All Supporting Information may be found in the online version of this article.
Corresponding author: S. Samy and M. D. Hays, National Risk Management Research Laboratory (NRMRL), U.S. Environmental Protection Agency, Research Triangle Park, NC 27711, USA. (firstname.lastname@example.org; email@example.com)
 The impacts of meteorology and air quality on the concentrations and relative distributions of free and combined amino acids (FAA; CAA) are evaluated during a month-long sampling campaign at a semiurban site in the southeastern U.S. The average FAA concentration in fine aerosols (PM2.5) was 11 ± 6 ng m–3, while CAA was found to be several times higher at 46 ± 21 ng m–3. Glycine and alanine were the most abundant amino acids, accounting for 48% of FAA and 58% of the CAA, while distinct differences were observed in compound distributions; glutamic acid, aspartic acid, serine, and threonine accounted for a further 29% of FAA and 30% of the total CAA. An intense rainfall event during the campaign demonstrated the significant impact of meteorological and air quality conditions on FAA-CAA concentrations and distributions. Correlative trends with atmospheric oxidant (ozone) and inorganic nitrogen levels suggest an important role for atmospheric processing. The liquid chromatography-mass spectrometry (quadrupole time-of-flight) technique used in this study allowed for detection of coextracted water-soluble organic compounds and characterization of a larger fraction of the organic nitrogen mass. N-heterocyclic compounds were detected in samples from this campaign, indicating a likely biomass burning source contribution for organic nitrogen.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Studies show that a significant fraction (20–80%) of the total nitrogen (N) in atmospheric particulate matter (PM) is bound in a complex organic mixture [Cornell et al., 2003; Ge et al., 2010; Mace et al., 2003; Zhang et al., 2002]. For example, at an experimental forest site in the southeastern U.S., organic nitrogen (ON) accounted for approximately one third of the total N load in atmospheric PM2.5 [Lin et al., 2010]. The nutrient enrichment resulting from dry and wet deposition of bioavailable ON is a topic of concern for ecologists, environmental regulators, and policy strategists [Castro et al., 2001; Dennis et al., 2007; Galloway et al., 2008; McFarland et al., 2010]. The current scarcity of ON data in the mass balance modeling approaches that determine critical N loads for a given ecosystem is a significant source of uncertainty (for further discussion and references see [Cornell, 2011]). Therefore, more accurate information about atmospheric ON and its influence on ecosystem health and productivity is required to support science-based policy decisions [Cape et al., 2011; Neff et al., 2002].
 Past work emphasizes the need to further understand regional N cycles and related biogeochemistry through quantitative speciation of ON [Burns, 2003; Cornell et al., 2003; Neff et al., 2002; Shi et al., 2010]. Many identified compounds are bioavailable and can be taken up by plant and soil communities. Advanced speciation and analysis techniques have begun to elucidate contributing sources, the impacts of atmospheric processing, and sinks of ON [Ma and Hays, 2008; Ozel et al., 2011]. A good example of this progress is evident in newly developed methods that take advantage of high-resolution quadrupole time-of-flight (Q-TOF) mass spectrometry (MS) to improve accuracy, precision, and unknown compound identification [Samy et al., 2011]. However, much of the analytical methodology is novel or evolving so there is limited ON information for spatially diverse sampling sites (e.g., marine versus continental) or at high temporal resolution. In turn, this has limited our determination of ON source-sink relationships, allowing only minimal interpretation of speciated data in the context of regional N cycling [Cornell, 2011; Ge et al., 2010]. In the study presented here, fine aerosol (PM2.5) sampling was conducted at a site receiving both marine and continentally influenced air masses over a campaign period long enough to include a shift in meteorology.
 In addition to the above mentioned ecological concerns, recent studies suggest that atmospheric transformations involving N-containing organic compounds can influence aerosol optical properties and contribute to light-absorbing “brown carbon” formation [Andreae and Gelencser, 2006; DeHaan et al., 2011; A. Laskin et al., 2009; J. Laskin et al., 2010]. These processes affect the Earth's radiative balance. For example, brown carbon can change the aerosol radiative forcing efficiencies that describe the net heating/cooling of the atmosphere and influence climate model predictions [Chakrabarty et al., 2010; Forster et al., 2007; Moosmuller et al., 2009]. Hence, a better understanding of the chemical composition of ON at a molecular level may also contribute to an improved framework for estimating the direct and indirect aerosol effects on climate [Lohmann and Feichter, 2005; Ramanathan and Carmichael, 2008].
