Characterization of aerosol nitroaromatic compounds: Validation of an experimental method

Abstract

The analytical capabilities associated with the use of silylation reactions have been extended to a new class of organic molecules, nitroaromatic compounds (NACs). These compounds are a possible contributor to urban particulate matter of secondary origin which would make them important analytes due to their (1) detrimental health effects, (2) potential to affect aerosol optical properties, and (3) and usefulness for identifying PM_2.5 from biomass burning. The technique is based on derivatization of the parent NACs by using N,O-bis-(trimethylsilyl)-trifluoro acetamide, one of the most prevalent derivatization reagent for analyzing hydroxylated molecules, followed by gas chromatography-mass spectrometry using electron ionization (EI) and methane chemical ionization (CI). This method is evaluated for 32 NACs including nitrophenols, methyl-/methoxy-nitrophenols, nitrobenzoic acids, and nitrobenzyl alcohols. Electron ionization spectra were characterized by a high abundance of ions corresponding to [M⁺ ] or [M⁺ - 15]. Chemical ionization spectra exhibited high abundance for [M⁺ + 1], [M⁺ - 15], and [M⁺ + 29] ions. Both EI and CI spectra exhibit ions specific to nitro group(s) for [M⁺ - 31], [M⁺ - 45], and [M⁺ - 60]. The strong abundance observed for [M⁺ ] (EI), [M⁺ - 15] (EI/CI), or [M⁺ + 1] (CI) ions is consistent with the high charge stabilizing ability associated with aromatic compounds. The combination of EI and CI ionization offers strong capabilities for detection and identification of NACs. Spectra associated with NACs, containing hydrogen, carbon, oxygen, and nitrogen atoms only, as silylated derivatives show fragment/adduct ions at either (a) odd or (b) even masses that indicate either (a) odd or (b) even number of nitro groups, respectively. Mass spectra associated with silylated NACs exhibited 3 distinct regions where characteristic fragmentation with a specific pattern associated with (1) –OH and/or –COOH groups, (2) –NO₂ group(s), and (3) benzene ring(s). These findings were confirmed with applications to chamber aerosol and ambient PM_2.5.

Graphical Abstract

INTRODUCTION

Ambient aerosols are complex mixtures of water, carbonaceous compounds, inorganic salts, metals, and mineral dust.1 These particles can have direct environmental impacts by scattering light or absorbing solar radiation, leading to visibility degradation2, 3 and altering the amount of solar radiation reaching the surface of the earth.4 Ultrafine particles can also indirectly affect cloud properties and the hydrological cycle by acting as cloud condensation nuclei.5, 6 In addition, it is fairly well established that extended exposures to fine particulate matter (PM_2.5) can lead to adverse health effects.7, 8 It is expected that understanding the molecular composition will be required to help explain the toxic properties associated with PM_2.5.

While the inorganic fraction of PM_2.5 is well established, the organic fraction, which can comprise up to 80% of the aerosol mass,9, 10 has proven difficult to understand at the molecular level.10 Organic compounds in the atmosphere encompass a wide range of chemical classes (eg, alkanes, alkenes, aromatics, alcohols, carbonyls, carboxylic acids, and compounds containing heteroatoms and functional groups, such as –COOH, –OH, –NH, and –SH). It has been speculated that 10⁴ to 10⁵ volatile organic compounds are present in the atmosphere, and many are unsaturated or aromatic, leading to highly oxidized compounds and secondary organic aerosol (SOA) formation.11, 12 An understanding of the physical properties of SOA, such as optical extinction, typically requires a grasp of the detailed structure at the molecular level.13–15 However, the chemical analysis of polar compounds from SOA is difficult due to polar interactions with the chromatographic liquid phase, which leads to peak broadening resulting in poor separation and calibration difficulties. The use of derivatization has been found to be a useful method for analyzing polar compounds in complex samples. A common method uses the silylation reagent N,O-bis-(trimethylsilyl [TMS])trifluoro acetamide (BSTFA), to produce a [–Si(CH₃)₃)] moiety for each reactive hydrogen atom [eg, –COOH, –OH, –NH, and –SH] present.16

A literature review shows that polar nitroaromatic compounds (NACs) are partially responsible for optical properties of ambient urban particles, thereby influencing their aerosol radiative forcing.15 Nitroaromatic compounds are typically associated with products from aromatic oxidations,17 biogenic oxidations,18 and biomass burning.17 In addition, a broad range of NACs have been detected in chamber irradiations of these precursors, as well as from field samples.19, 20 In a series of aromatic hydrocarbons with oxides of nitrogen irradiations, Xie et al measured individual NACs by liquid chromatography-mass spectrometry and determined their mass absorption efficiency at wavelengths from 375 to 550 nm.20 However, the low chromatographic resolution of the liquid chromatography-mass spectrometry method did not permit many of the NACs to be completely resolved, for example, isomeric compounds. While Morville et al,21 Schummer et al,22 and Irei et al23used silylation reactions on a small number of nitrophenols, derivatives were analyzed only in electron ionization (EI) which lead to only limited identifications of unknown NACs due to the likely absence of a molecular ion.

