Investigation of the ionisation and fragmentation ...

RAPID COMMUNICATIONS IN MASS SPECTROMETRY Rapid Commun. Mass Spectrom. 2006; 20: 2293–2302 Published online in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/rcm.2591

Investigation of the ionisation and fragmentation behaviour of different nitroaromatic compounds occurring as polar metabolites of explosives using electrospray ionisation tandem mass spectrometry Anne-Christine Schmidt*,y, Rainer Herzschuh, Frank-Michael Matysik and Werner Engewald University of Leipzig, Faculty of Chemistry and Mineralogy, Institute of Analytical Chemistry, 04103 Leipzig, Germany Received 21 March 2006; Revised 29 May 2006; Accepted 29 May 2006

In order to develop a liquid chromatography/electrospray ionisation tandem mass spectrometry (LC/ ESI-MS/MS) method for identification and quantification of polar metabolites of explosives using a triple quadrupole system, the mass spectrometric ionisation and fragmentation behaviour of different nitrophenols, nitro- and aminonitrobenzoic acids, nitrotoluenesulfonic acids, and aminonitrotoluenes was investigated. Due to their different molecular structures, the substances concerned showed a very different ionisation efficiency in the ESI process. Interestingly, 2,4-dinitrobenzoic acid yielded no mass signals in the Q1 scan suggesting a thermal decarboxylation in the ion source, whereas the corresponding 3,5-isomer showed a high ionisation yield. Using negative ionisation polarity, carboxylic, phenolic, and sulfonic acid groups were deprotonated resulting in molecular anions, which could be fragmented in a collision cell. A pronounced dependency of the produced fragment ion series on the kind and position of substituents at the nitrobenzene ring (ortho effects) was observed and exploited for the development of substance-specific detection methods in the multiple reaction monitoring mode. In case of benzoic and sulfonic acids, decarboxylation and desulfonation, respectively, were observed as the most frequent fragmentation reactions. Furthermore, besides loss of NO2, NO fragmentation occurred and preceded a decarbonylation of the benzene ring. The expulsion of the open-shell molecules NO and NO2 led to a variety of distonic radical anions. Copyright # 2006 John Wiley & Sons, Ltd. Even 60 years after the Second World War, in the surroundings of former ammunition plants significant problems can arise due to environmental pollutions caused by explosives.1 A variety of degradation and conversion products results from biological2,3 and chemical metabolisation4 of the original compounds, as well as from technical byproducts of the production of explosives. In many studies, the toxicological effects have been ascertained for explosives and their metabolites.5–8 There is still a considerable interest in the determination of highly water-soluble acidic compounds, because a number of new polar substances have been found at explosivescontaminated sites.9,10 However, the analysis of environmental samples from contaminated areas of former ammunition plants is still challenging. A lot of different substances can occur in a wide concentration range. For toxic, highly *Correspondence to: A.-C. Schmidt, Technical University Bergakademie Freiberg, Faculty of Chemistry and Physics, Institute of Analytical Chemistry, Leipziger Str. 29, 09596 Freiberg, Germany. E-mail: [email protected] y Present address: Technical University Bergakademie Freiberg, Faculty of Chemistry and Physics, Institute of Analytical Chemistry, Leipziger Str. 29, 09596 Freiberg Germany.

polar metabolites of explosives, mass spectrometry (MS) especially when coupled to liquid chromatography (LC) is regarded as a suitable analytical method for identification as well as for quantification.11–16 Because numerous isomers of explosives-related compounds occur in environmental samples,13,17,18 a MS differentiation of the analytes on the basis of their molecular ions is not possible even in the case of LC coupling due to non-resolved co-elutions.19,20 Even though single quadrupole instruments produce fragment spectra due to in-source collision-induced dissociation (CID), co-eluting analytes and matrix components form fragments leading to mixed mass spectra.21 Therefore, in the present study it was intended to investigate and to summarise the ionisation and fragmentation behaviour of important metabolites and technical by-products of common explosives (dinitrotoluenesulfonic acids, aminonitrobenzoic acids, nitrobenzoic acids, aminonitrotoluenes, and nitrophenols) under identical experimental conditions using electrospray ionisation tandem mass spectrometry (ESI-MS/MS). The knowledge of characteristic fragment ions produced during

Copyright # 2006 John Wiley & Sons, Ltd.

2294 A.-C. Schmidt et al.

CID of selected precursor ions was considered as a prerequisite for MS/MS method development. The fragmentation behaviour of the different substances is discussed in terms of the exploitation for isomer-specific detection using highly structure-specific fragmentation experiments.

