Revolutionizing n‐type Co3O4 Nanowire for Hydrogen Gas Sensing

This study presents conductometric sensors based on Co3O4 nanowires for hydrogen detection at ppb levels. The nanowires are synthesized through thermal oxidation of a 50 nm cobalt layer, exhibiting diameters between 6–50 nm and lengths of 1–5 μm, primarily growing along the (311) direction of spinal Co3O4. Raman investigation reveals five characteristic peaks at 195, 482, 521, 620, and 692 cm−1, corresponding to symmetric phonon modes of crystalline Co3O4. Electron paramagnetic resonance measurements confirm the presence of a ferromagnetic phase, attributed to incomplete cobalt oxidation, which disappears after 8 h of thermal aging at 400 °C. Conductometry measurements are performed in the temperature range of 300–500 °C. At temperatures above 300 °C, sensors exhibit abnormal n‐type semiconducting behavior due to lattice oxygen's involvement in the hydrogen sensing mechanism. Operating at 450 °C in dry air, the sensor shows a higher 232% response to 100 ppm H2 compared to ethanol, acetone, methane, carbon monoxide, and nitrogen dioxide. Remarkably, the sensor maintains a consistent conductance baseline even under high humidity (90%) for 25 d, with three‐cycle repeatability. This distinctive gas‐sensing capability is attributed to the catalytic activity and elevated operating temperature.

research focuses on achieving materials less sensitive to humidity variations.
Semiconducting n-type MOXs have always been considered the primary choice for the fabrication of chemoresistive sensors.Even though p-type MOX sensors have lower gas-sensing capabilities compared to n-type MOX sensors, they are still an interesting subject of study due to their lower dependency on humidity, superior catalytic properties, and high chemical stability. [12]In this context, Co 3 O 4 is considered one of the most stable and promising p-type materials for gas sensing. [13,14]he gas-sensing mechanism of the MOXs relies on surfaceassisted phenomena.Therefore, enhancing the surface area of Co 3 O 4 through decoration, [15,16] functionalization, [17,18] doping, [19,20] and control of growth to achieve specific morphologies was studied.Additionally, composites, [21] heterostructures, [22,23] and metal-oxide frameworks (MOFs) [24][25][26] were used to improve the gas-sensing properties of Co 3 O 4 .Also, many studies focused on improving the use of Co 3 O 4 for gas sensing in high-humidity environments. [19,27,28]espite the progresses made in the use of Co 3 O 4 , improving the gas-sensing properties and developing more efficient synthesis methods for the commercialization of Co 3 O 4 gas sensors still pose a challenge.In this context, we present the synthesis of Co 3 O 4 nanowires at low temperatures, allowing the growth of Co 3 O 4 nanowires on flexible plastics such as polycarbonate and nylon 6/6.The report also delves into the applicability of the fabricated Co 3 O 4 sensor for hydrogen sensing in high-humidity conditions, while reporting an abnormal, yet interesting, conductive behavior at high temperature (<300 °C).

Morphological, Structural, and Magnetic
Field-emission scanning electron microscope (FESEM) images of the prepared Co 3 O 4 are shown in Figure 1a,b.Formation of nanowires is clearly visible, with a higher density of nanowires on the material prepared from the 50 nm layer (Figure 1b).
Besides, long and nonhomogeneous distribution of nanowires is observed in the 100 nm-thick sample.The cross-sectional FESEM micrographical view of prepared material with an initial thickness of 200 nm is shown Figure 1c, which shows the existence of three layers.Furthermore, the nanowires are grown on the small grain of the topmost layer.Accordingly, the growth mechanism can be described as follows.

Growth Mechanism of Co 3 O 4 Nanowires
The Gibbs free energy of the reaction is crucial to determine the feasibility of reactions.Li et al reported that the reaction between Co and O 2 has a negative value, which suggests that this reaction is the primary driving force behind the formation of CoO and Co 3 O 4 . [29]Accordingly, a cobalt oxide film is formed leading to the formation of nanowires. [30,31]However, it would be necessary to describe the growth process in detail.[34] The schematic of mass transport phenomena is illustrated in Figure 2. The formation of the nanowires is prominent on the grain rather than the grain boundaries. [29,35]The lattice mismatch at the interface of the Co/CoO and the CoO/Co 3 O 4  generates significant stress at the interface.Subsequently, this stress acts as the main driving force for the diffusion of the Co ions from higher-to lower-concentration Co 3 O 4 nanowires roots, that is, Co-CoO and CoO-Co 3 O 4 as shown in Figure 2. [35,36] In this context, grain boundaries offer a significant role by forming a path for the diffusion of Co ions. [35]ransmission electron microscopy (TEM) image in Figure 3a confirms the creation of nanowires with diameters between 6 and 50 nm and lengths of 1-5 μm (Figure S2, Supporting Information).The electron diffraction pattern in Figure 3b