The focus of the analytical technique in this study is the characterization of NACs bearing (1) aromatic rings, (2) nitro (–NO₂), and (3) hydroxyl –OH) or carboxylic acid –COOH) groups. Here, we have developed a single-step derivatization by using BSTFA to analyze a wide range of compounds containing not only hydroxyl/carboxylic groups16 but also benzene ring(s) and/or nitro groups (–NO₂). Tandem mass spectrometry (MS/MS) spectra are obtained for the molecular or fragment ions of some compounds, which are used for further identification and to elucidate dissociation pathways. The analytical approach leaves the –NO₂ groups unperturbed. The method has been applied to 39 model compounds and SOA samples produced from chamber irradiations of aromatic precursors, including benzene, toluene, ethylbenzene, o-, m-, and p-xylenes and other higher molecular weight compounds having 1 or more aromatic rings. Also included among the precursors were oxygenated aromatics, such as m-cresol. Several parameters, such as solvent, reaction temperature, and reaction time, were optimized for extraction and silylation efficiency.16

EXPERIMENTAL METHODS

Chemicals and solvents

All chemicals, including standards and derivatization reagent BSTFA, with 1% trimethylchlorosilane (hereafter simply BSTFA), were purchased from Aldrich Chemical Co. (Milwaukee, WI) at the highest purity available and were used without further purification. All solvents (gas chromatography [GC]² quality) were from Burdick and Jackson (Muskegon, MI). Glassware was washed with soap, rinsed with hot water, and dried overnight at 200°C. Before use, the glassware was rinsed 3 times with acetone and 3 times with methylene chloride and dried at 200°C.

Model compounds and standard preparation

A total of 39 commercially available, organic chemicals serving as the model standards were selected to test the analytical method (Table 1). The selected compounds were chosen for their relevance to atmospheric systems, for structural variety, and to cover a wide range of polarities. The compounds are associated with 4 functional groups including (1) (–OH/–COOH) groups and a single benzene ring without (–NO₂) groups; (2) (–OH), (–NO₂), and benzene ring(s); (3) (–COOH), (–NO₂), and benzene rings; and (4) methyl/methoxy/ethyl substituted benzene rings ( –CH₃; –OCH₃; –C₂H₅, –NO₂, –OH, and/or –COOH) groups. Also found in Table 1 are the molecular weights of the parent compounds (MW), molecular weights of their respective BSTFA-derivatives (MW_bstfa), nomenclature, and structure of each compound.

Table 1.

Model compounds analyzed in this study. The five ions reported in methane CI mode are given in decreasing abundance; Bold (M⁺ + 1); underline (M⁺ − 15); Italic (M⁺ − 45); bold and Italic (M⁺ − 60). All EI and CI mass spectra exhibit an ion at m/z 73 [TMS⁺]. EI and CI mass spectra are shown in supplementary information.

