The sensitivity enhancement of TiO2-based VOCs sensor decorated by gold at room temperature

Detection of hazardous toxic gases for air pollution monitoring and medical diagnosis has attracted the attention of researchers in order to realize sufficiently sensitive gas sensors. In this paper, we fabricated and characterized a Titanium dioxide (TiO2)-based gas sensor enhanced using the gold nanoparticles. Thermal oxidation and sputter deposition methods were used to synthesize fabricated gas sensor. X-ray diffraction analysis was used to determine the anatase structure of TiO2 samples. It was found that the presence of gold nanoparticles on the surface of TiO2 enhances the sensitivity response of gas sensors by up to about 40%. The fabricated gas sensor showed a sensitivity of 1.1, 1.07 and 1.03 to 50 ppm of acetone, methanol and ethanol vapors at room temperature, respectively. Additionally, the gold nanoparticles reduce 50 s of response time (about 50% reduction) in the presence of 50 ppm ethanol vapor; and we demonstrated that the recovery time of the gold decorated TiO2 sensor is less than 40 s. Moreover, we explain that the improved performance depends on the adsorption-desorption mechanism, and the chemical sensitization and electronic sensitization of gold nanoparticles.


Introduction
The ever-increasing use of automated vehicles in our era poses significant threats to human health and the environment.
Volatile organic compounds (VOCs) and hydrocarbons are examples of such threats. It has been demonstrated in many reports that the level of these pollutions can exceed the threshold limit value in residential areas. Monitoring and estimating these harmful toxic VOCs require highly precise, efficient, and reliable analytical devices. Metal oxide-based chemoresistive sensors are advantageous over other chemical sensors such as acoustic-based, optical and electro-analytical due to their ease of production, low cost and good portability [1]. Different metal oxides such as WO 3 , SnO 2 , ZnO, and TiO 2 have been reported as potentially promising candidates for gas sensing, whether applied individually or in composite structures [2]. TiO 2 , an n-type semiconductor material with high resistance and a bandgap of approximately 3 eV, has received great attention in the field of gas sensing as it is eco-friendly, chemically stable, and also has catalytic properties and allows for structural modulation [3]. The anatase phase of TiO 2 exhibits a high ability to react with gas molecules due to the large number of oxygen vacancies [4]. Thus, it is more commonly used in gas sensors than the Rutile and brookite phases. Various n-type TiO 2 structures -pristine or doped, composites-are employed to sense different types of VOCs such as ethanol, propanol, acetone, nitrogen oxide, carbon disulfide, and toluene [5].
Limitations of pristine TiO 2 sensors such as low sensitivity, high operating temperature, weak electrical stability, and nonselectivity are the most prominent remaining challenges. Thus, various methods have been proposed to overcome these limitations. A reliable and effective method is the addition of noble metals such as gold, silver, platinum, and palladium, which can be performed during the synthesis of TiO 2 or after its completion [6,7]. These noble metals change the electronic characteristics of TiO 2 , redesign the crystalline phases, and increase the density of surface defects. Thus, noble metals act as activators to improve gas response and selectivity and reduce the response and recovery time and operating temperature [8,9]. in addition, noble metals have a greater impact on the gas sensing of TiO 2 nanostructures than other structures [5,8].
One of the low-cost synthesis methods of Titanium dioxide nanoparticles in the form of thin-film is thermal oxidation. This growth method can lead to the formation of titanium dioxides in the anatase phase by controlling the air pressure in deposition systems [10]. The Titanium dioxide phase variance reported in the literature is generally attributed to the difference between the titanium layer nanostructure and its grain size [11,12]. Oxidation starts from the surface and boundaries of the grains that the titanium film is composed of. The oxidation process continues until the titanium grain turns into titanium dioxide. The induced stress in the oxide changes its surface energy and controls the phase determination process. The oxidation temperature, pressure, and grain size affect the stress level. The higher the stress level gets, the higher the surface energy of the oxide becomes, and the oxide phase becomes anatase [13,14]. Extensive research has been conducted regarding gold-decorated TiO 2 gas sensors [15][16][17]. However, a detailed review of existing studies demonstrates that nanoparticle-based gold-doped TiO 2 sensors still face challenges such as nanocomposites synthesis, operating temperature, response time, recovery time, and lifespan.
