Responses of Secondary Inorganic PM 2 . 5 to Precursor Gases in an Ammonia Abundant Area in North Carolina

Secondary inorganic fine particulate matter (iPM2.5) constitutes a significant amount of the atmospheric PM2.5. The formation of secondary iPM2.5 is characterized by thermodynamic equilibrium gas-particle partitioning of gaseous ammonia (NH3) and aerosol ammonium (NH4). To develop effective strategies for controlling atmospheric PM2.5, it is essential to understand the responses of secondary iPM2.5 to different precursor gases. In southeastern North Carolina, the amount of NH3 is in excess to fully neutralize acidic gases (i.e., NH3-rich conditions). NH3-rich conditions are mainly attributed to the significant NH3 emissions in the region, especially from the large amounts of animal feeding operation (AFO). To gain a better understanding of the impact of NH3 on the formation of secondary iPM2.5 in this area, the responses of iPM2.5 to precursor gases under different ambient conditions were investigated based upon three-year monitoring data of the chemical components in iPM2.5, gaseous pollutants, and meteorological conditions. The gas ratio (GR) was used to assess the degree of neutralization via NH3, and ISORROPIA II model simulation was used to examine the responses of iPM2.5 to changes in the total NH3, the total sulfuric acid (H2SO4), and the total nitric acid (HNO3). It was discovered that under different ambient temperature and humidity conditions, the responses of iPM2.5 to precursor gases vary. In general, iPM2.5 responds nonlinearly to the total NH3 but linearly to the total H2SO4 and the total HNO3. In NH3rich regions, iPM2.5 is not sensitive to changes in the total NH3, but it is very sensitive to changes in the total H2SO4 and/or the total HNO3. Reducing the total H2SO4, as opposed to the total HNO3 or the total NH3, leads to a significant reduction in iPM2.5 and is thus a more effective strategy for decreasing the concentration of iPM2.5. This research provides insight into controlling and regulating PM2.5 in NH3-rich regions.

