Introduction

Chemical structure determines the behavior and function of a compound, which in turn informs use and processing. A single stage of high-resolution and high mass accuracy mass spectrometry [1,2,3,4,5,6] enables confident characterization of the elemental compositions present (with uncertainties) [7], but not the specific structures. The confidence in these assignments is finite and varies based on instrument type, experimental approach, and individual sample [7]. Consequently, tandem mass spectrometry (MS/MS) [8] is subsequently employed to isolate a single component of the sample, which is then activated and fragmented into ideally diagnostic charged pieces which are detected. Based on the mass-to-charge ratios (m/z) of the detected charged fragments, the precursor ion, and any other known chemical information or evidence, structural assignments are inferred [7, 9].

The simplest means of interpretation of tandem mass spectra is direct comparison with spectra of known standards collected under the same experimental conditions [10, 11]. For many important classes of chemicals, libraries are currently either unavailable or are limited by the impracticality of synthesizing the enormous number of possibilities necessary. Consequently, other algorithmic approaches have been developed as alternative methods to identify some of these chemicals [12,13,14,15,16,17]. How effective these approaches are is a function of the general quality of the models utilized, which in turn is a function of the level of (sometimes indirect) knowledge of what the dissociation chemistries at play produce and the amount and breadth of data available for comparison [7]. Consequently, our ability to confidently identify a particular analyte is inherently biased by what has gone before; meaning that as a field, mass spectrometric structural identification methods are generally much more effective at identifying materials in areas in which substantial experimental and informatics work has previously occurred.

For example, in areas such as the study of protonated species such as oils/petroleum [2, 3, 18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33], weathered [34,35,36,37], or partially decayed organic materials [38,39,40,41], or even synthetic degradation of large polyaromatic hydrocarbon materials [42], tandem mass spectral libraries are limited or non-existent and there are far too many potential chemicals for widespread synthesis of standards to be practical. Researchers would benefit from additional, complementary algorithmic and experimental methods to aid compound identifications. At present, much of the literature on fundamentals is concerned with IR spectroscopy and/or statistical modeling of radical cation analytes [43,44,45,46,47,48,49,50,51], or fixed charge “thermometer” ions [52,53,54,55,56,57,58,59,60,61,62], rather than protonated ones. However, recent work from the Vala Group examined protonated 1,2,3,4-tetrahydronapthalene with detailed electronic structure calculations of the many isomerization pathways on the way to fragmentation of this cation [63]. The authors spectroscopically characterized the precursor ions providing evidence for population of two sites of protonation. Additional theoretical evidence of feasible formation of a benzylium ion and the classically invoked tropylium C7H7+ structure via a series of 1,2-H-shifts was provided from density functional calculations. Differential mobility data from the Campbell and Hopkins groups [64] provided evidence of differing tautomeric populations of protonated aniline as a function of conditions. These populations in turn produced differing abundances for fragment ions. Tandem mass spectrometry of larger protonated or radical cation analytes which include one or more alkyl substituents is commonly applied [18, 19, 32, 34, 46, 47, 65, 66]. Typically, these studies are concerned with identifying broad information on complex samples rather than specifics on individual structures. Consequently, many of these spectra are from mixtures of precursor ions, so the provenance of “individual” fragment peaks is obscured. A better understanding of which protonated precursors generated which fragments and the energetic dependence of these processes would lead to more detailed and thus effective characterization.

