Understanding Sensitivity in Nanoscale Sensing Devices

In nanoscale sensors, understanding and predicting sensor sensitivity is challenging as the physical phenomena that govern the transduction mechanism are often highly nonlinear and highly coupled. The sensitivity of a sensor is related to both the magnitude of the analyte-caused signal change and the random error-caused fluctuation of the sensor's output. The extent to which these can be controlled, by carefully designing either the geometric or operating conditions of the sensor, determines the difference in signal output between the presence and absence of the analyte, as well as the impact of random errors on the distribution of these signal outputs. Herein, we use ion-current-rectifying nanopore sensors as a simplified case study to show how geometric and operating parameters can enable sensitivity optimization. Finite element analysis is used to obtain distributions of the sensor output, and then, Sobol analysis is used to highlight the most important contributions to sensor output errors. Furthermore, the magnitude of the signal change is considered alongside the spread of the output to calculate and optimize the sensor sensitivity. We highlight that the most important parameters contributing to the output variance are geometric. We observed that as the sensor is operated at smaller pore radii and lower electrolyte concentrations, the influence of the cone angle errors increases, the influence of the pore radius errors decreases, and the output becomes broader. We also show that the highest sensitivity is expected for larger pores operated at low electrolyte concentrations, and our simulation results are validated by experimental results. Recommendations to achieve optimum sensitivity are given for a range of nanopore scenarios in which ion-rectifying nanopore sensors may be used. This work aims to provide a framework for the nanoscale community to optimize sensitivity using simulations, as the analysis highlighted herein is viable for any system that can be modeled using continuum physics.