Fragmentation Analysis of Water-Soluble Atmospheric Organic Matter

Mar 21, 2012 - Isolated water-soluble atmospheric organic matter (AOM) analytes extracted from radiation fogwater samples were analyzed using collisio...
0 downloads 11 Views 1MB Size
Download PDF
Article pubs.acs.org/est

Fragmentation Analysis of Water-Soluble Atmospheric Organic Matter Using Ultrahigh-Resolution FT-ICR Mass Spectrometry Jeffrey P. LeClair,† Jeffrey L. Collett,‡ and Lynn R. Mazzoleni†,§,* †

Department of Chemistry, Michigan Technological University, 1400 Townsend Drive, Houghton, Michigan 49931, United States Department of Atmospheric Science, Colorado State University, 1371 Campus Delivery, Fort Collins, Colorado 80523, United States § Atmospheric Science Program, Michigan Technological University, 1400 Townsend Drive, Houghton, Michigan 49931, United States ‡

S Supporting Information *

ABSTRACT: Isolated water-soluble atmospheric organic matter (AOM) analytes extracted from radiation fogwater samples were analyzed using collision induced dissociation with ultrahigh-resolution Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS). Tandem mass analysis was performed on several mass ranges between 100 and 400 Da to characterize the functional groups of AOM species. Compounds containing nitrogen and/or sulfur were targeted because of the high number of oxygen atoms contained in their molecular formulas. Due to the large number of isobaric ions in the precursor isolation ranges, large numbers of product ions resulted from collision induced dissociation. Common neutral losses were assigned by matching the molecular formulas of the expected product ions with the detected product ions within the appropriate mass spectra. Since polar functional groups are expected to affect the hygroscopic properties of aerosols, the losses of H2O, CO2, CH3OH, HNO3, CH3NO3, SO3, SO4 and combinations of these were specifically targeted. Among the 421 compounds studied, the most frequently observed neutral losses were CO2 (54%), H2O (43%) and CH3OH (40%). HNO3 losses were observed for 63% of the studied nitrogen containing compounds and 33% of the studied compounds containing both nitrogen and sulfur. SO3 losses were observed for 85% of the studied sulfur containing compounds and 42% of studied compounds containing both nitrogen and sulfur. A number of molecular formulas matching those of monoterpene ozonolysis SOA were observed; they include organonitrates, organosulfates, and nitroxyorganosulfates. Overall, the results of fragmentation analysis of 400+ individual molecular precursors elucidate the complexity and multifunctional nature of the isolated water-soluble AOM.

■

reactions, such as: hydration, esterification,13 hemiacetal/ acetal,14 and aldol condensation.11 The aldol condensation mechanism is significant at higher pH.11 Thus, aqueous phase reactions can contribute to the production of SOA.7 The possibilities that exist for reaction products and mechanisms contribute to the complexity of and difficulty of the characterization of atmospheric organic matter (AOM). The compounds observed in polluted fogwater are from a variety of emission sources and the secondary reactions that may have occurred before and/or during the sampled fog event. In addition to the effects upon aerosol processes, fog events are important because of the effects they can have on human and environmental health.2,5,15−17 Preliminary studies have shown that AOM is quite complex.8,18,19 The use of ultrahigh-resolution mass spectrometry is

INTRODUCTION Radiation fog events are common during the winter in the California Central Valley.1−4 The events form during stagnant and humid conditions. Under a clear night sky, long wave radiation losses promote fast cooling of the moist surface air which promotes water vapor condensation onto pre-existing aerosol particles. A variety of emissions from agricultural, industrial, and residential activities in the valley accumulates and undergoes secondary chemical processes. The suspended fog droplets represent an aqueous phase reactor which allows secondary reactions between water-soluble gases, the watersoluble portion of scavenged aerosol particles, and atmospheric oxidants.5,6 In the aqueous phase, compounds undergo further oxidation and subsequent accretion, likely contributing to enhanced secondary organic aerosol (SOA) production.7 The oxidation products include organic acids and poorly characterized higher molecular weight accretion products, multifunctional compounds, organosulfates, and organonitrates.8−12 In the absence of photo-oxidation11 secondary aqueous products may form via non-oxidative accretion © 2012 American Chemical Society