EXPERIMENTAL The explosives-related substances (itemised in Table 1) were either purchased from Promochem (Wesel, Germany), Merck (Darmstadt, Germany), Aldrich (Milwaukee, WI, USA), Fluka (Buchs, Switzerland), or kindly donated by Dr. K. Steinbach, University of Marburg, Germany. Standard solutions of the explosives-related compounds were prepared by weighing the solid substances to 1 mg mL1 in acetonitrile (ACN) and stored in glass flasks at 48C. Test and calibration solutions for MS were prepared by diluting the stock solutions with ACN/H2O mixtures (1:1) and with different ACN/H2O mixtures buffered at pH 4.0 with 10 mM ammonium acetate/acetic acid buffer (20:70:10, 50:40:10 – ACN/H2O/buffer) which represent typical LC eluent compositions. A triple quadrupole mass spectrometer (API 2000, ABI, MDS Sciex, Canada) equipped with an electrospray ionisation interface (turbo ion spray) was used. For optimisation of the ionisation conditions and of fragment ion spectra analyte concentrations of 0.1 to 5 mg mL1 depending on the ion intensities were employed. For screening of mass signals of the different compounds and to search for diagnostic precursor ions for MS/MS experiments, Q1 scans were performed in the mass range of m/z 100–600. For detection of specific precursor/product ion transitions the triple quadrupole system was operated in the multiple reaction monitoring (MRM) mode. After ionisation in the ion source, the precursor ion is selected in the first quadrupole (Q1), and fragmented in the second quadrupole (Q2), which acts as a collision cell using nitrogen as collision gas. The

third quadrupole (Q3) extracts one of the produced fragment ions, which is than transferred to the detector. The different MRM methods optimised for each individual compound were then merged to one method, in which for each single MRM experiment a dwell time of 50 ms was chosen. Different modes for the injection of the analytes into the mass spectrometer were applied. Firstly, syringe pump injection at a flow rate of 15 mL min1 was used for the preliminary screening of the substances regarding their ion formation and resulting mass signals. Further, this lower flow mode with a continuous analyte introduction was used for optimisation of the MS/MS fragmentation conditions. Secondly, the flow injection mode was used for optimisation of the flow rate for LC coupling and for optimisation of the corresponding ion spray conditions. Because of the flow rate dependency of the ESI process, for each injection mode the ion source specific parameters were readjusted. The relevant parameters optimised for a flow injection rate of 100 to 400 mL min1 were as follows: curtain gas, 2.068 bar; collision gas, 2 units; ion spray voltage, 5.5 kV; temperature, 4508C; ion source gas 1 and 2, 2.068 and 0.483 bar, respectively. In addition to the ion source specific parameters the compound-specific parameters such as declustering potential, focusing potential, and entrance potential, which are important for MS/MS fragmentation experiments and for detection in the MRM mode, were optimised for each compound separately. For LC coupling, a HP 1100 LC system (Agilent Technologies, USA) and a reversed-phase (RP) separation column (SunFire C18, 3.5 mm, 2.1 50 mm; Waters, USA) were used. A gradient elution program was applied using H2O as eluent A and ACN as eluent B. A constant buffer concentration of 10% 0.01 M ammonium acetate, pH 4.0, was added to both eluents. The gradient program is described in the following: 5 min, 90% A and 10% B isocratic; 15 min, linear increase to 100% B; 10 min, constant 100% B; 2 min, linear decrease to 10% B; and finally 5 min, reequilibration of the initial conditions 90% A and 10% B. A sample volume of

Table 1. Compound-specific precursor (Q1)/product ion (Q3) pairs selected for multiple reaction monitoring (MRM) detection of 14 explosives-related compounds and collision energies (Q2) applied for fragment ion generation using a triple quadrupole system with ESI in negative polarity. Mass signals were optimised using a concentration of 2 mg mL1 each in 50% ACN/40% H2O/10% 0.01 M ammonium acetate buffer (v/v), pH 4.0. CE: collision energy Compound 2-methyl-3-nitrophenol (2-M-3-NP) 4-nitrobenzoic acid (4-NBA) 2-amino-4-nitrobenzoic acid (2-A-4-NBA) 2-amino-4,6-dinitrobenzoic acid (2-A-4,6-DNBA) 3,5-dinitroaniline (3,5-DNA) 3,5-dinitrophenol (3,5-DNP) 2,4-dinitrophenol (2,4-DNP) 4-amino-2,6-dinitrotoluene (4-A-2,6-DNT) 2-amino-4,6-dinitrotoluene (2-A-4,6-DNT) 3,5-dinitrobenzoic acid (3,5-DNBA) 2,4,6-trinitrophenol (2,4,6-TNP) 2,4-dinitrotoluenesulfonic acid-3 (2,4-DNTSA-3) 2,4-dinitrotoluenesulfonic acid-5 (2,4-DNTSA-5) 2,4,6-trinitrobenzoic acid (2,4,6-TNBA)

Q1-mass m/z 152 166 181 182 182 183 183 196 196 211 228 261 261 292

Q3-mass m/z 122 122 137 94 94 95 109 119 136 167 63 123 181 212

Precursor/product ion

CE [V]

[M–H] /[M–H–NO] [M–H] /[M–H–CO2] [M–H]/[M–H–CO2] [M–COOH]/[M–COOH–NO–CO–NO] [M–H]/[M–H–NO–CO–NO] [M–H]/[M–H–2NO–CO] [M–H]/[M–H–NO2–CO] [M–H]/[M–H–OH–2NO] [M–H]/[M–H–2NO] [M–H]/[M–H–CO2] [M–H]/[M–H–NO2] [M–H]/[M–H–SO3–NO–CO] [M–H]/[M–H–SO3] unknown/[M–COOH]

16.0 20.0 16.0 24.0 24.0 30.0 32.0 20.0 20.0 12.0 36.0 34.0 34.0 20.0

All masses are derived from monovalent ions.


