Obtaining of ZnO/Fe2O3 Thin Nanostructured Films by AACVD for Detection of ppb-Concentrations of NO2 as a Biomarker of Lung Infections

ZnO/Fe2O3 nanocomposites with different concentration and thickness of the Fe2O3 layer were obtained by two-stage aerosol vapor deposition (AACVD). It was shown that the ZnO particles have a wurtzite structure with an average size of 51–66 nm, and the iron oxide particles on the ZnO surface have a hematite structure and an average size of 23–28 nm. According to EDX data, the iron content in the films was found to be 1.3–5.8 at.%. The optical properties of the obtained films were studied, and the optical band gap was found to be 3.16–3.26 eV. Gas-sensitive properties at 150–300 °C were studied using a wide group of analyte gases: CO, NH3, H2, CH4, C6H6, ethanol, acetone, and NO2. A high response to 100 ppm acetone and ethanol at 225–300 °C and a high and selective response to 300–2000 ppb NO2 at 175 °C were established. The effect of humidity on the magnitude and shape of the signal obtained upon NO2 detection was studied.


Introduction
Semiconductor gas sensors based on metal oxides (MOS sensors) are widely used to detect toxic and explosive gases in air [1]. Devices with gas sensors of this type have a relatively low cost, high speed, high accuracy, high miniaturization potential, and are easy to operate [2].
Zinc oxide (n-type semiconductor with E g = 3.37 eV) is one of the most widely used gas-sensitive materials in MOS sensors [3]. ZnO thin films exhibit good electrical conductivity, chemical stability, high ultraviolet (UV) absorption, and low toxicity [4,5]. Despite the unique physical and chemical properties, individual ZnO in the composition of chemoresistive gas sensors has some drawbacks: limited maximum sensitivity, low selectivity, high operating temperature, and long response and recovery times [6].
Various approaches to ZnO modification are used to improve its gas sensing properties. Increase in sensitivity can be achieved by using various transition metals (Fe, Cu, Ni, Y, etc.) as dopants [7][8][9][10]. Doping with lanthanides (Ce, Er) can not only improve the sensor response, but it also reduces the operating temperature [11,12]. Decoration with noble metals (Pd, Pt, Au, etc.) allows a significant increase in the response and selectivity to certain gases due to chemical and electronic sensitization [13][14][15][16]. Creating nanocomposites with other metal oxides (SnO 2 , CuO, NiO, etc.) is also an effective method to improve purposes must be highly sensitive to ultra-low gas concentrations, selective, and have minimal dependence on humidity.
We found no data on the use of the AACVD process to produce ZnO/Fe 2 O 3 nanocomposites for chemical gas sensing, so the aim of this work was to obtain thin nanostructured ZnO/Fe 2 O 3 films by the AACVD method for chemoresistive gas sensors, as well as to study the influence of the Fe 2 O 3 amount on the gas-sensitive properties, including the detection of NO 2 , a biomarker of various diseases.  3 ] in methanol CH 3 OH (>99%, Chimmed) were used in the AACVD synthesis. Zinc and iron acetylacetonates were synthesized using zinc nitrate Zn(NO 3 ) 2 6H 2 O (>98%, "Reachem"), iron chloride, FeCl 3 , 6H 2 O (>98%, "Reachem"), and acetylacetone (>98%, "Ecos-1") by neutralization of their solutions with 5% ammonia hydrate solution (NH3·H2O, >99%, "Ecos-1"). ZnO films were obtained using a special laboratory setup consisting of a gas flow meter, an ultrasonic generator (Albedo IN-7), a furnace with a flow chamber, and an aerosol trapping system. To maintain constant conditions of aerosol formation, flow-through water cooling of the ultrasonic generator was performed. Aerosol deposition was performed in a flow furnace chamber on various substrates (glass, aluminum oxide, and an Al 2 O 3 sensor with platinum counter-pin electrodes and a microheater on the reverse side [40]) at a decomposition temperature of 400 • C. Nitrogen (high purity grade "6.0") was used as a carrier gas, with a flow rate of 300 mL/min. At the first stage, using zinc acetylacetonate solution, ZnO film was obtained (sample Z) with a deposition time of 60 min. In the second step, a Fe 2 O 3 film was deposited on the surface of the ZnO film using iron acetylacetonate solution for 10, 20, and 30 min at 400 • C, which are called Z1Fe, Z2Fe, and Z3Fe, respectively, in the further text. For additional analysis by Raman spectroscopy, a film of individual Fe 2 O 3 was synthesized on an Al 2 O 3 substrate under the following conditions: sputtering time of 45 min at 400 • C. The scheme of the laboratory setup used for AACVD synthesis of ZnO/Fe 2 O 3 thin nanostructured films used in this work is shown in Figure 1. After synthesis, additional heat treatment of the coatings was performed at 350 • C for 2 h in the air for additional removal of organic compounds from the surface of the films.