Ammonia is the major alkaline gas that may react with acidic gases to form iPM 2.5 in ambient air, and this process is also called gas-particle partitioning of NH 3 -NH 4 + .The neutralization degree of NH 3 can be characterized by gas ratio (GR), which is in Eq. (1) (Ansari and Pandis, 1998): where TA is total available ammonia, including NH 3 and NH 4 + (in the unit of µmole m -3 ).TS is total sulfate including SO 4 2-, bisulfate (HSO 4 -) and H 2 SO 4 (in the unit of µmole m -3 ).TN is total available nitrate, including NO 3 -and HNO 3 (in the unit of µmole m -3 ).When GR > 1, the total available ammonia exceeds the amount needed to fully neutralize both total sulfate and total available nitrate, and this is defined as NH 3 -rich condition; under this condition, the changes of total available ammonia may not be a key factor to affect the concentration of iPM 2.5 .When 0 < GR < 1, the amount of total available ammonia is adequate to fully neutralize total sulfate but not total available nitrate, and only NH 4 NO 3 formation is limited by NH 3 ; under this condition, the decrease of NH 3 may lead to corresponding decrease of NH 4 NO 3 .When GR < 0, the amount of total available ammonia is not enough to fully neutralize either total sulfate or total available nitrate, and both (NH 4 ) 2 SO 4 and NH 4 HSO 4 are limited by NH 3 (Wang-Li, 2015).
Based on USEPA's National Emission Inventory (NEI), animal feeding operation (AFO) contributed to more than 70% of the total NH 3 emissions in the United States (U.S.) (U.S. EPA, 2015b).While the AFO NH 3 emissions present a great potential to the formation of secondary iPM 2.5 in some regions where a significant amount of AFO facilities are located, the dynamic contribution of such emissions to the ambient iPM 2.5 is not well understood spatially and temporally.To gain holistic understanding of atmospheric PM 2.5 , it is essential to understand the dynamic responses of atmospheric iPM 2.5 to the AFO NH 3 emissions under different atmospheric conditions and geographical locations (Wang-Li, 2015).
To study the thermodynamic equilibrium processes of iPM 2.5 and its precursor gases, thermodynamic equilibrium model such as ISORROPIA was developed to simulate the gas-particle partitioning of NH 3 -NH 4 + (Nenes et al., 1998(Nenes et al., , 1999)).In ISORROPIA, the phase changes (e.g., gas, liquid, and solid) and interaction of different chemical species (NH 4 + , NO 3 -, SO 4 2-, Cl -, and Na + ) as well as the impacts of temperature (T) and relative humidity (RH) on partitioning of NH 3 -NH 4 + are simulated (Fountoukis and Nenes, 2007;Fountoukis et al., 2009).In an application of ISORROPIA to assess the formation of iPM 2.5 at an agricultural site located in eastern North Carolina (NC), Walker et al. (2006) examined the change of iPM 2.5 concentration in response to the 50% reduction of total NH 3 (gas + aerosol), total HNO 3 (gas + aerosol) and total H 2 SO 4 (aerosol) in winter and summer of 1999-2000.It was discovered that the 50% reduction of total NH 3 had the least impact on iPM 2.5 concentration as compared to the 50% reductions in total HNO 3 and H 2 SO 4 .This research suggested that at the agricultural sites with elevated atmospheric NH 3 concentration, the iPM 2.5 is more sensitive to acidic gases rather than NH 3 .In another iPM 2.5 study, Goetz et al. (2012) investigated the effect of NH 3 emissions from swine production facilities on iPM 2.5 concentration at three locations in eastern NC.The iPM 2.5 chemical composition data obtained from air quality monitoring stations of NC Division of Air Quality and gaseous pollutant concentrations measured or cited from literature were used to conduct iPM 2.5 simulation using ISORROPIA under three T and RH conditions.The results indicated that the simulation results of iPM 2.5 by ISORROPIA agreed well with the observation.Furthermore, this research revealed that high precursor gas concentrations, low T, and high RH led to higher chance for secondary iPM 2.5 formation.
In order to gain advanced understanding of the formation of iPM 2.5 as impacted by AFO NH 3 emissions, Li et al. (2014b) investigated the formation of secondary iPM 2.5 in response to total NH 3 inside a production house and in the vicinity of an egg farm in the southeastern U.S. Onsite measurements of NH 3 concentrations and PM 2.5 chemical components at in-house and ambient locations were used to conduct ISORROPIA II simulation to predict gas-particle partitioning of NH 3 -NH 4 + .Li et al. (2014b) confirmed that the most significant reduction of iPM 2.5 can be caused by the reduction of total H 2 SO 4 instead of NH 3 and this is because the formation of iPM 2.5 is limited by the availability of acidic gases when NH 3 exceeds the amount needed to fully neutralize acid gases.
In addition to the local-scale simulation, ISORROPIA II has been embedded into chemical transport model (CTM) to study more complicated atmospheric gas-particle partitioning on regional and/or global scales.Paulot and Jacob (2014) estimated the contribution of agricultural NH 3 emissions to the ambient PM 2.5 in the U.S. using GEOS-Chem global CTM coupled with ISORROPIA II.This modeling practice reported that there is a 0.36 µg m -3 increase of ambient PM 2.5 concentration caused by NH 3 emissions associated with the U.S. food export activities.
As the research gap exists in quantifying the formation of secondary iPM 2.5 experimentally and/or through model simulation in AFO region, the objectives of this research were as follows: (1) investigation of the neutralization degree of NH 3 ; (2) examination of seasonal variation of PM 2.5 mass closure; and (3) study of the responses of secondary iPM 2.5 to the changes of total NH 3 , total HNO 3 , and total H 2 SO 4 in an animal production area of NC under different meteorological conditions.

Since NH 4
+ , SO 4 2-, and NO 3 -account for the majority of atmospheric iPM 2.5 (Bell et al., 2007), this research focuses on the responses of NH 4 + , SO 4 2-, and NO 3 -to the changes of total NH 3 , total HNO 3 and total H 2 SO 4 at a site where atmospheric NH 3 is abundant due to NH 3 emissions from AFO.

