(Invited) A Novel Computational Framework for Simulations of Bio-Field Effect Transistors

Abstract

In this paper, we present the simulation results of a BioFET using a novel simulation methodology implemented in a computational framework called the Biomolecule-Oxide Simulator (BOxSim). Our novel simulation methodology is based on Gouy-Chapman-Stern and the modified Site-binding models [1][2]. Moreover, our “BOxSim” framework is the computational simulation framework that works on the hierarchical inclusion of interdependent functions to operate multiple models simultaneously. BOxSim is flexible simulation environment which considers the problems of reactive sites from oxide surfaces or non-specific bindings along with the noise present in the experimental signal to calibrate and extract the required device output characteristics [3]. It can help in generating the sensor response which can be experimentally calibrated to confirm the surface density of the specific and non-specific binding as a yield of the process while filtering the noise with the help of the developed methodology. BOxSim is designed in a way [shown in the attached block diagram] that can interact with the commercial tool (e.g., Synopsys TCAD) and our inhouse-simulator called Nano-electronic Simulation Software (NESS) to simulate the main device characteristics for designing either Ion-sensitive FET (ISFET) or Chem/Bio-FET based on the application. Industry-standard compact (BSIM) models can also be used to interact with the BOxSim which takes the generated surface potential from the BOxSim simulator and transfers it into a circuit simulator to evaluate various ISFET circuit designs.To show the capabilities of our BOxSim framework, as input we have considered several oxides, amino acids (AA) and short peptides which generate distinct signatures for arch AA or peptide. We have investigated the impact of various high-k dielectric materials [4][5] as the gate oxide on important Figures of Merit (FoM) such as current in the channel of the device, surface potential (Ψ o ), sensitivity and intrinsic buffer capacitance [6]. More importantly, the variation of differential capacitance (CT ) with the second gradient of drain current (d2 I o /dpH2 ) and surface potential (d2 Ψ o /dpH2 ) are used to uniquely identify the signatures of different amino acids or peptides. Fig. 1 shows the characteristics of different oxides with respect to the pH variation. The surface potential of oxides is dependent on the affinity constants and the zero-crossover point represents the zwitterion state. Similarly, we have simulated the detection process of different AA such as Glutamic Acid (E) and Lysine (K) immobilized with carboxyl (C-Imm.) or amine (N-Imm.) terminal [Fig. 2]. The affinity constants, isoelectric point, and density of amino acids can be extracted from the d2Ψo/dpH2 variation with the bulk pH. Other than amino acids, we have shown the distinct features of short a peptide Phenylalanine-Proline (FP) immobilized with carboxyl-terminal and the addition of Glutamic Acid to the sequence [Fig. 3]. The sensor output can be observed as the variation of drain current with the pH for the immobilized peptide sequence on the sensing surface. We have also confirmed the reliability of the designed model by calibrating the simulated results with the experimental data [7] for different physical conditions [Fig. 4]. The proposed method can be helpful in defining an efficient method for label-free protein sequencing.