Ion Composition Elucidation (ICE)
Determining Elemental Compositions from Exact Masses and Relative Abundances of Ions
Andrew H. Grange1 and William C. Brumley2
1Lockheed Engineering Systems and Technologies,
980 Kelly Johnson Drive, Las Vegas, NV 89119
2Characterization Research Divison, NERL, U.S. EPA,
PO Box 93478, Las Vegas, NV 89193-3478
A new scanning strategy permits on-line determination of exact masses and relative abundances of ions produced from analytes as they are separated chromatographically and analyzed by a high-resolution mass spectrometer. Tentative identifications of compounds based on library matches of their background subtracted mass spectra can now be confirmed or refuted by applying 6 criteria based on the molecular ion (M), M+1, and M+2 mass peak profiles. This article reviews the technique, demonstrates its most common application, and lists some of its other uses.
The U.S. Environmental Protection Agency (EPA), National Exposure Research Laboratory, Characterization Research Division in Las Vegas conducts research in separation and identification of compounds in hazardous wastes and other environmental samples. Gas chromatography-mass spectrometry (GC-MS) analysis generally provides reliable analyte identification when background subtracted mass spectra free of mass peaks from coeluting compounds are compared with library mass spectra. But coelution of components from complex environmental mixtures is a common problem, and identification of analytes from low resolution mass spectra is often inconclusive. Multiple library matches for compounds lead to misidentifications, especially if an analyte's mass spectrum is not in the library. In addition, identification is less certain for highly aromatic compounds such as polyaromatic hydrocarbons which provide only a few mass peaks in their mass spectra.
Because ions observed in mass spectra are composed of elements and their isotopes, all having different exact masses, the exact masses of ions having different elemental compositions are unique. If the error limit for an exact mass determination for a molecular ion (M) is small enough, only one composition corresponding to a limited number of isomers is possible for the compound. On-line, exact mass determinations for trace-level GC peaks would provide confirmation or rejection of tentative identifications by establishing the composition of the molecular ion or other prominent ions. Unfortunately, exact masses of ions from analytes introduced into mass spectrometers by chromatographic techniques have rarely been determined because quadrupole mass spectrometers lack the resolution necessary to exclude interferences from target ion signals, while high-resolution mass spectrometers lacked the scan speed necessary to track narrow chromatographic peaks.
A new technique, mass peak profiling from selected ion recording data (MPPSIRD), overcame these limitations by providing 3-fold faster scanning and 170 times more sensitivity than conventional full scan modes when using a VG70-250SE high-resolution mass spectrometer. Ten scans were made across chromatographic peaks 8 s wide and 6 fg of a tetrachlorobiphenyl was the detection limit at 10,000 resolution using MPPSIRD.
Full and partial profiles plotted from SIR data
For dioxin analyses (EPA Methods 8280, 8290, and 1613) , selected ion recording (SIR) is used to monitor the apex of mass peak profiles for multiple ions from several congeners. The chromatographic peak areas in the ion chromatograms provide quantification to very low detection limits. By monitoring, instead, several m/z ratios across a single mass peak profile, the exact mass of an ion can be determined as the weighted average mass of the chromatographic peak areas. In Figure 1a-e are shown 5 ion chromatograms for m/z ratios monitored across the mass peak profile in Figure 1f, which was plotted from the areas of the chromatographic peaks.
Figure 1. Five ion chromatograms (a-e) that display chromatographic peaks, the areas of which were used to plot the mass peak profile in (f).
Areas were plotted, rather than peak heights, to improve the signal-to-noise ratio and provide a smoother profile. Dividing the exact mass by the mass difference between the points at 5% of the maximum provides the mass resolution.
Although data have been acquired at nearly 40,000 resolution, triplicate exact mass determinations for an ion are routinely made at resolutions of 20,000 to 25,000, which provide a mass accurate to within 2.5 ppm. Usually, only one elemental composition corresponds to the exact mass within this error range for low mass ions. However, the number of compositions possible within an exact mass error range increases rapidly with the mass of the ion, the error range of the measurement, and the number of elements and atoms of each element considered. Thus, within a mass range of + to - 2.5 ppm, multiple compositions are possible for large mass ions and additional criteria are necessary to eliminate all but the correct composition.
