Highly Responsive Pd-Decorated MoO3 Nanowall H2 Gas Sensors Obtained from In-Situ-Controlled Thermal Oxidation of Sputtered MoS2 Films.

Among transition metal oxides, MoO3 is a promising material due to its layered structure and different oxidation states, making it suitable for different device applications. One of the methods used to grow MoO3 is radio frequency magnetron sputtering (RFMS), which is the most compatible method in industry. However, obtaining nanostructures by RFMS for metal oxides is challenging because of compact morphology film formation. In this study, α-MoO3 with vertical nanowalls is obtained by a two-step process; deposition of magnetron-sputtered MoS2 vertical nanowalls and postoxidation of these structures without changing the morphology. In situ transmittance and electrical measurements are performed to control the oxidation process, which shed light on understanding the oxidation of MoS2 nanowalls. The transition from MoS2 to α-MoO3 is investigated with partially oxidized MoS2/MoO3 samples with different thicknesses. It is also concluded that oxidation starts from nanowalls perpendicular to the substrate and lasts with oxidation of basal planes. Four different thicknesses of α-MoO3 nanowall samples are fabricated for H2 gas sensors. Also, the effect of Pd deposition on the H2-sensing properties of sensors is deeply investigated. An outstanding response of 3.3 × 105 as well as the response and recovery times of 379 and 304 s, respectively, are achieved from the thinnest Pd-loaded sample. Also, the gas-sensing mechanism is explored by gasochromic measurements to investigate the sensor behaviors under the conditions of dry air and N2 gas as the carrier gas.

Introduction

Transition metal oxides such as WO3, MoO3, and V2O5 are functional materials that has been attracting the attention of scientists in the last few years due to their wide and tunable band gaps, variable electronic and optical properties due to reduction and oxidation properties, as well as their layered structure in hydrate or anhydrate crystal phases.[1] Among them, MoO3 and its substoichiometric forms have a wide range of applications such as in electrochromic devices,[2−4] gas sensors,[5−7] optoelectronics,[8] catalysis,[9] supercapacitors,[10] and solid lubricants.[11] Stoichiometric MoO3 is a wide band gap semiconductor (∼3 eV) that might be narrowed by reduction into MoO (2 < x< 3).[12,13] In the reduction process, oxygen defects introduce more electrons into the conduction band. This presents itself as the narrowed band gap and better conductivity, which become possible by Mo6+ reduction to Mo5+ and finally to Mo4+ to achieve semimetal MoO2.[13] Through this process, the optical properties of MoO3 change significantly. Therefore, it is among the most studied materials for various devices, especially electrochromic and gasochromic devices.

MoO3 has two thermodynamically stable common crystal phases, stable orthorhombic α-MoO3 and metastable monoclinic β-MoO3.[12] The α-phase attracts much attention due to its layered crystal structure and its ability to form two-dimensional (2D) atomically thin morphologies, which is a state-of-the-art material after discovering graphene.[13] The fundamental building blocks of both α-MoO3 and β-MoO3 crystal structures are MoO6 octahedrons that bond differently.[14] In orthorhombic α-MoO3, distorted MoO6 octahedrons link with edge-sharing zigzag rows and are corner-sharing along [001] and [010] directions, respectively, to form planar double-layer sheets by weak van der Waals force between them.[15,16] Unlike the α-phase, monoclinic β-MoO3 does not form a layered crystal structure, and MoO6 octahedrons only share corners to form a ReO3-like crystal structure by corner-sharing.[13,15] Depending on the applications, α-MoO3 and β-MoO3 have their own advantages and disadvantages. The transition from the β-phase to the α-phase at temperatures above 350 °C has been reported.[17]

Understanding the shape and size effects of materials on their properties and tracing their behaviors in contact with the environment around them have enabled scientists fabricate different nanostructures with improved optoelectronic properties, which led to the fabrication of materials with a high aspect ratio. In the case of MoO3, nanostructures such as nanosheets,[18] nanorods,[19] nanowires,[20] quantum dots,[21] nanotubes,[22] and nanoparticles[23] have been reported so far via different material growth methods such as chemical vapor deposition,[24] radio frequency (RF)/DC magnetron sputtering,[25] thermal evaporation,[26] electron beam evaporation,[27] spray pyrolysis,[28] hydrothermal synthesis,[29] electrochemical anodization,[30] electrochemical deposition,[15] and exfoliation.[31] All the material growth methods mentioned above have their advantages and disadvantages. However, the sputtering method has some advantages that distinguish it from the growth methods mentioned above, such as its compatibility with industry, large area coatings, and the possibility of mass production. In addition, it is easy to control growth parameters to perform uniform and repeatable depositions. On the other hand, it is also inexpensive compared to other high vacuum-required growth methods. Besides these advantages, it is challenging to grow nanostructures by tuning the controllable growth parameters in magnetron sputtering. It is common that sputtered materials form a compact thin film morphology. In order to achieve nanostructures via the magnetron sputtering method, some extra post-treatments, usage of catalysis, and substrate treatments might be required. For example, Zhang et al. prepared monolayer hollow spherical tungsten oxide films by combining magnetron sputtering with the colloidal crystal template.[32] Instead of growing MoO3 nanostructures by the methods mentioned above, there are also some reports on the full or partial oxidation of MoS2 to achieve MoO3 or MoS2/MoO3 hybrid composite structures. In one of them, 2D MoO3 nanosheets were synthesized by chemical oxidation of exfoliated MoS2.[3] Similarly, MoS2/MoO3 heterostructures were prepared by plasma oxidation of MoS2 layers.[33] On the other hand, the thermal oxidation method is preferred by some researchers.[34−37] However, most of them did not successfully preserve the morphology during the conversion from MoS2 to MoO3 material.[38,39]

