Ion Composition Elucidation (ICE)
Mass Peak Profiling from Selected Ion Recording Data (MPPSIRD) as a Tool for Regulatory Analyses
Andrew H. Grange and G. Wayne Sovocool
Environmental Sciences Division, NERL, U.S. EPA, PO Box 93478, Las Vegas, NV 89193-3478
Identification of compounds in mixtures of environmental contaminants or synthetic products is essential for regulatory analyses. Exact masses of ions determined using high resolution mass spectrometry provide unique elemental compositions only for low-mass compounds (< 150 g/M). Using Mass Peak Profiling from Selected Ion Recording Data (MPPSIRD) to acquire additional mass spectral data and a Profile Generation Model (PGM) for automated interpretation of the data provide elemental compositions for ions with m/z up to 600 based on incontestible properties of atoms, their exact masses, isotopic abundances, and valences. In this study, MPPSIRD and the PGM were used to identify intended and unintended products resulting from attempted syntheses of two thermolabile, non-ionic, phosphorothioate compounds. The products were volatilized from a probe inserted into a VG70-250SE double focusing mass spectrometer. High mass resolution substituted separation in the mass domain for the temporal separation of most components provided by chromatographic techniques. MPPSIRD and the PGM identified the correct composition for M+ by rejecting all other compositions that were possible within the error limits of the exact mass determinations for M+. MPPSIRD was used with 10,000-24,000 resolution to determine exact masses of ions prominent in mass spectra and to isolate signals from different ions with the same nominal mass. Superposition of volatilization peaks of ions and linked scans (constant B/E) correlated fragment ions with the molecular ion. The PGM determined the compositions of fragment ions using the number of atoms of each element in the molecular ion as limits. Fragmentation schemes based on these ions and the tables of exact masses and relative abundances provided a preponderance of evidence for the product identities.
In the regulatory context, the ability to unequivocally identify compounds is essential in forensics, drug syntheses, food contamination, or environmental contamination cases. For legal proceedings, incontrovertible data documenting the elemental compositions of components in samples would contribute substantially when compiling a preponderance of evidence. In addition, a dependable analytical methodology for regulatory purposes must be capable of detecting trace quantities of compounds from complex sample matrices.
High resolution mass spectrometry (HRMS) is used to measure exact masses of ions. For ions formed from compounds with masses less than 150 Da that contain C, H, N, O, P, or S atoms, a unique elemental composition usually corresponds to the ion's exact mass. However, most compounds of regulatory interest yield molecular ions with masses larger than 150 Da. In addition, other elements must be considered when the origin of a sample is unknown. For synthetic products, the list of elements can be limited to those in the reactants, solvents, catalysts, and atmosphere in the reaction vessel when they are known. For unidentified compounds in contaminated sites, at least 10 elements should be considered initially (1). Food contaminants and illicit drugs or by-products of unknown origin might contain other elements. Generally, the exact mass of the molecular ion formed from these types of compounds cannot provide a unique elemental composition, because the number of compositions possible for an ion increases rapidly for a given error limit with the ion's mass and the number of elements considered (2).
Recently, the Environmental Sciences Division of the U.S. EPA developed powerful tools for regulatory analyses. These tools determine the unique elemental composition of ions with masses up to 600 Da (containing C, H, N, O, P, or S atoms) based on physical properties of atoms (their masses and isotopic abundances) and their valences, without further recourse to chemical arguments or knowledge about the history of a sample. These analytical tools have been used to characterize a Superfund site (3) and to identify pollutants in municipal well water from Toms River, NJ.(1) The selectivity and wide dynamic range of these tools was demonstrated by determining elemental compositions for 51 apparent molecular ions formed from both minor and major components in a tar-like sample from the Superfund site.(3) The sensitivity of these tools for a real-world sample was illustrated when five isomers of 1:2 styrene:acrylonitrile adducts present at levels too low for study by FTIR or NMR were identified in the well water extract (1). In addition, a detection limit of 6 fg for a tetrachlorobiphenyl standard was established using 10,000 resolution (4). The tools are a new HRMS technique for acquiring data, Mass Peak Profiling from Selected Ion Recording Data (MPPSIRD) (4,5), and a Profile Generation Model PGM) (2) that is used to plan experiments and interpret the data. The PGM considers exact masses and their error limits for an ion and ions heavier by 1 and 2 Da that arise from heavier isotopes, such as 13C and 18O. Also considered are the abundances of these heavier ions relative to the ion containing only the most abundant isotope of each element. Testing criteria based on five quantities, three exact masses and two relative abundances, rather than a single exact mass, are responsible for the increased mass range for which unique elemental compositions can be determined.
