Effects of UV-A Light Treatment on Ammonia, Hydrogen Sulfide, Greenhouse Gases, and Ozone in Simulated Poultry Barn Conditions

Gaseous emissions, a side effect of livestock and poultry production, need to be mitigated to improve sustainability. Emissions of ammonia (NH3), hydrogen sulfide (H2S), greenhouse gases (GHGs), and odorous volatile organic compounds (VOCs) have a detrimental effect on the environment, climate, and quality of life in rural communities. We are building on previous research to bring advanced oxidation technologies from the lab to the farm. To date, we have shown that ultraviolet A (UV-A) has the potential to mitigate selected odorous gases and GHGs in the context of swine production. Much less research on emissions mitigation has been conducted in the context of poultry production. Thus, the study objective was to investigate whether the UV-A can mitigate NH3, H2S, GHGs, and O3 in the simulated poultry barn environment. The effects of several variables were tested: the presence of photocatalyst, relative humidity, treatment time, and dust accumulation under two different light intensities (facilitated with fluorescent and light-emitting diode, LED, lamps). The results provide evidence that photocatalysis with TiO2 coating and UV-A light can reduce gas concentrations of NH3, CO2, N2O, and O3, without a significant effect on H2S and CH4. The particular % reduction depends on the presence of photocatalysts, relative humidity (RH), light type (intensity), treatment time, and dust accumulation on the photocatalyst surface. In the case of NH3, the reduction varied from 2.6–18.7% and was affected by RH and light intensity. The % reduction of NH3 was the highest at 12% RH and increased with treatment time and light intensity. The % reduction of NH3 decreased with the accumulation of poultry dust. The % reduction for H2S had no statistical difference under any experimental conditions. The proposed treatment of NH3 and H2S was evaluated for a potential impact on important ambient air quality parameters, the possibility of simultaneously mitigating or generating GHGs. There was no statistically significant change in CH4 concentrations under any experimental conditions. CO2 was reduced at 3.8%–4.4%. N2O and O3 concentrations were reduced by both direct photolysis and photocatalysis, with the latter having greater % reductions. As much as 6.9–12.2% of the statistically-significant mitigation of N2O was observed. The % reduction for O3 ranged from 12.4–48.4%. The results warrant scaling up to a pilot-scale where the technology could be evaluated with economic analyses.

preliminary assessment of the broader impact of proposed treatment on important ambient and indoor air quality parameters.

Experimental System
An experimental system to evaluate gas emission reduction efficacy under UV-A irradiation was based on a modified setup from previous research, Figure 1 [21,30]. Three mass flow controllers were used to control the dilution of the standard gases and pure air and the RH. A 500 mL glass gas sampling bulb (Supelco, Bellefonte, PA, USA) was installed before and after the UV treatment reactor. The standard gases flowing through the 200 mL reactor were irradiated with UV-A through a quartz window. The reactor bottom was made from an ordinary glass that was coated with a photocatalyst (nanostructured TiO 2 at 10 µg·cm −2 from PureTi, Cincinnati, OH, USA). The reactor temperature was maintained at 25 ± 3 • C while the heat generated by the UV lamps was discharged from the UV chamber by circulating-cooling tubes connected to the isothermal water bath.
An experimental system to evaluate gas emission reduction efficacy under UV-A irradiation was based on a modified setup from previous research, Figure 1 [21,30]. Three mass flow controllers were used to control the dilution of the standard gases and pure air and the RH. A 500 mL glass gas sampling bulb (Supelco, Bellefonte, PA, USA) was installed before and after the UV treatment reactor. The standard gases flowing through the 200 mL reactor were irradiated with UV-A through a quartz window. The reactor bottom was made from an ordinary glass that was coated with a photocatalyst (nanostructured TiO2 at 10 μg•cm −2 from PureTi, Cincinnati, OH, USA). The reactor temperature was maintained at 25 ± 3 °C while the heat generated by the UV lamps was discharged from the UV chamber by circulating-cooling tubes connected to the isothermal water bath.
The gas flow rate into the reactor ranged from 60-300 mL•min −1 , resulting in a range of 40 s to 200 s treatment time. The treatment time was selected to represent typical air exchange rates inside poultry barns. NH3 standard gas (70.5 ppm in N2, ultra-high-purity, UHP, grade, Praxair, Ames, IA, USA) was diluted to 30 ppm, a typical concentration reported inside poultry barns [31][32][33]. Similarly, H2S standard gas (5.2 ppm in N2, UHP grade, Praxair) was diluted to 0.5 ppm. The humidifier was used to adjust to three RH levels (approximately 0, 12%, 40%, and 60%). GHGs and O3 concentrations were measured simultaneously with changes to NH3 and H2S. Because ambient air was used as a source, a certain naturally occurring level of GHG and O3 naturally exists in the background of all experiments in the absence of all treatments. Typical values of NH3, H2S, CH4, CO2, N2O, and O3 were 30 ppm, 0.5 ppm, 2.2 ppm, 350 ppm, 0.2 ppm, and 23 ppb in the absence of any photolytic treatment, i.e., background control runs. The detection methods for each are described below. These environmental parameters were consistent with those observed in typical USA poultry and livestock production barns [34,35]. All experiments were performed in triplicate.