Received: Revised: Accepted: Published: 4312

October 3, 2011 March 15, 2012 March 21, 2012 March 21, 2012 dx.doi.org/10.1021/es203509b | Environ. Sci. Technol. 2012, 46, 4312−4322

Environmental Science & Technology

Article

methanol (Fisher Optima grade) and then rinsed with 1% formic acid (Fluka LC/MS grade) in high purity water (UV treated and deionized). A sample volume of 100 mL of previously filtered fogwater (with a dissolved organic carbon concentration of 12.7 ppm) was pH adjusted with formic acid to pH 4.5 and then applied to the SPE cartridge at a flow rate of 1 mL/min. After the sample had passed through the cartridge, the cartridge was washed with 1% formic acid in high purity water to remove inorganic compounds. Two mL of alkaline (pH adjusted to 10.4 with ammonium hydroxide) high-purity water, methanol, and acetonitrile (10/45/45 vol/vol/vol) was used to elute the organic compounds from the SPE cartridge. A brown color band was observed moving down and off the SPE material into the sample vial. The extraction recoveries of fog AOM components have not been evaluated. Low molecular weight organic acids and sugars are expected to be lost during the wash step. Likewise, high molecular weight compounds tightly retained by the Strata-X stationary phase may not have been recovered during extraction. Instrumental Parameters. Samples were analyzed with a hybrid 7 T FT-ICR MS (LTQ FT Ultra, Thermo Scientific) equipped with an ESI source. Diluted samples were infused at 4 μL min−1 into the ESI interface for negative ion analysis. The ESI probe was placed in position “B” and the needle voltage was set between at −3.7 and −3.8 kV (blanks were −4.0 kV). Mass resolving power, m/Δm50%, in which Δm50% is peak full width at half-maximum peak height, was set at 200 000 (at m/z 400) for all spectra. Automatic gain control was used to consistently fill the linear ion trap with the same number of ions (n = 1 × 106) for each acquisition and to avoid space charge effects from overfilling the mass analyzer. Target precursor ions were isolated and fragmented with helium collision induced dissociation in the linear ion trap and then the ions were transferred to the FT-ICR MS. This tandem mass analysis was done on several selected scan ranges with different isolation widths between 270 and 360 Da. A full list of the mass ranges and instrumental parameters are given in Table S-1 of the Supporting Information, SI. Due to the sample complexity, the isolation of target nominal masses resulted in the fragmentation of several precursor ions simultaneously. Mass spectra with and without collision induced dissociation were collected for each mass range. The mass spectra were collected and stored as transients by use of Thermo Xcalibur software. Prior to mass analysis, the instrument was externally calibrated in the negative ion mode with a standard calibration solution from Thermo; the resulting mass accuracy was better than 2 ppm. Data Processing and Assignment of Molecular Compositions. 100+ transients were recorded in the time domain for each scan range and were coadded25,50 with Sierra Analytics Composer software. Chemical formulas were assigned to the masses of singly charged ions 100 < m/z < 400 with relative abundances (RA) ≥ 0.1% after internal recalibration. A list of the recalibration masses for each mass range is given in Table S-1 of the SI. The chemical formula calculator was set to allow up to 30 carbon (C), 60 hydrogen (H), 20 oxygen (O), 3 nitrogen (N), and 1 sulfur (S) per elemental composition. Data filtering for quality assurance of the assigned formulas was done as described previously.19 The detection limit for the observed ions was set at 0.1% RA (Composer minimum), despite the much lower signal-to-noise threshold. Due to this high method detection limit, the precursor ions without any detected product ions were removed from the main data