One approach to improving our understanding of a chemical class is systematic generation of model compounds which can then be analyzed with high-resolution tandem mass spectrometry experiments. Simultaneously, electronic structure calculations can be utilized to help elucidate the key diagnostic fragmentation chemistries at play and provide evidence in support of fragment identities [65, 66]. The present article describes our targeted approach concerned with a class of derivatized polyaromatic hydrocarbon analytes with the formula CcHhN1 (Scheme 1) analyzed by electrospray ionization MS/MS and theory. Anthracene in addition to being a crude oil component [2, 3, 18,19,20,21,22,23,24,25,26,27,28,29,30,31, 33] has been implicated in fields as wide ranging as environmental chemistry [67,68,69,70] and astrochemistry [43, 44, 71]. Here, we investigate protonated imine anthracene derivatives. The formation of protonated imines in the gas-phase has been studied as a mechanistic tool [72, 73], and as part of peptide fragmentation chemistry (an ion chemistry) with theory and multiple types of experimental approach [74,75,76,77,78,79,80,81,82]. Such analyses of protonated imine precursors have yet to be performed. Here, we address this gap by systematically altering the substituent’s (Scheme 1) chain length and degree of branching to gain insight into the affect this has on the gas-phase ions in terms of stability, dissociation mechanism, and energetics. This knowledge of leaving-group effects [83, 84] provides a direct means of chemical classification for this class of analyte which complements earlier work on related systems [19, 32, 46, 47, 65, 66, 85]. The present, initial data are part of a much wider study; the results of which will be communicated in due course.

Scheme 1
scheme 1

Generic protonated imine model compound, [CcHhN1+H]+, investigated in this study. R1 is a linear (C4H9, C5H11, C6H13, and C7H15) or branched alkyl group (cyclohexane or tertiary-butyl)

Full size image

Experimental Methods

The acetonitrile, 9-anthracenecarboxaldehyde, amines, formic acid, and deuterated methanol (CH3OD) were purchased from Sigma-Aldrich Chemical Company (St. Louis, USA). The anthracene derivatives (Scheme 1) were synthesized based on published experimental procedures [86, 87]; details of which are provided in the supporting information.

Tandem mass spectrometric work was carried out using a MaXis plus quadrupole (hQh) electrospray time-of-flight mass spectrometer (Bruker, Billerica, MA). The instrument configuration has a hexapole then quadrupole followed by an enclosed hexapole collision cell pressurized with dry nitrogen (~ 10−2 mbar). Conditions are such that analyte ions experience multiple collisions in each experiment. MS/MS spectra were obtained by quadrupole isolation of the precursor ion followed by collision-induced-dissociation (CID) in the collision cell, then product ion dispersion by the time-of-flight analyzer. For further analysis, pseudo-MS3 experiments were performed for specific fragment ions (m/z 206, and m/z 179). In this approach, the desired ions are generated in the source by adjusting the potential difference between the two ion funnels located at the front of the instrument. The fragment ions are then isolated in the quadrupole for CID followed by mass analysis.

Ionization was by electrospray with the samples infused into the instrument in ~ 5 μM acetonitrile (100/0.1% formic acid) solutions at a flow rate of 3 μl min−1. For comparison with the protonated analyte, [M+H]+, data, [M+D]+ cations were generated by diluting each analyte in acetonitrile/CH3OD (50/50%) to a final concentration of ~ 5 μM, prior to electrospray ionization. Data were collected as a function of collision energy. Breakdown graphs expressing the relative fragment ion signals as a function of collision energy were obtained for all protonated analytes. Sixty spectra were averaged for each data point. Nitrogen was used as nebulizing, drying, and collision gas.

Theoretical Methods

Simulations were performed using density functional theory. Geometry optimizations of multiple candidate conformations were performed with the Gaussian 09 software [88] package culminating in calculations at the M06-2X/6-311G(2d,2p) [89] level of theory. The minima, transition structures, and separated products of these analytes were characterized. Initial explorative investigations were performed at the M06-2X/6-31G(d) level of theory. Multiple transition structures (TSs) were calculated for each potential fragmentation pathway. Minima and TSs were tested by vibrational analysis (all real frequencies or 1 imaginary frequency, respectively). The potential energy surface generated combined the zero-point energy (ZPE) correction to the electronic energy (Eel, 0 K) for improved accuracy (ΔEel+ZPE,0K). The related, standard enthalpy (ΔH298K), Gibbs free energy (ΔG298K), and entropy (ΔS298K) corrections to 298 K were also determined. The reaction pathway through each TS was determined by intrinsic reaction coordinate (IRC) calculations with up to eight steps in each direction. The terminating points of these calculations (one on product-side, one on reactant-side) were then optimized at the same level of theory to determine which minima were connected to each TS.

