Study of a Charge-Based Biosensor and Reconfigurability using BESOI MOSFET

Abstract

Sensors for biological components detection have been attracting attention due to the numerous application possibilities, for instance, the food industry and the point of care health(1). In this context, field-effect-transistors (FET) based sensors bring to the biosensing field: a reduced mass production cost, smaller size, signal processing circuit integration capability and higher sensitivity(2).The back-enhanced-silicon-on-insulator (BESOI) metal-oxide-semiconductor (MOS) FET(3) was developed and fabricated at Integrated System Laboratory (LSI) of University of Sao Paulo (USP), Brazil. One of the main attribute of this device is the reconfigurability, i.e., it can operate as an n-type or as a p-type transistor, depending on the back-gate bias. This work explores the characteristics of the BESOI MOSFET as a charge-based biosensor(4) through numerical simulations, using Synopsys Sentaurus TCAD(5), based on experimental measurements.Figure 1 shows the BESOI schematic profile. Experimental details can be seen in(3). The two underlap regions between the drain and gate electrodes and between the source and gate electrodes are the areas where the biological material of interest is deposited, whose charge concentration (Qoxbio) can affect the drain current conduction (figure 2). The BESOI’s operation principle is based on the back-gate bias (VGB). If VGB is positive (n-type BESOI), electrons are induced at the back interface of the silicon channel (between the silicon and the buried oxide). The carrier layer at the back interface enables the drain current conduction that can be modulated by the front-gate voltage (VGF). A negative enough VGF fully depletes the silicon channel, interrupting the current, whereas a positive enough VGF can induce electrons also at the front interface (between the silicon channel and the gate oxide), decreasing the channel resistance(6). An analogous operation occurs for the p-type BESOI, however, a negative VGB is needed and holes are induced at the back interface. In the figure 2, positive Qoxbio values, VGB=25V and VDS=100mV were used in the n-type BESOI while the p-type was biased with VGB=-25V and VDS=-100mV and negative Qoxbio values were considered. The drain current increased as a function of |Qoxbio| because the biomaterial charges electric field increases the induced carriers in the silicon channel.In order to evaluate the influence of the charge concentration of the biological material and the transistor dimensions on the drain current, the simulations were performed varying Qoxbio, the gate electrode length L, the underlap length LUD, the buried oxide thickness tBOX, the gate oxide thickness tox and the silicon film thickness tSi. The relative permittivity of the electrolyte was set to 80(7) and positive gate oxide effective charges were considered to fit experimental results. A sensitivity parameter was defined in equation 1, adopting the lowest Qoxbio (1x1010q/cm2) curve as the reference. A constant absolute front-gate overdrive voltage |VGT=VGF-VT| of 3V was considered in the calculations in order to compare the results of the n-type and p-type biased devices, where VT is the threshold voltage. Figure 3 shows the sensitivity(S) as a function of each parameter for Qoxbio=1x1012q/cm2. The L parameter presents small influence on the S, once the front-gate electrode length only contributes to alter the channel resistance. The underlap influences on the S because its length LUD modifies the area in which the biological material charges affect the channel. The buried oxide thickness tBOX influences inversely the S as the thicker buried oxide diminishes the electric field effect in the channel due to VGB. The tox and tSi parameters presented different trends for the n-type and p-type biases due to the positive gate oxide effective charges. In the n-type case, the gate oxide charges induce electrons also at the front interface, thus the biomaterial charges affect the front interface conduction and the thinner tox and thicker tSi present higher sensitivity. In the p-type bias, the positive gate oxide charges induce electrons at the front interface, suppressing the front interface conduction. The biomaterial charges affect mostly the carriers at the back interface, thus the thicker tox and thinner tSi present higher sensitivity. This fact also led to a higher sensitivity for p-type BESOI MOSFET than for n-type ones.The sensitivity as a function of the Qoxbio is shown in figure 4, in which negative charges were also considered for the n-type biased and positive charges for the p-type biased devices. The reconfigurable characteristic of the BESOI MOSFET means that it is possible to obtain a higher sensitivity in a single device if the biological material presents positive or negative charges. N-type biasing can be more suitable for detecting positive charges, while p-type biasing may present better sensitivity to negative charges. Figure 1