The Effect of Doping rGO with MnO₂ Nanomaterial on Its Gas Sensing Properties

1. Introduction

Technological and economic advances increasingly threaten the environment and air quality, necessitating urgent solutions. The demand for low-cost and effective gas sensors, crucial for detecting toxic agents such as nitrogen dioxide NO₂ [1], ammonia NH₃ [2], and carbon monoxide CO [3] in various fields, has never been more pressing. While traditional methods like infrared spectrophotometry (IRSP) [4], non-dispersive infrared analysis (NDIR) [5], and gas chromatography-mass spectrometry (GC-MS) [6] have been widely used for detecting toxic gases, they are not without significant drawbacks. Their high cost and complexity have underscored the need for a new, more efficient approach. This realization has led to the development of chemoresistive devices, known for their ease of operation, low production cost, fast response time, and ease of miniaturization [7].

Since their first use in the 1960s, metal oxides (MOX) such as ZnO [8], SnO₂ [9], WO₃ [10], and many more [11] have been studied as sensitive films to be used in the fabrication of chemoresistive devices for gas sensing and especially for detecting NO₂. Even though MOX are effective and sensitive, the need for a pervasive, widespread monitoring of air pollutants, has oriented research to the discovery of new, less power-hungry, gas sensitive materials. This has, for example, led to the emergence of graphene-based chemoresistive sensors. The unique and exceptional properties of graphene that made it a solid gas sensing material candidate are thermal stability, mechanical robustness, high conductivity, high carrier mobility at low to room temperatures, a large surface area of up to 2630 m²/g for single-layer graphene, low electrical noise [12] and, most importantly, the fact that its electronic properties are easily affected by the adsorption of gas molecules [13].

One of the graphene derivatives is reduced graphene oxide (rGO). It was reported in the literature to be the best and most used graphene-based sensing material for detecting NO₂ [14] and NH₃ [15] due to its numerous defect sites and functional groups, which facilitate gas adsorption. rGO has been reported able to detect chemical warfare agents and explosives at trace levels (ppb) [16]. Moreover, the synthesis of rGO can be achieved via straightforward and inexpensive processes that reduce GO via chemical and thermal routes or even using UV light [17]. Studies have shown that pristine rGO gas sensors exhibit slow response and recovery dynamics. Hence, the hybridization of rGO with MOx has been a solution often explored to enhance its sensing properties (e.g. ameliorating response dynamics and extending the number of gases that can be detected), D.Tripathi et al, explored this process and an enhancement of the sensitivity and selectivity of the rGO material towards ammonia by incorporating WO₃ nanomaterial in the sensing layer was achieved [18]. Another reported work of D. Milad et al, where they showed that the synthesis of a TiO₂/rGO composite exhibited an improved gas sensing properties towards methanol and ethanol [19]. This is achieved via the MOX nanoparticles supported on rGO behaving as catalysts or as electronic sensitizers, favouring the occurrence of heterojunctions at the MOX/rGO interface.

Some of the reported MOx nanoparticles used for the loading of the rGO layers for gas sensing are ZnO [20], SnO₂ [21], and WO₃ [22]. Still, in recent years, MnO₂ has gained popularity due to its low toxicity, low cost, high stability, and ease of fabrication. It was also used in a wide range of applications and fields such as energy storage [23], biomedical field [24], and in developing gas sensors [25]. However, in the gas sensing field, there is a limited amount of reported work discussing the incorporation of the MnO₂ nanomaterial into rGO to achieve a sensitive layer towards different toxic gases. One of the few works reported is of Hui Zhang et al., where they successfully synthesized an rGO-coated Ni foam-supported MnO₂ for the enhanced detection of NO₂ at a concentration of 50 ppm while the sensor was operated at room temperature [26]. Meanwhile, Alexander et al. modified rGO with doping of MnO₂ nanoparticles and tested the rGO/MnO₂ composite as gas sensor for different gases such as 25 ppm of NO₂, 500 ppm of H₂ and 1000 ppm of CH₄ under dry conditions with heating at 85 ^◦C [27]. Another reported work was of Ghosal et al. where they prepared different hybrids for alcohol vapor detection, being one of the hybrids rGO/MnO₂ nanoflowers binary composite and it showed good responses towards ethanol and methanol vapours in the range of 5-100 ppm while heating at 150 ^◦C [28]. Lastly, Ahmad et al. made a ternary nanocomposite of PANI@rGO@MnO₂ using a multi-step process for NH₃ detection. The tests were made at 100 °C under dry conditions [29]. The fact that very few works have been reported on rGO@MnO₂ gas sensors so far, makes exploring further its gas sensing properties interesting and worthwhile.