Research Site Selection and Data Collection
To achieve the research objective, the research site has to be in an area where a significant amount of AFO facilities are present.In addition, simultaneous measurements of gasphase pollutants (e.g., NH 3 and HNO 3 ) and particle-phase ions (e.g., NH 4 + , SO 4 2-, and NO 3 -) are required input data to conduct ISORROPIA II simulation.Thus, the availability of the required measurement data is another criterion for the site selection.Based upon these two site selection criteria, a monitoring site (35.23146 N, 77.568792 W) in Lenoir County of NC was selected for this research.Fig. 1 shows this site marked on the AFO distribution map developed by Zhao and Wang-Li (2015).The majority of the swine and poultry farms are in the south-central area of NC where the research site was also located.
At the selected site, 24-hr average measurements of PM 2.5 chemical components (e.g., NH 4 + , SO 4 2-, NO 3 -) and hourly measurements of NH 3 , reactive oxides of nitrogen (NO y ), T and RH were taken in 2002-2004 and reported at the EPA's Air Quality System (AQS) website (https://aqs.epa.gov/api).The 24-hr PM 2.5 chemical composition data were measured every sixth day while the hourly NO y concentrations, T, and RH were measured continuously for three years.To keep these data on the same time scale, hourly data on specific days when the PM 2.5 chemical composition data were taken were converted into 24-hr average data to match daily measurements of PM 2.5 chemical components.While the HNO 3 gas concentration is a required input for ISORROPIA II model, the HNO 3 gas measurements were not available.Thus, the NO y concentration may be used to indirectly determine the HNO 3 gas concentration.In general, NO y includes nitric oxide (NO), nitrous oxide (N 2 O), HNO 3 , peroxyacetyl nitrate (PAN), nitrous acid (HONO), nitrate radical (NO 3 ), dinitrogen pentoxide (N 2 O 5 ), organic nitrates etc.In addition, NO, NO 2 , HNO 3 , and PAN are the major components of NO y (Finlayson-Pitts and Pitts, 2000;Seinfeld and Pandis, 2006).In atmosphere, HNO 3 is mainly formed through the oxidation of NO 2 following Eq.( 2) in the daytime (Pun and Seigneur, 2001;Jacobson, 2005): where reaction (2) mainly happens in the daytime in which the photochemical reaction can provide adequate hydroxyl radical (OH).
To estimate the HNO 3 concentration through NO y concentration, NO 2 concentration should be first determined.Luke et al. (2010) discovered that the fractional contribution of NO 2 in NO y varied with the time of day.While no NO 2 measurement data were available in Lenoir County, there were simultaneous measurements of hourly NO y and NO 2 concentrations in 2014 in Wake County, which is another county in southeastern NC (Fig. 1).The median NO 2 /NO y ratios for each of 24 hours in Wake County were used to estimate NO 2 /NO y ratios in Lenoir County.The fractional contributions of NO 2 to the total NO y are set at different values in each of 24 hours based on the measurement data in Wake County, and this trend is consistent with the measurements performed by Luke et al. (2010).

Mass Closure Profile
The contribution of inorganic PM 2.5 to total PM 2.5 mass contribution was analyzed using mass closure profile of PM 2.5 chemical components.Major ions including iPM 2.5 anions/cations, up to 47 crustal elements, EC and OC as well as PM 2.5 mass concentration were simultaneously measured by MetOne SASS sampler.To develop a mass closure profile, PM 2.5 mass concentration measured by MetOne SASS Teflon was used to analyze the mass closure of PM 2.5 chemical components.When the sum of all the chemical component concentrations were greater than the measured PM 2.5 mass concentration, those data were excluded from mass closure analysis.
According to Dillner et al. (2012) and Weber et al. (2003), organic carbon matter (OCM) can be calculated using Eq. ( 3) to account for elements other than carbon: where OCM = organic carbon matter, OC m = organic carbon measurement, and OC b = field blank.
The contributions of various chemical components to PM 2.5 were calculated using Eq. ( 4): where P i = percentage of the chemical component i in PM 2.5 mass concentration, C i = concentration of chemical component i, and C m = measured PM 2.5 mass concentration from MetOne SASS Teflon filter.

ISORROPIA II Settings
For this study, all the iPM 2.5 chemical components are assumed to be internally mixed, and the thermodynamic equilibrium is also assumed to be established very rapidly.The ISORROPIA II allows user to specify the problem type (forward or reverse) and thermodynamic state (stable or metastable); in this study, ISORROPIA II is set as forward + stable.As NH 3 -NH 4 + -SO 4 2--HNO 3 -NO 3 -system is determined as the research focus, all the other species concentrations, including total sodium (Na + ), total hydrochloric acid (HCl), total calcium (Ca 2+ ), total potassium (K + ), and total magnesium (Mg 2+ ) are set as 0.
The examination of responses of iPM 2.5 to the total NH 3 , total HNO 3 , and total H 2 SO 4 was based on the minimum, median and maximum concentrations of the input parameters measured in 2002-2004 using ISORROPIA II (Table 1).
Five T and RH conditions were used to test the responses of secondary iPM 2.5 to the changes of total NH 3 , total H 2 SO 4 , and total HNO 3 .These combinations include maximum T + minimum RH, minimum T + maximum RH, median T + median RH, maximum T + maximum RH, minimum T + median RH, of which, maximum T + maximum RH and minimum T + median RH represent the summer and winter conditions, respectively.

