Sub-ppm Formaldehyde Detection by n-n TiO2@SnO2 Nanocomposites

Formaldehyde (HCHO) is an important indicator of indoor air quality and one of the markers for detecting lung cancer. Both medical and air quality applications require the detection of formaldehyde in the sub-ppm range. Nanocomposites SnO2/TiO2 are promising candidates for HCHO detection, both in dark conditions and under UV illumination. Nanocomposites TiO2@SnO2 were synthesized by ALD method using nanocrystalline SnO2 powder as a substrate for TiO2 layer growth. The microstructure and composition of the samples were characterized by ICP-MS, TEM, XRD and Raman spectroscopy methods. The active surface sites were investigated using FTIR and TPR-H2 methods. The mechanism of formaldehyde oxidation on the surface of semiconductor oxides was studied by in situ DRIFTS method. The sensor properties of nanocrystalline SnO2 and TiO2@SnO2 nanocomposites toward formaldehyde (0.06–0.6 ppm) were studied by in situ electrical conductivity measurements in dark conditions and under periodic UV illumination at 50–300 °C. Nanocomposites TiO2@SnO2 exhibit a higher sensor signal than SnO2 and a decrease in the optimal measurement temperature by 50 °C. This result is explained based on the model considering the formation of n-n heterocontact at the SnO2/TiO2 interface. UV illumination leads to a decrease in sensor response compared with that obtained in dark conditions because of the photodesorption of oxygen involved in the oxidation of formaldehyde.


Introduction
Formaldehyde, HCHO, a colorless gas with an unpleasant odor, is a toxic compound that causes, in trace concentrations of 0.1-0.5 mg/m 3 , serious diseases of the respiratory tract, gastrointestinal tract and eyes. Biochemical oxidation of HCHO in human body occurs with the formation of carbon dioxide and formic acid, which, with prolonged exposure, causes asthma, pulmonary edema, and cancer. Formaldehyde is widely used in the manufacturing of polymeric materials for flooring, furniture, heat and electrical insulation, artificial tissues, plastic windows, etc. In addition, a concentrated HCHO solution (formalin) is used in medicine for disinfection, and in the food industry for the preservation of fruits and vegetables. Formaldehyde is also included in some cosmetics and personal care products. A detailed description of the characteristics of formaldehyde and its effects on health can be found in Ref. [1]. According to the World Health Organization, the maximum permissible concentration of cover the electrodes. Thick films were sintered at 300 °C for 10 hours in air to remove the organic binder. The method used allows us to obtain the coatings which are continuous and uniform over the entire substrate with the thickness about 1 μm [30] and with the value of resistance in air at 300 °C, differing by no more than 10%. The current-voltage (I-V) characteristics of the sensors measured using Potentiostat P-8-NANO (Elins, Zelenograd, Russia) are shown in Figure 1c. All the samples exhibited linear I-V curves both for positive and negative applied bias voltages up to +2V and −2V, The schematic illustration of sensor measurements setup is shown in Figure 2. The DC conductivity was measured in situ using electronic module (10) providing control of sensor heating and high-precision measurement of the resistance of the sensitive layer. The sensors were placed into a Teflon airtight and light-tight flow chamber (3) connected to a computer-controlled (5) gas delivery system with electronic mass-flow controllers (8). The measurements were carried out under a controlled constant flux of 100 ml/min in the temperature range of 300-50 °C with a step of 50 °C, in the dark conditions and under periodic UV illumination. Miniature UV LED (4) (10 mW/cm 2 , λmax = 365 nm) located at a distance of 4 cm above the sensors was used as an illumination source. The power of the LED was effectuated using DC power supply (1). The illumination of the sensors was carried out in a pulsed mode with the period 2 min "on"-2 min "off". An automatic cyclic relay (2) periodically opened and closed the circuit, allowing measurements to be made with periodic illumination. The sensor properties of SnO2 reference sample and TiO2@SnO2 nanocomposites were investigated toward formaldehyde HCHO (0.06-0.08-0.15-0.3-0.6 ppm) in air (relative humidity at 25 °C RH25 = 30%). The gas mixtures containing preassigned concentration of HCHO were prepared by dilution of certified gas mixture (6) with background purified air (7). Flow through humidifier (9) was used to adjust the humidity (RH25 = 30%) of gases passing through the sensor chamber. The sensor response was calculated as = / , where is the resistance in background air, is the resistance in the presence of HCHO. The schematic illustration of sensor measurements setup is shown in Figure 2. The DC conductivity was measured in situ using electronic module (10) providing control of sensor heating and high-precision measurement of the resistance of the sensitive layer. The sensors were placed into a Teflon airtight and light-tight flow chamber (3) connected to a computer-controlled (5) gas delivery system with electronic mass-flow controllers (8). The measurements were carried out under a controlled constant flux of 100 mL/min in the temperature range of 300-50 • C with a step of 50 • C, in the dark conditions and under periodic UV illumination. Miniature UV LED (4) (10 mW/cm 2 , λ max = 365 nm) located at a distance of 4 cm above the sensors was used as an illumination source. The power of the LED was effectuated using DC power supply (1). The illumination of the sensors was carried out in a pulsed mode with the period 2 min "on"-2 min "off". An automatic cyclic relay (2) periodically opened and closed the circuit, allowing measurements to be made with periodic illumination. The sensor properties of SnO 2 reference sample and TiO 2 @SnO 2 nanocomposites were investigated toward formaldehyde HCHO (0.06-0.08-0.15-0.3-0.6 ppm) in air (relative humidity at 25 • C RH 25 = 30%). The gas mixtures containing preassigned concentration of HCHO were prepared by dilution of certified gas mixture (6) with background purified air (7). Flow through humidifier (9) was used to adjust the humidity (RH 25 = 30%) of gases passing through the sensor chamber. The sensor response was calculated as S = R air /R gas , where R air is the resistance in background air, R gas is the resistance in the presence of HCHO.

Characteristics of Nanocrystalline SnO2 and TiO2@SnO2 Nanocomposites
The elemental and phase composition of the samples under investigation are presented in Table 1. An increase in the number of ALD cycles led to a proportional increase in the titanium content (presented as ([Ti]/([Ti]+[Sn]) ratio, mol%) in nanocomposites.