In this work, a VOC vapor sensor based on the decoration of TiO 2 nanoparticles was fabricated. The growth of TiO 2 nanoparticles was performed using the thermal oxidation method and the decoration of gold nanoparticles using a sputter deposition. Metal contacts were used to measure and analyze the electrical properties of the fabricated sensor. The structure, morphology, and topology of Au-TiO 2 layers were studied using the SEM, XRD, AFM, and EDX characterizations. The absorbance of the active layer was presented to evaluate the    bandwidth. Moreover, the gas sensor's sensitivity, selectivity, and transient response were studied.

Growth of titanium dioxide
A silicon wafer substrate with the dimensions of 5 mm × 10 mm was ultrasonically cleaned in acetone, ethanol, and distilled water consecutively for 5 min. A titanium thin film was deposited on the surface of Si/SiO 2 using an e-beam evaporator system. The Si/SiO 2 substrate was fixed on a rotating holder. The device was cleaned under vacuum conditions for 15 min by argon ion beam and with 100 V DC voltage and 2 A current. A titanium layer of 100 nm was grown at a rate of approximately 0.05 nm per second. The growth temperature was kept at 100°C before the deposition process and then at 150°C during the growth process. During the growth process, the holder's spin speed was determined at 5 rpm so that a uniform growth is performed. Next, the sample was placed inside a furnace in the presence of air and at 600°C for 4 h. This annealing process resulted in the growth of a titanium dioxide layer with a thickness of approximately 50 nm on a titanium layer with a thickness of about 70 nm. Obtained SEM top and cross-sectional images of TiO 2 are shown in figure 1. As shown in figure 1(a), the growth of the oxide on the surface of titanium has increased the thickness of the sample, which corresponds to the positive coefficient of titanium's thermal oxidation growth rate [18]. Figure 1(b) displays high porosity on the surface of the obtained oxide. The mentioned porosity and the existence of grain boundaries are important factors in gas sensing [12,19,20].
XRD analysis (Inel, EQUINOX3000, 40 kV and 25 mA) was employed to determine the crystalline phase of the obtained TiO 2 layers. In XRD equipment monochromatic optic Kα1 or Kα1/2 were used. It was found that the oxide grown on the titanium layer at 600°C exhibits an anatase phase. Figure 2 shows the XRD pattern of this sample.

Growth of gold nanoparticles
Decoration of the fabricated TiO 2 sample with gold nanoparticles was accomplished using a sputtering device under certain conditions. For this purpose, a 99.99% purity 2-inch diameter gold target was used. The chamber's temperature was kept at room temperature, and the pressure was held at 1 × 10 -5 mTorr. The voltage and current for the argon beam were set to 100 V and 3 A, respectively. The distribution and size of gold nanoparticles were controlled by the sputtering time which was set to 2, 4, 6, 8 or 10 s.
After the deposition process, the sample was annealed at 100°C for 5 h. Figure 3 shows SEM images of the decorated TiO 2 samples. As shown in figure 3, gold metal nanoparticles have grown irregularly and randomly on the porous surface of TiO 2 .
Identification of the elements present in the sample is possible through energy dispersive x-ray (EDX) analysis. The fabricated TiO 2 sensor was characterized by EDX, before and after its decoration with gold, using an FEI Nova Nanosem equipped with EDAX Octane detector. As shown in figure 4, the EDX spectrum of the sample containing gold nanoparticles has an energy peak of approximately 2 eV, which demonstrates the presence of gold in the sample. From the EDX analysis, it is roughly estimated that the atomic ratio percentage of gold to TiO 2 is 10%. This recognition was made because of the registered energy, which is characteristic of gold nanoparticles. The Si signals originate from the substrate. The homogenous distribution of constituent element can be seen in elemental mapping presented in figures 4(c)-(d).