Results and Discussion

Scheme 1 shows the family of anthracene-derivatized imine analytes investigated here. We systematically varied the alkyl chain length to enable an assessment of substituent size effects on the protonated imine analytes MS/MS spectra. We then addressed whether the degree of branching in the alkyl substituent has a noticeable effect on the MS/MS spectra with secondary and tertiary alkyl substituents.

Tandem Mass Spectra

All precursor ions with linear alkyl substituents initially produced a substantial peak at m/z 206 [C15H12N]+ (Figure 1) that corresponds to loss of an alkene with concomitant transfer of a proton. We found that the series of precursor ions with linear alkyl substituents all behaved similarly, producing onsets of fragmentation at ~ 15 eV (laboratory collision energy). Comparisons made at the collision energies necessary to achieve 50% dissociation of the precursor peak (E50%, Table 1) suggest a small increase in E50% with alkyl chain length. However, crude normalization of this value for the systematic change in the (3*number of atoms-6) degrees of freedom (E50%/DOF) produced similar values with alkyl chain length. At higher collision energies, the following general observations can be made: (1) consecutive loss of 27 u (HCN or CNH) generates the abundant peak at m/z 179, [C14H11]+; (2) radical cations are increasingly prevalent (m/z 205 and 178) resulting from either consecutive loss of hydrogen radicals or higher energy, radical-based fragmentations of the precursor ion; (3) production of more highly conjugated species occurs via additional losses of hydrogen radicals or hydrogen molecules and (4) the extent of the latter two radical-based processes was reduced as the alkyl side chain length increased (data not shown).

Figure 1
figure 1

Example of MS/MS spectra of C4H9 alkyl chain subtituent, [C19H20N]+, m/z 262, at different laboratory collision energies, (a) 15 eV, (b) 25 eV, (c) 30 eV, respectively

Full size image
Table 1 Experimental and Theoretical Thresholds of the Precursor Ions (Anthracene Ring Substituents Listed in This Study)
Full size table

The branched precursor ions (cyclohexyl and tertiary-butyl) are substantially more labile than the linear alkyl imine substituted precursor ions (Figures 2 and 3; Table 1). The trend in ease of fragmentation is thus linear < secondary < tertiary alkyl substituent (Table 1; Figures 1, 2, and 3), i.e., the opposite of the E50% values. The abundant alkene loss from the imine nitrogen substituent followed by consecutive loss of 27 u is consistent with the preceding linear congeners. However, the degree of radical-based direct and/or consecutive processes are greatly reduced for the branched precursor ions. These results provide a potential means of distinguishing compounds that contain branched alkyl chain substituents from those with linear ones.

Figure 2
figure 2

Example of MS/MS spectra of cyclohexane substituent [C21H22N]+, m/z 288, at different laboratory collision energies, (a) 15 eV, (b) 25 eV, (c) 30 eV, respectively

Full size image
Figure 3
figure 3

Example of MS/MS spectra of tertiary butyl [C19H20N]+, m/z 262, at different laboratory collision energies, (a) 15 eV, (b) 25 eV, (c) 30 eV, respectively

Full size image

We also performed parallel experiments with deuterated precursor ions, [M+D]+. These data (Figures S1–S3) show a systematic shift of 1 u in both the precursor ions and the major dissociation pathway product peaks (m/z 206 → m/z 207 and m/z 179 → m/z 180). Consequently, these products all contain a single deuterium which any subsequent mechanistic proposals will need to account for. This simple labeling experiment shows (1) that the initial loss of alkene is solely from the substituent R1 and (2) that the subsequent loss of 27 u does not include the ionizing proton (deuteron). Additionally, these analytes showed highly similar dependencies on collision energy to the protonated congeners.

