TiO2-x/TiO2-Structure Based ‘Self-Heated’ Sensor for the Determination of Some Reducing Gases

In this research we report the gas-sensing properties of TiO2-x/TiO2-based hetero-structure, which was ‘self-heated’ by current that at constant potential passed through the structure. Amperometric measurements were applied for the evaluation of sensor response towards ethanol, methanol, n-propanol and acetone gases/vapours. The sensitivity towards these gases was based on electrical resistance changes, which were determined by amperometric measurements of current at fixed voltage applied between Pt-based contacts/electrodes deposited on the TiO2-x/TiO2-based layer. X-ray diffraction (XRD) analysis revealed the formation of TiO2-x/TiO2-based hetero-structure, which is mainly based on Ti3O5/TiO2 formed during the hydro-thermal oxidation-based sensing-layer preparation process. Additionally, photoluminescence and time-resolved photoluminescence decay kinetics-based signals of this sensing structure revealed the presence of TiO2 mainly in the anatase phase in the TiO2-x/TiO2-based hetero-structure, which was formed at 400 °C annealing temperature. The evaluation of TiO2-x/TiO2-based gas-sensing layer was performed at several different temperatures (25 °C, 72 °C, 150 °C, 180 °C) and at these temperatures different sensitivity to the aforementioned gaseous materials was determined.


Introduction
Among many other inorganic semiconductor-based structures, TiO 2 -based structures are often used for the development of gas-sensing devices due to their sensing properties [1]. Titanium dioxide (TiO 2 ) is an n-type semiconductor, which exists in three main phases (i) anatase, (ii) rutile and (iii) brookite with bandgaps of 3.02, 3.23, and 2.96 eV, respectively [2]. All phases of TiO 2 can be relatively easily synthesized and/or converted from each to another by relatively simple thermal treatment (annealing).
The attractiveness of TiO 2 has significantly increased since its water-splitting ability under ultraviolet (UV) light irradiation was discovered [3]. The discovery of the latter effect facilitated the exploitation of catalytic TiO 2 properties in the design of sensors for gases and gaseous materials. Due to remarkable properties, many different TiO 2 -based structures have found applications in various technological areas including biosensors [4,5] and chemical sensors [6][7][8][9][10].
TiO 2 -based gas sensors are not expensive. They mostly require simple analytical signal detection and evaluation systems and offer good sensitivity towards many gases and gaseous materials,

Formation of TiO 2 Sample
A Si wafer (1) was oxidized in electric oven with increased concentration of oxygen to form a few micrometers thick oxide (SiO 2 ) layer (2) over the Si wafer. After that the metallic titanium (Ti) layer (3) of 100 nm thickness was sputtered by a magnetron. Amorphous non-stoichiometric titanium oxide (TiO 2-x ) and titanium dioxide (TiO 2 ) based hetero-structure (TiO 2-x /TiO 2 ) was formed by hydrothermal oxidation of Si/SiO 2 /Ti-based wafer in aqueous alkaline solution. Finally, platinum (Pt) electrodes/(contact zones) (Figure 1, Section 3) were formed on the top of the crystalline TiO 2 by magnetron sputtering. sensitivity of the here designed gas sensor depends on the temperature of TiO2-x/TiO2-based sensingstructure.

Formation of TiO2 Sample
A Si wafer (1) was oxidized in electric oven with increased concentration of oxygen to form a few micrometers thick oxide (SiO2) layer (2) over the Si wafer. After that the metallic titanium (Ti) layer (3) of 100 nm thickness was sputtered by a magnetron. Amorphous non-stoichiometric titanium oxide (TiO2-x) and titanium dioxide (TiO2) based hetero-structure (TiO2-x/TiO2) was formed by hydrothermal oxidation of Si/SiO2/Ti-based wafer in aqueous alkaline solution. Finally, platinum (Pt) electrodes/(contact zones) (Figure 1, Section 4) were formed on the top of the crystalline TiO2 by magnetron sputtering.  The next step in sensor development procedure was based on annealing of Ti-based structures, which was performed at three different temperature regimes: (i) 50 • C/1 h + 400 • C/2 h temperature for the formation of TiO 2-x /TiO 2 (400 • C) structure; (ii) 50 • C/1 h + 600 • C/2 h temperature for the formation of TiO 2-x /TiO 2 (600 • C) structure; (iii) 50 • C/1 h + 800 • C/2 h temperature for the formation of TiO 2-x /TiO 2 (800 • C) structure. All the aforementioned procedures enabled us to form mixed crystal phase TiO 2-x /TiO 2 -based hetero-structures ( Figure 1, layers 4 and 5) at over-layer containing significant amount of TiO 2 in the form of anatase and/or rutile phases and TiO 2-x , which was formed in the deeper layers of the structure below fully oxidized stoichiometric TiO 2 .
Finally, platinum-based electrodes were deposited on the formed TiO 2-x /TiO 2 (400 • C), TiO 2-x /TiO 2 (600 • C) and TiO 2-x /TiO 2 (800 • C) layers by magnetron sputtering. The dimensions of the structure were: 8 mm-length of the structure and 3 mm-distance between platinum electrodes deposited over the TiO 2-x /TiO 2 -based layer. The thickness of formed titanium layer was 100 nm, the measurements were performed at 25 • C (room temperature), 72 • C, 150 • C, 180 • C. The humidity of the supplied aforementioned gaseous materials containing air stream was constant during all parts of measurements with methanol, ethanol, n-propanol, and acetone vapour: partial pressure of water in the air stream was 3.170 kPa, which corresponds to 100% of relative humidity at 25 • C and at 101.325 kPa pressure. During the measurement of signals towards water, initially a dry air stream was supplied and mixed with 1170 ppm of H 2 O containing air stream at corresponding ratio.
Platinum (Pt) contacts were formed by magnetron sputtering using sputter from VSTSER (Tel-Aviv, Israel) in direct current (DC) regime under Argon (Ar) atmosphere (20 mTorr pressure) using 1" diameter and 99.99% purity Pt target. Contacts geometry was determined by a mask. In order to improve adhesion between TiO 2-x /TiO 2 -based layer and platinum at first a thin (20 nm) titanium (Ti) layer was sputtered on which Pt contacts were formed. Sputtering power for Ti layer formation was 2.55 W/cm 2 with the growth of 0.13 nm/s. During the formation of the Pt procedure, magnetron power was 3.06 W/cm 2 , with layer growth of 0.08 nm/s.

