2. Experiment

Zinc films where deposited on *c*-plane sapphire substrates via direct current sputter deposition. Metallic Zn targets were obtained commercially and had a purity of 99.99%. Before deposition, substrates were cleaned via immersion in acetone, ultrasonically cleaned in methanol and rinsed in deionized water. The chamber base pressure was maintained between 1.0 × 10−<sup>5</sup> and 2.5 × 10−<sup>5</sup> mbars. A gate valve between the chamber and the pump was utilized as a throttle to maintain an Ar pressure of approximately 2 × 10−<sup>2</sup> mbars. Sputtering power was maintained between 10 and 20 W with the substrate located approximately 10 cm from the sputter source. Deposition times ranged from 15 to 40 min, resulting in film thicknesses between 100 nm and 600 nm as measured via the AFM profile and reflectometry. Thermal oxidation of the Zn metal films was carried out in an air-ambient muffle furnace. For all samples, Zn films were initially annealed at 300 ◦C for 9–24 hours to ensure complete oxidation. Some films were then re-annealed for 1 hour at 600 ◦C, 900 ◦C and 1200 ◦C.

Zinc oxide films were characterized via X-ray diffraction (XRD), atomic force microscopy (AFM) and photoluminescence (PL). Structural properties of the ZnO films were measured using Cu-Kα radiation in the range from 30◦ to 50◦. The morphology of the films was determined via dynamic-mode AFM using an Anfatec Level AFM and approximately 300 kHz resonant aluminum backside silicon tips. Photoluminescence spectra were obtained at room temperature using a HeCd laser as an excitation source and a power of *P* = 0.3 W/cm<sup>2</sup>. UV illumination was provided by a deuterium lamp.

Following morphological and optical characterization, the photocatalytic activities of the films were characterized by measuring the degradation of Rhodamine B dye (rhoB) in solution. RhoB was used to simulate an organic environmental contaminant, because its concentration in solution can be accurately measured spectrophotometrically. Oxidized rhoB products do not absorb visible light, so the concentration and the resulting absorbance of the rhoB solutions decrease as the rhoB is oxidized by the ZnO photocatalysts. The overall photocatalytic activities of the materials therefore correlate to the rate of change in the concentration of the rhoB solutions as measured by UV-Vis spectrophotometry.

The UV-Vis spectrophotometer was first calibrated in order to determine the relationship between absorbance measurements and rhoB concentrations. This was accomplished by measuring the absorbance of known concentrations of rhoB and constructing a calibration curve from the resulting measurements (not shown). The concentrations in the rest of the experiment were then calculated from the absorbance measurements. The linear range of the absorbance-concentration relationship was determined to be between 1 and 8 ppm rhoB, so an initial concentration of 8 ppm rhoB in solution was used for the photodegradation experiments.

Each ZnO thin film was incubated in an 8 ppm rhoB solution while irradiated with UV light at a constant power density of 80 μW/cm. The absorbance of the solution was taken after 10, 20, 30, 45, 60, 90, 120, 150 and 180 min of UV irradiation at the peak absorbance wavelength (553 nm) for maximum sensitivity, and the corresponding rhoB concentration was calculated using the equation of the rhoB calibration curve. In order to account for differences in the surface areas of the films, the total change in concentration of rhoB at each time interval was divided by the specific surface area of the film and plotted against the time of UV irradiation.

## 3. Film Growth and Characterization

As described earlier, greater catalytic activity is expected with high surface roughness and high optical efficiency. In this section, we discuss the morphology, structure and optical properties of the ZnO thin films under study. Specifically, we describe the morphological and structural evolution with annealing temperature and film thickness, concentrating on crystal grain sizes and surface roughness. We also describe how growth parameters affect the optical properties of the films. In particular, we discuss the relative excitonic emission efficiency and yellow-green band emission.

