Methanol falls under the category of volatile organic compounds (VOCs). It is used as an organic solvent in many industries, such as automotive fuel, biomedical, chemical, pharmaceutical, etc. Moreover, methanol fuel cells are considered as a good choice for clean and efficient power generation [1, 2]. However, it is highly flammable with a low explosive limit of 6% in air mixed with interfering gases, and can make humans intoxicated when a person is exposed to more than 200 ppm of methanol during 8 h with adverse effects [3]. Worse still, methanol oxidizes slowly and accumulates in human body, unlike other alcohols leading to poisoning [4]. Thus, the development of a methanol sensor with good sensitivity and selectivity, which can work at room temperature is important for environment and healthcare.

Investigations of VOC sensors mainly focus on metal oxide semiconductors (MOS), which usually work at a temperature of over 100 ℃ to achieve their optimum sensing properties including high sensitivity, the best selectivity, the shortest response/recover time, etc. However, a high operating temperature not only means a high power consumption, but also has hidden trouble in safety issues when detecting VOC gases. What is more, the high operating temperature limits the potential application in the Internet of Things and wearable sensing device that are difficult to power [5-9].

Up to now, different kinds of sensing materials have been explored in room temperature gas sensors [10-20]. Among these materials, two-dimensional (2D) materials including graphene [21], phosphorene [22], and metal disulfides [23, 24], have attracted interest due to their flexibility, and capability of working at room temperature. Ti₃C₂T_x, a well-studied 2D early-transition metal carbide, has metallic conductivity exceeding that of other 2D materials and possesses various surface functional groups denoted by T_x here, such as –OH, –O or –F, which are introduced during the etching of Al layers from Ti₃AlC₂ bulk with acid. Ti₃C₂T_x is the most widely studied MXene and it is relatively stable in various environments [25]. Besides its potential applications in supercapacitor, Li-ion battery, catalysis, electromagnetic shielding and medicine [26-33], Ti₃C₂T_x has been demonstrated recently to be a promising gas-sensing material, which can be used to detect ultralow concentration of gases at room temperature due to its large specific surface area and the plenty of surface functional groups which endow Ti₃C₂T_x with abundant active sites for interacting with gas molecules [34-36]. Additionally, thanks to its intrinsic metallic conductivity, the gas-sensing signal of Ti₃C₂T_x exhibited low electrical noise. However, to date, investigations focused on MXenes for gas sensing are limited and the selectivity is moderate, especially between methanol and ethanol [35-37].

Poly(3, 4-ethylenedioxythiophene) doped with poly(styrene sulfonic acid) (PEDOT:PSS), a conductive polymer, is widely used in chemical sensors because of its electrochemical activity, environmental stability and ability of working at room temperature [38, 39]. The mixture of PEDOT:PSS with 2D materials is expected to enhance their gas-sensing properties due to increasing the interlayer spacing of each other, while maintaining a high conductivity. Graphene has recently been incorporated into PEDOT:PSS by different methods and characterized for gas sensors [40, 41]. Compared with graphene, MXenes possess abundant functional groups. The Ti₃C₂T_x showed higher response to VOCs than reduced graphene oxide [35]. Moreover, the conductivity of MXene could more easily be changed in PEDOT:PSS-MXene hybrids due to its metallicity. PEDOT:PSS-MXene hybrids were well explored in capacitors [42]. Herein, we demonstrate the Ti₃C₂T_x and PEDOT:PSS hybrids having better gas sensing properties compared to pure Ti₃C₂T_x and pure PEDOT:PSS, including enhanced responses toward VOCs and a high response ratio of more than 5.54 to methanol than to other gases with ultra-low power consumption.

Ti₃C₂T_x MXene was synthesized by etching Al from Ti₃AlC₂ powder [43, 44]. Firstly, we added 1.6 g of LiF powder to 20 mL of 9 mol/L HCl while stirring with a Teflon magnetic stir bar. Secondly, one gram of Ti₃AlC₂ powder was added to the homogeneous solution little by little with continuous stirring. The etching process continued under stirring the solution for 24 h at room temperature. After the etching, the product was washed with deionized water and centrifuged until the pH reached around 6. At last, the near-neutral solution was centrifuged at 3500 rpm for 1 h, and the dark green supernatant was the monolayer Ti₃C₂ solution which was kept for further experiments.

Ti₃C₂T_x/PEDOT:PSS hybrid solutions were prepared through adding 5 mg/mL PEDOT:PSS solution to 5 mg/mL Ti₃C₂T_x solution with continuous stirring for 1 h at room temperature (Fig. 1a). The composition of the material was controlled through different weight ratios of PEDOT:PSS to Ti₃C₂T_x particularly of 10:1, 8:1, 4:1, 2:1, 1:1 and 1:2, which were labeled as P/T-10:1, P/T-8:1, P/T-4:1, P/T-2:1, P/T-1:1 and P/T-1:2 respectively.