To demonstrate the broad applicability of MPPSIRD and the PGM, they have been used to determine elemental compositions of ions from compounds present at trace and higher levels in samples introduced in the gas phase by capillary gas chromatography (GC) (1,3-5) and direct probe (3) and in the liquid phase by liquid chromatography (4) and flow injection.(6) The advantages of MPPSIRD relative to conventional full scanning techniques include greater mass accuracy, greater relative abundance accuracy, 100-fold greater sensitivity, shorter sampling cycles to permit delineation of chromatographic peaks, and unequivocal assignment of elemental compositions to higher mass limits for a given set of elements.
Scientific journals specify criteria to confirm identities of synthetic products by HRMS. The Journal of the American Chemical Society requires agreement of the experimentally determined exact mass to within 5 ppm of the theoretical mass for compounds with molecular weights of 1000 or less (7) and reporting of the experimental and theoretical relative abundances of ions in the "molecular envelop", although no error limits are specified. Historically, the Journal of Organic Chemistry also used a 5 ppm criterion for the exact mass error,(8) but in 1994 adopted criteria (9) in accord with those suggested by the Journal of the American Society for Mass Spectrometry.(10) The "acceptable uncertainty" in the measurement must be assessed, and all elemental compositions possible for an ion within that error range must be considered. In this study, to illustrate the preponderance of evidence beyond these limited criteria that can be established for determining elemental compositions of compounds, MPPSIRD and the PGM are used to conduct a more thorough mass spectrometric investigation leading to identification of two synthesis products.
Synthesis was attempted of two phosphorothioate compounds for development of immunoassays to screen for organophosphorus compounds used as pesticides.(11) These pesticides impair nerve function by inhibiting the enzyme acetylcholinesterase. While toxic to humans, these compounds are usually hydrolyzed in the environment.(12)
The thermolability and polarity of some of these compounds can make their analysis by GC difficult. Although liquid separation and introduction could be used, and would be necessary if the compounds were ionic, considerable time would be spent in optimizing buffers, column packings, and other variables associated with LC-MS, CE-MS, or CEC-MS. Here, simple probe introduction was used to obtain mass spectral information. Direct insertion probes are commonly available and data collection is rapid, since analytes volatilize within a few minutes.
Two synthetic products were provided for analysis. One was a clear, yellow oil, and the other was a dark brown oil. Small portions of the samples diluted with ~1 mL of toluene (Burdick and Jackson, High Purity Solvent, Muskegon, MI) provided abundant ions for analysis using direct probe introduction into the ion source of a VG70-250SE double focusing mass spectrometer (Micromass, Danvers, MA). Depending on the scan type, mass resolution, and abundance of the target ion(s), 0.1 to 1 mL of each solution was transferred into a cup with a volume of 3 mL formed on the end of a 2-mm diameter, glass rod. The cup-and-rod was inserted into a machinable glass and boron nitride tip that in turn fit into the stainless steel probe (all parts from Micromass). The solvent boiled off when the probe was inserted through a valve into a roughing vacuum. A second valve was opened after the source region ion gauge read < 10-5 mbar, and the probe was fully inserted until the tip rested against the block of the ion source, which was at 250 oC. The thermocouple that monitors the tip temperature was heated by the block from < 40 oC to < 55 oC as most of the analytes volatilized over about 1.5 min. Slow heating lessened or eliminated the risk of thermal degradation. Use of the small cup at the end of a glass rod rather than a capillary avoided observation of multiple volatilization peaks that result from droplets clinging to the side of the capillary. Using the cup, the sample solution was localized and heated uniformly.
Four types of scanning were used. Full scans (m/z 50-700) at low mass resolution (1,000) with an accelerating potential of 8 kV were acquired at 3.6 sec/scan to provide nominal masses for the presumed molecular ion (M+) and for abundant fragment ions. MPPSIRD was used to determine exact masses and relative abundances for molecular ions and exact masses of smaller-mass ions at resolutions of 10,000 to 24,000. Selected Ion Recording scans (fixed magnet current with proportional accelerating potential and electrostatic sector potential jumps between m/z ratios) were made at 10,000 to 24,000 resolution to acquire volatilization peaks of ions after exact masses of ions were determined using MPPSIRD. The cycle time was 0.27 to 0.42 sec to ensure slight shifts in volatilization peak maxima were detectable. The ratio of the mass range limits for SIR descriptors never exceeded 1:1.81, since sensitivity drops drastically at accelerating potentials below 4 kV, which corresponds to a ratio of 1:2. Three SIR descriptors were used to cover each mass range of 97 to 332 and 97 to 408 Da. Some ions were monitored by two SIR descriptors to provide indirect comparison of volatilization peaks of molecular ions with those for the lowest-mass ions. Linked scans with a constant B/E ratio were used to observe daughter ions in order to confirm or refute correlations based on the similarity of volatilization peaks of ions. The pressure of the collision gas, He (Air Products, Allentown, PA), was adjusted so that the abundance of a calibrant ion was reduced to 50% of its normal abundance before linked scans were made. The photomultiplier voltage was set at 300 V to 500 V, depending on the sensitivity required by the type of scan, mass resolution, and abundance of the ions observed.