UV-A Irradiation Sources
Fluorescent lamps (Spectroline, Westbury, NY, USA) and an LED lamp (ONCE, Plymouth, MN, USA) were used; both UV lamps have a primary wavelength of 365 nm (Table 1). The lamps were installed 0.20 m above the UV treatment reactor. The light intensity was measured at 0.20 m distance from the source with an ILT-1700 radiometer equipped with an NS365 filter and SED033 detector (International Light Technologies, Peabody, MA, USA). The LED had ~4× greater intensity compared The gas flow rate into the reactor ranged from 60-300 mL·min −1 , resulting in a range of 40 s to 200 s treatment time. The treatment time was selected to represent typical air exchange rates inside poultry barns. NH 3 standard gas (70.5 ppm in N 2 , ultra-high-purity, UHP, grade, Praxair, Ames, IA, USA) was diluted to 30 ppm, a typical concentration reported inside poultry barns [31][32][33]. Similarly, H 2 S standard gas (5.2 ppm in N 2 , UHP grade, Praxair) was diluted to 0.5 ppm. The humidifier was used to adjust to three RH levels (approximately 0, 12%, 40%, and 60%). GHGs and O 3 concentrations were measured simultaneously with changes to NH 3 and H 2 S. Because ambient air was used as a source, a certain naturally occurring level of GHG and O 3 naturally exists in the background of all experiments in the absence of all treatments. Typical values of NH 3 , H 2 S, CH 4 , CO 2 , N 2 O, and O 3 were 30 ppm, 0.5 ppm, 2.2 ppm, 350 ppm, 0.2 ppm, and 23 ppb in the absence of any photolytic treatment, i.e., background control runs. The detection methods for each are described below. These environmental parameters were consistent with those observed in typical USA poultry and livestock production barns [34,35]. All experiments were performed in triplicate.

UV-A Irradiation Sources
Fluorescent lamps (Spectroline, Westbury, NY, USA) and an LED lamp (ONCE, Plymouth, MN, USA) were used; both UV lamps have a primary wavelength of 365 nm (Table 1). The lamps were installed 0.20 m above the UV treatment reactor. The light intensity was measured at 0.20 m distance from the source with an ILT-1700 radiometer equipped with an NS365 filter and SED033 detector (International Light Technologies, Peabody, MA, USA). The LED had~4× greater intensity compared with the fluorescent lamp for nearly identical power consumption (measured w/ P3 wattage meter, Lexington, NY, USA). Lamps position inside the UV chamber (4 fluorescent lamps and an array of 108 LED chips on an Al board sources were used) Atmosphere 2019, 10, x FOR PEER REVIEW 4 of 32 with the fluorescent lamp for nearly identical power consumption (measured w/ P3 wattage meter, Lexington, NY, USA). Lamps position inside the UV chamber (4 fluorescent lamps and an array of 108 LED chips on an Al board sources were used)

Ammonia and Hydrogen Sulfide
NH3 concentrations were measured in real-time using a Drager Xam 5600 portable gas analyzer (Luebeck, Germany) with NH3 sensors (range: 0-300 ppm). The Drager analyzer was calibrated using Drager calibration software and standard gases. H2S was measured (also in real-time) using a gas monitoring system (OMS-300, Smart Control & Sensing Inc., Daejeon, Rep. of Korea) equipped with the H2S/C-50 electrochemical gas sensors from Membrapor Co. (Wallisellen, Switzerland; range: 0-50 ppm. The H2S gas sensor was calibrated using standard gases. The flow rate used in this study was 60,300 mL•min −1 , NH3, and H2S samples were collected in 5 L Tedlar bags to overcome the limitations associated with the sample collection flow rates required by the portable analyzers (NH3: 0.5 L•min −1 and H2S: 2 L•min −1 ).

Greenhouse Gases
GHGs samples were collected using syringes by drawing gas from the sampling bulbs and injecting them into evacuated 5.9 mL Exetainer vials (Labco Ltd, UK). Samples were analyzed on a gas chromatography (GC) equipped with a flame ionization detector (FID) and electron capture detector (ECD) detectors (SRI Instruments, Torrance, CA, USA). Standard calibrations were constructed daily using 10.3 ppm and 20.5 ppm CH4; 1,005 ppm and 4,010 ppm CO2; 0.101 ppm and 1.01 ppm N2O; and pure helium was used at 0 ppm (Air Liquide America, Plumsteadville, PA, USA) [36]. Samples were stored at 4 °C immediately after collection and analyzed within one day after sampling.

Ozone
A real-time O3 detector (Gas Sensing, Hull, IA, USA) was connected to the monitoring system (Series 500 monitor, Aeroqual, New Zealand) and installed in the UV treatment chamber. The O3 concentration was analyzed by measuring the concentration of O3 collected in a 0.5 L glass gas sampling bulb connected downstream from the UV reactor. The sensor was factory-calibrated before use. The detection range was from 0-0.05 ppm.

Dust Collection in A Poultry Barn
The presence of accumulated dust could potentially compromise the effectiveness of photocatalyst. Thus, in order to evaluate the effect of dust on mitigation efficiency, dust was collected with the fluorescent lamp for nearly identical power consumption (measured w/ P3 wattage meter, Lexington, NY, USA). Lamps position inside the UV chamber (4 fluorescent lamps and an array of 108 LED chips on an Al board sources were used)

Ammonia and Hydrogen Sulfide
NH3 concentrations were measured in real-time using a Drager Xam 5600 portable gas analyzer (Luebeck, Germany) with NH3 sensors (range: 0-300 ppm). The Drager analyzer was calibrated using Drager calibration software and standard gases. H2S was measured (also in real-time) using a gas monitoring system (OMS-300, Smart Control & Sensing Inc., Daejeon, Rep. of Korea) equipped with the H2S/C-50 electrochemical gas sensors from Membrapor Co. (Wallisellen, Switzerland; range: 0-50 ppm. The H2S gas sensor was calibrated using standard gases. The flow rate used in this study was 60,300 mL•min −1 , NH3, and H2S samples were collected in 5 L Tedlar bags to overcome the limitations associated with the sample collection flow rates required by the portable analyzers (NH3: 0.5 L•min −1 and H2S: 2 L•min −1 ).