Statistical Analysis
All the data analyses, analysis of variance (ANOVA) test, paired t-test, and Tukey's Honest Significant Difference (HSD) test were performed using R.

Statistical Characterization of the Field Measurements
Table 2 lists the summary of three-year measurements.Median NH 3 gas concentrations in winter and summer were 0.74 and 3.17 µg m -3 , respectively, and median NH 4 + concentrations in winter and summer were 1.15 and 1.64 µg m -3 , respectively.Research performed by Walker et al. (2006) at a site in the neighboring county, Sampson County (Fig. 1), reported the median NH 3 gas concentrations of 2.60 and 6.18 µg m -3 in winter and summer, respectively, and median NH 4 + aerosol concentrations of 1.90 and 1.69 µg m -3 in winter and summer, respectively.Comparatively, the NH 3 gas and NH 4 + aerosol concentrations at the site in Sampson County were greater than the concentrations  Goetz et al. (2012).

Responses of Total iPM 2.5 to Different Conversion Percentages of NO 2 to HNO 3
As it has been stated, 0%, 0.5%, 1%, 5%, 10%, 25%, 50% and 100% of NO 2 were assumed to be converted through Eq. (1) to HNO 3 ; in each case scenario, gas-phase HNO 3 and total HNO 3 concentrations were calculated based upon the different conversion ratios.The calculated total HNO 3 along with other model inputs (median values) were then used to simulate the formation of iPM 2.5 using ISORROPIA II.The predicted iPM 2.5 was compared with the measured iPM 2.5 concentrations under different NO 2to-HNO 3 conversion percentages (Fig. 2).
Fig. 2 shows that as the percentage of NO 2 to HNO 3 conversion was increased from 0% to 100%, the ratio of predicted iPM 2.5 over measured iPM 2.5 was increased from 0.95 to 1.80, correspondingly.This can be explained that in Lenoir County, ambient NH 3 is abundant (GR > 1); while NH 3 preferentially reacted with H 2 SO 4 to form (NH 4 ) 2 SO 4 and ammonium bisulfate (NH 4 HSO 4 ), excessive NH 3 could react with HNO 3 to form NH 4 NO 3 (Tanner et al., 1981;Yu et al., 2005).When NH 3 concentration in the atmosphere exceeds the amount needed to fully neutralize both total HNO 3 and total H 2 SO 4 , more HNO 3 can favor the formation of NH 4 NO 3 , then increase the concentration of iPM 2.5 .
When the conversion percentage of NO 2 to HNO 3 was set at 3.63%, the ratio of predicted iPM 2.5 over measured iPM 2.5 was close to 1, which may be indicative of a reasonable estimate of the conversion ratio.Moreover, Goetz et al. (2012) reported the minimum, median, and maximum concentrations of total HNO 3 were 0.26, 1.07, and 5.24 µg m -3 , respectively, in eastern NC in 2001-2004.The estimated minimum, median, and maximum total HNO 3 in this research were 0.24, 1.17, and 4.24 µg m -3 , respectively in 2002-2004.Thus, the estimation of the HNO 3 concentration generally agreed with the previous research.

Neutralization Degree of NH 3 : GR
The neutralization degree of NH 3 was characterized by The concentration of HNO 3 is calculated based on the assumption that 3.63% of NO 2 was converted to HNO 3 through Eq. ( 1).TNH 3 = NH 3 + NH 4 + ; THNO 3 = HNO 3 + NO 3 -; TH 2 SO 4 = H 2 SO 4 ; TNH 3 , THNO 3 , TH 2 SO 4 are all expressed as the equivalent concentration; iPM 2.5 is the sum of NH 4 + , NO 3 -and SO 4 2-; T is temperature; RH is relative humidity.

Fig. 2.
The ratio of predicted iPM 2.5 concentration over measured iPM 2.5 concentration against conversion percentage of NO 2 to HNO 3 .
GR.The ranges of GR in four seasons are listed in Table 3.This area was classified as the NH 3 -rich area since the GR values were greater than 1 in most of the time.Specifically, in summer, 29 out of 30 data points were characterized as GR ≥ 1; even in winter, 23 out of 28 data points were GR > 1.In addition, the median GRs in four seasons were all larger than 1 with the highest GR-12.69 in summer.This observation indicated that the research site in Lenoir County was dominated by NH 3 -rich condition such that there was a great potential for neutralizing acidic gases with excessive NH 3 in atmosphere.