Scanning Electron Microscopy (SEM)-Based Characterization of Formed TiO 2-x /TiO 2 -Based Hetero-Structure
The structural properties of the obtained TiO 2 samples on silicon substrates were determined using a scanning electron microscope (SEM) Helios NanoLab 650 from FEI (Eindhoven, The Netherlands).

X-ray Diffraction (XRD) Characterization of TiO 2-x /TiO 2 -Structure
The phase composition of thin films was determined by X-ray diffractometer D8 Advanced from Bruker (USA) with grazing-incident X-ray diffraction (XRD) over a 2θ range of 22 • -65 • using Cu Kα (= 1.5046). Films for XRD investigations were based on: (i) metallic Ti (100 nm) layer deposited over Si substrate covered by 300 nm SiO 2 layer (ii) TiO 2−x /TiO 2 (400 • C) hetero-structure, which was formed by the oxidation of the same metallic Ti layer (mentioned in part 'i') according to the aforementioned protocol of hydrothermal oxidation at 400 • C.
The diffraction pattern was generated by X-ray beam at grazing incidence angle of 2 • . XRD data library 'CDD Database PDF2010-PDF-2/Release 2010 RDB' was applied for the analysis of XRD patterns.
TiO 2 -powder, which by producer was declared as TiO 2 anatase (TiO 2(anatase) ) phase, was received from Sigma-Aldrich (St. Louis, United States) and was used as XRD control while was investigated at the same grazing incidence angle of 2 • over similar 2θ range of 22 • -65 • .

Photoluminescence (PL)-Based Characterization of TiO 2-x /TiO 2 -Structure
Optical properties of TiO 2-x /TiO 2 -based hetero-structures deposited on oxidized silicon substrates were investigated by photoluminescence (PL) studies using Edinburgh-F900 spectrophotometer (Edinburg Instruments Ltd., Livingston, UK). The photoluminescence spectra of TiO 2-x /TiO 2 -based hetero-structures were excited by solid-state laser with an excitation wavelength of 375 nm (the average pulse power was about 0.15 mW/mm 2 , the pulse duration 70 ps) and photoluminescence was measured in the range of 400-700 nm. For comparison, the photoluminescence spectra of similar oxidized silicon substrates were also registered. All fluorescence spectra were corrected accounting the sensitivity of the instrument. Photoluminescence measurements of 'self-heated' structure were performed at different voltages that heated the sensor up to particular temperatures. The exact temperature was followed with a thermocouple attached to the TiO 2-x /TiO 2 hetero-structure.
The position and intensity of the photoluminescence maximum was determined as corresponding characteristics of a Gauss function using the Origin program. Measurements of current passing through the sensor structure were performed by potentiostat/galvanostat Autolab 30 Eco Chemie Gmbh (Utrecht, The Netherlands), which was controlled by NOVA software. The TiO 2−x /TiO 2 (400 • C) sample was investigated under different constant voltages, which were applied by potentiostat on platinum-based electrodes deposited on the TiO 2-x /TiO 2 (400 • C) layer.
Evaluated gas concentrations were fixed at: 105 ppm for water, 118 ppm for methanol, 53 ppm for ethanol, 18 ppm for n-propanol, 220 ppm for acetone.
Electrical resistance of TiO 2−x /TiO 2 (400 • C)-based gas-sensitive structure decreased due to their increased conductivity, therefore, current passing through TiO 2-x /TiO 2 (400 • C)-based structure increased. In present research we have registered a variation of current (∆I) at constant potential. These variations of measured current (I) ( Figure 2) were converted into resistance (R) of TiO 2-x /TiO 2 -based structure according to Ohms law: where V is applied voltage.
fluorescence spectra were corrected accounting the sensitivity of the instrument. Photoluminescence measurements of 'self-heated' structure were performed at different voltages that heated the sensor up to particular temperatures. The exact temperature was followed with a thermocouple attached to the TiO2-x/TiO2 hetero-structure.
The position and intensity of the photoluminescence maximum was determined as corresponding characteristics of a Gauss function using the Origin program.

Determination of Analytical Signal towards Reducing Gases by TiO2-x/TiO2-Structure Based Sensor
Measurements of current passing through the sensor structure were performed by potentiostat/galvanostat Autolab 30 Eco Chemie Gmbh (Utrecht, The Netherlands), which was controlled by NOVA software. The TiO2−x/TiO2 (400 °C) sample was investigated under different constant voltages, which were applied by potentiostat on platinum-based electrodes deposited on the TiO2-x/TiO2 (400 °C) layer.
Evaluated gas concentrations were fixed at: 105 ppm for water, 118 ppm for methanol, 53 ppm for ethanol, 18 ppm for n-propanol, 220 ppm for acetone.
Electrical resistance of TiO2−x/TiO2 (400 °C )-based gas-sensitive structure decreased due to their increased conductivity, therefore, current passing through TiO2-x/TiO2 (400 °C )-based structure increased. In present research we have registered a variation of current (ΔI) at constant potential. These variations of measured current (I) (Figure 2) were converted into resistance (R) of TiO2-x/TiO2based structure according to Ohms law: where V is applied voltage. Then relative response (ΔR, %) was calculated as: where Rinitial is initial resistance of TiO2-x/TiO2 (400 °C )-based structure calculated from Iinitial at baseline, Rfinal-final resistance calculated from Ifinal ( Figure 2).  Then relative response (∆R, %) was calculated as: where R initial is initial resistance of TiO 2-x /TiO 2 (400 • C)-based structure calculated from I initial at baseline, R final -final resistance calculated from I final (Figure 2).