Raman-active vibrations (
), infrared-active vibrations (4F 1U ), and inactive modes (F 1g , 2A 2U , 2E U , 2F 2U ). [37]aman spectra of Co 3 O 4 reveal five characteristic peaks at 195, 482, 521, 620, and 692 cm À1 , which correspond to the F 2g1 , E g , F 2g2 , F 2g3 , and A g1 symmetric phonon modes of crystalline Co 3 O 4 . [38,39]The strongest peak at 195 cm À1 is attributed to the vibrations of Co 2þ at tetrahedral sites and the weakest peak at 692 cm À1 is attributed to vibrations of Co 3þ at octahedral sites (Figure S6, Supporting Information).Additionally, the Co 3þ species are located at octahedral sites (16a Wyckoff sites) and the Co 2þ species are located at tetrahedral sites (8a Wyckoff sites) in the spinel structure of Co 3 O 4 . [40]The bands at 195 and 692 cm À1 can be assigned to the Raman vibrations of Co 2þ -O 2À and Co 3þ -O 2À respectively, [41] which confirms the formation of Co 3 O 4 and suggests that tetrahedral sites are dominant compared to octahedral sites in the prepared structures. [29]igure 4a,b shows the X-and Q-band electron paramagnetic resonance (EPR) spectra of samples of Co 3 O 4 nanowires deposited on alumina (Al 2 O 3 ) substrates, measured before and after a thermal aging treatment at 400 °C in air for 8 h.Reference spectra of the alumina (Al 2 O 3 ) substrates are also displayed in both figures.The X-band spectrum of the Co 3 O 4 sample before thermal aging exhibits a very broad line at low magnetic fields, which further broadens as the temperature is decreased to 100 K (Figure 4c).After the thermal aging of the Co 3 O 4 sample, this line disappears and the X-band spectrum shows no difference from the spectrum of the alumina substrate (Figure 4a).The Q-band spectrum of the untreated Co 3 O 4 sample also consists of a broad line at low magnetic field (up to 550 mT), which disappears after thermal aging (Figure 4b).This broad line is associated with the presence of a minority ferromagnetic phase, consisting probably of Co and/or CoO clusters [42] due to the incomplete oxidation of the cobalt layer.The disappearance of the line after the 400 °C annealing could be due to the dissolution of the ferromagnetic clusters or oxidation processes that affected the ferromagnetic coupling.A comparison of the X-and Q-band spectra of the Co 3 O 4 sample with the reference spectra of the alumina substrate shows that all the narrower features superposed on the broad line are due to paramagnetic centers in the substrate.The absence of EPR signals from other paramagnetic centers in Co 3 O 4 [43] can be explained by the very small amount of Co 3 O 4 present in the measured sample.Given that the Co 3 O 4 volume in the EPR samples is of the order of 10 À3 mm 3 , such centers could be below the detection limit.

