*5.1. Reaction Order*

Figure 3a,b shows that concentrations of rhoB fit to the first- and half-order integrated rate law, respectively, for increasing annealing temperature and constant 200 nm-thick films. Half-order reaction kinetics (Equation (10)) better approximate the overall degradation reaction than the more commonly applied first-order rate law (Equation (8)), as indicated by the higher coefficient of determination (R<sup>2</sup>) values for the linear fits (shown in Table 1). The orders of the individual reactions that ultimately lead to the degradation of organic compounds, such as rhoB by semiconductor photocatalysts, vary between zero and one, depending on the reaction. When these reactions are coupled, the order of the overall reaction is somewhere between zero and one, and so the best approximation is most likely to be the half-order rate law [19]. For reasons of simplicity and uniformity between studies, the first-order rate law has commonly been applied to quantify the photocatalytic activities of semiconductor photocatalysts. However, because the half-order rate law is a better and more logical approximation, the photocatalytic activities were calculated in this study using the half-order integrated rate law (Equation (10)).

Figure 3. Concentrations of rhoB fit to the (a) first order and (b) half-order integrated rate law for varying annealing temperature (300–900 ◦C). The best fit curves are indicated by the solid lines, with the slope of the best fit curves representing the photocatalytic activity. The control consists of degradation measurements made in the absence of catalyst films.


Table 1. Regression statistics for first- and half-order reaction kinetics.

However, it should be noted that the order of the overall reaction is determined by which reaction types predominate at any given moment, which itself is dependent on a multitude of factors. One factor that determines the predominant reaction that is taking place at any given moment is the concentration of the various molecules involved in the degradation reactions, which varies significantly over time. The order of the overall reaction can thus change over the course of the photocatalysis experiment due to decreasing concentrations of rhoB or increasing concentrations of free radicals and peroxides in solution. The change in overall reaction order is apparent from inspection of Figure 4, which is a semi-log plot representing the first-order rate law. The semi-log concentration is linear during the first 60 min of irradiation time and significantly non-linear thereafter, suggesting a change in overall reaction order. Therefore, there is a limit to the ability of the reaction to follow any integrated rate law with time, which is not necessarily a result of the catalyst film itself. For studies involving long-term film stability, activities are measured in shorter 60–120-min time intervals before the rhoB solution is replenished.

Figure 4. Semi-log plot of rhoB concentration with time. Within the first 60 min of irradiation, first-order reaction kinetics fit the observed degradation as indicated by the solid line. However, the change in overall reaction order is apparent after the first 60 min.

## *5.2. Photocatalytic Activity and Annealing Temperature*

As shown in Figure 3b, the half-order photocatalytic activity decreases with increasing annealing temperature, as indicated by the decreasing slope of the best-fit lines with increasing temperature. A control measurement was made, which consisted of a sapphire substrate without a ZnO film incubated in an 8-ppm rhoB solution, while irradiated with UV light (hollow squares, Figure 3b). We observed no difference in photodegradation between rhoB solutions alone and rhoB solutions with immersed sapphire, suggesting that the substrate plays no significant role in the degradation of the rhoB. It is clear that some degradation of rhoB occurs as a result of UV illumination alone. To determine the contributions to degradation by the ZnO films and, therefore, the photocatalytic activity of the films, we subtract the control curve from measurements made in ZnO-incubated solution, with the resulting slope of the best-fit representing the photocatalytic activity of the metal-oxide films. The photocatalytic activities of 200 nm-thick ZnO films determined from half-order kinetics are reported in Table 2.

The activity of the films annealed at 600 ◦C was reduced by 20% compared to the films annealed at 300 ◦C, and the activity of the 900 ◦C annealed films was reduced by 30% compared to the 600 ◦C annealed films. Interestingly, this is the same proportional reduction as is observed for peak PL excitation and annealing temperature, as shown in Figure 2b. This suggests that excitation efficiency, specifically in the excitonic band, can significantly affect the photocatalytic activity of metal-oxide films.

Table 2. Photocatalytic activity at various annealing temperatures.


Since PL excitation is directly correlated to electron-hole pair formation on the semiconductor surface, the films with higher optical quality have higher rates of electron-hole pair formation. Furthermore, the oxidation and reduction reactions that lead to the degradation of rhoB and other organic compounds by the semiconductor photocatalysts are dependent on the availability of electrons and holes on the surface of the semiconductor, so the optical quality of the films is the most important factor in the photocatalyzed degradation by the semiconductor, at least when morphological features are similar.

