The structure, morphology, conductivity, and sensor response to hydrogen of ZnO-In2O3 composites obtained by the three approaches described in the previous section were determined by modern physicochemical methods. The analysis of these data enables the determination of the effect of the synthesis method on the phase composition, structural characteristics, as well as the conductive and sensor properties of the composites.
3.1. Structural Characteristics of ZnO-In2O3 Composites
The XRD spectra of ZnO-In
2O
3 composites obtained by the three synthesis methods are presented in
Figure 1. Analysis of the XRD spectra of composites obtained by mixing powders and impregnation methods shows that the two synthesis methods lead to the formation of two-phase composites—ZnO-In
2O
3. Diffraction peaks at 21.5, 30.6, 35.5, 37.7, 41.8, 45.7, 51.0, 55.9, and 60.7° correspond to planes (211), (222), (400), (411), (332), (431), (440), (611), and (622) of indium oxide, respectively. Peaks at 31.7, 34.1, 36.1, 47.3, 56.5, 62.7, and 67.7° can be attributed to the (100), (002), (101), (102), (110), (103), and (112) planes of the oxide zinc, respectively. For composite ZnO-In
2O
3 obtained by mixing powders and impregnation methods, all diffraction peaks correspond to the cubic structure of bixbyite In
2O
3 (JCPDS no. 06-0416) and hexagonal ZnO with a wurtzite structure (JCPDS no. 36-1451).
On the XRD spectra of ZnO-In
2O
3 composites synthesized by the hydrothermal method, peaks were recorded that relate to both the cubic and rhombohedral phases of In
2O
3 (
Figure 1). The data were compliant with JCPDS #22-0336. A similar pattern was observed in the ZnO-In
2O
3 and In
2O
3-ZnO systems synthesized by the two-step chemical precipitation methods [
27]. There are no diffraction peaks of ZnO in the XRD spectra of hydrothermal composites containing up to 20% ZnO. This may be due either to the introduction of Zn ions into the crystal lattice of indium oxide or the formation of an X-ray amorphous oxide phase. Our previous studies have shown that SnO
2-In
2O
3 composites also exhibit this behavior [
10].
The rhombohedral phase of In
2O
3 is indexed on XRD spectra only at concentrations of 10, 20, and 40% ZnO in the composites. The formation of the rhombohedral phase of In
2O
3 occurs due to the partial dissolution of Zn ions, which have a smaller size than In ions, in the lattice of basic cubic In
2O
3 crystals. In addition, this phase has concentrations of 15, 22, and 9 wt.% for 10, 20, and 40% ZnO, respectively. Interestingly, the concentration of the rhombohedral phase of indium oxide is maximum in hydrothermal composites containing 20% ZnO. Also, only the cubic phase of indium oxide is observed in the composites with a 65% ZnO concentration. It has been shown that the ratio between the rhombohedral and cubic phases of indium oxide can be controlled by the concentration of zinc in structures synthesized by a simple solvothermic method [
32].
In the spectra of ZnO-In2O3 composites containing 20 to 85% ZnO, peaks appear corresponding to reflexes of ZnO with a wurtzite structure (JCPDS no. 36-1451). The compositions of zinc oxide in these composites, calculated by the Rietveld method, are 9.7, 29.5, 64.9, and 100%. Note that the concentration calculated from the XRD data corresponds to that used for synthesis only for the 65%ZnO-35%In2O3 composite. It can be assumed that the ZnO-In2O3 heterostructure is formed only at this concentration. At other concentrations of zinc, one component is introduced into the structure of another. The absence of any extraneous peaks indicates the high purity and crystallinity of the samples synthesized by the three methods used in this study.
In contrast to mixed and impregnated composites, the main peak of indium oxide characterizing the plane (222) in hydrothermal samples is located in the region of large angles compared with the situation in pure In
2O
3 (see
Figure 1). The observed displacement of the indium oxide peak (222) to large angles indicates that zinc ions are embedded in the crystal lattice of indium oxide. In addition, the lattice parameter in hydrothermal composites with concentrations ranging from 0 to 65% ZnO decreases from 1.013 nm to 1.089 nm. This is because the size of the ionic radius of indium (0.81 Å) exceeds the radius of zinc (0.74 Å). It should be noted that in the 65%ZnO-35%In
2O
3 composite, the In
2O
3 lattice parameter reverts to the value of the initial indium oxide.