very useful for analyzing complex samples. Molecular formula assignment of its thousands of organic components requires interpretation of well resolved and accurate masses. Electrospray ionization (ESI) coupled with a Fourier-transform ion cyclotron resonance mass spectrometer (FT-ICR MS) is ideally suited for analysis of complex organic matter.20−24 Although molecular formulas have been identified for water-soluble organic compounds (WSOC) in AOM, investigations of the structure of these compounds are needed. Tandem mass spectrometry (MS/MS) with ultrahigh resolution mass analysis may provide structural characterization of the complex mixture. Several studies have been done to characterize natural organic matter using high- and ultrahigh-resolution MS,8,18−20,22,24−40 however far fewer have been done using high- and ultrahigh-resolution tandem mass analysis.26,27,29,34−38,41 In our previous work,19 measurements were obtained from a hybrid linear ion trap FT-ICR mass spectrometer. We observed a remarkable complexity in the mass spectra of water-soluble AOM isolated from Fresno fogwater. More than 1300 “monoisotopic” compounds containing only the most abundant isotopes (12C, 1H, 14N, 16O, 32S) were assigned to the negative ions measured in the mass range of 100 < m/z < 400. Of these, nearly 500 of them contained nitrogen. The assigned molecular formulas provide very little information related to the structure of the AOM components, thus additional analyses are needed. Due to the complexity of the AOM components FT-ICR MS was selected to obtain ultrahigh-resolution tandem mass analysis of organic anions of fogwater samples. The sample complexity hinders concrete structural elucidation, however it can be used to identify functional groups. Polar functional groups, such as hydroxyl, carboxyl, carbonyl, nitrate and sulfate, contain a significant amount of oxygen shown to be correlated with the hygroscopic properties of the atmospheric aerosols.42,43 Additionally, improvement of the structural characterization of atmospheric aerosol components will allow for a better understanding of the aerosol aging process. In the current study, we present the structural characterization of selected water-soluble organic compounds in polluted fogwater. Ultrahigh-resolution tandem mass analysis was conducted over selected mass ranges between 100 and 400 Da to target nitrogen and sulfur containing organic compounds. Identification of the nitrate, methyl-nitrate, sulfate, carboxyl, and hydroxyl functional groups associated with AOM is presented. Suspected monoterpene derived molecular formulas are presented; some of which have been reported in SOA chamber experiments and/or ambient samples.8,31,44−47

■

EXPERIMENTAL METHODS Sample Collection and Preparation. Fog sample collection and preparation were previously described in Mazzoleni et al.19 Briefly, radiation fog samples were collected in Fresno, CA in January 2006. A large stainless steel Caltech Active Strand Cloudwater Collector48,49 was set up at the California State University experimental farm in an open field. The site represents a polluted urban fog environment, influenced by emissions of residential and industrial activities and transportation. Fog samples were collected over 1−2 h time intervals and were stored in prebaked amber glass jars under refrigeration immediately after collection. Solid phase extraction (SPE; Phenomenex Strata-X) was used for sample preparation of fogwater samples. Strata-X is a hydrophilic sorbent made of surface modified styrene-divinylbenzene polymer (similar to Oasis HLB). The 1 g SPE cartridges were preconditioned with 4313

dx.doi.org/10.1021/es203509b | Environ. Sci. Technol. 2012, 46, 4312−4322

Environmental Science & Technology

Article

Figure 1. An excerpt of the mass spectra from 343.02 < m/z < 343.24 is shown to demonstrate the complexity within each nominal mass (top). Note the mass spectrum shown here is one without coaddition of transient data to improve the signal-to-noise ratio. The compound subgroups are distinguished by the symbol shape and series within the group are distinguished by color. #C vs m/z of the same nominal mass range 343.02 < m/z < 343.24 is shown (bottom). Compound subgroups are denoted by colored relative abundances (from the mass spectrum after coaddition of transient data) scaled symbols. The Δ36 mDa series (−O, +CH4) fall onto diagonal lines with slope of ∼0.04 (a line is drawn to show an example). The AOM complexity indicates the possibility for multiple matches between product ions and precursor ions (e.g., CO2 vs C2H4O).

