Synthesis of Nanoporous TiO 2 with the Use of Diluted Hydrogen Peroxide Solution and Its Application in Gas Sensing

: The chemical routes of metal oxidation in presence of hydrogen peroxide solutions are tailor-made for the synthesis of biocompatible metal oxide surfaces with clean intermediate and end products, such as oxides, hydroxides, hydrogen and water. The hydrolysis of titanium in hydrogen peroxide solutions is particularly interesting for medical applications, forming micro- and nanoscale titania surfaces. In this paper, the content of the hydrolysis solution is revised, allowing the fabrication of gas sensor devices based on nanoporous titania. Nanopore and microcrack formations were discussed in detail by monitoring the structural changes on the thin ﬁlm surface with ﬁeld-emission scanning electron microscopy (FE-SEM). A stable rutile crystalline phase was detected by glancing incidence X-ray di ﬀ raction (GI-XRD) measurement after repetitive hydrothermal processes. Electrical conductance measurements were carried out at high temperatures (400–600 ◦ C) under humid airﬂow (40% RH@20 ◦ C) with the injection of various concentrations of a wide set of test compounds (C 2 H 3 N, CO, H 2 , NO 2 , C 2 H 6 O), to observe the sensing capabilities of the material. Furthermore, the humidity e ﬀ ects on the sensing properties toward H 2 , CO, and C 2 H 6 O have been discussed.


Introduction
Metal oxides (MOXs) have played a big role in the development of technologies in various industrial applications. The huge progress in nanofabrication methods has boosted their use by introducing various cost-effective and reliable synthesis routes, especially with bottom-up approaches [1]. Nanofabrication has not only exploited their surface properties, but also allowed them to be integrated in small and portable devices. Among these MOX nanostructures, intrinsically multi-functional ones have attracted much more attention than others. For instance, titanium dioxide (TiO 2 ) nanostructures are one of these extremely useful materials. Titanium dioxide nanostructures have found various applications, as photocatalysts [2], biosensors [3], optoelectronics [4] and many more [5][6][7][8]. In most cases, the synthesis method for these structures is the hydrothermal process, because it is a low-cost, low-temperature and environmentally friendly process [9]. In particular, the hydrolysis of Ti in alkaline solutions and the hydrations of Ti salts are common ways for the hydrothermal synthesis of TiO 2 nanostructures [10]. The availability of various alkaline solutions (NaOH, KOH, etc.) and Ti salts (TiCl 4 , etc.) offers the possibility of obtaining TiO 2 nanostructures with various morphologies, exhibiting different crystalline structures and surface properties, making them perfect candidates for gas sensor applications. On the other hand, the single-step hydrolysis of Ti in hydrogen peroxide (H 2 O 2 ) solution has been drawing attention, especially in biomedical applications, since having multi-step processes involving various chemicals always carries some contamination risks. Tengvall  extensively on the interaction between Ti and peroxide solutions for biomedical implants [11][12][13][14][15][16]. In other studies, nanostructured titania gels and their crystallization processes have been investigated according to the effects of the solvent [17,18], additional chemicals [19][20][21], and temperature [22]. Radtke et al. studied in detail the photocatalytic properties and the biocompatibility of titania and titania alloy surfaces synthesized by hydrogen peroxide solutions and its derivatives [23][24][25][26]. Wu and Qi reported the synthesis of titania nanorods with the help of precursors [27]. Recent studies have shown that it is possible to detect some organic molecules by using the photocatalytic properties of nanostructured titania, synthesized in hydrogen peroxide [28,29]. All these findings have revealed the possibility of fabricating biocompatible titania-based organic sensors in near future. In this study, conductometric gas sensor application has been presented as a pioneer example of the use of these biocompatible sensing surfaces. The amounts of hydrogen peroxide solution and the titanium film thickness have been optimized to fabricate nanoporous TiO 2 directly on the active area of the sensors. This optimization is essential for obtaining proper device integration, avoiding the transfer of the nanoporous layer, and thus increasing the mechanical stability of the devices.