Seasonal Variations of PM 2.5 Chemical Speciation
In Lenoir County, PM 2.5 mass concentration was measured in 2002-2004 using two methods, Federal Reference Method (FRM) and MetOne SASS chemical speciation sampler; both datasets came from gravimetric analysis.To check the data quality, the comparison between these two datasets were performed and the results are listed in Table 4.
As indicated by Table 4, the PM 2.5 mass concentration measurements from MetOne SASS sampler are significantly greater than the measurements from FRM method.The seasonal mass closure profile is shown in Fig. 3.As can be seen from Fig. 3, iPM 2.5 accounted for a large proportion of the total PM 2.5 , and this fraction was highest in spring (51%) and lowest in summer (40%).For iPM 2.5 alone, three chemical species, SO 4 2-, NO 3 -and NH 4 + , also accounted for different fractions in four seasons.The results of Tukey's test were shown in Table 5.
As indicated by Table 5, the PM 2.5 concentration in summer was significantly higher than in winter; this can be explained by the seasonal variations of the major chemical components.The OCM and SO 4 2-together accounted for 41.4-54.2% of PM 2.5 mass concentration in four seasons; both species' concentrations were higher in summer than in winter.In addition, SO 4 2-was also the major component of iPM 2.5 and it accounted for 47.9-67.0% in the whole iPM 2.5 .The NO 3 -concentration was significantly higher in winter than in summer.This may be due to semi-volatile characteristic of NH 4 NO 3 , according to Olszyna et al. (2005), SO 4 2-salts and NO 3 -salts own different levels of thermal stability, the NH 4 NO 3 is not thermally stable so that it may decompose to gaseous HNO 3 and NH 3 when environmental conditions (high T and low RH) do not favor the particle phase.On the other hand, SO 4 2-salts are thermally stable compared with NO 3 -salts.In summer, the ambient conditions do not favor the formation of NH 4 NO 3 .However, the (NH 4 ) 2 SO 4 can still form in summer; thus, SO 4 2-was the year-round major component of iPM 2.5 and this is consistent with the finding from Holt et al. (2015).However, as for NH 4 + , there is no significant difference between four seasons.The NH 3 -rich conditions dominated in four seasons; NH 3 exceeded the amount needed to fully neutralize both HNO 3 and H 2 SO 4 .The NH 4 + in the form of (NH 4 ) 2 SO 4 was higher in summer than in winter and NH 4 + in the form of NH 4 NO 3 was higher in winter than in summer; thus, both seasonal variations offset the change of the NH 4 + concentration in winter and summer.

