Improved Hydrogen-Sensing of TiO2 Schottky Device Through Schottky Barrier Height Modulation

TL;DR

This study improves hydrogen sensing in TiO2 sensors by adjusting the Schottky barrier height through controlled annealing.

Contribution

A novel method of modulating the Schottky barrier height in TiO2 sensors via annealing temperature is introduced.

Findings

Annealing at 500 °C produced the highest gas-sensing response of 242 to 1 ppm H2.

The sensor's performance is linked to the Schottky barrier height and oxygen vacancy concentration.

Three distinct sensing regimes were identified based on the gas-sensing mechanism analysis.

Abstract

Adjusting the Schottky barrier height is an important approach to enhancing the gas-sensing performance of TiO2 Schottky sensors. In this study, micro TiO2 nanotube Schottky sensors were fabricated via magnetron sputtering and anodic oxidation, with their Schottky barrier height adjusted by varying the annealing temperature. The morphology, phase composition, oxygen vacancy concentration, band structure, and Schottky junction of the samples were investigated using SEM, GIXRD, EPR, Hall effect measurements, XPS, I-V curves, and AC impedance. The sensor annealed at 500 °C demonstrated the highest gas-sensing response, outperforming sensors treated at other temperatures by over 100 times. Its response value to 1 ppm H2 was 242. The annealing temperature significantly affects the TiO2 phase and oxygen vacancy concentration, resulting in the highest Schottky barrier height in the 500 °C-annealed sensor, which contributes to its superior sensing performance. AC impedance measurements revealed no significant Fermi-level pinning in TiO2. Based on the gas-sensing mechanism analysis, the response of the TiO2 sensor can be divided into three regimes: Schottky junction control, TiO2 resistance control, and co-control.

Figures

Full text

1. Introduction

Hydrogen energy is increasingly acknowledged as a sustainable and clean energy source. However, it becomes flammable and potentially explosive when its concentration exceeds 4%, creating significant risks during storage, transportation, and usage [1]. Therefore, the development of highly efficient trace-level hydrogen detection is essential. In Schottky-type hydrogen sensors, the Schottky barrier formed between the metal electrode and the semiconductor material has an exponential relationship with current. Hydrogen can modify the height of this barrier, resulting in substantial changes in the sensor’s current, which facilitates the effective detection of low-concentration hydrogen [2,3]. Titanium dioxide (TiO_2_), as a metal oxide semiconductor, can form Schottky-type hydrogen sensors with metal electrodes such as Pt, Pd, and Au [4,5]. Additionally, TiO_2_ possesses a suitable bandgap, is non-toxic, and demonstrates high chemical stability, making TiO_2_-based Schottky sensors highly promising for practical applications [6].

The TiO_2_ nanotubes produced through the anodic oxidation of Ti sheets display a highly ordered structure and a large specific surface area. When using Pt as electrodes, the response to 1000 ppm hydrogen can reach up to 10^8.7^, significantly surpassing that of TiO_2_ sensors with other morphologies [7,8]. We developed miniaturized TiO_2_ nanotube-based sensors on SiO_2_/Si substrates, achieving a response of 5.3 × 10^6^ to 200 ppm hydrogen at room temperature by optimizing the device architecture and adjusting the oxygen vacancy concentration in the TiO_2_ nanotubes [9,10]. However, these anodically oxidized TiO_2_ nanotubes are typically amorphous and have low structural strength, necessitating high-temperature treatment. At around 300 °C, TiO_2_ undergoes a phase transformation from amorphous to anatase, and when the temperature exceeds 550 °C, the anatase phase begins to convert to rutile. To preserve the structural integrity of the nanotubes, the annealing temperature is generally kept below 700 °C [11,12,13,14,15]. Research indicates that anatase-phase TiO_2_ typically exhibits a higher gas-sensing response. To maximize this response while maintaining the anatase phase, a sintering temperature of 500 °C is commonly selected [7,15,16,17,18]. The reasons behind this phenomenon warrant further exploration.

