Abstract: TiO2 gas sensors have unique advantages such as low operating temperature, high sensitivity, fast response speed and simple preparation, and have become a hot topic of concern. This paper summarizes the conductivity type and gas sensing mechanism of TiO2 materials, introduces the research progress of TiO2 gas sensors in two aspects: using traditional doping methods and using nanotechnology to improve sensor performance; and looks forward to the future research and development direction of TiO2 gas sensors. Keywords: gas sensor; titanium dioxide; metal oxide semiconductor; gas sensing mechanism; nanosensor I. Introduction In recent years, coal mine explosions and household gas accidents have occurred frequently, causing great damage to people's lives and property. In addition, the problem of air pollution is becoming increasingly serious, and the emissions of automobile exhaust and industrial equipment have become important sources of pollution. In order to effectively control the emission of these gases, people use gas sensors to detect the content of these gases. In the detection system, the wide bandgap n-type metal oxide gas sensor has become increasingly important because of its advantages such as low cost, high sensitivity, easy operation and control, and easy compatibility with microelectronic systems [1]. The n-type wide bandgap metal oxides used as sensing materials include SnO2, Fe2O3, TiO2, In2O3, WO3, etc. [2]. Among them, SnO2 gas-sensitive material was the first to be put into production and widely used. However, with the deepening of research, people have found that TiO2 gas-sensitive material has unique advantages such as low working temperature, good performance and simple preparation [3], and has therefore gradually become another hot spot of attention. This article starts from the structure of TiO2 material, reviews the gas-sensing mechanism of TiO2, the current research progress and existing problems, and makes a prospect for future development. II. Conductivity type of TiO2 sensing material Actual crystals do not have strict periodicity like ideal crystals. In crystals, atoms vibrate slightly near their equilibrium positions. This vibration causes the energy of their thermal vibration to fluctuate. Once the atom obtains enough kinetic energy, it may leave its normal position and jump into the gap between neighboring atoms to form interstitial atoms and form a vacancy in the corresponding position [4]. The generation of such point defects introduces defect energy levels into the band gap of oxide materials, making it relatively easy for electrons to jump from the defect energy level to the conduction band or from the valence band to the defect energy level, thus enabling metal oxides with high resistivity to exhibit semiconductor properties [4, 5]. For TiO2, there are no electrons in the conduction band of pure stoichiometric TiO2. Its low conduction band energy level is mainly contributed by the 3d orbital electrons of titanium ions, and the top energy level of the valence band is mainly contributed by the 2p orbital electrons of oxygen ions. Its band gap width is about 3.0 eV to 3.20 eV. Therefore, at room temperature, pure stoichiometric TiO2 has a very high resistivity [6, 7]. When there is a certain pressure and temperature, due to the requirements of thermal equilibrium conditions, the composition of TiO2 will deviate significantly from the stoichiometry [4], resulting in a certain concentration of point defects. For example, when TiO2 is reduced in a vacuum furnace, a large number of oxygen vacancies or a large number of titanium interstitials will appear under low oxygen partial pressure. TiO2 deviating from the stoichiometric ratio can be represented by the general formula TiO2-δ, where δ≠0. The reason is that oxygen deficiency leads to oxygen vacancies or titanium excess leads to titanium interstitial atoms[6]. TiO2 belongs to ionic crystals. In ionic crystals, ionic defects are all charged, and interstitial atoms carry their own charge. As for ionic vacancies, since the corresponding charge is missing in the normal charge distribution, the vacancy is equivalent to an effective charge with the opposite sign. The introduction of such interstitial atoms or vacancies causes an excess of a certain charge. In order to meet the requirement of electrical neutrality of ionic crystals, charged vacancy site defects will capture the effective charge with the opposite sign, thereby changing its charged state. In TiO2 materials, oxygen vacancies mainly play a role[5]. Oxygen vacancies