 Amino acids (AA) are an important compound group in the atmospheric ON fraction [Ge et al., 2010; Milne and Zika, 1993]. AA are generally classified as water-soluble organic compounds (WSOC) and contribute to secondary organic aerosol formation [Chan et al., 2005; de Haan et al., 2009]. WSOC has notable consequences for atmospheric chemistry and cloud formation as discussed previously [Collett et al., 1999; Samy et al., 2010; Saxena and Hildemann, 1996]. However, relatively brief focus on this research topic has produced limited information about the sources of key compound classes such as AA [Dittmar et al., 2001; Mace et al., 2003; Mopper and Zika, 1987; Zhang et al., 2002]. Although biomass burning was suggested as a significant source [A. Laskin et al., 2009; Mace et al., 2003], the marine environment also hosts potential sources of atmospheric AA (e.g., bursting sea bubbles, suspension of algae) [Kuznetsova et al., 2005; Mopper and Zika, 1987; Wedyan and Preston, 2008]. Other likely sources include the suspension of bacteria, fungi, pollen, and spores [Dittmar et al., 2001; Sattler et al., 2001; Scheller, 2001; Tong and Lighthart, 2000; Yu et al., 2002].
 Detection of native (chemically underivatized) AA in atmospheric aerosols using liquid chromatography (LC) interfaced with an MS (Q-TOF) was recently reported [Samy et al., 2011]. In this work, quantitation of dissolved free AA (FAA) (amino acids present in a dissolvable state) and combined AA (CAA) (amino acids present in peptides, proteins, and/or humic complexes) is achieved by LC-MS (Q-TOF) with positive electrospray ionization (+ESI). The accurate mass measurements (<2 ppm) of the MS (Q-TOF) not only allow selective detection of AA, but also provide the potential added value of selective and sensitive determination of coextracted polar organic compounds (POC) in aqueous samples [Hollender et al., 2010; Zwiener and Frimmel, 2004].
 For the present study, we provide an expanded southeastern U.S. ambient data set for FAA and CAA in PM2.5 quantified by LC-MS (Q-TOF). In addition, we demonstrate the potential for further chemical speciation of important ON compounds (N-heterocycles) with known biomass burning sources, using advanced LC-MS/MS techniques. To our knowledge, the only existing data set for atmospheric FAA and CAA in the region is from a 2010 study at the Duke Forest monitoring station outlined in Samy et al. , which is a more rural forested site approximately 20 miles NW of the semiurban sampling site in this study. Previously posed questions in relation to FAA and CAA are readdressed and evaluated. For example, we ask if trends in relative ratios of CAA/FAA can be linked to photochemical processing [Milne and Zika, 1993]. Further analysis of sampling conditions and air mass history provides important insight for ON processes. Simultaneous sampling of inorganic N (IN) was conducted during the study and is discussed in relation to ON results.
2.1 Sample Collection
 Sampling was performed at the U.S. Environmental Protection Agency (U.S. EPA) ambient air innovation research site on the Research Triangle Park (RTP), NC campus (35.89°N, 78.87°W) during September–October 2010. PM2.5 was collected on 90 mm Zefluor filters (1 µm pore size, Port Washington, NY) between 9 September and 13 October 2010 (N = 30) at the U.S. EPA campus in RTP. Sampling duration ranged from 24–48 h (105–270 m3), with an average sampling volume of 145 m3. Additional 48 h samples were collected at the same site on 1 December (220 m3) and 8 December (217 m3) 2010, which are discussed separately. Validated and certified ozone data from the Durham Armory (35.99°N, 78.89°W) was provided by the NC Division of Air Quality.
 Sampling of gas-phase IN compounds (HNO3, NH3), particle-phase IN (NO3–, NH4+), and SO2/SO42– was conducted using duplicate MARGA (Monitor for AeRosols and GAses, Model ADI 2080 1S, Metrohm Applikon B.V., Schiedam, The Netherlands) instruments at the same location between 8 September and 8 October 2010. During the sampling time, the performance of the MARGA was assessed under the Environmental Technology Verification program. Further details on the IN results and analysis during the Environmental Technology Verification assessment are provided in U.S. EPA .
 The MARGA is an on-line ion-chromatography based analyzer that semicontinuously measures gases and soluble ions in aerosols at an hourly temporal resolution [ten Brink et al., 2007]. Air is drawn through a custom inertial separator type inlet with a nominal particle aerodynamic size cutoff of approximately 26 µm and then through 4 m of 1.25 cm outer diameter polyethylene tubing to the analyzer. At the analyzer, sample air first flows into the wet rotating denuder (WRD), which has borosilicate glass walls, that are coated with double deionized water (DDI) causing the gases to diffuse into the liquid film [Keuken et al., 1998; Wyers et al., 1993]. Particles pass through the WRD and are collected directly downstream in the steam jet aerosol collector (SJAC) [Khlystov et al., 1995; Slanina et al., 2001]. Within the SJAC, a supersaturated environment is created in which particles grow by deliquescence, and are subsequently collected by inertial separation. Air is drawn through the WRD and SJAC at approximately 16.7 L min–1 using a vacuum pump (KNF Model N840FT.18, KNF Neuberger, Inc., Trenton, NJ) and mass flow controller (Brooks Smart Mass Flow Controller, Brooks Instrument, Hatfield, PA).