In this paper, rGO@MnO₂ sensitive layers were successfully synthesized and deposited on different transducing substrates (Kapton and silicon with gold electrodes). The gas sensing performance of the rGO@MnO₂ sensors was studied for different reducing and oxidizing species. The effect of ambient moisture on sensor response was evaluated as well. Results are presented and thoroughly discussed. A sensing mechanism for the detection of ammonia and nitrogen dioxide is presented.

2. Materials and Methods

In this section a detailed explanation of the preparation of the doped reduced graphene oxide with MnO₂ nanomaterial with 95/5 wt.% alongside the process of depositing it on the Kapton and silicon substrates via the spray coating technique is presented. The section also describes the techniques employed in the study of the morphology and composition of the hybrid material and in the test of its gas sensing properties.

2.1. Preparation of Reduced Graphene Oxide Doped with MnO₂

rGO doped with MnO₂ nanomaterial (rGO/MnO₂ 95/5 wt.%) was synthesized using a process based on patented procedures (Patent number ES2678419A1). Briefly, reduced graphene oxide was dispersed in oxalic acid, in which the starting Mn₃O₄ had been previously dissolved at 50 °C. After homogenization, MnO₂ nanomaterials were slowly precipitated on reduced graphene oxide by adding a basic solution (NaOH 5 M) under vigorous agitation. The solid was filtered and dried at 90 °C overnight. Synthesis parameters such as temperature, stirring speed, addition rate, or MnO₂/rGO proportion, were controlled to obtain the desired crystallinity that provides the material the optimal properties. Manganese oxide phase was checked via XRD, Figure S1 (in Supplementary Materials) shows the XRD diffractogram.

2.2. Substrates Preparation and Material Deposition

10 mg of the rGO@MnO₂ nanomaterial were weighted and suspended in a 10 ml ethanol solution via a 30-min sonication. Subsequently, suspended nanomaterials were deposited by spray coating onto two different substrates (i.e., silicon, and Kapton). During the coating process, substrates were heated at 50 °C to promote the evaporation of the solvent and the formation of a homogeneous film. The interdigitated gold electrodes were deposited on the substrates using different processes. For Kapton substrates, 9 nm of gold were sputtered using a shadow mask to form the electrodes. In contrast, for the silicon substrates, a two-step approach took place. At first, a laser lithography technique (DWL 66fs, Heildelberg Instruments) was used to pattern a photoresist that coated an oxidized silicon wafer in the shape of the electrodes. In the second step, a titanium adhesion layer was sputtered with a thickness of 10 nm, and then a gold layer was sputtered on top with a thickness of 100 nm. A final lift off process was conducted to obtain the electrodes. The silicon wafer was then diced. Figure 1 shows two sensors on the two types of substrates used.