Greenhouse Gases
GHGs samples were collected using syringes by drawing gas from the sampling bulbs and injecting them into evacuated 5.9 mL Exetainer vials (Labco Ltd, UK). Samples were analyzed on a gas chromatography (GC) equipped with a flame ionization detector (FID) and electron capture detector (ECD) detectors (SRI Instruments, Torrance, CA, USA). Standard calibrations were constructed daily using 10.3 ppm and 20.5 ppm CH4; 1,005 ppm and 4,010 ppm CO2; 0.101 ppm and 1.01 ppm N2O; and pure helium was used at 0 ppm (Air Liquide America, Plumsteadville, PA, USA) [36]. Samples were stored at 4 °C immediately after collection and analyzed within one day after sampling.

Ozone
A real-time O3 detector (Gas Sensing, Hull, IA, USA) was connected to the monitoring system (Series 500 monitor, Aeroqual, New Zealand) and installed in the UV treatment chamber. The O3 concentration was analyzed by measuring the concentration of O3 collected in a 0.5 L glass gas sampling bulb connected downstream from the UV reactor. The sensor was factory-calibrated before use. The detection range was from 0-0.05 ppm.

Dust Collection in A Poultry Barn
The presence of accumulated dust could potentially compromise the effectiveness of photocatalyst. Thus, in order to evaluate the effect of dust on mitigation efficiency, dust was collected 2.3. Ammonia and Hydrogen Sulfide NH 3 concentrations were measured in real-time using a Drager Xam 5600 portable gas analyzer (Luebeck, Germany) with NH 3 sensors (range: 0-300 ppm). The Drager analyzer was calibrated using Drager calibration software and standard gases. H 2 S was measured (also in real-time) using a gas monitoring system (OMS-300, Smart Control & Sensing Inc., Daejeon, Rep. of Korea) equipped with the H 2 S/C-50 electrochemical gas sensors from Membrapor Co. (Wallisellen, Switzerland; range: 0-50 ppm. The H 2 S gas sensor was calibrated using standard gases. The flow rate used in this study was 60,300 mL·min −1 , NH 3, and H 2 S samples were collected in 5 L Tedlar bags to overcome the limitations associated with the sample collection flow rates required by the portable analyzers (NH 3 : 0.5 L·min −1 and H 2 S: 2 L·min −1 ).

Greenhouse Gases
GHGs samples were collected using syringes by drawing gas from the sampling bulbs and injecting them into evacuated 5.9 mL Exetainer vials (Labco Ltd, UK). Samples were analyzed on a gas chromatography (GC) equipped with a flame ionization detector (FID) and electron capture detector (ECD) detectors (SRI Instruments, Torrance, CA, USA). Standard calibrations were constructed daily using 10.3 ppm and 20.5 ppm CH 4 ; 1,005 ppm and 4,010 ppm CO 2 ; 0.101 ppm and 1.01 ppm N 2 O; and pure helium was used at 0 ppm (Air Liquide America, Plumsteadville, PA, USA) [36]. Samples were stored at 4 • C immediately after collection and analyzed within one day after sampling.

Ozone
A real-time O 3 detector (Gas Sensing, Hull, IA, USA) was connected to the monitoring system (Series 500 monitor, Aeroqual, New Zealand) and installed in the UV treatment chamber. The O 3 concentration was analyzed by measuring the concentration of O 3 collected in a 0.5 L glass gas sampling bulb connected downstream from the UV reactor. The sensor was factory-calibrated before use. The detection range was from 0-0.05 ppm.

Dust Collection in A Poultry Barn
The presence of accumulated dust could potentially compromise the effectiveness of photocatalyst. Thus, in order to evaluate the effect of dust on mitigation efficiency, dust was collected at Poultry Teaching Farm (Ames, IA, USA). Three Styrofoam boxes that held two glass plates (blank and coated with TiO 2 ) were placed inside the barn horizontally and accumulated dust over time ( Figure 2, part a). Also, three aluminum (Al) foil coupons were attached to simultaneously measure the weight of accumulated dust per area. The Styrofoam boxes were then removed from the barn, one by one, at one-week intervals for three weeks. Then, the same glass plates were mounted into the UV reactor (as the 'Bottom layer: Glass in Figure 1) for testing. The weight of accumulated dust was estimated by subtracting the final from the initial Al foil coupon weight and extrapolated to the entire bottom layer glass area of the reactor. In addition, the effect of accumulated dust on light absorption at the glass with and without TiO 2 was measured using a 300-lumen bulb and a radiometer equipped with an XRD340B detector (International Light Technologies, Peabody, MA, USA), Figure 2 (part b).
Atmosphere 2019, 10, x FOR PEER REVIEW 5 of 31 one by one, at one-week intervals for three weeks. Then, the same glass plates were mounted into the UV reactor (as the 'Bottom layer: Glass in Figure 1) for testing. The weight of accumulated dust was estimated by subtracting the final from the initial Al foil coupon weight and extrapolated to the entire bottom layer glass area of the reactor. In addition, the effect of accumulated dust on light absorption at the glass with and without TiO2 was measured using a 300-lumen bulb and a radiometer equipped with an XRD340B detector (International Light Technologies, Peabody, MA, USA), Figure 2 (part b).

