Ion Composition Elucidation (ICE)
An Investigative Tool for Characterization and Identification of Compounds of Regulatory Importance
Andrew H. Grange, Lantis I. Osemwengie, George M. Brilis, and G. Wayne Sovocool
Environmental Sciences Division, NERL, U.S. EPA, PO Box 93478, Las Vegas, NV 89193-3478
Phone: (702) 798-2137
Fax: (702) 798-2142
Keywords: Mass Peak Profiling from Selected Ion Recording Data, MPPSIRD, Profile Generation Model, PGM, exact mass, high resolution mass spectrometry
Ion Composition Elucidation (ICE) often leads to identification of compounds and provides high quality evidence for tracking compounds to their sources. Mass spectra for most organic compounds are not found in mass spectral libraries used to tentatively identify analytes. In addition, multiple matches are common. ICE provides the numbers of atoms of each element in the ions in the mass spectrum, greatly limiting the number of possible compounds that could produce the mass spectrum. Review of chemical and commercial literature then limits the number of possible compounds to one or a few that can be purchased to confirm tentative compound identifications by comparison of mass spectra and chromatographic retention times. ICE is conceptually simple relative to other analytical techniques and more easily explained to a judge or jury. ICE is based on sums of the exact masses of atoms and their isotopic abundances. Several applications of ICE are demonstrated for ultra-trace-level compounds in an extract of the effluent from a tertiary sewage treatment plant including: (i) measurement of five values to determine an ion's composition and to generate evidence for the compound's identity, (ii) rejection of incorrect library matches, (iii) rapid screening for a target compound in an extract, and (iv) a strategy for tracking unidentified compounds to their sources.
Unequivocal identification of compounds provides an essential link in the chain of evidence for legal proceedings pertaining to crime scenes, drug syntheses, food contamination, or environmental contamination. A preponderance of evidence to support this link, including incontrovertible data documenting the elemental compositions of components in samples, would encourage both sides to accept compound identifications without argument.
Numerous laws regulate the permissible levels of many compounds in waste disposal sites, discharge water, and drinking water. For example, the Clean Water Act covers all discharges to surface waters including 126 priority toxic pollutants (Valente and Valente, 1995). However, total ion chromatograms provided by gas chromatography with mass spectrometric detection (GC/MS) for concentrated extracts from large volumes (at least 10 L) of discharge water from tertiary sewage treatment plants display dozens of chromatographic peaks, each of which corresponds to one or more compounds, because coelution is common when so many compounds are present. Although the concentration of each unidentified compound is probably too low to cause observable biological effects within the aquatic ecosystem into which the treated water is discharged, the summed effect of multiple compounds might have such effects (Daughton, Ternes, 1999). These compounds or byproducts arising from drinking water disinfection down stream might have deleterious affects on human fetuses (Coburn et al., 1997). Thus, it is important to identify these compounds so that their toxicology, alone and in mixtures, can be reviewed or studied. Remedial measures to reduce use of a few biologically potent compounds or to better remove them from sewage treatment plant effluents might be deemed necessary. If so, irrefutable compound identifications would be required.
GC/MS is often used to identify compounds based on pattern matching. A compound's mass spectrum contains ions with different mass-to-charge ratios (m/z). If three or more ions are present, the compound has been isolated from all others, and the compound's mass spectrum is in the library of mass spectra on the mass spectrometer's data system, a single library match can lead to purchase of the tentatively identified compound for confirmation based on its mass spectrum and retention time. Unfortunately, compared to the universe of organic compounds (at least 630,000 are commercially available [CAS, a] and more than 16 million organic and inorganic compounds have been synthesized or characterized [CAS, b]), the mass spectra of only a limited number of organic compounds are available in mass spectral libraries (107,886 in the NIST library and 390,000 in the Wiley library [Wiley Registry], which include the NIST mass spectra [Wiley, http]). In addition, the majority of organic compounds are ionic, too polar, too thermolabile, or too high in mass to traverse a GC column or to volatilize from a probe inserted directly into the ion source of a mass spectrometer. These compounds are usually introduced into mass spectrometers in solution rather than in the gas phase, and electrospray ionization or atmospheric pressure chemical ionization creates the ions for analysis. These are "soft" ionization techniques that provide few, if any, of the fragment ions required for pattern matching with mass spectra from standards and formation of adduct ions often complicates the mass spectra. Consequently, low resolution mass spectrometry is often inadequate for compound identification.
