Programmable Graphene e-Nose Sensor Arrays for Fast, Sensitive, and Label-Free Identification of Chemical Vapors

Abstract

Conventional nanoelectronic chemical vapor sensors are mostly based on charge transfer mechanisms. The presence of vapor molecules interacts with the sensing material, resulting in changes in charge density in the channel; the resulting conductance change can subsequently be detected as changes in the current signals. This charge transfer-based mechanism naturally requires strong molecule-sensor interaction in order to achieve high sensitivity. However, the strong binding of analytes to the sensor surface will inevitably lead to slow response time and slow sensor recovery, making it unsuitable for applications requiring rapid monitoring of chemical vapors.To address this challenge, our groups previously demonstrated a new type of fast, sensitive, and broad-spectrum electronic vapor sensor by exploiting the fringing field capacitance effect in the graphene field-effect transistor (GFET). The typically trivial fringing field capacitance change due to analyte absorption is greatly amplified by both the graphene transistor and a micro-flow channel covering the surface of graphene. Different from the charge-transfer process in traditional sensors, when molecules of analytes flow through the surface of the graphene transistor, its local dielectric environment is altered which results in changes in the fringing capacitance. This molecular fringing gate effect can be amplified intrinsically by the large transconductance of the GFET. Instead of using AC spectroscopy, this capacitance change can be measured conveniently as a peak in DC current.1 In this work, a programmable graphene e-nose sensor array with electrical gates design and enhanced packaging will be demonstrated to achieve true label-free sensing and identification of chemical vapors. Individual graphene sensors within the 1D sensor array (up to 1 x 9) can be optimized to offer sensitive (down to picogram) and fast (sub-second) detection toward a variety of analytes. Figure 1 demonstrates the prototype of the graphene sensor array. Each sensor chip consists of two parts: an array of nine GFETs with a dimension of 15mm by 15mm (Figure 1(A)), and a 40cm long micro-column chip with a column width of 20-60 μm fabricated on a silicon wafer with 200nm of SiO2 by using DRIE (Deep Reactive Ion Etching) (Figure 1(B)). These two components were then bonded by epoxy and the assembled sensor chip is shown in Figure 1(C). The graphene transistor array was fabricated by transferring CVD graphene onto a silicon substrate with 200nm of thermally grown SiO2. Individual gold metal back gates were fabricated underneath the graphene film with 200nm Al2O3 deposited by ALD (atomic layer deposition) as the gate dielectric. Devices with single or multi-layer (up to 4 layers) graphene films were fabricated and tested.The fabricated graphene sensors have near-perfect device yield and a resistance range within 2 ~ 10 kΩ. Figure 1(D) and 1(E) shows the DC current signal changes of a typical sensor within the array when vapor-phase analytes of different masses were injected through a benchtop gas chromatography (GC) setup. 9 ng and 6 ng of Hexanal injection generated current peaks with peak heights of 0.8 μA and 0.5 μA, and peak width (full width at half maximum) of 1.8 seconds and 2 seconds, respectively. The high signal-to-noise ratio (~ 80 to 1) and the short response time demonstrate the high sensitivity and fast detection capability of the graphene sensor. Figure 1(E) shows a detailed sensitivity test from the injections of hexanal vapor of five different masses. The graphene sensor exhibits good linearity with a sensitivity of 79nA per nanogram of Hexanal and a 3σ detection limit of 380 picograms.We further tested the electrical gate tunability of the graphene sensor’s response. In order to eliminate the effect of graphene conductance modulation by the gate voltage, we normalize the peak height with the baseline current. As shown in Figure 1(F), when the gate voltage is programmed from -10V to +10V, the hexanal peak height is strengthened by approximately 30 percent. Similar gate voltage tuning is also observed on other analytes, such as acetone and hexanol.Our results clearly show the gate voltage tunability of the graphene sensor, suggesting a novel graphene sensor array using electrical programming rather than chemical functionalization to achieve selectivity. Combined with advanced data analysis tools, the graphene e-nose array has the potential to offer label-free chemical vapor sensing without the need to individually functionalize each sensor as in traditional e-nose devices. Furthermore, we will discuss the integration of the graphene e-nose sensor array with a micro-gas chromatography chip and a smartphone-sized custom PCB board for programming, readout, and Bluetooth data communication. W. Zang et al., Nano Lett., 21, 10301–10308 (2021). Figure 1