Comparison of Thermal Characterization and Simulation Results

The circular microhotplate was thermally characterized, and the results were compared with simulations carried out according to the approach discussed in Chap. 3. Applying FEM simulations as described in Sect. 3.3 generate a temperature field, and the temperature in the membrane center represents the overall membrane temperature according to Eq. (3.21). The values that have been used for the simulation are summarized in Table 4.2. 

The relationship between the temperature difference, AT, and the input power is shown in Fig. 4.5 for microhotplate simulations and measurements. The simulated values are plotted together with the mean value of the experimental data for a set of three hotplates of the same wafer. The experimental curve was fitted with a second-order polynomial according to Eq. (3.24). As a result of the curve fit, the thermal resistance at room temperature, tjo, is 5.8 °C/mW with a standard deviation of 0.2 °C/mW, which is mainly due to variations in the etching process. 

The experimentally determined time constant was Tq = 9.7 0.2 ms. The calculated time constant is approximately 11% higher than the measured one. This is a good agreement given the fact that an error of 15% was assumed as a consequence of the uncertainty of 0.2 MJ/°C m in the heat capacities. 

In conclusion, simulated and measured values are in good agreement, and the achieved accuracy is sufficient for system-level simulations. The experimental results for the characteristic data of a circular microhotplate design are listed in Table 4.3. 

The deviation between the time constants for membrane heating and cooling was measured as well (Eq. (3.37)). The heater of a single microhotplate was driven with a rectangular-shape current pulse. The pulse amplitude was adjusted to produce a temperature rise of 50 °C. In this case the measured time constant for cooling was 

---