Instrumentation
XRD patterns of the coatings on glass substrates were recorded on a D8 Advance X-ray diffractometer (Bruker) with CuKα emission in the range of 2θ 25 • -65 • in 0.02 • increments with signal accumulation at the point for 1.8 s. The morphology and microstructure of the films on Al 2 O 3 substrates were studied using a NVision 40 three-beam workstation (Carl Zeiss). Electronic UV transmission spectra of ZnO/Fe 2 O 3 films on glass substrates were recorded using a SF-56 UV-Vis spectrophotometer. Raman spectra were obtained with a Renishaw inVia Reflex spectrometer using a 405 nm diode laser as the excitation source. All spectra were recorded in the 100-2000 cm −1 range with a spectral resolution of~3 cm −1 through a 50× (NA 0.5, FN 26.5) magnification lens, with an irradiated site diameter of 2 µm. The incident radiation power was less than 2.5 mW. Grating: 2400, and the signal accumulation time was 300 s.
The gas-sensitive properties were measured on a specialized precision setup [41][42][43]. The gas environment in the quartz cell was created using four Bronkhorst gas flow controllers with maximum flow rates of 1, 50, 100, and 200 mL/min. The temperature of the sensor element was regulated using a built-in platinum microheater pre-calibrated using a Testo 868 thermal imager. The resulting film was studied for sensitivity to the following analyte gases: CO, H 2 , CH 4 , NH 3 , benzene, acetone, ethanol, and NO 2 . The corresponding test gas mixtures in air were used as a source of the analyzed gases, and synthetic air was used to construct a baseline. To measure the signal at different relative humidity (RH), we used a special unit with a bubbler, and the RH of the gas mixture was controlled by a digital flow-through hygrometer «Excis». Electrical resistance of oxide films was measured using a Fluke 8846A (6.5 Digit Precision Multimeter) with an upper limit of 1 GOhm. The response to CO, H 2 , CH 4 , NH 3 , benzene, acetone, and ethanol was calculated using the formula: where R Air is the resistance of the oxide film in the synthetic air medium; R-in the environment with a given concentration of the analyte gas. The response to NO 2 was calculated by the inverse relation (1). The selectivity coefficient was calculated by the following equation: where S NO2 is the response when detecting 2 ppm NO 2 ; S 2 is the response to a given gas concentration (in our case it was 100 ppm ethanol or acetone), to which the highest response value after NO 2 was observed.
The following ratio was used to calculate «response drop» (RD) in a humid atmosphere where R 0 is the response at 0% RH, R RH is the response at the corresponding RH.