The effective surface area on the nano-scale should also affect the photocatalytic activities of the films, though the precise extent of its significance has been difficult to determine for metal-oxides due to the connection between optical properties and grain size, surface roughness and other morphological properties [43]. Since the films annealed at 300 ◦C in this study have larger grain sizes than those annealed at higher temperature, their surface area is effectively smaller, resulting in lower rates of contact with molecules in solution and ultimately leading to potentially lower reaction rates, as well. However, this was not observed, and we speculate that this change in grain size with increasing temperature was too small to have a significant contribution to the photocatalytic activity. Furthermore, there is no observed change in grain size with increasing annealing temperature past 600 ◦C, while significant reductions in photocatalytic activity are still observed, suggesting that optical properties dominate the contribution to activity in this case.

## *5.3. Photocatalytic Activity and Film Thickness*

Figure 5a shows the half-order concentration curves for films annealed at 300 ◦C and having thickness varying from 200 to 600 nm. Again, a control measurement was made, which consisted of a sapphire substrate without a ZnO film incubated in an 8-ppm rhoB solution while irradiated with UV light (hollow squares, Figure 5a). We subtracted the control curve from measurements made in ZnO-incubated solution (filled triangles, diamonds, squares and circles in Figure 5a), with the resulting slopes of the best-fit representing the photocatalytic activities of the metal-oxide films. The half-order photocatalytic activities are plotted as a function of film thickness in Figure 5b and are shown in Table 3. An approximately linear increase in photocatalytic activity is observed with increasing film thickness, as shown via the solid line in Figure 5b.

In the previous section, we investigated photocatalytic activity with film annealing temperature while keeping film thickness constant. For increasing annealing temperature, there was a significant decrease in photoemission efficiency, but little change in overall surface morphology, at least with respect to the observable height of ZnO grain protrusions, as seen in Figure 1d,e. In this section, we discuss the photocatalytic activity as a function of film thickness with constant annealing temperature. A comparison of the excitonic peaks for films annealed at 300◦C at a thicknesses of 600 and 200 nm shows no significant difference in excitonic photoemission efficiency with thickness, as seen in Figure 2a,b, respectively. However, inspection of the AFM images shown in Figure 1b,d indicates a doubling in the observable ZnO grain protrusion height when going from 200 to 600 nm-thick films.

Figure 5. Half-order concentration curves for films annealed at 300 ◦C and having thicknesses varying from 200 to 600 nm. Data is fit to the half-order integrated rate law, with the curves of best-fit shown as the solid lines.

The significant increase in protrusion height could explain the increase in photocatalytic activity with increasing film thickness. As mentioned, the effective surface area on the nano-scale may also affect the photocatalytic activities of the films, though the precise extent of its significance has been under question [43]. The polycrystalline nature of our films results in approximately 100–150 nm-diameter ZnO columns that protrude upward from the surface, with an inter-column spacing on the order of 10–20 nm. Increasing the height of these columns results in an effective increase in surface area, resulting in higher rates of contact with molecules in solution and ultimately leading to higher reaction rates. The observed increase in green and yellow band emission with thicker films appears to have no effect on the photocatalytic activity of the films. Therefore, as expected, the excitonic photoemission efficiency and effective surface area of the catalytic material are the main contributors to activity. For films grown via a two-step thermal annealing process, low temperature annealing combined with relatively thick films results in materials demonstrating greater photocatalytic activity, due mainly to high optical quality and the porous nature of the morphology.


Table 3. Photocatalytic activity at various film thicknesses.

## *5.4. Film Stability*

To determine film stability, we looked at the evolution of the photocatalytic activity over long periods of time (up to two days of continuous incubation). The order of the overall reaction can change over the course of the photocatalysis experiment due to decreasing concentration of rhoB or increasing concentrations of free radicals and peroxides in solution, which could manifest as a decrease in activity. We are interested in the photocatalytic activity of the films, so activity was determined via rhoB concentration measurements over one-hour intervals. The rhoB solution was replaced after each interval to ensure sufficient concentrations of rhoB and relatively low concentrations of free radicals and peroxides.

Figure 6 shows the photocatalytic activity as a function of time for a ZnO film having a thickness of 600 nm and annealed at a temperature of 300 ◦C. The photocatalytic activity decreases approximately linearly over 50 hours until reducing to zero. Interestingly, after 50 hours of continued incubation, the morphology of the sample is consistent with the morphology of the sapphire substrate (not shown), suggesting that the cause of the observed decreasing activity is the degradation of the ZnO film. During the incubation period, ZnO at the interface dissociates, forming Zn<sup>+</sup> and O<sup>−</sup> ions that leave the film structure and enter the solution. This is consistent with reports in the literature concerning the stability of bulk and thin-film ZnO in contact with aqueous solutions [4,17]. Preferential etching should occur at termination sites, such as at the tips of ZnO grain protrusions seen in Figure 1b. This results in a slow decrease in film thickness as the reactions take place. As shown in the previous section, when the photoemission efficiency is similar, the predominant contributor to the photocatalytic activity is the specific surface area and, in the context of this study, the film thickness. Therefore, the photocatalytic activity decreases with time due to surface etching of the film, resulting in decreasing film thickness.