## *3.1. Morphology and Structure*

The evolution of the surface morphology with increasing annealing temperature for 200 nmand 600 nm-thick ZnO films is shown in the AFM images presented in Figure 1a–e. As shown in Figure 1a, the as-grown zinc film demonstrates a high surface roughness and approximately 100-nm diameter protrusions. Figure 1b,c shows the surface morphology of resulting 600 nm-thick ZnO films annealed at 300 ◦C and 600 ◦C. Figure 1d,e shows the surface morphology of the resulting 200 nm-thick ZnO films annealed at 300 ◦C and 600 ◦C. For both thicknesses, there is very little change in the underlying characteristics between zinc-metallic films and ZnO films annealed at 300 ◦C; however, surface roughness is observed to increase, and protrusions grow in size by approximately 50 nm. This is consistent with previous studies, where Gupta *et al.* show that the preferred orientation of ZnO thermally oxidized on glass can depend on the Zn film texture and oxidizing agent [31]. For 600 nm-thick films, an increase in surface roughness is observed with increasing temperature (Figure 1b,c). Interestingly, at temperatures above 600 ◦C and a thickness of 600 nm, long vertically-aligned nanorods are visible, which is consistent with previous studies [32]. For a 200 nm-thick films, a decrease in protrusion diameter and surface roughness is observed with increasing temperature (Figure 1d,e). Nanorods are not seen for the thinner films annealed at any temperature. For both film thicknesses, there is little change in surface morphology observed at higher temperatures (not shown). Specifically, the protrusion size does not significantly change.

Grain size was characterized by both AFM and XRD. For films having thicknesses between 400 nm and 600 nm, grain size was observed to increase with increasing annealing temperature from 300 ◦C to 1200 ◦C. Interestingly, for films having thicknesses between 100 nm and 200 nm, grain size decreased with increasing temperature up to a certain point. Figure 1d,e shows a decrease in the protrusion diameter from approximately 150 nm to approximately 100 nm at annealing temperatures of 300 ◦C and 600 ◦C, respectively, with no further significant change in size as the temperature was further increased. Grain size, as determined by the full width at half maximum in the XRD spectra (not shown), was also found to decrease with increasing temperature up to 600 ◦C, with relatively small increases at higher temperatures, which is consistent with AFM measurements. This observation appears inconsistent with some reports in the literature for thermally-oxidized ZnO films [33,34]. This discrepancy may be the result of differences in studied temperature regimes, the variations in film thickness and/or our two-step thermal annealing process, where metallic zinc films are all initially oxidized at low temperature. Furthermore, our metallic zinc films display a significantly different texture and larger initial particle size in comparison to films grown via other methods, which has been shown to affect resulting film morphology and structure [31].

## *3.2. Optical Properties*

Figure 2 shows the PL spectra of ZnO films with a thickness of (a) 600 nm and (b) 200 nm thermally annealed at various temperatures. The spectra for the 600 nm-thick films all show asymmetric broad bands in the yellow-green region, with no significant yellow-green emission observed for the 200 nm-thick films. For all thicknesses, a more narrow band in the UV is observed, which is associated with excitonic emission; however, the 200-nm film demonstrates an asymmetric and broad band in the blue-UV region at higher temperatures. For all thicknesses, the strongest UV excitonic emission is observed at low temperature, with decreasing UV emission observed with increasing annealing temperature. Interestingly, for thinner films, increasing temperature results in a significant redshift (0.15 eV) in the UV excitonic peak and an asymmetrical peak broadening. This change in UV peak position could be explained by a corresponding shift in bandgap energy, which would be consistent with transmission studies of ZnO films grown via the sol-gel method and previous studies of thermally annealed Zn films [28,35]. Wu *et al.* and Dijken *et al.* both demonstrate that an increase in particle size should result in a redshift in energies, which appears inconsistent with our results, since in this temperature and thickness regime, we see a decrease in grain size [36,37]. However, these studies discuss systems where quantum size effects become relevant, and the particle sizes in this study are sufficiently large, such that the shift in bandgap cannot be explained via a similar mechanism. Jain *et al.* speculate that this red shift is the result of an increase in interstitial zinc atoms, which we demonstrated to be the case in a previous study [28,35].

Figure 1. AFM topography images of (a) Zn-metal films before oxidation (grayscale range = 200 nm) and the resulting polycrystalline ZnO films after annealing in ambient air; Six hundred nanometer-thick films annealed at (b) 300 ◦C (grayscale range = 500 nm) and (c) 600 ◦C (grayscale range = 1000 nm). Two hundred nanometer-thick films annealed at (d) 300 ◦C (grayscale range = 300 nm) and (e) 600 ◦C (grayscale range = 250 nm).