"Exact" masses of ions weighing less than 150 Da measured to within 6 parts-per-million (ppm) that contain C, H, N, O, P, or S atoms usually provide unique ion compositions, i.e., the list of elements and the number of atoms of each element that compose the ions. High resolution mass spectrometers can measure "exact masses" of ions to within several parts-per-million (ppm), but most compounds of regulatory interest yield ions heavier than 150 Da. The number of possible compositions for an ion increases rapidly with its mass, the number of elements considered, and the error limit of its exact mass determination (Grange and Brumley, 1997). Although most organic compounds are composed of C, H, N, O, P, or S atoms, other elements including F, Cl, Br, and Si must be considered for compounds of unknown origin. For commercial or illicit, synthetic products, all elements in the reactants, solvents, and catalysts must be considered. For unidentified compounds in contaminated sites, wastewater, drinking water, or foods, the 10 elements cited above should be considered initially.
An analytical methodology that characterizes or identifies compounds that produce ions heavier than 150 Da could have regulatory benefits. Recently, the Environmental Sciences Division (ESD) of the National Exposure Research Laboratory of the US EPA developed Ion Composition Elucidation (ICE) to determine the unique compositions of ions with masses up to 600 Da (containing C, H, N, O, P, or S atoms) based on atomic masses, relative abundances of higher isotopes, and the valences of elements (Grange et al., 1996a, 1997). A sample's history, chemical properties such as solubility and stability in the environment, transport properties through soils, and commercial aspects such as the amount of chemicals produced that could yield the observed mass spectrum need not be considered until the composition of the major ions in the mass spectrum have been determined. ICE was used to characterize a tar-like sample from a Superfund site (Grange and Brumley, 1996b) and to identify isomeric pollutants in municipal well water from Toms River, NJ (Grange and Sovocool, 1998). The tar-like sample consisted primarily of benzothiazole based compounds, which are used by dye and rubber producers. A nearby dye plant was responsible for the waste. For the well pollutants, tables of the exact masses of 10 fragment ions and the corresponding composite neutral losses from the molecular ion were used to build a detailed fragmentation pattern for the isomers. (Similar tables and fragmentation patterns were compiled for two synthetic products introduced into the mass spectrometer on a probe [Grange and Sovocool, 1999]). This data provided a preponderance of evidence and the principal responsible party accepted the compound identifications without argument and voluntarily provided authentic standards from the polymerization process that produced these byproducts for confirmation through comparison of retention times and mass spectra. This last step is required to conclusively identify a compound, but some polluters might not be this cooperative. In this paper, evidence sufficient to obtain a court order to obtain samples from a potentially responsible party is developed using ICE.
In legal proceedings, judges and juries are best persuaded by scientific arguments that are conceptually simple and concise. ICE has been used in a manner that minimized the data acquisition required to determine ion compositions. Alternative data collection and interpretation strategies that are simpler to explain will be demonstrated for tentative identification of several compounds found in effluent from a tertiary sewage treatment plant.
All data presented herein were obtained for a hexane extract of 12 L of water collected beneath the outflow pipe of a tertiary sewage treatment plant that services an urban area. The water sample was transported in an ice cooler to an ESD laboratory and extracted within hours onto a 47-mm diameter, C18 solid phase extraction disk using a CPI Accu*prep 7000 extractor (both CPI International, Santa Rosa, CA). The analytes were eluted from the disk with 20 mL of n-hexane (Burdick & Jackson, B&J GC2, Muskegon, MI), which was then dried over oven-baked Na2SO4 (EM Science, Gibbstown, NJ). The extract was concentrated to 1 mL with a Turbo Vaporization 500 Workmate (Zymark, Hopkinton, MA) and cleaned up on 1 g of silica gel within a Bond ElutTM extraction cartridge (Varian, Harbor City, CA). Ten mL each of hexane and ethyl acetate (Burdick & Jackson, B&J GC2, Muskegon, MI) eluted organic compounds from the cartridge. This eluent was concentrated to 300 mL. Fifty mL of hexane were added and evaporated to 1 mL to remove most of the ethyl acetate.
