196 Semom and Actuators A, 43 (1994) 196-201 A micromachined pressure sensor with fiber-optic interferometric readout M.A. Chan BioM edical Engineering Graduate Group, Univets@ zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHG of California, Davis, CA 95616 (USA) S.D. Collins and R.L. Smith Depaninent of Electrical and Computer Engineering, Univ~ity of California, Dawk, CA 95616 (USA) zyxwvutsrqponmlkjihgfedcbaZYX Abstract A high sensitivity, batch fabricated, fiber-optic pressure sensor has been fabricated using silicon micromachining technology. The transducer consists of a fiber positioning v-groove, a 45” stationary mirror and a silicon membrane, micromachined in silicon by anisotropic etching in KOH solution, and a single mode optical fiber. The membrane and optical fiber end form a Fabry-Perot cavity whose length varies with pressure. The generated optical interference fringes are used to detect and measure the change in membrane deflection. Pressure range of operation is dictated by the thickness, size and material of the membrane. The sensor described here was designed for low pressure range (O-25 mm Hg) applications. Temperature sensitivity and stability problems which are commonly encountered with currently available piezoresistive and capacitive pressure sensors are significantly reduced by the inherent differential nature of interferometric measurement and the use of all silicon construction. The fabrication, packaging and testing of the sensor are described in this paper. The performance of the sensor was evaluated and found to compare favorably with theoretical predictions. Introduction Pressure sensing is currently the most lucrative market for solid state microsensors. Some of the more common applications include biomedical blood pressure sensing, industrial process monitoring and automotive engine control. However, there remains tremendous demand for pressure sensors in other areas or specialized niches which are outside the functional limits of current solid state pressure sensors. One such market is pressure monitoring at elevated temperatures where electronic pressure sensors fail to function. Another is biomedical, low pressure range monitoring, where signal to noise ratio, cost or size are excluding factors. Commercial, solid-state pressure sensors can be classified into one of two general categories: piezoresistive or capacitive. Although both types rely on the deflection of a dia- phragm to measure pressure, the transduction mech- anism to convert the magnitude of the diaphragm deflection into a quantifiable output differs. Piezoresistive pressure sensors [l, 21 are the most widely used. Their detection mechanism employs the well known piezoresistance of silicon (strain-induced change in resistivity) to transduce diaphragm deflection into an electronic signal. Piezoresistive pressure sensors are relatively sensitive (typically 10-100 ppm/mm Hg) [3], have a low fabrication cost and are reasonably well behaved and stable. Their principal shortcoming is an inordinately high and non-linear sensitivity to temper- ature. For example, it is well known that diffused silicon piezoresistors have at least six different temperature drift mechanisms [3], some of which are extremely non- linear, making temperature compensation difficult. Sen- sitivity at low pressures can be accomplished by reducing the membrane thickness, but this results in an increased fragility and complexity of fabrication and associated cost. In a capacitive pressure sensor, the capacitance be- tween a deflected diaphragm and a stationary ‘back plate’ is used to quantify the pressure dependent dia- phragm deflection. Capacitive pressure transducers gen- erally have higher sensitivity and lower temperature dependence and drift than their piezoresistive coun- terparts [4]. Unfortunately, the change in capacitance for a solid state microsensor is extremely small, typically in the order of lo-” F, and necessitates the integration of sensor and detection circuitry for meaningful mea- surements. With integration comes increased unit cost and fabrication difficulties. In addition, any advantage gained in using the low temperature dependence of the capacitive method is usually lost in the inherent O!X%-4247194/$07.00 0 1994 Elsevier Science S.A. All rights resewed SSDI 0924-4247(93)00694-Y