1. Introduction
Metal oxides are produced in various forms for applications across many industries. The metal oxide may be produced by several methods, such as thermal, liquid-phase, and physical milling methods. Photocatalytic experiments on the bleaching of Prussian blue by ZnO were first carried out in 1924. The observations inspired many researchers to use ZnO as a photocatalyst for other reactions, such as the reduction of Ag
+ to Ag [
1]. Subsequently, metal-oxide-based photocatalysts were reported several times, but they were not included in the list of light-sensitive photocatalysts, and many studies were not conducted due to limitations in practical application [
2,
3,
4]. However, in the 1970s, the conception of photocatalysts began to change. Firstly, as we progress through the industrial age, research into alternative energy has become urgent due to the reckless use of fossil fuels. In addition, as environmental problems due to large-scale industrial operations and the acceleration of industry began to emerge, the need of eco-friendly treatment methods to reduce these problems grew [
5,
6]. Many innovative studies in the field of photocatalysis have been carried out. Among them was a groundbreaking study on metal-oxide-based photocatalysts, published by Fujishima and Honda in 1972,which proved that TiO
2 and Pl electrodes contained in aqueous electrolytes can generate oxygen and hydrogen by photocatalytic activity (i.e., the Honda–Fujishima effect) [
7]. Moreover, in 1977, photocatalytic water splitting was reported to generate H
2 and O
2 in a stoichiometric ratio of 2:1 through light irradiation alone, without the application of external energy, under an argon atmosphere [
8]. Following these numerous studies, more extensive studies, such as those on the enhancement of photocatalytic efficiency, the discovery of new photocatalyst materials, and the utilization of light sources, were conducted in the 1980s. Moreover, regarding the development of new photocatalysts, many studies have been carried out on metal-oxide-based photocatalytic materials with higher activity than the existing TiO
2 [
9,
10]. In the 21st century, research at the forefront of the next generation of renewable energy, such as the reduction of air pollutants, the decomposition of polluted water, and hydrogen generation, is being conducted in line with current trends [
11,
12,
13].
Recently, a synthetic method for the production of metal oxides using plasma has been introduced as an inexpensive and rapid method [
14,
15].Low-temperature, non-equilibrium plasmas provide many advantages for synthesizing metal-oxide nanomaterials with specific size and shape, even at room temperature. Plasma-utilizing equipment also has many advantages, such as low energy consumption, easy and cheap installation, atmospheric pressure discharge, and potential use in various industrial fields. Types of plasma sources include plasma reactors such as the dielectric barrier discharge (DBD), corona discharge, micro discharge, and torch systems [
16,
17]. Plasmas are routinely used for ozone generation, gas reforming, biological sterilization, surface functionalization, and medical treatment in plasma discharge equipment. Operating parameters, designs, and configurations vary widely for each reactor, each with different pros and cons [
18,
19,
20,
21].
Various techniques for producing nanomaterials in the presence of atmospheric plasma have been reported in the literature. For example, gold nanoparticle synthesis using micro-plasma for silicon nanomaterials, plasma jet for carbon nanomaterials, and plasma reaction with solutions have been previously performed [
22,
23].
Since ZnTiO3 has various crystalline phases and is applied in various fields, in this study Zn-Ti oxide nanocomposites, including stoichiometric ZnTiO3, were synthesized using an atmospheric plasma jet method, and their applicability for photocatalysis was assessed.
2. Results and Discussion
A scanning electron microscope (SEM) was used to determine the particle morphology and average size of the Zi-Ti oxide nanocomposites synthesized by a soft plasma jet. As shown in
Figure 1, the metal-oxide nanoparticles synthesized by soft plasma jet appeared to have an irregular shape. The size of the particles varied within the range of 50~100 nm, and increased as the amount of titanium butoxide was increased; while the particle size generally decreased, except that of Zn@Ti-D.
Table 1 shows the atomic percentage of plasma-synthesized Zn-Ti oxide nanocomposites, which were obtained from the energy-dispersive spectroscopy (EDS) data. As shown in
Figure 1f–j and
Table 1, the synthesized Zn-Ti oxide nanocomposites showed an increment in the ratio of Ti to Zn as the amount of the titanium precursor was increased, but the ratio decreased in Zn@Ti-E. The ratio of oxygen increased similarly by the corresponding amount, and maximum oxygen content was achieved in the sample of Zn@Ti-C, with the Zn/Ti ratio of 2/1and a stoichiometry of Zn
2.0Ti
1.0O
3+X.