Injections of 1 or 2 mL were made onto a 30-m, 0.25-mm i.d., 0.25-mm film thickness, XTITM-5 GC column (Restek, Bellefonte, PA) within an HP 6890 gas chromatograph (Agilent Technologies, Palo Alto, CA) interfaced to a Finnigan MAT 900S-trap double focusing/ion trap hybrid mass spectrometer (Finnigan MAT, Bremen, Germany). The ion trap was not used in these experiments. The temperature program was 60 oC for 1 min, 5 oC/min to 140 oC, 25 oC/min to 200 oC, 3 oC/min to 240 oC, 10 oC/min to 300 oC, hold 1 min. The He flow rate was 0.1 mL/min to widen chromatographic peaks to 6 s. The injector temperature was 250 oC, and the transfer line was maintained at 300 oC. For one experiment, 2 mL of the extract were injected onto an insertion probe. The hexane solvent was removed in a roughing vacuum region before the probe was rested against the ion source block. The probe was then heated from 25 oC to 250 oC at 200 oC/min to volatilize the previously dissolved compounds.
Ions produced by electron impact (70 eV) were detected with 1.6, 1.7, and 2.1 kV applied to the secondary electron multiplier when 1000, 3000, and 10,000 mass resolutions were used, respectively. Full scans (m/z 50-300 and m/z 100-600) at 1.76 s/scan were acquired with low mass resolution (1000) using an accelerating potential of 5 kV to provide nominal masses for the presumed molecular ion (M+) and for abundant fragment ions. MPPSIRD was used to survey wide mass ranges (2000 ppm) with 3000 resolution and then to determine exact masses and relative abundances for ions with 10,000 resolution.
RESULTS AND DISCUSSION
ICE has two facets, data acquisition and data interpretation. A high resolution mass spectrometric technique, Mass Peak Profiling from Selected Ion Recording Data (MPPSIRD) (Grange et al., 1994, 1996), acquires the data. Full scanning and selected ion recording are the most common types of mass spectrometric data acquisition. Full scanning samples all m/z ratios across a wide mass range. When high mass resolution is used, each scan requires a number of seconds, and 6-s wide chromatographic peaks cannot be delineated. Selected ion recording (SIR) monitors sequentially, only the maxima of targeted mass peak profiles. For each SIR cycle, a lock mass is first scanned to determine its maximum, which continually compensates for calibration drift. The data system uses this information to ensure the correct m/z ratios are measured each cycle. The full mass spectrum is no longer obtained, but sensitivity is increased 100-fold and six or more samplings during elution of each 6-s wide chromatographic peak are provided for each m/z ratio. Detection limits are lower, and higher mass resolution can be used to discriminate against interferences. However, no mass measurements are made, and if a mass interference is present, a single point monitored atop each profile provides no indication of its presence. A mass interference arises from a similar or greater number of ions with a different composition having nearly the same exact mass as the analyte ion.
MPPSIRD is a hybrid of these scan techniques that realizes the advantages of both. Multiple m/z ratios are monitored sequentially across each of two calibrant profiles and each of up to four analyte profiles to provide a series of ion chromatograms. In Figure 1a, the left-most ion chromatogram is for an analyte ion and the right-most is for a calibrant ion.
Figure 1. (a) Ion chromatograms for the m/z ratio corresponding to the maxima for the m/z 198 and lock mass partial profiles. (b) Partial mass peak profiles for PFK calibrant ions with m/z 193 and 205 and for m/z 198, 198+1, and 198+2 analyte ions.
Integrating and plotting the areas under the multiple ion chromatograms monitored for each profile provide the full or partial profiles. Partial profiles are shown in Figure 1b. The weighted average of the top several areas used to plot a profile provides its exact mass. The ratio of the sums of the areas used to plot two profiles provides the abundance of one relative to the other. A Gaussian (bell-shaped) profile usually indicates a lack of interferences (Grange et al., 1994). Ion chromatograms of analytes (left trace in Figure 1a) delineate chromatographic peaks as they elute from the gas chromatograph into the ion source of the mass spectrometer, while the ever-present calibrant, perfluorokerosene (PFK), provides steady signals in the ion chromatograms of the lock mass and calibration ions (right trace in Figure 1a). Baseline excursions are induced in the ion chromatograms by reversing the polarity of an ion source voltage for 5 s before and after each analyte elutes. The data system then integrates the area between the baseline excursions for all monitored m/z ratios before they are plotted to provide the profiles.