Annealing temperature influences not only the phase of TiO_2_ but may also increase the concentration of oxygen vacancies [19,20]. A higher concentration of oxygen vacancies promotes the formation of adsorbed oxygen on the TiO_2_ surface, which positively impacts the gas-sensing performance of the sensor [21]. Nonetheless, some studies suggest that high-temperature sintering can alter the interface characteristics of Schottky junctions. For example, in Ni/ZnO/Ag Schottky junctions, increasing the annealing temperature from 400 °C to 700 °C can lead to the disappearance of the unidirectional conductivity due to an elevated defect concentration [22,23]. Furthermore, excessively high defect concentrations may result in a high density of surface states in TiO_2_, causing Fermi-level pinning and diminishing the gas-sensing performance of TiO_2_ Schottky sensors [24,25]. It can be seen that adjusting the annealing temperature to alter the TiO_2_ phase and oxygen vacancy concentration can modulate the Schottky barrier height at the metal–TiO_2_ interface, thereby enhancing the gas-sensing performance of the sensor.

This study establishes the relationship among annealing temperature, TiO_2_ phase, oxygen vacancy concentration, and the TiO_2_ band structure. By controlling the annealing temperature to optimize the Pt-TiO_2_ Schottky barrier height, the gas sensing performance of the TiO_2_ sensor has been significantly enhanced. Furthermore, detailed analyses reveal that the TiO_2_ nanotube Schottky sensor exhibits no Fermi-level pinning and shows a remarkable response to low-concentration hydrogen.

2. Materials and Methods

2.1. Material Preparation and Sensor Fabrication

A 2.5 μm thick Ti film was deposited onto a SiO_2_/Si wafer using magnetron sputtering. The deposition occurred under a base pressure of 2.0 × 10^−4^ Pa, with argon as the sputtering gas at a pressure of 0.30 Pa. The substrate was maintained at a temperature of 200 °C, with a target-to-substrate distance of 100 mm and a sputtering power density of 3.18 W/cm^2^, resulting in the creation of Ti/SiO_2_/Si. The deposited Ti/SiO_2_/Si was then cut into samples measuring 10 mm × 30 mm.

The cut Ti/SiO_2_/Si samples underwent anodization at a constant DC voltage of 45 V for 15 min in an electrolyte comprising 0.2 M NH_4_F and 3 vol.% water in ethylene glycol. The ambient temperature is 20 °C with a relative humidity of 30%. After anodization, the samples were thoroughly cleaned with deionized water and anhydrous ethanol, then dried to obtain the TiO_2_/Ti/SiO_2_/Si samples.

Pt interdigitated electrodes (Pt IDEs) were then deposited on the surface of the TiO_2_/Ti/SiO_2_/Si samples via magnetron sputtering, under a base pressure of 2.0 × 10^−4^ Pa. During this process, argon was introduced at a sputtering pressure of 0.70 Pa, with a power density of 1.28 W/cm^2^. The thickness of the Pt interdigitated electrodes was 300 nm, resulting in Pt/TiO_2_/Ti/SiO_2_/Si samples.

The Pt/TiO_2_/Ti/SiO_2_/Si samples were subsequently heat-treated in air at temperatures of 300 °C, 400 °C, 500 °C, 600 °C, and 700 °C for 2 h each, with heating and cooling rates set at 2 °C/min. The resulting sensors were designated as T-300, T-400, T-500, T-600, and T-700, respectively.

2.2. Characterization

The morphology of the samples was characterized using scanning electron microscopy (SEM, JSM-7610F PLUS, JEOL Ltd., Tokyo, Japan) with an acceleration voltage of 5 kV. The phase composition of TiO_2_ following different thermal treatments was analyzed by grazing-incidence X-ray diffraction (GIXRD) utilizing a Bruker D8 Advance diffractometer with Cu Kα radiation at a grazing incidence angle of 1°. Oxygen vacancies in TiO_2_ sintered at various temperatures were examined by electron paramagnetic resonance (EPR) on a Bruker EMX Plus spectrometer at a measurement temperature of 77 K. The carrier concentration, mobility, and conductivity type of TiO_2_ were assessed using an Accent HL5500 Hall Measurement System under a magnetic field strength of 0.5 T and a test current ranging from −200 mA to 200 mA. Surface chemical states and band structure of TiO_2_ were investigated through X-ray photoelectron spectroscopy (XPS, Nexsa G2 Surface Analysis System, Thermo Fisher Scientific Inc., Waltham, MA, USA) with Al Kα radiation (1486.68 eV), with the instrument work function set at 4.38 eV. The current-voltage (I-V) characteristics of the sensor in various atmospheres were measured using a semiconductor analyzer (Keithley 4200-SCS, Tektronix Inc., Solon, OH, USA). AC impedance measurements were performed on a CHI760E electrochemical workstation (CH Instruments Shanghai, Inc., Shanghai, China) over a frequency range of 1 Hz to 100 kHz, with an amplitude of 100 mV and a bias voltage of up to 5 V.