are negative ion vacancies with positive effective charge. The state in which it captures electrons and becomes neutral is called a donor-like state. It can provide electrons, so the conductivity type of TiO2 materials is generally n-type. However, research has found that the conductivity type of this oxide semiconductor material is also related to the oxygen partial pressure [8]; in addition, doping can change the conductivity type of TiO2 material. For example, by doping Cr in TiO2 and controlling the doping concentration of Cr and the heat treatment temperature of the sample, TiO2 sensing material that is converted into p-type conductivity can be obtained [9]. III. Gas Sensing Mechanism of TiO2 Gas Sensor The recognition of the gas to be measured by the sensor using TiO2 as the sensing material is first caused by oxygen adsorption on the surface. Oxygen has a strong adsorption capacity. The adsorbed oxygen first exists on the TiO2 surface in the form of physical adsorption. When it obtains a certain activation energy, it enters the form of chemical adsorption. Chemically adsorbed oxygen generally has three forms, and the specific form is determined by the ambient temperature. Experimental studies have shown that at low temperatures, the oxide surface exists in the form of "molecular ions". As the temperature increases, it transforms into the form of "atomic ions". Above 450K, surface adsorption of oxygen is dominant [10]. In this process, oxygen in the air takes away surface electrons to become chemically adsorbed oxygen. The chemical reaction equation is: Oxygen adsorption becomes oxygen ions, and electrons will be transferred from the bulk to the surface. The bulk and the surface deviate from electrical neutrality, induce a space charge layer, and generate band bending. The appearance of the space charge layer reduces the number of surface charge carriers, thereby increasing the resistance of the material. In fact, equation (1) is just a simple description of the surface reaction, and this reaction can also be completed by means of it. However, the chemical activity is usually high, and it is easy to desorb into a state that is not easy to participate in the reaction in the medium temperature range. The adsorbed oxygen ions exist in the band gap as electron acceptors and are localized on the surface of the material [12]. When there is a reducing gas R in the ambient atmosphere, the pre-adsorbed oxygen reacts with the reducing gas on the surface of the sensing material as follows: The reducing gas reacts with the pre-adsorbed oxygen ions on the surface, removes an electron and releases it back to the conduction band, which increases the conductivity of the TiO2 material and plays the role of sensing. IV. Research progress of TiO2 gas sensors The performance indicators of gas sensors mainly include sensitivity, response time, selectivity, reliability and stability. The current research on TiO2 gas sensors mainly focuses on two aspects: one is to improve the sensing performance by changing the properties of the material through traditional doping and improved preparation methods; the other is to use the rapidly developing emerging nanotechnology to make nanoparticle or nanotube sensors to significantly improve the sensor performance. 1. Traditional research methods Metal oxide semiconductor surfaces will have uniform and non-uniform reactions. When there is no catalyst, the gas will have a uniform reaction on the oxide surface, while a non-uniform reaction will occur when there is a catalyst. The energy required for a uniform reaction is greater than that for a non-uniform reaction, so appropriate doping can accelerate the reaction rate and reduce the operating temperature. In addition, the surface structure and morphology of the material also affect the performance of the sensor. For example, the sensitivity, selectivity and stability of gas sensors made of sintered bodies, thin films, single crystals or polycrystalline materials are significantly different. Traditionally, the sensitivity and selectivity of the sensor are generally improved by doping and improved preparation methods [13]. Doped substances include noble metals such as Pt, Pd, etc., as well as rare earth elements, lanthanide elements, etc. At present, the operating temperature of sensors is relatively high, and the most stable structure of TiO2 at high temperature is the rutile structure. Therefore, TiO2 with rutile structure is usually selected as the sensing material. However, after pure TiO2 is heat-treated to transform into rutile structure, the grain size increases significantly, which is extremely unfavorable for identifying the gas to be measured. Studies have found that introducing certain foreign atoms into TiO2, such as Cr, Nb, Ta, La, Mo[14-17], can control the grain size of TiO2 after the two-phase transformation. Comini et al.