Chemical and Phase Composition
The chemical composition of the obtained ZnO/Fe 2 O 3 films was studied by energy dispersive X-ray spectroscopy (EDX). The EDX mapping of zinc and iron shows ( Figure 2) that both metals are evenly distributed on the surface of the obtained films. The calculated EDX values showed that the iron content in Z1Fe, Z2Fe and Z3Fe films was 1.3, 4.8 and 5.8 at.%, respectively. Thus, when the sputtering time of the Fe 2 O 3 film increases from 10 to 30 min, the iron content of the ZnO/Fe 2 O 3 nanocomposite films increases proportionally. Due to the small thickness of the synthesized films, the diffractograms obtained for films on polycrystalline Al 2 O 3 substrates are uninformative. Against the background of intense substrate reflexes, low-intensity film reflexes are practically indiscernible. Therefore, X-ray phase analysis of ZnO/Fe 2 O 3 thin films was performed on X-ray amorphous glass substrates. According to the data of X-ray phase analysis (Figure 3a) of ZnO/Fe 2 O 3 thin films on glass substrates it was found that, for all samples, the presence of characteristic reflexes of hexagonal ZnO wurtzite phase (spatial group P63mc, PDF 01-070-8070) is observed [44]. The most intense reflex (002) is observed at 34.4 • 2θ, which indicates a strong preferential orientation in the [001] direction. Crystallite growth variations are related to solvent polarity, so the intensity of reflex (002) may be due to the use of methanol, which has one of the highest polarities of all organic solvents, in the AACVD process [30]. The mean sizes of ZnO crystallites calculated by the Scherrer formula for samples Z, Z1Fe, Z2Fe, and Z3Fe were 40, 42, 30, and 35 nm, respectively. No separate iron-containing phases were detected on the diffractograms, which may be due to the low concentration of iron oxide, the small thickness of the Fe 2 O 3 layer, and the instrumental feature of the diffractometer in which the copper tube was installed. It is also worth noting that, for samples Z1Fe, Z2Fe, and Z3Fe, no shift in reflexes relative to individual ZnO is observed, which is an indirect sign of the formation of a two-phase nanocomposite rather than a solid solution.
Additionally, synthesized ZnO/Fe 2 O 3 thin films on Al 2 O 3 substrates were studied by Raman spectroscopy. As can be seen from Figure 3b, besides the peaks related to the α-Al 2 O 3 substrate (A1g at 418 and 644 cm −1 and Eg at 378, 430, 450, 576 and 749 cm −1 ) [45], an intense ZnO peak at 577 cm −1 , corresponding to A1(LO) mode, is observed in all samples. In addition, peaks assigned to the (E 2 high -E 2 low ), E 2 high , 2TO, and [2A1 (LO), E1(LO), 2LO] modes were observed at 332, 438, 982, and 1153 cm −1 , respectively [46]. Two peaks at 1356 and 1607 cm −1 can be attributed to the D-and G-bands of residual carbon that remained after the synthesis. Figure 3c shows the spectrum of the individual iron oxide film. As can be seen, besides the α-Al 2 O 3 substrate peaks, bands at 226, 501 (A 1g ) and 296 (E g ) cm −1 with low intensity, characteristic of α-Fe 2 O 3 (trigonal syngony, hematite phase, space group R3c), are observed [47]. Figure 3d shows that the α-Fe 2 O 3 peaks at 226 and 296 cm −1 in the spectra of Z1Fe, Z2Fe, and Z3Fe samples are also low in intensity. The weak signal from α-Fe 2 O 3 may indicate the low concentration and thickness of the iron oxide layer. Slightly more intense peaks are observed for the Z3Fe sample due to the higher concentration of iron oxide. The obtained peaks related to Fe 2 O 3 , although low-intensity, allow us to establish the formation of the ZnO/α-Fe 2 O 3 nanocomposite.  Figure 4a shows an overview XPS spectrum of Z and Z3Fe samples. As can be seen, the spectrum contains signals of the element peaks: Zn, O, and C (for the sample ZnO), as well as weakly intense signals from Fe (for the sample Z3Fe). It is likely that carbon is in the composition of CO 2 sorbed on the surface of the materials, as well as residual products after the synthesis. Figure 4b shows the XPS spectra of Zn 2p. Two peaks of Zn 2p 1/2 and Zn 2p 3/2 can be seen on the plot: at 1045.4 and 1022.3 eV, 1045.2 and 1022.1 eV for ZnO, and Z3Fe samples, respectively. The spin-orbit splitting was 23.1 eV, which is typical for ZnO with the wurtzite structure and agrees well with the literature data [48,49]. Figure 4c shows the XPS spectra of the Fe 2p sample Z3Fe. Two pronounced peaks of Fe 2p 1/2 and Fe 2p 3/2 at 723.5 and 710.3 eV can be seen on the plot. The signals obtained can equally be attributed to both Fe 2 O 3 and Fe 3 O 4 [50]. The main difference between Fe 2 O 3 and Fe 3 O 4 on the XPS spectra is the presence of a satellite peak for Fe 2 O 3 between Fe 2p 1/2 and Fe 2p 3/2 [51,52], which was also observed in our case at 717.5 eV. Therefore, based on the XPS data obtained, we can assert the formation of the Fe 2 O 3 phase within the thin films, which agrees well with the Raman spectroscopy data.   [53,54]. When Fe 2 O 3 is modified, the maxima of the main peaks of all elements are shifted by 0.2-0.5 eV, which may be related to the effect of the modifying additive and the formation of the Schottky barrier in the nanocomposite.