2.3. Material Characterization and Gas Sensing Measurements

The obtained sensors were characterized using different techniques, such as Raman via a Raman spectrometer (Renishaw, plc., Wotton-under-Edge, UK), with a laser wavelength of 514 nm to check the crystallinity of the materials. A Field Emission Scanning Electron Microscope (FESEM) using a Carl Zeiss AG-Ultra 55 (ZEISS, Jena, Germany) to study the surface morphology and to check the distribution of the nanomaterial on the graphene layer. A JOEL F200 TEM ColdFEG operated at 200 kV was used for the high-resolution transmission electron microscopy (HRTEM) characterization. TEM images and electron diffraction patterns were acquired with a Gatan OneView camera, a CMOS-based and optical fibre-coupled detector of 4096 by 4096 pixels. Gatan Digital Micrograph program was used to process the (S)TEM images. STEM images (1024 x 1024 pixels) were recorded from the JEOL bright-field (BF) and high-angle annular dark-field (HAADF) detectors with a camera length of 200 mm. Samples were inserted in a JEOL beryllium double-tilt holder for energy-dispersive X-ray spectroscopy (EDS). STEM-EDS mapping was recorded from an EDS Centurio detector (silicon drift) with an effective area of 100 mm² and 133 eV of energy resolution. STEM-EDS maps (512 x 512 pixels) were processed with the JEOL Analysis software to check the shape of the MnO₂ nanomaterial and its incorporation in the graphene layers. Gas sensing measurements were conducted by placing the different sensors in an airtight Teflon chamber with a volume of 35 cm³. A continuous stream of dry air (Air Premier, 99.995% purity) and diluted gases were passed through the testing chamber with a 100 mL/min flow via a set of Bronkhorst mass-flow controllers. The target gases from calibrated bottles (NO₂-1 ppm, CO-100 ppm, NH₃-100 ppm, and ethanol-20 ppm balanced in dry air) were further diluted using the mass flow controllers set. The resistance changes were continuously acquired using an Agilent HP 34972A multimeter. The humidity effect on the sensing performance was assessed by humidifying the gas stream through a controller evaporator mixer (CEM) to obtain low humidity levels of maximum 50% RH at 25ºC. For higher values of relative humidity, the flow of the dry air with the corresponding concentration of gas was humidified passing through a bubbling water system at room temperature. The sensing responses were calculated using the formula R (%) = ((R_g − R_a)/R_a) × 100, where R_g and R_a correspond to the resistance level after and before gas exposure, respectively.

3. Results

This section, presents and discusses at first the results of the characterization tests made for the prepared sensitive layers (rGO and rGO@MnO₂), which are based on the RAMAN, FESEM and HRTEM techniques. Secondly, the gas testing results for these sensors towards NO₂ and NH₃ at room temperature under dry conditions are presented, which is followed by the results gathered at different humidity levels. Additionally, selectivity tests are reported. The results of the gas sensing tests are compared to those found in the literature. Finally, a sensing mechanism for the detection of NO₂ and NH₃ is introduced.

3.1. Sensitive Layer Characterization

3.1.1. Raman

The study of the molecular structure of carbon products and the assessment of disorders and defects in the material can be done using Raman spectroscopy analysis. Two specific peaks always appear when analysing graphene: the G-band and the D-band. The first one, placed at around 1500 cm^-1_, corresponds to the first-order scattering of the E₂g phonons at the Brillouin zone centre and originates from the in-plane vibrations of the sp² carbon atoms [30]. Meanwhile, the D-band is observed around 1300 cm^-1 and represents the formation of j-point photons of A₁g symmetry; it is also associated with double bonds C=C, meaning the more intense the band is, the higher the presence of sp² domains is. Furthermore, D-band peak intensity depends highly on the presence of disorders and defects like vacancies and edges in the carbon lattice and grain boundaries [31]. To determine the degree of oxidation of the graphene, a simple calculation of the intensity ratio of both the D and G band peaks is enough, i.e., ID/IG; the higher this ratio is, the lower the oxidation level is [32].

Finally, the second-order bands are observed from 2500 cm^-1 to 3200 cm^-1, containing one always visible peak at around 2700 cm^-1, known as the 2D band. They are used generally to determine the layers of the graphene since graphene is susceptible to stacking [33]. The chosen name of this band comes from the fact that it is the overtone of the D band, and two of the same phonons responsible for the D band are involved in the 2D band. Two other bands are sometimes reported when studying graphene Raman spectra, which are the D+G band, that can be seen around 2900 cm^-1_, and the combined overtone of the D and G bands, the 2G band around 3200 cm^-1, which is attributed to the overtone of the G band [34]. In our case, we are working with rGO, which means that the stacking of the layers is random, and since the width of the peaks is relative to the disorder, that can lead to an overlapping of the 2D band peak with the D+G and 2G, resulting in a bump like peak observed in the range of 2600 cm^-1 to 3100 cm^-1 [35].