The high electrical resistance of bare TiO 2 nanostructures has restricted their applications in the gas sensing field [30]. In some studies, the use of pure nitrogen or argon as carrier gases, instead of synthetic air, has been a practical solution to keep electrical resistance in a favorable range for the investigation of the sensing properties. On the contrary, high percentages of oxygen, which are normally present in air, increase the electrical resistance of bare titania nanostructures to very high values, which requires complex and accurate electronics [30][31][32][33][34]. These studies have definitely contributed in explaining the gas sensing properties of titania, and may also have found some interesting application areas, but it is essential to understand how these sensors work in real-life conditions for practical usage. In other studies, TiO 2 nano-heterostructured materials (by doping, surface modifications, etc.) have been attracting attention, since most of them have been successful in reducing their high intrinsic electrical resistance to feasible values for device integration [31,[35][36][37][38][39]. Despite these developments, the gas sensing properties of bare TiO 2 nanostructures have not been studied properly for practical use, and there are still some missing parts that prevent real-life applications. Herein, we report the intrinsic sensing properties of nanoporous titania, which have been investigated by conductometric measurements at high temperatures (400-600 • C) in dry and humid (40% RH@20 • C) air flows towards a wide set of target compounds (C 2 H 3 N, CO, H 2 , NO 2 , C 2 H 6 O).

Synthesis of Nanoporous TiO 2
Polycrystalline alumina substrates (2 mm × 2 mm, 99% purity, Kyocera, Japan) were used for the deposition of Ti films and the fabrication of chemical sensing devices. Substrates were cleaned in an ultrasonic bath using acetone, ethanol and deionized water, for 5 min each one. Ti films of two different thicknesses (100 nm and 1 µm) were deposited on the substrates by RF magnetron sputtering (Kenotec, Italy). The depositions were carried out at 300 • C and a pressure of 6.6 × 10 −3 mbar, with 75 W RF magnetron power (Ar plasma). The synthesis of the nanoporous surface was performed by soaking the samples into a 15 mL H 2 O 2 solution, systematically diluted from 15% to 2% at 80 • C for 20 h in a home-made hydrothermal reactor. Oxidation processes on each sample were repeated three times to get totally bleached surfaces. The samples were finally annealed in a furnace, working at atmospheric pressure, at 450 • C for 4 h, to enhance their crystallization and stability. The experimental process is summarized in Table 1.

Sample Characterization
The surface morphologies were examined using a FE-SEM (Leo 1525 Gemini model; Carl Zeiss AG, Oberkochen, Germany) connected to an energy dispersive X-ray spectrometer (EDX, Oxford, UK), operated in the potential range of 0-10 kV. The crystalline phase was identified by XRD measurements (Empyrean diffractometer, PANalytical, Netherlands) carried out using Cu-LFF tube (λ = 1.5406 Å) operated at 40 kV/40 mA. The incident angle in glancing angle mode was 1.5 • and the spectra were recorded by using a proportional Xe detector in the range of 20 • -50 • .

Preparation and Testing of Gas Sensors
To perform the gas functional measurements, the samples were equipped with electrical contacts on the top of the titania layer, while heaters were deposited on the backside of the alumina substrates by magnetron sputtering technique. Firstly, TiW and Pt contact pads were deposited to increase the adhesion, at 75 W DC power (Ar plasma), 6.6 × 10 −3 mbar working pressure at 300 • C for 3 min, respectively. Afterwards, Pt interdigitated transducers (IDT) were deposited under the same temperature, pressure and DC power conditions for 20 min (≈1 µm thickness). The same procedure was used for the deposition of the heating element on the backside of the samples. The functional testing system is characterized by a climatic chamber (Angelantoni, Italy, model MTC 120), whose role is to keep the stainless-steel test chamber at a constant ambient temperature (20 • C). To control the working temperatures of the sensors, the heating element of each device is powered by Thurlbly Thandar PL330DP power supplies. A Dreschel bottle, placed in a thermostatic bath set at 25 • C, generates a saturated humid airflow, which is then mixed with synthetic dry air to achieve the desired level of relative humidity (RH). The measurements were carried out at a humidity level of 40% with a flow of 200 sccm. The devices were exposed to several injections of different concentrations of hydrogen (H 2 ), ethanol (C 2 H 6 O), carbon monoxide (CO), nitrogen dioxide (NO 2 ) and acetonitrile (C 2 H 3 N), at working temperatures between 400 to 600 • C. Each gas was kept flowing inside the chamber for 30 min and then, to recover the baseline, the airflow was restored for 60 min. A schematic configuration of experimental process and measurement set up is given in Figure 1.