Fabrication of the gas sensor
Gas sensing equipment and gas sensing test methods have been explained in our previous studies [21,22]. Two ∼3 mm wide, 200 nm thick gold stripe electrodes and a tantalum adhesive layer were deposited on the TiO 2 sample. Silver paste was used to ensure the electrical connections of the platinum wires. A sample of the prepared sensor and the schematic diagram of the measurement system is shown in figures 5(a) and (b), respectively.
The gas response was obtained by measuring the resistance of the two electrodes using a Sanwa handheld digital   The sensitivity of the gas sensor is expressed by the ratio of the measured resistance before exposure to gas to the measured resistance after it and is defined by the expression: where R a and R g are the sensor's resistance under room conditions (clean air) and in the presence of gas, respectively.

Results and discussion
3.1. Structural and morphological characteristics N 2 adsorption-desorption isotherms were applied to the sensor. The surface area was investigated using a Brunauer-Emmet-Teller (BET) method. Figure 6 displays N 2 adsorption-desorption isotherms and the pore size distribution diagram (BJH) of Au/TiO 2 and pure TiO 2 samples. The sample shows an isotherm similar to type IV, which is representative of mesoporous solid [23]. Data regarding the pores' structure have been listed in table 1. Both samples demonstrate similar N 2 adsorption-desorption isotherms. Although surface areas with wide pore distribution can be seen in both samples, the gold-decorated sample has a larger pore volume and a smaller pore size compared to the TiO 2 sample. This difference in size and pore volume leads to an increase in the surface-to-volume ratio in the sensor. Increasing the volume-to-surface ratio course increases the sensitive surface to detect gas molecules. Therefore, gas response increases.
The output characteristics (I-V ) of TiO 2 sensors before and after the decoration of gold on the TiO 2 are presented in figure 7(a). Both I-V curves show Schottky contact which indicates same contact before and after the decoration with gold nanoparticles. Due to the sparse nature of the gold nanoparticles, they had only a small impact on the conductivity of the sample [9,24]. The I-V characteristic of the Au-TiO 2 sample at different temperatures has been demonstrated in figure 7(b) in order to investigate the electrical contact of the electrodes [25]. Since gold's work function (W Au = 5.1 eV) [26] is larger than titanium dioxide's work function (W TiO2 = 4.6 eV) [27], the contact between TiO 2 and Au causes the charge to transfer from the semiconductor (TiO 2 ) to gold (Au), and hence, a Schottky barrier forms at the junction. It has been demonstrated that the barrier height of the gold-titanium dioxide contact corresponds to the Schottky-Mott model, and the barrier's value varies between 0.9 and 1.2 eV [28,29]. This difference is attributed to undesired surface states [30]. With due attention to the formed Schottky diode in the contact, all samples' resistance measurements were recorded at a voltage of 1 V to ensure reliable gas sensing results.
The optical response and optical bandgap of the layer are other criteria that help determine gold's dispersion in the TiO 2 thin film. Figure 8(a) shows the UV-vis absorption spectrum of TiO 2 and Au/TiO 2 at room temperature. To removal substrate absorption effect in obtained UV-vis spectrum, original samples, samples on the glass substrate and Si substrate were measured separately. As can be observed in figure 8(a), TiO 2 exhibits a sharp absorption edge at about 400 nm, which is related to the bandgap excitation of TiO 2 [31]. Due to the uneven distribution of gold nanoparticles on the surface of Titanium dioxide, gold does not exhibit any evident absorption. Moreover, as SEM and EDX characterizations demonstrate, the actual Au content of Au/TiO 2 is not much, which might lead to the lack of a prominent absorption peak. Nevertheless, as can be seen in figure 8(a), a slight increase in light absorption is apparent in the Au/TiO 2 sample compared to the TiO 2 sample.