Gas-Sensing Properties
Figure 5a shows the temperature dependence of the conductance of Co 3 O 4 nanowires in both dry and humid air (40 RH%).Both sensors displayed low conductance, but it increases with increasing temperature when operating in dry air, which is the typical semiconducting behavior of metal oxides (MOXs).The conductance value is also found to be lower in humid air compared to dry air.This is because water molecules can be absorbed onto MOX surfaces through two processes, known as physisorption and chemisorption, at low and high operating temperatures respectively.In the case of physisorption, water molecules are absorbed on the MOX surface in the form of a molecule, which hinders the baseline conductance. [44,45]It has been observed that there is a significant decrease in conductance when sensors are operating at 500 °C in wet air, which will be further explained in later sections of the study.
The dynamic sensing performance of prepared Co 3 O 4 toward various hydrogen concentrations in the temperature range of 300-500 °C was evaluated (Figure 5b).Prior to taking readings, sensors were stabilized for 4 h in a testing chamber at each operating temperature with 200 sccm synthetic dry to obtain a stable baseline and remove interferents.The H 2 sensing measurements were evaluated at three different concentrations: 100, 250, and 500 ppm.Besides, each cycle was extended to 90 min; 30 min of exposure to gas; and followed by 60 min of exposure to synthetic air for the recovery.Also, note that 15 min are required to completely change the environment of the testing chamber (1L) with a flow rate of 200 sccm in our system, due to chamber size.
The sensors demonstrated an inversion of sensors signal when the temperature is above 300 °C (Figure 5c).Generally, the inversion of the electrical conductance type, p-n or n-p, occurred due to composition, [46] additives (decoration), and reducing species, [47] humidity, and the high operating temperature. [48]In a report, Vladimirova et al. showed the inversion of conductance of Co 3 O 4 from p-n when operating at a temperature higher than 250 °C in dry air due to the decomposition of the of Co T -O 2 2À -Co O superficial adducts. [49]Also, Lin et.al. reported the participation of lattice oxygen of Co 3 O 4 in the oxidation process which helps to invert the conductance change. [50]ccordingly, in this context, the inversion of the conductance change is thermally driven.Furthermore, an exceptional drop in the conductance value at 500 °C in the presence of humidity can be due to the desorption of absorbed oxygen from Co 3 O 4 and a percent of water molecules. [51]he sensor with higher density of nanowires (50 nm) showed a higher response compared to lower density of nanowires (100 nm), where the highest response is 2.32 toward 100 ppm at the operating temperature of 450 °C (Figure 5c). Figure 5d shows the dynamic gas-sensing nature of sensors at the operating temperature of 450 °C.Both the sensors are recovered to the baseline conductance at the 40 RH% air while Co 3 O 4 (50 nm) recover to the baseline conductance even in dry air.The estimated response and recovery times of the Co 3 O 4 (50 nm) sensor toward 100 ppm H 2 are ≤330 s and ≤850 s, respectively, whereas are ≤300 s and ≤900 s, respectively, for Co 3 O 4 (100 nm) (Table 1).
The response of the sensors is not only investigated by means of the operating temperature, but also by introducing different gases.Accordingly, the sensors with higher density of nanowires are tested for C 2 H 5 OH, C 3 H 6 O, NO 2 , CH 4 , and CO at the best operating temperature (450 °C) in dry air for the selectivity investigations.The sensor selectively detects H 2 among the tested gases (Figure 6a).Concerning the performance of the sensors, repeatability of the sensors is also a key fact.Furthermore, the estimated detection limit (LOD, simple 10%) of the Co 3 O 4 (50 nm) in dry air is %360 ppb.Sensors have higher sensitivity when operating in dry air (Figure 6b), which is due to the higher concentration of free electrons in this environment.This leads to more ionosorbed oxygen, which in turn enhances the sensor's ability to react with hydrogen and allows detection at lower concentrations. [52]Figure 6c shows that the sensor holds its initial H 2 -sensing characteristics for three consecutive cycles of 100 ppm H 2 at its best operating temperature.Also, the sensor showed good stability up to 25 d with a decrement in sensor response by 10% when operating at 450 °C (Figure 6d).

Effect of Humidity in Response
Humidity plays a crucial role in the functioning of metal oxide gas sensors, as many such sensors have been created specifically to detect humidity.The adsorption of water on the MOX surface can negatively impact the sensor's performance in various ways, such as decreasing the baseline resistance of the sensor, reducing the active surface area for chemisorption of oxygen, and acting as a barrier for adsorbing analyte gas molecules. [53,54]owever, there are also studies that show improved sensor performance in humid environments and neutral effects on sensor performance. [28,55,56]These findings illustrate the intricate relationship between target gases and MOXs when exposed to water vapor.In this study, we have expanded our examination to evaluate sensor performance at various humidity levels, ranging from dry air to 90% relative humidity (RH%).Figure 7a illustrates the sensor's dynamic response at different humidity levels (10, 20,  40, 60, 80, 90 RH%) after 25 d of initial testing.The conductance baseline is insignificantly affected by humidity, which is unusual for MOX sensors.In general, the baseline resistance is decreased by the hydroxyl ions (OH À ) and protons (H þ ) formed by the dissociation and adsorption of water molecules on the MOX surface.Typically, OH À attaches to the metal cation and H þ attaches to the oxygen in the MOX. [57]However, this mechanism can change to a hopping mechanism, where H 2 O is physiosorbed on the MOX surface by attaching to the H þ already generated due to chemisorption of H 2 O at low-humidity conditions. [57,58]herefore, the formation of H 3 O þ on the MOX hinders the baseline conductance of the MOX. Figure 7 illustrates the strong and stable baseline conductance in our sensors even at high humidity (90% RH%).Additionally, Figure 7b shows the sensor's response to 100 ppm H 2 at different humidity levels (0-90% RH%).Notably, no significant difference is found in the response at dry air (2.32) and 90% RH% (2.26) at the optimum operating temperature.The obtained excellent gas-sensing performance at the optimum working temperature can be ascribed to the higher operating temperature, the catalytic activity of the prepared material, as well as the surface modifications due to the formation of the hydroxyl groups on the surface resulting in active sites that enhance the H 2 sensing response. [59]