2.2 Sample Preservation and Preparation
 Sample filters were stored in precombusted (550°C) aluminum foil envelopes inside separately sealed Petri dishes, placed inside a secondary bag, and frozen (–40°C) prior to extraction. In the laboratory, samples were handled using cleaned (1:1 IPA:H2O followed by MeOH) stainless steel forceps in a filtered air laminar flow clean hood. Sample filters were folded into precombusted (550°C) centrifuge tubes, spiked with a midlevel internal standard mix and extracted with high-performance liquid chromatography (HPLC) grade H2O (50–60 mL). Contents were sonicated (two times for 30 min each), and transferred to an acid-washed evaporation tube (RapidVap; Labconco, Kansas City, MO) using a precombusted Pasteur pipette. More details on the sample preparation are provided in Samy et al. . Extracts were concentrated to approximately 2–3 mL under vacuum, mild heating, and rotary motion (RapidVap method parameters: 90 mb, 60°C, 28% vortex speed). Extracts were transferred into 7 mL acid-washed amber vials, followed by a 3–4 mL wash of the evaporation tube using 2% formic acid (in HPLC water, pH = 2.1), and concentrated under a gentle stream of ultrapure nitrogen with mild heating (60°C) to approximately 500 μL. Samples were then filtered (0.2 µm Polytetrafluoroethylene) and transferred using an additional 2% formic wash (Vf = 1–2 mL) into a 2 mL silanized amber autosampler vial (National Scientific). Sample extracts were capped and immediately analyzed for FAA by LC-MS (Q-TOF) or frozen (−40°C) for subsequent analysis (within 24 h).
 Following FAA analysis, a subset of extracts were dried under ultrapure nitrogen, hydrolyzed, and resuspended using 2% formic acid for total AA (TAA) analysis. As an antioxidant, 500 µg of ascorbic acid (25 μL of a 20 µg/µL stock solution) was added to each extract prior to drying. Vapor phase hydrolysis (30 min with 6 M HCL, 10% trifluoroacetic acid (TFA), 0.1% phenol) in this study was performed with a Discover protein hydrolysis microwave system (CEM Corp., Charlotte, NC), which has a single mode cavity (uniform electromagnetic field intensity) and allows continuous control of input power for precise fiber-optic temperature control of the hydrolysis chamber. The detailed hydrolysis parameters and method validation are provided elsewhere [Samy et al., 2011].
2.3 Sample Analysis
 The advantages of native (underivatized) detection of AA in aqueous extracts was described earlier [Samy et al., 2011]. Briefly, an Agilent high-performance liquid chromatograph (HPLC 1200 Series) interfaced to a Q-TOF MS (Model 6520) was used for the sample analyses (Agilent Technologies, Palo Alto, CA). Target AA analytes were separated using ion-pairing chromatography with a Zorbax Eclipse XDB-C18 column (4.6 × 50 mm, 1.8 µm). The column temperature was maintained at 40°C, and a mobile phase ramp of H2O (0.8 mM perfluoroheptanoic acid/0.05% TFA/0.05% formic acid) and MeOH (0.03% formic acid) was performed at a flow rate of 0.40 mL min–1, increasing MeOH from 0% to 100% in 23 min with a 10 min postrun. Recent adoption of a narrow bore (2.1 mm) Zorbax Eclipse XDB-C18 column has reduced peak widths and improved resolution of isomers (leucine, isoleucine).
 A multimode ion source was operated in positive (+) ESI mode. The gas temperature (30 °C), vaporizer (250°C), drying gas (6 L/min), nebulizer (60 psi), capillary voltage (2400 V), fragmentor (100 V), and skimmer (65 V) parameters were optimized using individual AA compound injections. The Q-TOF MS was operated in scan mode with the spectral mass range set at 40–1000 m/z and an acquisition rate of 1.4 spectra s–1 (acquisition time = 714 ms/spectrum). Internal mass tuning and calibration for the Q-TOF was performed daily with a directly infused 10-point mass reference solution prior to initiation of all sample analyses. Evaluation of a dedicated dual ESI source indicated improved S/N for later eluting compounds (ornithine, histidine) and was therefore used for posthydrolysis CAA analysis. Because the multimode source was operated in +ESI, all parameters, except drying gas (12 L/min), remained the same. The discussed MS/MS data were produced with 40 V collision energy.
