Low Temperature CVD Grown Graphene for Highly Selective Gas Sensors Working under Ambient Conditions

: In this paper we report on gas sensors based on graphene grown by Chemical Vapor Deposition at 850 °C. Mo was used as catalyst for graphene nucleation. Resistors were directly designed on pre-patterned Mo using the transfer-free process we recently developed, thus avoiding films damage during the transfer to the target substrate. Devices operating at room temperature and relative humidity set at 50% were tested towards NO 2 . The sensors resulted to be highly specific towards NO 2 and showed current variation up to 6%. The performances were compared with those of gas sensors based on graphene grown at 980 °C, which represents the usual growth temperature for such material. The findings show that by lowering the graphene growth temperature and consequently the energy consumptions the sensing benefits of these devices are still preserved.


Introduction
Among the many attractive properties graphene has, the strong stability, the highest surface-to-volume ratio (~2600 m 2 g −1 ) and the interaction with only the surface atoms, make graphene the ideal candidate for gas sensors operating in ambient conditions [1]. CVD graphene, in particular, reveals itself particularly promising in terms of high quality and large scale production [2]. However, for practical applications, the bottleneck related to the CVD technique is inherent to the graphene transfer from the catalyst substrate to the target one [3]. On this regard, we have recently reported a transfer-free process (TFP) that prevents any issue related to the graphene transfer [4]. In our previous work, graphene-based gas sensors prepared through TFP were found to be able to achieve extremely low limit of detection (LOD), in the range of a few hundred ppb of NO2, and resulted scarcely sensitive towards NH3 [4]. Here, we present the performances of the gas chemi-resistors based on graphene fabricated through a process specifically developed to lower the growth temperature down to 850 °C. This value is significantly much lower than the 980 °C usually adopted as graphene growth temperature [2,4,5]. The choice to reduce the growth temperature is motivated by the fact that a lower growth temperature can reduce the energy consumption during the graphene growth and facilitate large scale production of these gas sensors. We demonstrate that the sensors performance is not affected by lowering the material growth temperature, thus preserving the above mentioned benefits.

Sensors Preparation
The graphene-based gas sensors presented in this study were fabricated on 4" Si (100) wafer covered by thermally grown SiO2 (90 nm). A thin film of Mo (50 nm) was sputtered from a pure (99.95%) Mo target. Afterwards, dry etching was used to pattern the Mo layer, as described in our previous work [5]. The graphene growth on the patterned Mo catalyst was carried out in an AIXTRON BlackMagic Pro at 850 °C, using Ar/H2/CH4 as feedstock at a pressure of 25 mbar. The Mo catalyst was then etched away following the transfer-free process (TFP) we developed [5] and the graphene film laid on the SiO2. Evaporated Cr/Au (10/100 nm) electrical contacts were defined on the top of graphene film using a lift-off process.

Sensors Characterization
The devices were electrically characterized by a semi-automatic probe-station with an Agilent 4156C semiconductor parameter analyzer.
Three different tests were performed on the gas sensors in a Gas Sensor Characterization System (GSCS, Kenosistec equipment) setting temperature and RH at 22 °C and 50%, respectively.

Sensors Test-Protocol Description
The first test, in the following addressed as Test 1, consists of a single exposure at 1 ppm of NO2 10 min long, preceded and followed by 20 min of baseline and recovery phases, respectively, in N2 atmosphere.
The second test, Test 2, consists of 5 sequential pulses of NO2 at 1 ppm, similarly to Test 1. Finally, Test 3 is made of 12 sequential pulses of NO2 at different concentrations ranging from 1.5 down to 0.12 ppm, each step being 4 min long. The baseline and recovery phases lasted 20 min, respectively. Only the baseline preceding the first step was set 10 min longer than in the other steps in order to further stabilize the devices in the test chamber.

Results and Discussion
In Figure 1, the I-V characteristics of the fabricated devices (inset) are showed. Red and black lines are referred to devices based on graphene grown at 850 °C and 980 °C, respectively. The linear behavior of the two curves proves that the Ohmic contact was successfully realized between graphene and the Cr/Au contacts. The different resistance value can be ascribed to the differences in the material crystallinity.
Such prepared chemi-resistors were tested upon the abovementioned three different protocols. In Figure 2, the real-time current behaviors of the chemi-resistors upon exposure to a single ( Figure 2a) and five sequential pulses of NO2 (Figure 2b) I0 and I represent the current values at the inlet and outlet of the NO2 flow, respectively. For sensors named "850 °C", ΔI/I0 was estimated to be roughly equal to 6%. As comparison, ΔI/I0 for sensor named "980 °C" was around 7%, providing the first indication that the sensors performance is not significantly affected by lowering the temperature of graphene growth. Figure 2b also attests the substantial equivalence between the two devices behavior, showing the overall same kinetics upon Test 2. In Figure 3, this comparison is further addressed. For both sensors, the current recorded during each gas pulse of Test 2 is compared. The signal is normalized to the value I0 at the gas inlet of each step. It is worth to note that, in both cases, same trend of the signal and current variation decreasing are noticed. For instance, for both sensors, the difference between the first and second step is about 2%. On the other side, the other steps do not present appreciable differences between the two sensors' performances.  Finally, the sensors were undertaken to Test 3 and the results are showed in Figure 4. Black and red lines refer to device "850 °C" and device "980 °C", respectively. Curves in Figure 4b, extrapolated from Figure 4a, show the plot ΔI/I0 as function of NO2 concentration, where I0 represents the current value at the gas inlet for each gas pulse.   Therefore, within the error bar, Figure 4b definitively discloses the substantial comparability of the findings, although the different growth temperature for the graphene.

Conclusions
In this wok, we presented gas sensors based on graphene grown by CVD at 850 °C. Such sensors were tested towards NO2 under environmental conditions, i.e., room temperature and relative humidity set at 50%. The performances were compared with those of devices based on graphene grown at 980 °C, which is usually adopted as growth temperature for such material. The different tests carried out on both sensors definitely uncovered that the sensing behavior is not affected by lowering the growth temperature of graphene. Therefore, these results indicate that energy consumption can be significantly reduced during the large scale production of graphene and graphene-based sensors by CVD technique.