Response of Secondary iPM 2.5 to Total NH 3
In this analysis, median total H 2 SO 4 concentration and median total HNO 3 concentration were used under five T and RH case scenarios (Table 1) to simulate the iPM 2.5 formation with total NH 3 concentration ranging from 0.54 to 37.94 µg m -3 .The responses of iPM 2.5 to the change of total NH 3 are shown in Fig. 4.Under different conditions, the responses of iPM 2.5 to total NH 3 were different.The concentration of iPM 2.5 responded to the total NH 3 nonlinearly.
Under maximum T + minimum RH (305.95K + 39%), the change of iPM 2.5 concentration is caused by the change of SO 4 2-, HSO 4 -, and NH 4 + .When the total NH 3 concentration is changed from 0.54 µg m -3 to 37.94 µg m -3 , the GR is also changed from -2.17 to 116.29; during this process, the response of iPM 2.5 to total NH 3 is changed from sensitive to insensitive.When GR ≤ 0, increasing total NH 3 concentration can increase the concentration of particle-phase SO 4 2-and NH 4 + while at the same time decrease the concentration of HSO 4 -.This is because that when NH 3 gas is insufficient for fully neutralizing total H 2 SO 4 , both (NH 4 ) 2 SO 4 and NH 4 HSO 4 exist in the system and adding more NH 3 can react with the available H 2 SO 4 and at the same time convert NH 4 HSO 4 to (NH 4 ) 2 SO 4 .When GR ˃ 0, NH 3 is adequate to fully neutralize total H 2 SO 4 , and there is excessive NH 3 in the system.In the whole  process, NO 3 -aerosol is close to 0, and HNO 3 concentration remains the constant.This is because that high T and low RH do not favor the formation of semi-volatile compound NH 4 NO 3 ; even when there is excessive NH 3 existing in the system, HNO 3 will not react with NH 3 to be converted to NO 3 -.Thus, in the end, increasing NH 3 will not change the concentration of any component in the system, and the response of iPM 2.5 to total NH 3 is insensitive.
Under minimum T + maximum RH (271.95K + 95%), the change of iPM 2.5 concentration is caused by the changes of SO 4 2-, HSO 4 -, NH 4 + and NO 3 -.When GR ≤ 1.63, increasing total NH 3 concentration can increase the concentration of SO 4 2-, NH 4 + and NO 3 -while at the same time decrease the concentration of HSO 4 -.This is owing to the fact that low T and high RH can favor the formation of NH 4 NO 3 ; when GR ≤ 0, adding more NH 3 can react with H 2 SO 4 and at the same time convert NH 4 HSO 4 to (NH 4 ) 2 SO 4 .When 0 < GR ≤ 1.63, the excessive NH 3 can react with HNO 3 to form NH 4 NO 3 until all the HNO 3 is depleted.After GR ˃ 1.63, adding more NH 3 will not change the concentration of iPM 2.5 significantly; this is due to lack of available acidic gases, which have been fully neutralized by NH 3 gas.
Under median T + median RH (292.05K + 77%), changes in the iPM 2.5 chemical composition are similar to those under the condition of minimum T + maximum RH (271.95K + 95%); the difference is the peak iPM 2.5 concentration value and the GR at which the system reaches the peak iPM 2.5 concentration.Fig. 4 shows that the difference in peak iPM 2.5 concentration is caused by the NO 3 -concentration.
Lower T and higher RH can favor the formation of NH 4 NO 3 ; thus, the peak iPM 2.5 under median T + median RH is about 6.13 µg m -3 , which is less than the peak iPM 2.5 concentration (6.24 µg m -3 ) under minimum T + maximum RH (271.95K + 95%) and greater than the peak iPM 2.5 concentration (4.76 µg m -3 ) under maximum T + minimum RH.According to Table 2, at summertime, the ambient meteorology was characterized by high T + high RH; at wintertime, the ambient meteorology was characterized by low T + median RH.In order to capture the seasonal variation of iPM 2.5 , these two combinations of T and RH are added to the analysis.As shown in Fig. 4, under maximum T + maximum RH (305.95K + 95%), the change of iPM 2.5 concentration is similar to the condition of median T + median RH.The difference is the peak iPM 2.5 concentration; the peak iPM 2.5 concentration under maximum T + maximum RH is close to 6.00 µg m -3 , which is lower than the peak iPM 2.5 concentration, 6.13 µg m -3 , under median T + median RH.This is owing to the fact that higher T does not favor the formation of NH 4 NO 3 and this case scenario can represent ambient condition in the summertime of this region.Under minimum T + median RH (271.95K + 77%), the change of iPM 2.5 concentration is caused by the change of SO 4 2-, NH 4 + , NO 3 -, and HSO 4 -, and this trend is similar to the condition of minimum T + maximum RH; the difference is the threshold value of total NH 3 concentration that indicates the transition from sensitive to the insensitive region.
Upon the above analysis, the threshold values of total NH 3 concentration at which transition from sensitive to the insensitive happens are identified for all the case scenarios (Table 6).Greater than threshold values, the iPM 2.5 becomes insensitive to the change of total NH 3 concentration.
Table 6 shows that under condition of median T + median RH, the threshold values of total NH 3 concentration and GR are the largest.Under condition of minimum T + maximum RH, the threshold values of total NH 3 concentration and GR are the smallest.According to Table 3, the ambient condition is dominated by NH 3 -rich condition; thus, under most of the cases, the response of iPM 2.5 to total NH 3 is insensitive.Thus, reducing total NH 3 concentration will not lead to significant decrease of iPM 2.5 concentration.