 The liquid samples from the MARGA WRD and SJAC are collected separately in an automated syringe pump module, which contains a third set of syringes for an internal standard. The syringe module is configured such that during each hour the WRD and SJAC samples from the previous hour are being analyzed while new samples are collected. Each sample is mixed with an internal standard solution of Li+ and Br–. The samples are analyzed by anion and cation ion chromatography (IC; Metrohm USA, Inc., Riverview, FL). The anion IC employs a Metrosep A Supp-10 75 mm column (Metrohm USA, Inc.) and 250 μL injection loop. The cation IC employs a Metrosep C2 150 mm column (Metrohm USA, Inc.) and 500 μL injection loop.
 A standard reference material containing 17 target AA (standard reference material 2389) was purchased from the National Institute of Standards & Technology. An additional six AA were purchased from Sigma-Aldrich (St. Louis, MO, USA) and added to the standard mixture (see Table S1 in the auxiliary material). This mixture was used to produce multilevel calibration standards. Deuterated internal standards were purchased from Cambridge Isotope Laboratories (Andover, MA). N-heterocyclic compounds were purchased from Sigma-Aldrich (St. Louis, MO, USA). HPLC grade water and methanol (MeOH) were used for instrument operation. Pure (99%) perfluoroheptanoic acid, TFA (≥98%), and formic acid (98%, ACS grade) were used for the ion-pairing chromatography. Nitric acid (Trace Metal Grade) and isopropanol (LC/MS grade) were used for glassware cleaning procedures discussed below.
 All MARGA solutions were prepared in 18.2 MΩ·cm DDI water. This DDI water, with 10 ppm of H2O2 added to control bacterial growth, was used as absorbance solution for the WRD and SJAC. Anion eluent (7.0 mmol/8.0 mmol Na2CO3/ NaHCO3) was prepared from pure material (>99.5%, ACROS Organics, Fisher Scientific, Pittsburgh, PA, USA). 2 M HNO3 (Sigma-Aldrich) was used to prepare the cation eluent (3.2 mmol). Pure LiBr (>99%, ACROS Organics) was used to prepare the internal standard (320 µg L–1 Li– and 3680 µg L–1 Br+). Pure NH4NO3 (99.9%, certified ACS, Fisher Scientific) and (NH4)2SO4 (99.0% certified ACS, Fisher Scientific) were used to prepare additional liquid standards of NH4+, NO3–, and SO42–.
2.5 Data Processing and Quality Assurance
 Target compound (23 total listed in Table S1) calibration was conducted by on-column injection of 6–8 multilevel standards. Each target compound was quantified via the internal standard method using a deuterated analogue or similar. All mass measurements of individual compound standards were performed on a Sartorius microbalance (Edgewood, NY) calibrated with National Institute of Standards & Technology traceable weights.
 Linearity (R2 > 0.99) was observed over the calibration range (0.007–5 ng/μL), and replicate (N = 9) injections of a midlevel standard confirmed previously reported method detection limit (MDL) values (1.7–27.4 pg/μL) and allowed determination of an MDL for ornithine (40 pg/μL), which was not previously reported [Samy et al., 2011]. Response factors were determined using the [M + H]+ ion, while additional ions were monitored for qualitative confirmation. Midlevel calibration checks were run at least once every eight samples, with typical accuracy within ±10%. The reporting thresholds or limits of quantitation (LOQ) were set at three times the MDL. Compounds reported as “trace” have been detected above the MDL but below the LOQ. All laboratory media blanks were determined to be below the MDL, while field blanks did display higher background levels with a typical concentration of less than 15% of the detected compounds. All samples underwent blank subtraction. Following hydrolysis, significant degradation of internal standards (<10% relative response) was observed in a subset of the samples. These samples were not included in the quantitation of CAA (CAA = TAA – FAA). Observed degradation may be due to insufficient antioxidant addition (ascorbic acid) and/or catalyzed reactions in the complex aerosol matrix (e.g., with available transition metals). Experiments to evaluate potential sampling breakthrough using in series filters and identical sampling conditions (N = 4) indicated an average 12 (5–15)% breakthrough of FAA. Values reported here were not corrected for breakthrough.
 Additional information on the MARGA Quality Assurance procedures, raw data adjustments, and processing used to calculate the IN concentrations are provided in the auxiliary material.