MPPSIRD's advantages relative to full-scanning, high-resolution mass spectrometry are greater mass accuracy and precision, greater relative abundance accuracy and precision, 100-fold greater sensitivity, 1-s sampling cycles, and mass recalibration each cycle to compensate for calibration drift.
Profile Generation Model
A Profile Generation Model (PGM) is used to plan experiments and interpret the data (Grange and Brumley, 1997). The PGM lists the possible compositions corresponding to an exact mass and its error limits. When multiple compositions are possible, the exact masses of the ions heavier by 1 and 2 Da that arise from heavier isotopes, such as 13C, 2H, 15N, 17O, 18O, 33S, and 34S and the abundances of these heavier ions relative to the ion containing only the most abundant isotope of each element are also considered. By testing criteria based on five quantities (three exact masses and two relative abundances), rather than on a single exact mass, the PGM greatly reduces the number of possible compositions for an ion.
Together, MPPSIRD and the PGM are responsible for the 4-fold increase in the mass range for which unique ion compositions can be determined from data acquired as chromatographic peaks elute.
ICE simplifies mass spectral interpretation. Once ion compositions are determined for the ions in a mass spectrum, the number of possible structures for each ion is reduced. However, a knowledge of organic chemistry is still important to predict which possible compounds would survive in water, air, soil, or during waste treatment. The software for MPPSIRD performed on a Finnigan MAT 900 or MAT 95 mass spectrometer and the PGM are available from the authors at no cost.
An ion's composition is determined through multiple experiments. First, a low resolution mass spectrum is obtained. An example is displayed in Figure 2a.
Figure 2. (a) A low resolution mass spectrum. (b) Full profiles each defined by three points acquired with 3000 resolution over a 2000-ppm mass range to survey a wide mass range about the analyte ion. The ion chromatogram inserts correspond to the maxima in the two profiles. (c) A full profile for the m/z 133 analyte ion obtained with 10,000 resolution and a mass increment of 10 ppm to provide 10 points across the profile. (d) Full m/z 133 and 133+1 profiles acquired with 10,000 resolution.
The first ion chosen for investigation is usually the apparent molecular ion, i.e., the highest-mass ion containing only the most abundant isotopes of the elements. For Figure 2b, m/z ratios were monitored across a wide mass range with 3000 mass resolution to ensure the most abundant analyte ion (at m/z 133) was further investigated and to obtain an estimate of its exact mass. If higher mass resolution were used initially, the ion investigated could be a less abundant m/z 133 ion from a partially coeluting compound rather than from the more abundant analyte ion of interest outside the mass range monitored. This data eliminates that argument. Two profiles, each delineated by three points, were evident. The ion chromatograms corresponding to the maxima of the two profiles shown in Figure 2b (displayed as insets) indicated that the lower-mass profile corresponded to an ever-present ion (possibly 13C2CF5+ from PFK) and to a compound that eluted from the GC to provide a chromatographic peak.
Next, the exact mass estimate from the analyte profile in Figure 2b was used as the center mass in a SIR descriptor to acquire data using 10,000 resolution with a mass increment of 10 ppm to provide 10 points across the full profile in Figure 2c. The exact mass determined as the weighted average of eight points used to plot the profile in Figure 2c was entered into the PGM to provide Table 1 with nine possible compositions.
a Rings and double bonds.
Atoms of C, H, O, N, F, P, S, and Si were considered. Cl and Br were not, because the relative abundance of the +2 profile in the low resolution mass spectrum was much less than 33%. The isotopic abundances of 37Cl and 81Br are 24.2% and 49.3%, respectively (CRC Handbook, 1994). These isotopic abundances provide relative abundances in the mass spectrum of 33% and 98% for the +2 profile compared to the profile for the ion containing only the 35Cl and 81Br isotopes.