Results and Discussion
3.1. Characteristics of Nanocrystalline SnO 2 and TiO 2 @SnO 2 Nanocomposites The elemental and phase composition of the samples under investigation are presented in Table 1. An increase in the number of ALD cycles led to a proportional increase in the titanium content (presented as ([Ti]/([Ti]+[Sn]) ratio, mol%) in nanocomposites.  Figure 3 shows the TEM image of SnO 2 powder used as a substrate in ALD synthesis confirming formation of SnO 2 nanocrystals with approximate size of 5-10 nm. Lattice-resolved TEM image presented in Figure 3b highlights the crystalline phase of SnO 2 with a plane spacing of 0.34 nm  Figure 3. shows the TEM image of SnO2 powder used as a substrate in ALD synthesis confirming formation of SnO2 nanocrystals with approximate size of 5-10 nm. Lattice-resolved TEM image presented in Figure 3b highlights the crystalline phase of SnO2 with a plane spacing of 0.34 nm belonging to the (110) plane of cassiterite SnO2. The lattice-resolved TEM image of SnTi-1 nanocomposite (Figure 3c) makes it possible to detect crystalline regions with interplanar distances of 0.34 and 0.29 nm, corresponding to (110) SnO2 (cassiterite) and (121) TiO2 (brookite), respectively. According to the results of the XRD analysis (Figure 4), the SnO2 annealed at 500 o C (equivalently to the post-synthetic annealing used for the synthesis of TiO2@SnO2 nanocomposites) crystallizes in a tetragonal cassiterite structure (ICDD 41-1445) with a crystallite size of 9-10 nm. Intense reflections from the cassiterite phase are observed in the diffraction patterns of all the samples. The titaniumcontaining phases, formed during post-synthetic annealing, change with an increase in the number of ALD cycles. The TiO2 with brookite structure (ICDD 29-1360) presents in SnTi-1 and SnTi-2 nanocomposites, as evidenced by the appearance of the intense (121) diffraction peak at 2θ = 30.8°. The most intense (120) at 2θ = 25.3° and (111) 2θ = 25.7° diffraction reflections of brookite coincide with the most intense (101) diffraction reflection (2θ = 25.3°) of the TiO2 anatase phase (ICDD 21-1272). Therefore, the presence of the anatase phase in these nanocomposites cannot be excluded. The intensity of diffraction peak at 2θ = 25.3° increases with increasing titanium content. In the diffraction pattern of SnTi-3 nanocomposite a new diffraction peak at 2θ = 48.1° corresponds to the (200) reflection of the anatase phase. At the same time, the diffraction peak at 2θ = 30.8° does not appear that indicates the absence of the brookite phase in this sample. The least intense and wide peaks in the ranges of 2θ = 36.5°-40.0° and 2θ = 52.6°-64.5° correspond to the superposition of the reflections of almost all of the above phases, so their deconvolution and assignment to a certain phase is a difficult task. The broadening of the SnO2 reflections in the diffraction patterns of nanocomposites and overlap of intense diffraction peaks corresponding to SnO2 and TiO2 in SnO2-TiO2 (anatase) and SnO2-TiO2 (brookite) systems (in contrast to the TiO2-ZnO [32]) do not allow us to make a positive or negative conclusion about the incorporation of Ti into the SnO2 lattice.
The crystallite size of detected Sn-and Ti-containing phases depending on titanium content in the samples is presented in Table 1. The crystallite size of tin dioxide, used as a substrate for the deposition of titanium dioxide by the ALD method, is 3-4 nm [30,31]. The post-synthetic annealing at 500 °C, necessary for the removal of organic residues of the titanium precursor, leads to an increase in the size of SnO2 crystallites to 9-10 nm. With an increase in the titanium content in TiO2@SnO2 nanocomposites, a decrease in the size of SnO2 crystal grains is observed as compared with the reference sample. A similar effect was repeatedly noted for nanocomposites based on semiconductor metal oxides [33][34][35]. The presence of impurities on the surface of growing crystallites slows down their growth rate under isothermal annealing due to the so-called Zener pinning [36]. The maximum crystallite size of the main phase is determined by the volume fraction and the size of particles (crystalline or amorphous) segregated on the surface of growing crystallites. The deposition of TiO2 The crystallite size of detected Sn-and Ti-containing phases depending on titanium content in the samples is presented in Table 1. The crystallite size of tin dioxide, used as a substrate for the deposition of titanium dioxide by the ALD method, is 3-4 nm [30,31]. The post-synthetic annealing at 500 • C, necessary for the removal of organic residues of the titanium precursor, leads to an increase in the size of SnO 2 crystallites to 9-10 nm. With an increase in the titanium content in TiO 2 @SnO 2 nanocomposites, a decrease in the size of SnO 2 crystal grains is observed as compared with the reference sample. A similar effect was repeatedly noted for nanocomposites based on semiconductor metal oxides [33][34][35]. The presence of impurities on the surface of growing crystallites slows down their growth rate under isothermal annealing due to the so-called Zener pinning [36]. The maximum crystallite size of the main phase is determined by the volume fraction and the size of particles (crystalline or amorphous) segregated on the surface of growing crystallites. The deposition of TiO 2 layer on the surface of nanocrystalline SnO 2 reduces the area of SnO 2 intergranular contacts, that prevents recrystallization of SnO 2 particles. The greater the thickness of the deposited TiO 2 layer, the less the coarsening of SnO 2 particles occurs. For the TiO 2 brookite phase in SnTi-1 and SnTi-2 nanocomposites the crystallite size was estimated using the (121) reflection that does not overlap with another diffraction peaks. For the size of the crystallites of the TiO 2 anatase phase, such an estimate is possible only in the case of SnTi-3 nanocomposite, in which there is no brookite phase. For the Ti-containing phases, the increase in the crystallite size is observed. Since during the ALD cycles the Ti-containing precursor has been deposited on the outer surface of SnO 2 agglomerates, the sintering and coarsening of Ti-containing particles will be easier. That is why an increase in the titanium content leads to an increase in the crystallite size of the Ti-containing phases. layer on the surface of nanocrystalline SnO2 reduces the area of SnO2 intergranular contacts, that