As shown in
Figure 2, the X-ray diffraction (XRD) patterns of the five different samples showed a ZnO hexagonal crystal phase (JPDS 36-1451). Nine diffraction peaks were observed at 31, 34, 36, 47, 56, 62, 66, 67, and 69 degrees, due to the diffractions of (100), (002), (101), (102), (110), (103), (200), (112), and (201) planes, respectively. In addition, the samples of Zn@Ti-C, Zn@Ti-D, and Zn@Ti-E exhibited the rhombohedral crystal phase of ZnTiO
3 (JPDS 26-1500) with typical four diffracted peaks at 32, 35, 48, and 61 degrees, reflecting the diffractions of (104), (110), (024), and (214) planes, respectively. This means that the crystal structures of Zn@Ti-A and Zn@Ti-B differed from those of the samples of Zn@Ti-C, Zn@Ti-D, and Zn@Ti-E, suggesting that a different behavior of photocatalytic activity could be expected. This indicates that Zn@Ti-A and Zn@Ti-B had a single-crystalline (hexagonal crystal structure) nature, while Zn@Ti-C, Zn@Ti-D, and Zn@Ti-E had mixed phases with both hexagonal and rhombohedra crystal structures. The XRD data also show different crystallinity. Among five different samples, the Zn@Ti-A sample exhibited the best quality of crystallinity, while Zn@Ti-B exhibited the second-best quality. The Zn@Ti-C and Zn@Ti-D samples were of similar quality, while the sample of Zn@Ti-E showed a poor quality of crystallinity, considering both peak intensity and the FWHM values of the main diffracted peak.
Figure 3 shows the Brunauer–Emmett–Teller (BET) graphs of five different samples. Among them, the Zn@Ti-A sample shown in
Figure 3a, made only of zinc nitrate precursor, was composed of ZnO nanoparticles and had a surface area of 30.556 m
2/g. The samples of Zn@Ti-B (b) and Zn@Ti-C (c),which were synthesized using zinc nitrate and titanium butoxide precursors, had increased specific surface areas(both over 60 m
2/g),corresponding to the increase of titanium butoxide. The Increase in the specific surface area of Zn@Ti-B (b) and Zn@Ti-C (c) was attributed to the formation of smaller metal-oxide nanoparticles, because the amorphous Zn-Ti metal-oxide complex interferes with the aggregation of zinc oxide particles. The surface areas of Zn@Ti-D (d) and Zn@Ti-E (e) were reduced because ZnTiO
3, which was larger than the zinc oxide nanoparticles, was produced. ZnTiO
3 was produced in a molar ratio of zinc nitrate to titanium butoxide gradually increasing to 1:1.Based on our XRD data, we calculated the grain sizes of the samples by utilizing the Scherrer equation. The obtained values of grain sizes were 21.8 nm (Zn@Ti-A), 92.1 nm (Zn@Ti-B), 100.7 nm (Zn@Ti-C), 55.9 nm (Zn@Ti-D), and 75.8 nm (Zn@Ti-E). This means that there was little relationship between the BET surface area and grain size. However, we found that, apart from the grain size of crystalline ZnTiO
3 (i.e., the Zn@Ti-A sample), the obtained grain sizes were comparable with the sequences of the average particle sizes shown in the SEM images.