and compounds containing C, H, N, O, and S (CHNOS)] for presentation. The total number and types of precursor compounds presented below are: 132 CHO, 166 CHNO, 75 CHOS, and 48 CHNOS compounds. As mentioned, the mass spectra are quite complex, for example there were 8−46 individual isobaric masses (with RA > 0.1%) identified within 1 nominal mass. Note isobaric masses have identical nominal mass but different exact masses. Several patterns of mass difference are evident within the isobaric masses. The most commonly repeating mass difference is 36 mDa. Ions with this mass difference have molecular formulas that differ by O and CH4. Due to the complexity of the sample, the molecular formulas of precursor and product ions were assigned with variable C and O content. Similar complexity has been observed in rainwater AOM,8 aerosol components,27,51 and fulvic acid samples.50,52 An excerpt of the mass spectrum at m/z 343 is shown in Figure 1(A). In the mass range of 343.02 to 343.23 Da, 10 series with 36 mDa mass differences are present. They include 2 CHO, 4 CHNO, 2 CHOS, and 2 CHNOS series. The compound groups are denoted with uniquely shaped symbols and the 36 mDa series within each group are distinguished by color. Alternatively, isobaric compounds are shown in Figure 1(B) with the number of carbon atoms vs m/z.53 The most oxidized compounds in each series are at the lower left of the plot, consistent with the lower numbers of carbon atoms, moving to the right with the exchange of O each for

presentation. The full list of precursors and the identified neutral losses is available in Table S-2 of the SI.

■

RESULTS AND DISCUSSION

As reported previously by Mazzoleni et al.,19 the mass spectra of water-soluble AOM isolated from Fresno fogwater is very complex. More than 1300 “monoisotopic” compounds containing only the most abundant isotopes (12C, 1H, 14N, 16O, 32S) were identified in the mass range of 100 < m/z < 400. All assigned ions were singly charged (as evident from the unit m/z separation between the 12Cc and 13C12Cc‑1 isotopomers for each elemental composition), even-electron ions, [M−H]- (due to electrospray ionization). Several compounds with N and S were identified with high relative abundances (RA). Note in ESI−MS analysis, RA is the product of initial neutral concentration and ionization efficiency. The occurrence and frequency of N- and S-containing compounds led to the selection of precursor mass ranges for fragmentation analysis. Since, the targeted N- and S-containing species could not be isolated from the other isobaric compounds, the precursor mass ranges included a wide number of additional compounds. All of the precursor molecular formulas were grouped by elemental composition [i.e., compounds containing C, H, and O (CHO), compounds containing C, H, N, and O (CHNO), compounds containing C, H, O, and S (CHOS), 4314

