterior mass loading. Both silicon-glass anodic bonding and tin-gold
solder bonding are used to form the structure, including the 5 micron
The fluid-structure system is computationally modeled using both
approximate analytic and numerical techniques. Model results indicate
a 28 dB displacement gain between the motion of the "input"
membranes and the "sensing" membranes. An off-chip charge
amplifier, with a 10 pF integrating capacitor, is used to convert
membrane motion into an electrical signal. Mean measured system
sensitivity is 0.8 mV/Pa (-180 dB re 1 V/microPa) from 300 Hz-15 kHz
with a 1.5 volt applied bias and a 26 dB preamplifier gain. The
predicted low frequency sensitivity is 0.3 mV/Pa. The measured
sensitivity exhibits considerable scatter below 7 kHz, with a standard
deviation of 80%. Laser vibrometry measurements indicate that this
scatter may be caused by compliance of the chip mounting scheme.
Above 10 kHz, the quiescent noise is -100 dB re 1 V/rtHz.
Noise characteristics exhibit a 1/f character below 10 kHz, rising to
a maximum of -50 dB re 1 V/rtHz at 100 Hz.
The mammalian cochlea achieves remarkable acoustic transduction characteristics in a compact and robust design. For this reason, its mechanics have been extensively studied, both mathematically and experimentally. Recently, a number of researchers have attempted to mimic the cochlear function of the basilar membrane in micromachined mechanical devices. This paper presents a design for a silicon cochlea which extends previous work by utilizing a micromachined liquid-filled two duct structure similar to the duct structure of the biological cochlea. Design issues related to both mechanical structure and electrical transduction will be discussed, particularly with regard to optimization of transducer performance. A parallel beam array structure is proposed as a model for an orthotropic membrane. Fabrication procedures and results are also presented. Challenging Fabrication issues related to through-wafer etching, adhesive wafer bonding, device release, and fluid injection are emphasized.