Figure 6. Photocatalytic activity as a function of time for a ZnO film having a thickness of 600 nm and annealed at a temperature of 300 ◦C. Measurements where made spectrophotometrically over one-hour increments, with the rhoB solution being replaced after each one-hour measurement.

## 6. Discussion

Extensive work has gone into investigating the photocatalytic properties of ZnO nanoparticles and colloid suspensions [20]. However, separating the catalyst from solution is challenging, which makes their use in these applications potentially cost-prohibitive [22]. As mentioned, Zn toxicity is also a concern for these systems, especially if allowed to remain in the environment. Fujishima *et al.* suggest that the nanofilm form of these semiconductors is preferable to particles for use in fluid decontamination exactly because the nanoparticles need to eventually be collected and removed from the fluid [23,24]. There has been little discussion in the literature concerning the sorts of structural and photophysical properties that can directly affect photocatalytic activity for film-based catalysts. For the porous polycrystalline films discussed in this study, both optical and morphological properties have been shown to significantly affect the photodegradation of organics at the interface.

Polycrystalline films demonstrating high surface roughness and with small crystal grain size do typically have a high effective surface area, which increases the number of available surface states to serve as reaction sites. The concern with respect to the engineering of effective metal-oxide-based catalysts, though, is that the decreased crystallinity associated with these surfaces typically results in reduced optical efficiency and, therefore, lower electron-hole pair production efficiency [23,27,28]. Both high surface area and optical efficiency are required for high photocatalytic activity with metal-oxides. For TiO<sup>2</sup> and ZnO thin, polycrystalline films, a decrease in grain size typically results in increased surface roughness and surface area; however, small grain size typically also corresponds to an increase in deep-level defects that could lower the number of photo-induced holes at the surface available for catalysis [28]. These competing mechanisms must be balanced. Thermal oxidation of Zn-metal films at low annealing temperatures applied over many hours results in a balance between high effective surface area and optical quality. Furthermore, the porous nature of these films results in increased effective surface area with increased film thickness, without a corresponding decrease in optical quality, at least with respect to electron-hole pair production efficiency.

However, a significant problem with metal-oxide systems, such as ZnO, is long-term stability. The most photoactive films discussed in this study decrease in effectiveness by 50% in approximately 24 hours. There is also some concern about the dissolution of ZnO particles and resulting Zn toxicity in marine environments [4,17]. Although most of the work on ZnO dissolution has been with respect to nanoparticles and colloid suspensions, there is increasing evidence that Zn toxicity should be a concern for film-based systems and even for the interface of the single-crystal bulk [4,25]. Continued work needs to be done to find metal-oxide systems that exhibit high catalytic activity and that are passivated from preferential etching at termination sites, resulting in increased stability.

## 7. Conclusions

In summary, the photocatalytic activity and stability of thin, polycrystalline ZnO films was studied. The oxidative degradation of organic compounds at the ZnO surface results from the ultraviolet (UV) photo-induced creation of highly oxidizing holes and reducing electrons, which combine with surface water to form hydroxyl radicals and reactive oxygen species. Therefore, the efficiency of electron-hole pair formation is of critical importance for self-cleaning and antimicrobial applications with these metal-oxide catalyst systems. In this study, the lower annealing temperature of the fabricated ZnO thin films resulted in decreased protrusion size and specific surface area, as well as increased UV excitonic emission. The films oxidized at lower annealing temperatures exhibited higher photocatalytic activity, which is attributed to the increased optical quality. Photocatalytic activity was also found to depend on film thickness, with lower activity observed for thinner films due to a decrease in effective surface area. Decreasing activity with use was found to be the result of decreasing film thickness due to preferential surface etching.

## Acknowledgments

This work was supported by the U.S. National Science Foundation Division of Materials Research (NSF-DMR 1104600) and the II-VI Foundation's Undergraduate Research Program. The photoluminescence data presented in this paper were obtained at Virginia Commonwealth University (Richmond, VA, USA) in the laboratory of Mikhail Reshchikov.

## Author Contributions

This work was completed in Christopher Moore's research laboratory under his direction. Data collection and analysis was completed by all three authors. Robert Louder contributed writing to Sections 4 and 5. Cody Thompson contributed writing to Section 5. Both Robert Louder and Cody Thompson provided their main contributions to this work while undergraduate students at Coastal Carolina University. Robert Louder is currently a biophysics graduate student at the University of California, Berkeley. Cody Thompson is currently a supervisor in the Research and Development laboratory at Wellman Engineering Resins.