prevents recrystallization of SnO2 particles. The greater the thickness of the deposited TiO2 layer, the less the coarsening of SnO2 particles occurs. For the TiO2 brookite phase in SnTi-1 and SnTi-2 nanocomposites the crystallite size was estimated using the (121) reflection that does not overlap with another diffraction peaks. For the size of the crystallites of the TiO2 anatase phase, such an estimate is possible only in the case of SnTi-3 nanocomposite, in which there is no brookite phase. For the Ticontaining phases, the increase in the crystallite size is observed. Since during the ALD cycles the Ticontaining precursor has been deposited on the outer surface of SnO2 agglomerates, the sintering and coarsening of Ti-containing particles will be easier. That is why an increase in the titanium content leads to an increase in the crystallite size of the Ti-containing phases. The IR spectra of nanocrystalline SnO2 and TiO2@SnO2 composites are compared in Figure 5a. In all spectra, a large broad peak in the region of 2700-3500 cm -1 is observed, which is related to the stretching vibrations of OH groups and a peak at 1628 cm -1 , which is related to the deformation vibrations of adsorbed water [37,38]. The wide and intense absorption bands at 530 cm -1 and 620 cm -1 in the case of SnO2 correspond to stretching vibrations of the Sn-O and symmetric vibrations of the O-Sn-O bonds, respectively [39]. The metal-oxygen (M-O) oscillation modes for SnO2 and TiO2 in TiO2@SnO2 composites overlap in the range of 400-650 cm -1 , nevertheless different components can be distinguished. All the FT-IR spectra of the composites show vibration modes at 450 and 610 cm -1 , which can be attributed to oscillations of Ti-O-Ti and Ti-O bonds, respectively [40][41][42]. With an increase in the titanium content an increase in the intensity of the Ti-O oscillation modes is observed, which appears as a broadening of the M-O absorption peak in this region. The IR spectra of nanocrystalline SnO 2 and TiO 2 @SnO 2 composites are compared in Figure 5a. In all spectra, a large broad peak in the region of 2700-3500 cm -1 is observed, which is related to the stretching vibrations of OH groups and a peak at 1628 cm -1 , which is related to the deformation vibrations of adsorbed water [37,38]. The wide and intense absorption bands at 530 cm -1 and 620 cm -1 in the case of SnO 2 correspond to stretching vibrations of the Sn-O and symmetric vibrations of the O-Sn-O bonds, respectively [39]. The metal-oxygen (M-O) oscillation modes for SnO 2 and TiO 2 in TiO 2 @SnO 2 composites overlap in the range of 400-650 cm -1 , nevertheless different components can be distinguished. All the FT-IR spectra of the composites show vibration modes at 450 and 610 cm -1 , which can be attributed to oscillations of Ti-O-Ti and Ti-O bonds, respectively [40][41][42]. With an increase in the titanium content an increase in the intensity of the Ti-O oscillation modes is observed, which appears as a broadening of the M-O absorption peak in this region. Figure 5b shows the Raman spectra of nanocrystalline SnO 2 (SnTi-0) and TiO 2 @SnO 2 composites. Raman spectrum of the SnO 2 clearly shows three characteristic modes at 480.5, 631 and 772.8 cm −1 that correspond to the E g , A 1g and B 2g vibrational modes, respectively [43,44]. The A 1g and B 2g modes are associated with symmetric and asymmetric Sn-O stretching, respectively, orthogonally to the c axis.
The translational E g mode is related to the motion of oxygen anions along the c axis [43,44]. The B 1g oscillation mode appears only in the spectra of nanocrystalline SnO 2 in the range of 100-184 cm −1 [45,46]. In the SnO 2 spectrum presented in Figure 5b the band at 137 cm −1 is due to B 1g mode, the shift may be associated with the nanoparticles size effect.  Figure 5b shows the Raman spectra of nanocrystalline SnO2 (SnTi-0) and TiO2@SnO2 composites. Raman spectrum of the SnO2 clearly shows three characteristic modes at 480.5, 631 and 772.8 cm −1 that correspond to the Eg, A1g and B2g vibrational modes, respectively [43,44]. The A1g and B2g modes are associated with symmetric and asymmetric Sn-O stretching, respectively, orthogonally to the c axis. The translational Eg mode is related to the motion of oxygen anions along the c axis [43,44]. The B1g oscillation mode appears only in the spectra of nanocrystalline SnO2 in the range of 100-184 cm −1 [45,46]. In the SnO2 spectrum presented in Figure 5b the band at 137 cm −1 is due to B1g mode, the shift may be associated with the nanoparticles size effect.
J. Zuo et al. studied the size effects in SnO2 nanoparticles [47]. They showed that, in addition to the characteristic vibrational modes of bulk SnO2, the Raman spectrum of nanocrystalline SnO2 has two additional Raman scattering bands at 358 (B1) and 572 cm −1 (B2). The B2 band corresponds to surface modes and is very sensitive to the changes in crystallite size for nanoscale SnO2. The appearance of surface modes is associated with a small particle size of SnO2 and can be due to the appearance of oscillations that are forbidden by symmetry due to the breaking of long-range order in the systems of reduced dimension. That is why a decrease in crystallite size leads to the formation of a highly defective surface layer, the contribution of which will be the highest for the materials with the smallest particle size [48]. Based on the above considerations, the wide band located at 563 cm −1 corresponds to the surface modes associated with in-plane oxygen vacancies of the nanocrystalline cassiterite SnO2 [48][49][50]. Consequently, the small size and surface defects of the SnO2 nanoparticles may have a positive effect on the gas sensitivity of the sensor. This assumption was experimentally confirmed in by the authors of [31], where it was shown that the relative intensity of Raman surface modes IS/IV taken as the ratio of the sum of their intensities IS to the intensity of A1g mode IV demonstrates the best linear correlation with gas response of SnO2 nanocrystalline materials to CO.
Changes in the crystal's local symmetry produce changes in some of the components of the polarizability tensor, even for usually forbidden vibration modes [49]. That is why the A2u IR active and Raman forbidden modes are found to transform into Raman active modes [48]. In this case the bands at 309 and 352 cm −1 (Eu) and the band at 444 (B1u) are related to transformation of an IR to Raman active modes.