The particle size of zinc oxide calculated by the Debye-Scherrer formula is 45–50 nm in mixed composites and 10–30 nm in impregnated samples. The size of these nanoparticles in mixed composites does not depend on the zinc oxide concentration. On the contrary, the particle size of ZnO in impregnated composites decreases from 30 to 10 nm as its concentration in the mixture with indium oxide increases. It is probable that with an increase in the concentration of zinc salt, a large number of crystallization centers appear in the process of impregnation, which prevent the growth of embryos and, consequently, the formation of large crystals. In addition, the size of indium oxide nanoparticles in samples prepared by these two methods was calculated from the data from X-ray diffraction analysis. Calculations have shown that the size of indium oxide particles is approximately 70 nm and does not depend on the zinc oxide concentration in such samples.
According to XRD data, an increase in the concentration of zinc oxide in hydrothermal composites is accompanied by a decrease in the size of indium oxide nanoparticles from 25 to 9 nm. This result indicates that during synthesis, Zn atoms replace In atoms in the lattice, which creates deformations in the composites [
33]. Thus, the introduction of Zn ions into the In
2O
3 structure hinders crystal growth. In composites containing from 20 to 85 wt.% zinc oxide, the particle size of ZnO is almost three times larger than that of In
2O
3 and decreases from 72 to 25 nm. The observed difference in the sizes of nanoparticles of zinc oxide and indium oxide may indicate that zinc oxide completely covers the surface of indium oxide during the formation of crystalline zinc oxide in hydrothermal composites, resulting in the formation of a core-shell structure.
According to TEM data, the mixed ZnO-In
2O
3 nanocomposites are a mixture of nanocrystals of the corresponding oxides, have a diverse shape, and are characterized by a wide particle size distribution from 50 to 100 nm, which practically coincides with the particle sizes of the initial commercial nanopowders (
Figure 2). Another picture is observed for impregnated composites. As a result of impregnation of indium oxide with zinc nitrate and subsequent heat treatment, small ZnO nanoparticles of almost spherical shape appear on the surface of relatively large indium oxide nanocrystals (up to 100 nm) (
Figure 2).
The particle size varies from 5 to 30 nm, depending on the zinc oxide concentration in the composite. Thus, in impregnated samples of ZnO-In
2O
3, the particle size is approximately 20–30 nm for zinc oxide concentrations of 10–60%, but 5–10 nm for ZnO concentrations of more than 60%. According to XRD data, zinc ions are introduced into the indium oxide structure in hydrothermal composites. This result is confirmed by SEM and EDX data. The 5%ZnO-95%In
2O
3 hydrothermal composite consists of indium oxide particles with an inhomogeneous, loose structure in which zinc ions are fairly evenly distributed (
Figure 2).
To analyze the specific surface area and porosity of the synthesized samples, nitrogen adsorption-desorption experiments were carried out at a temperature of 77 K (see
Figure 3). According to the IUPAC classification, the adsorption-desorption isotherms of the hydrothermal composite have a type IV shape with a hysteresis loop H3. Such a shape is typical of the mesoporous nature of the samples. At the same time, the desorption and adsorption isotherms for samples obtained by mixing nanopowders and the impregnation method (
Figure 3) have a form characteristic of macroporous samples.
The surface area calculated by the BET method was 25 m
2/g for hydrothermal samples and five times less for mixed and impregnated samples (
Figure 3). To study the mesoporous structure of hydrothermal samples, we used the BJH method for the desorption branch of the isotherm. The pore size of hydrothermal samples is 18–20 nm. It should be noted that the introduction of zinc oxide leads to an increase in the specific surface area of the sample, the volume of pores, and the pore size. The largest specific surface area (32 m
2/g) is observed for the hydrothermal composite containing 65% ZnO. A higher specific surface area and a smaller average pore size of hydrothermal composites compared with mixed and impregnated composites facilitate the adsorption of O
2 and the analyzed gas, which should lead to an acceleration of the response-recovery processes during gas detection.