Because isotopic abundances are different for each element having multiple isotopes, the abundances of the M+1 and M+2 profiles relative to the M profile differ between compositions. The mass differences between isotopes are also different for each element. Hence, the relative abundances and exact masses of the M+1 and M+2 profiles provide 4 criteria for distinguishing between compositions. If a possible composition contains one or more S or Si atoms, the M+2 profile is broadened substantially due to the similar abundances of M+2 ions containing 34S (4.2% per S) and 30Si (3.1% per Si) and M+2 ions containing 2 13C atoms (e.g., 3.6% for C28). For some compositions, double maxima are observed in the M+2 profile at 20,000 resolution. Thus, M+2 profile shapes can often be used to reject compositions. In sum, 5 criteria used to identify the correct composition become available when the M+1 and M+2 profiles are plotted.
The mass increment between m/z ratios is chosen as (M/R x N), where M is the ion's mass, R is the resolution, and N is the number of points monitored across the profile. A SIR descriptor can contain up to 25 m/z ratios (22 were used in this work). In the absence of profile broadening, setting N to 10 provides 10 points across the profile and 7 points to establish a baseline. The remaining 5 points provide a partial profile for the calibration ion from perfluorokerosene (PFK), which is always present in the ion source.
To determine relative abundances, multiple analyte profiles must be monitored simultaneously by a split SIR descriptor, i.e. one that monitors multiple analyte ions. The top 60% of the M, M+1, and M+2 profiles are each monitored by 6 m/z ratios and the apex region of the calibration ion is monitored at 4 points. In Figure 2 are shown example M, M+1, and M+2 partial profiles.
Figure 2. Partial profiles for M, M+1, M+2, and calibration ions that were plotted from data acquired with a split SIR descriptor.
The data system cannot plot profiles. To make MPPSIRD practical, automated procedures were written in RSXTM and VGTM code on the data system and in Lotus 1-2-3TM (2.2), ReflectionTM 4.2, and WordPerfectTM 5.1 on an ancillary personal computer to prepare SIR descriptors and to plot and print the profiles.[5-7] After the user enters the mass resolution and 1 center mass for full profiles or 3 center masses for partial profiles, a Lotus TM 1-2-3 procedure picks the PFK ion closest in mass as the reference ion, calculates the 22 m/z ratios, and writes an ASCII file that is executed in ReflectionTM 4 to automatically enter the m/z ratios, resolution, and other parameters into the preparation menu for the SIR descriptor. After data acquisition, the user enters only the start and end times for chromatographic peaks of interest in the data file to obtain hard copies of the corresponding full or partial profiles. Multiple analytes are often monitored during a single data acquisition, since the transition time between SIR descriptors is only 5 s.
(Note: These tasks are now performed by code written in Lotus 123 version 9.0, which is commercially available. In addition, 31 m/z ratios are now used to monitor a lock mass, a calibration mass, and up to four analyte profiles using a Finnigan MAT 900 double focusing mass spectrometer.)
Low mass ion example
Four library matches of low quality were found for the background subtracted mass spectrum of a trace component in an amber colored extract from a Superfund site sample that contained large amounts of polychlorobiphenyls. In the total ion chromatogram, the component provided a peak only 5% of full scale, atop a baseline at 64% of full scale due to much larger amounts of other components. The largest ion in each match spectrum appeared at m/z 266, but each composition was different. The exact masses of the compositions covered a mass range of 148 ppm and all differed from each other by more than 5 ppm. Hence, the exact mass of the ion would identify the compostion.
At 20,000 resolution, a mass range of only 80 ppm is observed when 10 points are monitored across a full profile. Unless the correct composition is assumed, the M profile will usually not be observed within the mass range monitored. Instead, a SIR descriptor written to survey a broad mass range certain to include all ions with the integer mass under investigation provided an estimate of the exact mass, which was then entered as the center mass in the SIR descriptor used to acquire data at 20,000 resolution. In Figure 3a-b are shown two of the ion chromatograms used to plot the survey data in Figure 3c.
Figure 3. Ion chromatograms (a-b) corresponding to the maxima in the profiles in (c).