2.3. Gas Sensing Measurement

The sensor was placed in an aluminum-sealed chamber with a diameter of 80 mm and a depth of 30 mm for gas-sensing tests. A Keithley 6487 picoammeter was employed to measure the sensor’s current under various operating conditions, allowing for the evaluation of its gas-sensing performance. The temperature range for testing was from 27 °C to 300 °C. Prior to each measurement at a specific temperature, synthetic air was introduced into the chamber for a sufficient duration until the current stabilized. The gas-sensing test involved alternately introducing synthetic air and a H_2_/N_2_ mixed gas into the sealed chamber for 5 min each, repeated over five cycles, with a gas flow rate maintained at 500 sccm. The sensor response was defined as I_H_2/I_Air_, where I_H_2 and I_Air_ represent the current measured in hydrogen and dry synthetic air, respectively. Additionally, a M4070 LCR meter from Dongguan Jingyan Instrument Technology Co., Ltd. (Dongguan, China) was used to measure the sensor’s capacitance in real time under different atmospheric conditions, enabling an assessment of the atmosphere’s influence on gas-sensing performance. Both the synthetic air and the mixed gas used in the experiments were supplied by Air Products and Chemicals, Inc., Beijing Branch (Beijing, China).

3. Results and Discussion

3.1. TiO2 Characterization

Figure 1 illustrates the morphology of TiO_2_ nanotubes for sensors T-300, T-400, T-500, T-600, and T-700. The TiO_2_ nanotubes annealed at temperatures from 300 °C to 700 °C demonstrate similar morphological characteristics, all displaying a highly ordered structure. The length of the TiO_2_ nanotubes measures 4.55 μm, while the thickness of the unoxidized Ti film at the base of the nanotubes is approximately 1 μm.

The as-anodized unannealed TiO_2_ nanotubes exhibit an amorphous phase, as shown in Figure S1. Figure 2 presents the GIXRD patterns of TiO_2_ nanotubes from the T-300, T-400, T-500, T-600, and T-700 samples. After high-temperature processing, the phase of TiO_2_ nanotubes transforms from amorphous to crystalline. The GIXRD patterns of TiO_2_ annealed at 300 °C, 400 °C, and 500 °C correspond to the characteristic peaks of the anatase phase TiO_2_. At an annealing temperature of 600 °C, a diffraction peak begins to appear at 27.440°, which corresponds to the characteristic peak of the (110) crystal plane of rutile phase TiO_2_. This indicates that partial anatase phase TiO_2_ starts to transform into rutile phase TiO_2_ at 600 °C. Both anatase and rutile phases of TiO_2_ coexist at 700 °C. However, the intensity of the characteristic peak (101) of anatase TiO_2_ is significantly higher than that of the characteristic peak (110) of rutile TiO_2_; therefore, the TiO_2_ phase remains predominantly anatase. The variation in the phase of TiO_2_ nanotubes with temperature is generally consistent with that reported in the literature [11].

Figure 3a presents the EPR spectrum of the TiO_2_ nanotubes, where the symmetric signal centered at g = 2.0036 is attributed to oxygen vacancies in TiO_2_ [26]. A quantitative analysis of the resonant absorption peak area shown in Figure 3a indicates that the concentrations of unpaired electrons in TiO_2_ nanotubes for T-300, T-400, T-500, T-600, and T-700 are 6.0 × 10^17^, 6.7 × 10^17^, 7.2 × 10^17^, 8.2 × 10^17^, and 9.0 × 10^17^, respectively (Figure 3b). Since the concentration of oxygen vacancies in TiO_2_ is proportional to the number of unpaired electrons, it can be concluded that the oxygen vacancy concentration in TiO_2_ nanotubes increases with annealing temperature. At higher temperatures, lattice oxygen in the TiO_2_ nanotubes vacates its site, leading to the formation of oxygen vacancies and the release of electrons and an oxygen atom—an endothermic process [20,21]. Consequently, the concentration of oxygen vacancies in TiO_2_ nanotubes rises with increasing annealing temperature.