One of the most used applications of MoO3 materials is the gas sensor devices.[40−42] Metal oxide semiconductors are preferred because of their resistance to a harsh environment, ease of production, low cost, and repeatability. Also, due to the ease of production and the ability to miniaturize, chemiresistive semiconducting metal oxide gas sensors and their heterostructures attract much attention among other types of gas sensors.[43] In particular, materials for H2 gas sensing have gained importance because they are supposed to be one of the materials with the highest potential as alternative energy sources. Increasing the usage of H2 gas as an alternative to fossil fuels and considering its explosive potential, detecting H2 gas leakages in storage containers and gas tanks at low concentrations (under 1000 ppm) and relatively low temperatures becomes highly important. H2 is an odorless gas, which has a high propagation constant, a low ignition energy, and a wide burning range (4–75%) in air.[44] A variety of metal oxides such as SnO2,[45] WO3,[46,47] MoO3,[18,48,49] ZnO,[50] and In2O3[51] have been utilized for H2 gas sensors. Because of the safety concern of H2 gas, it is very important to achieve high responses against low concentrations at low working temperatures of H2 gas. It is important to achieve easily readable current values in the case of chemiresistive gas sensors. The oxide-based sensors generally operate at relatively high temperatures with low sensitivity.[52] Pd-decorated MoO3 has become a promising material among the metal oxide semiconductor gas sensors for detecting H2 gas due to the chromic nature of MoO3 and its ability to undergo reversible reduction and oxidation reactions by ion injection and extraction.

Here, we report thermal oxidation of sputtered MoS2 to achieve α-MoO3 without a morphological change of the film to use in highly responsive H2 gas sensors. In situ conversion of MoS2 to MoO3 was performed through optical and electrical measurements. Furthermore, the potential of the converted materials was explored by fabricating H2 gas sensors with different thicknesses to study the effect of thickness and nanostructures on the gas-sensing properties of achieved MoO3. An outstanding sensitivity of 3.3 × 105 for 1000 ppm was achieved from the thinnest MoO3 sample, one of the highest reported responses for MoO3-based H2 gas sensors at a relatively low operating temperature (100 °C). Furthermore, the effect of Pd deposition on H2 gas-sensing performance is discussed in depth in the H2 Gas-Sensing Performances and Mechanism section.

Experimental Section

The Experimental Section in this study is divided into four parts: (a) film deposition, (b) dynamic in situ transmittance and electrical measurements during the oxidation process, (c) film characterization, and (d) H2 gas sensor measurements.

Film Deposition

MoS2 nanowalls were deposited via a RF magnetron sputtering (RFMS) system using a MoS2 target with a diameter of 2 in. and a purity of 99.99%. Quartz and ITO-coated glasses with a sheet resistance of 20 Ohm/cm were used as substrates for gas sensor measurements and material characterizations. Both quartz and ITO substrates were cleaned ultrasonically in isopropanol, ethanol, and deionized water (10 min for each step) and dried under N2. The distance between the target and substrates was about 7 cm, and the gun position was fixed perpendicular to the substrates. The chamber was evacuated to a base pressure of 2 × 10–6 Torr. Sputtering was performed at a substrate temperature of 300 °C and an argon pressure of 16 mTorr with a sputtering power of 120 W. Four different thicknesses of MoS2 were sputtered. Depending on the growth time, films were labeled Z0.5, Z2.5, Z7.5, and Z30, corresponding to 30 s, 2.5 min, 7.5 min, and 30 min growth time with thicknesses of 40, 115, 370, and 1440 nm, respectively, as shown in the cross-sectional field emission scanning electron microscopy (FESEM) images presented in Supporting Information, Figure S1.

In Situ Transmittance and Electrical Measurements

The dynamic in situ transmittance and two-probe electrical measurements were performed during the oxidation process in a chamber having an optical window. A tungsten–halogen lamp was used as the light source. MoS2 samples to be oxidized were heated up to 380 °C, and oxygen gas was let in during preheating up to 380 °C and during the oxidation process. For two-probe electrical measurements, interdigitated electrode (IDE) contacts (Pt contacts deposited with a RF-magnetron sputter) with a 400 μm distance between them were used.