3.2. Conductivity of ZnO-In2O3 Composites
The temperature dependence of the conductivity in air of composites obtained by the various methods investigated exhibits the same characteristics. In the range of 300 to 520 °C, the conductivity of all synthesized systems increases with rising temperature, which is typical for n-type semiconductors. In all the cases considered, the temperature dependence of the resistance of synthesized composites is not described by the Arrhenius law. The deviation from the Arrhenius law is obviously due to the fact that an increase in temperature leads to the desorption of chemisorbed active oxygen particles located on the surface of the sensitive layer.
At the same time, as can be seen from the experimental data shown in
Figure 4, the character of the dependence of the ZnO-In
2O
3 composite conductivity on the concentration of zinc oxide is largely determined by the synthesis method.
The nature of the change in the resistance of the impregnated composite, depending on the composition, differs little from the situation observed for the mixed composite. Specifically, the resistance in these systems increases sharply with an increase in the ZnO concentration. An increase in resistance is also observed in ZnO-In
2O
3 systems when a heterojunction occurs [
31,
34]. Impregnated and mixed composites contain phases of indium oxide and zinc oxide, i.e., such nanocomposite films, as well as films of individual oxides, have an electronic type of conductivity.
Due to the large difference in the values of the electron work function between In
2O
3 and ZnO (4.3 eV and 5.2 eV, respectively), electrons are transferred from In
2O
3 to ZnO nanocrystals at contacts between nanocrystals of these oxides. This circumstance should be taken into account when considering the conductivity process in this system using the percolation model [
35].
As a result of the percolation transition, the path of the current flow changes. The conducting particles become nanoparticles of zinc oxide rather than indium oxide nanoparticles. Such a transition causes a sharp decrease in conductivity. It is important to note that the percolation transition in the systems investigated depends on the method used for sample preparation and occurs at different concentrations of zinc oxide. While in order to change the current path, the impregnated composite must contain 40 wt.% ZnO, the percolation transition in a composite formed by mixing nanopowders is observed at a concentration of at least 65 wt.% zinc oxide. This difference is mainly because the size of ZnO nanoclusters formed on the surface of indium oxide does not exceed 10–30 nm. This size is several times smaller than that of ZnO nanoparticles in a mixed composite, the average size of which is approximately 70 nm. The specific surface area of ZnO and, consequently, the contact area between ZnO and In2O3 particles in the composites investigated differ even more.
All this implies that the transition of electrons from indium oxide to zinc oxide in contact with it occurs more efficiently in an impregnated composite than in a mixed composite. As a result of such a transition of conduction electrons, the current in the system will flow through aggregates of zinc oxide molecules with much lower conductivity. This trend also explains the observed decrease in the conductivity of the impregnated composite at a lower concentration of ZnO than in the mixed composite. With ZnO concentrations above 40%, the conductivity of the impregnated composite becomes significantly less than that of a mixed composite of the same composition in which current flows through In2O3 nanocrystals.
The curve of change in the resistance of a hydrothermal composite is fundamentally different from the curves observed for ZnO-In
2O
3 composites formed by mixing nanopowders and impregnation (
Figure 4). When discussing the data on the conductivity of hydrothermal composites, it is necessary to consider the fact that in such composites containing up to 20% zinc oxide, individual particles of this oxide are not formed. Indeed, only zinc ions embedded in the crystal structure of indium oxide are present in the system.
The electronic structure of indium oxide changes as a result of the introduction of zinc into the In
2O
3 structure, and an increase in In
2O
3 resistance is observed due to a decrease in the concentration of conduction electrons in this oxide (
Figure 4). The formation of “holes” in the system during the substitution of indium ions in the In
2O
3 lattice for zinc ions, which are acceptors, affects the growth of resistance in composites containing up to 20% zinc oxide. In this case, the electrons are compensated by holes generated by the acceptors, and the resistance consequently increases. Thus, the introduction of zinc into the In
2O
3 structure leads to the appearance of
p-type defects in the system, which results in an increase in the resistance of hydrothermal composites prepared from mixtures of oxides containing up to 20% ZnO. A similar result was obtained for Zn-doped In
2O
3 systems synthesized by the sol-gel method, for which enhancement of resistance was observed with an increase in Zn to 3 mol.% [
33].