Pseudo-MS3 Mass Spectra

We took pseudo-MS3 mass spectra from each of the analytes to provide an assessment of whether our consecutive fragmentation and common fragment hypotheses were plausible. Following dissociation of each protonated precursor in the source, we successively isolated the m/z 206 and m/z 179 peaks and collisionally activated them. The resulting pseudo-MS3 spectra of the m/z 206 population primarily produced the m/z 179 and 178 peaks in extremely similar abundances in all cases (Figure S4). Consistent with this finding, the pseudo-MS3 spectra of the m/z 179 population overwhelmingly produced the m/z 178 peak (loss of H•) followed by a low abundance m/z 152 peak (subsequent loss of C2H3·) at higher collision energies (Figure S5). Again, the data were extremely similar for each of the six analytes.

Mechanisms of Loss of the Alkyl Substituents

We performed multiple series of calculations in order to identify the precursor ion minima and major fragmentation pathways of each of the analytes. Experimentally, the most facile degradation pathway was loss of the alkyl substituent as an alkene to produce the anthracen-9-ylmethaniminium peak at m/z 206. For the linear alkyl chains, this reaction could in principle proceed directly (Scheme 2a) or via a more complex process of isomerization, followed by dissociation (Scheme 2b). Our electronic structure calculations indicate that a small energetic preference (8–14 kJ mol−1) exists for the more complex process in which the rate-determining step is isomerization of the linear alkyl side chain to form a branched intermediate (Figure 4; Figures S6 and S7; Tables 2, S1–S3). This mechanism is consistent with the deuterated, [M+D]+, spectra (Figure S1; Scheme S1). Additionally, we performed RRKM calculations to provide a better understanding of the relative competitiveness of these two pathways as a function of energy and time (Figure S8). The RRKM calculations indicate that the rate-determining step is always isomerization, prior to dissociation, and that the larger alkyl chains generally had lower rate constants than did the shorter ones. This result is much more consistent with our E50% values than the degree of freedom scaled ones.

Scheme 2
scheme 2

Lowest energy mechanisms of but-1-ene loss illustrated for the analyte with R1 = C4H9. (a) Direct C–N bond cleavage of the alkyl chain to form C15H12N+, m/z 206; (b) isomerization of the alkyl chain, followed by C–N bond cleavage. The relative energies (ΔEel+ZPE,0K (ΔG298K) in kJ mol−1) of the transition structures calculated at the M06-2X/6-311G(2d,2p) level of theory are provided for illustration

Full size image
Figure 4
figure 4

Minimum energy reaction pathway of the competing alkene loss pathways (but-1-ene loss) from the C4H9 alkyl chain substituent [C19H20N]+. (a) Lowest energy pathway of alkene loss in which the alkyl chain is isomerized to a branched butyl substituent, followed by C–N bond cleavage and the product ion of the anthracen-9-ylmethaniminium product ion. (b) Competing, higher energy, direct loss of but-1-ene to form the same product ion. Values of the rate-determining TSs are provided for clarity (kJ mol−1).

Full size image
Table 2 Relative Energies of the Minima, Transition Structures, and Separated Products of Alkyl Chain C4H9, Substituent Calculated at the M06-2X/6-311G(2d,2p) Level of Theory
Full size table

This rearrangement process involved both a 1,2-H-shift and a 1,2-N-shift, and required at least 263–270 kJ mol−1 to initiate (Scheme 2; Figure 4a; Figure S6 and S7; Tables 2, S1–S3), depending on alkyl chain length. Dissociation of the branched intermediate formed occurred by a concerted mechanism where the C–N bond cleavage and proton abstraction accompany alkene formation (227–235 kJ mol−1). The competing, more direct process (Scheme 2a; Figure 4b, Figure S9) essentially involved this same reaction type but produced higher energy transition structures (277–279 kJ mol−1) due to the reduced charge stabilization of the transition structure available from the linear substituents. This discovery prompted the logical examination of branched systems to test whether this lowering of dissociation threshold in branched systems was general.