The electron released during the formation of oxygen vacancies from lattice oxygen in TiO_2_ at high temperatures significantly affect its electrical properties. Table 1 summarizes the Hall effect measurement results for TiO_2_ nanotubes from the T-300, T-400, T-500, T-600, and T-700 samples. The Hall coefficients of all samples are negative, indicating that TiO_2_ functions as an n-type semiconductor [27]. Consequently, oxygen vacancies serve as donor impurities, enhancing the carrier concentration in the TiO_2_ nanotubes. This finding aligns with the data presented in Table 1, where the carrier concentration increases with higher sintering temperatures. Since the carrier mobility among the samples sintered at different temperatures shows minimal variation, the resistivity decreases as the annealing temperature increases.

Oxygen vacancies in TiO_2_ not only function as donor impurities that influence its electrical properties but also serve as active adsorption sites for atmospheric oxygen and other species, impacting the gas-sensing characteristics of TiO_2_ [20,28]. Figure 4 presents the O1s XPS spectra of TiO_2_ nanotubes for unannealed, T-300, T-400, T-500, T-600, and T-700 samples. The peaks at 529.87 eV and 532.35 eV correspond to lattice oxygen (O_L_) in TiO_2_ and adsorbed water molecules (O_H_2_O_) on the TiO_2_ surface, respectively; the peak at 531.44 eV also indicates hydroxyl oxygen (O_OH_) or surface-adsorbed oxygen (O_S_) [29,30,31,32]. As the annealing temperature increases, the concentration of oxygen vacancies in TiO_2_ rises (Figure 3), resulting in a decrease in the binding energy of lattice oxygen and a corresponding downward shift in the binding energies of other oxygen species. The quantitative analysis results from Figure 4 are summarized in Table S1. The unannealed TiO_2_ nanotubes exhibit the lowest O_L_ content and the highest concentrations of O_OH_ and O_H_2_O_, attributed to the adsorption of water molecule from the anodization electrolyte onto the TiO_2_ surface. High-temperature treatment not only promotes the desorption of surface groups, such as water molecules and hydroxyls, from TiO_2_ but also facilitates the formation of oxygen vacancies, which can act as adsorption sites for oxygen and water molecules from the air. Under the influence of these vacancies, water molecules can convert into hydroxyl groups [21,30,33]. As a result, the contents of O_OH_ and O_H_2_O_ in unannealed TiO_2_ are significantly higher than in sintered TiO_2_. The proportion of O_L_ gradually increases from T-300 to T-600 and then decreases again in the T-700 sample. Meanwhile, O_S_, O_OH_, and O_H_2_O_ exhibit complex variation trends.

There are differences in the band structures between anatase-phase and rutile-phase TiO_2_. Oxygen vacancies, which act as donor impurities in TiO_2_, can also modify its band structure. These variations in the band structure subsequently influence the gas-sensing performance of TiO_2_ [8]. The minimum kinetic energies of photoelectrons from TiO_2_ in samples T-300, T-400, T-500, T-600, and T-700, derived from the kinetic energy mode of XPS valence band spectra shown in Figure 5a, are 1478.10 eV, 1478.14 eV, 1478.28 eV, 1478.17 eV, and 1478.04 eV, respectively. Based on references [34,35] and Equation (S1), the Fermi levels (E_F_) of TiO_2_ for these samples are calculated to be −4.20 eV, −4.16 eV, −3.98 eV, −4.13 eV, and −4.26 eV, respectively. As indicated in Figure S2, the Fermi level (E_F_) of TiO_2_ without high-temperature annealing is −4.98 eV. From the binding energy mode of the XPS valence band spectra presented in Figure 5b, the differences between the valence band maximum (E_V_) and the Fermi level (E_F_) for samples T-300, T-400, T-500, T-600, and T-700 are determined to be 2.72 eV, 2.81 eV, 2.89 eV, 2.84 eV, and 2.82 eV, respectively [36]. Consequently, the valence band maxima (E_V_) for the samples are calculated as −6.92 eV, −6.97 eV, −6.87 eV, −6.97 eV, and −7.08 eV, respectively. As shown in Table 1, the carrier concentrations of the different samples are on the order of 10^14^, significantly lower than the effective density of states in the conduction band of TiO_2_, which is 9.93 × 10^18^ cm^−3^ [10]. Therefore, applying the Boltzmann distribution (Equation (S2)) results in the conduction band minima (E_C_) for samples T-300, T-400, T-500, T-600, and T-700 being −3.93 eV, −3.89 eV, −3.71 eV, −3.86 eV, and −4.00 eV, respectively. Schematic diagrams of the band structures for TiO_2_ in samples T-300, T-400, T-500, T-600, and T-700 are illustrated in Figure 5c.