3 Results and Discussion
 The average late summer-early fall (9 September 2010 to 13 October 2010) FAA concentration was 11 ± 6 ng m–3 (140 pmol N m–3; N = 30) at Research Triangle Park, which is approximately half of the average FAA recently reported from samples collected at the Duke Forest experimental station (16 July 2010 to 16 August 2010) [Samy et al., 2011]. The percent distribution of FAA from RTP (Figure 1) is similar to the earlier Duke Forest results, with fewer compounds (13 > LOQ) detected above the reporting threshold (16 > LOQ at Duke Forest). The seven most abundant FAA (glycine > alanine = aspartic acid = arginine > glutamic acid > serine > glutamine) account for 90% of the total average. Individual compound results are provided in Table S2. Although the RTP site is better described as an urban-suburban site within the Raleigh-Durham (RDU) urban zone and Duke Forest is a more rural forested site approximately 20 miles NW [Geron, 2009], other factors (discussed below) may have contributed to the observed differences in concentration between the sites. A subset of RTP samples (N = 17) were analyzed for CAA following microwave-assisted vapor-phase hydrolysis. The mean CAA values were determined to be several times higher than FAA (46 ± 21 ng m–3; 470 pmol N m–3). Figure 1 displays the percent distribution of the most abundant compounds at RTP. In contrast to FAA results that revealed glycine as the most abundant compound, the most abundant CAA compound is alanine (40%). The seven most abundant CAA (alanine > glycine > glutamic acid > threonine > aspartic acid > tyrosine > serine) represent 92% of the total average. Relative to FAA results, five additional compounds (methionine, histidine, phenylalanine, leucine, isoleucine) were quantified above the reporting thresholds, and two more compounds (lysine, ornithine) were determined to be above the MDL (Table S2). Note that the FAA percent distribution for the identical subset of samples analyzed for CAA (N = 17) is provided in Figure S2 and is similar to the FAA distribution displayed in Figure 1.
 Figure 2 displays the daily CAA/FAA ratios along with the total amino acid (TAA = FAA + CAA) results for the individual samples. While the TAA concentrations (mean = 54 ± 22 ng m–3) at RTP are generally lower than those previously determined at Duke Forest (70 ± 35 ng m–3), the mean CAA/FAA ratios (RTP = 5.5) are higher (DF = 3.1). This trend suggests less aging of the collected PM2.5 at RTP, and may indicate a more local source for the RTP TAA [Milne and Zika, 1993]. Although a difference in the sampling site characteristics (i.e., proximity to potential sources) may partially explain these results, a closer look at the meteorological conditions and atmospheric oxidant (ozone) trends during the RTP campaign can provide valuable insight.
3.1 Meteorological Shift
 Midway through the sampling campaign (25 September 2010), a low-pressure frontal system in the upper plains (www.mmm.ucar.edu/imagearchive) shifted to the southwest (SW) of the RDU area resulting in high levels of moisture advection and convective instability with significant prolonged precipitation (5.7 cm reported at RDU airport on 26 September 2010). The event lasted several days with 16.6 cm of precipitation recorded at the RDU airport between 26 September 2010 and 1 October 2010. Mean daily temperatures at RDU dropped significantly following the passage of the front (28.3°C on 25 September 2010 compared to 16.1°C on 1 October 2010) with generally cooler conditions prevailing for the second half of the campaign (first half mean temperature = 25.9 ± 2.1°C, second half mean temperature 19.3 ± 3.3°C). NOAA Hybrid Single-Particle Lagrangian Integrated Trajectories (HYSPLIT) were calculated for preprecipitation and postprecipitation dates to determine the origin of air masses sampled at RTP [Draxler and Rolph, 2012]. Figure 3A provides the Hybrid Single-Particle Lagrangian Integrated Trajectories back trajectories for a preprecipitation sampling day (23 September 2010) and the change in flow coinciding (30 September 2010) with the frontal event (Figure 3B). The more continental (westerly) flow prior to the southeasterly postprecipitation shift indicates a distinct change in air mass history. This shift provides an opportunity to evaluate our results in relation to preprecipitation and postprecipitation conditions, and determine if significant changes in compound profiles and concentrations are observed.
 For this analysis, data were initially grouped into preprecipitation (9–25 September 2010) and postprecipitation (26 September to 13 October 2010) sets and significance evaluations (Student's t-test) were performed to determine changes in both compound distributions and mass concentrations. Several FAA displayed significant (p < 0.001) changes in percent distributions following precipitation (include: glycine, aspartic acid, glutamic acid, serine, glutamine, and threonine). Further evaluation of the time segment (5 days) just prior to precipitation when clear hot conditions prevailed, relative to an equivalent time frame following precipitation, revealed significant changes (p < 0.03) in both average mass concentrations and percent distributions (% of total FAA) for glycine, aspartic acid, and glutamine. Figure 4 displays the FAA distributions for the associated 5 day time frames. While glycine and aspartic acid significantly decrease following precipitation, other compounds (glutamine, serine, threonine) significantly increase. The two other major compounds, alanine and arginine, do not display significant changes in distribution. However, a decrease (p < 0.04) in alanine mass concentration is observed following the precipitation event. These results are in agreement with past studies where FAA concentrations in precipitation paralleled FAA in aerosols, and the scavenging ratios were estimated to be relatively high; even greater than ratios observed for NH4+ (>500) [Gorzelska and Galloway, 1990].