Microstructure and Morphology
The microstructure and dispersity of ZnO/Fe 2 O 3 thin films on Al 2 O 3 substrates were studied by scanning electron microscopy (SEM). As can be seen ( Figure 5), all films uniformly cover the substrate surface, forming a continuous coating. The lower layer of ZnO in all films consists of spherical nanoparticles. In sample Z, the size of ZnO nanoparticles is 66 ± 13 nm, Z1Fe: 50 ± 14 nm, and Z2Fe: 51 ± 10 nm, and, in Z3Fe, the iron oxide covers the ZnO particles tightly, due to which it is not possible to calculate their size. The data obtained correlate well with the calculated crystallite values estimated using the Scherrer method on the obtained X-ray diffraction patterns. Z2Fe and Z3Fe samples show Fe 2 O 3 nanoparticles on the ZnO surface, their size in the Z2Fe sample being 23 ± 4 nm, and, with increasing sputtering time (Z3Fe), their size increases to 28 ± 5 nm. Due to the low concentration of iron in the Z1Fe sample, it is not possible to estimate the size of Fe 2 O 3 nanoparticles.    Table 1 summarizes the properties of the obtained ZnO/Fe 2 O 3 nanostructured thin films.

Selectivity
The obtained ZnO/Fe 2 O 3 films showed good electrical conductivity, which allowed us to study chemoresistance responses to various gases (H 2 , CH 4 , CO, C 6 H 6 , NH 3 , ethanol, acetone, and NO 2 ) in a wide temperature range: from 150 to 300 • C (Figure 8). At lower operating temperatures, the resistance of the nanocomposite films was high enough that it was not possible to evaluate the gas sensitivity on the equipment used in the present work (electrical resistance measurements were limited to 1 GOhm). At elevated temperatures, the greatest response for all analyzed films was observed for VOCs: ethanol and acetone. The greatest sensitivity to the above gases was observed at 300 • C, with response values being: 11.2, 13.6, and 11.8 (100 ppm ethanol), as well as 12.1, 13.0, and 14.9 (100 ppm acetone) for samples Z, Z1Fe, and Z2Fe, respectively. For the sample with the highest iron content (Z3Fe), the maximum response (11.8 and 12.3 for ethanol and acetone, respectively) was observed at 275 • C. The response to other gases at 275-300 • C did not exceed 7.4, 8.7, 4.9, and 4.7 for Z, Z1Fe, Z2Fe, and Z3Fe, respectively. Upon lowering the operating temperature to 150 • C, a significant decrease in the responses to ethanol and acetone (to 2.1-3.2) was observed. When detecting NO 2 , a different dependence was observed: when the operating temperature was changed from 300 to 150 • C, the response significantly increased. Such phenomena are quite typical for ZnO-based receptor nanomaterials, which are characterized by increased sensitivity to various VOCs at elevated temperatures and to NO 2 at moderate ones [58,59].