Determination of Electrical Resistance Variation with Temperature
Measurements of resistance vs. temperature, were performed by system based on closed cycle helium cryostat made by Sumitomo Heavy Industries (Tokyo, Japan) combined with RDK-408D 4K Sensors 2020, 20, 74 6 of 17 cold head and SRDK Series cryocooler and CSA-71A compressor unit. Temperature was controlled by Lakeshore 340 temperature controller (Lake Shore Cryotronics, Inc., Westerville, OH, USA). Resistance was measured by Tektronix DMM 4050 multimeter (Tektronix UK Ltd., Bracknell, UK). This closed-cycle helium cryostat was used for measurement of resistance and was performed only after exact equilibration of temperature at each selected point in the temperature range from 4.2 K to 310 K, with 5 K intervals between measurement points. Temperature was changed in cyclic manner in two ways: black cycles shows points measured by cooling down, red squares shows points by increasing temperature. Measured performed in vacuum 10 −3 Torr.

SEM-Based Structural Characterization of TiO 2-x /TiO 2 -Based Layer
SEM images of TiO 2-x /TiO 2 (400 • C)-based hetero-structure at different at some extent s ( Figure 3A,B) illustrate that the sample has a highly porous surface with nanostructures in the form of nano-plates and nano-sponges, which significantly enhanced both (i) surface area and (ii) surface to volume ratio of the gas-sensitive area of sensor. Therefore, such formations are very beneficial in order to obtain increased surface area, which is available for gas adsorption and enhances the sensitivity of such a structure.

Determination of Electrical Resistance Variation with Temperature
Measurements of resistance vs. temperature, were performed by system based on closed cycle helium cryostat made by Sumitomo Heavy Industries (Tokyo, Japan) combined with RDK-408D 4K cold head and SRDK Series cryocooler and CSA-71A compressor unit. Temperature was controlled by Lakeshore 340 temperature controller (Lake Shore Cryotronics, Inc., Westerville, OH, USA). Resistance was measured by Tektronix DMM 4050 multimeter (Tektronix UK Ltd., Bracknell, UK). This closed-cycle helium cryostat was used for measurement of resistance and was performed only after exact equilibration of temperature at each selected point in the temperature range from 4.2 K to 310 K, with 5 K intervals between measurement points. Temperature was changed in cyclic manner in two ways: black cycles shows points measured by cooling down, red squares shows points by increasing temperature. Measured performed in vacuum 10 −3 Torr.

SEM-Based Structural Characterization of TiO2-x/TiO2-Based Layer
SEM images of TiO2-x/TiO2 (400 °C )-based hetero-structure at different at some extent s ( Figure  3A,B) illustrate that the sample has a highly porous surface with nanostructures in the form of nanoplates and nano-sponges, which significantly enhanced both (i) surface area and (ii) surface to volume ratio of the gas-sensitive area of sensor. Therefore, such formations are very beneficial in order to obtain increased surface area, which is available for gas adsorption and enhances the sensitivity of such a structure.

XRD Characterization of TiO2-x/TiO2-Structure
The XRD pattern of metallic Ti ( Figure 4, A(metallic Ti)), which was observed for initial metallic Ti layer of 100 nm thickness that was formed by magnetron sputtering and was used for further hydrothermal oxidation at 400 °C into a TiO2-x/TiO2 (400 °C )-based hetero-structure, represents all characteristic peaks of metallic Ti well (according to PDF '00-044-1294 for metallic Ti').
The XRD pattern of TiO2-x/TiO2 (400 °C) ( Figure 4, B(TiO2−x/TiO2-structure)) represents relatively high dispersion in TiO2-x/TiO2 (400 °C)-based hetero-structure, but according to Fukushima et al. [32] even the presence of broad 'peak area' in titanium oxide based pattern between 27° and 37° can be assigned to the presence of Ti3O5, Ti4O7 and/or Ti8O15. Thus, in our research we have observed a much better expressed XRD peak between 34° and 37° ( Figure 4, Inset, area 1), which is in good agreement with γ-Ti3O5 reported in the PDF '00-040-0806 for γ-Ti3O5' and also in relatively good agreement with the XRD pattern presented by Yoshimatsu et al. [33] for γ-Ti3O5 formed by pulsed laser deposition. Results obtained by Yoshimatsu et al. [33] showed low-temperature superconductivity in both Ti4O7