Rapid Commun. Mass Spectrom. 2006; 20: 2293–2302 DOI: 10.1002/rcm

Study of polar metabolites of explosives by ESI-MS/MS

10 mL was injected. The flow rate of the mobile phase was optimised by means of flow injection measurements with a typical eluent composition (50% ACN/40% H2O/10% 0.01 M ammonium acetate buffer, pH 4, v/v).

RESULTS AND DISCUSSION Effect of the LC conditions on the MS detection Because the main focus of the presented studies was directed to highly polar metabolites of explosives (nitrophenols, nitrobenzoic acids, nitrotoluenesulfonic acids), electrospray ionisation (ESI) was chosen for ion generation. Regarding the LC separation the influence of the solvent composition on the intensity of detectable ions was studied. The ratio between the aqueous and the organic eluent was varied and the effect of adding a volatile buffer (ammonium acetate, pH 4.0) was tested. An increasing ratio of water-to-organic medium led to a lower ion yield and therefore to a diminished detection sensitivity. In most cases the addition of the ammonium acetate buffer had no influence on the ion yield. The addition of the buffer was necessary in order to suppress the dissociation of the concerned acidic analytes facilitating the RP-HPLC separation. A clear decrease in the peak intensities of the ion signals with increasing flow rate was observed (Fig. 1). A further reduction of the flow rate to 50 mL min1 improved the sensitivity of the MS detection, but resulted in a significant peak broadening. Consequently, 100 mL min1 was regarded as the optimum flow rate for the LC eluent.

ESI behaviour of different explosives-related compounds Initially, the compounds of interest were analysed regarding emerging ion signals in the Q1 scan mode in the mass range of m/z 100–600. The ionisation yield of the selected substances was considerably higher using negative ionisation mode in comparison to positive polarity, which can be attributed to the electronegative nitro groups. Positively charged ions were either not formed or showed a very low intensity in the

2295

mass spectra. This is in agreement with other MS investigations and is ascribed to the low proton affinity of nitroaromatic compounds.22 In most cases, in the Q1 spectra deprotonated molecular anions dominated or appeared as exclusive ion signals. In the case of nitrobenzoic acids, the decarboxylation products of the molecular anions were observed in addition to the deprotonated anions. Compared to mononitro- and dinitrobenzoic acids, where the decarboxylated ions had a lower intensity than the corresponding molecular anions (Figs. 2(a) and 2(b)), for amino-substituted dinitrobenzoic acids the decarboxylation products yielded a higher intensity (Fig. 2(c)). Adduct ions formed with the buffer additive were found only in the case of 4-A-2,6-DNT and 3,5-DNA. However, the signal intensities for the buffer anion adducts of one or two analyte molecules are considerably lower than for the deprotonated molecular anions (Fig. 2(d)). The nitramine compounds hexogene and octogene could be detected as adduct ions only. Single molecular ions were not present in the mass spectra. According to Gapeev et al.,23 under ESI and atmospheric pressure chemical ionisation (APCI) conditions with negative polarity, characteristic ions of 1,3,5-trinitrotriazacyclohexane (RDX or hexogene) are generated basically by addition of an anion originating from the sample or from the solvent, e.g. formate, acetate, hydroxyacetate, and chloride. Formations of adducts are often not reproducible, since such adducts are also produced by contamination of the MS system. In order to control the adduct formation and thereby to ensure the reliability of measurements, a defined addition of large amounts of a dedicated ion for a reproducible adduct formation is necessary. Unfortunately, an additional solvent additive has a negative effect on the detection sensitivity of the whole system. In the Q1 scan mode NO2 or NO groups were hardly released from the various nitroaromatics, whereas collisioninduced fragmentations in the course of MS/MS experiments generally led to a loss of NO and NO2 (see below). For sulfonic acids in addition to molecular anions desulfonated

1.4E+07

Peak Area [c ount s ]

1.2E+07 2,4-DNP

1.0E+07

3,5-DNP

8.0E+06

2,4-DNTSA-5

6.0E+06

2-A-4,6-DNBA octogene

4.0E+06 2.0E+06 0.0E+00

100

150

200

Flow Rate [µL min-1] Figure 1. Peak areas of ion signals (ESI-MS, Q1 scan) of selected nitroaromatic substances in dependence on the LC flow rate. LC eluent composition: ACN/H2O/ 0.01 M ammonium acetate buffer (50:40:10, v/v); analyte concentration: 1 mg mL1 for 2,4-DNP and 2,4-DNTSA-5, 2 mg mL1 for 3,5-DNP, 2-A-4,6-DNBA, and octogene. Copyright # 2006 John Wiley & Sons, Ltd.