MPPSIRD and the PGM
MPPSIRD has been defined and demonstrated previously,(4,5) but will be reviewed briefly. In Figure 1a are displayed volatilization peaks obtained at different m/z ratios across the mass peak profile in Figure 1b.
Figure 1. (a) Volatilization peak areas acquired at five m/z ratios across the top of a mass peak profile. (b) A full mass peak profile plotted from volatilization peak areas, including those in (a). (c) Partial mass peak profiles for the M, M+1, and M+2 ions.
The top portion of the profile was plotted from the areas of the shaded volatilization peaks. Figure 1b is a full mass peak profile acquired at 20,000 (+ to -10%) resolution with 5 ppm mass increments between the points. The volatilization peak areas used to plot the three partial profiles in Figure 1c were acquired using a single SIR descriptor, which can monitor no more than 25 m/z ratios. From the partial profiles, the exact masses of the M+1 and M+2 partial profiles and the abundances of the M+1 and M+2 partial profiles relative to the M partial profile are obtained, the exact masses as weighted averages of the top several points and the relative abundances as the appropriate ratio of the sum of the six points for each partial profile. Partial profiles were used only to identify molecular ions. The partial profile of a single calibration ion (not shown) from perfluorokerosene-H (PCR, Inc., Gainesville, FL) is also monitored for both full and partial profiles.
To obtain the exact mass of an ion, up to three data acquisitions were required. First, a wide mass range, usually 1600 ppm, was observed at a mass resolution of 3000. Profiles defined by only three points resulted and provided estimates of the exact masses of one or more ions that were used to acquire data at 10,000 resolution for each ion observed. In those cases where multiple compositions for ions or neutral losses from the molecular ion were possible, a third profile was plotted from data acquired at 20,000 resolution to reduce the error limits of the exact mass determination. Narrower error limits rejected more compositions.
MPPSIRD provides about 100 times more sensitivity and 3-fold faster cycle times (4) than electric full scanning (fixed magnet current with proportionally scanned accelerating and electrostatic sector potentials) at 10,000 resolution. These advantages mandate its use when investigating minor components in complex mixtures that enter the mass spectrometer as narrow chromatographic peaks.(6) These advantages were not important in these experiments. Here, MPPSIRD was used rather than electric full scanning for three reasons. (i) With MPPSIRD, data interpretation was automated, whereas electric full scan data requires manual interaction to locate profile maxima and to type in reference masses. (ii) The error limits for exact mass determinations and relative abundances for the M+1 and M+2 profiles by MPPSIRD have been established and are incorporated in the PGM,(2) which is used to reject incorrect compositions using criteria based on these quantities automatically. (iii) The computer memory requirement for data is an order of magnitude less for MPPSIRD.
RESULTS AND DISCUSSION
The two target products were O,O-diethyl O-4-phenylbutyric acid phosphorothioate and O,O-diethyl O-6-carboxyhexyl phosphorothioate. Synthesis of these and other compounds was part of the National Exposure Research Laboratory's development of immunoassay-based analytical methods for field screening of contaminants. The syntheses and specific rationale for producing these compounds is presented in Reference 11. The techniques discussed here established that the first target compound was synthesized, but that the second was not. The identity of the second synthetic product and why it was formed were questions answered by this study.
The low resolution mass spectrum corresponding to the maximum in the volatilization peak for the majority species in the first synthesis product is shown in Figure 2a.
Figure 2. Low resolution mass spectra with probe introduction of synthesis products. (a) Mass spectrum corresponding to the maximum in the volatilization peak for the m/z 332 ion from the first synthesis product and (b) mass spectrum corresponding to the maximum in the volatilization peak for the m/z 408 ion from the second synthesis product. The ion abundances were multiplied by 10 between 350 and 415 Da.