Sample Characterization
The surface morphologies were examined using a FE-SEM (Leo 1525 Gemini model; Carl Zeiss AG, Oberkochen, Germany) connected to an energy dispersive X-ray spectrometer (EDX, Oxford, UK), operated in the potential range of 0-10 kV. The crystalline phase was identified by XRD measurements (Empyrean diffractometer, PANalytical, Netherlands) carried out using Cu-LFF tube (λ = 1.5406 Å) operated at 40 kV/40 mA. The incident angle in glancing angle mode was 1.5° and the spectra were recorded by using a proportional Xe detector in the range of 20°-50°.

Preparation and Testing of Gas Sensors
To perform the gas functional measurements, the samples were equipped with electrical contacts on the top of the titania layer, while heaters were deposited on the backside of the alumina substrates by magnetron sputtering technique. Firstly, TiW and Pt contact pads were deposited to increase the adhesion, at 75 W DC power (Ar plasma), 6.6 × 10 −3 mbar working pressure at 300 °C for 3 min, respectively. Afterwards, Pt interdigitated transducers (IDT) were deposited under the same temperature, pressure and DC power conditions for 20 min (≈1 µm thickness). The same procedure was used for the deposition of the heating element on the backside of the samples. The functional testing system is characterized by a climatic chamber (Angelantoni, Italy, model MTC 120), whose role is to keep the stainless-steel test chamber at a constant ambient temperature (20 °C). To control the working temperatures of the sensors, the heating element of each device is powered by Thurlbly Thandar PL330DP power supplies. A Dreschel bottle, placed in a thermostatic bath set at 25 °C, generates a saturated humid airflow, which is then mixed with synthetic dry air to achieve the desired level of relative humidity (RH). The measurements were carried out at a humidity level of 40% with a flow of 200 sccm. The devices were exposed to several injections of different concentrations of hydrogen (H2), ethanol (C2H6O), carbon monoxide (CO), nitrogen dioxide (NO2) and acetonitrile (C2H3N), at working temperatures between 400 to 600 °C. Each gas was kept flowing inside the chamber for 30 min and then, to recover the baseline, the airflow was restored for 60 min. A schematic configuration of experimental process and measurement set up is given in Figure 1.  A voltage of 1 V was applied to the active layer using an Agilent E3631A power supply. The electrical conductance value of each device, crucial to determine the sensors response, was recorded by dedicated picoammeters (Keithley 486). The following equations (Equations (1) and (2)) Coatings 2019, 9, 681 4 of 12 highlight the relationship between the response of the sensors and the variations of the electrical conductance caused by the presence of oxidizing or reducing gas, respectively (for n-type): G air and G gas represent the conductance values of devices under synthetic airflow and in the presence of injected gas, respectively. Sensor response time (t res ) is defined as the time interval needed to reach a conductance value of 90% of [G gas − G air ] after the gas injection. Similarly, recovery time (t rec ) is the time interval needed to reach a conductance value reach of 90% of [G gas − G air ] after restoring the airflow. The calculations for R, t res, and t rec were performed by an analysis program written according to the above equations and methods.