Data analysis, Accounting for Sample Losses due to Adsorption
Gas samples were collected after a 1 h of equilibration time under each treatment condition. Small, yet consistent losses to target gases were observed over the course of experiments with the photocatalyst. Thus, the standard gas recoveries were also measured and reported as 'adsorption' series in the Results. The adsorption to the photocatalyst was assumed to be responsible for the losses and accounted for in the analyses. The overall mean reduction for each measured gas was estimated using: where: CCon and CTreat are the mean measured concentrations in control and treated air, respectively.

Statistical Analysis
The R program (version 3.4.2) was used to analyze the effects of the catalyst, lamp-type, and environmental parameters on the reduction of the target gases by one way ANOVA. This statistical analysis generated p-values for evaluating whether a specific parameter/factor had a significant influence on treatment. A significant difference was defined for a p-value <0.05.

Data analysis, Accounting for Sample Losses Due to Adsorption
Gas samples were collected after a 1 h of equilibration time under each treatment condition.
Small, yet consistent losses to target gases were observed over the course of experiments with the photocatalyst. Thus, the standard gas recoveries were also measured and reported as 'adsorption' series in the Results. The adsorption to the photocatalyst was assumed to be responsible for the losses and accounted for in the analyses. The overall mean reduction for each measured gas was estimated using: where: C Con and C Treat are the mean measured concentrations in control and treated air, respectively.

Statistical Analysis
The R program (version 3.4.2) was used to analyze the effects of the catalyst, lamp-type, and environmental parameters on the reduction of the target gases by one way ANOVA. This statistical analysis generated p-values for evaluating whether a specific parameter/factor had a significant influence on treatment. A significant difference was defined for a p-value <0.05.

Ammonia
In general, longer treatment time, use of photocatalyst, increased light intensity, and the presence of moisture in treated air improved the % NH 3 reduction. The highest reduction was 18.7% for 200 s treatment, LED photocatalysis at 12% RH, and no dust conditions. Dust decreased the performance of the photocatalyst. Detailed summaries and statistical significance of the effects of each treatment are presented in the subsections below.
3.1.1. Effect of the Photocatalyst, Relative Humidity, Light Intensity, and Treatment Time The controlled NH 3 concentration used in the control group using the standard gas was 29.8 ± 1.2 ppm. Figure 3 illustrates the NH 3 reduction at three treatment conditions (direct photolysis, photocatalysis, and adsorption (by TiO 2 )) under different RH, and 40 s (part a) and 200 s (part b) treatment time, respectively. Photocatalysis resulted in a 2.6-18.7% reduction, which was statistically significant for nearly all conditions (Tables A1 and A2, Appendix A). In comparison, direct photolysis resulted in no treatment or negligible % reduction and was not statistically significant. The controlled NH3 concentration used in the control group using the standard gas was 29.8 ± 1.2 ppm. Figure 3 illustrates the NH3 reduction at three treatment conditions (direct photolysis, photocatalysis, and adsorption (by TiO2)) under different RH, and 40 s (part a) and 200 s (part b) treatment time, respectively. Photocatalysis resulted in a 2.6-18.7% reduction, which was statistically significant for nearly all conditions (Tables A1 and A2, Appendix A). In comparison, direct photolysis resulted in no treatment or negligible % reduction and was not statistically significant.
Closer inspection of the patterns in the effectiveness of photocatalysis showed that it was affected by RH, light (type) intensity, and treatment time. The LED lamp (having ~4× higher intensity) facilitated a higher % reduction, the greatest (~2×) improvement observed at 12% RH. Moreover, the s (a) (b)  Closer inspection of the patterns in the effectiveness of photocatalysis showed that it was affected by RH, light (type) intensity, and treatment time. The LED lamp (having~4× higher intensity) facilitated a higher % reduction, the greatest (~2×) improvement observed at 12% RH. Moreover, the statistical difference in this improvement was shown for both RH 12% and 40% at 200 s treatment ( Figure 3, part b). Figure 4 highlights the % reduction with different treatment times and RH. The LED-based photocatalysis at lower RH (12% and 40%) outperformed the fluorescent-based treatment for NH 3 mitigation.
Atmosphere 2019, 10, x FOR PEER REVIEW 7 of 32 statistical difference when comparing different light intensities (fluorescent vs. LED) at the same relative humidity (p < 0.05). The % reduction was the highest at 12% (7.3-18.7%, p < 0.05). Error bars signify ± standard deviation. Figure 4 highlights the % reduction with different treatment times and RH. The LED-based photocatalysis at lower RH (12% and 40%) outperformed the fluorescent-based treatment for NH3 mitigation. Superscript (*) signifies a statistical difference compared to the control (p < 0.05), and the different characters (a, b, c, d) signify the statistical difference between treatments associated with one UV lamp type (p < 0.05). Error bars signify ± standard deviation.

Effect of Poultry Dust
Dust accumulation on TiO2 had a detrimental effect on the effectiveness of photocatalysis ( Figure  5, Tables A3 and A4), particularly at low RH (12%). In addition, accumulated poultry dust absorbed light, and the linear increase (from 14.1 to 40.1%) in light absorption with dust accumulation on the photocatalyst surface over time (from 6.9 to 16.3 mg•cm −2 ) was observed. The average light absorption was 27 ± 12%, and the dust accumulation was 11 ± 4 mg•cm −2 ( Table 2). Superscript (*) signifies a statistical difference compared to the control (p < 0.05), and the different characters (a, b, c, d) signify the statistical difference between treatments associated with one UV lamp type (p < 0.05). Error bars signify ± standard deviation.