Film Characterization

A Zeiss Sigma 300 field emission scanning electron microscope was used to investigate the morphological characteristics of MoS2 and MoO3 films. A WITech alpha 300R was used for micro-Raman measurements of MoS2, MoS2/MoO3, and MoO3 films. For crystal structure characterization of films, a PANalytical Empyrean X-ray diffraction (XRD) system was used. X-ray photoelectron spectroscopy (XPS) measurements for determining the material composition were performed with a Specs Flex-Mod XPS system equipped with a 150 mm radius hemispherical energy analyzer with a 2D charge-coupled detector. The measurements were performed using an Al anode with a Kα energy of 1486.71 eV. The depth profile measurements were conducted by 3 keV Ar+ ion sputtering with 300 s of etching for each cycle.

H2 Gas Sensors (Sample Preparation and Performance Measurements)

For H2 gas sensor measurements, Pt IDE contacts were deposited first on a quartz substrate by sputtering using a shadow mask. Then, MoS2 films with different thicknesses were deposited on contacts. Next, oxidation of samples of different thicknesses was performed in tube furnaces for 2 h. At the final step of the device fabrication, sample surfaces were activated by Pd deposition under 27 mTorr and 25 W power conditions for 7 s by RFMS. Gas sensor measurements were performed using a homemade gas sensor system with a chamber of 0.5 L volume. All the samples were held under an applied voltage of 0.5 V using a KEITHLEY 487 picoammeter/voltage source. H2 gas with various standard cubic centimeters per minute (sccm) was mixed with 500 sccm of dry air to achieve the desired H2 part per million (ppm) levels between 100 and 1000 ppm at three different measurement temperatures (100, 200, and 300 °C), which were set before the gas-sensing measurements. Measurements between 10 and 90 ppm were performed for sample Z0.5 at an operating temperature of 200 °C.

Results and Discussion

In Situ Measurements

The thermal oxidation process was used to convert sputtered nanowall morphology MoS2 films into MoO3 without changing the surface morphology to take advantage of the large surface-to-volume ratio of the films in sensor applications. After many trials to keep the surface morphology with no change (Supporting Information, Figure S2), it has been realized that certain oxidation conditions are needed to control during the thermal oxidation, such as the oxidation temperature and time. It has also been confirmed that these parameters depend on the MoS2 thickness. Therefore, it is decided to control the oxidation process by in situ transmittance and two-probe electrical measurements. These in situ measurements allow us to trace the sulfur-to-oxygen ratio by simply checking the samples’ opaqueness because the decreasing sulfur to oxygen ratio presents itself as increasing transparency and decreasing conductivity. The predictable chemical equation for the thermal oxidation process of MoS2 is shown below

In some applications, it is inevitable to use transparent conductive oxides (TCOs) as a substrate for MoO3 films, that is, electrochromic devices, in which the transparency and conductivity of electrodes are essential parameters. Moreover, a longer oxidation time means the disappearance of oxygen vacancies in TCO, which presents itself as a decrease in conductivity.[53,54] Therefore, for such situations, controlling the oxidation time might be an important parameter.

MoS2 films with two different thicknesses, namely, Z2.5 and Z30 were used for in situ measurements. Changes in optical and electrical properties of MoS2 films with the transmittance during the oxidation process are shown in Figure a,b. The data collection time for the transmittance change is every 60 s. In order to avoid crowd plots, sparse data is used in the plot. As seen in Figure a,b, there is a significant change in the transparency of the films, which turned from opaque to transparent (see the inset image). This indicates that the oxidation process is successful. Figure c shows the transmittance change of films versus time monitored at a wavelength of 550 nm. As seen from the figures, both samples’ transparency increase as the oxidation time gets longer. The films’ starting transmittance values change due to the thickness differences in the samples. This is so crucial for the research that we have conducted because it allows us to find approximate oxidation time. The term approximation is used here because finding the exact oxidation time from the graph is hard. The change in transmittance becomes slower after some duration, as seen from the plot. At this point, it is not easy to understand whether all sulfurs are gone or some remain. On the other hand, this plot, in its way, gives a strong idea about the approximate oxidation time. As seen from Figure c, the oxidation times for Z2.5 and Z30 are about 13.2 and 23.0 min, respectively. The differences in oxidation times between Z2.5 and Z30 are predictable due to the thickness differences between the films. It can be seen that the slope of Z30 is greater than that of Z2.5, which means faster oxidation. In addition, the tendency of dynamic transparency of Z30 gives rise to flattening quicker than that of the sample Z2.5, but transmittance of Z30 still increases very slowly.

Figure 1

In situ optical and electrical measurements during oxidation. (a) Transmittance change for sample Z2.5, (b) transmittance change for sample Z30, (c) transmittance changes of Z2.5 and Z30 vs time at a wavelength of 550 nm, and (d) current changes of Z2.5 and Z30 during the oxidation vs time.