In the structure of hydrothermal composites containing more than 20% ZnO, zinc oxide crystals appear, which, on interaction with modified In2O3 crystals, leads to a sharp decrease in the resistance of the composite. The substitution of In3+ ions for Zn2+ ions in the crystal structure of indium oxide is accompanied by the appearance of new oxygen vacancies in the system. A further increase in the ZnO concentration makes the structure even more unstable due to the large number of oxygen vacancies formed. This can lead to structural defects, such as the formation of a solid solution in the embedding. The change in the type of solid solution from substitution to introduction, which contributes to maintaining the stability of the cubic structure of In2O3, is accompanied by a decrease in the resistance of the composite.
It is important to note that the interaction between ZnO-In2O3 components plays a significant role in changing the structure and conductivity of the composites. This interaction is manifested to the greatest extent in impregnated and hydrothermal composites. In impregnated samples, specific contacts between indium and zinc oxides lead to a redistribution of electrons in the oxides and, as a result, a change in the conductivity of the system. On the other hand, the interaction of components in composites formed by the hydrothermal method is accompanied by the incorporation of zinc ions into the indium oxide lattice and also leads to a change in the phase composition of the system.
3.3. Sensor Properties of ZnO-In2O3 Composites
The sensor properties of ZnO-In
2O
3 composites also depend on the synthesis method and the observed differences in the interaction between the components. The character of the dependence of the maximum sensor response to hydrogen on the ZnO concentration in composites obtained by impregnation and mixing commercial oxide powders using screen-printing technology [
14] is nearly similar (for example,
Figure 4 shows the maximum response of composites to 1100 ppm H
2).
According to the current generally accepted model of sensor processes in the detection of reducing compounds, the sensor response is determined by two processes: the adsorption of oxygen and the analyzed gas on the surface of the sensor layer and the reaction of adsorbed compounds with oxygen centers O
– on this surface. The formation of O
– oxygen centers occurs during the dissociation of chemisorbed oxygen on the surface oxygen vacancies of a metal oxide nanoparticle. In the ZnO–In
2O
3 composites investigated, oxygen dissociation with the formation of chemisorbed oxygen atoms occurs mainly on the surface of ZnO nanoparticles. This is because ZnO has a higher catalytic activity than In
2O
3. The electrical current in both mixed and impregnated composites containing up to 80% ZnO flows through electron-rich In
2O
3 nanoparticles. The current flow path is determined not only by the ratio of component concentrations in the composite but, in accordance with the percolation theory, also depends on the size of oxide particles [
18,
35].
Upon contact between In2O3 and ZnO nanoparticles, oxygen atoms formed on the surface of ZnO nanoparticles pass to the surface of In2O3 nanoparticles, capturing conduction electrons and transforming into anion radicals, O-. These anion radicals participate in the detection reaction of reducing compounds, in particular H2. The ZnO also catalyzes the dissociation of adsorbed H2. Consequently, these nanocrystals are chemical sensitizers of the sensor reaction in the ZnO-In2O3 composite. The effect of ZnO on the sensor response in the ZnO-In2O3 nanocomposite can be attributed to the flow of H atoms from ZnO into In2O3 nanocrystals, where they react with O− and return electrons to the In2O3 conduction band. The chemical sensitization of the process is most effective in the composite when conducting aggregates of In2O3 are in contact with ZnO nanocrystals along the entire current path.
On the curve of the dependence of the maximum response of mixed and impregnated composites to 1100 ppm of hydrogen on composition, there are two maxima at zinc oxide concentrations of 15–20 and 85 wt.%. Note that the maximal response to ethanol in the ZnO-In
2O
3 system is also observed at 20% ZnO [
26]. The maximum sensor response of impregnated composites for all compositions is 30–50% higher than the response of composites obtained by mixing commercial nanopowders.
The main reason for the different sensor activity of the composites, as well as their structural characteristics discussed above, is that the size of zinc oxide nanoparticles in the mixed composite is several times larger than the size of ZnO nanoclusters deposited on the surface of In
2O
3 nanoparticles during impregnation. The low bond strength in the lattice of such clusters and their developed surfaces contribute to a decrease in the energy required for the formation of active oxygen vacancies. As a result, the concentration of active centers in the cluster significantly increases, which causes a higher sensor response when detecting hydrogen with an impregnated composite. The role of contacts between zinc and indium oxide nanoparticles having different electronic structures and morphologies was also observed in the detection of ethanol [
21,
25,
28].