dx.doi.org/10.1021/es203509b | Environ. Sci. Technol. 2012, 46, 4312−4322

Environmental Science & Technology

Article

The AI values range from 0 to 1.5, where the higher AI values indicate more C−C double bonds. Using AI classifications, 57% of the precursors were defined as aliphatic (AI = 0); 37% were defined as olefinic (0 < AI ≤ 0.50); and 6% were defined as aromatic (AI > 0.5) (Table 1; Figure S-1 of the SI). The observed aromatic and condensed aromatic compounds (AI > 0.5) had the lowest RA values (Note RA refers to the ratio of the individual peak area to the base peak (100%) in the mass spectrum). RA is affected by the molecular ionization efficiencies in ESI, thus it is not a direct measure of concentration. Elemental ratio plots with RA scaled symbols and AI classifications are shown in Figures S-1, S-2, and S-3 of the SI. Interestingly, the assigned molecular formulas of the product ions have similar AI characteristics to the molecular formulas of the precursor ions (Figure S-1 of the SI). This confirms that the observed product ions resulted from simple losses of functional groups and not major structural units. Similar elemental ratio plots of the precursor ions are shown for each compound group, CHO, CHNO, CHOS, and CHNOS in Figures S-2 and S-3 of the SI. Overview of the AOM Neutral Losses. Neutral losses of CO2, H2O, CH3OH, HNO3, CH3NO3, SO3, and SO4 and combination losses were determined by matching the molecular formulas of assigned product ions to the expected molecular formulas for the isolated precursors with respect to each mass spectrum. The method of matching was selected after evaluation of the product ion mass spectra, which contained obvious clusters of ions associated with the loss of H2O, CH3OH, and CO2 (Figure 2; Figure S-4 of the SI). Note product ion spectra were subject to a low-mass cutoff due to the ion trap fragmentation and the transfer between mass analyzers. The frequencies of the matched neutral losses by compound subgroup and AI value are given in Table 1 and Figure 3 (See also Table S-2 of the SI for a full list of precursors and the identified losses). Neutral losses were not identified for approximately half of the total number of isolated precursors. The primary reason for this was related to the high method detection limit imposed by the software minimum RA of 0.1%. Typically, this value was 100 times the signal-to-noise threshold value. Thus, we retained only the precursors with 1 or more identified neutral loss (n = 421). The loss of H2O, is likely associated with hydroxyl functional groups and was observed with a high frequency. Note the interpretation of functional groups associated with the neutral losses was not tested with known compounds. A single loss of water was observed from 64% of the CHO and 40% of the CHNO compounds. Two H2O losses (H4O2) were observed for a large fraction of these. Multiple H2O losses can only occur from multifunctional components. Losses of CH3OH were observed from 50% of CHO and 42% of CHNO compounds. The loss of CH3OH indicates the presence of methoxy groups possibly from hemiacetal/acetal structures. Hemiacetal reactions between methanol and reactive carbonyl functional groups may occur in fogwater.11,14,55 Note that methanol was used during the sample extraction and analysis. Thus, it is possible the losses are associated with hemiacetal/acetal structures that may have resulted from either ambient aqueous or artificial processes.14 CH3OH losses were primarily associated with aliphatic and olefinic compounds, but the loss was also associated with several aromatic CHNO compounds. Losses of CO2 were especially prevalent with the CHO and CHNO olefinic and aromatic compounds and olefinic CHNOS compounds (Figure 3).

CH4 (Δ36 mDa) in the series. Additionally, there is a decrease of the double bond equivalent (DBE) as the series moves right, corresponding to each exchange. Interestingly, the highest DBE values correspond to the most oxidized compounds and have the lowest mass defects (note mass defect is the difference between the noninteger exact mass and the integer mass). The isobaric complexity of the water-soluble AOM components presents a substantial challenge for structural identification methods. To overcome this challenge, we conducted MS/MS analysis with ultrahigh resolution MS analysis to identify the AOM functional groups. The molecular formulas of the precursor and product ions were analyzed for several neutral losses, including: CO2 (44 Da), H2O (18 Da), CH3OH (32 Da), HNO3 (63 Da), CH3NO3 (77 Da), SO3 (80 Da), and SO4 (96 Da). The AOM complexity presents the possibility of multiple neutral loss matches between product and precursor molecular formulas (e.g., CO2 vs C2H4O). We found up to 10 oxygen atoms in some of our assigned molecular formulas and it is likely that the AOM compounds are multifunctional with a combination of hydroxyl, carbonyl, carboxyl, ester, nitrate, and sulfate functional groups. Thus, the likelihood for the matched neutral losses with higher O is expected to be higher. The presented neutral losses indicate the most likely functional groups associated with the precursor structures. Of the 421 compounds studied, the most frequently observed losses were CO2 (54%), H2O (43%) and CH3OH (40%) (Table 1). The losses are prominently observed in Table 1. Percent Frequency of Selected Neutral Losses neutral losses (%)a AI class

b

Aliphatic

Olefinic

Aromatic

group

H2O

CH4O

CO2

HNO3

SO3

CHO (n = 65) CHNO (n = 70) CHNOS (n = 43) CHOS (n = 62) CHO (n = 62) CHNO (n = 78) CHNOS (n = 3) CHOS (n = 13) CHO (n = 5) CHNO (n = 18) CHNOS (n = 2) CHOS (n = 0)