[17] studied the characteristics of Mo-doped TiO2 thin films. They used RF magnetron reactive sputtering to deposit 0.5% Mo-doped TiO2 thin films on (100) oriented single crystal silicon wafers in an atmosphere of 50% Ar + 50% O2, and analyzed the surface morphology and structure of the thin films. The test found that the crystal structure of the thin film after heat treatment at 600°C was a mixture of anatase and rutile, with an average grain size of about 31 nm. After heat treatment at 800°C, it was found that the crystal structure of the film had been completely transformed into the rutile phase structure. Mo doping has a better effect on controlling the grain size than the previously reported W doping. Metal doping such as Ta, La, Nb and Cr can effectively control the grain size during the two-phase transformation of TiO2, but as the research progressed, it was found that there were still problems: the doping of these metals hindered the transformation of anatase to rutile, which increased the heat treatment temperature of the transformation. Based on this, Carotta et al. [18] added V to TiO2 and found that V doping can not only control the grain size, but also promote the two-phase transformation. They compared Ta-doped samples and V-doped samples in their study. The results showed that the pure TiO2 sample or Ta-doped sample after calcination at 650°C had anatase structure and the grain size was 30nm to 50nm. Usually, the pure TiO2 sample after calcination at 850°C has been transformed into the rutile structure, but its grain size has increased significantly to about 400nm. When the Ta-doped sample was calcined at the same temperature, no signs of grain enlargement were found. However, the Ta-doped sample still had anatase structure after calcination at this temperature. In contrast, the V-doped sample not only did not show signs of grain enlargement, but also transformed into a rutile structure at 650℃. The study by Ruizd et al. [19] also solved this problem. They used the sol-gel method to add La and Cu simultaneously during the preparation of TiO2. The results showed that after La was added to TiO2, the two-phase transformation was delayed, and the anatase structure still existed until 800℃. However, La doping can control the grain size. The grain size of the 10% La-doped sample was 7 nm, and the grain size of the 5% La-doped sample was 11 nm. When 2% Cu was added to the La-doped sample, the two-phase transformation was accelerated, and the grain size increased, but not significantly. Samples doped with both La and Cu retain a pure anatase phase structure at 700°C, but exhibit a distinct rutile structure at 800°C. Gas-sensing performance tests show that a 5% La-doped sample annealed at 900°C has high sensitivity to CO. Further doping with 2% Cu improves the selectivity for CO. Ruizd et al.'s work provides new ideas and methods for studying the microstructure of materials to improve sensor performance. We can move beyond the traditional single-element doping method and try simultaneously incorporating two complementary elements to improve sensor performance. TiO2 is a typical n-type conductive material; its resistance decreases when exposed to reducing gases. Traditionally, this type of conductive material is used to detect reducing gases, but when exposed to certain oxidizing gases, such as NO2, the resistance of TiO2 increases, sometimes exceeding the detection limit of traditional circuits. This phenomenon is even more pronounced when TiO2 is in thin film form. Therefore, p-type conductive materials should be selected to detect NO2, because under appropriate conditions, the resistance of p-type conductive materials decreases when they encounter oxidizing gases [14]. Studies have found that by doping TiO2 with Cr and controlling the doping concentration of Cr and the heat treatment temperature of the sample, TiO2 sensing materials that are converted to p-type conductive types can be obtained. Ruizd et al. [9] conducted a systematic study on this work. They prepared Cr-doped TiO2 samples on alumina substrates using the sol-gel method. They controlled the Cr doping concentration at 5% to 30% and the heat treatment temperature of the samples at 600°C to 900°C. The XPS test results of the samples