Silicon Micro-mechanics For Solid State Sensors

Since even before the first successful demonstration of the transistor in 1947, evolving semiconductor processing technology has lent itself well toward use in sensing of various physical phenomena. The first record of silicon used as a sensor was in 1940, when Robert Oh1 of Bell Labs discovered the photoelectric effect in silicon. In 1953, Philco was already using chemical micromachining to shape semiconductors. This paper explores the history of the semiconductor developments which have led to today's state of the art monolithic silicon sensors with on-chip self test, self calibration, and signal conditioning. These devices are part of the continuing evolution of smart monolithic sensors which have been and will continue to be made possible and economical by the continuing development of silicon manufacturing technology. Sensor Backaround There are two classes of silicon sensors considered here. The first utilizes only the electrical properties of silicon, and is included here primarily for reference. The second class of sensors utilizes the mechanical, and sometimes the electrical properties, and is the real subject of the discussion. Within each of these classes, there are several different sensing techniques (piezoresistive, capacitive, etc.), as well as various stimuli (pressure, acceleration, etc.) that may be sensed with each sensing technique. Silicon Sensors Utilizina Electrical ProDerties This class of sensors utilizes the Hall effect for measuring magnetic fields, and the band gap of silicon for measuring heat and light. Although they often utilize special layouts or modified manufacturing processes to enhance the sensitivity to the desired stimulus, these sensors historically have relied upon integrated circuit processing without further machining or mechanical enhancements. As such, they are typically very economically produced, and can inherently combine the functions of circuit and sensor, making them inexpensive and elegant solutions for many sensing and control problems.