3.2. Hydrogen-Sensing Performance of TiO2 Sensors

Figure 6a,b display the response and recovery curves of the T-300, T-400, T-500, T-600, and T-700 sensors to 5 ppm H_2_ and 200 ppm H_2_. As depicted in Figure 6c, the initial current of the sensors in air decreases exponentially with increasing annealing temperature. Figure 6d illustrates that the response values of the sensors to both 5 ppm H_2_ and 200 ppm H_2_ initially increase and then decline with increasing annealing temperature. The T-500 sensor, annealed at 500 °C, demonstrates the highest gas response, with values of 1.6 × 10^3^ for 5 ppm H_2_ and 6.9 × 10^7^ for 200 ppm H_2_, which are approximately 1000 times greater than those of the T-300 sensor. This trend of first increasing and then decreasing response with annealing temperature aligns with the changes in the Fermi level of TiO_2_ as the annealing temperature rises (Figure 5c). This suggests that the alterations in the Fermi level may fundamentally influence the differences in gas-sensing performance related to annealing temperature.

Figure 6e presents the response and recovery curves of the T-500 sensor to H_2_ concentrations ranging from 1 to 1000 ppm at 150 °C, showing clear and stable responses across all concentrations. The response value of the T-500 sensor in a nitrogen background gas is 10, while other sensors have shown no detectable response to nitrogen. The relationship between the response of the T-500 sensor and gas concentration is depicted in Figure 6f. The response values in Figure 6d–f have already been subtracted from the response values in nitrogen. With both axes on a logarithmic scale, linear relationships are observed in the ranges of 1–10 ppm and 100–1000 ppm, with slopes of 1.23 and 0.16, respectively. This indicates that although the sensor adheres to a power law in both concentration ranges, its responses to H_2_ are associated with different gas sensing mechanisms in each range. The T-500 sensor shows a response value of 242 at 1 ppm H_2_.

3.3. Tuning Gas-Sensing Performance via Modulation of Pt-TiO2 Schottky Barrier Height

The gas-sensing response process of TiO_2_ sensors fundamentally involves the conversion of gas signals into electrical signals, with carrier transport playing a crucial role in determining the sensors’ response characteristics [28]. The energy band structure of TiO_2_ significantly influences the carrier transport properties in TiO_2_ Schottky sensors by affecting the interfacial characteristics of the Pt-TiO_2_-Ti [10]. Thus, to elucidate how annealing temperature impacts the gas-sensing performance of the sensors, a detailed analysis of the Pt-TiO_2_-Ti interface is essential.

Using the half of Pt interdigitated electrodes on the surface of TiO_2_ nanotubes as the anode and the Ti film at the bottom of the nanotubes as the cathode, current-voltage (I-V) curves were recorded to characterize the properties of the Pt-TiO_2_-Ti interface in the sensor. The I-V curve for the Pt/TiO_2_/Ti sample without high-temperature annealing, shown in Figure S3, demonstrates ohmic behavior. In contrast, the I-V curves for sensors annealed at higher temperatures—designated T-300, T-400, T-500, T-600, and T-700—are presented in Figure 7a. These curves indicate that after annealing, the Pt/TiO_2_/Ti samples exhibit unidirectional conductivity, forming a Schottky junction at the Pt-TiO_2_ interface, while maintaining ohmic contact at the TiO_2_-Ti interface. This distinction arises from the differences in work functions between the metals (Pt and Ti) and the TiO_2_ semiconductor. Specifically, the work function of Pt is 5.3 eV, significantly higher than that of TiO_2_ (as illustrated in Figure 5c), while Ti has a work function of 3.9 eV, which is closer to that of TiO_2_ [16,37]. The threshold voltages for the T-300, T-400, T-500, T-600, and T-700 sensors are 1.5 V, 1.4 V, 2.4 V, 3.5 V, and 4.1 V, respectively, while the reverse breakdown voltages are 5.2 V, 6.0 V, 14.4 V, and greater than 20 V. Both the threshold voltage and the reverse breakdown voltage increase with rising annealing temperature. This leads to an exponential decrease in the initial current of the sensor in air as the annealing temperature increases (Figure 6c).

However, the built-in potential formed between the Pt electrode and TiO_2_ initially increases with annealing temperature before decreasing, peaking at 1.32 eV at 500 °C (Figure 7b). This suggests that the magnitude of the built-in potential does not primarily influence the I-V characteristics of the sensors.