annealed at 600°C in the valence band region showed that with the increase of Cr, the electronic conductivity type tended to change from n to p. The electrical performance test found that the working temperature of this Cr-doped sample was 250°C, and the response time of the 90% Cr-doped sample to NO2 was less than 1 min. In this work, Ruizd et al. analyzed the reasons for the conversion of the material conductivity type. Based on the test results, they assumed that the change in conductivity type was caused by Cr doping. However, this is only an experimental result, and there is no definite theoretical basis to support this assumption. Therefore, further in-depth research is still needed. However, this method has opened the door for us to develop new sensors. 2. Using nanotechnology to develop new sensors Nanotechnology is a science and technology that studies the motion laws and interactions of material composition systems with sizes ranging from 0.1 nm to 100 nm, as well as technical issues in possible practical applications. In recent years, the rapid development of nanotechnology has also brought broad prospects for the advancement of sensing technology. Compared with traditional sensors, the use of nanotechnology to prepare gas sensors has advantages that conventional sensors cannot replace: First, nano solid materials have a huge interface, providing a large number of gas channels, thereby greatly improving sensitivity; second, the working temperature is greatly reduced; and third, the size of the sensor is greatly reduced. Vargese et al. [20] used the anodic oxidation method to prepare a self-assembled titanium nanotube sequence as a gas sensor for detecting hydrogen gas. The preparation method is as follows: a titanium foil with a thickness of 0.25 mm is used as the anode, a Pt electrode is used as the cathode, and it is placed in a 0.5% hydrofluoric acid solution for electrolysis to grow a titanium nanotube array. When voltages were applied to the anode at 20, 14, and 10 V, the resulting nanotubes had inner diameters of 76 nm, 53 nm, and 22 nm, and wall thicknesses of 27 nm, 17 nm, and 13 nm, respectively. At 290°C, with 1000 ppm hydrogen gas introduced, the sensitivities of the 76 nm, 53 nm, and 22 nm titanium nanotube samples were 70, 260, and 10000, respectively, indicating that the finer the titanium nanotube, the higher its gas sensitivity. The 22 nm sample showed a sensitivity 10 times higher than the previously reported highest sensitivity to hydrogen. At 290°C, as the hydrogen concentration increased from 0 to 500 ppm, the resistance of the 22 nm titanium nanotube changed by four orders of magnitude, demonstrating that this sensor not only has high sensitivity but also operates at a low temperature. Its working mechanism is that during the chemical adsorption process, hydrogen exists as a surface state near the titanium conduction band, electrons migrate from hydrogen atoms to titanium atoms, and a space charge layer with potential energy V0 is generated on the surface, which reduces the resistance of the nanotube. Wang Yuan et al. [21] prepared a TiO2/PtO-Pt double-layer nanofilm for detecting hydrogen. The preparation method is to first cover a porous film composed of Pt nanoparticles on a glass substrate, wherein the diameter of the Pt nanoparticles is about 1.3 nm and the film thickness is about 100 nm. Then, a TiO2 film is covered on the PtO-Pt film, wherein the diameter of the TiO2 nanoparticles ranges from 3.4 nm to 5.4 nm and the average diameter is 4.1 nm. Its working temperature is 180°C～200°C. The PtO-Pt porous film acts as a catalyst to partially reduce the hydrogen in the TiO2 nanofilm, so that the sensor exhibits high sensitivity and selectivity to hydrogen in the air, even in the presence of reducing gases such as CO, NH3, and CH4. In the research of nanotechnology, there are few reports on the preparation of new materials such as nanotubes and nanoribbons based on TiO2. Recently, Wang Zhonglin et al. [22] synthesized nanoribbons of metal semiconductor oxides such as zinc oxide, tin oxide, indium oxide, cadmium oxide, and gallium oxide. They synthesized them by high-temperature thermal evaporation of bulk materials and then low-temperature vapor deposition onto alumina substrates. Titanium oxide has many amazing properties. Can we choose appropriate methods to synthesize new materials such as nanoribbons to enrich the field of TiO2 gas sensors? This is not only an opportunity for us, but also a challenge. It is foreseeable that these new materials will have a bright application