Figure 2a,b show the Raman spectra of bare rGO and rGO doped with MnO₂ nanomaterial, respectively. For both cases, D bands are located at 1355 cm^-1 and G bands around 1590 cm^-1. For the second-order bands, three visible peaks corresponding to the 2D, D+G, and 2G bands at respectively 2697 cm^-1, 2940 cm^-1, and 3188 cm^-1 are present in the Raman spectra of the reduced graphene oxide; meanwhile, in the Raman spectra of the MnO₂ doped reduced graphene oxide we observe a bump-like peak around 2926 cm^-1, which is in agreement with the explanation made previously. Moreover, the ID/IG intensity ratio is 0.89 for rGO@MnO₂ and 0.85 for rGO, showing a slightly lower oxidation level in the doped rGO and the presence of higher number of defects, that can be caused by the doping process. Finally, it was noticed that the Raman signal in the rGO@MnO₂ sample shows a higher intensity than the pristine rGO sample, which can be correlated to the presence of the MnO₂ nanomaterial, since the presence of MOX usually leads to this increase in the intensity [36].

3.1.2. FESEM

Figure 3.a. shows the obtained FESEM images of the layers present on the surface of the graphene loaded with MnO₂ on the silicon substrate using a back-scattered electron detector (BSE). A very homogenous layers is observed covering the totality of the surface inspected. MnO₂ cannot be clearly seen, even when using a BSE detector, because of the low concentration of the nanomaterial and the small size but, when performing an EDS analysis of the surface, its presence was detected (see Figure s2 in the supplementary materials). Figure 3.b shows the FESEM image of the graphene doped with MnO₂ deposited on Kapton. Again, a very good coverage of the substrate surface is seen. However, in this case the surface of the substrate is getting rapidly charged because of the effect of the magnetic field coming from the BSE detector, leading to the formation of very bright areas. Similarly to the samples on silicon, an EDS analysis (see Figure s3 in the supplementary materials), the presence of MnO₂ was detected on the samples deposited over Kapton. Still, the MnO₂ crystals are too small to be seen in the FESEM images, just as for the sensing layer deposited on silicon.

3.1.3. HRTEM

An HR-TEM analysis was conducted to examine better the morphology of the MnO₂ nanomaterial and its incorporation in the graphene layer. Figure 4 shows an HR-TEM image of layers of graphene in addition to an interesting structure on the top right side; zooming in on this structure (Figure 4.b), a sponge-like shaped nanomaterial was observed, which was attributed to the MnO₂ after performing an EDS analysis. Moreover, when chemically mapping the chosen area's surface, a high concentration of Mn is located in the same position as the sponge-like structure, proving the presence and the likely shape of the MnO₂ nanomaterial. Figure 4.d shows an EDS map spectrum of the same area, showing the elements present in the mapping appearing Mn with the highest concentration; the other elements, except for C and O, come from the grid of the HRTEM.

3.2. Gas Sensing Results

A selection of different toxic gases and vapours was used to study the sensing properties of the pristine rGO and rGO@MnO₂ sensors. First, NO₂ was thoroughly studied with different dilutions ranging from 200 ppb up to 1000 ppb under dry air as well as under ambient moisture conditions (close to real conditions). Sensors were always operated at room temperature. Then, NH₃ was also tested as an interferent gas with a concentration of 50 ppm under the same conditions used for NO₂. Figure 5.a shows the response of the different sensors towards different concentrations of NO₂. It was noticed that the type of substrate used does not affect the response of the sensitive layer towards the analyte. rGO on silicon and Kapton have almost the same response through the studied range, with an average difference of 0.8%. The same behaviour was also seen for the rGO@MnO₂ sensors, where the average difference between the responses was 0.4%. Moreover, the most important aspect to notice is that the sensors based on rGO incorporating MnO₂ show a superior response than the pristine ones (2-fold increase in response). The loading of rGO with MnO₂ is effective at increasing sensitivity towards NO₂. In fact, rGO@MnO₂ on Kapton exhibits a higher sensitivity of 3 % ppm^-1 compared to the 1 % ppm^-1 for the pristine rGO, meanwhile for the materials deposited on silicon the sensitivity of the doped material is slightly better than its pristine counterpart with 1.8 % ppm^-1 for rGO@MnO₂ on silicon and 1.5 % ppm^-1 for on rGO silicon. The sensitivity values were evaluated from the slope of the line obtained from the linear regression of the responses of the sensor towards different concentrations of the gas. Figure 5.b shows the resistance changes of the rGO@MnO₂ on silicon substrate for 600 ppb of NO₂ and Figure 5.c shows the resistance changes of the rGO@MnO₂ on Kapton substrate for 600 ppb of NO₂ at 25% RH.