Figure 2. Selected mass spectra of polar metabolites of explosives obtained in the first quadrupole (Q1) of a triple quadrupole mass spectrometer after ESI: (a) 3,5-dinitrobenzoic acid; (b) 4-nitrobenzoic acid; (c) 2-amino-4,6-dinitrobenzoic acid; and (d) 3,5-dinitroaniline. Concentration: 2 mg mL1 in ACN/H2O/0.01 M ammonium acetate buffer (50:40:10, v/v). ESI conditions: voltage, 5.5 kV; source temperature, 4508C; syringe pump injection, 15 mL min1.

species were recorded with a low intensity. Fragmentations occurring in the Q1 scan spectra originate from fragmentation reactions in the ion source or in the entrance region before the vacuum system. Because of their different chemical structures the investigated compounds possess a very different ionisability in the employed ESI source. The peak areas of the individual substances obtained from measurements in the MRM mode differed considerably (see Fig. 3). A high yield of ion formation was obtained for 3,5-dinitrophenol, 2,4-dinitrophenol, 3,5-dinitrobenzoic acid, and 2,4,6-trinitrobenzoic acid, followed by 2,4-DNTSA-3, 2-A-4,6-DNBA and 3,5-DNA (measured at the same mass transition), 2-M-3-NP, and 2-A4-NBA. For picric acid, 4-nitrobenzoic acid as well as both aminodinitrotoluenes 2-A-4,6-DNT and 4-A-2,6-DNT, sufficient ionisation yields could be attained. The second investigated sulfonic acid isomer, 2,4-DNTSA-5, showed a considerably lower ion yield than 2,4-DNTSA-3. A very interesting result was noticed regarding the comparison of 2,4-DNBA and 3,5-DNBA. Despite varying the ion source conditions, during Q1 sanning only very small mass signals could be obtained for the former compound not exceeding the background signals. It can be assumed that a thermal decarboxylation of 2,4-DNBA occurs in the ion Copyright # 2006 John Wiley & Sons, Ltd.

source producing the corresponding uncharged nitrobenzene (Scheme 1(b)), which yields no ions in the ESI process. The high source temperature of 4508C was necessary for spraying and nebulising of the sample and to evaporate the LC eluent using the turbo ion spray source coupled to LC. The same problem was reported in a previous study concerned with 2,4-DNBA and ESI.24 In the case of 2,4DNBA an ortho effect of the nitro substituent influences the decarboxylation of the benzoic acid. The nitro group located in the ortho position to the carboxyl group draws electrons leading to a destabilised C–COOH bond. In the Q1 spectrum of the second dinitrobenzoic acid isomer, 3,5-DNBA, the deprotonated as well as the decarboxylated molecular ion show high intensities (Fig. 2(a)). Moreover, 3,5-DNBA shows a clear fragment spectrum, which can easily be interpreted (Scheme 1(a) and Fig. 4(a)). A similar thermal decarboxylation was assumed for 2,4,6-TNBA and 4-A-2,6-DNBA (see below). The other benzoic acids studied (e.g. Fig. 2(b)) possess distinct ion signals for decarboxylation products. Such a decarboxylation originates from the negatively charged deprotonated ion and produces a decarboxylated anion yielding a mass signal. In contrast, a thermal decarboxylation would originate from the uncharged molecule. Rapid Commun. Mass Spectrom. 2006; 20: 2293–2302 DOI: 10.1002/rcm


2297

Figure 3. Substance-specific mass spectrometric sensitivity for the ESI of polar explosivesrelated compounds. MRM detection in a triple quadrupole system (for detected masses, see Table 1); substancespecific optimised conditions; substance concentration 100 ng mL1 for all compounds in ACN/ H2O (1:1, v/v). Additionally, some less polar nitroaromatic compounds (1,3-dinitrobenzene, 1,3,5-trinitrobenzene, 2,4-dinitrotoluene, 2,6-dinitrotoluene, 3,5-dinitrotoluene), investigated due to their occurence in contaminated water samples,20 could not be ionised using ESI. These less polar nitroaromatic compounds can hardly be deprotonated because of a lack of acid functions which can form anions. Using APCI, a dissociative or a non-dissociative electron capture for the MS detection of nitrobenzenes and nitrotoluenes can be realised, which is induced by the corona discharge of the high voltage needle25 leading to a strong increase in detector response.16 It should be noted that compared to the nitrotoluenes mentioned above diamino-substituted dinitrotoluenes produce negative ions in the ESI process (Figs. 3 and 4(h)), presumably by deprotonation of the methyl groups.

Fragmentation behaviour of polar explosivesrelated compounds in MS/MS Most of the nitroaromatics studied regarding their fragmentation behaviour in MS/MS experiments show a clear product ion pattern, which is suited for MRM method development. The most frequent fragmentations were losses of NO and NO2 and decarbonylation (Schemes 1 and 2). A release of the open-shell molecules NO and NO2 from a negatively charged precursor ion affords the formation of radical anions. These fragmentation reactions of nitric oxide and nitric dioxide break the even-electron rule. In the resulting distonic structures appearing during fragmentation of nitrobenzoic acids, nitrophenols as well as nitrotoluenesulfonic acids (Schemes 1(a), 1(c), 1(d), 1(e)) a negative charge and an unpaired electron are localised at different sites of the molecule. Kenttamaa and co-workers characterised positively charged distonic ions, the phenyl radical cations which can be formed in the gas phase.26,27 Copyright # 2006 John Wiley & Sons, Ltd.

