Modeling and simulation of temporal and temperature drift for the development of an accurate ISFET SPICE macromodel

Modeling of non-idealities in ion-sensitive field-effect transistors (ISFET) is crucial for obtaining precise pH-sensing characteristics. This paper presents an accurate Simulation Program with Integrated Circuit Emphasis (SPICE) macromodel of a \(\hbox {Si}_{3}\hbox {N}_{4}\)–gate ISFET pH sensor which takes into account the temperature and temporal drift. The robust model includes surface site densities of both the silanol and primary amine sites, which were approximated in previously reported ISFET macromodels. We use a thermodynamic model to incorporate the variation of dissociation constants with temperature to achieve a better fit with experimental data reported in the literature. Several regression iterations are performed, and an average adjusted R-squared value of 0.9992 is obtained. The transfer characteristics of the device obtained from the extracted parameters are in good agreement with the experimentally reported values, with an error of 3.53% and 2.29% between the reported and modeled isothermal points for gate voltage (\({V}_{\mathrm{GS}}\)) and drain current (\({I}_{\mathrm{DS}}\)), respectively. The developed macromodel is imported in a constant voltage/constant current (CVCC) readout circuit as a subcircuit block and simulated in a temperature range of 15–\(45\, ^{\circ }\hbox {C}\). Simulations of the CVCC readout circuit show that operating the device near the isothermal point reduces the temperature drift to approximately 0.015 pH for the pH range of human blood, i.e., pH 7.35 to 7.45. Temporal drift modeling and simulation are performed, and show a stretched exponential dependence on time, which is in good agreement with experimental results reported in the literature. Finally, we demonstrate the combined effect of temporal and temperature drift, where it is found that the temporal drift rate increases with temperature as a stretched exponential function, validating the experimentally reported values.