Further studies were conducted where ambient moisture was introduced via two different methods to check its effect on the sensing properties of the sensors. The first method consisted of using a controller evaporator mixer (CEM) to obtain 25% RH and the second methods consisted of using a bubbling water glass bottle that was installed between the mass flow and the chamber to humidify the air and the gas to reach a maximum RH of 70%. Ambient temperature was kept constant at 25ºC throughout the measurement period. Figure 6a,b show the calibration curves for the studied sensors at 25 and 70 % of relative humidity, respectively. Comparing the results shown in Figure 6.a (dry conditions) and 6a (25 % RH), one can notice that the response of the MnO₂-doped rGO sensors under humid conditions increases by a factor of 2.5 than when under dry conditions. For example, the rGO@MnO₂ on Kapton sensor response for NO₂ 1000 ppb at 25% humidity is 17.6%, while it is 6.4% under dry conditions. Interestingly, the responses of the pristine rGO sensors at 25% RH were enhanced by factors of 3.5 and 4, reaching similar response intensities than those recorded for MnO₂-doped rGO sensors. For example, rGO on silicon and rGO on Kapton responses to NO₂ 1000 ppb were 13.8 % and 12.7 % respectively, whereas under dry atmosphere their responses were 4.1 % and 3.2 %, respectively. Sensitivity values were calculated following the slope of the linear regression of the responses values towards different dilutions of NO₂ and compiled in Table 1.

Meanwhile, Figure 6.b reveals the calibration curves of the sensors under 70% ambient moisture. It is noticed that when increasing the concentration of water vapor, the response of the pristine rGO sensors is much more enhanced than the corresponding doped ones but the increase in the sensitivity is not so significant. When measuring 1000 ppb of NO₂, increasing the RH levels from 25 to 70 %, the response the pristine rGO sensors are doubled, while the sensitivity just increased in a factor of around 1.3. In the case of the rGO@MnO₂ sensors the increase in the response is only in the order of a factor of 1.2, but the increase in the sensitivity is higher than in the previous case especially for the sensor on silicon substrate, as can be seen in Table 1.

This behaviour of the pristine rGO layers is expected, since the same material was already reported in the literature as a humidity detector, such as in the work of Muhammed et al. where they fabricated a rGO and rGO/Fe₂O₃ components for humidity detection and the pristine material showed a high sensitivity towards RH and it increased more with the incorporation of Fe₂O₃[37]. Zhou et al. managed also to make humidity sensors with the sensitive layer of rGO/SnO₂, initially they tested the pristine SnO₂ sensitivity and response towards 75% RH and they saw these results improve by adding rGO and making rGO/SnO₂ porous film indicating the fact that rGO is a very sensitive material towards humidity [38]. Although in this work pristine rGO response towards RH increases with the increase of the humidity level, the doping of rGO with MnO₂ made the response less affected by the RH levels but the sensitivity is increased when the level of humidity increases.

Table 2 compares the results reported here with those of the literature. The sensors we report are more sensitive to NO₂ than those found in the literature. In addition, the concentrations tested in the literature are generally higher than the ones reported here, which indicates that our material is more sensitive in the low ppm concentration range. While most works totally overlook the effect of ambient humidity in the sensing properties, our material is shown to be able to detect NO₂ in a wide range of ambient moisture levels.