As shown in Figure 2, 600 nm-thick films demonstrate increasing yellow-green band intensity with increasing annealing temperature when compared to UV emission. This can be attributed to a rapid increase in V<sup>+</sup> <sup>o</sup> and O<sup>−</sup> <sup>i</sup> ion centers at high temperatures [34,36]. In contrast, 200 nm-thick films demonstrate little green and yellow band emission at any temperature (Figure 2b). Deep level emission is attributed to bulk defects; therefore, it is possible that decreased bulk volume results in the formation of relatively fewer deep-level states. If green and yellow emission results from the recombination of a delocalized electron close to the conduction band with a deeply trapped hole in the V<sup>+</sup> <sup>o</sup> and O<sup>−</sup> <sup>i</sup> centers in the bulk, respectively, then a decrease in film thickness would decrease the bulk with respect to the depletion region, resulting in weaker bulk-related, deep-level emission [36]. Reaction kinetics could also contribute, with thinner films having shorter diffusion paths for reactive oxygen species during oxidation [38]. Thin films would therefore demonstrate fewer V<sup>+</sup> <sup>o</sup> and O<sup>−</sup> <sup>i</sup> ions at any annealing temperature, as observed.

Figure 2. PL spectra of ZnO films grown at 300 ◦C, 600 ◦C, 900 ◦C and 1200 ◦C at thicknesses of (a) 600 nm and (b) 200 nm. For thicker films, increasing green band emission relative to UV emission is seen with increasing temperature. Thinner films demonstrate little green band emission; however, a significant redshift in UV emission and asymmetrical band broadening is observed at higher temperatures.

For thin films at high temperatures, the asymmetric and broad UV excitonic emission bands result from increasing blue emission and corresponding decreasing UV emission [28]. As shown in Figure 2b, the PL spectra for the 200 nm-thick film exhibited the most dramatic blue band emission and very low green band emission at all temperatures. Wang *et al.* found that the intensities of the green and yellow cathodoluminescence peaks were strongly affected by the width of the free-carrier depletion region near the surface [34]. They argue that single ionized oxygen vacancies exist only in the bulk, so the magnitude of the depletion region in relation to the bulk directly affects the intensity of the green emission in the cathodoluminescence spectrum. It has been suggested that blue emission results from zinc interstitials found in the depletion region, so an approximately 400-nm blue emission should only be observed in a sample with a wide depletion region relative to the bulk. Otherwise, deep-level green emission will dominate [28,39,40]. Therefore, the emergence of blue emission with decreasing film thickness results from a low ratio of the bulk to the depletion region.

Temperature and grain size also contribute to the PL spectra, since no blue emission is observed for 200-nm films annealed at 300 ◦C (see Figure 2b). As discussed, we observe a larger grain size via XRD and AFM for 200-nm films annealed at 300 ◦C, resulting in a larger bulk to depletion region ratio, which could contribute to the weaker blue emission. Furthermore, thinner films have shorter diffusion paths for reactive oxygen species during oxidation, and lower temperatures slow the Zn/O<sup>2</sup> reaction. This results in fewer zinc interstitials, which would manifest as weaker blue emission and a red shift in both blue and UV emissions, as is observed.

For photocatalysis applications, thin ZnO films show significant promise due to their potential for balancing high optical quality with high surface roughness. In particular, 200 nm-thick films annealed at low temperature (300 ◦C) demonstrate a high UV-to-green emission ratio with relatively high surface roughness. Low-temperature annealed 600-nm thick films also show strong excitonic photoemission combined with a tall protrusion height within the porous film structure, which could result in greater effective surface area for photochemical reactions. In this study, we investigate photocatalysis with these thin films, because they show the most promise for high photocatalytic activity.

## 4. Surface Catalysis and Reaction Kinetics

Metal-oxide semiconductors use light energy to catalyze oxidation-reduction reactions via electron-hole pair production at the material surface [41,42]. The photogenerated holes on the semiconductor surface have a high oxidation potential, while the photogenerated electrons have a high reduction potential [19,23]. Several types of aqueous reactions catalyzed by electron-hole pairs lead to the formation of the hydroxyl radicals and reactive oxygen species that are directly responsible for the oxidative degradation of organic compounds. One of these reactions involves the oxidation of water (H2O) by a hole (h<sup>+</sup>) into hydrogen ions (H<sup>+</sup>) and a hydroxyl radical (O∗):