# cpd	Nomenclature	Structure	MW (MWbstfa) (g mol−1)	5 Most intense ions (methane CI)
1	Phenol(C6H6O)		94 (166)	167, 151, 195, 207, 73
2	Catechol (C6H6O2)		110 (254)	73, 255, 239, 283, 295
3	p-Toluic acid		136 (208)	193, 209, 155, 273, 73
4	2-naphthoic acid (C11H8O2)		172 (244)	229, 245, 225, 193, 311
5	2-naphthol C10H8O)		144 (216)	217, 201, 245, 173, 145
6	2-methyl-1-naphthoic acid (C12H10O2)		186 (258)	169 (M+ − 89), 243, 259, 73,197
7	4-hydroxy benzoic acid		138 (282)	283, 267, 225, 193, 311
8	2,4-dinitrophenol		184 (256)	257, 241, 211, 195,285
9	2,5-dinitrophenol		184 (256)	257, 241, 73, 285, 225
10	4-methyl-2-nitrophenol (C7H7NO3)		153 (225)	226, 209, 166, 254, 181
11	3-methyl-4-nitrophenol (C7H7NO3)		153 (225)	210, 226, 254, 136, 180
12	4-nitrobenzoic acid (C7H5NO4)		167 (239)	224, 240, 150, 268, 178
13	2-hydroxy-5-nitrobenzyl Alcohol (C7H7NO4)		169 (313)	314, 181, 226, 270, 211
14	2-methoxy-5-nitrophenol (C7H7NO4)		169 (241)	242, 181, 226, 270, 211
15	4-nitroguaiacol (C7H7NO4)		169 (241)	242, 181, 226, 270, 196
16	4,5-dimethoxy-2-nitrobenzyl Alcohol (C9H11NO5)		213 (285)	196(M+ −89), 270, 286, 224
17	4-nitrotoluene (C7H7NO2)		137	138, 92, 122, 166, 178 not a TMs derivative
18	2-nitrophenol (C6H5NO3)		139 (211)	196, 212, 240, 122, 166
19	3-methyl-2-nitrophenol (C7H7NO3)		153 (225)	210, 226, 136, 254, 194
20	2-nitrobenzyl alcohol (C7H7NO3)		153 (225)	226, 180,165, 254, 208
21	2-methyl-3-nitrophenol (C7H7NO3)		153 (225)	136(M+ − 89), 210, 226, 194
22	5-methyl-2-nitrophenol (C7H7NO3)		153 (225)	210, 226, 254, 136, 194
23	2,6-dimethyl-4-nitrophenol (C8H9NO3)		167 (239)	240, 194, 268, 224, 280
24	4-methoxy-2-nitrophenol (C7H7NO4)		169 (241) (2 peaks)	226, 152, 124, 270, 242
25	3-methoxy-4- nitrobenzyl alcohol (C8H9NO4)		183 (255)	256, 284, 239, 166, 210
26	4-nirophenol (C6H5NO3)		139 (211)	212, 195, 182, 240, 166
27	2-methyl-4-nitrophenol (C7H7NO3)		153 (225)	226, 209, 254, 180, 165
28	4-nitrocatechol (C6H5NO4)		155 (299)	300, 328, 284, 254, 340
29	2-methyl-4-nitroresorcinol (C7H7NO4)		169 (313)	314, 298, 224, 342, 282
30	2-nitrophloroglucinol (C6H5NO5)		171 (315)	272, 388, 298, 416, 356
31	2-methyl-5-nitrobenzoic acid (C8H7NO4)		181 (253)	254, 238, 164, 282, 294
32	2,6-dinitrophenol (C6H4N2O5)		184 (256)	241, 226, 257, 196, 210
33	4,6-dinitro-o-cresol (C7H6N2O5)		198 (270)	241, 255, 299, 239, 199
34	2-nitro-1 -naphthol (C10H7NO3)		189 (261)	172(M+. − 89), 246, 262, 290
35	2,5-dimethyl-4- nitrobenzoic acid (C9H9NO4)		195 (267)	268, 252, 178, 296, 236
36	4,6-dinitro-2-methylresorcinol (C7H6N2O6)		214 (358)	357(M+.−1), 431, 313, 297, 223 [Note (M+. - 1) was observed instead of (M+. + 1)]
37	2-(2-nitrovinyl)furan (C6H5NO3)		139 (139)	140, 97, 123, 110,83 not a TMs derivative
38	5-nitrosalicylic acid		183 (327)	328, 312, 368, 372, 73
39	4-nitro-2-furoic acid		157 (229)	230, 215, 258, 270, 184

Primary stock solutions containing 1 to 10 analytes were prepared gravimetrically at about 300 to 1000 μg cm⁻³ in acetonitrile, methanol, or in a mixture of methanol and methylene chloride (1:1 vol/vol). The estimated precision for formulating these standards is 0.1%. The primary solutions were used to prepare mixtures at concentrations consistent with the sensitivity of the GC-MS. Thus, inlet injection volumes were selected to give on-column masses from 1 to 350 ng. To adapt this method to samples collected in air, an optimization of the filter extraction technique was conducted. Blank 47-mm glass fiber filters were spiked with NAC stock solutions, and known amounts of internal standards ketopinic acid (KPA), p-menth-6-ene-2,8-diol (PMD), and d₅₀-tetracosane (TCS). In our previous work,16 a 50:50 methanol/methylene chloride solvent mixture was found to quantitatively extract aromatic components in SOA and has been adopted for filter extractions of NACs. In the extraction technique, filters underwent sonication for 1 hour, followed by evaporation of the extracts under a gentle stream of ultrapure N₂ by using an N-Evap evaporation bath (Organomation Associates, Inc., Berlin, MA).

The derivatization of the model compounds and filter extracts was conducted by using a method similar to that found in Jaoui et al.16 For the model compounds, a 25-μL aliquot of each primary solution was transferred to a 5-mL conic-bottom tube with a Teflon-lined cap also containing 50 μL each of KPA at 581 μg mL⁻¹, PMD at 1001 μg mL⁻¹, and TCS at 569 μg mL⁻¹, then evaporated to dryness as for filter extracts. The procedure was repeated for each solution of the model compounds. Each dried extract was dissolved in 200 μL of BSTFA and 100 μL of pyridine. The tubes were sealed with a Teflon cap and Parafilm and allowed to react for 1 hour at 70°C. After cooling to room temperature, the solutions were transferred to GC-MS vials and analyzed by GC-MS in the EI and methane-chemical ionization (CI) modes. In some cases, depending on the derivative concentration, dilutions were made to avoid overloading the column. Three replicates were injected to check instrumental reproducibility. Integrated peak areas for recovery calculations were referenced to KPA peak areas, and the percent recovery was calculated for each compound by comparing extracted spiked filters to the standards prepared above.