A distonic didehydrophenoxide biradical anion (m/z 91) is assumed to appear as an intermediate in the fragmentation series of 3,5-DNBA (Scheme 1(a)) and 3,5-DNP (Scheme 1(c)). In addition, substituted dehydrophenoxide ions were assumed for the collision-induced fragmentation of 3,5DNBA and 3,5-DNP (m/z 137, Schemes 1(a) and 1(c)) and for 2,4-DNTSA-3 (m/z 151, Scheme 1(d)). Distonic dehydrophenoxide ions were also described to be produced in gas-phase reactions of benzyne anions with NO228 and as CID fragmentation products of nitrobenzoic acids.29 In the current study the fragmentation of 4-NBA (Fig. 4(b) and Scheme 2(a)), producing the p-nitrophenide ion (m/z 122) and the distonic p-dehydrophenoxide radical anion (m/z 92), has been confirmed. Moreover, Reed et al.29 investigated the reactivity of such dehydrophenoxide ions created by CID of o-, m- and p-nitrobenzoic acid. These radical anions are of special interest because in such structures the s-system of a phenyl radical and the p-system of a phenoxide ion are combined. The loss of NO is associated with a rearrangement of bonds from C–N to C–O as exemplified in Schemes 1(a), 1(c), 1(d), and 1(e). The formation of the new C–O bond precedes a subsequent decarbonylation. Such a fragmentation series occurs in many cases (3,5-DNBA, 2-A-4,6-DNBA, 2,4DNTSA-3, 2-A-4,6-DNT, 2,4-DNP, 3,5-DNP, 3,5-DNA, 2,4,6-TNP, 2-M-3-NP) and ends presumably in the formation of a cyclopentadienyl radical ion that can be mono- or didehydrogenated and substituted with NO2 dependent on the precursor ion (Schemes 1(c) and 1(d)). In the latter case nitric oxide is eliminated subsequently. The dehydrocyclopentadienyl ion is assumed to be formed by decarbonylation of dehydrophenoxides.28 In contrast to nitro functions, a loss of amino groups was not observed because the C–NH2 bond is very stable. An Rapid Commun. Mass Spectrom. 2006; 20: 2293–2302 DOI: 10.1002/rcm


3,5-DNP lacks such a fragmentation sequence because there are no nitro groups in the ortho position to the OH group. The [M–H–NO2] fragment ion of 3,5-DNP is subjected to a second cleavage of NO2, generating the ion signal at m/z 91, which does not appear in the product ion series of 2,4-DNP. The same fragment ion sequence (–NO2/–CO) resulting from the ortho effect as described for 2,4-DNP was also observed for 2,4,6-TNP, but to a lesser extent. In the product ion spectra of 2-M-3-NP only two fragment ions appeared resulting from the loss of NO and the subsequent CO fragmentation (Scheme 2(b)). For aromatic isomers, ortho effects resulting from the position of the substituents at the aromatic ring can influence the MS fragmentation.30–32 Especially in the case of electron impact ionisation (EI) ortho effects of molecular radical

overview of the fragmentation behaviour of nitrobenzoic acids and nitrophenols is given in Scheme 2. For CID of mononitrobenzoates a sequential loss of carbon dioxide and nitric oxide is described.29 The same fragment ion sequence was obtained for all mono- and dinitrobenzoic acids tested including isomers containing amino groups (Scheme 2(a)). The comparison of the product ion spectra of 2,4-DNP and 3,5-DNP (Figs. 4(c) and 4(d)) demonstrates a different fragmentation depending on the position of the substituents at the benzene ring which suggests the development of isomer-specific MRM methods for MS differentiation of isomers of polar explosives-related compounds. In the case of 2,4-DNP an ortho effect accounts for the CO fragmentation after fragmentation of the nitro group in the ortho position.

a

COOH

COO - H+

NO2

O2N

C-N/C-O bonding rearrangement

- CO2 O2N

O2N

NO2 [M-H]carboxylate anion

3,5-DNBA

NO2

O

N

O

NO2 - NO

charged nitrobenzene m/z 167

m/z 211

• m-nitrodehydrophenoxide radical anion, m/z 137

O

NO2 - NO2 •

didehydrophenoxide radical anion, m/z 91

•

O

- CO •

didehydrocyclopentadienyl radical anion, m/z 63

b

COOH NO2

•

H

T

NO2

- CO2

NO2

NO2

2,4-DNBA

uncharged nitrobenzene

c

O

O

- NO2 OH

O

- NO2

•

- H+ O2N

NO2 3,5-DNP

m/z 137

O2N

NO2

- NO

- CO •

NO2

•

•

•

m/z 91

m/z 63

O

[M-H]-

-NO

- CO

m/z 183

NO2

•O m/z 153

•O

NO2 m/z 125

O

•

O•

m/z 95

Scheme 1. MS/MS fragmentation series (triple quadrupole system) of the investigated substance classes: nitrobenzoic acids (3,5-dinitrobenzoic acid (a), 2,4-dinitrobenzoic acid (b)); nitrophenols (3,5-dinitrophenol (c)); nitrotoluenesulfonic acids (2,4-dinitrotoluenesulfonic acid (d)); and aminonitrotoluenes (4-amino-2,6-dinitrotoluene (e)). Copyright # 2006 John Wiley & Sons, Ltd.