 The production of FAA in aged aerosols by direct photolysis (UV radiation), photocatalytic hydrolysis and/or enzyme-based hydrolysis of CAA, has been suggested in the past [Milne and Zika, 1993; McGregor and Anastasio, 2001]. More recent work has associated glycine and alanine (as FAA) with increased chemical aging and long-range transport (more distant sources) when evaluated relative to other FAA [Barbaro et al., 2011]. With a focus on the significant shift in meteorological conditions that occurred on 25 September 2010 and the high scavenging ratios reported for individual FAA species [Gorzelska and Galloway, 1990], the absence of scavenging from cloud droplets and precipitation combined with significant concentration changes in lower molecular weight FAA (glycine) suggests a more aged aerosol character (longer range transport) prior to the frontal event. Formation of aspartic acid via hydroxyl radical attack of asparagine during the predominantly clear (high actinic flux) preprecipitation days may also contribute to the significant change in concentrations. However, past work has determined that FAA in suspendable plant material (pollen) increases when exposed to atmospherically relevant levels of ozone [Mumford et al., 1972]. More recent work by Shiraiwa et al.  has indicated a significant increase in chemical transformation of amorphous protein by condensed phase ozone uptake, resulting in depletion of reactive amino acids in the combined protein form (CAA) [Shiraiwa et al., 2011]. This decomposition of the more complex forms of CAA (proteins and peptides) can result in higher concentrations of simpler reaction products detected as FAA [Milne and Zika, 1993]. Figure 5 displays the daily mass concentrations of glycine and aspartic acid along with the average ozone concentrations. Interestingly, the beginning of the peak FAA concentrations prior to precipitation (22 September 2010) corresponds to relatively high ozone and the lowest CAA/FAA ratio observed during the campaign (CAA/FAA = 2.6 on 22 September 2010, see Figure 2), followed by a relatively low ratio on 24 September 2010 (CAA/FAA = 3.4). These trends are consistent with the aerosol aging-photochemical transformation hypothesis of amino acids presented by Milne and Zika .
 Due to the potential complexity of the precursors and temporal resolution of the data, modest linear correlation is observed for FAA mass concentrations (glycine + aspartic acid) with ozone (R2 = 0.49), with a higher correlation when evaluated as percent of total FAA (R2 = 0.58). These ambient observations support the hypothesis that a combination of higher actinic flux and ozone levels promotes photochemically driven oxidative reactions that transform CAA (proteins and peptides) and potential precursor FAA (asparagine) into the dominant FAA determined in this study (glycine and aspartic acid). In contrast, the significant increases in concentrations of three FAAs (serine, glutamine, threonine) observed following the precipitation event (Figure 6) may be indicative of differing sources and photostability. An inverse linear correlation (R2 = 0.55) was observed between mean ozone and the sum of the three amino acids (as percent FAA), which suggests these three compounds have distinct sources and/or precursors that more significantly contribute to the total FAA following the precipitation event.
 To further investigate the linkage of FAA with secondary atmospheric processing in the southeastern U.S. [Lee et al., 2007], IN gas (NH3, HNO3), IN particle (NH4+, NO3–), and SO42– results collected at the site were examined for correlative trends [Finlayson-Pitts and Pitts, 2000]. Significant positive correlation (p < 0.004) was observed between total measured IN (IN = NH3 + NH4+ + HNO3 + NO3–) and FAA (R2 = 0.30), while NH4+ displayed the highest linearity with total FAA (R2 = 0.40). Concentration averages for IN and SO42– are provided in the auxiliary material. CAA did not display a significant correlation with individual or total IN species.