Characterization Results
Hydrogen peroxide (H 2 O 2 ) solutions were diluted more than in previous studies in order to protect the structural integrity of the Ti thin films [40]. A dense and homogeneous formation of nanoporous TiO 2 was observed on the surface of the samples prepared using a 1 µm Ti thin film in 15% H 2 O 2 solution, as reported in Figure 2a. Similar results were obtained using a 100 nm Ti thin film in 2% H 2 O 2 solution, as reported in Figure 2b. Micro-cracks on the surface were observed in thicker thin films with higher peroxide concentrations. The use of thinner films and a lower concentration of hydrogen peroxide solution seemed to be effective on reducing the size of the cracks. Moreover, more dollops were attained on the surface with higher concentrations and thicker Ti films, as shown in Figure 2a. Therefore, nanoporous TiO 2 layers synthesized by the oxidation of 100 nm Ti thin films in 2% H 2 O 2 solution were selected as the optimal morphology for gas sensor devices. The hydrothermal process was repeated three times (20 h each) until the nanoporous TiO 2 films were totally bleached. Sun and Wang described the possible chemical reactions (Equations (3)-(8)) that could occur during H 2 O 2 oxidation [40]. Those reactions are reported below: At low temperature, Ti reacts with hydrogen peroxide, forming titania phases (Equation (3)). At higher temperatures, some TiO 2 is hydrolyzed due to the effects of hot water and H 2 O 2 (Equations (4)-(5)). Hydration reaction with TiO 2 takes place simultaneously, forming Ti-OH functional groups (Equation (6)). Hydrated and hydrolyzed TiO 2 cause the formation of negatively charged surfaces (Equations (7)-(8)), which are involved in the formation of nanoporous structures.
In addition to morphological analysis, EDX measurements were carried out on the samples to determine elemental composition and to detect possible contaminants. Figure 3 displays the quantitative elemental analysis results and the dispersion field of prepared samples using a 2% H 2 O 2 solution. The inset table (Figure 3) gives the atomic percentages of existing elements. The rest of atomic content (5.92%) belongs to carbon, which originated from natural carbon contamination and from carbon tape used to mount the substrate on the holder. Carbon was not shown in EDX analysis in order to prevent misunderstandings. Oxygen, aluminum, and titanium peaks are consistent with titania (TiO 2 ) and alumina (Al 2 O 3 ) presence. The high amount of O (83.16%) could be linked to hydroxide (OH) species on the surface (Equations (7) and (8)). In addition to morphological analysis, EDX measurements were carried out on the samples to determine elemental composition and to detect possible contaminants. Figure 3 displays the quantitative elemental analysis results and the dispersion field of prepared samples using a 2% H2O2 solution. The inset table (Figure 3) gives the atomic percentages of existing elements. The rest of atomic content (5.92%) belongs to carbon, which originated from natural carbon contamination and from carbon tape used to mount the substrate on the holder. Carbon was not shown in EDX analysis in order to prevent misunderstandings. Oxygen, aluminum, and titanium peaks are consistent with titania (TiO2) and alumina (Al2O3) presence. The high amount of O (83.16%) could be linked to hydroxide (OH) species on the surface (Equations (7) and (8)). The crystalline structure of the synthesized nanoporous TiO2 layer was determined by GI-XRD measurement. There are two overlapping measurement profiles in Figure 4, which belong to a wet oxidation sample before and after a thermal annealing in the furnace. In addition to alumina substrate peaks (+), the main characteristic peaks of TiO2 rutile phase (R) belong to (110) plane, visible at 27.3°, as well as other rutile crystalline planes (101), (111) visible at 36° and 41.3° (JCPDS 