When multiple ion compositions are possible, a final experiment is required to obtain additional data. In past experiments, partial profiles had been plotted as illustrated in Figure 1b for a different analyte ion with m/z 198. To acquire the necessary data, seven m/z ratios were monitored for each analyte profile across 60% of the mass range of the ion's profile and the profiles heavier by 1 and 2 Da, and five m/z ratios were monitored across 40% of the mass range of each of two calibration ions. The m/z ratios monitored bracket the profile maxima and check for interferences across the parts of the profiles that provide the most signal. More closely spaced points across the profiles provide better estimates of the exact masses and ensure signals are measured at each m/z ratio. All 31 m/z ratios provided by the data system were used. The mass resolution was 10,000 and the mass increment was 10 ppm. The three exact masses and two relative abundances were entered into the Profile Generation Model (PGM) to provide Table II.
Table 2. Profile generation model output for five measured values for fragment ion m/z 198.05527 + to - 6 ppm; 10,000 resolution; elements considered: C, H, N, O, F, P, S, and Si
The list of 32 compositions are those possible based on the exact mass of the ion within the error limit of its determination. Each "X" next to an entry indicates that the measured and calculated values are inconsistent for the composition in that row. Only for the last composition are all five pairs of values consistent. All other possible compositions have been rejected based on failure of one or more comparisons. Beyond a reasonable doubt, only the last composition can be correct. A single excellent NIST library match for 2,2'-dinitro-1,1'-biphenyl suggested this ion is a fragment ion resulting from loss of NO2 from one of the rings. The molecular ion is absent. Unknown to the mass spectroscopist during these analyses, this compound had in fact been added as a surrogate standard.
Six of the nine compositions in Table 1 have negative numbers of rings and double bonds (RDB). These compositions would not be listed if elemental valences had been considered by the PGM as for Table 2. Ranges of RDB occur when N, P, or S atoms and O atoms contribute to a possible composition. Without O atoms present in an organic compound, N, P, and S have valences of 3, 3, and 2, respectively. If enough O atoms are present, N, P, and S bound to O atoms can have valences of 5, 5, and 4 or 6. In Table 1, elemental valences were not considered to simplify arguments. In Table 3, the RDB column has been removed and only sums of exact masses and relative abundances of isotopes would need to be explained to a judge or jury. The exact mass and relative abundance of the +2 profile were not required to establish the correct composition.
Table 3. Profile generation model output (with RDB column removed) for measured values from Figure 2(d). m/z = 133.06408 + to - 6 ppm; 10,000 resolution; elements considered: C, H, N, O, F, P, S, and Si
Full or Partial Profiles
For data acquisitions, exact mass error limits of 6 ppm and 3 ppm for profiles obtained with 10,000 and 20,000 resolution, respectively, are used by the PGM. The relative abundance ranges in Table 2 result from consideration of (i) instrumental precision, (ii) isotopic abundance variation for each composition, (iii) offset error in monitoring the profiles if the wrong hypothetical composition is chosen to calculate the center masses of the profiles to be monitored, (iv) up to one mass increment of offset error by the data system, and (v) up to 10% error in the resolution. The last three errors are the most difficult to explain and are eliminated when full profiles are monitored. Thus, for legal proceedings, only full profiles should be obtained. Two full profiles are shown in Figure 2d for the m/z 133 ion and its +1 profile.
Mass ranges sufficient to plot the full profiles were monitored. The horizontal line corresponds to 5% of the profile maximum. When this line can be drawn, a full profile has been monitored that is free of significant interferences. The disadvantages of obtaining full profiles over partial profiles are (i) two data acquisitions are required rather than one to obtain all five values, (ii) interferences in the tail regions of full profiles can occur, and (iii) broadened +1 or +2 profiles due to the analyte are observed for low mass ions at 10,000 resolution. However, for ultra-trace concentrations of analytes in extracts containing dozens of compounds that contain no Cl or Br atoms, the +2 profile signal is generally lost in the chemical noise or obscured by ions from coeluting compounds. When only tail interferences occur, a full profile can be treated as a partial profile by considering only the top five or six points. However, the most easily defended exact masses and relative abundances are obtained from full profiles free of interferences.
A background subtracted mass spectrum of an analyte in the extract (Figure 3a) and the best NIST library match (Figure 3b) for terbutylazine, an herbicide, are shown.
Figure 3. (a) A background subtracted mass spectrum for a compound in a tertiary sewage treatment plant extract. (b) The best NIST library match for terbutylazine. (c) Full m/z 229 and 229+1 profiles. (d) Full m/z 229 and 229+2 profiles. The resolution was 10,000.