Figure d shows the current versus time changes of MoS2 films during the oxidation process. It can be noted that the graph is semilogarithmic. As seen from the plot, the current value decreases as the oxidation time gets longer, which shows basically that converted MoO3 films have a larger resistance than MoS2. As seen from the figure for sample Z2.5, current increases from 1.2 to 12.0 μA. The increase in the current value in this interval is related to the increase in the temperature from RT to 380 °C. However, the temperature reached at the maximum point of the current is 375 °C. The current starts to decrease rapidly at a temperature of 375 °C, indicating that the oxidation process has started handling the current decrease. After 586 s, the current value reaches 1.7 nA for the sample Z2.5. For sample Z30, the current starts from 1.2 μA and reaches a value of 14.2 μA. Unlike the sample Z2.5, the current starts to decrease for Z30 at a temperature of 360 °C. The Z30 sample takes about 890 s to achieve a current value of 1.7 nA. This experiment proves a successful oxidation process with the in situ measurements and gives some information about the conductivity of the samples. It also allows us to stop the oxidation process by checking the sample’s transparency or conductivity to obtain a semioxide composite of MoS2/MoO3.

Material Characterization

In situ optical and electrical measurements confirm the successful thermal oxidation process of MoS2 into MoO3. In order to follow the morphology, as well as the structural and optical properties after the MoS2–MoO3 conversion, Raman, XRD, and the FESEM measurements were performed on the end materials together with the MoS2/MoO3 composite materials in which the thermal oxidation process was completed at some point. Figure a–h shows the results of these measurements. However, before we elaborate on how oxidation takes place in MoS2 nanowalls, first, we offer a glimpse on the growth mechanism of MoS2 nanowalls. Figure a,b presents FESEM images of 15 and 30 s grown MoS2 films on the ITO substrate. As seen from Figure a, no vertically grown MoS2 nanowall-shaped morphology is seen in 15 s of deposition, but ITO film grains can be seen. The result is because of horizontally grown MoS2 with a 2D-like morphology at the first step of deposition. As the growth time continues, the nanowall-like vertical structures start to appear, as seen in 30 s grown MoS2. The appearance of 3D structures might be due to accruing stress between boundaries of grown sheets or defect sites that start during the first steps of deposition. After a while, it appears to be the beginning of vertical growth.[55] A detailed study of transition from 2D to 3D growth forming nanowalls in the WS2 film deposited by RFMS, which was predicted to be the same for MoS2, was reported in our previous study.[55]

Figure 2

(a,b) FESEM images of 15 and 30 s MoS2 grown on the ITO film, respectively. (c,d) XRD and Raman shifts of MoS2, MoS2/MoO3, and MoO3 for 2.5 and 30 min grown films, respectively. (e,f) FESEM and cross-sectional FESEM images of 2.5 min grown MoS2 and MoO3 after thermal oxidation, respectively. (g,h) FESEM and cross-sectional FESEM images of 30 min grown MoS2 and MoO3 after thermal oxidation, respectively.

Figure c shows the XRD graphs of MoS2, partially oxidized MoS2/MoO3, and MoO3 films of Z2.5 and Z30 samples. The (002) plane peak intensity, related to basal planes oriented parallel to the substrate, has decreased with a longer deposition time. However, the peak intensity of the (100) plane, which is related to edge planes perpendicular to substrates, increases as the deposition time gets longer.[56] This might also prove the 2D and 3D growth stages in sputtered MoS2. A similar observation of 2D and 3D growth stages has been reported earlier.[57] This structural anisotropy in MoS2 films also gives rise to anisotropic oxidation. It is known that oxidation of MoS2 starts from edge crystal planes, and oxidation is very easy for these planes compared to the basal planes.[58] The (002) basal plane is more protected against oxidation. However, as already reported, the edge plane, that is (100), is in more contact with the environment and is more reactive to oxidation.[59] As shown in Figure c, there is no linear change in the oxidation time with increasing thickness. The oxidation time for Z2.5 is 13.2 min, while it is 23.0 min for Z30.

This result is because of the formation of the same (002) basal plane in both thin and thick films, where the first nanowalls are extremely fast oxidized, followed by very slow oxidation of the (002) basal plane. As shown in Figure e,g, the thickness of walls is approximately about 10 nm making oxidation easier. In addition, sputter-grown MoS2 films are highly sulfur-deficient,[60] and this might be helpful for the oxidation process. XRD patterns show that the edge plane (100) is almost diminished in the partially oxidized MoS2/MoO3 sample. However, the (002) plane can still be observed, belonging to the MoS2 basal plane, which indicates that the oxidation starts from the edge planes. Furthermore, it can be seen that MoO3-related XRD peaks start to appear. Overall, all the MoS2-related XRD peaks entirely disappear, and the appearance of the (020) peak and several other XRD peaks belonging to the oxide indicates the successful transformation into α-MoO3.