Fig. 1b shows the fabrication process of a gas sensor device. Alumina plates with silver interdigitated electrodes were used as substrates. The width and interspacing of the electrodes were 0.28 mm and 0.12 mm, respectively. The prepared Ti₃C₂T_x/PEDOTPSS nanocomposite solution was then drop-casted over the interdigitated electrode. Then, the nanocomposite gas sensor was dried for 1 h in vacuum at 150 ℃.

The morphology and microstructure of the as-prepared samples were studied by field-emission scanning electron microscope (SEM, FEI Nova NanoSEM 450) with an accelerating voltage of 15 kV. Raman spectra were recorded with a Renishaw in Via Raman spectrometer. The prepared gas sensors were mounted in a homemade gas-sensing chamber (Fig. 2). Target gases were diluted with N₂, and the flow rates were appropriately controlled by mass flow controllers to obtain desired gas concentrations. The resistance signal of the mounted sensor was measured under 0.1 V by a Keithley 2400 source meter.

Figs. 3a and b show cross-sectional SEM images of a Ti₃C₂T_x film and a P/T-4:1 film produced by vacuum filtration of the same solutions used for fabricating the gas-sensing devices. PEDOT:PSS separated MXene layers provide a favorable architecture for the adsorption of gas molecules on both PEDOT:PSS and MXene surface (Fig. 3b). Moreover, the high conductivity and weak contact between layers of Ti₃C₂T_x allow wide fluctuation of sensor resistance upon adsorption leading to high sensing response.Fig. 3c shows the Raman spectra of pure Ti₃C₂T_x and the prepared P/T-4:1 hybrid. As for the pure Ti₃C₂T_x, the peaks at 202 and 720 cm^–1 are attributed to the Ti-C and C-C vibrations (A_1g symmetry) of Ti₃C₂T_x with the oxygen terminal. The peaks at 392 and 575 cm^–1 come from the E_g and A_1g vibrations of the O atoms. The peaks at 275 and 512 cm^–1 are correspondingly due to the E_g and A_1g vibrations of H atoms in the OH groups of Ti₃C₂T_x. The peak at 618 cm^–1 is attributed to the E_g vibrations of the C atoms in the OH-terminated Ti₃C₂(OH)² [45]. The Raman spectra of the P/T- 4:1 hybrid exhibited a combined character of Ti₃C₂T_x and PEDOT: PSS, but the peaks of Ti₃C₂T_x are weak because of the large quantity of PEDOT:PSS in the hybrid. The bands at 1564, 1531 and 1503 cm^–1 are related to the asymmetric C_α=C_β stretching, 1434 cm^–1 to symmetric C_α=C_β(–O) stretching, 1373 cm^–1 to C_α=C_β stretching, 1255 cm^–1 to C_α–C_α inter-ring stretching, 1097 cm^–1 to C-O-C deformation, 990 and 578 cm^–1 to oxyethylene ring deformation, 696 cm^–1 to symmetric C-S-C deformation, and 436 cm^–1 to SO₂ bending [46, 47]. The bands at 854 cm^–1 is ascribed to C-O bending vibrations [45].

As the conductivity of gas sensors affects the energy consumption of the devices, we performed resistance measurements of the sensing devices at room temperature. Fig. 4a indicates that all the gas sensors possessed excellent conductivity but Ti₃C₂T_x/PEDOT:PSS gas sensors have a higher initial resistances than pristine Ti₃C₂T_x and PEDOT:PSS. The maximum resistance belongs to the hybrid whose mass ratio of PEDOT:PSS and Ti₃C₂T_x is 8:1. The result can be explained as follows: PEDOT:PSS enlarges the interlayer distance of MXene, separating MXene flakes and destroying conducting network of MXene. Thus, with the increasing amount of PEDOT:PSS added in MXene, the resistance of the hybrids increases. With the PEDOT:PSS content dominating the hybrid, the main conductive pathways become contributed by the polymer matrix instead of MXene. The presence of MXene in the polymer matrix probably increases the interchain distance of PEDOT making electron hopping process difficult [40]. The high conductivity enables all the sensors to work under 0.1 V at room temperature, leading to a low power consumption.