prospect in the field of sensors. V. Prospect of the development direction of TiO2 gas sensors With the continuous improvement of traditional research methods and the application of nanotechnology, the development of TiO2 gas sensors has made great progress, and their selectivity, sensitivity and stability have been greatly improved. However, some problems still need to be solved in the research, such as how catalysts affect the sensing mechanism; although the operating temperature of gas sensors made using traditional improvement methods and nanoparticles has been greatly reduced, it is still too high, around 200℃; and although most of the performance of nanosensors is superior to that of traditional ones, their recovery time is relatively long. With the further development of theoretical research and nanotechnology, these problems will surely be well solved, and the research and application of TiO2 gas sensors will reach a new level. References: [1] K. Zakrzewska. Mixed oxide as gas sensors[J]. Thin solid films, 2001, 391: 229-238. [2] Xu Yulong, G. Heiland. Metal oxide gas sensor (I)[J]. Journal of Sensor Technology, 1995, 4: 59-64. [3] Tian Liqiang, Pan Guofeng, Zhao Yanxiao et al. Research and development of TiO2 thin film oxygen sensing element[J]. Sensor World, 2002, 7. [4] Feng Duan, Shi Changxu, Liu Zhiguo. Introduction to Materials Science - A Coherent Discussion[M]. Beijing: Chemical Industry Press, 2002. [5] Yang Shuren, Wang Zongchang, Wang Jing. Semiconductor Materials[M]. Beijing: Science Press, 2004. [6] Xu Yulong, G. Heiland. Metal oxide gas sensor (III) [J]. Journal of Sensor Technology, 1996, 2: 56-61. [7] Zhou Wei, Sun Chengwen, Yang Zhizhou. Research and development overview of TiO2 oxygen sensing element [J]. Journal of Inorganic Materials, 1998, 13(3): 275-281. [8] Xu Yulong, G. Heiland. Metal oxide gas sensor (II) [J]. Journal of Sensor Technology, 1996, 1: 63-67. [9] Xu Yulong, G. Heiland. Metal oxide gas sensor (IV) [J]. Journal of Sensor Technology, 1996, 3: 72-78. [10] A. Ruiz, G. Sakai, A. Cornet, et al. Cr-doped TiO2 gas sensor for exhaust NO2 monitoring [J]. Sens.Actuators B, 2003, 93: 509-518. [11] Tian Jingmin, Li Shouzhi. Exploration of the gas-sensitive mechanism of metal oxide semiconductor [J]. Journal of Xi'an University of Technology, 2002, 18(2): 144-147. [12] DE Williams. Semicoducting oxides as gas-sensitive resistors [J]. Sens.Actutators B, 1999, 57: 1-16. [13] PY Liu, JF Chen and WD Sun. Characterizations of SnO2 and SnO2:Sb thin films prepared by PECVD [J]. Vacuum. 2004, 76: 7-11. [14] A. Ruiz, A.Cornet, G. Sakai, et al. Preparation of Cr-doped TiO2 thin film of p-type conduction for gas sensor application [J]. Chem.Lett, 2002, 9: 892-893. [15] A. Ruiz, D. Dezanneau, J. Arbiol, et al. Study of the influence of Nb content and sintering temperature on TiO2 sensing films[J]. Thin Solid Films, 2003, 436: 90-94 [16] N. Bonini, M. Carotta, A. Chiorino, et al. Doping of a nanostructured titania thick film: structural and electrical investigations[J]. Sensors &. Actuators B, 2000, 68: 274-280. [17] E. Comini, G. Sberveglieri, M. Ferroni, et al. Response to ethanol of thin films based on Mo and Ti oxides deposited by sputtering[J]. Sensors &. Actuators B, 2003, 93:409-415. [18] M. Carotta, M. Ferroni, S. Gherardi, et al. Thick-film gas sensors based on vanadium-titanium oxide powders prepared by sol-gel sythesis[J]. J. Eur. Ceram. Soc., 2004, 24: 1409-1413. [19] A. Ruiz, A. Cornet, J. Morante, et al. Study of La and Cu influence on the growth inhibition and phase transformation of nano- TiO2 used for gas sensors[J]. Sensors &. Actuators B, 2004, 100: 256-260. [20] O. Varghese, D. Gong, M. Paulose, et al. Extreme changes in the electrical resistance of titania nanotubes with hydrogen exposure[J]. Adv Mater. 2003, 15: 7-8. [21] X. YDu, Y. Wang, YY Mu, et al. A new highly selective H2 sensor based on TiO2/PtO-Pt dual-layer films[J]. Chem. Mater. 2002, 14: 3953-3957. [22] ZW Pan, ZR Dai, ZL Wang. Nanobelts of semiconducting oxides[J]. Science, 2001, 209:1947-1949. About the authors: Zhai Lin, Master's student at the School of Science and Engineering, Jinan University, specializing in functional thin film materials and sensor devices. Zhong Fei, Master's student at the School of Science and Engineering, Jinan University, specializing in functional thin film materials and display devices. Liu Pengyi, Professor, Master's supervisor, and Department Head at the School of Science and Engineering, Jinan University, specializing in functional thin film materials and sensing technology.