¹Department of Plant Science, The Pennsylvania State University, USA
²Department of Plant Pathology and Environmental Microbiology, The Pennsylvania State University, USA
³Department of Meteorology and Atmospheric Science, The Pennsylvania State University, USA

^*Corresponding author: Decoteau DR, The Pennsylvania State University, University Park, PA, USA, Email: drd10@ psu.edu

Copyright: © Lloyd KL, et al. 2019. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Article Information: Submission: 31/07/2019; Accepted: 04/09/2019; Published: 07/09/2019

Corona discharge ozone (O₃) generators provide valuable data on the response of vegetation to O₃ exposures. Systems that use dried air as a feed gas, instead of pure or concentrated oxygen (O₂), are known to produce trace nitrogen (N) oxidant byproducts that may be toxic to plants. This study quantified the concentration of total N oxidants, including nitrogen oxides (NO_x, the sum of NO and NO₂), dinitrogen pentoxide (N₂O₅), and nitric acid (HNO₃ ), relative to O₃ levels in a continuous stirred tank reactor (CSTR). The CSTR was part of computer-controlled O₃ delivery and monitoring system used to study effects of O₃ on vegetation within a greenhouse with charcoal-filtered air. Ozone was generated via corona discharge with dried air as a feed gas, and the system was operated at different O₃ output levels and environmental conditions in seven separate trials. At O₃ levels up to 330 ppb, total N oxidant concentrations in the CSTR did not exceed 9.2 ppb, when averaged over 60-sec intervals. Across all trials, the relationship between total N oxidants and O₃ was described by the equation: N oxidants (ppb = 0.0108 [O₃(ppb)] + 3.37 (R² = 0.46; n = 205). In this system, trace N oxidant levels produced under typical experimental conditions are not expected to cause direct toxicity to vegetation. Therefore, corona discharge O₃ generators provide a suitable, inexpensive method of O₃ production for vegetation exposure studies.

Air pollution; Ozone; Oxides of nitrogen; CSTR exposure chambers; Corona discharge

Tropospheric ozone (O₃): Ambient tropospheric O₃ is one of the most phytotoxic air pollutants in the U.S., if not the world [1-3]. Ozone is a secondary air pollutant formed from photochemical reactions of nitrogen oxides (NO_x, the sum of NO and NO₂) and volatile organic compounds (VOCs). The U.S. Environmental Protection Agency (EPA) has designated O₃ as one of six criteria air pollutants regulated by the National Ambient Air Quality Standards (NAAQS) to protect human beings, agricultural crops, forest ecosystems, and other resources in the U.S. from ambient exposure [4]. Ozone is of regional-scale importance in the U.S. due to its multi-day lifetime within slowmoving, stagnant high-pressure systems and, as a result, may cause damage to vegetation many miles downwind from the origin of its precursors, NO_x and VOCs [3].

Exposing vegetation to O₃ in chambers: Ozone generators are important tools to study effects of O₃ on vegetation. Since O₃ cannot be stored, it must be created on-demand at the application site. In vegetation studies, O₃ generators have been essential for controlled studies evaluating the harmful effects of Ov on vegetation [5,6], including the U.S. EPA’s National Crop Loss Assessment Network, which established dose-response relationships between O₃ and crop yields using a network of open-top chambers [7]. Current research relies on O₃ generators to evaluate the impacts of O₃ on different crop species [8,9], at different times of day [10,11], and interacting with climatic changes [8].

Generation of ozone: Ozone generators dissociate molecular oxygen (O₂) into atomic oxygen (O). Subsequently, the O atoms produced by the generator combine with O₂ to form O₃ [12,13].

The most common O₃ generation method, corona discharge, uses a high-voltage electric arc to split O₂ (i.e., similar to lightning), but if air is used as a feed gas instead of pure O₂ , NO_x and N₂O₅ also form [14]. Corona discharge or high-voltage electric arc generators produce electrons that collide with and dissociate molecules of O₂ and N₂ in the air, resulting in formation of O and NO. As a byproduct, NO is then oxidized by O₃ until it reaches the highest possible oxidation states as N₂O₅ or HNO₃ . If water vapor is present, the N₂O₅ is hydrated to HNO₃ [15]. Nitrous oxide (N₂O), another byproduct, is not formed via dissociation but rather from an excited N₂ molecule, which reacts with an O molecule. N₂ O is chemically stable and not further oxidized [15]. To prevent byproducts, pure O₂ is the ideal feed gas for corona discharge, providing up to twice the O₃ output of dried air. In addition to compressed O₂ , oxygen concentrators can be used to increase O₂ levels in a pressurized ambient air supply. However, both options raise production costs. Ambient air is therefore the least expensive feed gas but necessitates frequent corona cell maintenance. When ambient air is dried (i.e., to a dewpoint ≤-60 ° C), O₃ output is more consistent, and maintenance needs are reduced relative to humid air [14,16].