As shown in
Figure 4 (the left-hand figure), ZnTiO
3crystals(such as Zn@Ti-E) as in point group C
3i in group theory had ten Raman active modes: 5A
g + 5E
g(A
g: 177, 267 347, 490, 717 cm
−1and E
g: 129, 234, 393, 472, 618 cm
−1). The Raman spectra of the ZnTiO
3crystal were first reported by Baran and Botto in 1979, without mode assignment. The frequency of the mode obtained from the Raman spectra of Zn@Ti-E that we synthesized, corresponds exactly with the Raman spectra of the ZnTiO
3crystal reported by Baran and Botto [
24]. However, in the Raman spectra of Zn@Ti-A, shown in the right-hand figure of
Figure 4, the hexagonal phase of ZnO was also identified from the characteristic sharp Raman peak at 440.0 cm
−1, which is the E
2 optical phonon mode. The E
2H-E
2L and A
1T modes of the multi-phonon process were observed at 333.0 and 389.0 cm
−1, respectively. The presence of structural defects in the form of oxygen defects in zinc epilepsy was evident from the E
1L mode at 591.2 cm
−1, which was weak compared to the strong Raman active peak at 440.0 cm
−1 [
24,
25,
26]. As the amount of titanium butoxide was increased, the Raman spectra of Zn@Ti-B, Zn@Ti-C, and Zn@Ti-D gradually changed to the mode of ZnTiO
3 from the ZnO mode. An important distinction between the Raman and XRD data is that the XRD signal is the result of long-range structural ordering among crystalline lattice planes, whereas the Raman spectrum reflects local molecular bond vibrations, making the Raman highly sensitive even for detecting nanocrystallite formations in those materials with a high Raman cross section. In the case of the pristine TiO
2 sample, for example, the Raman spectra showed the only phase determined is rutile, with peaks at 611 cm
–1 (
A1g), 447 cm
–1(
Eg), and 143 cm
–1(
B1g). A noticeable increment of the anatase phase weight percentage was observed for an Al-doped TiO
2 sample with the Al-dopant concentration above 0.7 wt%, and was characterized by the apparent growth of the bands at 197 and 640 cm
–1(E
g), 398 and 515 cm
–1(B
1g), and 515 cm
–1(A
1g). The energy band gaps of TiO
2were 3.0 eV (rutile phase) and 3.2 eV (anatase phase), while the ZnO and ZnTiO
3 crystals showed energy band gaps such as 3.29 eV (wurtzite hexagonal ZnO), 3.10 eV (blende cubic ZnO), and 3.06 eV (ZnTiO
3) [
27,
28,
29]. This means that in the case of Ti-Zn oxide nanocomposites, it is possible to control the energy band gaps between 3.0 and 3.3 eV.
Studies on both the photocatalytic degradation of methylene blue (MB) dye and its kinetics were carried out using the five different photocatalysts (i.e., Zn-Ti oxide nanocomposites). The degradation experiment with MB dye solution was carried out to determine the possibility of removing hazardous substances with environmentally friendly industrial applications.
Figure 5a shows the results of photocatalytic decomposition with 10 mL of 10 ppm MB dye for 2 h. During the degradation experiments with the MB dye, we measured the changes of dye concentrations with an UV-visible spectrophotometer every 5 min. The first results of relative concentration (C/C
o) with the five different photocatalysts (Zn@Ti-A~E) obtained after UV light irradiation for 5 min were 0.868, 0.802, 0.756, 0.820, and 0.835, respectively. This means that the amount of MB dye decomposed under UV light for the initial 5 min was not significantly different among the samples. However, the degree of photo-degradation began to show greater differences between 5 and 15 min under UV light. The sample of Zn@Ti-C in particular showed a large dye decomposition tendency compared with the other samples. The photocatalytic efficiency of the nanocomposites was good in the decreasing order of Zn@Ti-C, Zn@Ti-B, Zn@Ti-D, Zn@Ti-E, and Zn@Ti-A. After 120 min of irradiation under UV light, the system containing the Zn@Ti-A composite retained approximately 85% of the MB dye, while the systems with Zn@Ti-C and Zn@Ti-E nanocomposites retained about 68% and 73% of the dye, respectively. This suggests that when photodegrading MB dye with UV light for 120 min, the photocatalytic efficiency of the Zn-Ti oxide nanocomposites could depend on several parameters, such as crystallinity (Zn@Ti-A), surface area (Zn@Ti-C), surface functionality (Zn@Ti-E), crystal size and shape, surface defects, etc. For example, based on the data in
Figure 3, the surface areas of the Zn@Ti-B and Zn@Ti-C samples were double those of the Zn@Ti-A, Zn@Ti-D, and Zn@Ti-E samples. The crystallinities of both the Zn@Ti-A (based on XRD) and Zn@Ti-E (based on Raman) samples were relatively better than those of the other samples. Moreover, the Zn@Ti-E sample contained a Ti- and oxygen- rich constituent compared with other samples, signifying different surface functionality. However, in this study we obtained the best photocatalytic efficiency when we used the photocatalyst Zn@Ti-C, which contained the highest oxygen content among the five different samples with the Zn/Ti ratio of 2. We compared this result with the photocatalytic efficiency of a commercially available TiO
2 photocatalyst (namely Degussa P-25, which contains a 70% anatase crystalline phase and a 30% rutile crystalline phase) that we had tested before [
30].The Zn@Ti-C in this study showed a higher efficiency within the first 30 min (
Table 2). However, the efficiency of crystalline ZnTiO
3 (i.e., Zn@Ti-A) was very poor compared with that of both Degussa P-25 TiO
2 and the oxygen-rich Zn
2.0Ti
1.0O
3+x (i.e., Zn@Ti-C). This suggests that surface functionality, together with oxygen defects and surface area, are more important factors than crystallinity in influencing photocatalytic activity. For this reason, our inexpensively synthesized photocatalyst can be considered as a replacement for the alternative catalysts in various fields of application.