XRD Characterization of TiO 2-x /TiO 2 -Structure
The XRD pattern of metallic Ti ( Figure 4, A (metallic Ti) ), which was observed for initial metallic Ti layer of 100 nm thickness that was formed by magnetron sputtering and was used for further hydrothermal oxidation at 400 • C into a TiO 2-x /TiO 2 (400 • C)-based hetero-structure, represents all characteristic peaks of metallic Ti well (according to PDF '00-044-1294 for metallic Ti'). different, therefore, the composition and stoichiometry of both TiO2-x-based structures discussed here can be different at some extent. Another broad peak area (Figure 4, Inset, area 2) between 36° and 39° is assigned to TiO2(anatase) according to PDF '01-075-2545 for TiO2(anatase)' and control XRD pattern ( Figure 4, C(TiO2 powder)) registered at the same experimental conditions for TiO2(anatase) powder purchased from Sigma Aldrich, where according to match between XRD patterns (Figure 4. C(TiO2 powder)) with PDF '01-089-0552 for TiO2(rutile)' we have detected the presence of some TiO2(rutile) phase. . X-ray diffraction (XRD) patterns of: A(metallic Ti)-metallic Ti layer of 100 nm thickness, which was formed by magnetron sputtering; B(TiO2-x/TiO2-structure)-TiO2-x/TiO2 (400 °C)-based hetero-structure, which was formed from above mentioned metallic 100 nm thick Ti layer; C(TiO2 powder)-TiO2-powder, which was used as 'control sample' and by supplier (Sigma-Aldrich) was declared as 99.3% TiO2 in the anatase phase.
Titanium pentoxide (Ti3O5) with polymorphisms (α-, β-, γ-, δ-, and λ-phases) is 'a close neighbour' of the Magnéli phase [34][35][36][37][38], and sometimes it is designated as the first member of the Magnéli phase, because their chemical formula is consistent with that of the Magnéli phase (TinO2n-1 at n = 3). According to PDF '00-040-0806 for γ-Ti3O5′, Ti3O5 possesses a monoclinic cell (a = 9.9701 Å , b = 5.0747 Å, c = 7.1810 Å, β = 109.865°), which is superconducting at low temperatures (below 3 K) similarly to Ti4O7-based Magnéli phase [33]. Differently from the 'most respectful' member of Magneli phase-Ti4O7, which in crystal structure is having TiO2(rutile)-based shear planes [39,40], in Ti3O5 there are no such TiO2-rutile based shear planes [33]. Therefore, in our XRD patterns we are observing only signs of M3O5 and TiO2(anatase) without any presence of rutile. Such a composition was formed because for the formation of TiO2-x/TiO2-based hetero-structures we have applied 400 °C temperature at which the formation of M3O5-TiO2(anatase) 'intergrowths' is observed, as it has been reported and investigated by other research teams [41] and very recently has been confirmed by some other research group in different conditions [42]. The formation of above mentioned M3O5-TiO2(anatase) 'intergrowths' is in well agreement with our results based on photoluminescence ( Figure 5A) and photoluminescence decay ( Figure 5D) measurements, which are discussed in the next chapter, where we are clearly observing effects induced by the presence of TiO2(anatase). . X-ray diffraction (XRD) patterns of: A (metallic Ti) -metallic Ti layer of 100 nm thickness, which was formed by magnetron sputtering; B (TiO2-x/TiO2-structure) -TiO 2-x /TiO 2 (400 • C)-based hetero-structure, which was formed from above mentioned metallic 100 nm thick Ti layer; C (TiO2 powder) -TiO 2 -powder, which was used as 'control sample' and by supplier (Sigma-Aldrich) was declared as 99.3% TiO 2 in the anatase phase.
The XRD pattern of TiO 2-x /TiO 2 (400 • C) (Figure 4, B (TiO2−x/TiO2-structure) ) represents relatively high dispersion in TiO 2-x /TiO 2 (400 • C)-based hetero-structure, but according to Fukushima et al. [32] even the presence of broad 'peak area' in titanium oxide based pattern between 27 • and 37 • can be assigned to the presence of Ti 3 O 5 , Ti 4 O 7 and/or Ti 8 O 15 . Thus, in our research we have observed a much better expressed XRD peak between 34 • and 37 • (Figure 4, Inset, area 1), which is in good agreement with γ-Ti 3 O 5 reported in the PDF '00-040-0806 for γ-Ti 3 O 5 ' and also in relatively good agreement with the XRD pattern presented by Yoshimatsu et al. [33] for γ-Ti 3 O 5 formed by pulsed laser deposition. Results obtained by Yoshimatsu et al. [33] showed low-temperature superconductivity in both Ti 4 O 7 and γ-Ti 3 O 5 films. In addition, these authors reported relatively broad XRD peaks at 36-38 • for γ-Ti 3 O 5 and at 42-43 • for Ti 4 O 7 . The peak observed in our research ( Figure 4, B (TiO2−x/TiO2-structure) and Inset, area -1) according to the shape-like-features and signal to noise ratio is very similar to that at 36-38 • for γ-Ti 3 O 5 , but just some shift is observed because preparation procedures of both films were very different, therefore, the composition and stoichiometry of both TiO 2-x -based structures discussed here can be different at some extent. Another broad peak area (Figure 4, Inset, area 2) between 36 • and 39 • is assigned to TiO 2(anatase) according to PDF '01-075-2545 for TiO 2(anatase) ' and control XRD pattern (Figure 4, C (TiO2 powder) ) registered at the same experimental conditions for TiO 2(anatase) powder purchased from Sigma Aldrich, where according to match between XRD patterns (Figure 4. C (TiO2 powder) ) with PDF '01-089-0552 for TiO 2(rutile) ' we have detected the presence of some TiO 2(rutile) phase.
Titanium pentoxide (Ti 3 O 5 ) with polymorphisms (α-, β-, γ-, δ-, and λ-phases) is 'a close neighbour' of the Magnéli phase [34][35][36][37][38], and sometimes it is designated as the first member of the Magnéli phase, because their chemical formula is consistent with that of the Magnéli phase (TinO 2n-1 at n = 3). According to PDF '00-040-0806 for γ-Ti 3 O 5 , Ti 3 O 5 possesses a monoclinic cell (a = 9.9701 Å, b = 5.0747 Å, c = 7.1810 Å, β = 109.865 • ), which is superconducting at low temperatures (below 3 K) similarly to Ti 4 O 7 -based Magnéli phase [33]. Differently from the 'most respectful' member of Magneli phase-Ti 4 O 7 , which in crystal structure is having TiO 2(rutile) -based shear planes [39,40] are no such TiO 2 -rutile based shear planes [33]. Therefore, in our XRD patterns we are observing only signs of M 3 O 5 and TiO 2(anatase) without any presence of rutile. Such a composition was formed because for the formation of TiO 2-x /TiO 2 -based hetero-structures we have applied 400 • C temperature at which the formation of M 3 O 5 -TiO 2(anatase) 'intergrowths' is observed, as it has been reported and investigated by other research teams [41] and very recently has been confirmed by some other research group in different conditions [42]. The formation of above mentioned M 3 O 5 -TiO 2(anatase) 'intergrowths' is in well agreement with our results based on photoluminescence ( Figure 5A) and photoluminescence decay ( Figure 5D) measurements, which are discussed in the next chapter, where we are clearly observing effects induced by the presence of TiO 2(anatase) .