Raman peaks observed in the spectra of TiO2@SnO2 composites at 145, 198, 397, 516 and 635 cm −1 (Figure 5b), refer to the Eg, Eg, B1g, A1g + B1g and Eg modes of the anatase phase, respectively [51]. For TiO2 nanoparticles the Eg Raman peak is mainly caused by symmetric stretching vibration of O-Ti-O groups, B1g peak is caused by symmetric bending vibration of O-Ti-O and A1g peak is caused by antisymmetric bending vibration of O-Ti-O [52]. The presence of an intense 145 cm −1 mode (a  [47]. They showed that, in addition to the characteristic vibrational modes of bulk SnO 2 , the Raman spectrum of nanocrystalline SnO 2 has two additional Raman scattering bands at 358 (B 1 ) and 572 cm −1 (B 2 ). The B 2 band corresponds to surface modes and is very sensitive to the changes in crystallite size for nanoscale SnO 2 . The appearance of surface modes is associated with a small particle size of SnO 2 and can be due to the appearance of oscillations that are forbidden by symmetry due to the breaking of long-range order in the systems of reduced dimension. That is why a decrease in crystallite size leads to the formation of a highly defective surface layer, the contribution of which will be the highest for the materials with the smallest particle size [48]. Based on the above considerations, the wide band located at 563 cm −1 corresponds to the surface modes associated with in-plane oxygen vacancies of the nanocrystalline cassiterite SnO 2 [48][49][50]. Consequently, the small size and surface defects of the SnO 2 nanoparticles may have a positive effect on the gas sensitivity of the sensor. This assumption was experimentally confirmed in by the authors of [31], where it was shown that the relative intensity of Raman surface modes I S /I V taken as the ratio of the sum of their intensities I S to the intensity of A 1g mode I V demonstrates the best linear correlation with gas response of SnO 2 nanocrystalline materials to CO.
Changes in the crystal's local symmetry produce changes in some of the components of the polarizability tensor, even for usually forbidden vibration modes [49]. That is why the A 2u IR active and Raman forbidden modes are found to transform into Raman active modes [48]. In this case the bands at 309 and 352 cm −1 (E u ) and the band at 444 (B 1u ) are related to transformation of an IR to Raman active modes.
Raman peaks observed in the spectra of TiO 2 @SnO 2 composites at 145, 198, 397, 516 and 635 cm −1 (Figure 5b), refer to the E g , E g , B 1g , A 1g + B 1g and E g modes of the anatase phase, respectively [51]. For TiO 2 nanoparticles the E g Raman peak is mainly caused by symmetric stretching vibration of O-Ti-O groups, B 1g peak is caused by symmetric bending vibration of O-Ti-O and A 1g peak is caused by antisymmetric bending vibration of O-Ti-O [52]. The presence of an intense 145 cm −1 mode (a characteristic oscillation mode of anatase) indicates that TiO 2 nanocrystals have a certain degree of long-range order, while weaker and wider peaks in the high-frequency region indicate the absence of a short-range order in the anatase phase [53][54][55]. According to the factor group analysis, TiO 2 brookite phase has a total of 36 Raman active modes (9A 1g + 9B 1g + 9B 2g + 9B 3g ). Raman spectra of the SnTi-1 and SnTi-2 composites show both anatase and brookite bands. In total, 4 brookite bands were readily identified, including A 1g (255 cm −1 ), B 3g (285, 443 cm −1 ), B 2g (581 cm −1 ). Also, the A 1g (153, 194 cm −1 ) and B 2g (395 cm −1 ) brookite bands may be overlapped by anatase modes, which are very close to  [56,57]. The full assignment of the vibrational modes in the Raman spectra of nanocomposites is presented in Table 2. Table 2. Assignment of Raman vibrational modes (cm −1 ).
Thus, the Raman spectra confirm the data obtained by the XRD method. The formation of the metastable brookite phase is observed in SnTi-1 and SnTi-2 composites. Anatase can play the role of a stabilizer for this phase [58]. A correlation among the surface enthalpies of the TiO 2 three polymorphs and their particle size was found by Zhang and Banfield [59]. The formation energies of anatase, brookite and rutile are sufficiently close that they can be reversed by small differences in surface energies. Zhu et al. [60] developed an empirical expression on a critical grain size of brookite, which determines the transition sequence between anatase and brookite. These transformations become noticeable with prolonged isothermal annealing at temperatures above 500 • C. To avoid the changes in phase composition and crystallite size of nanocomposites subjected to post synthesis annealing at 500 • C, the maximum temperature during the manufacture of sensor elements and gas sensor measurements did not exceed 300 • C.

Gas Sensor Properties
As discussed in the review [8], depending on the operating temperature the change of the resistance of n-type semiconductor oxides when interacting with formaldehyde is due to HCHO oxidation with chemisorbed oxygen O α− β(ads) to HCOOH or CO 2 : At constant temperature and HCHO concentration, the value of the sensor response will depend on the concentration and the predominant form of chemisorbed oxygen on the surface of the semiconductor oxide.
According to the literature [20,[61][62][63][64], the interaction of UV light with TiO 2 involves the following processes: electron-hole pairs generation (3), oxygen photodesorption (4), formation of "active" chemisorbed oxygen (5), which is able to oxidize formaldehyde even at room temperature (6) hν Under UV light, hydroxyl groups presented on the TiO 2 surface can also pass into the "active" form (7) and then participate in the oxidation of formaldehyde (8): Figure 6 demonstrates the change in the resistance of nanocomposites with the cyclic changes in the composition of the gas phase "air (15 min)"-"0.6 ppm HCHO in air (15 min)". The measurements were carried out in the temperature range of 300-50 • C in dark conditions (Figure 6a) and under periodic UV illumination (Figure 6b). At a fixed temperature, TiO 2 @SnO 2 nanocomposites have a higher resistance than unmodified SnO 2 (SnTi-0). UV illumination leads to reduced resistance of nanocomposites by 1.5-3 times. In all cases in the presence of HCHO, the resistance of SnO 2 and TiO 2 @SnO 2 nanocomposites decreases due to the oxidation of formaldehyde by chemisorbed oxygen. In dark conditions at low measurement temperatures (T < 200 • C), the baseline resistance drift is observed for all samples. This may be due to the accumulation of formaldehyde oxidation products on the surface of the sensitive layer under these conditions. The use of UV illumination reduces the drift of the resistance at T = 150 • C. It can be assumed that UV light stimulates the desorption of the products of formaldehyde oxidation.