GC-MS analysis

Two GC-MS systems were used in this study. The first analysis was conducted by using a ThermoQuest GC instrument coupled with an ion trap mass spectrometer (Austin, TX). The injector was operated in splitless mode and at 270°C. Compounds were separated on a 60 m long, 0.25-mm ID RTx-5MS column (Restek, Inc., Bellefonte, PA) with 0.25-μm film thickness. The oven temperature program started isothermally at 84°C for 1 minute followed by a temperature ramp of 8°C min⁻¹ to 200°C, followed by a 2-minute hold, then 10°C min⁻¹ramp to 300°C and a 15-minute hold. The ion source, ion trap, and interface temperatures were 200, 200, and 300 °C, respectively. Tandem mass spectrometry measurements were made for some compounds by the collisionally activated dissociation of the most abundant ion by using He as the collision gas. To investigate the mass spectrum reproducibility across instruments, a set of samples were also analyzed with an Agilent GC-MS instrument equipped with a Model 7693 automated sampler, a Model 7890 B gas chromatograph with a splitless injector, and a Model 5977 B mass selective detector. A DB-5MS capillary column (30 m, ID 0.25 mm, film thickness 0.25 μm) was used for the analysis. For these measurements, the inlet temperature was set to 280°C. The temperature program started at 84°C for 1 minute, raised to 200°C at 10°C min⁻¹, then to 300°C at 30°C min⁻¹ where it was held for 5 minutes. The mass spectrometer was operated in EI (70 eV) and methane-CI full-scan modes. For quantification of the target analytes, a single-ion monitoring mode was used for the most abundant ions without interferences. The transfer line, source, and quadrupole temperature were set to 280, 230, and 150°C respectively.

RESULTS AND DISCUSSION

Gas chromatography parameters

Poor chromatography is often a problem when analyzing ambient polar organic compounds (POCs); therefore, derivatization procedures are an important approach for generating structural information at the molecular level. While BSTFA derivatization has been used extensively for nonnitro-containing hydroxylated compounds,24 only limited studies have been associated with NACs analysis, mainly nitrophenols.20–23 Therefore, the silylation parameters including reaction temperature and time have been optimized for the NACs. The results show that temperatures between 60 and 90°C and reaction times between 60 and 120 minutes have negligible effects on the analytical responses, similar to results reported by Jaoui et al16 for compounds having C, H, and O-atoms only. It was also found that the exact choice of the extraction solvent, methanol, acetonitrile, or methanol/methylene chloride (50:50) has little, if any, influence on the analytical response factors. For quantitative analysis, the analytical responses were normalized to the analytical response of TCS, as well as KPA or PMD. Good separation of silylated NACs was achieved for the different mixtures by using a 30-m or 60-m DB5 column. Specific optimizations of GC parameters were necessary for the organically complex ambient samples to minimize coelution (see Supporting Information).

MASS SPECTROMETRY PARAMETERS FOR CI AND EI

Mass spectrometry has proven to be a valuable tool for identifying POCs found in ambient aerosol. For this study, the method allows for the detection of extensive molecular fragmentation despite numerous molecular rearrangements which can occur. These rearrangements can be characteristic of particular classes of compounds, thus aiding the identification of unknown compounds. As noted, the mass spectra from derivatives formed by the silylation reactions are essential for identification.

To demonstrate the method, the mass spectra of silylated model compounds having nitro groups were evaluated to establish molecular weights and structural information of the model compounds. In particular, benzene rings and nitro groups play an important role in directing the fragmentation. During the silylation reaction, hydroxyl groups are converted into their corresponding TMS via an SN₂ substitution reaction, yielding a single derivative for each acidic and nonacidic OH. Depending on the number of (–NO₂) groups in the underivatized molecules, the mass spectra show (1) odd ions for zero (–NO₂) group, (2) even ions for odd number of (NO₂) groups, and (3) odd ions for even number of (–NO₂) groups. This behavior of silylated NACs is generalized in Scheme 1 by using phenol as a representative compound for Group 1, 4-nitrophenol for Group 2, and 2,5-dinitrophenol for Group 3. In analytical responses, characteristic CI and EI mass spectra of the derivatized compounds are considered in more detail below. First, we consider the CI spectra of NACs.

Scheme 1.

Generalized silylation reaction of nitroaromatic compound (NAC); MW, molecular weight of the underivatized NACs

CI mass spectra of NAC compounds

Under positive methane-CI ionization, new ion species [M + H]⁺, [M + C₂H₅]⁺, and [M + C₃H₅]⁺, corresponding to (M⁺ + 1), (M⁺ + 29), and (M⁺ + 41), respectively, are formed through chemical reactions between methane reagent ions CH₅⁺, C₂H₅⁺, and C₃H₅⁺ and neutral analyte molecules (M). Subsequent fragmentation of [M⁺ + 1] involves the loss of a moiety to form further ionized species. Scheme 2 (top) shows bimolecular reactions leading to CH₅⁺, C₂H₅⁺, and C₃H₅⁺ formed in the ion source by using methane as the reagent gas. To rationalize the fragmentation of [M⁺ + 1] ions associated with NACs, one must consider which sites of the target NAC the proton is attached. The spectrum may then be rationalized in the fragmentation of the different types of [M⁺ + 1] ions. In this case, protonation occurs on N- and O-heteroatoms having lone pairs of electrons. This process is frequently followed by charge-induced elimination of the moiety containing the heteroatom. One additional protonation site is the aromatic ring. The quasimolecular ion [M⁺ + 1] in NACs, often representing the base peak, is formed through proton transfer reaction between the analyte molecule (M) and CH₅⁺ (Scheme 2, bottom). The [M⁺ + 29] and [M⁺ + 41] adducts are formed by electrophilic addition of the reagent ions C₂H₅⁺ and C₃H₅⁺, for example, 2,6-dimethyl-4-nitrophenol shown in Scheme 2 (bottom portion).