Response of Secondary iPM 2.5 to Total H 2 SO 4
In this analysis, median total NH 3 concentration and median total HNO 3 concentration were used under five T and RH case scenarios (Table 1) to simulate the iPM 2.5 formation with total H 2 SO 4 concentration changing from 0.59 to 14.60 µg m -3 .The results are shown in Fig. 5.
Fig. 5 shows that the response of iPM 2.5 to total H 2 SO 4 is linear.Increasing total H 2 SO 4 concentration can lead to the increase of iPM 2.5 concentration.Under maximum T + minimum RH (305.95K + 39%), the change of iPM 2.5 concentration is caused by the change of SO 4 2-, HSO 4 -, and NH 4 + .As the formation of NH 4 NO 3 is not favored under this condition, the concentration of NO 3 -is close to 0, and HNO 3 gas concentration remains the constant.While under the conditions of minimum T + maximum RH (271.95K + 95%), median T + median RH (292.05K + 77%), maximum T + maximum RH (305.95K + 95%), and minimum T + median RH (271.95K + 77%), the change of iPM 2.5 concentration is caused by the change of SO 4 2-, HSO 4 -, NH 4 + and NO 3 -.With the increase of total H 2 SO 4 , NH 3 gas concentration is decreased due to the reaction with H 2 SO 4 , and during this process, H 2 SO 4 also competes with HNO 3 to react with NH 3 to form (NH 4 ) 2 SO 4 and NH 4 HSO 4 .When total H 2 SO 4 concentration is greater than 9.0 µg m -3 , HSO 4 -begins to increase; this is because that if total NH 3 is not adequate to fully neutralize H 2 SO 4 , adding more H 2 SO 4 can convert (NH 4 ) 2 SO 4 to NH 4 HSO 4 .
In summary, the response of secondary iPM 2.5 to total H 2 SO 4 concentration is linear.Adding more H 2 SO 4 can increase iPM 2.5 concentration significantly; this is because of the excessive NH 3 existing in the system.When all the NH 3 gas is neutralized by H 2 SO 4 , SO 4 2-is converted to HSO 4 -; thus concentration of iPM 2.5 is increased.Response of secondary iPM 2.5 to total H 2 SO 4 is not determined by NH 3 -rich or NH 3 -poor conditions in this research.This is because that in NH 3 -NH 4 + -SO 4 2--HNO 3 -NO 3 -system, NH 3 gas preferentially reacts with H 2 SO 4 , and if total NH 3 is in excess of fully neutralizing H 2 SO 4 , excessive NH 3 can react with HNO 3 .In this research, when there is excessive NH 3 existing in the system, adding more H 2 SO 4 can directly increase the concentration of (NH 4 ) 2 SO 4 .After all the NH 3 is neutralized, (NH 4 ) 2 SO 4 is then converted to NH 4 HSO 4 .In total, the iPM 2.5 concentration keeps increasing but with lower increasing rate (there are two stages in the Fig. 5, higher increase rate in the first stage and lower increase rate in the second stage).

Response of Secondary iPM 2.5 to Total HNO 3
In this analysis, median total NH 3 concentration and median total H 2 SO 4 concentration were used under the five ) is less than 0.2, the response of iPM 2.5 to total NH 3 is defined as insensitive.Threshold GR value is calculated based on median total H 2 SO 4 concentration and median total HNO 3 concentration, 3.53 and 1.17 µg m -3 , respectively.T and RH case scenarios (Table 1) to simulate the iPM 2.5 formation with total HNO 3 concentration changing from 0.24 to 4.24 µg m -3 .The results are shown in Fig. 6.
Under maximum T + minimum RH (305.95K + 39%), iPM 2.5 is insensitive to the change of total HNO 3 .This is owing to the fact hat high T and low RH do not favor the formation of NH 4 NO 3 ; even if there is excessive NH 3 gas in the system, it will not react with the added HNO 3 to form NH 4 NO 3 .Thus, all the other chemical components remain the constant and NO 3 -concentration is close to 0. Under the conditions of minimum T + maximum RH (271.95K + 95%), median T + median RH (292.05K + 77%), maximum T + maximum RH (305.95K + 95%), and minimum T + median RH (271.95K + 77%), the changes of iPM 2.5 concentration is caused by the change of NH 4 + and NO 3 -.This is because that under these four conditions, the formation of NH 4 NO 3 is not limited by the ambient meteorology; thus, adding more HNO 3 can react with the available NH 3 to form NH 4 NO 3 , which leads to the increase of NO 3 -and NH 4 + simultaneously.The SO 4 2concentration remains the constant; this is due to the full neutralization of H 2 SO 4 by NH 3 .In summary, the response of secondary iPM 2.5 to total HNO 3 is linear; adding more HNO 3 to the system can increase the iPM 2.5 concentration linearly due to the formation of NH 4 NO 3 .Under different ambient conditions, the increase rate of iPM 2.5 is different; lower T and higher RH favor the formation of NH 4 NO 3 and lead to higher iPM 2.5 concentration.