 A closer look at individual FAA compounds revealed the most significant (p < 0.0001) positive NH4+ correlations with glycine (R2 = 0.69), aspartic acid (R2 = 0.60), and alanine (R2 = 0.58) concentrations. In contrast, significant (p < 0.0001) inverse NH4+ correlations with glutamic acid, glutamine, threonine, and serine (R2 range = 0.51–0.64) was observed. In all cases, sulfate (SO42–) correlations were statistically equivalent (p < 0.0001) to the observed FAA-NH4+ trends and ranges mentioned above. For example, the highest SO42– correlations were also observed with glycine (R2 = 0.68), aspartic acid (R2 = 0.62), and alanine (R2 = 0.53) concentrations. Ammonium and sulfate are established secondary inorganic aerosol species in the eastern U.S. [Seinfeld and Pandis, 1998; Lee et al., 2007], and more recently have been linked with a secondary organic compound; oxalate [Wan and Yu, 2007]. This suggests that more locally sourced (primary) FAA with relatively shorter atmospheric residence times would display negative (inverse) correlations with the associated ionic species (NH4+, SO42–), while more secondary FAA (glycine, aspartic acid, alanine) with similar residence times would display positive correlations. Additional comparison of FAA with nitric acid and nitrate (HNO3 + NO3–) concentrations reveals similar trends with the percent sum of the inversely correlated compounds (glutamine + threonine + serine), resulting in the most significant inverse linear relationship (R2 = 0.70). The associated plot is provided in Figure S2. These results further support the hypothesis that the relative FAA levels may be linked to aerosol aging processes.
 In order to evaluate FAA and CAA relative distributions in low-temperature winter conditions, two additional samples were collected at the site on 1 December (mean temp = 2.7°C) and 8 December (mean temp = –1.7 °C) 2010. While a more diverse distribution of both FAA and CAA is observed in the winter samples suggesting a shift in potential sources (see Figure S3) and processing mechanisms, glycine is still the dominant compound accounting for approximately one third of the totals. Glycine plays a central role in the triple helix molecular structure of the most abundant fibrous protein in animals (collagen), and accounts for one third of the sequence [Lodish et al., 2000]. Other glycine-rich proteins such as elastin and certain keratins (e.g., silk fibroins found in insect pupae and spiders) occur naturally, which may partially explain the observed distributions. Relative to the fall samples, a similar concentration of FAA was detected (mean = 14 ng m–3). However, the warmer (above freezing) sampling day (1 December) displayed three times the value (21 ng m–3) relative to the colder below freezing conditions (7 ng m–3). Interestingly, CAA values (9 ± 1 ng m–3) for these samples are five times lower than the fall values (46 ± 21 ng m–3) with an average CAA/FAA ratio of 0.9 (Figure S3). The suppression of biological activity during the winter sampling due to freezing conditions and reduced solar insolation may explain the lower CAA results. Future seasonal sampling should provide more insight on these trends.
4 Further Speciation of ON
 The ability to determine elemental composition of abundant MS spectral peaks has been exploited in recent years for various purposes. For example, determination of unknown compound elemental composition using accurate-mass spectral data has recently been adopted in fog, precipitation, and aerosol characterization studies [Altieri et al., 2009; J. Laskin et al., 2010; Mazzoleni et al., 2010]. One advantage of accurate-mass LC-MS (Q-TOF) detection of native AA in the aqueous extracts of atmospheric aerosols is the potential to identify additional coextracted compounds, and establish a “next step” in method development for WSOC quantitation. While the supporting information provides an in depth discussion of the MS spectra results and the associated methodology for qualitative examination, we briefly focus on quantitative results for a candidate compound group (N-heterocycles) linked with biomass burning aerosols.
 An evaluation of our ambient sample spectra followed by injections of authentic standards, confirmed the presence of several N-heterocyclic (N-het) compounds. These compounds were previously quantified in biomass burning samples in our laboratory using a two-dimensional GC-MS method [Ma and Hays, 2008], and were recently reported as present in experimental fire samples [A. Laskin et al., 2009]. Table 1 includes a list of the compounds with chemical formula, exact mass, [M + H]+ adduct mass, and isotopic mass along with the theoretical percent abundance of the associated isotope. Figure 7A displays the extracted ion chromatograms (EIC) for the quantitation ions in a midlevel standard mixture along with compound structures. Reinjection of three aqueous extracts (sample collection dates: 15 September 2010, 16 September 2010, 4 October 2010) resulted in estimated concentrations of individual compounds ranging from 5 to 215 pg m–3 with a total average of 173 pg m–3 (see Table 1 for individual compound averages). MS/MS confirmation of the detected compounds with selected parent-daughter ion transitions can be viewed in Figure 7B. See auxiliary material for further discussion on N-het MS/MS spectra and additional EIC (Figure S4).
Table 1. N-Heterocyclic Compounds Evaluated in Southeastern PM2.5
Relative theoretical percent abundance mainly due to natural occurrence of 13C and 15 N.
Percent relative standard deviation (RSD) calculated using replicate injections (30 μL) of authentic standard (N = 8), concentration range 3–16 pg/μL.
Average relative percent abundance of isotope (N = 8) with comparison to theoretical value in parentheses.
Figure 7 provides chromatograms, see auxiliary material for full list of confirmed transitions.
Total compound average = 173 pg m–3; ND = not detected.
Percent RSD from replicate injection of ambient sample.