Under UV light, hydroxyl groups presented on the TiO2 surface can also pass into the "active" form (7) and then participate in the oxidation of formaldehyde (8): Figure 6 demonstrates the change in the resistance of nanocomposites with the cyclic changes in the composition of the gas phase "air (15 min)"-"0.6 ppm HCHO in air (15 min)". The measurements were carried out in the temperature range of 300-50 °C in dark conditions (Figure 6a) and under periodic UV illumination (Figure 6b). At a fixed temperature, TiO2@SnO2 nanocomposites have a higher resistance than unmodified SnO2 (SnTi-0). UV illumination leads to reduced resistance of nanocomposites by 1.5-3 times. In all cases in the presence of HCHO, the resistance of SnO2 and TiO2@SnO2 nanocomposites decreases due to the oxidation of formaldehyde by chemisorbed oxygen. In dark conditions at low measurement temperatures (T < 200 °C), the baseline resistance drift is observed for all samples. This may be due to the accumulation of formaldehyde oxidation products on the surface of the sensitive layer under these conditions. The use of UV illumination reduces the drift of the resistance at T = 150 °C. It can be assumed that UV light stimulates the desorption of the products of formaldehyde oxidation. The obtained results allowed us to draw the temperature dependence of the sensor response = / (Figure 7a). The maximum sensor response of unmodified SnO2 (SnTi-0) is observed at T = 20 °C. The introduction of Ti-containing phases leads to a decrease in the temperature of the maximum sensor signal to 150 °C. In dark conditions, the maximum response was detected for the SnTi-2 nanocomposite. The use of UV illumination does not change the position of the maxima on the temperature dependence of the sensor response, but unexpectedly it leads to a slight decrease in the signal value in the low temperature range of 50-150 °C. Such a decrease in the sensor response can be caused by a decrease in the concentration of chemisorbed oxygen participating in the oxidation of formaldehyde by reactions (1) and (2), due to the partial desorption of oxygen from the surface of semiconductor oxides under UV light [65]. Figure 7b shows the temperature dependences of the effective photoresponse SPh of nanocomposites in background air calculated as SPh = Rdark/Rlight, where Rlight is the minimum resistance achieved during the sensor illumination, and Rdark is the maximum The obtained results allowed us to draw the temperature dependence of the sensor response S = R air /R gas (Figure 7a). The maximum sensor response of unmodified SnO 2 (SnTi-0) is observed at T = 20 • C. The introduction of Ti-containing phases leads to a decrease in the temperature of the maximum sensor signal to 150 • C. In dark conditions, the maximum response was detected for the SnTi-2 nanocomposite. The use of UV illumination does not change the position of the maxima on the temperature dependence of the sensor response, but unexpectedly it leads to a slight decrease in the signal value in the low temperature range of 50-150 • C. Such a decrease in the sensor response can be caused by a decrease in the concentration of chemisorbed oxygen participating in the oxidation of formaldehyde by reactions (1) and (2), due to the partial desorption of oxygen from the surface of semiconductor oxides under UV light [65]. Figure 7b shows the temperature dependences of the effective photoresponse S Ph of nanocomposites in background air calculated as S Ph = R dark /R light , where R light is the minimum resistance achieved during the sensor illumination, and R dark is the maximum resistance achieved in the dark period [66]. The maximum photoresponse corresponds to a temperature range of 150-200 • C. A decrease in the S Ph value with an increase in the measurement temperature up to 250-300 • C is due to the contribution of thermal oxygen desorption in dark conditions. resistance achieved in the dark period [66]. The maximum photoresponse corresponds to a temperature range of 150-200 °C. A decrease in the SPh value with an increase in the measurement temperature up to 250-300 °C is due to the contribution of thermal oxygen desorption in dark conditions.
(a) (b) The dynamic sensor characteristics, response time and recovery time * , are presented in Figure 8. Even though the absolute values of response time and recovery time are strongly dependent on the parameters of the testing system (geometry and size of the measurement cell, the method to switch the gases, the gas flow rate), they are useful to compare the dynamic sensor characteristics of materials if measurements are performed in identical conditions. With a decrease in the operating temperature, an increase in both the response and recovery times is observed. It should be noted that the sensor based on unmodified SnO2 is characterized by significantly worse dynamic characteristics than sensors based on TiO2@SnO2 nanocomposites, which are close to each other. Such a difference may be due to a decrease in the degree of sintering of SnO2 nanoparticles in nanocomposites as compared to unmodified tin dioxide.  The dynamic sensor characteristics, response time τ 90 . and recovery time τ * 90 . , are presented in Figure 8. Even though the absolute values of response time and recovery time are strongly dependent on the parameters of the testing system (geometry and size of the measurement cell, the method to switch the gases, the gas flow rate), they are useful to compare the dynamic sensor characteristics of materials if measurements are performed in identical conditions. With a decrease in the operating temperature, an increase in both the response and recovery times is observed. It should be noted that the sensor based on unmodified SnO 2 is characterized by significantly worse dynamic characteristics than sensors based on TiO 2 @SnO 2 nanocomposites, which are close to each other. Such a difference may be due to a decrease in the degree of sintering of SnO 2 nanoparticles in nanocomposites as compared to unmodified tin dioxide. resistance achieved in the dark period [66]. The maximum photoresponse corresponds to a temperature range of 150-200 °C. A decrease in the SPh value with an increase in the measurement temperature up to 250-300 °C is due to the contribution of thermal oxygen desorption in dark conditions.
(a) (b) The dynamic sensor characteristics, response time and recovery time * , are presented in Figure 8. Even though the absolute values of response time and recovery time are strongly dependent on the parameters of the testing system (geometry and size of the measurement cell, the method to switch the gases, the gas flow rate), they are useful to compare the dynamic sensor characteristics of materials if measurements are performed in identical conditions. With a decrease in the operating temperature, an increase in both the response and recovery times is observed. It should be noted that the sensor based on unmodified SnO2 is characterized by significantly worse dynamic characteristics than sensors based on TiO2@SnO2 nanocomposites, which are close to each other. Such a difference may be due to a decrease in the degree of sintering of SnO2 nanoparticles in nanocomposites as compared to unmodified tin dioxide.  To clarify the mechanism of formaldehyde oxidation on the surface of semiconductor oxides, the in situ DRIFT studies have been conducted. In situ DRIFT spectra of the unmodified SnO 2 (SnTi-0) and SnTi-2nanocomposite during HCHO adsorption at room temperature are shown in Figure 9. To clarify the mechanism of formaldehyde oxidation on the surface of semiconductor oxides, the in situ DRIFT studies have been conducted. In situ DRIFT spectra of the unmodified SnO2 (SnTi-0) and SnTi-2nanocomposite during HCHO adsorption at room temperature are shown in Figure 9.