d

CH3

CH3 NO2

CH3 NO2

- H+

SO3H

CH3 NO2

- SO2

SO3

NO2

2 C-N/C-O bond rearrangement

O

[M-H]sulfonate anion

CH3 O

- CO

O

NO2

O

m/z 197

m/z 137

NO2

2,4-DNTSA-3

O

- 2 NO

2299

O m/z 109

- CO2

m/z 261

Fragmentation series a) C H2 m/z 65

CH3

CH3 NO2

CH3 NO2

-SO3

O

- NO

SO3

- CO

•

C-N/C-O bond rearrangement

NO2

NO2

NO2

m/z 261

m/z 181

m/z 151

[C6H5NO2] •

-NO

m/z 123

CH3

m/z 93

[C7H5O]

- OH

-

m/z 105

[C6H4] • m/z 76

CH •

CH2 NO2

O2N

NO2

O2N - H+

NH2 4-A-2,6-DNT

-

-NO2

Fragmentation series b)

e

[C6H5O]

NO

O2N

• CH

CH •

- OH

O2N

•

- NO

- NO

O

C-N/C-O bond rearrangement

NH2

NH2

NH2

NH2

[M-H]-

m/z 179

m/z 149

m/z 119

m/z 196

Fragmentation series a)

CH2 NO2

O2N

CH2 - 2 NO

O

O

O

CH2

- CO

2 C-N/C-O bond rearrangement

NH2

NH2

[M-H]-

m/z 136

m/z 196

NH2 m/z 108

Fragmentation series b)

Scheme 1. (Continued)

cations of substituted aromatic compounds were observed in the gas phase.33 However, fragmentation series that are determined by ortho positions of functional groups are scarcely reported for atmospheric pressure ionisation (API) modes like ESI and APCI.34 Similarly to 2,4-DNBA, for 2,4,6-trinitrobenzoic acid a thermal decarboxylation leading to the uncharged trinitrobenzene can be assumed because of the ortho position of two nitro substituents to the carboxy group. However, an unidentified adduct ion dominated the Q1 spectra at m/z 292 corresponding to a mass difference of þ35 to the molecular mass. However, this unknown ion was stable and reproducible, and formed an intensive fragment ion at m/z 212 corresponding to the decarboxylated molecule [M– COOH]. Therefore, this precursor/product ion pair was used for MRM detection of 2,4,6-TNBA. Further fragmentation experiments of 2,4,6-TNBA would be necessary, however, in order to gain a deeper understanding of these Copyright # 2006 John Wiley & Sons, Ltd.

processes. Different solvent modifiers should be tested regarding the formation of adduct ions with 2,4,6-trinitrobenzoic acid. The two dinitrotoluenesulfonic acid isomers examined in the present study show a different fragmentation pattern emanating from the sulfonate anion (Figs. 4(e) and 4(f)). For 2,4-DNTSA-5 the most intensive fragment ions resulted from a SO3 fragmentation, whereas for 2,4-DNTSA-3 both losses of SO2 and SO3 determined the subsequent product ion series (Scheme 1(d)). The elimination of SO2 is accompanied by a rearrangement of the C–S bond to a C–O bond. The intensities of the corresponding mass signals suggest that the fragmentation products of 2,4-DNTSA-5 and 2,4DNTSA-3 resulting from SO3 and SO2 fragmentation differ regarding their stability (Figs. 4(e) and 4(f)). The product ion at m/z 181 originating from desulfonation of 2,4-DNTSA-3 showed a very low intensity, but a pronounced fragment ion at m/z 123 derived from this ion. In contrast, the expulsion of Rapid Commun. Mass Spectrom. 2006; 20: 2293–2302 DOI: 10.1002/rcm


SO2 led to a more stable fragment ion at m/z 197. The analogous ion derived from 2,4-DNTSA-5 had a minimal intensity. For the desulfonation product of 2,4-DNTSA-5 a higher stability is assumed because the corresponding mass signal at m/z 181 achieves a relative intensity of 30% relating to the molecular anion. A MS study of the CID of some benzenesulfonic acids and sulfobenzoic acids in the gas phase has been presented by Ben-Ari et al.35 The phenoxy anion is described as the final fragmentation product of both benzenedisulfonic and sulfobenzoic acids. Characteristic ions resulting from the loss of NO, NO2 and SO2 from the [M–H] ions were previously reported to occur in mass spectra obtained from LC/QTOFMS analysis of different nitrotoluenesulfonic acid isomers.10 The aminodinitrotoluenes show a loss of OH in addition to NO and NO2 fragmentations due to an ortho effect caused by a methyl group in the ortho position to the nitro group (Fig. 4(h) and Scheme 1(e)). 3,5-Dinitroaniline is formed as a decarboxylation product of 2-A-4,6-DNBA. Therefore, the fragmentation patterns from 3,5-DNA and 2-A-4,6-DNBA are very similar (Scheme 2(a) and Fig. 4(g)).