The presumed molecular ion of the target product at m/z 332 and numerous possible fragment ions were observed. Ions with m/z 360 and m/z 389 were also of interest, since one could have been the parent ion of the m/z 332 ion.
Two maxima in the summed ion volatilization peaks and in numerous single ion volatilization peaks indicated that two compounds comprised most of the second synthesis product. The largest mass ions observed from the two compounds were m/z 171 and m/z 408. The mass spectrum with the maximum abundance of the m/z 408 ion is shown in Figure 2b. A significant abundance for the molecular ion of the target compound, C10H21O5PS, at m/z 284 was not observed, which suggested that the target product was not formed or further reacted to form other products. The product that produced the m/z 408 ion was investigated to determine its elemental composition and structural features of the molecule.
Determination of Elemental Compositions of the Molecular Ions
The exact mass of the m/z 332 ion was determined to be 332.0855 + to -0.0010 Da (+ to -3 ppm) from the mass peak profile in Figure 1b obtained at 20,000 resolution. The PGM determined that 53 compositions were possible within the error limits of the exact mass determination. In Table 1, 17 of the compositions are shown with the calculated mass defects (non-integer part of the mass) for the M profile and the M+1 and M+2 partial profiles, as well as the calculated abundances of the M+1 and M+2 partial profiles relative to the M partial profile.
Table 1. Elemental compositions and quantities useful for distinguishing among them
aRings and double bonds:
minimum with valences of 3, 3, and 2 for N, P, and S and maximum
with valences of 5, 5, and 6 for N, P, and S;
bbased on partial profiles that provide maximum area;2
cbased on full profiles;
dbased on partial profiles centered about the calculated mass of the hypothetical composition, + to -1 mass increment at + to -10% of resolution, and isotopic abundance error.2
An "X" indicates application of this criterion will reject this composition if the hypothetical composition is correct.
For a single determination at 20,000 resolution, the theoretical and experimental exact masses must agree within 3 ppm. The permissible ranges for %M+1 and %M+2 calculated by the PGM are shown in parentheses. The experimental values were determined from the partial profiles in Figure 1c. The error limits exceeded by the experimental results are marked with "X"s. Based on the assumption that only C, H, N, O, P, and S atoms were present during synthesis, only the expected composition, C14H21O5PS was consistent with the five quantities; the other 52 compositions were not.
Because phosphorus has a valence of 5 in the first target product, the PGM was modified to provide a range of rings-and-double-bonds (RDB). The minimum number of RDB was calculated by the formula RDB = x - y/2 + z/2 + 1, where x is the number of C, Si, or other tetravalent atoms, y is the number of H, halogen, or other univalent atoms, and z is the number of N, P, or other trivalent atoms.(13) The maximum possible number of RDB was determined by assuming N and P could have a maximum valence of 5, rather than 3, and S a valence of 6, rather than 2. One RDB was added for each N and P atom and 2 RDB were added for each S atom. Using these RDB ranges, compositions for which these three atoms had their maximum valences would not be overlooked. Before this adjustment, the correct composition for four fragment ions was overlooked, since compositions with fewer than 0 rings and double bonds were discarded after the formula above calculated their RDB values as -0.5 based on the lowest valences for N, P, and S.
Using full and partial profiles acquired at 10,000 and 20,000 resolution, the compositions of the m/z 360 and m/z 389 ions seen in low abundances in Figure 2a were determined using the PGM to be C16H25O5PS+ and C17H26O4PS2+, respectively, based on the five criteria. The first composition is consistent with ethylation of the carboxyl group of the target compound and the second composition is consistent with a fragment (M-CH3) resulting from a product with both carboxyl and ring ethylations and dithiophosphoryl contamination of a reagent (14). Because the m/z 332 ion contains one more O atom than the m/z 389 ion, the m/z 332 ion was not formed from the m/z 389 ion.
Also shown in Table 1 is a portion of the PGM output obtained after entering the experimental values of the mass defects of the m/z 408 profile and the 409 and 410 partial profiles and the relative abundances of the 409 and 410 partial profiles. Assuming the compound contained only C, H, O, N, P, and S atoms, 151 compositions were possible (20 are shown) based on the exact mass of 408.0590 + to -0.0012 Da (+ to -3 ppm) determined from a single full profile plotted from data acquired at 20,000 resolution. Exact masses and relative abundances to test the other four criteria were determined from duplicate sets of partial profiles acquired at the same resolution. At least one criterion was failed by all but the correct composition, and C12H26O7P2S2 was identified as correct.