The bands of weakly bonded forms at 1202, 1742, 1766 and 3010 cm −1 for SnTi-0 and at 1768 cm −1 for SnTi-2 attributed to molecular adsorption form of HCHO with the formation of hydrogen bonds between its carbonyl oxygen and surface hydroxyl groups [67][68][69]. The formation of CO2 (at 2337 and 2365 cm −1 ) was observed on SnTi-0 after 5 min of HCHO adsorption, moreover these bands grew in intensity with the increase of the adsorption time [70]. In SnTi-2 spectra the bands at 1344 and 1592 cm −1 are assigned to COO symmetric stretching and COO asymmetric stretching of formate species [71,72]. The bands located at 2838, 2890 and 2590, 2868 cm −1 for SnTi-0 and SnTi-2, respectively, belong to CH stretching of formate species [68,73,74]. The bands at 2936 cm −1 for SnTi-0 and 2930 cm −1 for SnTi-2 were assigned to the characteristic peak of dioxymethylene (DOM) intermediate [71]. The other DOM bands also were identified at 1050, 1107, 1145 cm −1 for SnTi-0 and 1066, 1107, 1146, 1165, 1478 cm −1 for SnTi-2 [67,689]. Figure 10 shows the DRIFT spectra of the SnTi-2 nanocomposite obtained during the heating in dry air at T = 300 °C after formaldehyde adsorption at room temperature. From the results, it can be seen that the formats are intensively desorbed from the surface of the nanocomposite, as evidenced by decreasing of their characteristic vibrational modes at 1346, 1382 and 1553 cm −1 [68,71,72]. However, at such a high temperature, there is an increase in the intensity of dioxymethylene species, the bands of which are at 1050, 1108 and 1450 cm −1 [67,68], which indicates partially HCHO oxidation to DOM. Most likely, at T=300 °C, formaldehyde is oxidized to CO2 and H2O, as a result of which hydroxyl groups of water accumulate on the surface (increase in the intensity of OH bands at 3550 and 3630 cm −1 ) and CO2 desorption occurs (decrease in intensity at 2340 cm −1 ); at room temperature CO2 is accumulated on the SnTi-0 surface (Figure 9a). The assignment of IR bands for intermediates is summarized in Table 3.
So, for HCHO oxidation, oxygen ions adsorbed on the surface of the nanocomposites play a key role in the generation of the surface formates. According to the results of the in situ DRIFT analysis and literature review [67][68][69][70]75,76] we propose a reaction mechanism as depicted in (9) The bands of weakly bonded forms at 1202, 1742, 1766 and 3010 cm −1 for SnTi-0 and at 1768 cm −1 for SnTi-2 attributed to molecular adsorption form of HCHO with the formation of hydrogen bonds between its carbonyl oxygen and surface hydroxyl groups [67][68][69]. The formation of CO 2 (at 2337 and 2365 cm −1 ) was observed on SnTi-0 after 5 min of HCHO adsorption, moreover these bands grew in intensity with the increase of the adsorption time [70]. In SnTi-2 spectra the bands at 1344 and 1592 cm −1 are assigned to COO symmetric stretching and COO asymmetric stretching of formate species [71,72]. The bands located at 2838, 2890 and 2590, 2868 cm −1 for SnTi-0 and SnTi-2, respectively, belong to CH stretching of formate species [68,73,74]. The bands at 2936 cm −1 for SnTi-0 and 2930 cm −1 for SnTi-2 were assigned to the characteristic peak of dioxymethylene (DOM) intermediate [71]. The other DOM bands also were identified at 1050, 1107, 1145 cm −1 for SnTi-0 and 1066, 1107, 1146, 1165, 1478 cm −1 for SnTi-2 [67,68]. Figure 10 shows the DRIFT spectra of the SnTi-2 nanocomposite obtained during the heating in dry air at T = 300 • C after formaldehyde adsorption at room temperature. From the results, it can be seen that the formats are intensively desorbed from the surface of the nanocomposite, as evidenced by decreasing of their characteristic vibrational modes at 1346, 1382 and 1553 cm −1 [68,71,72]. However, at such a high temperature, there is an increase in the intensity of dioxymethylene species, the bands of which are at 1050, 1108 and 1450 cm −1 [67,68], which indicates partially HCHO oxidation to DOM. Most likely, at T=300 • C, formaldehyde is oxidized to CO 2 and H 2 O, as a result of which hydroxyl groups of water accumulate on the surface (increase in the intensity of OH bands at 3550 and 3630 cm −1 ) and CO 2 desorption occurs (decrease in intensity at 2340 cm −1 ); at room temperature CO 2 is accumulated on the SnTi-0 surface (Figure 9a). The assignment of IR bands for intermediates is summarized in Table 3.
adsorbed surface oxygen to yield formate HCOO-, which can proceed via formation of dioxymethylene H2COO -intermediate (10), (11) [69]. For SnTi-0 sample there are two weak bands at 2337 and 2365 cm −1 indicated the appearance of carbon dioxide [64,77]. In this case, CO2 and H2O were formed as the result of oxidation of formate ions (12) of chemisorbed oxygen on semiconductor oxides [78], reaction (10) proceeds without changing the electron concentration in the conduction band of the semiconductor, that causes the absence of a sensor response at T = 50 °C (Figure 6).    So, for HCHO oxidation, oxygen ions adsorbed on the surface of the nanocomposites play a key role in the generation of the surface formates. According to the results of the in situ DRIFT analysis and literature review [67][68][69][70]75,76] we propose a reaction mechanism as depicted in (9)-(12): At first, formaldehyde molecularly adsorbs as a whole via strong hydrogen-bonding interactions (9). Desorption of formaldehyde is in competition with the reaction between formaldehyde and co-adsorbed surface oxygen to yield formate HCOO-, which can proceed via formation of dioxymethylene H 2 COOintermediate (10),(11) [69]. For SnTi-0 sample there are two weak bands at 2337 and 2365 cm −1 indicated the appearance of carbon dioxide [64,77]. In this case, CO 2 and H 2 O were formed as the result of oxidation of formate ions (12) of chemisorbed oxygen on semiconductor oxides [78], reaction (10) proceeds without changing the electron concentration in the conduction band of the semiconductor, that causes the absence of a sensor response at T = 50 • C ( Figure 6). The observed effect of Ti-containing phases on the sensor response of SnO 2 toward HCHO in dark conditions should be explained based on the model considering the formation of n-n heterocontact at the SnO 2 /TiO 2 interface. The estimated band alignment of the SnO 2 (cassiterite) and TiO 2 (anatase) is presented on Figure 11a [24,79,80]. The available lower-energy conduction band states stimulate electron transfer to n-SnO 2 [81]. As a result, the depletion layer is formed at n-TiO 2 surface due to loss of electrons, and accumulation layer is formed at SnO 2 surface due to added electrons. In turn, an increase in the electron concentration stimulates the chemisorption of oxygen species enhancing the response formed due to reactions (1). The scheme in Figure 