The ion chromatograms correspond to the maxima for the 2 profiles observed. The first profile was due to a PFK ion and the ion chromatogram verified that it was always present in the ion source. The second profile was due to the analyte and a chromatographic peak was observed. A chromatographic peak was simulated for the PFK ion to permit automated integration of an area by closing and opening a switch located between the pre-amplifier and flash box before and after elution of the analyte peak. The wider mass range permitted monitoring of only 2 or 3 points across each profile, but provided evidence for the presence or lack of interferences with similar exact masses. In this example there were none.
The average exact mass from triplicate determinations made at 20,000 resolution correspondcd to C20H26 and all 4 compositions suggested by the library matches were rejected. The 5 compositions and the differences between the observed average exact mass and theoretical masses are shown in Table 1.
Table 1. Five compositions, their theoretical masses (TM), and differences between the average exact mass (EM) observed and the TMs (the correct composition is in blue)
The compound could be an alkylated biphenyl, but not any of the compounds matched in the library.
High mass ion example from an environmental sample
For the largest ion observed in the mass spectrum of a major component in an extract of a sample from another contaminated site, the average exact mass from triplicate determinations made at 20,000 resolution was 423.2955 Da. Within the error limit of 2.5 ppm, the 12 compositions listed in Table 2 were possible.
Table 2. Possible Elemental
Compositions for 423.2955 Da + to -2.5 ppm up to:
C35H74O6N6P4S4 Si4 (the correct composition is in blue)
To determine exact masses and relative abundances for the M+1 and M+2 profiles a split SIR descriptor was used to acquire data. To ensure the M+1 and M+2 profiles were centered in the mass windows monitored, the correct composition had to be chosen and exact masses had to be predicted for the profiles at the resolution used. A model based on the Gaussian distribution written in QuickBASICTM predicted the exact masses and relative abundances for M+1 and M+2 profiles expected when only the top 60% of the profiles were to be monitored. The model also plotted full M+1 and M+2 profiles. Calculated M+2 profiles for 5 of the 12 compositions at 20,400 resolution, the resolution used to acquire a full M+2 profile, are shown in Figure 4.
Figure 4. Calculated M+2 profiles (a-e) corresponding to 5 of the 12 possible compositions in Table 2.
The profiles in Figures 4a and b displayed double maxima and were unique in appearance, while the profiles for the other 10 compositions had shapes similar to those in Figure 4c-e. The calculated and observed M+2 profiles are shown in Figure 5. Their similarity identified the last composition in Table 2 as correct.
Figure 5. The calculated M+2 profile from Figure 4 (a) that most resembled the experimentally obtained M+2 profile in (b).
High mass ion example from a standard
When the appearance of the M+2 profile is not unique, the composition with predicted masses most similar to the exact masses determined from the M and M+2 profiles is chosen as the hypothetical composition and its predicted masses are used to write the SIR descriptor. If the apex of the M+1 profile were not observed or offset within the mass window monitored, another hypothetical composition would be chosen until the correct one was found.
As part of a quality assurance study, a standard with a molecular ion weighing 536.3151 Da (methylpyrroporphyrin XXI ethyl I ester) was obtained and introduced into the ion source by a heated probe. A single data acquisition and new calibration were made for each sample introduction. The average of 15 exact mass determinations was 536.3151 Da. Although 15 determinations for each of 4 standards provided an error limit of only 1.0 ppm, 2.5 ppm is used for this example, since only triplicate determinations are normally required. To narrow the list of possible compositions from 24 to 3, a %M+1 filter of + to - 4.8% (12% of %M+1 for the composition of the standard ) and a %M+2 filter of + to -1.0% (12% of %M+2) were applied. These filters are twice the error limit associated with triplicate determinations of %M+1 and %M+2. None of the remaining compositions contained S or Si atoms, which are primarily responsible for profile broadening. Hence, no significant broadening was observed in the M+2 profiles and none of their shapes were distinctive. The other 4 criteria were necessary to determine the correct composition. The differences between the theoretical and observed values for the masses and relative abundances are listed in Table 3. The entries in red did not meet the criteria of + to -2.5 ppm for exact masses and + to -6% for %%M+1 (2.4%) and %%M+2 (0.5%).