Photoluminescence Properties of Hybrid TiO 2-x /TiO 2 -Based Structures
All photoluminescence spectra of the evaluated TiO 2-x /TiO 2 -based structures were characterized by a wide photoluminescence maximum in the region of wavelengths between 415-500 nm. However, the quality of the photoluminescence signals of all three TiO 2-x /TiO 2 -based structures were very different. As it is demonstrated in the Figure 5, this depends on the temperature that is applied for the annealing of TiO 2-x /TiO 2 -based structures, because at different temperatures different phases of TiO 2 on the surface of TiO 2-x /TiO 2 -based structures were formed. TiO 2-x /TiO 2 (400 • C) structure, which was formed by annealing at 400 • C, generated the most intense photoluminescence signal ( Figure 5A). The upper layer generates the strongest photoluminescence signal and the spectrum of TiO 2-x /TiO 2 (400 • C) is characterized by photoluminescence peaks, which are observed at 415, 440, 470 nm of the main photoluminescence band ( Figure 5A). These photoluminescence peaks reveal the presence of TiO 2(anatase) in TiO 2-x /TiO 2 (400 • C) hybrid-structure. TiO 2-x /TiO 2 (600 • C) structure, which was formed by annealing at 600 • C, also has demonstrated some photoluminescent properties but the photoluminescence signal was about 10 times lower ( Figure 5B) and revealed presence of mixed TiO 2(anatase) and TiO 2(rutile) structures of TiO 2-x /TiO 2 (600 • C) sample.
TiO 2-x /TiO 2 (800 • C) structure, which was formed by annealing at 800 • C, was characterized by very weak, about 100 times lower photoluminescence signal, than that was observed for TiO 2-x /TiO 2 (400 • C) sample ( Figure 5C). Thus TiO 2-x TiO 2 (800 • C) structure is not suitable for further investigations required for optoelectronic sensors and, as revealed by the results below, this structure was also not suitable for the sensing of gases selected for this research.
TiO 2-x /TiO 2 (400 • C) structure showed the most interesting and the highest quality photoluminescence signal. Therefore, during the next experiment, which was also based on photoluminescence spectrum registration, this structure was investigated under different constant voltages, which were applied by potentiostat on platinum-based electrodes deposited on the TiO 2-x /TiO 2 (400 • C)-based layer. The plots of photoluminescence spectra ( Figure 6A) vs. applied voltage are shown in Figure 6B. The elevated voltage from 1 V to 5 V, results the decrease of intensity of the main photoluminescence maximum and a small shift of photoluminescence maximum position towards shorter wavelengths.
voltage are shown in Figure 6B. The elevated voltage from 1 V to 5 V, results the decrease of intensity of the main photoluminescence maximum and a small shift of photoluminescence maximum position towards shorter wavelengths.   Figure 6B) of the photoluminescence maximum was determined as a maximum of Gauss function, which represented the best fitting with registered photoluminescence spectra. At voltage values (from 1 V to 5 V) tested in this experiment the spectral position of the photoluminescence maximum (λmax) gradually shifts towards the infrared (IR) region ( Figure 6B). In general, the difference between the positions λmax for TiO2-x/TiO2 (400 °C )-based sample at 0 V and 5 V is about 5 nm. A similar trend related to the decrease in photoluminescence intensity of the main maximum with increasing voltage was observed ( Figure 6C). The decrease of photoluminescence intensity by the increase of applied voltage from 0 V to 5 V has been determined, and it shows that the concentration of photoluminescence emitting centers in TiO2-x/TiO2 (400 °C ) hetero-structure is reducing by increasing voltage. The plot of temperature vs. voltage of TiO2-x/TiO2 (400 °C )-based sample ( Figure 6D)  Figure 6B,C show the position and the intensity of the photoluminescence maximum, which is changing depending on the voltage applied to TiO 2-x /TiO 2 (400 • C)-based sample. The position ( Figure 6B) of the photoluminescence maximum was determined as a maximum of Gauss function, which represented the best fitting with registered photoluminescence spectra. At voltage values (from 1 V to 5 V) tested in this experiment the spectral position of the photoluminescence maximum (λ max ) gradually shifts towards the infrared (IR) region ( Figure 6B). In general, the difference between the positions λ max for TiO 2-x /TiO 2 (400 • C)-based sample at 0 V and 5 V is about 5 nm. A similar trend related to the decrease in photoluminescence intensity of the main maximum with increasing voltage was observed ( Figure 6C). The decrease of photoluminescence intensity by the increase of applied voltage from 0 V to 5 V has been determined, and it shows that the concentration of photoluminescence emitting centers in TiO 2-x /TiO 2 (400 • C) hetero-structure is reducing by increasing voltage. The plot of temperature vs. voltage of TiO 2-x /TiO 2 (400 • C)-based sample ( Figure 6D) illustrates dependence in tested potential interval, which can be predicted from Ohms law, followed by basic recalculations into the heat released by this system.
Such behavior of the photoluminescence spectra vs. applied voltage can be caused by an increased sample temperature, which is dependent on electrical current flowing through the TiO 2-x /TiO 2 (400 • C)-based hetero-structure. Thus, a rapid decrease in the photoluminescence intensity of TiO 2 with an increase in voltage at U ≥6 V is related to the increase of the sample temperature, which leads to the temperature-based quenching of photoluminescence of TiO 2(anatase) [43][44][45]. Photoluminescence studies of nanostructured TiO 2 at room-temperature were reported in some other research: (i) the photoluminescence of TiO 2(anatase) colloidal particles of different sizes occurs from the shallow trap levels, which are located between 0.41 and 0.64 eV below the conduction band [46]; (ii) the narrow photoluminescence emission bands of TiO 2(anatase) powder, which originated from the self-trapped exciton emission (STE) in crystalline TiO 2(anatase) containing TiO 6 octahedral sheet-based structures, was reported [47,48]. In addition to registration of photoluminescence spectra, we have measured the photoluminescence decay kinetics for TiO 2-x /TiO 2 (400 • C)-based hetero-structure at different narrow photoluminescence emission bands ( Figure 5D). The photoluminescence of TiO 2(anatase) decay non-exponentially with dominating fast component, which is characterized by a decay of about 0.6 ns. The photoluminescence decay kinetics presented here clearly verify the self-trapped exciton emission origin of photoluminescence, which from the broad range of non-stoichiometric titanium oxides (TiO 2−x ) and stoichiometric titanium oxides (TiO 2 ) is the most characteristic for crystalline TiO 2(anatase) [49].