11a is constructed for the heterocontact SnO 2 (cassiterite)/TiO 2 (anatase), since the anatase phase is present in all nanocomposites (as it follows from the data of Raman spectroscopy). Brookite (revealed in SnTi-1 and SnTi-2 nanocomposites) is the least studied polymorph of TiO 2 . However, from the data of Ref. [82] it follows that the bottom of the conduction band of brookite lies above the bottom of the conduction band of anatase. Thus, in the formation of a heterocontact SnO 2 (cassiterite)/TiO 2 (brookite), electron transfer will also occur from TiO 2 (brookite) to SnO 2 , increasing the chemisorption of oxygen on its surface. An increase in the concentration of chemisorbed oxygen on the surface of nanocomposites compared with unmodified SnO 2 is confirmed by the method of thermo-programmed reduction with hydrogen (TPR-H 2 ). Figure 11b shows the temperature dependencies of the hydrogen consumption during the reduction of unmodified SnO 2 and SnTi-1 nanocomposite. The reduction of SnO 2 with hydrogen occurs in the temperature range of 430-800 • C, the maximum H 2 consumption is observed at T = 620 • C. The introduction of Ti-containing phases leads to a shift of the H 2 consumption towards lower temperatures 570-600 • C. Such a change in the TPR-H 2 profiles may be due to the reduction of highly dispersed Ti-containing phases. According to the reports [83][84][85], the reduction of TiO 2 with hydrogen occurs at T > 400 • C, the maximum H 2 consumption is observed at T = 530-650 • C, depending on the TiO 2 dispersion. Another feature of the TPR-H 2 profiles is the presence of hydrogen absorption peaks in the low-temperature range of 100-400 • C. This corresponds to the interaction of hydrogen with oxygen-containing particles (chemisorbed oxygen, hydroxyl groups, etc.) on the surface of highly dispersed semiconductor oxides. The reduction of TiO 2 @SnO 2 nanocomposites is accompanied by more significant hydrogen consumption in the low-temperature region that indicates a higher concentration of oxygen-containing species on its surface. An increase in the concentration of chemisorbed oxygen on the surface of nanocomposites provides an increase in their resistance compared to unmodified SnO 2 . At the same time, due to the photodesorption of chemisorbed oxygen under UV light, TiO 2 @SnO 2 nanocomposites are characterized by the larger photoresponse in the background air and demonstrate a more significant decrease in the sensor response to HCHO.
Thus, we are forced to conclude that in the case of TiO 2 @SnO 2 nanocomposites, the amplitude of the sensor response when detecting HCHO in the sub-ppm range is determined mainly by the SnO 2 /TiO 2 interface. The role of UV light is to enhance the photodesorption of oxygen, and the processes of HCHO oxidation by photoactivated particles (reactions (6) and (8)) do not contribute to the sensor response.
The dependence of the sensor response on the HCHO concentration in air was studied in dark conditions and under periodic UV illumination at 150 • C during the stepwise increase and subsequent stepwise decrease in the HCHO concentration ( Figure 12). The results obtained allowed us to build the calibration curves ( Figure 13a) and to determine the lower detection limit (LDL). As in the case of measurements at different temperatures, when using UV illumination, a decrease in the sensor response toward HCHO is observed, that may be due to the photodesorption of chemisorbed oxygen under UV light. The LDL values were calculated from the calibration curves as the gas concentration corresponding to the minimum measurable sensor signal R av /(R av -3σ) , where R av is the average sensor resistance in air at 150 • C, σ is the standard deviation of the resistance in air. A decrease in the sensitivity of nanocomposites is accompanied by an increase in LDL from 9-15 to 30-32 ppb (Table 4).
an increase in their resistance compared to unmodified SnO2. At the same time, due to the photodesorption of chemisorbed oxygen under UV light, TiO2@SnO2 nanocomposites are characterized by the larger photoresponse in the background air and demonstrate a more significant decrease in the sensor response to HCHO.
Thus, we are forced to conclude that in the case of TiO2@SnO2 nanocomposites, the amplitude of the sensor response when detecting HCHO in the sub-ppm range is determined mainly by the SnO2/TiO2 interface. The role of UV light is to enhance the photodesorption of oxygen, and the processes of HCHO oxidation by photoactivated particles (reactions (6) and (8)) do not contribute to the sensor response. sensor response toward HCHO is observed, that may be due to the photodesorption of chemisorbed oxygen under UV light. The LDL values were calculated from the calibration curves as the gas concentration corresponding to the minimum measurable sensor signal Rav/(Rav -3σ) , where Rav is the average sensor resistance in air at 150 °C, σ is the standard deviation of the resistance in air. A decrease in the sensitivity of nanocomposites is accompanied by an increase in LDL from 9-15 to 30-32 ppb (Table 4). The comparison of the sensor response toward 0.06-0.6 ppm HCHO for SnTi-2 nanocomposite with the literature data [7][8][9][10][11][20][21][22][23]28, is shown in Figure 13b and Table 5. There are very few papers that consider the detection of formaldehyde in a practically important sub-ppm concentration range. TiO2@SnO2 nanocomposites exhibit high sensitivity to HCHO in the sub-ppm range. Higher sensor response values were obtained only by the authors of [86], where metastable In4Sn3O12 was used as sensitive material at 350 °C and in Ref. [7], where Si modified SnO2 was used at 400 °C. It should be noted that TiO2@SnO2 nanocomposites have comparable sensitivity at much lower measurement temperature (150 °C vs. 350 °C or 400 °C), which can provide a significant reduction in the power consumption of semiconductor gas sensor.  The comparison of the sensor response toward 0.06-0.6 ppm HCHO for SnTi-2 nanocomposite with the literature data [7][8][9][10][11][20][21][22][23]28, is shown in Figure 13b and Table 5. There are very few papers that consider the detection of formaldehyde in a practically important sub-ppm concentration range. TiO 2 @SnO 2 nanocomposites exhibit high sensitivity to HCHO in the sub-ppm range. Higher sensor response values were obtained only by the authors of [86], where metastable In 4 Sn 3 O 12 was used as sensitive material at 350 • C and in Ref. [7], where Si modified SnO 2 was used at 400 • C. It should be noted that TiO 2 @SnO 2 nanocomposites have comparable sensitivity at much lower measurement temperature (150 • C vs. 350 • C or 400 • C), which can provide a significant reduction in the power consumption of semiconductor gas sensor.