High-temperature treatment enhances atomic diffusion, which helps eliminate defects at the Pt-TiO_2_ interface, facilitating atomic bonding and promoting the formation of a Schottky junction. As the temperature rises, this process accelerates, establishing a positive correlation between annealing temperature and both the threshold voltage and reverse breakdown voltage [38].

The I-V curves of the sensors indicate the formation of a Pt-TiO_2_ Schottky junction. This junction establishes a depletion layer capacitance at the Pt-TiO_2_ interface [39]. As a result, the electrical characteristics of the Schottky junction cannot be fully characterized by DC-collected I-V curves, necessitating further investigation using AC impedance measurements to explore how annealing temperature affects the Schottky junction in the T-300, T-400, T-500, T-600, and T-700 sensors.

Using half of the Pt interdigitated electrodes on the surface of the TiO_2_ nanotubes and a Ti film at the bottom as electrodes, reverse biases of 0 V, 1 V, 3 V, and 5 V were applied. Figure 8a,b display the Nyquist and Bode plots for the T-300, T-400, T-500, T-600, and T-700 sensors under 0 V bias. The Nyquist plot reveals an approximate semicircle, with test frequency decreasing from left to right. Additionally, AC impedance measurements were conducted on the sensors under reverse biases ranging from 0 V to 5 V, and the resulting Nyquist and Bode plots are shown in Figures S4–S8.

As the applied reverse bias increases, the semicircle diameter, total impedance, and peak phase angle all decrease. The intersection of the low-frequency portion of the semicircle with the real axis represents the total resistance of the circuit, while a phase angle closer to −90° indicates more pronounced capacitive reactance characteristics [40,41]. An increase in reverse bias reduces the depletion layer capacitance of the Pt-TiO_2_ Schottky junction, potentially leading to junction breakdown, alongside an increase in reverse saturation current density through the junction, resulting in decreased capacitive reactance and resistance [39]. Thus, the Nyquist and Bode plots from AC impedance measurements provide valuable insights into the Pt-TiO_2_ Schottky junction.

A higher built-in potential at the Pt-TiO_2_ interface corresponds to a smaller depletion layer capacitance. Among the sensors, the T-500 sensor exhibits the largest built-in potential at the Pt-TiO_2_ interface (Figure 7b), resulting in the largest semicircle diameter, total impedance, and peak phase angle (Figure 8). Interestingly, although the built-in potential at the Pt-TiO_2_ interface in the T-700 sensor is lower than that in the T-300 sensor, its semicircle diameter and total impedance are higher, while the phase angle peak is lower. This suggests that the phase angle peak is more sensitive to variations in built-in potential height, whereas total impedance reflects the bonding properties at the Pt-TiO_2_ interface.

The impedance spectra were analyzed using equivalent circuit models. The Nyquist plot obtained from AC impedance measurements of metal oxide semiconductor gas sensors displays a semicircle that results from the superposition of three individual semicircles, which are often not distinctly separable [41]. Additionally, since the phase angle peaks are all below −90°, this suggests the presence of non-ideal capacitive behavior in the circuit. Consequently, a constant phase element (Q) was employed to represent the non-ideal capacitance in the fitting of the impedance spectra [41]. The fitting process utilized a circuit model consisting of a series combination of resistor R_1_ and a parallel sub-circuit of R_2_ and Q_1_, followed by another parallel sub-circuit of R_3_ and Q_2_ in series. The corresponding equivalent circuit is illustrated in the inset of Figure 8a. The fitted parameters are detailed in Table S2. The constant phase elements listed in the table have been converted to true capacitances using the Brug method [42]. In this model, R_1_ represents the bulk and contact resistance; R_2_ and C_1_ correspond to the resistance and depletion-layer capacitance of the Pt-TiO_2_ Schottky junction, respectively; while R_3_ and C_2_ denote the resistance and capacitance associated with the Schottky barriers formed between the TiO_2_ nanotubes [41].

From Table S2, it is evident that the T-500 sensor exhibits the highest Pt-TiO_2_ Schottky junction resistance and the lowest depletion-layer capacitance. This is attributed to the excellent interfacial adhesion and the maximum built-in potential present at the Pt-TiO_2_ interface in the T-500 sensor.