Electrical Resistance Variation with Temperature
Electrical resistance measurements of TiO 2-x /TiO 2 -based structure revealed the oxidation of metallic titanium-based layer, because by thermal oxidation the resistance of the sample has increased from 0 Ω (for bare Ti-based layer) up to 72 Ω (for formed TiO 2-x /TiO 2 -based structure). The increase of resistance together with the shift of photoluminescence signals clearly shows the formation of TiO 2 in the phase of anatase and rutile. However, stoichiometric TiO 2 -based layers have relatively low electrical conductivity, which is typically of 10 −10 S/m, but it was demonstrated that the conductivity of TiO 2 -based layers can be significantly increased by heat-based treatment at a high temperature in a reducing gas-based environment [50]. Further investigations revealed the formation of TiO 2-x /TiO 2 -based hetero-structures, because only TiO 2 based structures have relatively high band-gap and, therefore, they do not conduct well while the presence of TiO 2-x in forms of Ti 2 O 3 , Ti 3 O 5 and/or Ti 4 O 7 significantly increases the conductivity of such hetero-structures. It should be noted, that an important issue related to TiO 2-x conductivity is that these oxides at the stoichiometry of Ti n O 2n-1 (with 3 < n < 10) are forming so called Magnéli phases [51], which possess some properties of metallic conductor [52,53]. As seen from resistance vs. temperature dependence (Figure 7, area 1), in the TiO 2-x /TiO 2 -based hetero-structures at temperatures below 150-180 K reported here, we have also determined some signs of such transition into metallic conductivity, which is characteristic of presence of Magnéli phases. and/or Ti4O7 significantly increases the conductivity of such hetero-structures. It should be noted, that an important issue related to TiO2-x conductivity is that these oxides at the stoichiometry of TinO2n-1 (with 3 < n < 10) are forming so called Magnéli phases [51], which possess some properties of metallic conductor [52,53]. As seen from resistance vs. temperature dependence (Figure 7, area 1), in the TiO2-x/TiO2-based hetero-structures at temperatures below 150-180 K reported here, we have also determined some signs of such transition into metallic conductivity, which is characteristic of presence of Magnéli phases.

Figure 7.
Temperature dependence of electrical resistance (R(T)) for the TiO2-x/TiO2 (400 °C )-based hetero-structure. Temperature was changed in two ways (indicated by black and red arrows): (i) black cycles shows points measured by cooling down, (ii) red squares shows points by increasing temperature. Measured was performed in vacuum using helium cryostat.
The R(T) dependence (temperature dependence of electrical resistance, Figure 7) is roughly temperature independent from 310 K down to relatively low temperatures 180-150. Such an R(T) dependence (almost constant R(T) with respect to T with some increase of conductivity when The R(T) dependence (temperature dependence of electrical resistance, Figure 7) is roughly temperature independent from 310 K down to relatively low temperatures 180-150. Such an R(T) dependence (almost constant R(T) with respect to T with some increase of conductivity when temperature is decreasing) is expected for a heterogeneous mixture of strongly disordered metal/(metal oxide)-based layer, and proves that the TiO 2-x /TiO 2 (400 • C)-based hetero-structure is strongly disordered and composed of many randomly oriented nanocrystals and of many different TiO 2-x phases and probably even some metal clusters randomly distributed in the volume of TiO 2-x /TiO 2 -based hetero-structures. It is also remarkable that the roughly constant R(T) dependence also exhibits a series of small resistance jumps in the range of a few percent. These resistance jumps might be induced by the metal-semiconductor transitions within the involved Magneli phases, which is actually a combination of a variety of different Ti n O 2n-1 phases rather than a single phase. Interestingly at 40 K the increase of electrical resistance is observed (Figure 7, area 2), similar increase, which was followed by the drop of conductivity below 4 K, observed by some other authors [33], which have reported superconductivity in Ti 4 O 7 and γ-Ti 3 O 5 films. Some earlier research illustrated that so called Magnéli phases can be observed in TiO 2-x -based layers [54] as planes based on Ti n O 2n-1 that are penetrating through a matrix of TiO 2 , and therefore, this shear plane based on Ti n O 2n-1 structure can conduct relatively well [52,53]. There are some indications that the conductivity of properly doped or reduced TiO 2 , which partly turns into Ti n O 2n-1 , at low temperatures is based on n-type conductivity along the aforementioned shear planes. Similar conductivity features are exploited in memristor-type devices based on TiO 2 , in which electrical resistance is changed by the oxidation/reduction of the TiO 2 -based layer by applied corresponding potentials [55]. In other research, it has been reported that the Ti 3+ -containing TiO 2-x /TiO 2 , has localized oxygen vacancies, which are beneficial for the electron mobility in n-type semiconducting TiO 2-x /TiO 2 structure [56]. We predict that the aforementioned oxygen vacancies are offering advanced gas-sensing ability for TiO 2-x /TiO 2 hetero-structure evaluated in our research. The reduction of TiO 2 -based layers leads to the formation of non-stoichiometric titanium oxides, which are represented by the general formula Ti n O 2n-1 (where n is in the range between 3 and 10 (3 < n < 10) and are known as Magneli phases [51]. In non-stoichiometric titanium dioxide (TiO 2-x ) with a low x (0 < x < 0.10), the dominating point defects in the structure are based mainly on Ti 3+ and Ti 4+ interstitials and on oxygen vacancies [57]. However, the Magneli phases in which x is in the range between 0.10 and 0.34 (0.10 < x < 0.34) extended planar defects and crystallographic shear planes, which are varying according to the oxygen deficiency are observed [52,58]. TiO 2-x /TiO 2 -based hetero-structure, which due to the formation of Ti 3+ has Ti n O 2n-1 doped TiO 2-x clusters ( Figure 1B, 3rd layer) with significantly advanced electrical conductivity, can be synthesized by several different methods: plasma treatment [59], metallic zinc-based reduction [60], high-energy particle bombardment [61], laser irradiation [62] and some reactions at higher temperatures [63]. In addition to these methods, in recent research we have demonstrated that the hydro-thermal approach applied here is also suitable for the formation of TiO 2-x /TiO 2 -based hetero-structures from initially deposited titanium-based layer. Due to good electrical conductivity and chemical stability, the aforementioned Magneli phases are applied in variety of applications, e.g., cathodic protection, batteries, catalyst support for fuel cells, waste and contaminated water treatment [50,64,65]. However, the majority of Magneli phases based research has been related to the fabrication of powders [19,20], only few attempts to form Magneli phase based fibers of~250 µm with tenable conductivity have been reported [54]. By contrast with those research projects, we have produced thin layer of TiO 2-x -based Magneli phases by oxidation of metallic titanium-based layer, which was deposited by magnetron sputtering. However, due to polymorphism of Ti 3 O 5 based difficulties the growth of a single Ti 3 O 5 crystal is still very challenging, therefore physical properties of Ti 3 O 5 are still under debate [33]. Only several studies have dealt with the structural phase transitions accompanying metal-insulator transition of Ti 3 O 5 , which were induced: (i) by irradiation with visible-light pulses for β ↔ λ transition [37], (ii) at 450 K temperatures for α ↔ β transition [34] and at 240 K temperatures for δ ↔ γ transition [35][36][37][38]. In addition, metal-insulator transition around 350 K was reported by Yoshimatsu et al. [33]. Such a temperature region (240-350-450 K), where the most significant variation of Ti 3 O 5 conductivity was observed, is in good agreement with our recent research, because we clearly observed the variation of conductivity based on the presence/absence of gaseous compounds in temperature region between room temperature (298 K) and 180 • C (453 K).