For all product ions proposed in Schemes 1 and 2 the final electronic structure as well as possible mesomeric effects could not be defined unequivocally only on the basis of the information obtained from the mass spectra. Therefore, the suggested electronic structures should be regarded only as formal assignments. In the molecular formulas mono- and biradicals are simply defined as radicals. A MRM detection could not be established for all the compounds studied. In contrast to 2-A-4,6-DNBA, for 4-A2,6-DNBA stable and reproducible fragmentation series could not be attained because possible precursor ions like [M–COOH] or [M–COO] exhibited insufficient intensities. This could be attributed to the same thermal decarboxylation process as observed for 2,4-DNBA caused by an ortho effect of a nitro group to the carboxyl function (see above). For 4-aminobenzoic acid, during fragmentation a decarboxylation was recorded but the intensities were too low for a MRM method to be utilised. The small ionisation efficiency could be ascribed to a betaine structure formed by the COO and the NH3þ group. The corresponding 4-nitrobenzoic acid yielded much higher ion intensities and distinct product spectra (Figs. 2(b), 4(b) and Scheme 2(a)).

Figure 4. Product ion spectra of polar metabolites of explosives obtained in the third quadrupole (Q3) of a triple quadrupole mass spectrometer after collision-induced fragmentation in the second quadrupole (Q2) of a precursor ion selected in the first quadrupole (Q1): (a) 3,5-dinitrobenzoic acid; (b) 4-nitrobenzoic acid; (c) 2,4-dinitrophenol; (d) 3,5-dinitrophenol; (e) 2,4-dinitrotoluenesulfonic acid-3; (f) 2,4-dinitrotoluenesulfonic acid-5; (g) 3,5-dinitroaniline; and (h) 4-amino-2,6-dinitrotoluene. ESI: source temperature, 4508C; voltage, 5.5 kV; collision energies see Table 1. Analyte concentration: 2 mg mL1 in ACN/H2O/0.01 M ammonium acetate buffer (50:40:10, v/v). Copyright # 2006 John Wiley & Sons, Ltd.



2301

Figure 4. (Continued)

Scheme 2. Overview of the fragmentation pathways of different isomers of nitrobenzoic acids (a) and nitrophenols (b) occuring in electrospray ionisation followed by collision-induced fragmentation in a triple quadrupole mass spectrometer. Copyright # 2006 John Wiley & Sons, Ltd.



On the basis of the discussed product ion spectra substance-specific precursor/product ion transitions were chosen in order to elaborate suitable MRM detection methods (Table 1). As precursor ion for MRM detection the most intensive ion from the Q1 scan was chosen. Typically, the most intensive ion was the deprotonated molecular ion. For 2-A-4,6-DNBA the decarboxylation product yielded a higher ion intensity (see above, Fig. 2(c)). For isomers which could possibly interfere in the MRM detection because they can produce the same product ions, standard solutions containing only one of these isomers were measured with the MRM method adapted to the other isomer. Regarding the sulfonic acid isomers, no interference emerged, since 2,4-DNTSA-3, which is detected at the mass transition 261/123, gave no signal at the mass transition 261/ 181 used for 2,4-DNTSA-5, and vice versa. The different fragmentation series of these two sulfonic acids are described above (Figs. 4(e) and 4(f)). At the precursor/product ion pair 183/109 selected for the detection of 2,4-DNP the competing isomer 3,5-DNP did not produce an interfering peak. Although for 2,4-DNP the predominant MRM trace was 183/95, it provided also a small peak at 183/109 chosen for MRM detection of 3,5-DNP. This behaviour is confirmed in the product ion spectra (Figs. 4(c) and 4(d)) showing a peak at m/z 109 only for 2,4-DNP and at m/z 95 a peak for both isomers, but for 2,4-DNP to a considerably lower extent. However, 2,4-DNP and 3,5-DNP were clearly separated by HPLC.20 The LC/ESI-MS/MS method described was successfully applied for identification and quantification of a large number of samples containing explosives-related compounds in a wide concentration range.20

CONCLUSIONS The various highly polar metabolites of explosives are characterised by a different mass spectrometric ionisation and fragmentation behaviour that can be exploited for the development of a LC/MS/MS method for the analysis of complex explosives-contaminated water samples. Mass spectra of highly polar metabolites of explosives obtained using electrospray ionisation and triple quadrupole mass analysis can be elucidated by means of ortho effects to a large extent. Product ion series caused by NO2, NO, CO2, CO, SO3, and SO2 fragmentations were observed in dependence on the position of the substituents on the benzene ring. Many intermediates of these CID sequences are radical anions showing a distonic structure. Thermal decarboxylations of various nitrobenzoic acids in the ion source can play an important role depending on instrumentation and should consequently be studied regarding the influence of the temperature of the ion source during API. Future studies should be directed to additional isomers of metabolites of explosives regarding their ionisation and fragmentation behaviour in order to gain a more complete comparison of isomer series.