The selectivity of the different sensors we tested was studied under the same experimental conditions used for NO₂ detection. Different species, namely, CO (50 ppm), NH₃ (50 ppm), H₂ (500 ppm) and Ethanol (20 ppm) under dry conditions for sensors operating at room temperature were measured. Figure 7 shows the responses to these different gases or vapours. As can be seen, none of the sensors showed any response to H₂. Noticeably, the inclusion of the MnO₂ reduced the response towards CO and Ethanol, making it utterly unresponsive to these interfering gases. Therefore, the incorporation of MnO₂ improved the sensors selectivity. Nevertheless, all the sensors showed very significant responses towards 50 ppm of NH₃.

Taking into account the good responses observed for NH₃, the effect of moisture in the sensor responses to this gas was analysed. The sensors were exposed to 50 ppm of NH₃ under different humidity conditions (dry, 25 % RH and 50 % RH), always at room temperature. Figure 8.a shows the responses of the sensors to 50 ppm of ammonia for the three different humidity conditions studied and Figure 8.b shows resistance changes of the sensor pristine rGO on Kapton when exposed to NH₃. As seen in the figure, when exposed to ammonia analyte, the sensors resistance decreases in contrast to what is expected for a p-type material, this behaviour was explained later in the mechanism part. It is noticed also that the response of the rGO@MnO₂ on silicon sensor is the highest throughout all the conditions. The response of this sensor reaches a value of 18.5 % at 50 % RH, which is 4 times higher than the response of the pristine rGO on silicon. rGO@MnO₂ and pristine rGO on Kapton show basically the same behaviour and the doped one shows a slightly higher response towards NH₃, with values of 6.7 % and 5.8 % for rGO@MnO₂ and rGO respectively under dry conditions, 7.5 % and 6 % for 25 % RH and 8 % and 6.7 % at 50 % RH. Pristine rGO on silicon shows the lowest response values towards ammonia with a value of 4.6 % at 50 % RH. In essence, pristine rGO on Kapton and on Silicon substrates shows a linear-like behaviour throughout the different RH levels tested with a very little increase in sensitivity with increasing moisture levels. To have a better understanding of the behaviour of the sensors towards ammonia under humidity, it could be explained as following: Since we are working in a humid environment, the sensing layers have already adsorbed water molecules on its surface, saturating to an extent the adsorption sites especially of the pristine rGO layers. Later on, when these layers are exposed to a NH₃ gas flow, another phenomenon happens in the working atmosphere, and it can be attributed to the characteristics of ammonia itself. In fact, both H₂O and NH₃ have a strong tendency to form H bonds. Moreover, the electronegativity of the atoms determines the possibility of forming hydrogen bonds, and since oxygen is more electronegative than nitrogen, the O atom from H₂O rapidly creates a hydrogen bond with NH₃ [45] as shown in Figure 8.c. Therefore, when considering the silicon substrate sensors, the response of the pristine rGO sensor remains unchanged practically because of the phenomenon previously explained preventing ammonia molecules from getting adsorbed on the surface. Meanwhile, for the rGO@MnO₂ sensor the significant increase in the response, despite the occurrence of the hydrogen bonding of the ammonia and water molecules, can be explained by the presence of the MnO₂ nanomaterial which plays a compensatory role by creating more adsorption sites in the layer, meaning more space for the ammonia and water molecules to be adsorbed also it has been previously reported as a good NH₃ adsorbing agent [46] which explains the increase of the response of the MnO₂@rGO sensor. As for the sensors deposited on Kapton, both pristine rGO and rGO@MnO₂ showing similar behaviour can be explained by the fact that the substrate is made of a very strong hydrophobic material. Therefore, water molecules are getting repelled off of the surface resulting in a poor H₂O adsorption hence the low dependency of these sensors to the ambient moisture.

