D4.2 - Investigation of DRIE etching performance on signal quality of a SOI based pressure sensors for harsh environments
- AMA Conferences 2015
2015-05-19 - 2015-05-21
- Proceedings SENSOR 2015
- D4 - New Technologies
- P. Mackowiak, F. Meinecke, O. Ehrmann, K. Lang - Technical University Berlin (Germany), B. Mukhopadhyay, T. Hoang, Q. Dao - Fraunhofer Institute for Reliability and Microintegration, Berlin (Germany), H. Ngo - University of Applied Sciences, Berlin (Germany)
- 570 - 574
Many silicon microsensors are using a thin silicon membrane with integrated piezoresistors as sensing element . The performance of these micromechanical silicon sensors is strongly depending on the quality of the manufactured membranes such as geometrical parameters, roughness of membrane surface, uniformity of thickness etc..
Silicon membrane structures for sensors and actuators could be realized by using bulk micromachining technologies like KOH, TMAH or DRIE (Deep reactive Ion Etching). DRIE method is a well-known and very often used etching method in manufacturing of silicon sensors as it has many advantages such easy to use, no contact with etching solutions etc. . The main disadvantages of this method are the non-uniformity of the etching rate over wafer, aspect ratio depending etching rate, tilting effect and notching .
Hence the signal quality of membrane-based MEMS sensors largely depends on his manufacturing process. Despite a maximum process control, it is not always possible to adjust the process parameters so that ideal manufacturing yield can be achieved. The dry etching does not deliver a fully homogenous result throughout the whole wafer. Caused by the inhomogeneity of the etching process different sensor signals are generated due to
chip position on the wafer. In order to understand the influence of the inhomogeneity effect of dry etching on the sensor signal, the processed wafer has been fully measured using a profilometer and a mathematical model has been generated. Knowing the sensor position and the etching profile it is possible to reduce the distortion, predict the sensor signal and perform proper signal compensation.