To select the optimal temperature for the detection of the analyte gas, several factors should be taken into account, among which the main ones are, definitely, selectivity and energy efficiency, which are determined by the operating temperatures at which the detection is performed. When analyzing the obtained array of responses of ZnO/Fe 2 O 3 thin films to various gases at 150-300 • C, a low working temperature of 175 • C was chosen, and pronounced sensitivity to ultra-low NO 2 concentrations was observed at this point. Next, the chemoresistive gas-sensitive properties upon NO 2 detection at 175 • C were studied in more detail (Figure 9). All samples of ZnO/Fe 2 O 3 thin films showed unusually high sensitivity to NO 2 . An increase in electrical resistance (which is typical for n-type MOS) was observed when detecting NO 2 . When the concentration of NO 2 increases above 2 ppm, the resistance upon response recording exceeds 1 GOhm, which does not allow its detection on the equipment used. Therefore, all further measurements upon NO 2 detection were limited to the concentration of 2 ppm. The sample (Z1Fe) with the lowest iron content had the highest response (S = 14.2) and selectivity (Sel = 3.4) when detecting 2 ppm NO 2 . For other samples, the responses to 2 ppm NO 2 are 4.0, 6.1, and 2.6, and the selectivity coefficients were 1.2, 1.7, and 0.6 for Z, Z2Fe, and Z3Fe, respectively. Thus, the sample (Z1Fe) with the lowest iron content has the highest selectivity at 2 ppm NO 2 . It should be noted that the comparison of responses to NO 2 with other gases was carried out at their various concentrations, which differ by 50 times or more (100 ppm vs. 2 ppm NO 2 ). Therefore, the obtained value (Sel = 3.4) for Z1Fe, as well as for other samples, is an extremely high value, showing good selectivity for NO 2 detection. Figure 10a shows the experimental data on the sensitivity to different NO 2 concentrations of Z, Z1Fe, Z2Fe, and Z3Fe samples. The zinc oxide-based films show a high response in a wide range of ultra-low NO 2 concentrations (300-2000 ppb), covering the threshold limit value (TLV) recommended by the National Institute for Occupational Safety and Health [60]. The response increases from 1.3 to 4.0, 4.0-14.2, 1.9-6.1, and 1.3-2.6 with increasing NO 2 concentrations from 300 to 2000 ppb for Z, Z1Fe, Z2Fe, and Z3Fe, respectively.  Figure 10b shows the dependences of the response value on the concentration of NO 2 in the gas mixture. The highest response over the entire NO 2 concentration range was observed for the Z1Fe sample. All samples can be described well by a linear function. This dependence is typical for many chemoresistive gas sensors and correlates well with the available literature data [1,61]. Figure 10c shows the reproducibility of the signal when detecting 500 ppb NO 2 at 0%RH. It can be seen that, for individual ZnO, there is a drift in the values. With the addition of iron, the signal becomes more stable with minimal drift of the resistance values.
The kinetic characteristics were evaluated: the response time changes nonlinearly with increasing analyte concentration, and, for concentrations of 300-2000 ppb, these values amount to 143-180, 118-203, and 121-160, as well as 58-176 s for Z, Z1Fe, Z2Fe, and Z3Fe, respectively. Figure 11 shows the responses of samples Z, Z1Fe, Z2Fe, and Z3Fe when detecting 500 ppb NO 2 at different relative humidity values (RH). For individual ZnO, when the RH increases from 0 to 75%, a consistent decrease in the response value from 1.8 to 1.5 is observed (Figure 11a). The shape of the signal has a sharply pointed appearance, which may indicate that the equilibrium on the surface has not been established, which is due to the long process of establishing a stable signal. When doped with iron oxide, the shape of the responses becomes close to rectangular. For the sample Z1Fe (Figure 11b), the smallest influence of humidity on the signal is observed. At RH in the 0-75% range, the response decreases linearly from 4.6 to 4.1. The Z2Fe sample (Figure 11c) shows a response of 3.2-2.4 with a general trend of decreasing response with increasing humidity, but the signal at 50% RH is slightly greater than at 25% RH. A different behavior is exhibited by Z3Fe (Figure 8d), for which the response in a humid environment (2.3-2.5) is higher than in a dry atmosphere (1.8).  Figure 12a summarizes the data on the effect of humidity on the response value of the obtained thin films. For samples Z, Z1Fe, and Z2Fe, the response dependence on RH can be described by the curve with a negative slope. The sample Z3Fe showed an increase in the response at 25% RH, and then a consistent decrease in its value was observed. Figure 12b shows the modulus of change in the response value relative to its value in dry air. As can be seen, the sample Z1Fe is the least affected by humidity among the samples, and the sample Z3Fe has the maximum sensitivity to humidity.

