RF Detection for Low Frequency cMUTS and its Comparison to Traditional Detection zyxw A. S. Ergun, S. T. Hansen, and B.T. Khuri-Yakub *Edward L. Ginzton Laboratory, Stanford University, Stanford, CA 94305-4085. zyx Abstract- Capacitive micromachined transducers have long been used as generators and detectors of ultrasound both in air and immersion applications [l-31. Typically, the frequency of operation has been in the MHz range. How- ever, they can still be operated at very low frequencies, where their mechanical sensitivity is poor, but the frequency response is flat. In this paper we show that, together with the RF detection method that provides a very sensitive and broadband acoustic detection zyxwvutsr [4], [5], it is possible to oper- ate the cMUTs at very low frequencies, which covers the microphone and sonar applications. Keywords- microphone, RF Detection, Capacitive Trans- ducers. I. INTRODUCTION Capacitive microphones are made up of a diaphragm or a plurality of diaphragms that are suspended over a conduc- tive backplate. The diaphragm is coated with a conductive electrode to create a parallel plate capacitor. The value of the capacitor is determined by the area of the electrode zyxwvu as well as the gap between the diaphragm and the backplate. The detection of sound with a capacitive microphone de- pends on the change in its capacitance when the diaphragm starts to vibrate upon receiving a sound pressure. Tradi- tionally, the change in the capacitance is detected by mea- suring either the output current under a constant bias volt- age or the output voltage under a constant-charge on the diaphragm electrode. There are two significant figures that determine the sen- sitivity of a microphone. The first one is the open circuit output signal generated in volts per unit sound pressure im- pinging on the microphone in Pascals. This sensitivity fig- ure can be decomposed into two as the product of mechan- ical sensitivity in angstroms per Pascal input pressure, and electrical sensitivity in volts per angstrom displacement. The research on capacitive microphones has been primar- ily based on optimizing these two components to maximize the open circuit output voltage per Pascal input pressure. However, this figure of sensitivity is not sufficient by itself to show the performance of a microphone because it does not include the noise generated by the microphone and the detection circuitry. Therefore, we consider the second fig- ure of sensitivity, which is the signal-to-noise ratio (SNR) in dBs obtained per Pascal input pressure, as a more com- plete performance meter for a microphone. Note that, the first amplification stage is the most important electronic stage that determines the overall noise performance of the microphone, and the SNR should be measured after the preamplifier. After we introduce our approach to the mi- crophone technology, we will be showing both calculation and measurement results in terms of SNR. 0-7803-6365-5/00/$10.00 zyxwvut 0 2000 IEEE Traditionally, the trend in capacitive microphone tech- nology has been to make the diaphragm softer, larger, and as close to the backplate as possible. The gap is filled with air to prevent the diaphragm collapsing on to the backplate due to the atmospheric pressure. This type of microphone has limitations due to squeeze film effects, air streaming resistance, nonlinearity, and durability. A fine review on capacitive microphones can be found in reference [6] as well as on other types of microphones. 11. A NEW MICROPHONE APPROACH Our approach to microphone technology is different from the traditional approach. Instead of making soft and large diaphragms with air backing, we build the microphone with small membranes and make them stiff enough to sustain the atmospheric pressure. Then, we can evacuate the gap and seal the membranes creating a vacuum between the mem- brane and the backplate [l], [2]. In this way, we eliminate the limitations of traditional microphones that are caused by air backing and soft membrane and create a more ro- bust microphone. Furthermore, these devices have their resonance frequency in the MHz range which means a per- fectly flat frequency response up to a good fraction (80%) of the resonance frequency. However, these advantages come with a disadvantage, which is the significant reduction in the mechanical sensitivity of the microphone. Note that, reduced mechanical sensitivity does not imply reduced me- chanical signal-to-noise ratio. Indeed, since the mechani- cal losses are minimized, the mechanical noise limit of the microphone is also minimized. That means, in an ideal situation with no electronic noise contribution, the SNR obtained from a microphone with our approach would be better than the one obtained with traditional microphones. In a situation where electronic noise is dominant, this may not be possible. The main focus of our work is to increase the electrical sensitivity of the detection circuitry so that the mechanical noise of the microphone dominates the elec- tronic noise. Then, we can detect sound pressure at the mechanical noise limit. To compensate for the poor mechanical sensitivity we boost the electronic sensitivity by using an RF detection method which was initially introduced for capacitive mi- cromachined ultrasonic transducers [4] [5]. In this method, the membranes are attached with a single continuous metal line as shown in Fig. 1, which is actually a transmis- sion line periodically loaded with membrane capacitances. It has a propagation constant that is a function of the membrane capacitance, so one can imagine the modula- tion of the propagation constant as the membranes start 2000 IEEE ULTRASONICS SYMPOSIUM zy - 935