 The potential link between FAA concentrations and biomass burning aerosols has been suggested in a past study [Mace et al., 2003]. Further evaluation of our sample set revealed the most significant positive correlations (p < 0.002) between 3-hydroxy-pyridine (3-HP) and total FAA concentrations (R2 = 0.31). Similar correlations were observed with 3-HP and free glycine, alanine, and aspartic acid (R2 = 0.30–0.34). The lack of historical data and the limited observations here require further simultaneous quantitation of N-het and FAA to achieve a more robust statistical correlation with conditional sampling (e.g., during regional fire episodes) that may identify a primary biomass burning component at our sampling site. However, these results provide guidance for future work.
4.1 Further Work
 The ability to simultaneously quantify important compound groups, while moving forward with verification of unknown compounds will be a crucial link in improving our understanding of ON. This process may provide additional insight for other pending atmospheric science inquires. For example, light absorption in the UV-Vis wavelengths by FAA, CAA, N-het, and other ON compounds may explain the observed optical characteristics of aerosols in past studies [Andreae and Gelencser, 2006; J. Laskin et al., 2010; Milne and Zika, 1993]. The abundance of these compounds and the resulting light absorption may be further evaluated to enhance future modeling efforts.
 Recent work by McFarland et al.  has confirmed that FAA plays an important role in forest soil microbial metabolism, and consumption of the most abundant FAA quantified in our aerosol samples (glycine) was rapid across all six forest ecosystems studied [McFarland et al., 2010]. Future work may focus on deposition of characterized FAA profiles in N-limited ecosystems where impacts are more significant and uptake rates are increased [Averill and Finzi, 2011]. Using determined profiles of FAA from sampling efforts such as the one described in this study, a mimicking of deposition rates in key N-limited environments (e.g., alpine) in conjunction with monitoring efforts described in the above noted studies could strengthen the ecological impact link. This research will be of interest to climate scientists because of the implication for carbon sequestration. Past work has indicated FAA uptake by microorganisms in natural systems influences CO2 (flux) respiration [Crawford et al., 1974; del Giorgio et al., 2011]. Therefore, ON deposition results in changes of the estimated radiative forcing associated with the indirect biogeochemical feedback effects on climate [Mahowald, 2011].
 Water solubility of β-carboline photosensitizers (harmane and norharmane) detected in this study have been determined to increase with decreasing pH [Varela et al., 1995]. Quantification of these compounds and additional N-het may further elucidate ON transformation and sources. For example, the presence of hydroxy-pyridines may help to catalyze proton-dependent reactions due to the likely tautomerization of these compounds in the aqueous phase [Finlayson-Pitts and Pitts, 2000; Scanlan et al., 1983]. Future work can exploit the capabilities of the LC-MS (Q-TOF) method to add these potential source tracers by establishing a conditional sampling regime based on meteorological parameters and likely contributors such as fire events.
 A robust relationship between chemically speciated and bulk ON has not been established, due to the lack of existing data sets. The increasing focus on the atmospheric ON fraction will require more spatially diverse speciation data, along with collocated bulk ON determination to establish ecological impacts and science-based policies [Cape et al., 2011; Cornell, 2011]. In this study, we demonstrate that accurate mass spectra can be further evaluated for coextracted WSOC to characterize a larger fraction of the ON mass. While further spectral data exploration to achieve additional qualitative verification of unknown compounds is beyond the scope of this article, forthcoming publications will focus on the topic in more detail.
 Expanded ambient data sets for FAA and CAA in the southeastern US were presented and discussed in relation to sampling conditions. To better source atmospheric ON and understand chemical transformations, recent analytical advancements using an LC-MS (Q-TOF) method were adopted to produce and further interpret the results. A progressing frontal system with the associated change in air-mass and intense rainfall event during the campaign demonstrated the significant impact of meteorological and air quality conditions on FAA-CAA concentrations and distributions.
 The compound-specific correlative trends observed in relation to ozone and IN suggest atmospheric processing plays an important role in understanding ON impacts on biogeochemical cycling, which determines bioavailability and ecological productivity. The correlations observed with inorganic compounds (IN, SO42–) at Research Triangle Park suggest that sourcing and transformations of ON can be better understood through collocated sampling efforts. Future analysis to determine potential ON sources and transformations may be guided by these results.
 The authors would like to thank Chris Geron for use of sample collection equipment. We would also like to thank Ken Cowan, Tom Kelly, Rob Proost and Elizabeth Hanft for their assistance with the IN sampling and data analysis. Additional funding for this research was provided through a U.S. Department of Energy-U.S. Environmental Protection Agency interagency agreement administered by the Oak Ridge Institute for Science and Education. No official agency endorsement of statements should be inferred.