Identifying Fragment Ions
The compositions of fragment ions and neutral loss fragments produced from the molecular ion reveal structural details of a compound.(1,15) In this study, such details provided greater confidence that the first targeted reaction product was produced and suggested where the second synthesis went awry. With probe introduction, composite mass spectra were observed due to impurities. Prominent ions observed in the mass spectra were investigated to determine if they were daughter ions from the m/z 332 or m/z 408 ions, and if so, to determine their compositions. A three-step process was employed. First, the exact masses of the fragment ions at each nominal mass were determined by acquiring data at 3000, 10,000, and in some cases 20,000 resolution as described in the experimental section. Second, volatilization peaks were acquired at 10,000 to 24,000 resolution to determine which ions fragmented from the molecular ion. Linked scans acquired with 1000 resolution were also used for this purpose. Finally, the PGM was used to determine elemental compositions of the fragment ions based on the maximum number of atoms of each element in the molecular ion. The ions labeled in Figure 2 were investigated.
Correlating Fragment Ions and the Molecular Ion
For compounds separated by GC/MS analyses, overlap of maxima in normalized ion chromatograms indicates which fragment ions are probably daughter ions of a molecular ion. Volatilization from a probe, however, provides much less separation capability than a GC. To compensate for limited separation in the time domain, high mass resolution was used to provide separation in the mass domain. Hence, volatilization peaks obtained with high mass resolution were still useful for correlating ions.
In Figure 3 are illustrated several situations encountered in this study for pairs of volatilization peaks acquired with 10,000 resolution.
Figure 3. Volatilization peaks acquired with 10,000 resolution for (a) m/z 107.0499 and m/z 96.9514, (b) m/z 107.0499 and m/z 97.1019, (c) m/z 201.9888 and m/z 115.0764, (d) m/z 408.0590 and m/z 293.9914, and (e) m/z 99.9936 and m/z 96.9903.
In Figure 3a, the two volatilization peaks overlap. Unless other compounds coelute perfectly, these two ions are produced from the same analyte. In Figure 3b, the volatilization peaks have very different maxima, and the ions are produced from different compounds. In Figure 3c, one volatilization peak displays two maxima, which indicates that an ion of the same mass is produced by two different compounds. The second maximum very nearly overlaps the single maximum in the other volatilization peak. Because the second peak in the m/z 115 trace is superimposed on the tail of the first peak, the retention time of the second peak is 3 sec earlier than the retention time for the peak in the m/z 202 trace. Even so, both m/z 115 and m/z 202 ions are produced by the compound that volatilized last. In Figure 3d, the elevated baseline for the lower-mass ion (m/z 293.9914) is due to interference from the 13CC6F11+ ion (m/z 293.9858) of perfluorokerosene (PFK), the ever-present calibrant. The overlapping maxima still indicate that the lower mass ion is probably a daughter ion. Finally, in Figure 3e, a pair of volatilization peaks produced from PFK are shown. The ion abundance for both ions is relatively constant. The two excursions to the baseline at 15 sec and 25 sec occurred when valves were opened to insert the probe into the roughing vacuum and then to rest it against the ion source.
The most important criterion for deciding that two ions are produced from the same compound is that maxima in their volatilization peaks overlap. Discerning how well volatilization peaks track each other and observing the divergence of tails and small displacements between maxima is somewhat subjective, and coelution of other compounds with similar molecular weights within the tolerances judged to be acceptable is possible. Thus, mis-assignments of daughter ions can result. The maxima in the volatilization peaks for m/z 332.0847 and m/z 314.0742 in Figure 4a were offset by 5 sec, and the two ions were thought to be formed from different compounds.
Figure 4. Computer-smoothed volatilization peaks acquired at 10,000 resolution for (a) m/z 332.0847 and m/z 314.0742 and (b) m/z 332.0847 and m/z 360.1160.
The volatilization peaks for m/z 332.0855 and m/z 360.1160 in Figure 4b overlapped well, and the m/z 332 ion was at first thought to be formed from the m/z 360 ion. These judgments were later proven to be incorrect. Hence, this approach is generally useful for correlating daughter ions with a molecular ion, especially by eliminating potential daughter ion candidates when major differences in the volatilization peaks occur. This approach is not as rigorous, however, as using tandem mass spectrometry or linked scans to establish daughter ions. On the other hand, tandem mass spectrometry or linked scans that provide nominal masses of the daughter ions cannot determine if more than one daughter ion with a given nominal mass is produced from a molecular ion. In these cases, multiple daughter ions can usually be established based on volatilization peaks acquired with high mass resolution.