Effect of Poultry Dust
Dust accumulation on TiO 2 had a detrimental effect on the effectiveness of photocatalysis ( Figure 5, Tables A3 and A4), particularly at low RH (12%). In addition, accumulated poultry dust absorbed light, and the linear increase (from 14.1 to 40.1%) in light absorption with dust accumulation on the photocatalyst surface over time (from 6.9 to 16.3 mg·cm −2 ) was observed. The average light absorption was 27 ± 12%, and the dust accumulation was 11 ± 4 mg·cm −2 ( Table 2). The values in the table report the mean ± standard deviation.   There was no statistical significance of the change in the reduction at RH 60% ( Figure 5, part b). The low (12%) RH had the most considerable decrease in mitigation (from 18.7% to 5.1%) under the LED light, yet it was still statistically significant even with the highest dust accumulation of 16.3 mg·cm −2 (p < 0.05). In other words, the LED-based treatment was still performing well, regardless of dust accumulation (p < 0.05).

Hydrogen Sulfide
The controlled H 2 S concentration in the control group using the standard gas was 0.52 ± 0.02 ppm. No statistically significant reduction in H 2 S concentration was observed under any experimental conditions (p > 0.05), even with the most favorable conditions of 200 s, photocatalyst usage, LED irradiation, and elevated moisture ( Figure 6). Similarly, there was no statistical difference associated with the dust accumulation at 12% RH regardless of the light type (intensity) and treatment time (Figure 7).
The values in the table report the mean ± standard deviation.
There was no statistical significance of the change in the reduction at RH 60% ( Figure 5, part b). The low (12%) RH had the most considerable decrease in mitigation (from 18.7% to 5.1%) under the LED light, yet it was still statistically significant even with the highest dust accumulation of 16.3 mg•cm -2 (p < 0.05). In other words, the LED-based treatment was still performing well, regardless of dust accumulation (p < 0.05).

Hydrogen Sulfide
The controlled H2S concentration in the control group using the standard gas was 0.52 ± 0.02 ppm. No statistically significant reduction in H2S concentration was observed under any experimental conditions (p > 0.05), even with the most favorable conditions of 200 s, photocatalyst usage, LED irradiation, and elevated moisture ( Figure 6). Similarly, there was no statistical difference associated with the dust accumulation at 12% RH regardless of the light type (intensity) and treatment time (Figure 7).

Greenhouse Gases
The treatment of target gases (NH3, H2S) was evaluated for a potential impact on important ambient air quality parameters, the possibility of simultaneously mitigating or generating GHGs. No GHGs were fed into the reactor; however, as noted previously, the air source naturally contained measurable amounts of these compounds. Thus, the GHGs concentrations in the treatment and control were compared.

Methane
The average concentration in controls was 2.2 ± 0.1 ppm. There was no statistically significant change in CH4 concentration under direct photolysis, photocatalysis, and adsorption to the catalyst. Moreover, there were no statistically significant changes, regardless of RH, light type (intensity), treatment time, and dust accumulation when NH3 and H2S standard gases were treated with UV (Figures 8 and 9).

Greenhouse Gases
The treatment of target gases (NH 3 , H 2 S) was evaluated for a potential impact on important ambient air quality parameters, the possibility of simultaneously mitigating or generating GHGs. No GHGs were fed into the reactor; however, as noted previously, the air source naturally contained measurable amounts of these compounds. Thus, the GHGs concentrations in the treatment and control were compared.

Methane
The average concentration in controls was 2.2 ± 0.1 ppm. There was no statistically significant change in CH 4 concentration under direct photolysis, photocatalysis, and adsorption to the catalyst. Moreover, there were no statistically significant changes, regardless of RH, light type (intensity), treatment time, and dust accumulation when NH 3 and H 2 S standard gases were treated with UV (Figures 8 and 9).

Greenhouse Gases
The treatment of target gases (NH3, H2S) was evaluated for a potential impact on important ambient air quality parameters, the possibility of simultaneously mitigating or generating GHGs. No GHGs were fed into the reactor; however, as noted previously, the air source naturally contained measurable amounts of these compounds. Thus, the GHGs concentrations in the treatment and control were compared.

Methane
The average concentration in controls was 2.2 ± 0.1 ppm. There was no statistically significant change in CH4 concentration under direct photolysis, photocatalysis, and adsorption to the catalyst. Moreover, there were no statistically significant changes, regardless of RH, light type (intensity), treatment time, and dust accumulation when NH3 and H2S standard gases were treated with UV (Figures 8 and 9). (a)

Carbon Dioxide
The average CO2 concentration in control samples (i.e., present naturally in air) was 350 ± 25 ppm, and no mitigation was observed under photolysis and adsorption. Interestingly, there was a 3.8% (mean) reduction at 200 s photocatalysis with LED at 12% RH ( Figure 10 and Table A5). Specifically, the mitigation was 3.2% and 4.4% when NH3 and H2S standard gases were treated, respectively. However, there was no statistical difference between the two standard gases (p > 0.05). There was no CO2 reduction under dust accumulation ( Figure 11).

Carbon Dioxide
The average CO2 concentration in control samples (i.e., present naturally in air) was 350 ± 25 ppm, and no mitigation was observed under photolysis and adsorption. Interestingly, there was a 3.8% (mean) reduction at 200 s photocatalysis with LED at 12% RH ( Figure 10 and Table A5). Specifically, the mitigation was 3.2% and 4.4% when NH3 and H2S standard gases were treated, respectively. However, there was no statistical difference between the two standard gases (p > 0.05). There was no CO2 reduction under dust accumulation ( Figure 11). Error bars signify ± standard deviation.