Table 3. Possible elemental
compositions up to C40H86O10N10P10S10
for the standard
(values in red failed a criterion and permitted rejection of a composition)
Only the last composition, that of the standard, passed all 6 criteria: 3 mass criteria, 2 relative abundance criteria, and the M+2 profile shape criterion. Similar results were obtained for 3 other standards with molecular ions weighing between 536 and 766 Da. In 2 cases, the error limits associated with 15 determinations (1.0 ppm and 2%) were required to eliminate the composition most difficult to reject. In these examples, a single composition remained after all criteria were applied and consideration of sample history, the extraction scheme, or compound stabilities would not be required to identify the compositions if the ions had been produced from compounds in a sample extract.
Other applications of MPPSIRD
MPPSIRD has also been used to determine exact masses for monosulfonated dyes introduced into the source in liquid chromatography eluents and ionized by cesium ion bombardment, rather than by electron impact; to demonstrate the resolution necessary to exclude interferences from analyte signals; and to document the mass resolution as each analyte entered the ion source. For polychlorinated and polybrominated analytes that coelute with other poly-halogenated compounds, a split SIR descriptor could be used to determine relative abundances of M+2 and M+4 (and if necessary, M+6 and M+8 profiles) from which the number of Cl and Br atoms can be determined (this point was later illustrated in a poster). For molecular ions so large that a unique composition cannot be determined, exact masses and relative abundances of fragment ion (F), F+1, and F+2 profiles can be determined, since at 20,000 resolution, interferences are usually excluded. Neutral loss fragments can be determined as exact mass differences between ions and the molecular ion can be identified as the sum of its parts. When partial coelution is troublesome, a split SIR descriptor can monitor M and suspected fragment ions at high resolution to verify that these ions are produced from the same analyte. High resolution distinguishes between fragment ions with the same integer mass, but with different compositions. Further applications are feasible, including use of MPPSIRD as a highly specific and sensitive qualitative and quantitative screening tool for environmentally important compounds.
Application of MPPSIRD need not be limited to the environmental arena. For example, composition identification of ions produced from components in combinational libraries and from impurities in batches of drugs is feasible for the pharmaceutical industry. Any analyte ions weighing less than 1000 Da are plausible targets for composition identification using the technique. In addition, metabolites in complex biological extracts could be quantified to low detection limits.
Addition to manufacturer's software for high-resolution mass spectrometers of automated procedures for performing MPPSIRD and a model for generating profiles could encourage widespread use of MPPSIRD.
MPPSIRD provides greater sensitivity and faster measurement than conventional mass scanning, and further exploits high mass resolution for performing analyses. MPPSIRD is a powerful new technique for determining elemental compositions for ions produced from major and trace components in complex mixtures, even when chromatographic separation is incomplete. Automated procedures rapidly prepare error-free SIR descriptors, process data, and provide plots of full and partial profiles. A simple model enables planning of experiments and interpretation of results. Qualitative and quality assurance applications have already been demonstrated and quantitative applications are feasible.
Split SIR descriptor: an SIR descriptor that simultaneously monitors multiple analyte ion profiles, in addition to part of the calibration ion profile.
Low mass ion: an ion for which only one elemental composition is possible within the error limit of the mass determination.
High mass ion: an ion for which multiple elemental compositions are possible within the error limit of the mass determination.
Exact mass: the mass determined experimentally as the weighted average of several points across the full or partial profile.
Center mass: the mass entered into a SIR descriptor thought to correspond to the apex of a profile.
Predicted mass: the weighted average mass across a full or partial profile calculated by the model.
Possible composition: a composition with a predicted mass that falls within the error range for an exact mass determination.
IUPAC, Pure and Appl. Chem., 63 (1991) 991.
A.H. Grange, Proc. 40th ASMS Conf. on Mass Spectrom. and Allied Topics, Washington, DC, (1992) 1145.
A.H. Grange and W.C. Brumley, Proc. 40th ASMS Conf. on Mass Spectrom. and Allied Topics, Washington, DC, (1992) 1143.
Note: The U.S. EPA, through is Office of Research and Development, funded and collaborated in the research described here. The article has been subjected to the agency's peer review and has been approved as an EPA publication