Because a Cl atom is included in the molecular ion, the +2 profile in Figure 3d had sufficient signal for the 5% of maximum line to be plotted. The five measured values were only consistent with the values calculated for C9H16N5Cl+ and its isotopically related ions (229.10942, 230.11187, 231.10663, 12.0%, and 32.6%). C9H16N5Cl+ is the composition of the molecular ion of terbutylazine. ICE provided evidence in addition to the library match to justify purchase of terbutylazine for probable confirmation of its identity.
The estimated errors used in the PGM were obtained using a VG 70SE double focusing mass spectrometer, which relies on only one calibration mass during data acquisition (Grange and Brumley 1997). The Finnigan MAT 900S uses both a lock mass and a second calibrant mass to provide lower error limits. Hence, the PGM provides generous error limits to ensure no compositions are erroneously rejected. For legal proceedings, seven or more full profile experiments to determine exact masses and relative abundances for an analyte are recommended to estimate the standard deviations (sigma) for the measurements. The Student's T test predicts the range for the population mean as xmean + to - sxt/-, where xmean is the measured mean, sx is the estimated standard deviation, t is 5.959 for 99.9% probability and 7 experiments, and n is the number of experiments (Calculator). There is only 1 chance in 1000 that the true mean falls outside the range, + to - 2.25sx. If the calculated value for a possible composition falls outside this range, the composition can be rejected with this level of certainty.
A control free of potential analytes (in this case 12 L of de-ionized water) was collected, extracted, and cleaned up exactly as were the samples. The absence of analytes in the blank would refute the argument that the compounds found in the treatment plant extract were laboratory contamination.
Examples of Analytical Problems Addressed by ICE
Non-volatile compounds entering urban sewage systems must pass through treatment plants. Influent samples provide a catalog of compounds discharged into the system, while the effluent provides a list of compounds remaining after treatment that will flow into aquatic ecosystems. The multitude of ultra-trace level compounds in the tertiary treated effluent provided numerous analytical problems for illustrating the utility of ICE.
Rejecting multiple library matches
In Figure 4 are seven similar mass spectra.
Figure 4. (a) A background subtracted mass spectrum for a compound in a tertiary sewage treatment plant extract. (b) to (g) NIST library matches corresponding to 10 compounds with three different compositions for the m/z 133 ion.
The top mass spectrum (the same as in Figure 2a) is a background subtracted mass spectrum for a compound in the effluent from a tertiary waste treatment plant; the others are from the NIST library for the same mass range (NIST). The isomers in parenthesis in Figure 4 also had similar NIST mass spectra. The compound that provided the mass spectrum was present in the extract at an ultra-trace level. Chemical noise and coelution of others of the dozens of compounds in the extract and septum bleed components generally result in background subtracted mass spectra containing extraneous ions or lacking low-abundance ions expected from the analyte. Hence, none of the NIST library matches can be ruled out without additional data. The full profiles in Figure 2d provided the exact mass of the molecular ion (M+) and the exact mass and relative abundance of the (M+1)+ profile. These values entered into the PGM provided Table 3, which identified C7H7N3+ as the composition of the molecular ion and thereby rejected compounds with the other two compositions, C8H7NO+ and C9H11N+. Only the first of the two isomers on Figure 4b, 5-methyl-1H-benzotriazole, is associated with a commercial use. It is a derivative of 1H-benzotriazole used as a corrosion inhibitor in airplane deicers. (Callender) . ICE narrowed the number of possible compounds from 10 to three, and commercial use considerations then suggested a single compound to be purchased for confirmation. Both this compound and possibly terbutylazine as discussed above were found in an extract of 100 L of water from the lake into which the treatment plant effluent empties (Snyder et al., 2001).
The question of why the three compositions for the similar mass spectra provide different relative abundances is answered simply by examining Figure 5.
Figure 5. With 10,000 resolution, calculated
individual contributions from ions containing an atom of different +1
isotopes to the composite +1 profile for compositions:
(a) C7H7N3+, (b) C8H7NO+, and (c) C9H11N+.