Scheme 2.

(top) Bimolecular reactions occurring in the ion source in methane-chemical ionization leading to CH₅⁺, C₂H₅⁺, and C₃H₅⁺; (bottom) formation of the quasimolecular ion [M⁺ + 1] and adducts [M⁺ + 29] and [M⁺ + 41] using 2,6-dimethyl-4-nitrophenol as an example

EI mass spectra of underivatized NAC compounds

The mass spectra of nitroaromatics, which do not form BSTFA derivatives, have been studied extensively mainly in EI mode, and to a lesser extent polar oxygenated aromatic containing –OH/–COOH groups.25, 26 One salient feature of these compounds is their high charge stabilizing ability, which is reflected in the presence of a molecular ion in their EI mass spectra. Due to limited amount of information on NACs,22, 23 the work on these classes of compounds becomes important in the interpretation of NACs because it might be expected that the NACs show similar characteristics. Because mass spectra of nitroaromatic and hydroxylated aromatic compounds were generated in EI mode, and mostly for the underivatized aromatic compounds, we have provided a number of mass spectra for silylated nitroaromatics and polar aromatics in EI and CI modes (Compounds 1 to 7, 17, and 37). The interpretation of silylated mass spectra for nonnitro/benzene ring containing compounds has been reported in several papers and therefore not discussed here.16, 24The information obtained from the interpretation of these mass spectra has been compared with mass spectra generated for NACs to see whether characteristic fragments were common to these classes of compounds.

Mass spectra of compounds with and without benzene rings (absence of (–NO₂) group: Rule M⁺_(−15/1/29))

Electron ionization and CI mass spectra of silylated model compounds were evaluated to determine whether characteristic fragmentation patterns of silylated aromatic and nonaromatic compounds of unknown oxygenates could be found from an interpretation of their mass spectra. Figure 1 shows methane-CI mass spectra of phenol (A), catechol (B), pentanedioic acid (C), 2-methylthreitol (D), undecanedioic acid (E), and 1,2,3,4-cyclobutanetetra carboxylic acid (F). Compounds containing benzene ring (eg, phenol and catechol) exhibit strong peaks at [M⁺ + 1], [M⁺ − 15] and [M⁺ + 29] (Figure 2A and B) and weak peaks between m/z 73 and [M⁺ − 15]. [M⁺ + 1] are formed through proton transfer reaction (M + CH₅⁺ ➔ MH⁺ + CH₄). The [M⁺ + 29] adduct is formed through electrophilic addition through attachment of reagent ion C₂H₅⁺ to the analyte molecule (Scheme 2 bottom). This pattern of mass spectra in CI mode is named Rule [M⁺_(−15/1/29)]. Electron ionization mass spectra of aromatic hydrocarbons usually show similar features but with a strong [M⁺] and/or [M⁺ − 15] which is consistent with the high charge stabilizing ability associated with benzene ring. This fragmentation trend (Rule M⁺_(−15/1/29)) was observed only in NAC-TMS, although other classes of species not containing benzene ring [M⁺ − 15], [M⁺ + 1], and [M⁺ + 29] were present with a strong ion at [M⁺ − 89] representing the base peak (eg, linear dicarboxylic acid with 9 carbon atoms or more: undecanedioic acid, Figure 1E, and 1,2,3,4-cyclobutanetetracarboxylic acid, Figure 1F). For compounds reported by Jaoui et al,16 Rule M⁺_(−15/1/29) was not observed due probably to the absence of benzene ring as can be seen in Figure 1C and D for pentanedioic acid and 2-methylthreitol. For compounds containing benzene ring, the base peak is either [M⁺ + 1] or [M⁺ − 15]. Results show in general that CI mass spectra associated with silylated compounds containing aromatic rings produce a low-abundant weak or no fragment ion corresponding to [M⁺ − 89], which is one of the most prevalent fragments associated with nonaromatic compounds.16

Figure 1.

Chemical ionization mass spectra of silylated compounds: phenol, catechol, pentanedioic acid, 2-methylthreitol, undecanedioic acid, and 1,2,3,4-cyclobutanetetracarboxylic acid

Figure 2.