In contrast to mixed and impregnated systems, several characteristic areas can be distinguished on the curve of the dependence of the maximum sensor response on the composition of hydrothermal composites ZnO-In
2O
3: 0–20 wt.%, 20–85 wt.%, and above 85 wt.% ZnO. These sites differ in the effect of zinc oxide on the sensor response of the composite. Note that a similar pattern was previously observed in SnO
2-In
2O
3 composites [
10]. The sensor response to hydrogen is reduced in composites containing up to 20 wt.% ZnO compared with pure indium oxide. According to X-ray data, no crystalline zinc oxide is formed in such composites, and zinc ions are embedded in the structure of indium oxide.
The magnitude of the sensor response of hydrothermal ZnO-In
2O
3 composites containing up to 20% zinc oxide is significantly influenced by the fact that during the synthesis of such composites, a change in the phase composition of In
2O
3, namely the transition of the cubic to the rhombohedral phase, occurs. Earlier, we found that the introduction of cerium into indium oxide, leading to its phase changes, causes a sharp drop in the sensor response to hydrogen compared with pure indium oxide [
36]. The introduction of zinc into cubic indium oxide also leads to an increase in the response to NO
2 compared with the undoped sample [
33].
An increase in zinc oxide concentration from 20 to 85 wt.% leads to a rise in the sensor response to hydrogen by almost two times. In this case, the formation of crystalline zinc oxide particles and the return of a stable cubic phase of indium oxide are already taking place. It should be noted that in mixed and impregnated composites, indium oxide has an exclusively cubic structure. That is, an increase in the sensor response to hydrogen in hydrothermal composites containing from 20 to 85% zinc oxide is due, on the one hand, to the formation of a heterojunction between ZnO and In2O3 and, on the other, to a decrease in the proportion of the rhombohedral phase of indium oxide.
Note that the composite containing 65% ZnO has the maximum sensor activity. According to the data obtained, only two phases (cubic indium oxide and hexagonal zinc oxide) are formed in such a system, and an increase in the sensory response is associated with chemical sensitization. A decrease in the sensor response of composites containing more than 85% zinc oxide is associated with the interruption of the flow of current between indium and zinc oxides. This is because nanoparticles of crystalline ZnO are formed on In2O3 during hydrothermal synthesis and, at such a concentration, completely cover the surface of indium oxide. Accordingly, the sensitivity of the nanocomposite decreases, approaching the sensitivity of nanocrystalline ZnO.
As for most metal oxide sensors, including mixed and impregnated ZnO-In
2O
3 composites, the temperature dependence of the sensor response of hydrothermal composites to hydrogen passes through the maximum. The maximum temperatures (T
max) in the detection of 1100 ppm of hydrogen for composites obtained by the hydrothermal method and the impregnation method are 100 and 125 °C, respectively, which are higher than for the mixed composite (
Figure 5).
The observed difference in T
max is because small ZnO nanoparticles in impregnated and hydrothermal composites contribute to an increase in sensor response, unlike large nanocrystals (larger than 50 nm) contained in mixed composites. This happens because of oxygen dissociation on the surface of nanoparticles, resulting in a large concentration of defects with catalytic activity. Such defects contribute to the formation of oxygen adatoms. The bond of the resulting atoms with defects leads to an increase in the energy required for oxygen desorption and its transition from ZnO particles to In
2O
3 particles, which determines the conductivity of the composite. Therefore, the sensitization of the sensor response by ZnO nanoparticles is manifested in such composites at higher temperatures than in mixtures of commercial nanopowders and is more effective due to the high concentration of defects [
11].
The main disadvantage of metal oxide sensors is their insufficient selectivity with respect to cross-sensitive gases. In this regard, we studied the influence of the method of obtaining ZnO-In
2O
3 composites on their selectivity in the detection of H
2 with respect to CO.
Table 1 shows the values of the sensor response of the composites, with the maximum response to hydrogen upon detection of 1100 ppm H
2 and CO.
The data presented show that although the composite synthesized by the hydrothermal method has the smallest value of the maximum response to both gases, its selectivity, i.e., the S
H2/S
CO ratio, significantly exceeds the selectivity of composites obtained by mixing oxide nanopowders or by the impregnation method. Such a difference in the selectivity of the samples obtained by different methods is probably due to the change in the structures of the resulting composites and, as a consequence, the sorption activity of the metal oxides included in their composition. The high selectivity of the hydrothermal composite may also be associated with the formation of
p-type defects [
37].