21-1276). No peaks belonging to other phases were detected. Overlapping spectral peaks prove the stability of rutile-TiO2 after the thermal process. In the literature, single rutile [27], single anatase [40], or dual rutile and anatase [41] phases were observed in samples fabricated by using similar solutions. Moreover, Wu observed single anatase or dual-phase in titania films when using a 30% H2O2 solution made by the oxidation of Ti powder and plates [21]. It was proposed that low Ti(IV) In addition to morphological analysis, EDX measurements were carried out on the samples to determine elemental composition and to detect possible contaminants. Figure 3 displays the quantitative elemental analysis results and the dispersion field of prepared samples using a 2% H2O2 solution. The inset table (Figure 3) gives the atomic percentages of existing elements. The rest of atomic content (5.92%) belongs to carbon, which originated from natural carbon contamination and from carbon tape used to mount the substrate on the holder. Carbon was not shown in EDX analysis in order to prevent misunderstandings. Oxygen, aluminum, and titanium peaks are consistent with titania (TiO2) and alumina (Al2O3) presence. The high amount of O (83.16%) could be linked to hydroxide (OH) species on the surface (Equations (7) and (8)). The crystalline structure of the synthesized nanoporous TiO2 layer was determined by GI-XRD measurement. There are two overlapping measurement profiles in Figure 4, which belong to a wet oxidation sample before and after a thermal annealing in the furnace. In addition to alumina substrate peaks (+), the main characteristic peaks of TiO2 rutile phase (R) belong to (110) plane, visible at 27.3°, as well as other rutile crystalline planes (101), (111) visible at 36° and 41.3° (JCPDS . No peaks belonging to other phases were detected. Overlapping spectral peaks prove the stability of rutile-TiO2 after the thermal process. In the literature, single rutile [27], single anatase [40], or dual rutile and anatase [41] phases were observed in samples fabricated by using similar solutions. Moreover, Wu observed single anatase or dual-phase in titania films when using a 30% H2O2 solution made by the oxidation of Ti powder and plates [21]. It was proposed that low Ti(IV) ion concentrations in the solution and an acidic environment (low pH value) could boost rutile phase formation [27,41,42]. In the present case, hydrogen peroxide has a weak acidity characteristic, The crystalline structure of the synthesized nanoporous TiO 2 layer was determined by GI-XRD measurement. There are two overlapping measurement profiles in Figure 4, which belong to a wet oxidation sample before and after a thermal annealing in the furnace. In addition to alumina substrate peaks (+), the main characteristic peaks of TiO 2 rutile phase (R) belong to (110)  No peaks belonging to other phases were detected. Overlapping spectral peaks prove the stability of rutile-TiO 2 after the thermal process. In the literature, single rutile [27], single anatase [40], or dual rutile and anatase [41] phases were observed in samples fabricated by using similar solutions. Moreover, Wu observed single anatase or dual-phase in titania films when using a 30% H 2 O 2 solution made by the oxidation of Ti powder and plates [21]. It was proposed that low Ti(IV) ion concentrations in the solution and an acidic environment (low pH value) could boost rutile phase formation [27,41,42]. In the present case, hydrogen peroxide has a weak acidity characteristic, so a natural growth of rutile phase could be derived with low degree of supersaturation because of small amount of H 2 O 2 (2%) concentration and Ti thin film layer (100 nm) [41]. so a natural growth of rutile phase could be derived with low degree of supersaturation because of small amount of H2O2 (2%) concentration and Ti thin film layer (100 nm) [41].