Carbon Dioxide
The average CO 2 concentration in control samples (i.e., present naturally in air) was 350 ± 25 ppm, and no mitigation was observed under photolysis and adsorption. Interestingly, there was a 3.8% (mean) reduction at 200 s photocatalysis with LED at 12% RH ( Figure 10 and Table A5). Specifically, the mitigation was 3.2% and 4.4% when NH 3 and H 2 S standard gases were treated, respectively. However, there was no statistical difference between the two standard gases (p > 0.05). There was no CO 2 reduction under dust accumulation ( Figure 11).

Nitrous Oxide
In general, mitigation of concentration was observed under both direct photolysis and photocatalysis, with greater reductions with TiO2 photocatalysts. However, there was no statistical difference between the two conditions. There was no apparent relationship between N2O % reduction and other controlled parameters.

Comparison of N2O Mitigation when Treating NH3 and H2S Standard Gas
The average N2O concentration in control was 0.24 ± 0.03 ppm. As much as 6.9% and 12.2% of the statistically-significant % reduction were observed for 200 s photocatalysis with LED at 12% RH when NH3 and H2S were treated, respectively (Table 3). In general, statistically-significant % reductions were found for more experimental conditions for H2S than NH3. However, there was no

Nitrous Oxide
In general, mitigation of concentration was observed under both direct photolysis and photocatalysis, with greater reductions with TiO2 photocatalysts. However, there was no statistical difference between the two conditions. There was no apparent relationship between N2O % reduction and other controlled parameters.

Comparison of N2O Mitigation when Treating NH3 and H2S Standard Gas
The average N2O concentration in control was 0.24 ± 0.03 ppm. As much as 6.9% and 12.2% of the statistically-significant % reduction were observed for 200 s photocatalysis with LED at 12% RH when NH3 and H2S were treated, respectively (Table 3). In general, statistically-significant % reductions were found for more experimental conditions for H2S than NH3. However, there was no Figure 11. CO 2 mitigation under different poultry dust levels at 12% of relative humidity in photocatalysis. Superscript (*) signifies a statistical difference compared to the control (p < 0.05), and the character (a) signifies there is no statistical difference under the two different treatment times (p > 0.05). Error bars signify ± standard deviation.

Nitrous Oxide
In general, mitigation of concentration was observed under both direct photolysis and photocatalysis, with greater reductions with TiO 2 photocatalysts. However, there was no statistical difference between the two conditions. There was no apparent relationship between N 2 O % reduction and other controlled parameters.

Comparison of N 2 O Mitigation when Treating NH 3 and H 2 S Standard Gas
The average N 2 O concentration in control was 0.24 ± 0.03 ppm. As much as 6.9% and 12.2% of the statistically-significant % reduction were observed for 200 s photocatalysis with LED at 12% RH when NH 3 and H 2 S were treated, respectively (Table 3). In general, statistically-significant % reductions were found for more experimental conditions for H 2 S than NH 3 . However, there was no significant difference between the % reduction resulting from the use of two standard gas treatments (i.e., p-values in Table 3). We further investigated the apparent mitigation of N 2 O (a potent GHGs) by averaging the results for H 2 S and NH 3 to elucidate possible mechanisms responsible for this finding. The statistically significant mitigation was observed in both direct photolysis and photocatalysis, at 3.3-6.5% and 2.8-9.5%, respectively ( Figure 12, Tables A6 and A7). In general, photocatalysis was more effective for reducing N 2 O than photolysis alone ( Figure 12, parts a and b, Tables A6 and A7). However, there was no statistical difference between the two treatments (i.e., at dry and 12% RH, Table A7) were compared. Similarly, no apparent statistical significance was found for variation of the treatment time, dust accumulation, lamp type, and RH (Figures 13 and 14, Tables A8-A10). This is because there are few statistically significant % reduction of N 2 O for variable parameters, without any apparent trend.
Atmosphere 2019, 10, x FOR PEER REVIEW 13 of 32 significant difference between the % reduction resulting from the use of two standard gas treatments (i.e., p-values in Table 3).

Effect of Photocatalyst, Light Type, Relative Humidity, Dust, and Treatment Time on N2O
We further investigated the apparent mitigation of N2O (a potent GHGs) by averaging the results for H2S and NH3 to elucidate possible mechanisms responsible for this finding. The statistically significant mitigation was observed in both direct photolysis and photocatalysis, at 3.3-6.5% and 2.8-9.5%, respectively ( Figure 12, Tables A6 and A7). In general, photocatalysis was more effective for reducing N2O than photolysis alone ( Figure 12, parts a and b, Tables A6 and A7). However, there was no statistical difference between the two treatments (i.e., at dry and 12% RH, Table A7) were compared. Similarly, no apparent statistical significance was found for variation of the treatment time, dust accumulation, lamp type, and RH (Figures 13 and   14, Tables A8-A10). This is because there are few statistically significant % reduction of N2O for variable parameters, without any apparent trend.
(a)     (a) % reduction at a relative humidity of 12% and 60% in direct photolysis; (b) % reduction at a relative humidity of 12% and 60% in photocatalysis. Superscript (*) signifies a statistical difference compared to the control (p < 0.05), and the different characters (a, b) signify the statistical difference between treatments associated with one UV lamp type and one relative humidity (p < 0.05). Error bars signify ± standard deviation.