In summary, the annealing temperature influences the Schottky barrier height of Pt-TiO_2_ by altering the phase composition and oxygen vacancy concentration of TiO_2_, leading to an initial increase followed by a decrease as the temperature rises. The T-500 sensor exhibits the highest built-in potential and Schottky barrier height in its Pt-TiO_2_ interface, with no significant Fermi-level pinning observed. As a result, the T-500 sensor shows the highest gas-sensing response. Furthermore, despite having a lower Schottky barrier height, the T-700 sensor demonstrates a higher gas-sensing response to low-concentration hydrogen compared to the T-300 sensor, due to its superior interfacial bonding properties at the Pt-TiO_2_ junction. This indicates that the annealing temperature influences the gas sensing performance of the sensors by jointly regulating the Schottky barrier height and the interfacial bonding characteristics of Pt-TiO_2_.

3.4. Analysis of Gas-Sensing Response Mechanism

To investigate the response mechanism of the Pt/TiO_2_Ti sensor, the T-500 sensor was selected for study. In this configuration, the half of Pt interdigitated electrodes positioned atop the TiO_2_ nanotubes functions as the anode, while the Ti film located at the bottom of the nanotubes serves as the cathode. The operating temperature was maintained at 150 °C. I-V curves for the T-500 sensor were recorded in air and in atmospheres containing 5 ppm, 10 ppm, 20 ppm, 50 ppm, and 100 ppm H_2_, as illustrated in Figure 9. As the hydrogen concentration increases, the unidirectional conductivity of the sensor gradually decreases and may even disappear. This suggests that the Pt-TiO_2_ Schottky junction functions effectively only at low hydrogen concentrations.

To investigate the response mechanism of the sensor, AC impedance analysis was performed on the T-500 sensor under varying hydrogen concentrations. Figure 10a–c present the Nyquist plots of the T-500 sensor in air and at hydrogen concentrations ranging from 5 ppm to 100 ppm, while Figure 10d shows the corresponding Bode plots.

After introducing hydrogen, the semicircle diameter in the Nyquist plots, the total impedance of the sensor, and the phase angle peak all exhibited a monotonic decrease. At a hydrogen concentration of 50 ppm, the complete semicircle in the Nyquist plot became incomplete. When the hydrogen concentration reached 100 ppm, the semicircle transformed into a straight line with increasing test frequency, indicating that the circuit’s resistive characteristics were enhanced while its capacitive reactance characteristics weakened. The capacitive reactance arises from the depletion layer capacitance of the Pt-TiO_2_ Schottky junction, while the resistance primarily stems from the Schottky junction resistance and the resistance of the TiO_2_ nanotubes.

This suggests that at low hydrogen concentrations, the depletion layer capacitance of the Pt-TiO_2_ Schottky junction is dominant, whereas at high hydrogen concentrations, the change in sensor current is mainly attributed to variations in resistance of TiO_2_ nanotubes. The equivalent circuit after hydrogen introduction is illustrated in the inset of Figure 10a. Unlike the equivalent circuit in air (Figure 8), this circuit incorporates an R_2_Q parallel component representing the Schottky junction, which is connected in parallel with an RL series circuit. Here, RL denotes the dipole layer formed by hydrogen at the Pt–TiO_2_ Schottky junction.

The capacitance is derived from the Pt-TiO_2_ Schottky junction. Therefore, by monitoring the capacitance variation in the sensor in different gases in real time, the role of the Pt-TiO_2_ Schottky junction under various atmospheres can be observed more directly. Based on this, the response and recovery curves of the T-500 sensor toward 1 ppm to 50 ppm H_2_ were collected in real time via the capacitance method, as shown in Figure 10e,f. The response and recovery curves of sensors annealed at other temperatures, obtained by the capacitance method, are presented in Figure S9.

When hydrogen is introduced, it reacts with the negatively charged oxygen adsorbed on the TiO_2_ surface, causing electrons to transfer from the adsorbed oxygen back to the TiO_2_ matrix. This interaction leads to secondary doping of TiO_2_ by H_2_, resulting in an increase in the donor impurity concentration (N_D_) of TiO_2_ [39]. Simultaneously, H_2_ dissociates at the Pt interface, and the hydrogen atoms diffuse to the Pt-TiO_2_ interface, forming a dipole layer and creating additional charge states. This process reduces the built-in potential (V_D_) and lowers the Schottky barrier height [24,39]. The depletion layer capacitance (C) can be expressed by Equation (1) [39]:

$$[eqn]$$

In the equation, q represents the electron charge, ε is the dielectric constant of TiO_2_, V_bia_ is the applied bias voltage, k is the Boltzmann constant, and T is the absolute temperature.