Detection Mechanism
The resistive response in the detection of NO 2 by individual ZnO results from various reactions, which can be described using the model for n-type semiconductors generally accepted in the literature. In an air environment at elevated temperatures, oxygen molecules adsorb onto the semiconductor surface, which leads to a change in the resistance of the material. Electrons from the conduction zone reduce O 2 to ionic form (at 175 • C, O − particles would be predominant) [62]. The presence of such ions on the ZnO surface contributes to the formation of the electronic structure of the core-shell type. The core is the inner region of the semiconductor particle, and the shell is the electron depletion layer (EDL) formed as a result of electron consumption during the reduction of O 2 to O − .
In the work by Mei Chen et al. [63], the process of interaction of NO 2 with zinc oxide was studied using a number of methods of in situ physicochemical analysis. The authors were able to show that, upon the intake of NO 2  ion [64]. In our previous work [59], while studying the thick zinc oxide films obtained by solvothermal method, we correlated the Mei Chen data with the results we obtained in the study of chemoresistive responses in NO 2 detection and summarized them. More details on the mechanisms of NO 2 detection by individual ZnO can be found in our work at the above reference.
Thus, based on the above data, we can conclude that the maximum response to NO 2 of an individual ZnO thin film at 175 • C can be explained by the maximum intensity of NO 2 adsorption in this temperature range. It is worth noting that, just as in our previous work, where a thick ZnO film was obtained by the solvothermal method, so in the present work, where a thin ZnO film was obtained by the AACVD method, zinc acetylacetonate was used as a precursor. In both cases, extremely NO 2 -sensitive receptor nanomaterials, based on ZnO, were obtained using completely different methods. Therefore, the choice of the precursor used probably also plays an important role in obtaining a highly sensitive receptor nanomaterial for chemoresistive gas sensors.
The highest response over the entire NO 2 concentration range for the Z1Fe sample can be attributed to the formation of a heterojunction ( Figure 13). The Fermi level of Fe 2 O 3 is lower than that of ZnO, so there is directional electron transfer from the conduction zone of ZnO to Fe 2 O 3 until level alignment is achieved. Thus, an electron-depleted layer is formed at the boundary of the two semiconductors, which leads to significant chemical and electronic sensitization. In addition, the Fe 2 O 3 nanoparticles can interact with NO 2 , which leads to the electron transfer from ZnO to Fe 2 O 3 and widening of the electron-depleted ZnO layer [65]. The obtained cluster structure of Fe 2 O 3 nanoparticles on the ZnO surface (sample Z1Fe) is suitable for the realization of the heterojunction, as described above, allowing us to improve the chemoresistive properties. As the thickness of the Fe 2 O 3 layer increases, the sensitivity to NO 2 decreases as the area covered by the Fe 2 O 3 nanoparticles increases. Zinc oxide stops being the main gas-sensitive material, and the heterojunction effect ceases to be the limiting one. In humid atmosphere, water molecules sorbed on the ZnO surface additionally participate in the detection mechanism. In many cases, humidity has a negative effect on the chemoresistive response and remains one of the main problems in gas sensing. After sorption, the H 2 O(ads) molecules react with point defects (zinc Zn(lat) and oxygen O(lat) in the crystal lattice, resulting in the formation of hydroxyl groups on the ZnO surface [3,66]. The released electrons enter the conduction band of ZnO, which leads to decrease in electrical resistance. The formed hydroxyl groups occupy the active adsorption centers on the ZnO surface, thus reducing the adsorption of NO 2 (ads) and O − (ads) and the intensity of the reactions, which contribute to a chemoresistive response.
For the sample Z1Fe with the Fe 2 O 3 cluster structure, moisture has almost no effect on the response value. Additionally, with further increase in the thickness of the Fe 2 O 3 layer, on the contrary, the largest drop in the response is observed. It can be assumed that, when doping with Fe 2 O 3 , the observed phenomena are related to the participation of iron cations directly on the surface of the ZnO film in the detection mechanism. Iron in oxides is known in two stable oxidation degrees: +2 and +3, which correspond to the oxides FeO, Fe 2 O 3 , and Fe 3 O 4 . After sorption of water molecules, we can assume the following reactions of iron cations with hydroxyl groups: Thus, a small iron content contributes to leveling the influence of moisture. With further increase in the iron content, reactions (3)(4) begin to compete with those resulting in a response to NO 2 , which leads to an increase in the response drop. Nevertheless, this mechanism has not been experimentally confirmed and requires further study.