Sensing Mechanism
Co 3 O 4 is typically a p-type semiconductor in which holes are the majority carriers.When Co 3 O 4 interacts with oxygen, oxygen species O À 2 (T op < 100 °C), O À (100 °C < T op < 300 °C), and O 2À (T op > 300 °C) are absorbed on the surface of the Co 3 O 4 increasing the hole concentration in the valence band and forming a hole-accumulating region as shown in Figure S5a,b, Supporting Information. [12,60]This leads to an increase in conductance when exposed to air (Figure S5d,e, Supporting Information).Additionally, when exposed to reducing gas such as H 2 , it reacts with adsorbed oxygen species releasing electrons (Figure S5c, Supporting Information) as shown in Equation ( 1) and ( 2). [61,62]As a result, the width of the hole accumulation layer decreases, causing a corresponding decrease in conductance (Figure S5f, Supporting Information).
In this study, the understanding of the gas-sensing mechanism is different from the typical explanations, as is based on the inversion of semiconducting behavior from p-type to n-type, which is driven by thermal energy and is caused by the involvement of lattice oxygen in H 2 oxidation.This leads to the formation of oxygen vacancies (Equation (3)).
If in the surface layer of Co 3 O 4 the electron concentration [n] begins to exceed the hole concentration [p], then the electrons become the main charge carriers ([n] > [p]) and the type of conductivity changes from p-type to n-type resulting in the inversion of sensing behavior. [49]When H 2 interacts with the oxygen species, it increases the number of electrons and thus the conductance.It is also being studied how the transformation of CoO into Co 3 O 4 at the intermediate layer between Co and Co 3 O 4 may affect the electron concentration on the surface. [49]

Conclusion
This research focuses on the fabrication of Co 3 O 4 nanowires at low temperature and the assessment of their conductometric gas-sensing properties.The most favorable morphology of Co 3 O 4 nanowires is achieved through thermal oxidation using a 50 nm-thick layer of metallic cobalt at 300 °C, 3 h, 2.2 mbar, and 100 sccm Ar flow.This method results in a higher nanowire density compared to using a 100 nm-thick cobalt layer.TEM analysis reveals that Co 3 O 4 nanowires have diameters ranging from 6 to 50 nm and lengths of 1-5 μm.
Performed gas-sensing tests initially indicate that a higher nanowire density leads to a more pronounced hydrogen-sensing response.Notably, the sensors demonstrate an unusual reversal of the sensor signal from p-type to n-type behavior when operated at temperatures exceeding 300 °C.This behavior is attributed to the participation of lattice oxygen in the gas-sensing mechanism, which is activated by the elevated operating temperature.Remarkably, the sensors maintain a stable baseline conductance even under severe RH conditions (90%) over a testing period of 25 d.This stability is attributed to the catalytic activity of Co 3 O 4 and the high operating temperature.In comparison to acetone, ethanol, carbon monoxide, methane, and nitrogen dioxide, the sensors exhibit exceptional selectivity toward hydrogen.The reported response is remarkable, reaching the value of 2.32 in the presence of 100 ppm of hydrogen, with an estimated detection limit of %360 ppb.

