Many today's microsystems like strain-gauge-based piezoresistive pressure sensors contain doped resistors. If one wants to predict correctly the temperature impact on the performance of such devices, the accurate data about the temperature coefficients of resistance (TCR) are essential. Although such data may be calculated using one of the existing mobility models, our experiments showed that we can observe the huge mismatch between the calculated and measured values. Thus, in order to investigate the TCR values, a set of the test structures that contained doped P-type resistors was fabricated. As the TCR value also depends on the doping profile shape, we decided to use the very thin, 340 nm thick SOI wafers in order to fabricate the quasi-uniformly doped silicon layers ranging from 2 × 10 17 at cm −3 to 1.6 × 10 19 at cm −3 . The results showed that the experimental data for the first-order TCR are quite far from the calculated ones especially over the doping range of 10 18 -10 19 at cm −3 and quite close to the experimental ones obtained by Bullis about 50 years ago for bulk silicon. Moreover, for the first time, second-order coefficients that were not very consistent with the calculations were obtained.

Electro-thermal models are commonly used in simulations and designing MEMS devices, also in case of microbolometers. These models allow to estimate sensors performance before fabrication that consequently impacts on fabrication cost. Coupling of electric and thermal domains and building appropriate model for carrying out simulations are crucial to detect all phenomena having impact for Readout Integrated Circuit. This article brings closer to the subject matter of circuit modelling (willingly implemented in MATLAB/SIMULINK or PSPICE tools) and to its theoretical background.