Possible Mitigation Strategies for Ambient iPM 2.5 Reduction
Analysis of the iPM 2.5 responses to the precursor gases may lead to the development of the iPM 2.5 mitigation strategy.Under different ambient conditions, the responses of iPM 2.5 to total NH 3 , total HNO 3 , and total H 2 SO 4 are different.In general, iPM 2.5 is more sensitive to the change of total H 2 SO 4 concentration.The change of total NH 3 and total HNO 3 may also lead to the change of iPM 2.5 but may be limited by the neutralization degree of NH 3 and ambient conditions.In NH 3 -rich condition, the change of total NH 3 concentration has the least impact on the iPM 2.5 concentration.Reducing total H 2 SO 4 is more effective to decrease iPM 2.5 as compared to the reduction of total NH 3 and total HNO 3 .
It needs to be noted that the two acidic gases used in ISORROPIA II, H 2 SO 4 and HNO 3 , are not directly emitted, but are transformed from the oxidation of nitrogen oxides (NO x ) and sulfuric dioxide (SO 2 ) in the atmosphere.Reducing the two acidic gases would require the reduction of their associated primary pollutants, NO x and SO 2 .Thus, in development of the total H 2 SO 4 and HNO 4 concentration reduction strategies, strategies for SO 2 and nitrogen oxide (NO x ) emissions reduction should be taken.

CONCLUSIONS
In this research, the responses of iPM 2.5 to changes in the total NH 3 , the total HNO 3 , and the total H 2 SO 4 were simulated by ISORROPIA II based upon three-year measurements of the chemical components in iPM 2.5 and gaseous pollutants as well as meteorological conditions.It was found that iPM 2.5 responds to precursor gases differently under different T and RH conditions.In general, the response of iPM 2.5 to the total NH 3 is nonlinear, whereas its response to the total H 2 SO 4 and the total HNO 3 is linear.In NH 3 -rich regions, iPM 2.5 is insensitive to the total NH 3 but highly sensitive to the total H 2 SO 4 and/or the total HNO 3 .This research determined that the threshold values for the total NH 3 concentrations, below which the iPM 2.5 was insensitive to changes in the total NH 3 , varied under different ambient conditions.Although dry and wet deposition and the nonlinear conversion of NO x and SO 2 to HNO 3 and H 2 SO 4 were not simulated with the thermodynamic model, inferences can still be drawn from our results about the dynamic changes in iPM 2.5 in response to changes in precursor gases, offering insight into controlling and regulating iPM 2.5 in NH 3 -rich regions.These findings also illuminate the impact of AFO NH 3 emissions on the formation of secondary iPM 2.5 , which may help to develop strategies for reducing ambient PM 2.5 .Future studies may examine the quantitative contribution of AFO NH 3 emissions to forming iPM 2.5 using the Chemical Transport Model, thus further exploring the dynamic contribution of these emissions to the gas-particle partitioning of NH 3 -NH 4 + .

Fig. 1 .
Fig. 1.Spatial distribution of poultry and swine farms across the entire NC (the black star: iPM 2.5 and NH 3 monitoring site; the red triangle: swine farms; the green solid circle: poultry farms).

Fig. 3 .
Fig. 3. PM 2.5 chemical speciation mass fractions in Lenoir County in four seasons.

Fig. 4 .
Fig. 4. Responses of different iPM 2.5 chemical components to total NH 3 concentration under five T and RH conditions.

Fig. 5 .
Fig. 5. Response of different iPM 2.5 chemical components to total H 2 SO 4 concentration under five T and RH conditions.

Fig. 6 .
Fig. 6.Response of different iPM 2.5 chemical components to total HNO 3 concentration under five T and RH conditions.

Table 1 .
Inputs of the ISORROPIA II simulation. in this research.This can be justified by the difference of AFO distribution and density in these two counties shown in Fig.1.The AFO density is much higher in Sampson County than in Lenoir County.In addition, minimum, median and maximum total H 2 SO 4 concentrations of this site (0.59, 3.53 and 14.60) are comparable with those (0.58, 3.43 and 14.30) in Sampson County reported by measured

Table 2 .
The statistics of concentrations of different gases and PM 2.5 chemical components by season.

Table 3 .
Numbers in different GR ranges in four seasons of Lenoir County.calculated based on the median total HNO 3 , median total H 2 SO 4 and median total NH 3 concentrations.N is the total data point number in each season.

Table 4 .
Comparison of FRM vs. MetOne SASS measured PM 2.5 mass concentration.
Comparison is performed using paired t-test, 0.05 significance level.

Table 5 .
Comparisons of different PM 2.5 chemical component concentrations by season.

Table 6 .
Threshold values of insensitive response of iPM 2.5 to total NH 3