Experimental Section
The experimental process began with the cleaning of alumina substrates (Al 2 O 3 , 99.9% purity, 2 Â 2 mm, Kyocera, Kyoto, Japan) by sonication in an acetone bath.Subsequently, thin layers of metallic cobalt (Co) were deposited on the cleaned Al 2 O 3 with a thickness of 50 nm and 100 nm using DC magnetron sputtering techniques at room temperature with an argon (Ar) plasma (50 W, Ar, 7 standard cubic centimeters (sccm)).Then the metallic Co films were thermally oxidized at a temperature of 300 °C under a pressure of 2.2 mbar with an Ar flow of 100 sccm for 3 h.Later, interdigitated Pt electrodes and Pt heater were DC magnetron sputtered on prepared Co 3 O 4 nanowires to investigate the conductometry performances of synthetized material. [63]Subsequently, devices were soldered to the TO-39 package to fabricate sensor chips for testing the capability to detect different gases (Figure S1, Supporting Information).
The conductometric measurements were performed in a customized climatic gas chamber, [59,64] with a constant flow of 200 sccm in the presence of hydrogen (H 2 ), ethanol (C 2 H 5 OH), acetone (C 3 H 6 O), nitrogen dioxide (NO 2 ), methane (CH 4 ), and carbon monoxide (CO) with a fixed voltage of 1 V applied to the sensing elements.The sensor's response (for n-type behavior) was calculated by comparing its conductance in synthetic air to its conductance in the presence of the target gas, defined as S (ΔG/G) = (G g ÀG a )/G a or (G a ÀG g )/G g for reducing and oxidizing gases, where G a is the conductance of the sensor in the synthetic air while G g is the conductance of the sensor in the presence of analyte gas. [65]he morphological and structural characteristics of the prepared nanostructures were examined using FESEM (model TESCAN MIRA 3), an analytical TEM (model JEOL JEM ARM 200F) equipped with a JEOL JED-2300T unit for EDS spectra, and Raman spectrometer (model HORIBA XploRA Nano).For TEM observation, the preparation consisted of grazing the nanowires from the substrate with a diamond pen and placing a droplet of ethanol at the area of the grazing.A copper grid with lacey carbon film was put on top of the drop and then it was left to dry at room temperature.Additionally, EPR measurements were conducted using Bruker ELEXSYS E580 and E500Q spectrometers operating in the X-band (9.8 GHz) and Q-band (34 GHz), respectively, equipped with CF935 continuous flow cryostats from Oxford Instruments.The samples used for the EPR investigations consisted of Co 3 O 4 nanowires grown on alumina substrates cut into 1.4 Â 1.4 mm 2 squares.For the X-band measurements, three such samples were inserted in a 3 mm (diameter) quartz tube, while for the Q-band measurements only one sample was inserted in a 2 mm (diameter) quartz tube.The samples were submitted to thermal aging at 400 °C in air for 8 h in a temperature stabilized (AE1°) furnace.

Figure 1 .
Figure 1.FESEM images of prepared Co 3 O 4 nanowires: a) 100 nm b) 50 nm, c) cross sectional view of Co 3 O 4 nanowires with initial Co thickness of 200 nm.

Figure 2 .
Figure 2. Schematic illustration of the mass flow phenomena of the Co 3 O 4 nanowires formation.
indicates the presence of four main reflection planes of Co 3 O 4 , (111), (220), (311) and (400) (CIF no.9005896).High-resolution transmission electron microscopy (HRTEM) images in Figure 3c,d indicate that the (311) plane is the most prominent, suggesting that the material is primarily grown in the (311) direction of the spinal Co 3 O 4 .TEM characterization of various sections of the Co 3 O 4 nanowires is presented in Figure S2a, Supporting Information and the corresponding electron diffraction patterns confirm that the Co 3 O 4 nanowires are grown along the (311) direction throughout the entire wire (Figure S3b-d, Supporting Information).Energy-dspersive X-ray (EDS) analysis of the grown Co 3 O 4 nanowires on TEM grid (Figure S4 and XRD in S5, Supporting Information) reveals that the nanowires are composed of only Co and O. Co 3 O 4 has a cubic spinel structure (space group (Fd3m)) in which Co 3þ and Co 2þ are located at octahedral and tetrahedral sites, respectively.The vibrational mode of Co 3 O 4 includes

Figure 3 .
Figure 3. a) Conventional TEM image of a grown Co 3 O 4 nanowire, b) electron diffraction pattern, and c,d) HRTEM image of the TEM image shown in (a).

Figure 4 .
Figure 4. EPR spectra of the Co 3 O 4 nanowires grown on alumina substrates, before and after annealing at 400 °C for 8 h, measured in the a) X-and b) Q-band at room temperature.The spectra of the alumina substrates are also given for reference.c) X-band EPR spectra of the untreated Co 3 O 4 sample measured at room temperature and at 100 K.The narrow lines marked with * are from paramagnetic centers in the alumina substrate.

Figure 5 .
Figure 5.The gas-sensing functionality of the sensors: a) variation of the conductance of the sensors, b) dynamic response of the sensors at the tested temperature range (300-500 °C), (c) response of the sensors for 100 ppm H 2 at dry air, and c) dynamic response of the sensors at the optimum operating temperature (450 °C) in dry air and 40 RH%.

Figure 6 .
Figure 6.Gas-sensing features of the best sensors (50 nm) at optimum operating temperature (450 °C): a) selectivity against the ethanol, acetone, carbon monoxide, methane, and nitrogen dioxide in dry air, b) calibration curve at dry (red) and 40 RH% (cyan), c) repeatability toward H 2 , and d) stability of the response (the error bars represent the average experimental error of four sensors).

Table 1 .
Various hydrogen sensors are fabricated with different MOX nanostructures.