63 31 23 27 71 49 33 0 0 39 50 0

51 29 30 27 53 56 33 0 0 33 0 0

66 36 16 27 94 68 67 0 100 89 0 0

n/a 83 37 n/a n/a 58 0 n/a n/a 50 0 n/a

n/a n/a 37 82 n/a n/a 100 100 n/a n/a 50 0

a

The full list of precursors (n = 820) and the associated losses are also provided in Table S-2 of the SI. bCompounds were classified by the aromatic index (AI): aliphatic (AI = 0); olefinic (0 < AI < 0.5); and aromatic (AI.0.5).

the product ion spectra (e.g., Figure 2). More than half of these have 2 or more neutral losses associated with them. To evaluate the data for patterns with respect to the molecular formula properties, the data were sorted by the Koch and Dittmar54 aromaticity index (AI). The AI provides a compound classification metric related to the number of DBE in a chemical formula after the contributions from O, N, and S are removed. An alternate DBE equation (DBEAI = 1 + C − O − S − 0.5H) and alternate number of C atoms (CAI = C − O − N − S) is used in the equation for AI, given by eq 1:54 AI = DBEAI/CAI = (1 + C − O − S − 0.5H)/(C − O − S − N)

(1)

If DBEAI ≤ 0 or CAI ≤ 0, then AI = 0. 4315

dx.doi.org/10.1021/es203509b | Environ. Sci. Technol. 2012, 46, 4312−4322

Environmental Science & Technology

Article

Figure 2. Excerpts of the product mass spectrum after collision induced dissociation of the precursor ions (339 < m/z < 345): (A) an excerpt of the product ions ranging from 250 to 330; (B) an excerpt of the product ions from 323.0 to 323.3; (C) an excerpt of the product ions from 309.0 to 309.25; and D) an excerpt of the product ions from 297.0 to 297.3. Note the clusters of ions in A associated with the loss of H2O (18 Da), CH3OH (32 Da), and CO2 (44 Da). See also Figure S-3 of the SI for a detailed example of the neutral losses associated with m/z 342.

absorbing products, from glyoxal, in the presence of ammonium sulfate and ammonium nitrate was observed. Ammonium ions have been shown to act as a catalyst for accretion reactions.61−63 Reactions with aldehydes and dicarbonyls (i.e., glyoxal) produce various imidazole compounds38,60,64 and reactions with polycarbonyls produce imines.36,38,60 The reactions of glyoxal and amino acids have shown imidazole and diamine products,59 while reactions involving methylglyoxal and amino acids/methylamine have shown imidazole and accretion products.64 Similarly, reactions have been shown to occur with monoterpene oxidation products.36,38 These reports, along with our observation of isolated colored fog AOM components, suggest the presence of secondary products from the reaction of NH4+or NH3 with reactive carbonyls in this sample. The pH of the fogwater at the time of collection was 6.5. The abundance of agricultural emissions in the California Central Valley and the high partitioning coefficient for NH3 suggests that it or NH4+ were available for reaction, potentially forming imidazoles. Large concentrations of ammonium in fog waters collected in Fresno and elsewhere in the Central Valley have been reported previously by Collett et al.56 We observed two molecular formulas, C16H23NO3 and C16H22N2O3, in common with the laboratory study of limonene/ozonolysis SOA aging in the presence of NH3 gas by Laskin et al.38 Only a