Gas Sensing
The isothermal dynamic responses of the nanoporous TiO2 layer towards hydrogen gas injections demonstrate the n-type semiconducting behavior of the material (Figure 5a). A visualization of the response and recovery times is presented in Figure 5b with the zoomed conduction curve of titania at 600 °C towards 500 ppm H2. The electrical conductance increases upon interaction with reducing hydrogen gas. Due to the increased temperature, the oxygen molecules are dissociated and adsorbed on the titania surface in different forms (Equations (9) and (10)), depending on the sample temperature. For example, in the temperature range of 400-600 °C there are two different possible adsorption phases of oxygen ions [43]: The trapped electrons are released with the interaction of the reducing gas with chemisorbed oxygen ions. The possible reactions (Equations (11)-(15)) for reducing H2 [44], C2H6O [45,46], C2H3N [47,48], CO [49] gases and oxidizing NO2 [31] are given by the following equations:

Gas Sensing
The isothermal dynamic responses of the nanoporous TiO 2 layer towards hydrogen gas injections demonstrate the n-type semiconducting behavior of the material (Figure 5a). A visualization of the response and recovery times is presented in Figure 5b with the zoomed conduction curve of titania at 600 • C towards 500 ppm H 2 . The electrical conductance increases upon interaction with reducing hydrogen gas. Due to the increased temperature, the oxygen molecules are dissociated and adsorbed on the titania surface in different forms (Equations (9) and (10)), depending on the sample temperature. For example, in the temperature range of 400-600 • C there are two different possible adsorption phases of oxygen ions [43]: Nanoporous TiO2 gas sensors were tested towards C2H3N, CO, H2, NO2, and C2H6O in humid air (40% RH) flows. A comparison of the sensor responses towards fixed concentrations of target gases with working temperatures in the range 400-600 °C is given in Figure 6a. Sensors do not exhibit any appreciable response (≤0.1) in the presence of 10 ppm NO2. It is worth reporting that there are no significant conductance changes in dynamic gas measurements towards 2 and 5 ppm NO2 injections. Rutile phase may be responsible for NO2 deafness since the crystalline phase of TiO2 can be decisive for the detection of oxidizing gases as previously reported in literature [31,50]. Also, high working temperatures (>250 °C) are not favorable for NO2 detection. The reducing gases (C2H3N, CO, H2, and C2H6O) show similar trends: a small decrease in the response values at 450 °C followed by an increasing response up to 600 °C. On the other hand, there is a drastic decrease in The trapped electrons are released with the interaction of the reducing gas with chemisorbed oxygen ions. The possible reactions (Equations (11)-(15)) for reducing H 2 [44], C 2 H 6 O [45,46], C 2 H 3 N [47,48], CO [49] gases and oxidizing NO 2 [31] are given by the following equations: Nanoporous TiO 2 gas sensors were tested towards C 2 H 3 N, CO, H 2 , NO 2, and C 2 H 6 O in humid air (40% RH) flows. A comparison of the sensor responses towards fixed concentrations of target gases with working temperatures in the range 400-600 • C is given in Figure 6a. Sensors do not exhibit any appreciable response (≤0.1) in the presence of 10 ppm NO 2 . It is worth reporting that there are no significant conductance changes in dynamic gas measurements towards 2 and 5 ppm NO 2 injections. Rutile phase may be responsible for NO 2 deafness since the crystalline phase of TiO 2 can be decisive for the detection of oxidizing gases as previously reported in literature [31,50]. Also, high working temperatures (>250 • C) are not favorable for NO 2 detection. The reducing gases (C 2 H 3 N, CO, H 2 , and C 2 H 6 O) show similar trends: a small decrease in the response values at 450 • C followed by an increasing response up to 600 • C. On the other hand, there is a drastic decrease in NO 2 response value at 450 • C. It is plausible that not only temperature, but also humidity should be taken into consideration to explain these changes. However, due to the complexity of the involved surface chemical reactions, more in-depth investigations are required to fully understand the roles of humidity and temperature and how they affect the sensing properties. In operando measurements, for example, could provide useful information to better understand the surface chemistry.   