$$h^{+} + H\_{2}O \to 2H^{+} + O^{\*}\tag{l}$$

Another reaction involves the reduction of molecular oxygen (O2) into a superoxide radical (O−∗ <sup>2</sup> ) by a photogenerated electron (e−):

$$e^- + O\_2 \to O\_2^{-\*} \tag{2}$$

The superoxide radical can be further reduced by another electron and then paired with two H<sup>+</sup> ions to form hydrogen peroxide (H2O2):

$$2O\_2^{-\*} + e^- + 2H^+ \to H\_2O\_2 \tag{3}$$

The hydrogen peroxide can then be reduced by an electron to form hydroxyl radicals:

$$2\text{ e }^- + H\_2O\_2 \rightarrow OH^- + OH^\* \tag{4}$$

By these reactions, either the electrons or the holes of the photogenerated electron-hole pairs can produce hydroxyl and superoxide radicals that can subsequently degrade organic compounds [19,23]. It has also been proposed that the electrons or holes themselves may be responsible for at least some of the degradation of organics. Whether degradation occurs predominantly due to reaction with free-radicals or directly with the holes themselves is controversial, with some groups even proposing a predominant electron-based catalytic pathway on ZnO single-crystal surfaces [23,25]. Most likely, however, a large combination of chemical reactions involving various intermediates ultimately leads to the degradation of the organic compounds, which complicates the mathematical descriptions of the rates at which the overall degradation occurs [19].

Simplified mathematical descriptions of the degradation reactions have been used to experimentally quantify and compare photocatalytic activities. For most purposes, especially when the concentration of the contaminant is less than 10 ppm, the reaction can be considered first order, meaning that the time t rate of change in concentration c of the contaminant, dc/dt, follows the general relationship:

$$-\frac{dc}{dt} = kc\tag{5}$$

where k is the rate constant [19]. This equation has been modified for solid photocatalysts to account for the surface area of the photocatalyst and the intensity of the incident light, as follows:

$$-\frac{dc}{dt} = kcA\_S\sqrt{I} \tag{6}$$

where I is the intensity of light used and A<sup>S</sup> is the total surface area of the photocatalyst [23]. The solution to this differential equation is as follows:

$$c(t) = c\_0 e^{-ktA\_S\sqrt{I}}\tag{7}$$

where c<sup>0</sup> is the initial concentration of the contaminant. In order to make the concentration-time relationship linear, the equation can be rewritten as the integrated rate law, as follows:

$$
\ln \frac{c}{c\_0} = -ktA\_S \sqrt{I} \tag{8}
$$

When the natural log of the normalized concentration c/c<sup>0</sup> is plotted as a function of time, the rate constant k can be easily determined from the slope of the linear best-fit. The rate constant is the figure of merit referred to in the literature for quantitative comparisons of photocatalytic activity [19].

Zero- and half-order rate laws have also been used to describe the photocatalyst degradation reactions. In particular, the half-order rate law has been shown to more accurately model photocatalytic reactions on metal-oxide surfaces, most likely due to the combination of zero- and first-order chemical reactions involved in photodegradation [19]. The differential and integrated forms of the half-order rate law are as follows:

$$-\frac{dc}{dt} = A\_S \sqrt{I}kt^{1/2} \tag{9}$$

and:

$$c^{1/2}(t) = c\_0^{1/2} - A\_S \sqrt{I} \frac{k}{2} t \tag{10}$$

respectively. In this study, we calculate (c<sup>1</sup>/<sup>2</sup> <sup>−</sup> <sup>c</sup> 1/2 <sup>0</sup> )/A<sup>S</sup> and plot this as a function of the UV irradiation time, with the half-order rate constant determined from the slope of the linear best-fit. While the light intensity is a factor in calculating the rate constant, the same light intensity was used for each experiment in this study, and its effect on the rate of each reaction can be negated.

## 5. Results

In this section, we discuss the photocatalytic activity of the fabricated ZnO thin films and its dependency on the annealing temperature and film thickness. We also discuss the stability of these films with continued use by investigating the time-dependence of the photocatalytic activity. As described in Section 3, the annealing temperature and film thickness have been shown to affect both the surface roughness and the efficiency of electron-hole pair production, which have been shown to affect photcatalysis. Furthermore, surface degradation of ZnO with exposure to aqueous environments has been reported, which could affect the long-term stability of ZnO thin films with respect to photocatalysis [4,17].