In addition to reducing power consumption, a sufficiently low measurement temperature allows one to achieve an increase in the selectivity of sensors when detecting various VOCs. The cross sensitivity of SnO 2 and TiO 2 @SnO 2 nanocomposites was studied at a measurement temperature of 150 • C in the detection of formaldehyde, benzene, and acetone ( Figure 14). It can be noted that the synthesized materials have low cross-sensitivity to benzene in the wide concentration range. As for acetone, the interference of signals can occur if acetone concentration is in the range of C > 1 ppm and at least 20 times higher than formaldehyde concentration. SnTi-2 nanocomposite is characterized by the lowest response to acetone.  In addition to reducing power consumption, a sufficiently low measurement temperature allows one to achieve an increase in the selectivity of sensors when detecting various VOCs. The cross sensitivity of SnO2 and TiO2@SnO2 nanocomposites was studied at a measurement temperature of 150 °C in the detection of formaldehyde, benzene, and acetone ( Figure 14). It can be noted that the synthesized materials have low cross-sensitivity to benzene in the wide concentration range. As for acetone, the interference of signals can occur if acetone concentration is in the range of C > 1 ppm and at least 20 times higher than formaldehyde concentration. SnTi-2 nanocomposite is characterized by the lowest response to acetone.
Comparing the sensor characteristics of unmodified SnO2 and TiO2@SnO2 nanocomposites in the detection of formaldehyde at 150 °C (Table 4) we can conclude that the SnO2 is inferior in the magnitude of the sensor response and the dynamic characteristics (response and recovery time). For HCHO detection at concentrations of 60 ppb and above, the SnTi-2 nanocomposite, which exhibits the highest sensor response and the lowest cross-sensitivity to acetone, seems to be preferred. However, for the measurements of lower HCHO concentrations, the SnTi-1 nanocomposite characterized by lower LDL value, lower base resistance in air, and the same dynamic characteristics, may be optimal.  [8][9][10][11][20][21][22][23]28, (squares), Ref. [7] (diamonds) and Ref. [86] (circles). The color of the symbol corresponds to the measurement temperature. For correct comparison all the data were recalculated as S = (R air /R gas -1)*100%.    In addition to reducing power consumption, a sufficiently low measurement temperature allows one to achieve an increase in the selectivity of sensors when detecting various VOCs. The cross sensitivity of SnO2 and TiO2@SnO2 nanocomposites was studied at a measurement temperature of 150 °C in the detection of formaldehyde, benzene, and acetone ( Figure 14). It can be noted that the synthesized materials have low cross-sensitivity to benzene in the wide concentration range. As for acetone, the interference of signals can occur if acetone concentration is in the range of C > 1 ppm and at least 20 times higher than formaldehyde concentration. SnTi-2 nanocomposite is characterized by the lowest response to acetone.
Comparing the sensor characteristics of unmodified SnO2 and TiO2@SnO2 nanocomposites in the detection of formaldehyde at 150 °C (Table 4) we can conclude that the SnO2 is inferior in the magnitude of the sensor response and the dynamic characteristics (response and recovery time). For HCHO detection at concentrations of 60 ppb and above, the SnTi-2 nanocomposite, which exhibits Comparing the sensor characteristics of unmodified SnO 2 and TiO 2 @SnO 2 nanocomposites in the detection of formaldehyde at 150 • C (Table 4) we can conclude that the SnO 2 is inferior in the magnitude of the sensor response and the dynamic characteristics (response and recovery time). For HCHO detection at concentrations of 60 ppb and above, the SnTi-2 nanocomposite, which exhibits the highest sensor response and the lowest cross-sensitivity to acetone, seems to be preferred. However, for the measurements of lower HCHO concentrations, the SnTi-1 nanocomposite characterized by lower LDL value, lower base resistance in air, and the same dynamic characteristics, may be optimal.
As a summary, we have to conclude, that compared to the sensor characteristics described in [7,86,118], the sensors based on TiO 2 @SnO 2 nanocomposites are inferior in sensitivity and selectivity. However, their advantage is significantly reduced operating temperature. The increase in sensitivity while maintaining a low operating temperature should be possible due to the addition of modifiers of different chemical nature on the surface of TiO 2 @SnO 2 nanocomposites. The increase in selectivity, in particular towards humidity, can be achieved using passive filters-selective membranes based on SiO 2 [6,[118][119][120] or metal organic frameworks [110,121] as well as by the creation of multi-sensor systems operating using mathematical signal processing [6,122].

Conclusions
Nanocomposites TiO 2 @SnO 2 obtained by ALD synthesis were investigated as sensitive materials for sub-ppm formaldehyde detection in dark conditions and under periodic UV (λ max = 365 nm) illumination. As observed by XRD and Raman spectroscopy, all nanocomposites contain nanocrystalline SnO 2 cassiterite and TiO 2 anatase as the main crystalline phases. Depending on Ti content in nanocomposites predetermined by the number of ALD cycles, the minor TiO 2 phases brookite, rutile and anosovite have been found. A thorough study of nanocomposites using FTIR and TPR-H 2 methods made it possible to establish that nanocomposites have a higher concentration of chemisorbed oxygen and surface hydroxyl groups compared to unmodified SnO 2 . When detecting formaldehyde in the sub-ppm range, TiO 2 @SnO 2 nanocomposites exhibit a higher sensor signal than SnO 2 and a decrease in the optimal measurement temperature by 50 • C. This result is explained based on the model considering the formation of n-n heterocontact at the SnO 2 /TiO 2 interface. TiO 2 @SnO 2 nanocomposites have high sensitivity toward HCHO at quite low measurement temperature 150 • C that can provide a significant reduction in the power consumption and an increase in the selectivity of semiconductor gas sensors when detecting various VOCs.
Unexpectedly, UV illumination leads to a decrease in sensor response compared with the results obtained in dark conditions. So, we have to conclude that in the case of TiO 2 @SnO 2 nanocomposites, the amplitude of the sensor response toward sub-ppm HCHO concentrations is determined mainly by the SnO 2 /TiO 2 interface. The UV light stimulates the photodesorption of oxygen, while the processes of HCHO oxidation by photoactivated particles do not contribute to the sensor response.