To check the position of this work in the literature regarding ammonia detection, a set of data such as response and sensitivity of other materials and sensors analysing NH₃ gas were collected and compiled in Table 3 and put in comparison with our results. Considering the same NH₃ concentration, NiFe₂O₄/rGO had a response of 1.17 meanwhile Pani@MnO₂@rGO had a response of 15.5 while heating up to 100°C. Both these materials showed lower responses than our work which is 18.6 % at 50 % RH. It is true that FeCo₂O₄/WO₃/rGO have a slightly higher response of 19.8 % at dry conditions, but in this work NH₃ concentration is 100 ppm and the working temperature is 200°C, meanwhile we are working at RT and half of NH₃ concentration.

3.3. Sensing Mechanism

Graphene and its derivatives, such as rGO, are p-type materials, which implies that usually, the interaction between rGO and oxidizing gases such as NO₂ causes a change in the local carrier concentration and, therefore, a decrease in graphene-based sensor resistance meanwhile when exposed to reducing gas such as NH₃ an increase in the resistance takes place [51]. Meanwhile, MnO₂ is an n-type nanomaterial, and when exposed to an ambient environment, a chemisorption of the oxygen molecules takes place, capturing electrons from it and releasing different oxygen species such as O₂, O^2- and O^-[52]. Moreover, the incorporation of the MOx nanomaterial (in our case, MnO₂) in the rGO results in the formation of a p-n heterojunction, causing the flow of the electrons from the MnO₂ to rGO, implying the formation of a depletion layer on the area of contact of both materials, also increasing the electron concentration in the rGO and the hole concentration in MnO₂ [53]. The exposure of the rGO@MnO₂ to air leads to the adsorption of oxygen on the surface of the p-n heterojunction material and the transfer of electrons from its conduction band to the oxygen, resulting in the formation of O₂⁻ ions following these equations [54]:

O₂ gas → O₂ (ads)

O₂ (ads)+e⁻ → O₂⁻ (ads)

When exposed to NO₂, it gets adsorbed on the rGO@MnO₂ surface and reacts with the oxygen ions and electrons from the layer following this equation, causing the decrease of the resistance of the sensor:

NO₂(ads) + O₂⁻ (ads) + 2e⁻ → NO₂⁻ + 2O⁻ (ads).

As expected, our material showed the exact same behaviour explained previously, where the baseline resistance of the sensors decreased when put in contact with NO₂ gas and recovered again when the gas flow stopped.

Although NH₃ is a strong reducing gas, the baseline resistance should increase when in contact with the gas but not in our case where the resistance of our sensors decreased. This kind of behaviour have been reported previously in the literature by A. Umar et al. and it was explained as following:

when exposed to NH₃, the interaction between the analyte and the sensitive layer results in the release of electrons back to the conduction band of the MnO₂ nanomaterial, which is believed to be the cause of the decrease of the resistance of the sensor [29] This abnormal behaviour has been observed also for pristine rGO and was reported in the work of X.Xiao et al. [55]. Finally, it is worth noting that ambient moisture usually enhances the sensitivity of graphene-based sensors [56]. Considering the room temperature detection, the water molecules probably act as a mediated adsorption site for the analyte, causing an increase in sensitivity towards the target gas [57] which is in accordance with the results we obtained where the responses of the sensors increased under the ambient moisture conditions.

4. Conclusions

Incorporating MnO₂ nanomaterial in rGO flakes to form a MnO₂@rGO nanomaterial to be integrated in chemoresistive sensors is a novel approach, since a very limited number of papers exist in the literature reporting the use of this material as a sensitive layer for gas detection. Our nanomaterial exhibits better performance and properties than other approaches previously reported in the literature, showing a set of very promising results and high responses towards low concentrations of NO₂ and NH₃. When deposited either onto rigid (silicon) or flexible (polyimide) transducing substrates, the material shows good response properties when operated at room temperature and even in the presence or ambient humidity. Furthermore, sensors show very small cross-sensitivity to other species such as hydrogen, ethanol vapours or carbon monoxide. All these properties make MnO₂@rGO an excellent candidate nanomaterial for the inexpensive, chemoresistive detection of nitrogen dioxide or ammonia in real life environments.

5. Patents

Patent number ES2678419A1