As shown in figure 8(a), the average adsorption in the range of 400-500 nm is below 10%. The bandgap of Au/TiO 2 can be determined using the adsorption of TiO 2 thin film in various wavelengths. Figure 8(b) shows the Tauc plot used to calculate the E g of the Au/TiO 2 sample. This plot was used to obtain E g and is based on equation (2): where α is the adsorption coefficient, A is the constant coefficient (band tailing parameter), hν is the energy of the incident photon, and E g is the bandgap energy. Extrapolation of the linear region obtains the bandgap on the hν axis. The obtained E g for the TiO 2 sample was calculated to be approximately 3.4 eV which corresponds to the literature [32]. Moreover, figure 8(b) demonstrates band gap energy obtain for the both samples. Based on the optical band-energy spectra shown in figure 8(b), it is clear that gold particles could change the light absorption behavior, as well as alter the optical bandgap of TiO 2 . The optical bandgap is found to be composed and there is a slight increase in the bandgap of the TiO 2 with gold decoration. The doping of various transitional metal ions into TiO 2 could shift its optical absorption edge from the UV into the visible light range. Unlike TiO 2 , which only absorbs light energy in the range of the UV spectrum, Au/TiO 2 absorbs additional light energy in the visible range due to the presence of the plasmonic phenomenon in the Au nanoparticles [33].

Gas sensing results
The fabricated sensors using bare TiO 2 and decorated TiO 2 were tested for sensing different VOC vapors. Figure 9 and the table inside it shows the sensing response of two gas sensor samples to 50 ppm of acetone, ethanol and ethanol vapors. As can be seen in figure 9, gold decoration does not lead to a significant change in sensitivity of the fabricated samples. As the gold-decorated TiO 2 has a sensitivity response of 1.13 to 50 ppm acetone vapor, while the intrinsic TiO 2 has a sensitivity of 1.1 in similar conditions. Gold decoration does significantly impacted the sensors' response time, such that the 100 s response time in the TiO 2 sensor decreased to about 50 s in the Au/TiO 2 sensor for 50 ppm of all tested VOC vapors. It should attribute to the catalytic promotion effect of Au, which leads to the decrease of activation energy and results in an increase of gas adsorption for the gassensing performance [34].
When the gas sensor is exposed to different concentrations of gas, the electrical conductivity of the active layer modulates the adsorption rate of gas molecules on the sensor surface. Gold nanoparticles in the form of discrete islands modify the surface structure of TiO 2 . In an oxygen-rich atmosphere, oxidation of the nanometer islands provides a pair of Au + and neutral Au in the equilibrium [35]. These redox pairs cause the formation of an electron depletion layer around the gold islands. The Schottky potential barrier formed at the Au 0 /Au + pairs boundary causes carriers to increase [36]. This increase in effective carriers in gas sensing leads to the spillover effect. Therefore, the number of free electrons in the Au/TiO 2 sample is much larger than the number of free electrons in TiO 2 . This fact is also confirmed by the current-voltage characteristic shown in figure 7(b). Because of the spillover effect, these trapped electrons cause the electron region to thicken and increase the oxygen adsorption sites [37]. Thus, the presence of gold nanoparticles on the surface of TiO 2 improves the Schottky barrier modulation in the oxidation process.
The sensing mechanism of the TiO 2 sensor is based on changes in the electrical conductivity of the sample at different atmospheric conditions. TiO 2 , as an n-type semiconductor, adsorbs oxygen molecules to its surface when exposed to air, which results in the trapping of the electrons from the conduction band as shown in figure 10(a) . Hence, the released h + creates a space charge and causes a larger potential barrier to form [3,38]. The increase in this potential leads to reduced electrical conductivity (see diagram on the left in figure 10(a)). This adsorption process can be defined according to equation (3): When the sensor is exposed to gas molecules (see diagram on the right in figure 10(a)), its conductivity changes due to the reaction of the adsorbed oxygen species with gas molecules, and it releases back electrons to the conduction band. This process results in a smaller potential barrier and, thus, increased conductivity. This process can be defined according to equation (4): O ads X gas e . 4 +  + -- As figure 10(b) demonstrates, the increase in gas response can be described by electronic sensitization and chemical sensitization of Au catalyst. The different work functions of Au and TiO 2 on the gas adsorbent surface can lead to electron interaction between Au and TiO 2 and the formation of an extra electron depletion layer on the joint surface. Therefore, when the surface oxygen molecules react with gas molecules, more carriers are released, which results in further change in electrical conductivity. Figure 11 demonstrates the dynamic performance of Au/TiO 2 towards different concentrations (1-200 ppm) of ethanol gas at room temperature. Moreover, this figure shows the transient sensing curves and stable sensing and recovery of the TiO 2 sample. As can be seen in figure 11, increasing ethanol concentration causes more molecules to react with the adsorbed oxygen molecules, which leads to an increase in the sensor sensitivity. The sensitivity of the sensor to 1 ppm, 5 ppm, 10 ppm, 50 ppm, 100 ppm and 200 ppm ethanol gas was found to be 1.0014, 1.0069, 1.02, 1.07, 1.138 and 1.263, respectively. The fabricated sensor shows an acceptable response to low concentrations of ethanol vapor at room temperature, hence making it an appropriate option for lowpower applications.