Using our experimentally obtained data, especially that shown in
Figure 5a, we carried out a theoretical kinetic study. Since the first-order kinetics for the photocatalytic reaction were well-known, we measured the changes of UV-visible absorption (i.e., the п-п* transition), and then obtained the variation ratio of the MB dye concentration, as shown in
Figure 5a. The MB dye has double bonds that would be broken on the catalyst surface by the absorption of UV light, resulting in degradation. In this work, therefore, we consider only that the type of reaction is A (MB dye) ↔P (degraded product), and the role of the catalyst is to allow a preferred adsorption site, and to lower the activation barrier and accelerate the reaction. The photocatalytic reaction constants of each sample were thus calculated using Equations (1) and (2) below, assuming first-order kinetics. We then plotted
Figure 5b by utilizing Equation (2).
where C
0 and C are the initial concentration and the concentration of the MB dye solution, respectively, t is the photocatalytic reaction time, and k is the reaction constant. Since equation (1) has the form of an exponential decay similar to
Figure 5a, a common feature of all first-order reactions is that the concentration of the reactant (i.e., the MB dye) decays exponentially with time. However, there are different exponential decay curves, depending on k values, as shown in
Figure 5a. The greater the rate constant, the more rapid the decay curve. Using Equation (1) or (2), we could extract the first-order rate constants (k
1/min) from each decay curve. For example, in the case of sample Zn@Ti-A, we obtained the value of k
1as 2.8
10
−2 min
−1. However, in the other cases we obtained k
obs values of 4.4
10
−2, 5.5
10
−2, 3.9
10
−2, and 3.6
10
−2 min
−1 for Zn@Ti-B, Zn@Ti-C, Zn@Ti-D, and Zn@Ti-E, respectively.
Figure 5b shows the kinetics curves (the plot of ln(C/C
0) as a function of the photocatalytic reaction time (t)) of the reaction of 10 ppm MB dye solution degraded by the photocatalysts Zn@Ti-A, Zn@Ti-B, Zn@Ti-C, Zn@Ti-D, and Zn@Ti-E. The curves indicate that the degradation reaction could be expressed by the first-order reaction kinetics model only for the Zn@Ti-A sample; the curves of the Zn@Ti-B, Zn@Ti-C, Zn@Ti-D, and Zn@Ti-E samples did not operate with first-order reaction rate models. This means that the perfect linearity indicates first-order kinetics with the first-order rate constant (k
1/min). However, the experimentally obtained data exhibit large deviations, suggesting a pseudo-first-order kinetics (i.e., not a perfect first-order reaction), with pseudo-first-order rate constants (k
obs/min). Therefore,
Figure 5 can provide us with some hints as to the effects of the initial MB concentration on the photocatalytic efficiency. Firstly, due to the large surface area, the Zn@Ti-C photocatalyst among the five samples could decompose the MB dye the most rapidly (5 times faster than the perfect crystalline sample). Secondly, the photocatalytic degradation of MB exhibited pseudo-first-order kinetics with different rate constants (k
obs), except in the case of the crystalline Zn@Ti-A. This suggests that the kinetics of photocatalytic degradation of MB in non-crystalline and crystalline photocatalysts might be different [
31]. For example, since the concentrations of radicals such as O
2−· (1D) and OH· are also closely related to the kinetics of dye degradation, it is not an exact fit with first-order kinetics. This means that a more detailed kinetic study, including the detection of key radicals such as O
2−· (1D) and OH· radicals, is highly desirable in order to clarify our important results. It is well-known that the average lifetime of OH· radical (τOH·) in ambient atmospheric conditions is around 0.01~1 s [
32], which is affected by the concentration of reactive gas components such as ozone, VOCs, and NOx. In conclusion, it is very important to note that a photocatalytic reaction is not simple and cannot in general be inferred from the chemical equation for the reaction. For this reason, further study is needed to systematically clarify the detailed reaction mechanism.