ZnO/Fe 2 O 3 Nanostructured Thin Films As Gas Biosensors
The obtained ZnO/Fe 2 O 3 thin-film nanocomposites showed extremely high sensitivity to ultra-low NO 2 concentrations (300-2000 ppb). Nitrogen dioxide is a gas biomarker for diseases associated with inflammatory processes in the lungs: bronchial asthma, bronchiectasis, and chronic obstructive pulmonary disease (COPD) [32,36,37]. The concentration of NOx in the exhaled air of a sick person, compared to the exhaled air of a healthy person, differs at the level of tens of ppb. Thus, the detection limit of NO 2 by the ZnO/Fe 2 O 3 thin-film nanocomposites obtained in the present work is not only much lower than the TLV, which is important in the context of environmental safety, but is also in a suitable range for medical purposes. Therefore, ZnO/Fe 2 O 3 nanocomposites can be effective receptor materials in the composition of analytical devices for noninvasive diagnosis of diseases related to lung disease.
Additionally, to enable selective signal separation, we used principal component analysis (PCA) to evaluate the correlation between the response of the four receptor nanomaterials used (Z, Z1Fe, Z2Fe and Z3Fe) and selectivity. Chemometric analysis is quite common in multisensor and E-nose systems for the detection of a wide group of analyte gases, including medical applications. Figure 14 shows the data obtained after mathematical processing of the PCA responses using the covariance matrix. As can be seen, the resulting graph consists of several distant clusters: 300-2000 ppb NO 2 , 100 ppm C 2 H 5 OH, C 3 H 6 O, and CO, and 1000 ppm CH 4 , H2, 100 ppm C 6 H 6 , and NH 3 (inset), which allows their concentrations to be determined with high accuracy. Thus, using a sensor array consisting of four thin-film ZnO/Fe 2 O 3 nanocomposites, signals can be separated with great selectivity when detecting the eight diverse analyte gases used in this work. Such results are extremely important in the transition to multisensor devices that are used for noninvasive disease diagnosis, where it is necessary to analyze a wide group of gases contained in human exhaled air.  Table 2 compares some gas-sensitive characteristics of ZnO/Fe 2 O 3 nanocomposites with analogues presented in the literature. It is worth noting that the ZnO/Fe 2 O 3 thin films obtained in our work have an improved set of gas-sensing properties: moderately low detection temperature, low detection limit, high response, and good selectivity.

Conclusions
ZnO/α-Fe 2 O 3 nanocomposites were prepared by two-step aerosol vapor deposition (AACVD). Zinc and iron acetylacetonates in methanol were used as precursors, and synthesis was carried out at 400 • C with sequential sputtering of ZnO (60 min) and Fe 2 O 3 (10, 20, 30 min). The iron content of the films with sputtering times of 10, 20, and 30 min was found to be 1.3, 4.8, and 5.8 at%, respectively, according to EDX data. The phase composition was determined by Raman spectroscopy, XPS and XRD were used for analysis, and the morphology was determined by SEM. ZnO particles have a wurtzite structure with an average size of 51-66 nm, and iron oxide particles with hematite structure on the surface of ZnO have an average size of 23-28 nm. All films have a large transmittance in the visible range. The optical band gap calculated by the Tauc method decreases with increasing Fe 2 O 3 concentration from 3.26 to 3.16 eV. The films obtained showed a high response to 100 ppm acetone and ethanol at 225-300 • C and a high and selective response to 300-2000 ppb NO 2 at 175 • C. The effect of humidity on the magnitude and shape of the signal obtained in NO 2 detection was studied. The sensor with 1.3 at% Fe has the highest response, selectivity coefficient, and it has lower dependence on humidity, which can be explained by the formation of a heterojunction at the boundary of the two semiconductors.

Conflicts of Interest:
The authors declare no conflict of interest.