In this study, linked scans provided nominal masses of daughter ions as low as m/z 134 and, as illustrated in Figure 5a, indicated that ions with m/z 314, 273, 260, 245, 217, and 171 were produced from the m/z 332 ion.
Linked scans that provide fragment ions from the (a) m/z 332 and
from the (b) m/z 360 ion.
Other linked scans revealed that the m/z 259 ion observed in Figure 2a was produced from the 314 ion and that the m/z 136 ion fragmented from the m/z 245 ion. Another linked scan demonstrated production of m/z 294, 266, 238, 202, and 171 ions from the m/z 408 ion. Also, it was determined from Figure 5b that the m/z 332 ion was not produced from the m/z 360 ion. The two mis-assignments based on the appearance of the volatilization peaks were corrected by the linked scans.
Necessity for High Mass Resolution
When a volatilization peak is acquired at low mass resolution, it can result from isobaric ions (ions with the same nominal mass) having different compositions. In this case, an ion associated with M+ might be overlooked as the shape of the ion's volatilization peak is altered by the presence of another, more abundant isobaric ion produced from a less or more volatile impurity. High mass resolution is necessary to establish the presence of single or isobaric ions and to determine their compositions. When isobaric ions exist, the volatilization peak of each must be acquired at a high enough mass resolution to separate their signals before comparison to the volatilization peak of M+ to establish correlations. This point is illustrated in Figures 3a, 3b, and 3e for ions with a m/z ratio of 97. Examination of a 3200 ppm mass range around 97 Da at 2000 resolution using MPPSIRD revealed the presence of ions centered at 96.9514, 96.9903, 97.0656, and 97.1019 Da as subsequently determined from full profiles acquired using 10,000 resolution. The volatilization peaks of two m/z 97 ions acquired with 10,000 resolution are shown with the volatilization peak of the m/z 107.0499 ion, which was similar in appearance to that of the molecular ion, in Figures 3a and 3b, and a third m/z 97 ion is shown with the volatilization peak of the m/z 99.9936 ion from PFK in Figure 3e. The overlap of the maxima in the volatilization peaks in Figure 3a suggested that the m/z 96.9514 ion was associated with the target compound. In fact, this ion, H2O2PS+, is present in the mass spectra of phosphorothioates (16). Early maxima in the traces for the m/z 97.0656 ion (C6H9O+ - not shown) and the m/z 97.1019 ion (C7H13+) shown in Figure 3b revealed that they were not associated with the compound of interest. After about 50-fold dilution, these ions were observed with similar abundances, while the signals of the m/z 96.9514 and m/z 107.0499 ions decreased as expected. In Figure 3e, the m/z 96.9903 ion provided a steady signal that indicated its origin was an ever-present compound. Its exact mass corresponded to the C2F3O+ ion formed probably from PFK and to no other compositions based on C, H, N, O, P, S, Si, or F.
For the m/z 408 compound, the same four ions were observed at nominal m/z 97, and pairs of ions were found for m/z 115 and m/z 143. After determining the exact masses of these ions at 10,000 resolution, volatilization peaks at the same resolution were acquired for each ion. Two of the m/z 97 ions, both m/z 115 ions, one m/z 143 ion, and the m/z 171 ion showed one or two maxima, one of which overlapped that of the volatilization peak of a higher-mass fragment ion produced from the m/z 408 ion. The volatilization peaks are shown for one example in Figure 3c. The experimental exact masses and assigned compositions were 96.9513 Da (H2O2PS), 97.0652 Da (C6H9O), 114.9619 Da (H4O3PS), 115.0764 Da (C6H11O2), 142.9929 Da (C2H8O3PS), and 171.0244 Da (C4H12O3PS).
Another case where high mass resolution was essential to correlation of ions is illustrated in Figure 6.
Figure 6. A full mass peak profile acquired with 10,000 resolution that provided 153.0115 Da as the exact mass, (b) a full profile acquired with 24,000 resolution showing partial separation of two ions with measured exact masses of 153.0105 Da and 153.0139 Da, (c) volatilization peaks acquired at 24,000 resolution for m/z 153.0139 and m/z 120.0573 and (d) for m/z 153.0105 and m/z 120.0573. The resolution was determined as the average from profiles for the m/z 154.9920 ion of PFK obtained before and after the profile in (b).