Ozone
Concerns about O3 generation when UV light is used were addressed in this research. In general, O3 concentrations were significantly reduced under direct photolysis and photocatalysis. The direct photolysis treatment did not result in a clear relationship between the controlled parameters and the O3 % reduction. Photocatalysis resulted in an improved mitigation dependent on the light type (intensity) at RH of 12% and 60%. However, there was no effect associated with treatment time and dust accumulation.  (a) % reduction at a relative humidity of 12% and 60% in direct photolysis; (b) % reduction at a relative humidity of 12% and 60% in photocatalysis. Superscript (*) signifies a statistical difference compared to the control (p < 0.05), and the different characters (a, b) signify the statistical difference between treatments associated with one UV lamp type and one relative humidity (p < 0.05). Error bars signify ± standard deviation.

Ozone
Concerns about O 3 generation when UV light is used were addressed in this research. In general, O 3 concentrations were significantly reduced under direct photolysis and photocatalysis. The direct photolysis treatment did not result in a clear relationship between the controlled parameters and the O 3 % reduction. Photocatalysis resulted in an improved mitigation dependent on the light type (intensity) at RH of 12% and 60%. However, there was no effect associated with treatment time and dust accumulation.

Comparison of Mitigation under NH 3 and H 2 S Standard Gas
The average concentration of O 3 in the control group was 22.6 ± 6.5 ppb. As much as 46.5% & 50.3% of the statistically-significant % reduction in O 3 concentrations were observed for 200 s photocatalysis with LED at 12% RH when NH 3 and H 2 S were treated, respectively (Table 4). However, there was no significant difference between the reduction resulting from the use of two standard gas treatments (i.e., p-values, Table 4). In addition, astatistically significant % reduction was found at more experimental conditions under 200 s treatment time compared with 40 s. O 3 % reduction for various parameters was evaluated using the average of % reductions for treatments of NH 3 and H 2 S standard gases. The was significant mitigation on direct photolysis and photocatalysis of 12.4-23.5% and 21.6-48.4%, respectively (Figure 15, Tables A11 and A12). In general, photocatalysis was more effective for reducing O 3 concentrations than was direct photolysis. However, there was no statistical difference between the direct photolysis and photocatalysis except the condition of 200 s at dry and 12% RH ( Figure 15, part b). Notably, the % reduction increased~4× for 200 s photocatalysis with LED at 12% RH (Table A12).
In the case of direct photolysis, there was no clear trend and statistical significance for the treatment time, dust accumulation, lamp type, and RH (Figures 15-17, Tables A11-A14). More statistically-significant % reductions were found for 200 s treatments than 40 s. Moreover, even the maximum accumulation of dust did not have a significant impact on % reduction (Figure 17, Tables A15 and A16) Figure 17. Comparison of O3 mitigation under different poultry dust levels. (a) % reduction at a relative humidity of 12% and 60% in direct photolysis; (b) % reduction at a relative humidity of 12% and 60% in photocatalysis. Superscript (*) signifies a statistical difference compared to the control (p < 0.05), and the different characters (a, b) signify the statistical difference between treatments associated with one UV lamp type and one relative humidity (p < 0.05). Error bars signify ± standard deviation.

Ammonia and Hydrogen Sulfide
In this study, NH3 mitigation was only effective when photocatalysis was used, regardless of the (a) % reduction at a relative humidity of 12% and 60% in direct photolysis; (b) % reduction at a relative humidity of 12% and 60% in photocatalysis. Superscript (*) signifies a statistical difference compared to the control (p < 0.05), and the different characters (a, b) signify the statistical difference between treatments associated with one UV lamp type and one relative humidity (p < 0.05). Error bars signify ± standard deviation.

Ammonia and Hydrogen Sulfide
In this study, NH 3 mitigation was only effective when photocatalysis was used, regardless of the type of UV lamp, which is generally consistent with previous research. Research [37,38] suggests that a shorter wavelength (e.g., 220 nm) is needed to mitigate NH 3 with photolysis. Other researchers have also reported on the weak adsorption of NH 3 to the TiO 2 coated surface at room temperature [39,40].
The greatest mitigation of NH 3 was at 12% RH in photocatalysis. The % reduction decreased with either dry air or increasing RH. In general, the higher % reduction is achieved under low (or dry) humidity conditions. This is due to the adsorption of water on the TiO 2 surface [39,41,42], which, in turn, inhibits the mitigation of the target substances [43][44][45]. A similar trend (at least for low RH) was observed in this study. However, the % reduction was found to be decreased in the dry condition, which was expected to show the highest % reduction. One explanation could be that the decreased % reduction in dry conditions is due to the absence of HO radicals produced by the photocatalysis of water. HO radicals make it easier to oxidize NH 3 [46]. The optimal RH for the % reduction is different depending on the type of target gas. The comparison of optimum RH for selected target gases in the photocatalysis is summarized in Table 5.  Table 6 summarize previous research on the % reduction of selected target gases important in animal production systems via photocatalysis with UV-A. The % reduction of NH 3 was~30%, but it required longer treatment (>6 min). In this study, the % reduction was~9% on average (max: 18.7%, min: 2.56%) and 200 s. In general, the % reduction increases as the UV light intensity and treatment time of photocatalysis action increase [38,47,51]. This study also showed an increase in NH 3 % reduction with increasing light intensity and treatment time. Photocatalysis was affected by dust accumulation. In particular, the increase in dust at high RH conditions canceled the NH 3 % reduction effect. This is because when dust accumulated, poultry dust continually increased the absorption of the UV light. Zhu et al. [21] reported that dust accumulation (in a swine barn) had no effect on the % reduction of VOCs. In this study, H 2 S showed no % reduction effect in the treatment system using UV-A light and TiO 2 based photocatalysis. Previous studies with higher TiO 2 coating density and light intensity have shown a mitigation effect [53,54].

