Apart from the required selectivity and sensitivity, a sensory system to be used by a non-trained person must produce a response clear enough to allow an outcome which is unambiguous and easy to interpret. Most systems reported in the literature [11, 15], use a small source drain potential to determine the response of the system. These small potentials result in relatively small shifts in the source drain current, which is used to detect the presence of the target. This is not an issue when using scientific equipment, but when used without, for example in a point of care application, the variation needs to be improved.
As such it is necessary to look at the IV-curve over a wide range of potentials to determine the most suited within a device. This is shown in Figure 2. It can be seen that there is significant variation in the current levels dependent on the concentration of the spike protein.
A number of distinct features can be observed in Figure 1. There is an initial rapid increase in current, which levels off at about ±1 V, which can be associated with the time dependence of the sweep. After this increase there is a constant increase in current. On the return of the potential towards zero, the current stays constant or even shows a small increase before returning to the same current value at zero. This hysteresis can be explained by the fact that when the potential is increased more trapped states become accessible and fill with electrons.
We make the following correction for experimental variation between measurement runs and samples. The relative change of the current is used to create a device-independent parameter, R, defined as
where I is the measured current and I0 is that measured at the same voltage in the IV characteristic, determined using pure PBS without the addition of the protein. In Figure 2a the ratio R is shown over the whole voltage range for the various concentrations. It is worth noting that R is constant for the returning sweep, and independent of the applied source drain potential.
The constant R, is an important observation as unlike other aptamer or antibody-based transistor sensors where the source drain potential is kept at a maximum of 100 mV [11, 15, 16], here a range of voltages was investigated. At low potentials, a wide variation of R can be seen as effectively, the numbers are small. At higher potentials, the ratio remains fairly constant, arising from large differences in absolute numbers, which is easier to differentiate.
Figure 2b shows the source drain current as a function of concentration. The main observation from this figure is that there is a significant variation, up to orders of magnitude, in the current levels as function of the concentration with an almost linear response between the 10 fM and 100 pM. Perhaps surprisingly, an increase in concentration of the spike protein causes the current to reduce compared to the zero concentration values. This effect has been observed before for the case of molecular detection using an indium oxide thin film transistor functionalised with aptamers  and was explained by the changes in conformation of the molecules within the aptamer strand. These conformational changes result in a modification of the local electric field and the effect can create either an increase or decrease of the film resistance. While the situation here is somewhat different in that a protein is detected rather than a molecule, it is not unreasonable to expect that there are certain similarities.
The starting position is an aptamer, which has an inherent charge distribution which is a key parameter for ensuring selectivity in the binding between aptamer and protein. This charge distribution will create a specific resistivity of the intrinsic silicon channel. While it has been reported  and proven by the response of the TFT sensor that the aptamer binds to the spike protein, what is unknown is the exact binding location of the protein and how the local charge distribution of the protein is arranged. The observed effect could be explained if the presence of the protein were to limit or shield the effect of the charges on the aptamer, while not adding any additional charge contribution of its own.
Another observation that can be made is that the current variation as a function of concentration spans the whole range of currents. The maximum current is observed when the samples are immersed in a 0 M spike protein solution, or pure PBS. At the maximum detectable concentration, the current decreases to the same level of that of a bare silicon TFT indicating that at that stage, there is no net charge in close proximity to the conduction channel.
A direct comparison with other proteins is not possible as the exact charge distribution near to the silicon channel is unknown and highly likely to be completely different. However it is interesting to compare our results with reported work on a graphene TFT binding to the same spike protein . There are two major differences in the experimental technique compared to the results presented here. The first is that the experiments were performed in liquid, the second is that an antibody was used for the attachment of the spike protein. When comparing the like for like results, the response of the aptamer-based system reported here, was significantly larger than that of the anti-body based system. This difference is related to the afore-mentioned Debye length, which governs the screening distance for the charges of the protein which in turn, modules the conduction channel. The Debye length of both experiments is approximately 1 nm, hence a protein that is further away from the channel due to being bound by an antibody will create less of a response.