The dependency of the response as a function of ethanol concentration for the Au/TiO 2 sample is shown in the inset of figure 11. As can be seen in the figure, linearly increasing the concentration from 1 to 200 ppm significantly increases the response sensitivity. This linear dependency can be modeled by equation (5): where S is the sensitivity and C is ethanol vapor concentration. Moreover, the detection limit of the sensor of 0.22 ppm was recorded. The limit of detection was obtained theoretically by using extrapolation of the sensitivity curve till it hits the noise level of the device [39].  Another important factor when studying sensor performance is selectivity, which is the ability of sensors to show a precise response to the target gas in the presence of different gases [22,40]. Both TiO 2 and Au-TiO 2 sensor structures were exposed to 50 ppm six different gases, including acetone, ethanol, methanol, 1-propanol, ammonia, and carbon monoxide. Figure 12 shows the sensitivity of the fabricated sensors in the presence of different gases. The decorated sample did not exhibit increased selectivity for any of the studied gases.
Long-term stability is another one of the gas sensor's important factors, and it demonstrates the reliability of the sensor [41]. The stability of the decorated TiO 2 sensor in the presence of 50 ppm ethanol was evaluated at room temperature. As is evident in figure 13, the relative deviation of the gas response sensitivity was less than 2% in the first 15 d and about 3% in the following 15 d.
Moreover, the effect of humidity on the sensing mechanism of the fabricated sensors was investigated. Figure 14 shows the response sensitivity as a function of the applied relative humidity to TiO 2 and Au/TiO 2 sensors when exposed to 50 ppm acetone vapor. it is found that the effect of humidity on tested TiO 2 and Au/TiO 2 gas sensors is similar to the impact of humidity on the nanowire TiO 2 ethanol vapor sensors [3]. To ensure the responses obtained from the gas testing results, two other samples with the same experimental  conditions were fabricated for each type of sensor. Recorded gas sensitivity parameters of other samples guarantee all results. Table 2 demonstrates a comparison of the current research with similar studies in key sensing parameters such as sensitivity, gas concentration, response and recovery time. As can be observed, different TiO 2 structures (such as TiO 2 nanobelt, Au-TiO 2 nanoparticles, etc) were employed to detect different VOC gases, especially ethanol. Among them, we employed a simple growth method to obtain porosity and Au decoration at room temperature, which demonstrated satisfactory results compared to other works.

Conclusion
A TiO 2 gas sensor was fabricated through thermal oxidation with process control to increase porosity. Gold nanoparticles were deposited on the surface of the fabricated sensor by the sputtering technique. The developed sensors are capable to detect low concentrations of VOC vapors at room temperature. Close observation indicates that gold decoration leads to an increased sensitivity towards VOC vapors. Pure TiO 2 and Au/TiO 2 exhibited a sensitivity percentage of 3% and 7% towards 50 ppm ethanol vapor, respectively. Moreover, it was found that the response time of the decorated sample had improved by 50% compared to pure TiO 2 . The sensors showed no degradation in their response over a period of 30 d. Thus, this research can provide the basis for further extensive studies on enhancing the gas sensor's performance of VOC sensors. To continue this research using other noble metals as catalysts are recommended.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).