There is a decrease of almost one order of magnitude in the response values for C 2 H 6 O and H 2 when introducing a humid airflow, while there is a small increase in response values for CO. These results again show that the interaction depends strongly on the chemical species involved. On a different semiconducting oxide (SnO 2 ), Barsan and Ionescu proposed that water molecules dissociate in H + and OH − , which are absorbed on SnO 2 surface as hydroxyl group [51]. The participation of water vapor increases the interaction between atmospheric oxygen and n-type semiconductor surface, acting like a catalyst for this reducing gas [51]. This may also hold for TiO 2 , and may explain the increase in the response value in presence of CO, since nanoporous TiO 2 shows a similar n-type behavior. In addition, the shorter response and recovery times in humid air for all reducing gases prove the existence of a catalytic effect for nanoporous TiO 2 . On the other hand, the effect of water vapor seems to work in an opposite way for H 2 and C 2 H 6 O gases. The humidity reduced the sensor responses of H 2 and C 2 H 6 O. In contradiction with our observations, Moon et al. presented that H 2 sensing of mesoporous titania was not effected by humidity variations (5%, 15%, 42%RH) at 140 • C [52]. Similarly, hydrothermally grown TiO 2 nanorods did not change the response values towards 1-100 ppm VOCs (ethanol, methanol, and propanol) at 75 • C under humidity variations (8%, 52%, 75%RH) [53]. The most probable reason for both of these cases is that the working temperatures were too low to initialize a strong chemisorption of water molecules on the surface of TiO 2 . The physisorption of water molecules has no effect on conduction [51]. In our case, the temperature was high enough for the chemisorption of water vapor on nanoporous titania surface. In addition to this, the hydrogen sensing mechanisms of titania surfaces were attributed mostly to spill-over effect [54,55]. Hydroxyl groups on the surface may behave as an anti-catalyst for dissociation and diffusion process of H 2 molecules, which is different from the adsorption of CO molecules. Ethanol sensing seems more complex because of its intermediate products during decomposition [56,57]. The authors indicate that similar observations for SnO 2 nanobelts were discussed and reported in a previous work [58]. The catalytic and anti-catalytic properties of hydroxyl groups might be effective for these dehydration (Equation (16)) and dehydrogenation (Equation (17) Table 2 summarizes the sensing performances of this work compared to some recent studies on bare-TiO 2 nanostructures, with numerical parameters. Some studies defined the sensor response value as the ratio of the electrical resistance value in air with the resistance value in the test gas (R air /R gas ). These are indicated in parentheses next to their response values in the table. Our rutile nanoporous titania sensor had better response values towards H 2 and C 2 H 6 O than similar nanoporous TiO 2 sensors [38,52,59], as well as most other nanostructures [53,[60][61][62]. Sensor response towards 25 ppm of C 2 H 3 N was 0.53 at 600 • C, which is an appropriate response value (≥0.1) to be used as an early detection system for acetonitrile exposure. Acetonitrile has toxic effects on the human body, and its odor threshold value is 42 ppm [63]. On the other hand, t res and t rec were longer than those reported in existing studies regarding H 2 and C 2 H 6 O gases. Higher working temperature and lower gas concentration values could be the reason for these long times. In general, high working temperatures are not preferable for smart gas sensor applications due to energy considerations. Nevertheless, high working temperatures are a natural requirement for bare semiconducting MOXs. Thanks to our synthesis method, which allows the easy fabrication of TiO 2 nanoporous sensing layers on 2 mm × 2 mm surface area, our sensors need only 0.5 W electrical power to work at 400 • C. Unfortunately, there are no available data to compare energy efficiency, although other similar nanoporous sensors have lower working temperature [52].

Conclusions
The hydrolysis process of titanium thin film in peroxide solution was optimized for fabrication of nanoporous TiO 2 gas sensors. The 2% hydrogen peroxide solution was determined to be an optimal solution for nanoporous layer formation on 100 nm titanium thin film. The formation kinetics of nanoporosity were discussed and demonstrated with chemical reactions. The lower hydrogen peroxide concentration offers better mechanical strength for the sensor surface. The reduction in surface cracks and the decrease in the formation of dollops through the titania surfaces were monitored using FE-SEM images. In addition, one of the main motivations for this study was to use a clean synthesis method to protect the biocompatibility of TiO 2 surface. The EDX profile of the nanoporous surface proved the purity of the synthesis process. The rutile crystal structure and its thermal stability were screened by GI-XRD measurements of TiO 2 nanoporous samples before and after thermal annealing. After device configuration, high-temperature sensor tests were conducted in humid airflow with respect to different concentrations of H 2 , C 2 H 6 O, CO, NO 2, and C 2 H 3 N. Our sensor reported better sensor response towards H 2 and C 2 H 6 O than similar nanoporous surfaces. The kinetics of chemical sensing was illustrated with reaction formulations for each gas. The catalytic and anti-catalytic effects of humidity were discussed in detail for reducing gases (H 2 , C 2 H 6 O, CO) with their response value comparisons in dry and humid air flows.