The mass peak profile in Figure 6a was acquired at 10,000 (+ to -10%) resolution and provided an exact mass of 153.0115 Da. No composition based on the elements in the molecular ion, C, H, O, P, and S, corresponded to this mass, yet its volatilization peak tracked that of a fragment ion formed from M+ (m/z 332). The mass peak profile in Figure 6b acquired at 24,000 resolution clearly indicated that two ions with similar abundance produced the composite profile in Figure 6a. The exact masses of the two profiles determined from the weighted average of the three top points defining each maximum, 153.0104 Da and 153.0135 Da, corresponded to compositions of C7H6O2P and C4H10O2PS. Thus, two ions with masses different by only 20 ppm were formed from the target compound. This conclusion was supported by the overlap in Figures 6c and 6d of volatilization peaks acquired at 24,000 resolution for the ions with m/z 153.0105 and m/z 153.0139, the theoretical masses, with that of the ion with m/z 120.0573, which tracked the apparent molecular ion. If only tandem mass spectrometry that provided nominal mass data were used, one composition might be overlooked and a less complete picture of the fragmentation pathways of the molecular ion would result.
Higher resolution was also necessary to determine the exact mass of the m/z 245 ion. At 10,000 resolution the exact mass was determined to be 245.0370 Da, which corresponded to C13H10O3P+. Loss of 1 C, 11 H, 2 O, and 1 S from C14H21O5PS+ to form this ion was unlikely. The mass peak profile acquired at 24,000 displayed a broad, low-mass tail due to another ion present with lower abundance. At the lower resolution, interference from this tail had provided a low estimate of the ion's exact mass. The exact mass obtained using 24,000 resolution was 245.0396 Da, which corresponded to C10H14O3PS+. This ion results from the more plausible loss of 4C, 7H, and 2O from M+ (CH2C[=O]OH + C2H4) and is illustrated in Figure 7a.
Figure 7. Plausible fragmentation schemes for the ions investigated that were produced from the (a) m/z 322 ion and from the (b) m/z 408 ion. The 7-digit masses are the theoretical masses for compositions determined using MPPSIRD and the PGM. The asterisks indicate fragmentations confirmed by the linked scans.
Compositions of the Fragment Ions
In Table 2 are listed the exact masses of 13 fragment ions from the m/z 332 ion and of 10 ions formed from the m/z 408 ion.
Table 2. Exact Masses Determined for the Ions Related to the m/z 332 and m/z 408 Ions
In Table 3 are listed the corresponding neutral losses determined from the mass difference between M+ and these fragment ions.
Table 3. Exact Masses of Neutral Losses Determined by Mass Differences Between the Molecular and Fragment Ions
aThe sum of the maximum errors for
determinations of the molecular and fragment ion exact masses.
bNumber of compositions possible based on 14 C, 21 H, 5 O, 1 P, and 1 S atoms for the first 13 rows, and on 12 C, 26 H, 7 O, 2 P, and 2 S atoms for the next 10 rows.
cNumber of compositions possible based on the nominal mass up to the same number of atoms of each element.
The compositions in the table of fragment ions are the unique elemental compositions that correspond to each exact mass based on the total number of atoms of each element in the molecular ion: 14 C, 21 H, 5 O, 1 P, and 1 S for the m/z 332 molecular ion and 12 C, 26 H, 7 O, 2 P, and 2S for the m/z 408 molecular ion.
The maximum error estimated for the neutral losses is the sum of the maximum errors for the molecular ion (3 ppm for a single determination at 20,000 resolution) and of the fragment ions (6 ppm for one determination at 10,000 resolution or 3 ppm for one determination at 20,000 resolution).(2) Larger mmu error limits result for the lower-mass neutral losses determined from the difference between two large-mass ions, which have larger mmu error limits for their exact mass determinations. The number of possible compositions is based on the maximum error limits. Full profiles for the four largest-mass fragment ions produced from the m/z 408 ion and for three fragment ions produced from the m/z 332 ion were acquired at 20,000 to reduce the total error limit of the neutral losses. This reduced the number of combinations of atoms possible for neutral losses and also reduced from two to one, the number of compositions possible for each of three fragment ions from the m/z 332 ion.
For all neutral losses from both the m/z 332 ion and the m/z 408 ion, only one composition was possible, the one that corresponded to subtraction of the composition of each fragment ion from the composition of the molecular ion. This agreement provided a check for consistency. As noted earlier, one implausible neutral loss indicated that greater resolution was required to isolate the target ion from an interfering ion. Afterward, all entries in Table 3 could be rationalized.