The branched analytes have no need to isomerize prior to cleaving the alkyl side chain. Consistent with this and the experimental data, our calculations predict that less energy was necessary to initiate fragmentation of the cyclohexyl and tertiary-butyl forms than the analytes with linear alkyl chains. Generation and expulsion of cyclohexene and 2-methylprop-1-ene, respectively, from these systems occurs via similar mechanisms (Scheme 3). The rate-limiting cyclohexene expulsion transition structure required at least 234 kJ mol−1 (Figure 5a, Figure S10; Table 3) whereas the 2-methylprop-1-ene expulsion required at least 199 kJ mol−1 to initiate (Figure 5b, Figure S11; Table 4). These barriers are distinct from the linear forms and each other reflecting the systematic stabilization of the rate-limiting transition structures in the increasingly branched analyte ions. The deuterated data (Figures S2 and S3; Scheme S2) and RRKM calculations (Figure S12) are consistent with these mechanisms too.

Scheme 3
scheme 3

(a) Mechanism of loss of cyclohexene illustrated for the analyte with R1 = cyclohexane, C6H11. Direct C–N bond cleavage produces C15H12N+, m/z 206. (b) Mechanism of loss of 2-methylprop-1-ene by direct C–N bond cleavage to produce C15H12N+, m/z 206. The relative energy (ΔEel+ZPE,0K (ΔG298K) in kJ mol−1) of the transition structure calculated at the M06-2X/6-311G(2d,2p) level of theory is provided for illustration

Full size image
Figure 5
figure 5

Minimum energy reaction pathway plots for loss of the respective alkenes from the branched substituents. (a) Cyclohexene loss from the cyclohexane substituent of [C21H22N]+, R1 = C6H11. (b) Loss of 2-methylprop-1-ene from the tertiary-butyl substituent of [C19H20N]+, R1 = C4H9. Values of the rate-determining TSs are provided for clarity (kJ mol−1)

Full size image
Table 3 Relative Energies of the Minima, Transition Structures, and Separated Products of Branched C6H11 Substituent (Cyclohexane), [C21H22N]+, Calculated at the M06-2X/6-311G(2d,2p) Level of Theory
Full size table
Table 4 Relative Energies of the Minima, Transition Structures, and Separated Products of Branched C4H9 Substituent (Tertiary-Butyl), [C19H20N]+, Calculated at the M06-2X/6-311G(2d,2p) Level of Theory
Full size table

Mechanisms of Consecutive Loss of 27 u (HCN or NCH Loss)

Each of the model system investigated here generates the anthracen-9-ylmethaniminium ion, [C15H12N]+, detected at m/z 206. This ion expels a 27 u fragment (HCN or CNH) to form the abundant m/z 179, [C14H11]+ peak (Figure S4, Figures 1, 2, and 3). A simple, direct mechanism for this dissociation was not immediately clear. Our calculations (Scheme 4; Figure 6; Tables 2, 3, and 4 and S1–S3) support a HCN loss lowest energy pathway in a multistep process beginning with rate-limiting proton (deuteron) transfer from the protonated imine nitrogen to carbon 9 of the ring. This transfer requires at least 328–338 kJ mol−1 in order to occur (depending on from which precursor the anthracen-9-ylmethaniminium ion was generated) and triggers delocalization of the positive charge over the ring (Figure 6a, Figure S13). Note that a formal positive charge has been drawn on the right-hand side of the anthracene ring in Scheme 4 for the purpose of mechanistic illustration. Overcoming a subsequent, trans-cis rotational barrier (294–304 kJ mol−1) places the remaining imine nitrogen proton in position for concerted transfer to carbon 10 on the anthracene ring and loss of HCN (TSs 294–303 kJ mol−1; Scheme 4; Figure 6a, Figure S13). This mechanism is entirely consistent with the [M+D]+ data (Figures S1–S3; Schemes S1–S3).