Examples of chromatograms of silylated model compounds used in this study

CI mass spectra of derivatized NACs: Rules M⁺_(−15/1/29) and M⁺_{(−60/−45/−31)}

Each NAC containing OH and/or COOH group was silylated to form a single peak (Figure 2). Figure 3 shows CI mass spectra of silylated 2-nitrophenol (A), 4-nitrocatechol (B), 2,5-dinitrophenol (C), and 2-nitrophloroglucinol (D). Phenol and catechol as model compounds containing benzene ring but not (–NO₂) groups are shown in Figure 1A and B, respectively. All CI mass spectra recorded in this study for NACs (Figure 3 and Supporting Information) follow Rule M⁺_(−15/1/29) similar to compounds containing aromatic ring but not nitro groups (see previous section). This observation shows that TMS and nitro group(s) do not affect the high charge stabilizing ability associated with benzene rings (Rule M⁺_(−15/1/29)). However, they contain all additional important characteristics specific to the presence of nitro groups, that is, ions at [M⁺ − 31], [M⁺ − 45], and [M⁺ − 60], denoted as Rule [M⁺_{(−60/−45/−31)}] (see for example phenol vs 2-nitrophenol, and catechol vs 4-nitrocatechol; Figures 1 and 3 and Supporting Information). Additional weak ions at m/z 73 and 75 and ions corresponding to [M⁺ − 89] and [M⁺ + 73] were detected and are common to all silylated model compound used in this study and Jaoui et al.16 The base peak in the CI mass spectra is either [M⁺ + 1] or [M⁺ − 15] ion. A fragmentation pattern following Rule [M⁺_{(−60/−45/−30)}] as well as Rule M⁺_(−15/1/29) provides important information that can be used to differentiate between NACs and other class of compounds, particularly when field samples are analyzed.

Figure 3.

Mass spectra of silylated compounds in chemical ionization mode: 2-nitrophenol, 4-nitrocatechol, 2,5-dinitrophenol, and 2-nitrophloroglucinol

Comparison between EI and methane CI mass spectra

Although EI spectral libraries are widely used for compound identification, few EI mass spectra associated with silylated compounds are available. Based on our experience generating methane-CI spectra of a wide variety of silylated compounds, we have found that these spectra are highly specific, reproducible, and produce characteristic fragments useful in determining structural information and molecular weight, when authentic standards are not available.16, 19, 24 In this study, 32 silylated NACs (Table 1) were analyzed in EI and CI modes in the full scan (Supporting Information). For example, Figure 4 shows EI, methane-CI, and MS/MS mass spectra of silylated 2,6-dimethyl-4-nitrophenol (26DMNP). Mass spectra of 26DMNP-TMS exhibit a strong ion abundance for [M⁺ + 1] (CI) (m/z 240), [M⁺] (EI) (m/z 239), and [M⁺ − 15] (EI) or [(M⁺ + 1) − 16] (CI) (m/z 224). The loss of methyl radical from silylated compounds is found to dominate the product ion spectra for all NACs. Characteristic fragment ions in EI, due to the nitro substituent in nitroarenes for [M⁺ − NO₂], [M⁺ − O] (weak), and [M⁺ − NO],27 are observed also in this study. The relative ion intensities in EI and CI depend on differences in the structures of NACs, although no pattern is observed because of the various type, position, and substituents on the nitroaromatic ring.

Figure 4.

A, Mass spectral fragmentation of 2,6-dimethyl-4-nitrophenol-trimethylsilyl methane-chemical ionization; B, electron ionization; C, product ion spectrum collision-induced dissociation MS² (m/z 240) yields product ions at m/z 223, 194, 179, and 149; D, further fragmentation of m/z 223 (collision-induced dissociation MS²) yields product ions at m/z 208, 193, 178, 177, 163, and 119

Collision-induced dissociation performed on the protonated molecule (m/z 240) in CI mode by using a quadruple mass filter resulted in an interesting product ion at m/z 223 (Figure 4C). This product ion is proposed to result from the loss of an OH group, as shown in Scheme 3A. For simplicity, Scheme 3A shows only fragmentation pathways originated from the quasimolecular ion [M⁺ + 1], in which proton transfer reaction occurred on the oxygen associated with (–NO₂) group (Scheme 2: bottom). The product ion at m/z 223 undergoes further reactions (Figure 4D) by loss of NO to form ions at m/z 193 or a CH₃ to form ions at m/z 208 (Scheme 3A). Scheme 3A shows also the formation of product ions at m/z 209 (Figure 4A) from the quasimolecular ion [M⁺ + 1] through a rearrangement followed by loss of HNO. The weak intensity of m/z 223 is consistent with a further decomposition, as shown in Figure 4D. In fact, collision-induced dissociation was conducted on this ion (Figure 4D) and resulted in a mass spectrum with major identifiable ions at m/z 208 [loss of 15 u: methyl], 193 (loss of 30 u: NO), 178 (loss of 45 u: NO + methyl), 177 (loss of 46 u: NO₂), 163 (loss of 60 u: 2 methyl + NO), and 119 (loss of 89 u: [–OSi(CH₃)₃]). Some of these ions (not shown in Scheme 3A) could be explained from the quasimolecular ion [M⁺ + 1], in which a proton transfer reaction occurred on the oxygen associated with the silyl group (Scheme 2: bottom).

Scheme 3.