In contrast, UV lamps use ambient air as a feed gas without generating trace N oxidants. Light emitted by mercury lamps, in the UV region at 185 nm, irradiates O₂ present in ambient air, similar to the photochemistry of the stratosphere, where O₂ absorbs radiation from 240 to 120 nm. In this process, one photon can generate up to two O₃ molecules when it dissociates one O₂molecule to two single O molecules, which then primarily combine with O₂ to produce O₃ [12]. Other types of lamps, such as xenon excimers, are also capable of dissociating O₂ and have been studied for practical O₃ production [17]. However, mercury remains standard, and new coatings have been developed to increase lamp lifetime [18]. In spite of advances in technology for UV lamps, corona discharge generators provide the most efficient, durable O₃ production, particularly for studies requiring high flow rates of O₃ and distribution to multiple exposure chambers from a single source [19]. However, the cost of pure O₂ as a feed gas can be prohibitive for long-term studies [20].

Potential toxicity of N oxidants to vegetation: When using ambient air as a feed gas, it is important to quantify the potentially phytotoxic N oxidant compounds that result from passing O₂ and N₂ through a high-voltage dielectric field. These include NO and NO₂ [20], as well as HNO₃ [22] formed from hydrated N₂O₅ [15].

Under the Clean Air Act, EPA has maintained the secondary NAAQS, which protect public welfare, for NO₂ in the form of an annual arithmetic mean of 53 ppb, which is considered sufficient to protect vegetation from direct effects of gaseous NO₂ [23]. However, EPA [4] acknowledged the causal relationship between gaseous NO_x and injury to vegetation. Further, EPA concluded that, at ambient exposure levels for NO₂ , exposure-response relationships were variable, due to differences in biological and environmental factors among experiments [24]. In some cases, low NO₂ levels increased growth, likely via foliar N fertilization. For continued (> 14 d) exposures of several hours per day, growth reductions generally appeared when NO_x levels exceeded 100 to 500 ppb, depending on the plant species [24]. EPA supported the conclusion that gaseous HNO₃ can cause “changes” to vegetation but did not find evidence of direct injury from HNO₃ exposure [4]. They noted that dry deposition of HNO₃ and resulting changes (e.g., degradation of epicuticular waxes) may increase adverse effects of other pollutants, such as O₃ , on vegetation [25]. However, Mortensen and Jørgensen [20] suggested that trace N oxidants produced by corona discharge can also protect vegetation against O₃ damage. Few studies have been performed since the 1993 EPA summary [24], leading to a lack of information on the long-term effects of low concentrations of HNO₃ and total atmospheric oxidized N (NO_y ) on plant species [4].

Terminology in this paper that defines inorganic N species is as follows:

Previous studies have measured the production of N oxidants relative to O₃ by corona discharge with dried air, as emitted directly from the generator. Notably, different systems and conditions (e.g., temperature, pressure) cause variation in relative yields [25]. Using infrared spectroscopy Harris et al. [14] and Kogelschatz and Baessler [17] estimated a molar ratio of HNO₃ to O₃ ranging from 0.007 to 0.010 per 1 mol O₃ . Bubbling the generator air stream through water and measuring dissolved NO₃ - resulted in higher HNO₃ : O₃ ratios, in the range of 0.020 to 0.025 [20,25].

The objective of this study was to quantify trace N oxidants, as NO_y , present in a charcoal-filtered-air greenhouse during O₃ production via corona discharge. Specifically, the relationship between O₃ and NO_y concentrations within continuous stirred tank reactor (CSTR) treatment chambers [6], used to study the response of vegetation to O₃ [11] was of interest. Quantification of N oxidant byproducts under experimental operating conditions was necessary to ensure that the use of pure air as a feed gas for the corona discharge generator would not produce injurious levels of N oxide byproducts, potentially confounding the effects of O₃ treatment on vegetation.

Ozone was generated via corona discharge, with dried air as a feed gas, and distributed among 16 separate CSTRs, each with a volume of ~2.6 m³, as described by Lloyd et al. [11]. Data were recorded within a single representative CSTR. In order to quantify NO_y, oxidation products were reduced via thermal dissociation at 650 ° C to NO₂ and measured using chemiluminescence (Model 42i-TL; Thermo Environmental Corp., Franklin, MA) as NO, NO₂ , and NO_x Ozone was generated via corona discharge, with dried air as a feed gas, and distributed among 16 separate CSTRs, each with a volume of ~2.6 m³, as described by Lloyd et al. [11]. Data were recorded within a single representative CSTR. In order to quantify NO_y, oxidation products were reduced via thermal dissociation at 650 ° C to NO₂ and measured using chemiluminescence (Model 42i-TL; Thermo Environmental Corp., Franklin, MA) as NO, NO₂, and NO_x, with a 60-sec averaging time. The thermal dissociation column was constructed as described by Wooldridge et al. [26] and placed in one of the CSTRs. Measurements recorded when the thermal dissociator was at ambient temperature and when heated to 650 ° C reflect NO_x and NO_y levels, respectively. Therefore, the difference between those quantities (i.e., NO_y – NO_x ) gives an approximation for NO_z.