NO, NO2, and NO3 functional groups were of interest in this study because of the numerous and highly abundant CHNO compounds in the sample.19 Neutral losses of HNO3 were observed for 83% and 37% of the aliphatic CHNO and CHNOS compounds, but only 58% of the olefinic CHNO compounds (Table 1). The loss was identified with 50% of the aromatic CHNO compounds. There were several aliphatic CHNO compounds with very high RA without an identified HNO3 or CH3NO3 neutral loss. The absence of HNO3 or CH3NO3 losses may suggest other structural forms of nitrogen, such as nitro, amine, or imine groups. NO and NO2 losses were observed from nitrophenols in the range of 138−197 Da, but not at the higher mass ranges (although product ions matching nitrophenols were observed). Additional nitrophenols and other CHNO compounds with nitro functional groups are expected to be present in Fresno fogwater, however the loss of NO and NO2 were rarely detected. Alternatively, the potential for reduced organic nitrogen species in aqueous samples has been well documented.33,36,38,56−64 Zhang and Anastasio reported the presence of amino acids and amino containing compounds in radiation fog samples from farther north in the Central Valley.57 Herckes et al.65 and Hutchings et al.66 observed several types of organic nitrogen compounds in fogwater collected in Fresno. In recent studies,60,62 the formation of light 4316

dx.doi.org/10.1021/es203509b | Environ. Sci. Technol. 2012, 46, 4312−4322

Environmental Science & Technology

Article

Figure 3. Percent frequency of neutral losses with respect to aromatic index (AI) classifications (aliphatic, olefinic, and aromatic): (A) the neutral losses associated with CHO compounds; (B) the neutral losses associated with CHNO compounds; and (C) the neutral losses associated with CHOS compounds; and (D) the neutral losses associated with CHNOS compounds. See also Table 2 and Table S-2 of the SI.

observed with a higher frequency than HNO3 for the CHNOS compounds. Selected SOA Components and Identified Neutral Losses. A large number of the selected precursor molecular formulas matched SOA molecular formulas. Thus, a comparison of the molecular formulas assigned to precursors in this study and those of a recent α-pinene/ozonolysis study67 was done to evaluate the presence of biogenic SOA in fog AOM. The fog episodes were sampled in Fresno, CA during the winter with an average temperature of 55 °F, thus emissions of terpenes were possible. Several common CHO formulas were found in the mass range of 292−356 Da (Table 2). Although the molecular formulas are identical to SOA molecular formulas, they may represent different isomeric structures. The C16 compounds range from O5−O8 with DBE = 5 and 6, with at

loss of CO2 was observed for the C16H22N2O3 precursor and no losses were observed for C16H23NO3. Organosulfate functional groups, determined by losses of SO3 and/or SO4, are important because they contribute to water-solubility and indicate the location of several oxygen atoms in the structures. Analysis of the precursor and product ions resulted in finding neutral losses representing SO3 and SO4 for 85% and 27% of the CHOS compounds and 42% and 0% of the CHNOS compounds. The loss of SO3 was identified for all of the CHOS and CHNOS olefinic compounds and a majority of the CHOS aliphatic compounds. Several aliphatic CHOS compounds of very high RA were observed without matching product ions representing losses of SO3 and/or SO4. It is possible that some of these compounds contain nonterminal sulfate groups.45 Interestingly, the losses of SO3 were 4317

dx.doi.org/10.1021/es203509b | Environ. Sci. Technol. 2012, 46, 4312−4322

Environmental Science & Technology

Article

Table 2. Selected Precursors and the Identified Neutral Lossesa selected precursors m/z

DBE

AI

molecular formula

293.1397 295.1554 309.1346 311.1505 325.1662 341.1245 341.1609 343.1401 353.1611 355.1767 274.0934 276.0727 276.1091 290.0884 292.1040 352.1770 354.1562 354.1926 275.0886 277.0674 289.156 279.0546 293.0703 295.0496 341.1431 355.1588 310.0604 326.0555 342.0503