Gas Sensing by TiO 2-x /TiO 2 (400 • C) Hetero-Structure-Based Sensor
Changes of electrical current passing through TiO 2-x /TiO 2 -based hetero-structure at fixed potential were evaluated and recalculated into changes of resistance (∆R %) using Equations (1) and (2). ∆R was evaluated as analytical signal of this sensor. Results presented in Figure 8 illustrate that already at room temperature (25 • C) the sensor shows some sensitivity towards humidity and to all four gaseous materials (methanol, ethanol, n-propanol, acetone) evaluated here. But the signals determined amperometrically at constant 0.5 V potential, which was required for the achievement of 25 • C, temperature towards humidity and towards all the aforementioned gases were very different. In the presence of water, the resistance decreased, while in the presence of methanol, ethanol, n-propanol and acetone the resistance of the sensor increased. It is observed due to different nature of these compounds. On the surface of TiO 2-x /TiO 2 -based structure adsorbed water tends to fill the boundaries between TiO 2-x and TiO 2 grains and therefore it enhances the conductivity of TiO 2-x /TiO 2 -based hetero-structure, while all other materials have much lower conductivity in comparison to species, which they are replacing during physical-and/or chemical-sorption of methanol, ethanol, n-propanol, acetone. Therefore, the conductivity of the TiO 2-x /TiO 2 -based structure at some extent decreases (in the range of 1% in comparison to initial conductivity of TiO 2-x /TiO 2 -based structure). There are clear indications that sensitivity towards humidity tends to decrease by elevation of temperature from 72 • C to 180 • C, because at higher temperatures H 2 O evaporates from the surface of TiO 2-x /TiO 2 -based hetero-structure. A similar tendency at 180 • C was observed for methanol and ethanol, but for n-propanol and acetone the highest sensitivity was observed at the highest evaluated temperature of 180 • C, while the best sensitivity towards methanol and ethanol was determined at 150 • C and the best selectivity towards methanol was achieved at 72 • C.
Different sensitivities towards various gaseous materials at different temperatures opens the avenue to apply the array based on similar TiO 2-x /TiO 2 -based hetero-structures, where between Pt electrodes different constant potential will be applied and this will heat these structures up to different temperatures, where TiO 2-x /TiO 2 -based structures will have different sensitivities. Therefore, the read-out signals from such arrays can be evaluated by analysis of variance (ANOVA)-based approaches and interpreted as analytical signals. With applied 'self-heating' of the sensor, the best selectivity towards methanol was achieved at 72 • C, and for ethanol the temperature was 150-180 • C. For n-propanol and acetone signals increase with heating and reaches its maximum at 180 • C.
Thus TiO 2 thin films are the best for methanol detection at 72 • C, for ethanol at 150 • C and for acetone at 180 • C as the response to other gases are significantly lower. There are some indications that to all other gaseous materials investigated here, TiO 2-x /TiO 2 -based hetero-structures were the most sensitive at even higher temperatures, but these temperatures were not available due to limitations of our experimental set up.
four gaseous materials (methanol, ethanol, n-propanol, acetone) evaluated here. But the signals determined amperometrically at constant 0.5 V potential, which was required for the achievement of 25 °C , temperature towards humidity and towards all the aforementioned gases were very different. In the presence of water, the resistance decreased, while in the presence of methanol, ethanol, npropanol and acetone the resistance of the sensor increased. It is observed due to different nature of these compounds. On the surface of TiO2-x/TiO2-based structure adsorbed water tends to fill the boundaries between TiO2-x and TiO2 grains and therefore it enhances the conductivity of TiO2-x/TiO2based hetero-structure, while all other materials have much lower conductivity in comparison to species, which they are replacing during physical-and/or chemical-sorption of methanol, ethanol, npropanol, acetone. Therefore, the conductivity of the TiO2-x/TiO2-based structure at some extent decreases (in the range of 1% in comparison to initial conductivity of TiO2-x/TiO2-based structure). There are clear indications that sensitivity towards humidity tends to decrease by elevation of temperature from 72 °C to 180 °C , because at higher temperatures H2O evaporates from the surface of TiO2-x/TiO2-based hetero-structure. A similar tendency at 180 °C was observed for methanol and ethanol, but for n-propanol and acetone the highest sensitivity was observed at the highest evaluated temperature of 180 °C , while the best sensitivity towards methanol and ethanol was determined at 150 °C and the best selectivity towards methanol was achieved at 72 °C . Different sensitivities towards various gaseous materials at different temperatures opens the avenue to apply the array based on similar TiO2-x/TiO2-based hetero-structures, where between Pt electrodes different constant potential will be applied and this will heat these structures up to different temperatures, where TiO2-x/TiO2-based structures will have different sensitivities. Therefore, the read-out signals from such arrays can be evaluated by analysis of variance (ANOVA)-based approaches and interpreted as analytical signals. With applied 'self-heating' of the sensor, the best selectivity towards methanol was achieved at 72 °C , and for ethanol the temperature was 150-180 °C . For n-propanol and acetone signals increase with heating and reaches its maximum at 180 °C . Thus TiO2 thin films are the best for methanol detection at 72 °C , for ethanol at 150 °C and for acetone at 180 °C as the response to other gases are significantly lower. There are some indications that to all other gaseous materials investigated here, TiO2-x/TiO2-based hetero-structures were the most sensitive at even higher temperatures, but these temperatures were not available due to limitations of our experimental set up.  It should be noted that for some volatile organic compounds evaluated here the response of sensing layer was even slower (e.g., towards the highest n-propanol concentration it exceeded 35 s). This fact can be related to: (i) some inertia of gas supply system and (ii) the dimensions of analyte-molecule larger molecules needs some more time to access deeper layers of TiO 2-x /TiO 2 (400 • C)-based hetero-structure.  (Figure 9, lines 3 and 4), while at 72 °C it was 5.3 ppm and at room temperature (25 °C ) it was 10.6 ppm, due to relatively low response in comparison to background noise. The coefficient of variation calculated from 12 measurements at 150 °C towards 53 ppm of ethanol, was 8.8%. During the measurements of different ethanol vapour concentrations, steady-state current was achieved within 4-20 s, dependently on ethanol concentration (4 s for the lowest measured ethanol concentration and 20 s for the highest measured ethanol concertation). It should be noted that for some volatile organic compounds evaluated here the response of sensing layer was even slower (e.g., towards the highest n-propanol concentration it exceeded 35 s). This fact can be related to: (i) some inertia of gas supply system and (ii) the dimensions of analyte-molecule larger molecules needs some more time to access deeper layers of TiO2-x/TiO2 (400 °C )-based hetero-structure.