Across seven trials, background NO_x levels in the CSTR, prior to operation of the O₃ generator, ranged from approximately 2 to 5 ppb, with about 45% in the form of NO (data not shown). For comparison, across the U.S., the average annual NO₂ concentration for ambient air is ≈15 ppb [27]. Production of O₃ from the generator decreased the proportion of NO, since O₃ reacts with NO to form NO₂ [25]. The minimum levels of NO recorded during ambient conditions and O₃ production were 0.77 and 0.22 ppb, respectively (data not shown).

Background NO_x levels during operation of the O₃ generator (and thermal dissociation column) can be inferred from the intercept term of least squares regression, with a mean of 3.37 ppb across trials (Figure 1). Background NO₂ was included in the analysis of the relationship between NO_y and O₃ to provide a maximum estimate of the level of NO_y plants may be exposed to in CSTRs.

Across all measurements, NO_y concentration was linearly related to both O₃ concentration and electric current, but electric current explained slightly more variation (R² = 0.54) than did O₃ (R² = 0.46, Figure 1). For the range of O₃ concentrations tested, up to 331 ppb, the maximum NO_y levels recorded did not exceed 9.2 ppb. Across the seven trials, the linear relationship between NO_y and O₃ in the CSTR was described by: NO_y (ppb) = 0.0108 [O₃ (ppb)] + 3.37 (R² = 0.46, n = 205 Figure 1).

For the seven separate trials, results of least squares regression are given in Table 1. R² values ranged from 0.32 to 0.95, slope coefficients ranged from 0.0074 to 0.0168 ppb NO_y ·ppb O₃^-1, and intercepts ranged from 1.83 to 4.65 ppb NO_y . Notably, the number of individual measurements and overall time period varied among trials (Table 1). Based on least squares regression, there was no relationship between slopes or intercepts and air temperature, relative humidity, or photosynthetically active radiation (PAR) across the seven trials (R² = 0.01 to 0.04, data not shown).

Comparison of the trials on two dates with the highest R² values, 9 September and 31 August, helps explain the variation in slopes and intercepts. Relative to 31 August, the regression from 9 September produced a larger intercept (3.43 vs. 1.83 ppb NO_y ) and smaller slope (0.0079 vs. 0168 ppb NO_y . ppb O₃^-1, Table 1). Figure 2 shows the O₃ concentration (ppb), ratio of O₃ (ppb) to NO_y (ppb), and power efficiency (ppb O₃ ·mA^-1) plotted relative to the elapsed measurement time (min) on both days. On 9 September, a larger number of measurements (n = 44 vs. 20) was recorded over a longer time period (185 vs. 67 min). On both dates, target O₃ levels in the CSTR were initially set at greater than 300 ppb and decreased over time. The rate was slower on 9 September than on 31 August, and the ratio of O₃ : NO_y was less variable, as well as the generator power efficiency. These differences reflect inherent “noise” in the O₃ distribution and monitoring systems, which can result during computerized feedback when adjustment of electrical current to the generator overshoots target levels. The air in each CSTR is replaced (via a blower system) approximately once per minute, leading to a time lag between O₃ input from the generator, equilibration of the gas composition in the CSTR, and travel distance for a sample parcel to reach the O₃ monitor [11]. Figure 3 shows that the more rapid decrease in target O₃ concentration and resulting “bumps” in generator output caused underestimates of predicted NO_y at low O₃ levels and overestimates at high levels, relative to the expected values based on regression of the overall data set.

The slope obtained via linear regression of all CSTR observations (0.0108 ppb NO_y ·ppb O₃^-1) falls within the range of prior measurements (0.007 to 0.025), confirming that present observations of NO₃ were within reported values [14,15,20,25].

Using the predictive equation derived from all seven trials, at a CSTR O₃ level of 300 ppb, the expected NO_y concentration was ≈6.6 ppb. With maximum CSTR concentrations far lower than the secondary NAAQS for NO₂ , set at 53 ppb [24], direct plant injury is unlikely. Further, Stripe et al. [22] treated two snap bean (Phaseolus vulgaris L.) genotypes with HNO₃ during the daytime for 6 weeks. Exposure to peak daily HNO₃ concentrations of 80 to 100 ppb did not significantly affect bean plant biomass. Therefore, NO₂ and HNO₃ generated by corona discharge in the CSTR system are unlikely to incite direct phytotoxic effects.

The system-specific estimates of NO_y production via corona discharge, with dried air as a feed gas, are in agreement with other studies, and these levels are not expected to be directly phytotoxic in the form of NO₂ or HNO₃ . Notably, Taylor et al. [28] suggested that elevated levels of both O₃ and HNO₃ are representative of ambient conditions in the outdoor growth environment. However, O₃ has a much higher phytotoxicity than NO_x [24]. Therefore, studies employing this method of O₃ generation should produce valid results testing the effect of O₃ treatment on vegetation, though actual N by product outputs will vary among exposure systems.