Each +1 profile is a composite resulting from contributions from ions containing +1 isotopes of the elements composing the molecular ion. For the +1 profile in Figures 2d and 5a, the probability that any one of the seven 12C atoms is instead a 13C atom is 1.1%. Because there are seven C atoms, this likelihood contributes 7.7% to the total +1 relative abundance. Likewise, the probability that one of seven H atoms could be a 2H atom contributes 7 x 0.015%, and the chance that one of three N atoms could be a 15N atom contributes 3 x 0.38% to provide a sum of 8.9%. The different numbers of these atoms and the presence of an O atom in C8H7NO+ yield different relative abundances (9.3% and 10.4%) for the other two compositions (Figures 5b and 5c). Figure 5a indicates that the enhanced low-mass tail of the +1 profile in Figure 2d is due primarily to the 15N containing +1 ion, rather than to an interference. The +1 profile in Figure 3c shows a similar low-mass tail for the same reason.
Providing confirmation for a mass spectrum having few ions
Shown in Figure 6a is a background subtracted mass spectrum for a compound in the same extract.
Figure 6. (a) A background subtracted mass spectrum for a compound in a tertiary sewage treatment plant extract with only one prominent ion. (b), (c) The two best NIST library matches for hydroquinone and resorcinol. (d) Full m/z 110 and 110+1 profiles acquired with 10,000 resolution.
Only one prominent ion is observed. In Figure 6b is the NIST library match for hydroquinone, used as a photographic developer and reducer, an anti-oxidant (Merck), and in skin ointments to treat hyperpigmentation (rxlist). In Figure 6c is the NIST library match for resorcinol, used similarly to treat acne and eczema, as a UV absorber in resins (chemicalland), in tanning, and in dyeing and printing textiles (Merck). But, based on addition of the integer masses of C, H, N, O, F, P, S, and Si, 276 compositions are possible for the m/z 110 ion. Four compositions are possible for the measured exact mass of 110.03669 Da. The right-most analyte profile in Figure 6d provided the exact mass and relative abundance of the +1 profile. The measured values entered into the PGM unequivocally confirmed the ion composition as C6H6O2+, which is consistent with the library matches. The theoretical exact masses and relative abundance for C6H6O2+ are: 110.03678 Da, 111.04018 Da, and 6.8%. Both hydroquinone and resorcinol should be purchased to determine if one of these compounds is the analyte. Because both hydroquinone and 5-methylbenzotriazole are components of x-ray film developer, a major source of these compounds could be hospitals, clinics, medical laboratories, and dental offices.
Rejecting erroneous matches
Figures 7a and 7b display a background subtracted mass spectrum and an NIST library match for a compound commonly called coumarin 1, a brightener for nylon, acetate, and wool fabrics (chemfinder). The full profile in Figure 7c provided an exact mass of 231.13897 Da, which was 56 ppm different than 231.12593 Da, the exact mass of the molecular ion of coumarin 1. Clearly, this analyte was not coumarin 1.
Figure 7. (a) A background subtracted mass spectrum for a compound in a tertiary sewage treatment plant extract. (b) The best NIST library match for coumarin 1. (c) A full m/z 231 profile acquired with 10,000 resolution.
Screening for a compound
MPPSIRD is a rapid, sensitive, and highly selective investigative tool for determining if compounds might be present when extracts are introduced into the mass spectrometer with a heated insertion probe. Figure 8a is a background subtracted mass spectrum for an analyte in the treatment plant extract and Figure 8b is the best NIST library match for tris(1,3-dichloro-2-propyl)phosphate.
Figure 8. (a) A background subtracted mass spectrum for a compound in a tertiary sewage treatment plant extract. (b) The best NIST library match for tris(1,3-dichloro-2-propyl)phosphate. (c) The ion chromatogram corresponding to the maximum point in the m/z 379 profile. A broad volatilization peak is present. (d) Partial lock mass profile, full m/z 379 profile, full +2 profile, and partial calibration mass profile plotted from data acquired with probe introduction of the extract.