Figure d shows the Raman spectra of MoS2, MoS2/MoO3 composite structure, and MoO3. Both E2g1 and A1g, which are related to in-plane and out-of-plane Raman optical phonon modes of MoS2, respectively, appear for each different thickness. B1g and A1g modes of α-MoO3 start to appear for the partially oxidized MoS2/MoO3 structure. After oxidation, all the MoS2-related Raman peaks disappear, indicating that fully oxidized α-MoO3 is achieved. The primary purpose of this study is to investigate the oxidation of MoS2 nanowalls without changing the morphology of films. Figure e–h shows the FESEM images of 2.5 and 30 min grown MoS2 films and MoO3 after oxidation. The FESEM images of MoS2 show the homogeneous distribution of nanowalls throughout the substrate. It is also seen from FESEM images that increasing the deposition time gives rise to extra branches growing on nanowalls. Cross-sectional SEM images also show vertically grown nanowalls (for cross-sectional FESEM images of Z2.5, Z7.5, Z10, and Z30, see Supporting Information, Figure S1). From transmission electron microscopy images of obtained MoO3, it can be seen that the Mo to Mo distance between two layers is about 1.37 nm (Supporting Information, Figure S3) which is close to the theoretical value of 1.4 nm.[13]Figure f,h shows the FESEM images of thermally oxidized MoO3 grown for 2.5 and 30 min. As seen from FESEM images, there is no significant change in the morphology of oxide films compared to the sulfide films.

Figure a shows the XPS survey spectra of 30 min grown MoS2, fully oxidized α-MoO3, as well as the partially oxidized MoS2/MoO3 sample. The figure shows only peaks belonging to the C, O, S, and Mo elements. The C element peak is attributed to surface contamination. This adventitious C–C binding energy is used to correct the spectra obtained. The inset figure shows the energy regions of O and S elements in the survey spectra. The intensity of the O element peak is the weakest in the MoS2 sample, which is due to the surface oxidation, and it is increased for the oxidized samples. On the other hand, the S element peak is the maximum for the MoS2 sample, as shown in the inset and the high-resolution spectra in Figure b, while it diminishes for the fully oxidized α-MoO3 sample. The partially oxidized MoS2/MoO3 sample shows minimal S element peak intensity as seen in the spectra, although we tried to obtain a more significant intensity by taking out the sample from oxidation in a few minutes. Nevertheless, it is realized that the surface is fully oxidized upon starting the oxidation because the nanowall morphology of MoS2 is highly active for oxidation, which is already discussed in the XRD analysis. The calculated atomic percentages of the elements are given in the inset table. This data clearly shows the successful oxidation of the sputtered MoS2 samples. Figure c shows the high-resolution Mo 3d region spectra of the three samples investigated in the study. An S 2s peak is seen for the MoS2 sample, while no signal appears for the fully and partially oxidized samples. It is apparent in the figure that the Mo binding energy peak appears on the lower energy side, indicating the Mo4+ oxidation state of the MoS2 sample. This confirms the formation of MoS2. On the other hand, partially and fully oxidized samples show similar binding energy peaks, shifting to a higher binding energy corresponding to the Mo6+ oxidation state. This confirms the formation of MoO3 after thermal oxidation.

Figure 3

30 min grown MoS2, partially oxidized MoS2/MoO3, and fully oxidized α-MoO3 samples. (a) XPS survey spectra, (b) high-resolution spectra of the S element, and (c) high-resolution spectra of the Mo element. Depth profile for the 30 min grown and partially oxidized MoS2/MoO3 sample, (d) atomic percentage variation with the depth profile for Mo, S, and O elements, (e) Mo 3d region, and (f) S element 2p region.

Figure d–f shows the depth profile of the 30 min partially oxidized sample, which shows how the oxidation process occurs. The Mo 3d high-resolution depth profile is obtained for a total of 30 cycles of measurements. The Mo 3d binding energy peaks appear at 233.1 and 236.1 eV, which are similar up to the sixth etch cycle. These binding energy values correspond to the Mo6+ oxidation state, confirming the formation of MoO3.[61] As seen in Figure e, the Mo peak displays a binding energy shift as the etch cycle number increases. At a lower binding energy, the shoulder starts to appear at the sixth etch cycle. The evolution of the shoulder peak can be clearly seen in the spectra in which it first appears as a shoulder, then becomes more prominent in terms of intensity, and finally becomes a separate dominant peak. Although some reports show the reduction of MoO3 to lower oxidation states with Ar+ ion sputtering, it can be seen in the S element depth profile in Figure f that the S element starts to appear at the sixth etch cycle, where exactly a shoulder peak appears first in the Mo 3d binding energy region. The inset in Figure e shows the Mo 3d spectra for some etch cycles. A shift of the binding energy peak for the entire depth of the sample to a lower binding energy region corresponding to the lower oxidation states of the Mo element can be seen in this inset figure clearly. In the S element’s depth profile, the S element appears at the sixth cycle and becomes highly dominant as the number of etch cycles is increased. These two Mo and S elements’ depth profiles indicate that oxidation of MoS2 starts from the surface and continues for the entire depth of the sample, as previously discussed in the XRD data. It is clear that oxidation did not definitely occur beyond the 14th etch cycle for the partially oxidized MoS2/MoO3; beyond that cycle, bare, not oxidized MoS2 is present, which is confirmed by the Mo 3d and S 2p photoelectron binding energy peaks.