In Tables 2 and 3, the number of possible compositions calculated by the PGM based on each nominal mass + to -0.5 Da is in parenthesis. Although the RDB calculations were applied to fragment ions, the number of RDB was not considered for the neutral losses, which can arise from sequential fragmentations. Thus, for the neutral losses, all subsets of atoms within the limits established by the molecular ion with the same nominal mass were calculated by the PGM. The alternative neutral losses based on nominal masses have a range of plausibilities. For example, a neutral loss of 18 Da probably corresponds to a loss of H2O, less plausibly to loss of CH6 (CH + 3H ), and very implausibly to loss of 18H atoms. However, among the dozens of possibilities in parentheses, some alternative, single and composite neutral losses would not be easily rejected. Obtaining exact masses eliminates other plausible neutral losses based on the exact mass of the atoms involved, which is a physical property, rather than a knowledge based probability. Simplified interpretation and greater certainty for fragmentation schemes are provided when exact masses are determined.
Possible Fragmentation Schemes
In Figure 7 are depicted possible fragmentation schemes that yield all of the ions shown to be associated with the molecular ions. The linked scans were used to establish some of the pathways. The compositions of the fragment ions and the neutral losses, and rational explanations for formation of all the ions provide additional evidence that the first target compound is correct and provide detailed structural information about the unintended product. Although other fragment ions and neutral losses could be isolated, the ions with high relative abundances that were investigated were sufficient to provide the structural details of both molecules.
The unexpected product shown in Figure 7b formed instead of the second target compound was a dithiopyrophosphate, specifically O,O,O'-triethyl-O'-6-carboxyhexyl dithiopyrophosphate, formed by condensation of the desired product with another thiophosphate group due to excessive heating. More gentle heating conditions yielded the expected product.
Traditionally, to provide mass spectrometric confirmation of the identity of a synthetic product, the exact mass of the molecular ion is determined and cited as consistent with the expected elemental composition within the error limits of the mass determination. In this study, the evidence that the first product was made is far more complete. All other possible compositions for the molecular ion based on the exact mass and its error limits have been rejected and structural details were deduced from the compositions of the fragment ions and neutral losses. For the unintended product, a unique composition, structural details, and knowledge of organic chemistry suggested why it was synthesized rather than the target compound.
Rapid analyses were performed using probe introduction of mixtures of products not likely to survive transit through a GC column. The loss in separation efficiency in the time domain was compensated by using high mass resolution to provide greater resolution in the mass domain. Volatilization peaks for each of multiple ions with the same nominal mass were obtained at 10,000 to 24,000 resolution to prevent mutual interferences. Linked scans revealed the origin of two ions for which the volatilization profiles in the volatilization peaks were ambiguous.
If the identities of the two compounds were critical to a legal proceeding, Tables 1, 2, and 3 and the fragmentation schemes in Figures 7a and 7b would provide a body of compelling evidence for the identity of the compounds. Although additional analytical effort was required to compile this evidence, such a preponderance of evidence might deter either party to a suit from disputing the issue of compound identities and thereby lead to a judgment based on sound science after less court time with a lower total cost to both sides.
Instructions for Authors (1993) J. Org. Chem. 58, 8A.
Instructions for Authors (1997) J. Org. Chem. 62, 14A.
Instructions for Authors (1997) J. Am. Soc. Mass Spectrom. 8, A14.
Johnson, J.C., Van Emon, J.M., Pullman, D.R., & Keeper, K.R. (1998) J. Agricultural & Food Chem. 46, 3116-3123.
Eto, M. (1974) Organophosphorus Pesticides: Organic and Biological Chemistry, CRC Press, Cleveland, OH.
McLafferty, F.W., & Turcek, F. (1993) Interpretation of Mass Spectra, 4rd Ed., Univ. Science, Books, Mill Valley, CA, P. 23.
Sovocool, G.W., Harless, R.L., Bradway, D.E., Wright, L.H., Lores, E.M. & Feige, L.E. (1981), J. Anal. Toxicol 5, 73-80.
Lee, M.S., Klohr, S.E., Kerns, E.H., Volk, K.J., Leet, J.E., Schroeder, D.R., & Rosenberg, I.E., (1996) J. Mass Spectrom. 31, 1253-60.
Damico, J.N. (1972) Biochemical Applications of Mass Spectrometry, Waller, G.R. Ed., Wiley-Interscience, New York, NY, p. 627.
The U.S. Environmental Protection Agency (EPA), through its Office of Research and Development (ORD), funded and performed the research described. This manuscript has been subjected to the EPA's peer and administrative review and has been approved for publication.