Scheme 4
scheme 4

Schematic representation of the mechanism of HCN loss from anthracen-9-ylmethaniminium, forming [C14H11]+, m/z 179. This reaction requires trans-cis rotation of the imine proton to facilitate HCN loss. Note that the calculations predict that the charge is delocalized in the final product.

Full size image
Figure 6
figure 6

Minimum energy reaction pathway plots for anthracen-9-ylmethaniminium ion degradation. (a) Loss of HCN to form [C14H11]+, m/z 179. (b) Loss of CNH to form an alternate [C14H11]+ isomer, m/z 179. The red line indicates where the two pathways differ. Values of the TSs and separated products are provided for clarity (kJ mol−1)

Full size image

The alternate loss of the higher energy isomer CNH (55 kJ mol−1) was also investigated. This process (Figure 6b) begins identically to the mechanism of loss of HCN and thus shares several structures (blue line, Figure 6b). Subsequent cleavage (red line, Figure 6b) of the imine to anthracene C–C bond in a twisting motion forms a proton bound dimer in which the formerly Cα hydrogen of the imine group is left pointing at C1 of the anthracene ring. Abstraction of this proton produces another dimer which then dissociates to yield anthracene protonated at C1 and CNH. This entire pathway is limited by the initial proton (deuteron) transfer transition structure to the ring, just like the HCN loss pathway so both product types are possible. This initial proton transfer is highly energetically demanding and thus rate-limiting as it removes the conjugation from the central anthracene ring. Again, the mechanism is entirely consistent with the [M+D]+ data (Figures S1–S3; Schemes S1–S3).

Other Fragmentation Processes

At higher collision energies, radical cations were detected from the precursor ions with linear alkyl substituents, and to a much smaller degree from the precursor ions with branched substituents. Consistent with experiment, where available, direct formation of these ions is predicted to be enthalpically unfavorable (≥ 354 kJ mol−1; Tables 2, 3, and 4 and S1–S3) compared with the closed-shell alkene losses (≤ 270 kJ mol−1). These pathways have higher but more similar thresholds to the pathways for closed-shell, consecutive loss of HCN or CNH, but are massively entropically disfavored (ΔS298 K = 17–29 J K−1 mol−1 versus > 170 J K−1 mol−1) as they are competing with consecutive processes. Alternate, consecutive formation processes to generate radical cation species are significantly enthalpically unfavorable too (> 500 kJ mol−1). Consequently, none of these processes were investigated further.

Conclusion

Our combined experimental and computational evidence indicates that the degree of branching of alkyl substituents is the key determining factor in their relative ease of dissociation (linear < secondary < tertiary). The lowest energy degradation pathways were common across compounds and involved loss of the branched alkyl substituent as an alkene. Linear substituents preferentially isomerized to branched forms prior to this loss. The chemistry underlying these related processes is dominated by the ability to stabilize the rate-determining transition structures, a task more effectively undertaken by branched substituents. The m/z 206 anthracen-9-ylmethaniminium ion generated from these alkene loss reactions undergoes rate-limiting proton transfer prior to expulsion of hydrogen cyanide or CNH. Proposed mechanisms were consistent with the deuterated analyte and pseudo-MS3 experimental findings. The combination of the differences in primary fragmentation thresholds and degree of radical-based fragmentation processes provide a potential means of distinguishing compounds that contain branched alkyl chain substituents from those with linear ones.

In subsequent work, we will expand these investigations to encompass a wider suite of analytes with a more diverse range of substituents and also modes of ionization. This will enable a broader understanding of CcHhN1 series including other functional groups (e.g., acridinic and pyrrolic nitrogen derivatives, and other amines), isomeric species, and the variation in their tandem mass spectra and fragmentation energetics.