Piezoresistive silicon pressure sensor samples were thermally cycled after being consecutively packaged to three different levels. These started with the absolute minimum to allow measurement of the output and with each subsequent level incorporating additional packaging elements within the build. Fitting the data to a mathematical function was necessary both to correct for any testing uncertainties within the pressure and temperature controllers, and to enable the identification and quantification of any hysteresis. Without being subjected to any previous thermal preconditioning, the sensors were characterized over three different temperature ranges and for multiple cycles, in order to determine the relative contributions of each packaging level toward thermal hysteresis. After reaching a stabilised hysteretic behaviour, 88.5% of the thermal hysteresis was determined to be related to the bond pads and wire bonds, which is likely to be due to the large thermal mismatch between the silicon and bond pad metallisation. The fluid-fill and isolation membrane contributed just 7.2% of the total hysteresis and the remaining 4.3% was related to the adhesive used for attachment of the sensing element to the housing. This novel sequential packaging evaluation methodology is independent of sensor design and is useful in identifying those packaging elements contributing the most to hysteresis.

Micro-electro-mechanical systems (MEMS) have allowed high precision pressure sensors to be manufactured at low cost, but they typically exhibit some level of hysteresis behaviour, which is believed to be primarily due to the materials used to package them. In order to evaluate the effects of the different packaging materials and processes, in this study, the performance of piezoresistive absolute pressure MEMS was measured for each of the first two packaging stages of the product assembly process. Sensing elements first packaged in a "floating exposed" configuration and then in an "attached exposed" configuration were compared. These two different packaging configurations exhibited distinct hysteresis behaviour when thermally cycled. After applying a third order polynomial fit to the raw data, stabilized and moving hysteresis loops were observed for the floating exposed and attached exposed configurations respectively. The stabilized hysteresis is believed to originate from the inelastic behaviour of the die metallisation, whereas the nonstabilized behaviour stems from the adhesive used to attach the die to the housing.