Acknowledgements We are indebted to Prof. Dietmar Heidrich, University of Leipzig, Wilhelm Ostwald Institute of Physical and Theoretical Chemistry, for the detailed suggestions concerning the chemical structures of the fragmentation intermediates.

REFERENCES 1. Lewin U, Wennrich L, Efer J, Engewald W. Chromatographia 1997; 45: 91. 2. Schmidt TC, Steinbach K, Buetehorn U. Environmental analysis of nitroaromatics. In Encyclopedia of Analytical Chemistry, Meyers RA (ed.). John Wiley: Chichester, 2000; 2946. 3. Spain JC. Bacterial degradation of nitroaromatic compounds under aerobic conditions. In Biodegradation of Nitroaromatic Compounds, Spain JC (ed.). Plenum Press: New York, 1995. 4. Dillert R, Brandt M, Fornefett I, Siebers U, Bahnemann D. Chemosphere 1995; 30: 2333. 5. Robidoux PY, Hawari J, Bardai G, Paquet L, Ampleman G, Thiboutot S, Sunahara GI. Arch. Environ. Contamin. Toxicol. 2002; 43: 379. 6. Lachance B, Renoux AY, Sarrazin M, Hawari J, Sunahara GI. Chemosphere 2004; 55: 1339. 7. Homma-Takeda S, Hiraku Y, Ohkuma Y, Oikawa S, Murata M, Ogawa K, Iwamuro T, Li S, Sun GF, Kumagai Y, Shimojo N, Kawanishi S. Free Radical. Res. 2002; 36: 555. 8. Steevens JA, Duke B, Maurice L, Guilherme R, Bridges TS. Environ. Toxicol. Chem. 2002; 21: 1475. 9. Preiss A, Elend M, Gerling S, Tranckner S. Magn. Reson. Chem. 2005; 43: 736. 10. Ma WT, Steinbach K, Cai Z. Anal. Bioanal. Chem. 2004; 378: 1828. 11. Nipper M, Qian Y, Scott CR, Miller K. Chemosphere 2004; 56: 519. 12. Crowson A, Beardah MS. Analyst 2001; 126: 1689. 13. Schmidt TC, Buetehorn U, Steinbach K. Anal. Bioanal. Chem. 2004; 378: 926. 14. Groom CA, Halasz A, Paquet L, Olivier L, Dubois C, Hawari J. Environ. Sci. Technol. 2002; 36: 112. 15. Zhao X, Yinon J. J. Chromatogr. A 2002; 946: 125. 16. Cassada DA, Monson SJ, Snow DD, Spalding RF. J. Chromatogr. A 1999; 844: 87. 17. Lewin U, Efer J, Engewald W. J. Chromatogr. A 1997; 730: 161. 18. Lewin-Kretschmar U, Efer J, Engewald W. Liquid chromatography. In Explosives. Encyclopedia of Separation Science, Academic Press: London, 2000. 19. Schmalz L, Tra¨nckner S. Vom Wasser 2004; 102: 7. 20. Schmidt AC, Niehus B, Matysik FM, Engewald W. Chromatographia 2006; 63: 1. 21. Zwiener C, Frimmel FH. Anal. Bioanal. Chem. 2004; 378: 851. 22. Casetta B, Garofolo F. Org. Mass Spectrom. 1994; 29: 517. 23. Gapeev A, Sigman M, Yinon J. Rapid Commun. Mass Spectrom. 2003; 17: 943. 24. Schreiber A. PhD thesis, University of Leipzig, Germany, 2001. 25. Hayen H, Jachmann N, Vogel M, Karst U. Analyst 2002; 127: 1027. 26. Chyall LJ, Kenttamaa HI. J. Mass Spectrom. 1995; 30: 81. 27. Thoen KK, Smith RL, Nousiainen JJ, Nelson ED, Kenttamaa HI. J. Am. Chem. Soc. 1996; 118: 8669. 28. Wenthold PG, Hu J, Squires RR. J. Mass Spectrom. 1998; 33: 796. 29. Reed DR, Hare MC, Fattahi A, Chung G, Gordon MS, Kass SR. J. Am. Chem. Soc. 2003; 125: 4643. 30. Blachut D, Danikiewicz W, Olejnik M, Czarnocki Z. J. Mass Spectrom. 2004; 39: 966. 31. Reddy PN, Srikanth R, Venkateswarlu N, Rao RN, Srinivas R. Rapid Commun. Mass Spectrom. 2005; 19: 72. 32. Weickhardt C, To¨nnies K. Rapid Commun. Mass Spectrom. 2002; 16: 442. 33. Ramana DV, Srinivas R, Mahalakshmi P. Org. Mass Spectrom. 1991; 26: 305. 34. Reemtsma T. J. Chromatogr. A 2001; 919: 289. 35. Ben-Ari J, Etinger A, Weisz A, Mandelbaum A. J. Mass Spectrom. 2005; 40: 1064.