H2 Gas-Sensing Performances and Mechanism

MoO3 films with four different thicknesses corresponding to their deposition times of 0.5, 2.5, 7.5, and 30 min, indicated as Z0.5, Z2.5, Z7.5, and Z30, respectively, are studied. Figure a–e shows dynamic responses, response, and recovery times, and response (IH/Iair) of MoO3 sensors at different operating temperatures and for different H2 gas concentrations. Figure a shows dynamic responses of MoO3 nanowalls different thicknesses for 1000 ppm H2 gas operating at a temperature of 100 °C. It can be observed that the sensor responses decrease as the thickness increases. In order to determine the saturation of the current in the presence of H2 gas, it took almost 42 min for sensor measurements for the Z2.5 sample, despite it being ∼7 min for Z0.5, which shows the current saturation is much faster for this sensor than that for other sensors used in the current study. The highest response is 3.2 × 107 for sensor Z2.5, followed by Z7.5 with a 2.2 × 106 response, as seen in Figure e. However, as shown in Figure a, the maximum current value reached for Z7.5 is smaller than that for Z0.5, which shows a response of 3.3 × 105. The difference comes from the conductivity change of the samples with the thickness, which leads to different Iair values. Apparently, this value is minimum for the thicker samples, which is on the order of picoamperes, while it is on the order of nanoamperes for the thinnest sample. Although Z2.5 displays the highest response, it should be noted that the responsivity and recovery times should also be considered. Therefore, sensor Z0.5, with the response and recovery times of 379 and 304 s, respectively, shows the best sensing performance at a temperature of 100 °C for 1000 ppm H2 gas. It should be kept in mind that the recovery current could not reach the starting current value, Iair, at an operating temperature of 100 °C. The starting current value of the Z0.5 sensor is 2.0 × 10–9 A, while after recovery, it reaches a current value of 3.6 × 10–7 A. Although the difference is about 2 orders between the starting and recovery values, this recovery value seems negligible when it is considered that the response of the sensor is more than ∼105.

Figure 4

(a) Dynamic responses of Z0.5, Z2.5, Z7.5, and Z30 for 1000 ppm H2 gas at an operating temperature of 100 °C. (b) Dynamic responses of Z0.5, Z2.5, Z7.5, and Z30 for 100, 500, and 1000 ppm H2 gas at an operating temperature of 200 °C. (c) Dynamic responses of Z0.5, Z2.5, Z7.5, and Z30 for 100, 500, and 1000 ppm H2 gas at an operating temperature of 300 °C. Insets show 200 and 300 °C semilogarithmic dynamic responses for the 500 ppm H2 level. (d) Response and recovery times of Z0.5, Z2.5, Z7.5, and Z30 for 1000 ppm H2 concentration at different operating temperatures. (e) Response bar of Z0.5, Z2.5, Z7.5, and Z30 for 1000 ppm H2 concentration at different operating temperatures. (f) Dynamic responses of Z0.5 at an operating temperature of 200 °C for H2 concentrations under 100 ppm. (g) 11 Cycles for Z0.5 at an operating temperature of 300 °C for 1000 ppm H2 concentration.

The dynamic responses of four sensors are demonstrated at an operating temperature of 200 °C in Figure b. Sensor Z0.5 shows the highest response of 2.4 × 104, followed by Z2.5 with a response of 1.8 × 104 under 1000 ppm H2 gas conditions. In addition, the figure shows that the Z0.5 sensor reaches the highest current value in the presence of H2 gas. It is preferable for a gas sensor to reach the current value as high as possible, and the Z0.5 sensor is therefore even better due to the ease of detection of higher current values. Sensor Z0.5, due to its responsivity and recovery times of 633 and 224 s, respectively, shows the best performance among other sensors. In addition, Z0.5 shows the maximum response for 100 ppm at an operating temperature of 200 °C. For that reason, measurements of H2 concentrations under 100 ppm were performed for the Z0.5 sensor as shown in Figure f. At a temperature of 200 °C, a response of 0.84 (%84) for 10 ppm H2 gas is measured for Z0.5. Figure c indicates the dynamic response of sensors for 100, 500, and 1000 ppm at an operating temperature of 300 °C. Z0.5 demonstrates a maximum response of 7.4 × 104 in comparison with other sensors. However, as shown in Figure b,c, the maximum current value that the Z0.5 sample reaches at 200 °C is about 3 times larger than that at 300 °C. Regarding the recovery point, the dynamic responses at a temperature of 300 °C show the best recovery. The sensors recover fully and reach the starting current before H2 is introduced into the chamber, as seen in the inset figure of semilogarithmic plots for 500 ppm H2 gas. Figure g shows the 11-cycle sensor response for sample Z0.5 at 300 °C for 1000 ppm H2 gas and the repeatability of the sensor. As seen in the figure, the response does not change significantly after 11 cycles of measurements.

