To establish the sensor s specificity and reproducibility fo

To establish the sensor’s specificity and reproducibility for Hg2+ detection in the field, the response to other potential contaminants must be minimized. The sensor was tested with a variety of other ions, including common environmental contaminants, e.g., Na+, Fe3+, Ca2+, Pb2+, and Cd2+ with the same procedure as that used for Hg2+. The DNA probe for highly selective sensing of Hg2+ using the formation of DNA–Hg2+ complexes has been reported while testing against a subset of these ions [19,33]. The additional ions resulted in slight decreases of IDS, as shown in Fig. 4b–f; this response may have resulted from the electrostatic interactions between the negatively charged DNA and the positive charged ions. Furthermore, the rGO/Al2O3/DNA sensor was tested for protein detection (Escherichia coli antibody and avidin), which did not show any response (Figure S4). Another control experiment was conducted using an rGO device with a 2nm-thick Al2O3 film, but without the decoration of Au NPs. The rGO/Al2O3 device was not responsive to the Hg2+ ions (Figure S5). The sensor relative sensitivity (relative resistance change, %) as a function of the metal ion concentration (nM) shown in Fig. 4g indicates the rGO/Al2O3/DNA devices show a much higher sensitivity to each Hg2+ concentration than that of other metal ions. (Note: R, device resistance before dropping of meal ion solutions at each time; ΔR, relative change of device resistance). In other words, it was confirmed that the binding between Hg2+ and probe DNA is specific through the interaction with thymine Histone Compound Library pairs in the DNA. The sensor’s capability was also investigated for selective detection of Hg2+ in complex solutions. While testing with a complex sample containing multiple ionic species (Na+, Fe3+, Cd2+, Pb2+, and Hg2+, the same concentration of each metal ion with Hg2+ in the mixture solution), the sensor showed a similar trend with the detection of Hg2+ alone, which means that sensing interference from other metal ions is negligible (Fig. 5a), as previously observed for detecting Hg2+ with carbon nanotubes [15]. Without Hg2+, the devices showed a weak response to a complex sample at 1nM, likely resulting from nonspecific binding between DNA and ions. Subsequent exposure of the devices to the complex sample lacking Hg2+ resulted in a minimal sensing signal. The response of the DNA-functionalized devices to the solution with Hg2+ is significantly higher than that to the solution without Hg2+ due to the rearrangement of the Hg2+-binding DNA sequence by Hg2+. Ultimately, the high selectivity of the sensing platform for Hg2+ demonstrates the ability for specific detection of a selective target, while the platform’s capability for sensing without any other ion interference in complex solutions demonstrates its viability for specific detection in a highly complex environment. The sensors also showed good performance in real-time water sensing (tap water from Milwaukee) (Figure S7). Understanding the function of a passivation layer on the sensor performance will lead to better sensor design. The sensitivity of rGO/DNA devices without coverage of Al2O3 thin films is shown in Fig. 6a. According to the gating effect of a p-type rGO FET, it is anticipated that positively charged Hg2+ will decrease the current during the sensing operation. Fig. 6a illustrates the drain-source voltage dependence of the drain current ISD of our sensors after adding selected concentrations of Hg2+ (10−8, 10−7, 10−6, and 10−5M) dissolved in DI water. On the contrary, the conductance of the devices increases with increasing concentrations of Hg2+, showing an opposite trend to that of rGO/Al2O3/DNA devices. This may result from ionic conduction across drain-source electrodes through the water droplet that was placed on top of the device for operation under aqueous conditions, or from doping the semiconductor rGO, which has been previously observed for detecting Hg2+ with thermally reduced GO sensor platforms [7]. We have further developed our sensor with coverage of 1nm-thick Al2O3 films, illustrated in Fig. 6b. The sensor with 1nm Al2O3 passivation layer showed less sensitivity than that with 2nm-thick Al2O3 films on the rGO surface, which showed a clear gating effect behavior (Fig. 4a).