Conclusions and Future Developments
In this research we have succeeded in forming TiO2-x/TiO2-based hetero-structure from a thin metallic titanium-based layer. This structure as a sensing layer was integrated into a gas sensor suitable for the determination of some reducing gases. Relatively high conductivity of the TiO2-x/TiO2based hetero-structure was exploited for 'self-heating' of this sensor. Such 'self-heating' is very beneficial for the development of TiO2-x/TiO2-based sensors, because only at higher temperatures (from 72 °C to 180 °C ) was advanced sensitivity of TiO2-x/TiO2-based sensor towards methanol,

Conclusions and Future Developments
In this research we have succeeded in forming TiO 2-x /TiO 2 -based hetero-structure from a thin metallic titanium-based layer. This structure as a sensing layer was integrated into a gas sensor suitable for the determination of some reducing gases. Relatively high conductivity of the TiO 2-x /TiO 2 -based hetero-structure was exploited for 'self-heating' of this sensor. Such 'self-heating' is very beneficial for the development of TiO 2-x /TiO 2 -based sensors, because only at higher temperatures (from 72 • C to 180 • C) was advanced sensitivity of TiO 2-x /TiO 2 -based sensor towards methanol, ethanol, n-propanol and acetone achieved. Analytical performance of the sensor proposed here can be adjusted by the optimization of TiO 2-x /TiO 2 -based sensing structure and analytical signal registration approach, e.g., different methods of analytical signal registration (e.g., various potentiostatic, potentiodynamic and galvanostatic methods), which at the same time will serve as heating protocols, can be also applied in order to change the sensitivity/selectivity of here proposed TiO 2-x /TiO 2 -based hetero-structure. Therefore, after some further investigations it can be applied to the design of sensors with different selectivity and sensitivity.
Despite some progress in the preparation of TiO 2 -based materials, practical exploitation of TiO 2 -based structures self-doped by Ti n O 2n-1 -based structures is still very limited, because according to the best of our knowledge such hetero-structures have still been used for a limited number of practical applications. We expect that the aforementioned titanium oxide TiO 2-x /TiO 2 -based hetero-structures can find more advanced applications in sensing devices, catalysts, electrode materials, energy storage devices, etc. However, better understanding of the structures formed by the procedure proposed here and their sensing ability, which will meet the requirements of other specific applications, is still required. In addition, more comprehensive understanding of the transition from the conducting metallic titanium-based layer towards the semiconducting TiO 2-x /TiO 2 -based structure is still crucial in order to achieve desirable semiconducting properties, which are optimal for the applications mentioned above. Therefore, these challenges will be on the agenda of our further studies related to the development of TiO 2-x /TiO 2 -structure-based materials and the application of these structures for practical purposes.