Further experiments were conducted to study the hydrogen-sensing mechanism and the effect of Pd catalysis on H2-sensing performances of MoO3 nanowalls. The H2 gas-sensing mechanism of metal oxide-based gas sensors is generally explained as follows: oxygen molecules get adsorbed on the surface of metal oxide, which extract electrons from the conduction band of the semiconductor and reduce the conductivity of the material, resulting in the formation of negatively charged oxygen species at different operating temperatures including O2–, O–, and O2–.[62] In the case of n-type semiconductors, this reduction in conductivity is reversed by exposure to reducing gases such as NH3 or H2. Hydrogen bonding with oxygen species that are adsorbed on the semiconductor surface removes surface-adsorbed oxygens that introduce electrons into the conduction band and lead to an increase in the conductivity of metal oxide.[42] The formation process of negatively charged oxygen ions can be summarized as follows[63,64]

Depending on the nature and structure of materials and the behavior with different target gases, each gas sensor material has a different optimum operating temperature. The H2 gas-sensing mechanism of chromogenic materials such as MoO3 cannot be explained just by the oxygen adsorption model because H2 molecules and H+ ions not only incorporate with adsorbed oxygens but also attack different coordinated oxygens in the structure of the material in order to reduce MoO3 to MoO3–. In order to improve the gas-sensing properties of metal oxide semiconductors, there have been different approaches applied so far. One of them is decorating or doping with noble metals such as Pt, Pd, or Au, which act as catalysts.[62] To find out whether noble metals change the work function of metal oxide, there are two approaches, which are “chemical sensitization” and “electronic sensitization”.[65] Chemical sensitization is the disassociation of the target analyte molecule called the spillover effect. In this case, the deposition of metal catalysts does not change the conductivity of the sensing material. Electronic sensitization refers to the temporary oxidation of a catalytic metal, which gives rise to the formation of a depletion layer. This depletion layer reduces the conductivity of the sensing material. Oxygens on the surface of the catalyst metal are removed when the H2 gas is introduced. This appears as thinning of the depletion layer, which increases material conductivity.[65]

Gasochromic measurements have been designed to determine the sensing mechanism related to hydrogen intercalation. The inset in Figure shows the schematic illustration of the sensor materials, explaining the sensing mechanism of gas sensors produced in the study. Black dots on the surface represent the Pd nanostructures as a catalytic material. FESEM images of Pd-decorated MoO3 nanowalls are shown in Figure S5. The left side of the figure represents the sensing mechanism at 100 °C, while the right side shows that at 300 °C. Figure a,b shows the transmittance changes of samples Z2.5 and Z30 at a wavelength of 700 nm for 1000 ppm H2 gas at operating temperatures of 100 and 300 °C, respectively. There are no significant changes in transmittance for both sample measurements performed at 300 °C; however, significant changes of 25.5 and 15.0% transmittance are observed for samples Z2.5 and Z30, respectively, at 100 °C. These results show similarity to those of the electrical sensor measurement in which the sensor response at 100 °C is higher than that at 300 °C. Gasochromic measurements indicate that more H+ ions intercalate at 100 °C temperature than those at 300 °C. There might be a couple of reasons for this to observe. First of all, it might be possible with more H2 gas desorption from the surface at higher temperatures by removing the active number of H2 gas molecules, as seen in the inset figure circle number 2. For this reason, the spillover zone at 300 °C is smaller than that at 100 °C, as seen in the schematic illustration. On the other hand, the MoO3 sites reduced by the H2 gas are oxidized easily by the oxygen molecules in dry air with rising temperature[66] because 300 °C is very close to the oxidation temperature of MoS2 as it is shown by the oxidation at around 380 °C from the in situ measurements. Actually, there is a trade-off between the oxidation of the sites by the oxygen in dry air and reduction by H2 gas. It is this trade-off that creates an equilibrium slab of active thickness, which basically determines the conductivity change of the sensors. In order to confirm this hypothesis, gasochromic measurements under H2 gas mixed with N2 were performed. As seen in Figure c, the transmittance change of Z30 under 1000 ppm H2 at a temperature of 100 °C and under N2 conditions at a wavelength of 700 nm is more than 70.0%, which is about 4.8 times higher than that of H2 mixed in the air. The inset figure shows the colored state of the sample. The response is much higher under N2 gas compared to that under dry air. It is known that N2 itself removes surface-absorbed oxygen species, which boosts the reduction of MoO3, and in the case of the nanostructure, this effect is stronger than that on compact films because more surface is exposed to N2 gas. This also indicates that the equilibrium slab thickness is not well enough to change the conductivity of the thick sensors compared to that of the thinner samples, indicating the higher response in the thinner samples such as Z2.5 and Z0.5.