Greenhouse Gases and Ozone
CH 4 was not affected by any treatment in this study. The results add to a mixed body of knowledge. Our previous study with PureTi coating [1] did not show a statistically significant % reduction. Another research reported photocatalytic CH 4 % reduction at a low product yield and low energy efficiency [55]. However, two previous studies [18,19] (UV-A light) reported an 11-27% reduction. The reasons for our lack of apparent treatment could be due to low TiO 2 coating density, and the possibility that the mitigation effect was offset by forming CH 4 from the reduction of CO 2 [56][57][58]. CO 2 showed a 3.8% mitigation only under the RH of 12% with 200 s of LED irradiation. Although there is previous research demonstrating photocatalytic reduction of CO 2 to CH 4 under specific conditions [59], there is no chemical reason to expect that photocatalysis under these conditions (aerobic atmosphere and standard TiO 2 catalyst) could reduce CO 2 ; in fact, CO 2 is the oxidative endpoint for photocatalytic oxidation of virtually all carbon-containing compounds. It is thus tempting to suggest that an indirect mechanism for any observed CO 2 mitigation must exist, such as conversion to carbonates or surface absorption. By whatever mechanism, similar CO 2 concentration reductions were also observed by previous studies [19,58]. N 2 O and O 3 were mitigated in both direct photolysis and photocatalysis. In general, N 2 O and O 3 are known not to absorb significantly in the UV-A range, meaning that they are not subject to direct photolytic degradation at these wavelengths. However, indirect effects through more complex reaction paths can certainly affect their observed concentrations. Similarly, under photocatalytic conditions, where direct absorption by the substrate is not required, reasonably direct removal can occur. Previous research [60] reported that the N 2 O photolysis rate was inhibited at >230 nm, and [61] reported that O 3 % reduction does not occur efficiently at >305 nm. However, this and our previous study [1] showed the mitigation effect of N 2 O and O 3 under 365 nm. In the case of N 2 O, the % reduction was about 3.3-6.5%, also in the case of O 3 , the % reduction was 12.4-24.1% (Table 7). As expected, photocatalysis showed higher mitigation for N 2 O than direct photolysis. However, this study result is quite different from those of previous studies. This is because previous research [20,37,38,51,52,62] indicates that by-products like N 2 O and N 2 are generated via photocatalysis reaction in the presence of NH 3 , butthe levels of N 2 O and N 2 vary according to treatment conditions and wavelength. Only one previous study showed a consistent trend with this study [1]. In theory, TiO 2 could only be activated by UV light with a wavelength of <387.5 nm due to its considerable bandgap energy [63,64]. Also, N 2 O has been reported to be mitigated with UV-A, although its efficiency is lower than that of UV-C [63]. How much of this process actually decrease the N 2 O produced by NH 3 decomposition has not been investigated in this study. However, the low but statistically significant % reduction could be due to other factors such as direct reduction by photocatalysis, indirect reduction by electrochemical reactions during the decomposition of other substances such as O 3 , and adsorption of by-products on the TiO 2 .
In the case of O 3 , the previous results are different depending on the type of TiO 2 coating material, but it is reported that O 3 reduction does not occur >290 nm with TiO 2 [65]. However, O 3 has been reported to increase the reduction of target gas through the formation of ozonide radicals during photocatalysis [66][67][68][69]. Thus, during photocatalysis, O 3 concentration can be reduced due to the formation of ozonide radicals that are beneficial for reducing other target gases. In this study, the % reduction was 3.4-9.7% for N 2 O and 20.4-48.4% for O 3 .

Conclusions
The results of the study provide evidence that photocatalysis with TiO 2 coating and UV-A light can reduce gas concentrations of NH 3 , CO 2 , N 2 O, and O 3; without significant effect on H 2 S and CH 4 . The particular % reduction depends on the presence of photocatalyst, RH, light type (intensity), treatment time, and dust accumulation on the photocatalyst surface. In the case of NH 3 , the % reduction varied from 2.6-18.7% and was affected by RH and light intensity. The % reduction of NH 3 was the highest at 12% RH and increased with treatment time and light intensity. The % reduction of NH 3 decreased with the accumulation of poultry dust. The % reduction for H 2 S had no statistical difference under any experimental conditions. The proposed treatment of NH 3 and H 2 S was evaluated for a potential impact on important ambient air quality parameters, the possibility of simultaneously mitigating/generating GHGs. There was no statistically significant change in CH 4 concentrations under any tested conditions. CO 2 was reduced at 3.8-4.4%. N 2 O and O 3 concentrations were reduced by both direct photolysis and photocatalysis, with the latter having greater % reductions. As much as 6.9-12.2% of the statistically-significant % reduction of N 2 O was observed. The % reduction for O 3 ranged from 12.4-48.4%. The results warrant scaling up to pilot-scale where the technology could be evaluated with economic analyses. It is necessary to investigate the practical applicability to the real system through large scale studies.

Conflicts of Interest:
The author does not declare a conflict of interest. The funders did not play any role in the study design, data collection, analysis, interpretation, and decision to write a manuscript or present results.   Table A16. O 3 mitigation under the different dust levels at a relative humidity of 40% and 60% in photocatalysis. Value in the table report % reduction ± standard deviation (p-value) and are an average between treatments of NH 3 and H 2 S standard gases. Bold font signifies statistical significance.