A, Proposed fragmentation pathways originated from the ion [M⁺ + 1] of 2,6-dimethyl-4-nitrophenol-trimethylsilyl (TMS) leading to the main product ions observed in methane-chemical ionization mode. B, Proposed fragmentation pathways originated from the ion [M⁺] of 2,6-dimethyl-4-nitrophenol-TMS leading to the main product ions observed in electron ionization mode

In the 26DMNP EI spectrum (Figure 4B), characteristic product ions were observed at m/z207 (loss of 32 u: oxygen atom and CH₄), 208 (loss of 31 u: methyl + CH₄), 178 (loss of 61 u: methyl + NO₂), 179 (loss of 60 u: loss of 2 methyl + NO), 163 (formed from m/z 224 through loss of HNO and 2 CH₃ groups), and 149 (loss of 90 u: HOSi(CH₃)₃ from [M⁺]). The mechanistic pathways leading to a subset of these ions are shown in Scheme 3B. The ion at m/z 91 observed with relatively low abundance in EI and CI has been proposed to be associated with (C₇H₇⁺) tropylium ion for nitro-substituted aromatics formed through α-cleavage of the nitro group.28, 29 The ion at m/z 223 can be explained by loss of oxygen from [M⁺] as proposed for a series of nonsilylated nitro/hydroxy-aromatics.27, 28, 30 An additional pathway associated with [M⁺ − 16] is proposed here through loss of CH₄: that is, first loss of methyl group followed by cyclization with loss of 1 hydrogen atom as shown in Scheme 3B. A similar cyclization reaction was proposed for the silylated 1-hydroxypyrene to explain the [M⁺ − 31] ion formation.22 Scheme 3B shows additional proposed pathways leading to the main product ions observed in the EI mode (Figure 4B).

The most common and abundant ions detected in silylated NACs are [M⁺ + 1], [M⁺ − 15], and [M⁺ + 29] in CI mode and [M⁺] and [M⁺ − 15] in EI mode. The base peak is either [M⁺ − 15] or [M⁺] in EI and [M⁺ − 15] or [M⁺ + 1] in CI. This finding is consistent with characteristic fragmentation features observed for non-TMS aromatics/nitroaromatics due to their high charge stabilizing ability associated with the aromatic ring. The charge stability was not affected by the presence of TMS groups. The nitro group has been reported to have poor charge stability for nitroaliphatics.28, 31, 32 Here, the TMS group in NAC-TMS appears to have charge stability, due to the weak [M⁺ − 89] ion in 26DMNP-TMS (Figure 4B), as well as strong [M⁺], [M⁺ + 1], and [M⁺ − 15] ions in EI or CI ionization. This finding is consistent also with all other NACs studied having a weak [M⁺ − 89] ion. Here, each class of compounds presents additional but weak characteristic fragments (eg, m/z 73, [M⁺ + 73], [M⁺ + 41], [M⁺ − 89], [M⁺ − 105], and [M⁺ − 117]), aiding in the mass spectral interpretation. The EI and CI spectra associated with silylated NACs show fragments at [M⁺ − 31], [M⁺ − 45], and [M⁺ − 60], characteristic of –NO₂ group(s), typically not detected in non-NACs. Depending of the structure of the NAC, additional specific ions may occur, usually of low abundance, including [M⁺ − 61], [M⁺ − 46], [M⁺ − 47], and [M⁺ − 30], which are interpreted as loss of (CH₃ + NO₂), NO₂, HNO₂, and NO, respectively. Other ions were observed for some silylated NACs (Supporting Information) for [M⁺ − 105] and [M⁺ − 117] and others being characteristic fragments from the presence of –OH and/or –COOH groups. The combination of EI, CI, and MS/MS analysis is complementary and of value for identifying NAC molecular weights, the presence of aromatic rings, nitro groups, and other functionalities (eg, OH and COOH). The CI analysis of the silylated compounds yielding a highly abundant protonated molecule is of special value given that NACs by EI often show a weak or negligible molecular ion. In addition, m/z 91 is observed for all standards in both modes and consistent with the tendency of aromatics to produce benzylic cations. Finally, the mass spectra of NACs show a [M⁺ − 31] ion from silylated hydroxy aromatics following the cleavage of the TMS ether moiety involving 2 methyl groups and 1 hydrogen followed by cyclization.22, 33 The 5 strongest CI fragment ions are shown in Table 1.

QUANTITATIVE ANALYSIS OF NACS

In addition to the molecular weight, the silylation method for NACs reported here gives 2 important pieces of structural information as seen by Rule M⁺_{(−60/−45/−30)} and Rule M⁺_(−15/1/29). The qualitative findings can serve as the basis for quantitative analysis by using recovery and internal standards. Previous quantitative analysis was performed on non-NACs containing hydroxyl and/or carboxylic groups.16 Calibration curves were prepared by dilution of a standard mixture (see Supporting Information). The compounds in Tables S1and S2 have been chosen due to their structural variation similar to NACs detected in chamber irradiations and in ambient aerosol. The reproducibility of the method was evaluated by comparing analyte response factors for each of 3 replicate solutions (Table S1). Calibration curves were developed for the 8 silylated compounds (Tables S2 and S3) by plotting the relative integrated peak areas versus relative amount. Analyte mass injected in the GC-MS instrument ranged from 72 to 621 ng (Table S3). Linearity was assessed by using least-square regression. Relative standard deviations, incorporating variation in the derivatization, and mass spectral stability ranged from 0.05 to 0.31. Table S3 gives slopes, linear ranges, and variances for the silylated compounds examined. In addition, the stabilities and recoveries for 22 silylated NACs were determined at 4 different concentrations. For this analysis, extracted spiked glass fiber filters were compared with the nonextracted silylated standards. Results associated with recovery/stability are reported in an accompanying paper where the method is applied to samples generated from chamber irradiations and found in ambient PM_2.5.19