As indicated by Equation (1), in the hydrogen concentration range of 1 ppm to 10 ppm, the Schottky barrier junction width (w_s_) of the T-500 sensor decreases, leading to an increase in the depletion layer capacitance at the Schottky junction (Figure 10e) and a reduction in capacitive reactance. Figure 10f illustrates that when the hydrogen concentration reaches 20 ppm or higher, the capacitance initially increases while the capacitive reactance decreases. However, after a period of gas exposure, the capacitance drops rapidly. This decline is not due to an increase in capacitive reactance; rather, it results from the pronounced secondary doping of TiO_2_ by high-concentration H_2_. As the carrier concentration in TiO_2_ continues to rise, the Schottky barrier width (w_s_) further narrows, leading to a significant increase in tunneling current through the Schottky junction and placing the junction in a “short-circuit” state. This indicates that at high hydrogen concentrations, the modulatory effect of the Schottky junction on the sensor current is significantly diminished. These observations align well with the I-V curves shown in Figure 9 and the AC impedance data presented in Figure 10a–d.

Based on the I-V curves, AC impedance tests, and real-time capacitance variation data, the response mechanisms of TiO_2_ Schottky sensors at different hydrogen concentrations are elucidated, as illustrated in Figure 11.

In hydrogen concentrations of 1 to 10 ppm, the sensor response is primarily dominated by the Pt-TiO_2_ Schottky junction. As the hydrogen concentration increases, both the built-in potential height (V_D_) and the Schottky barrier height (φ_b_) decrease, while the Schottky junction width (w_S_) narrows, leading to a reduction in capacitive impedance. During this stage, the sensor response exhibits a power-law relationship with hydrogen concentration, characterized by a slope of 1.22 (Figure 6f).

When the hydrogen concentration reaches 20 ppm, the narrowing of the Schottky junction width reduces its impedance, causing the resistance of the TiO_2_ nanotubes to begin influencing the sensor current. This results in a deviation from the power-law relationship observed between 1 ppm and 10 ppm H_2_, with the deviation becoming more pronounced as hydrogen concentration increases. In this regime, the sensor response is jointly influenced by the Pt-TiO_2_ Schottky junction and the TiO_2_ nanotubes.

In hydrogen concentrations of 100 ppm or higher, the Schottky junction width decreases significantly, leading to a substantial increase in tunneling current across the junction, effectively causing it to behave like a short circuit. Under these conditions, the sensor current is primarily influenced by the Schottky barriers between the nanotubes, which also follow a power-law relationship. However, since these inter-nanotube barriers are much lower than the Pt-TiO_2_ interfacial Schottky barrier, the corresponding slope is only 0.16 (Figure 6f). Thus, the sensor response in this high-concentration range is governed by the Schottky barriers between the TiO_2_ nanotubes. Due to their position and low height, these barriers are typically attributed to the resistance of TiO_2_.

Table 2 provides a comparison of the gas-sensing performance of the silicon-based micro TiO_2_ Schottky-type hydrogen sensor developed in this study with previously reported hydrogen sensors. The results indicate that our sensor demonstrates superior hydrogen sensing capabilities, especially in detecting low concentrations of hydrogen.

4. Conclusions

Pt/TiO_2_/Ti sensors were treated at annealing temperatures of 300 °C, 400 °C, 500 °C, 600 °C, and 700 °C. The combined effects of TiO_2_ phase and oxygen vacancy concentration result in a trend where the Pt-TiO_2_ Schottky barrier height initially increases and then decreases with rising annealing temperatures, peaking at 500 °C. Similarly, the sensor’s response to hydrogen shows an initial increase followed by a decline as the annealing temperature rises. The sensor treated at 500 °C demonstrates the highest gas-sensing response, achieving values of 1.6 × 10^3^ for 5 ppm H_2_ and 6.9 × 10^7^ for 200 ppm H_2_—over 100 times greater than those of sensors processed at other temperatures. The theoretical minimum detectable hydrogen concentration is 80 ppb. Throughout the tested hydrogen concentration range of 1 ppm to 1000 ppm, the relationship between gas-sensing response and hydrogen concentration (in logarithmic coordinates) exhibits two distinct linear regions: 1 ppm–10 ppm and 100 ppm–1000 ppm. This behavior can be attributed to the dominance of the Schottky junction at low hydrogen concentrations, while at high concentrations, the sensing mechanism is primarily governed by the resistance of TiO_2_.