The gas-sensing behaviors of the fabricated gas sensors toward acetone, ethanol and methanol were investigated next. The responses of the gas sensors were evaluated by [(R_g-R_N)/R_N]× 100%, where R_N is the resistance in pure N₂, and R_g is the resistance under exposure to the target gas. Fig. 4b illustrates the sensing responses of the eight sensors toward acetone, ethanol and methanol with a concentration of 300 ppm at room temperature. All the sensors exhibit a higher response toward methanol and the hybrid with the mass ratio of 4:1 between PEDOT:PSS and Ti₃C₂T_x shows the highest responses toward all the three gases. The enhancement of responses may be due to the variational sensing mechanism for different sensors. For the pure PEDOT:PSS, the increase in the resistance by introducing target gases should belong to the body-controlled behavior of the material [41]. The adsorbed reducing gases donate electrons to the polymer matrix, which is p-type semiconductor, increasing the resistance. In addition, the swelling from the insertion and diffusion of gas molecules into the polymer matrix, increases the resistance [48, 49]. As to the hybrids with the mass ratio of 10:1 and 8:1 between PEDOT:PSS and Ti₃C₂T_x, the main conductive channel still is the polymer matrix. The electrons transferred from adsorbed gas molecules weakly change the resistance because of the presence of metallic Ti₃C₂T_x. Then, only the swelling process increases the resistance of the two hybrids with the mass ratio of 10:1 and 8:1 between PEDOT:PSS and Ti₃C₂T_x. So, there is a decrease of response from pure PEDOT:PSS to P/T-8:1. With the increasing amount of Ti₃C₂T_x in the hybrid, the conductive channel would be contributed by the polymer-polymer, polymer-MXene and MXene-MXene connections. Elimination of MXene-MXene connection would increase the resistance obviously due to the high conductivity of MXene. The MXene-MXene connection of the hybrid with the mass ratio of 4:1 between PEDOT:PSS and Ti₃C₂T_x could most easily be destroyed due to a few contacts between MXene and MXene in this composition. This is why the sensor based on the hybrid with the mass ratio of 4:1 showed the highest response among all the eight sensors toward 300 ppm acetone, ethanol and methanol gases at room temperature. More interesting, the sensor based on the hybrid with the mass ratio of 4:1 also exhibited the highest response ratio of responses toward methanol and other gases among all the eight sensors. We defined E as R_largest/R_{second largest}, where R_largest is the largest response and R_second largest is the second largest response. After calculating, we obtain the result of E_polymer = 4.67, E_10:1 = 2.90, E_8:1 = 1.16, E_4:1 = 5.54, E_2:1 = 4.47, E_1:1 = 4.75, E_1:2 = 3.46 and E_MXene = 1.13 respectively. These results testified that a mixture of PEDOT:PSS and Ti₃C₂T_x can enhance the selectivity to methanol, and the E_4:1 is the largest one.

The response and recovery curves of the P/T-4:1 sensor toward 300 ppm acetone, ethanol and methanol at room temperature are shown in Fig. 4c. The response and recovery times were somewhat longer than those of conventional sensors, which could be further improved by optimizing the film thickness, architecture and device structure that could decrease the diffusion length. Fully MXenebased sensors with interdigitated printed MXene electrodes should be a choice to improve sensing performance [50]. In particular, the curve of methanol does not show a square shape manifesting the response time should be much longer. From another point of view, the concentration of 300 ppm might be too high for the sensor as the sensor could detect lower concentration of methanol, and the response toward 300 ppm methanol in steady state could be much higher than what is shown in the Fig. 4c. However, our ability to test lower concentrations of methanol was limited by our method of introducing gas from the vapor of organic liquid. Even so, the results can still help us understand the mechanism of gas-sensing process for the Ti₃C₂T_x/PEDOT:PSS. To further demonstrate the dynamic gas-sensing performance, we recurrently exposed the P/T-4:1 sensor to acetone, ethanol and methanol gases with different concentrations. As shown in Fig. 4d, the sensor exhibited excellent cyclic curves toward acetone and methanol at room temperature.

In conclusion, we investigated the gas sensing properties of the hybrid materials consisting of Ti₃C₂T_x and PEDOT:PSS at room temperature. Compared to the pure Ti₃C₂T_x and pure PEDOT:PSS, Ti₃C₂T_x/PEDOT:PSS hybrids exhibited enhanced sensing behaviors on acetone, ethanol and methanol. The sensor based on the hybrid with the mass ratio of 4:1 between PEDOT:PSS and Ti₃C₂T_x showed the highest response among all the eight sensors toward acetone, ethanol and methanol gases at room temperature and the highest selectivity to methanol due to the synergistic effect of PEDOT:PSS and Ti₃C₂T_x. This sensitive sensor based on Ti₃C₂T_x and PEDOT:PSS for detecting methanol can open up new opportunities for exploring more MXenes-based gas sensors with controlled composition and morphology towards higher performance.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

We thank Dr. Seon Joon Kim, Tyler Mathis and Kanit Hantanasirisakul for their help in setting up gas-sensing equipment. The research was supported by the National Natural Science Foundation of China (No. 51602035), State Scholarship Fund of China, Liaoning Provincial Natural Science Foundation of China (No. 20180510036), and the Fundamental Research Funds for the Central Universities (No. DUT19JC41).