COMPARISON WITH OTHER WORK

The method for measuring NACs presented here follows from earlier studies which have used silylation to improve the chromatography and identify phenols and nitrophenols in complex matrices. This study presents a single step derivatization and analysis approach, giving broad compounds coverage required for a comprehensive analysis of ambient particulate samples. For example, the chromatographic method is designed to detect underivatized compounds, such as nitrotoluene and 2-(2-nitrovinyl) furan, present in the matrix. While the matrix is not unique compared with that in other studies,22, 23 the identification methods presented here may be more robust. For example, work of Schummer et al22 focuses almost entirely on a comparison of the derivatization methods, n-(tert-butyldimethylsilyl)-n-methyltrifluoroacetamide with BSTFA for a broad range of compounds, some of which are important in ambient aerosol. However, this work relies on an array of programmed gas chromatographic methods which hamper large-scale analysis of field samples. This study, however, does provide an excellent comparison of the advantages of one derivatization method over another. Irei et al23 presents a method that uses stable ionization isotope mass ratio spectrometry to identify and quantify a limited number of compound types, alkylphenols and nitrophenols. However, the detection technique is not broadly used in the identification of atmospherically important compounds in aerosol.

The works of Li et al25 and Morville et al21 use techniques which, while similar, are difficult to compare with the present method. The former uses BSTFA in a sea water matrix to analyze alkylphenols, chloro alkylphenols, and bis-phenol-A. The rate of derivatization was found to be strongly influenced by the presence of acetone in the derivatizing solution. The work of Morville et al21 was focused on measurement of phenols and nitrophenols in the gas phase and led to a broad range of reported recoveries, 38 to 107% requiring constant calibrations within the matrix to obtain quantitative results.

The work of Xie et al20 presents a liquid chromatographic (LC) method for analyzing smog chamber aerosol filters. However, this work had a different focus, that is, the measurement of the optical properties of phenols and nitrophenols in ambient aerosol. The nondestructive LC technique also allowed the absorption spectrum of the individual compounds to be determined prior to ionization of the target analytes. However, the method suffered from poorer separation of the target compounds of the complex mixture compared with the GC methods given here. That said, the LC and GC methods must be considered complementary methods depending on the chemical and physical properties sought for the analyte compounds. Table S4 shows a comparison of these methods.

CONCLUSIONS

In this study, the identification and quantification of derivatization by silylation previously reported from this laboratory16 were extended to new class of compounds, NACs. The technique is based on derivatizing both the hydroxyl and carboxylic groups simultaneously, while conserving the specific fragmentation pattern associated with aromatic ring (Rule M⁺_(−15,1,29)) and nitro groups Rule M⁺_{(−60,−45,−30)}). Electron ionization mass spectra show strong ions corresponding to [M⁺], or [M⁺ − 15], and by weak fragment ions. Chemical ionization mass spectra exhibited strong ions for [M⁺ + 1], [M⁺ − 15], and [M⁺ + 29] and other weak ions and adducts. Both EI and CI mass spectra exhibit ions specific to nitro groups for [M⁺ − 31], [M⁺ − 45], and [M⁺ − 60]. Strong ions observed for [M⁺] (EI), [M⁺ − 15] (EI/CI), or [M⁺ + 1] (CI) are consistent with a high charge stability associated with aromatics. Spectra associated with compounds containing none or 2 nitro groups show fragments and adducts at even ion masses, and those with odd numbers of nitro groups show ions at odd masses. In general, mass spectra associated with NACs exhibited 3 distinct regions where characteristic fragmentation with specific pattern associated with (Region 1) –OH and/or –COOH groups, (Region 2) –NO₂ groups, and (Region 3) aromatic rings. The combination of EI and CI ionization as well as MS/MS capabilities shows that critical information is gained giving compositional and structural information for unknown NACs.19 The mass spectra associated with NACs show interesting details, including strong ions in CI corresponding to [M⁺ − 15], [M⁺ + 1], [M⁺ + 29] and reasonably abundant ions for [M⁺ − 60], [M⁺ − 45], and [M⁺ − 31], typically not appearing in spectra associated with other POCs studied in our laboratory.16 The findings from this study supplement compound identification for optical measurements and those associated with biomass burning using a reliable GC-MS method. These findings have been confirmed with applications to both chamber aerosol from aromatic oxidations and that found in ambient PM_2.5.19 This technique also offers the opportunity to analyze classes of compounds found in PM_2.5 by using a single filter extraction protocol.