6 5 6 5 5 6 5 5 6 5 4 4 3 4 3 6 6 5 4 4 7 3 3 3 5 5 3 3 3

0.1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.3 0 0 0 0 0 0 0 0

C16H22O5 C16H24O5 C16H22O6 C16H24O6 C17H26O6 C16H22O8 C17H26O7 C16H24O8 C18H26O7 C18H28O7 C11H17NO7 C10H15NO8 C11H19NO7 C11H17NO8 C11H19NO8 C18H27NO6 C17H25NO7 C18H29NO6 C10H16N2O7 C9H14N2O8 C16H22N2O3 C10H16O7S C11H18O7S C10H16O8S C17H26O5S C18H28O5S C10H17NO8S C10H17NO9S C10H17NO10S

neutral losses literature matchesb 67 67 67 67 67 67 67 67 67 67

H2O

CH4O H4O2 CO2

x

x

x

x

x x x x x

x x x x

x x

x

x x x x x x x x x x x

CH2O3 HNO3 C2H4O3 CH3NO3 x x x x x x x x x x x

x x x x

x x x

x x

38 46, 47

x x x

SO3 C2O4 SO4

x

x x x x x x x x x x x x x x x

x x

x x x x x x x x x x x x x

x x

x x x x x

x x

x

x x x x

x x

x

a

The full list of precursors (n = 820) and the associated losses are also provided in Table S-2 of the SI. bNumbers indicate the matched literature by reference number: 38 = Laskin et al., 2010; 46 = Surratt et al., 2007; 47 = Surratt et al., 2008; and 67 = Putman et al., 2012.

of HNO3 and CH3NO3 were found for all five of these analytes. These molecules appear to have methyl-nitrate functional groups (−CH2NO3). Although methyl-nitrate functional groups have been previously proposed,47 previous studies have also reported a neutral loss of CH3NO3 associated with nitrate functional groups (−NO3; 46). The NO7 compounds differ by only 2 H atoms which equates to one DBE. Losses of CO2, and CO2 + H2O were observed for C11H17NO7 and a loss of two H2O molecules was observed for C11H19NO7. This indicates that the double bond in the carboxylic acid functional group is likely responsible for the difference in DBE values. Two of the three NO8 compounds, C10H15NO8 and C11H17NO8, differ by a CH2 unit and C11H19NO8 differs from C11H17NO8 by two H atoms. In addition to the nitrate losses, a loss of CO2 + H2O was observed for C10H15NO8 and a loss of two H2O molecules was observed for C11H19NO8. A few CHNO compounds with two N atoms were also observed with low RA. Only a loss of HNO3 was observed corresponding to C10H16N2O7 and C9H14N2O8. The N2O7 and N2O8 compounds may represent monoterpene oxidation products with two nitrate groups. A few of the CHNO compounds have similar numbers of C and H atoms (C17−C18 and H25−H28) compared to “dimer” monoterpene oxidation products. For example, several molecular formulas with NO6 and NO7 and DBE = 5 or 6 were assigned (Table 2). Product ions corresponding to losses

least two compounds per O number, each differing by two H atoms (or one DBE). Product ions were observed for most of the C16 compounds corresponding to neutral losses including H2O, CH3OH, double H2O (H4O2), CO2, CO2 + H2O (CH2O3), and CH3OH + CO2 (C2H4O3). Neutral losses corresponding to H2O, CH3OH, CO2, CO2 + H2O, CH3OH + CO2, and two CO2 were observed for both C17H26O7 and C18H28O7. It is likely they have similar structures because the same neutral losses were observed for each of them and they differ by only CH2. C18H26O7 appears to be similar to the previous two compounds, with fragment ions corresponding to losses of H2O, CO2, and CO2 + H2O. Losses of CO2 and CO2 + H2O were identified for C17H26O6. Several additional common molecular formulas were found with low RA (