Figure 5

Transmittance changes of Z2.5 and Z30 for 1000 ppm H2 gas in air (a) at an operating temperature of 100 °C, (b) at an operating temperature of 300 °C, (c) transmittance changes of Z30 for 1000 ppm H2 gas in nitrogen at an operating temperature of 100 °C, and (d) dynamic sensor responses of Z0.5 with and without Pd catalysis at different operating temperatures. (e) Average current changes in sample Z0.5 at different operating temperatures for 1000 ppm H2 gas at different time intervals. The figure in the middle shows the schematic illustration of the MoO3 gas sensor.

In order to understand the effect of Pd catalysis, sensor measurements were performed for the sample Z0.5 without Pd loading. Figure d shows the dynamic sensor responses of Z0.5 with and without Pd catalysis at 100 °C temperature. Sensor responses obtained for Z0.5 under 1000 ppm H2 in air without Pd loading are 4074, 114, and 98 fold at 100, 200, and 300 °C operating temperatures, respectively. These values are 79.7 (at 100 °C), 207 (at 200 °C), and 754 (at 300 °C) times smaller than that obtained at the same operating temperature (see Supporting Information, Figure S4) for Pd-loaded samples. This indicates that the spillover effect occurs at different temperatures and boosts the sensor sensitivity and response time. At all operating temperatures for Pd-loaded samples, sensor responses increase significantly compared to those of bare samples. On the other hand, the sensor responsivity of Pd-loaded sample with considering temperature, shows a similar trend to that without Pd catalysis. Figure e shows the average current change rates of Pd-loaded Z0.5 sample at different operating temperatures for different time intervals of sensing times. As shown in Figure e, for the first 100 s of gas sensing, the average current change rate for 300 °C operating temperature is 2 nA/s, and it reaches the maximum value of 31 nA/s between 200 and 250 s. The current variation by time is almost constant through 50–400 s time intervals. However, the average current change rate starts with 150 pA/s for the first 100 s and then it reaches the maximum value of 6.5 μA/s at a time interval of 250–300 s. The rate variation is much higher than the result obtained at 300 °C. These different current change rates at different temperatures at the beginning and in the middle of the gas-sensing process can be explained by the following steps of sensing mechanism: H2 molecules first reach the surface, then H+ ions are formed by the spillover effect, these ions play a role in removing adsorbed surface oxygens, causing the conductivity change which is relatively small at the beginning of the process. Then, more H+ ions are formed due to the spillover effect, and the H+ intercalation begins in the inner sections of MoO3 due to the field-effect caused by the H+ in contact with each other at the spillover intersection zones, as seen in the schematic representation of the gas sensor in Figure . The reduction of MoO3 starts at the bulk, forming water vapor that leaves the structure, causing a dramatic increase in the current change rate as illustrated in the inset figure circle number 3 of Figure . Room-temperature (RT) H2 gas-sensing measurement for sample Z0.5 for 1% H2 gas concentration is shown in Supporting Information, Figure S6. A response of 7.5 × 107 was achieved at RT. The data are included in the Supporting Information due to the long sensing and recovery times.

Conclusions

In conclusion, MoS2 nanowall films with four different thicknesses (40, 115, 370, and 1440 nm) deposited by RFMS were successfully turned into α-MoO3 by thermal oxidation without changing the morphology of the films, which are confirmed by XRD, Raman, FESEM, and XPS measurements. In situ measurements during the oxidation process were performed to control the oxidation. They also show that the nanowalls perpendicular to the substrate turn to the oxide material faster than the basal planes. It is concluded from the depth profile XPS measurement that the oxidation starts from the nanowalls found on the surface and continues through the depth of the material. The conversion from MoS2 to MoO3 is also confirmed by Raman and XRD measurements. H2 gas-sensing performances of MoO3 films were determined on films with different thicknesses at 100, 200, and 300 °C operating temperatures. The thinnest sample with a thickness of ∼40 nm shows the outstanding performance with a response of 3.3 × 105-fold. Response and recovery times of 379 and 304 s, respectively, are achieved at an operating temperature of 100 °C. In addition, gasochromic measurements were performed to explain the H2-sensing mechanism of MoO3 at 100 and 300 °C. Transmittance changes of about % 25.5 and % 15.0 are achieved for samples with thicknesses of 115 and 1440 nm, respectively, at a temperature of 100 °C. Also, a further gasochromic experiment was performed for 1000 ppm H2 in a N2 atmosphere to understand the intercalation mechanism of the H+ ions. Elimination of the H+ ions by dry air is also confirmed.