Operando monitoring of a room temperature nanocomposite methanol sensor

The sensing of volatile organic compounds by composites containing metal oxide semiconductors is typically explained via adsorption–desorption and surface electrochemical reactions changing the sensor's resistance. The analysis of molecular processes on chemiresistive gas sensors is often based on indirect evidence, whereas in situ or operando studies monitoring the gas/surface interactions enable a direct insight. Here we report a cross-disciplinary approach employing spectroscopy of working sensors to investigate room temperature methanol detection, contrasting well-characterized nanocomposite (TiO2@rGO-NC) and reduced-graphene oxide (rGO) sensors. Methanol interactions with the sensors were examined by (quasi) operando-DRIFTS and in situ-ATR-FTIR spectroscopy, the first paralleled by simultaneous measurements of resistance. The sensing mechanism was also studied by mass spectroscopy (MS), revealing the surface electrochemical reactions. The operando and in situ spectroscopy techniques demonstrated that the sensing mechanism on the nanocomposite relies on the combined effect of methanol reversible physisorption and irreversible chemisorption, sensor modification over time, and electron/O2 depletion–restoration due to a surface electrochemical reaction forming CO2 and H2O.


S3: Thermal analysis
To examine a potential use of the prepared sensor materials at higher temperatures, their thermal properties were analyzed by TG, DTG and SDTA ( Fig. 1E and Fig. S4). The temperature of maximum decomposition (Tn) within one weight loss step, the relative weight loss (Δm) and the decomposition temperature (Td), i.e., the temperature of 5% weight loss, are summarized in Table S3 for each material. To evaluate the effect of oxygen in the air, corresponding studies were first performed in N2.
The thermal behavior of TiO2-NPs in N2 (Fig. S4) can be divided into three stages. The first endothermic stage from room temperature to 200 °C with a mass loss of 2.6% can be attributed to the evaporation of small molecules such as physically adsorbed water (dehydration) and removal of organic solvent residues. The further weight loss from 200°C to 600°, with a global exothermic behavior, may include TiO2 phase transformation from anatase to rutile 5,6 , thermal decomposition of the peroxo titanium complex and the oxidative decomposition of carbon based compounds with maximum degradation at T = 498°C. Beyond 600° C, the endothermic peak in the DTA curve (T3 ~750°C) without weight loss reflects the phase transformation of TiO2 from anatase to rutile 6,7 . Overall, the sample had a total of 8.8% weight loss with a Td value of 425°C.
In air, the TG curves of TiO2-NPs showed a similar behavior, but with less evident endothermic peaks since the presence of oxygen favors exothermic processes. In this case, the first weight loss attributed to the loss of small molecular compounds was 2.5%. Then, the thermal decomposition of the titanium complex and organic residues gave rise to 2.1% and 1.4% weight loss at Tn´s of 287°C and 452°C, respectively, followed by the phase transition from anatase to rutile without significant weight loss. In total, the TiO2-NPs lost 7.35% weight in air and Td was 404°C. For rGO, in the presence of air (Fig. S4), the mass loss occurred in a single main step. First, up to 400°C, a slight weight decrease occurred due to the loss of VOCs and water with possibly back-toback endo and exothermal processes. Then, at T˃400°C, due to the presence of oxygen, a sharp oxidative decomposition occurred with an exothermic peak at T = 531°C and weight loss of 88%. At T˃600°C, no further weight loss was recorded. The total mass loss of rGO in air was 97% and Td was 295°C.
In N2, the weight loss of rGO due to pyrolysis was more gradual then under air. At first, VOCs and water were lost up to 150°C (T1 = 49°C). Then, thermal decomposition of organic compounds occurred which overlapped with exothermic peaks with a maximum weight loss at T = 270°C. Td was 290°C. At 1000°C, the residual weight (75%) in N2 was higher than for RGO in air due to additional residuals of unburned char. Moreover, TG in N2, differently from TG in air, showed a continuous weight decrease even at T >1000°C. Decomposition of rGO in N2 primarily depended on the available functional oxygen groups. When compared to literature, the rGO of the current study was a highly reduced form of GO with least available O2 species and stronger van der Waals forces between the layers 8 thus providing higher thermal stability and some resistance to pyrolysis at high temperature. For thermal analysis results of TiO2@rGO-NC in air see the main text (Fig. 1E). In brief, the TiO2@rGO-NC was thermally stable up to at least 300°C, making it compatible with high temperature sensing. In N2, only the Δm at T3 was significantly lower due to the absence of oxygen.

S4: BET Surface area
For rGO, the hysteresis loop appeared (Fig. S5) at relative high pressure (p/p 0 ), fitting best a type-III isotherm model, which reflects the hydrophobic nature of rGO as adsorbent. Such adsorption behavior is typical for adsorbate-adsorbate interaction being larger than adsorbate-sorbent interaction, e.g., water adsorption on hydrophobic surfaces (zeolites or activated carbon) 9 . On the other hand, for TiO2-NPs and 50% TiO2@rGO-NC the characteristic H3-hysteresis loop fits the type-IV isotherm model 10 . The BET surface area of TiO2-NPs, rGO and 50% TiO2@rGO-NC were 101.5 m²/g, 36.4 m²/g and 70.8 m²/g, respectively. Moreover, BJH cumulative pore volume and pore width of TiO2-NPs, rGO and 50% TiO2@rGO-NC were 0.11 cm³/g -5.9 nm,0.13 cm³/g -37.4 nm and 0.08 cm³/g -6.8 nm, respectively. A high surface area, small and homogenous particle size of the green synthesized TiO2-NPs are beneficial for their methanol gas sensing performance. Usually, metal oxide NPs with high surface area exhibit better sensitivity towards methanol sensing, which comes, however, with the limitation of higher working temperature 11 . Contrary, our study reports that TiO2-NPs with high surface area (101.5 m²/g) can indeed be efficiently utilized for better understanding room temperature methanol sensing.