This flame retardant was also found in a 100-L extract of lake water collected downstream from the sewage treatment plant (Snyder et al., 2001) and on pine needles in the Sierra Nevada mountains (Aston, 1996). M/z ratios were monitored with 10,000 resolution across the profiles of the m/z 379 fragment ion (378.89941 Da) and its +2 profile (380.89652 Da) as the compounds in 2 mL of the extract were volatilized from an insertion probe into the ion source of the mass spectrometer. Figure 8c shows a broad volatilization peak for the m/z ratio corresponding to the maximum in the m/z 379 profile shown in Figure 8d. The measured exact mass of the fragment ion (378.90091 Da, +3.9 ppm error) and the measured exact mass and relative abundance of the +2 profile (380.89732 Da, +2.1 ppm error and 157.4% vs. 161.1%) were consistent with the target compound. Based on their exact masses, the high mass resolution with high sensitivity afforded by MPPSIRD permitted separation of the weak signals from this compound's ions to be isolated from the signals of more abundant ions formed from all of the compounds in the extract as they volatilized from the probe. This approach could be used to track the extent of plume travel. A negative finding for an extract would indicate that quantification for the target compound is unnecessary. To estimate the detection limit for a target compound and to verify that the compound would remain after sample extraction and clean-up, a blank should be spiked with a low level of the compound and processed identically to the samples.
Tracking an unidentifiable compound to its source
In Figure 9a is shown a background subtracted mass spectrum that provided no NIST library matches.
Figure 9. A background subtracted mass spectrum for a compound in a tertiary sewage treatment plant extract. (b) to (e) Ion chromatograms for m/z ratios corresponding to the profile maxima for ions determined to have the compositions shown. The composition of (d) was assumed to be that of (e) after loss of a methyl group.
ICE determined the ion compositions of three of the prominent ions, and the m/z 252 ion was assumed to result from loss of a methyl group from the m/z 267 ion. In Figures 9b - 9e are shown ion chromatograms for the exact masses of these ions acquired with 10,000 resolution and a cycle time of 0.35 s. To reduce the cycle time from 1 s, only the maxima of the calibration and the four analyte ions were monitored. The four analyte ion chromatograms track each other very well suggesting that all four ions are produced from the same compound. This compound could be present in the influent or be a decomposition product of the biodegradation treatment process. The influent to the treatment plant could be checked for this compound, and samples collected along the sewage system could locate a single or limited number of sources of the compound or determine if the compound is discharged ubiquitously by households. For a single source, the sewage trail, similar mass spectra and retention times, overlapping of all four ion chromatograms obtained with high mass resolution, and the same ion compositions for the compound at the suspected source and in the effluent would provide sufficient evidence to obtain a court order (if necessary) to acquire samples within the property of the suspected source. Amounts of the compound sufficient to acquire data using other analytical techniques such as NMR, FTIR, or X-ray crystallography would lead to its identification if the compound was an unidentified byproduct of an industrial synthesis. This body of evidence would be compelling in any subsequent legal proceeding.
Availability of ICE
The Environmental Chemistry Branch of the EPA-Las Vegas will consider assisting in the characterization of "mystery" compounds in a limited number of extracts others have prepared from environmentally important samples. In addition, the computer code and manuals describing ICE are available for Finnigan MAT900 or MAT95 double focusing mass spectrometers.
Ion Composition Elucidation (ICE) provides unequivocally the elemental compositions of ions in a mass spectrum. This knowledge greatly reduces the number of possible compounds that could produce a mass spectrum not found in mass spectral libraries. One can conclusively refute as candidate compounds those that provide similar low resolution mass spectra containing a molecular ion or a fragment ion with a different ion composition. Review of the chemical and commercial literature can further limit the probable identity of an analyte to one or a few compounds. To be found in the environment, large amounts of the compound or a precursor compound are probably synthesized commercially and used by industrial plants (point sources), numerous businesses (multiple sources in a region), or the public (ubiquitous sources). The remaining candidate compounds can be obtained for confirmation of the analyte's identity through comparison of mass spectra and retention times for chromatographic techniques. If a compound is suspected to be a waste product from a single source, ICE should provide sufficient evidence to obtain a court order directing the generator of the compound to provide a sample for analysis. ICE evidence is based on sums of atomic masses and isotopic abundances, and is therefore simpler to explain to judges or juries than evidence derived from energy-based spectroscopic techniques.
(Aston, 1996) Aston, L.S., Noda, J., Seiber, J.N. and Reece, C.A. 1996. "Organophosphate Flame Retardants in Needles of Pinus ponderosa in the Sierra Nevada Foothills" J. Environ. Contam. Toxicol., 859-866.
(CAS, b) http://www.cas.org/casdb.html
(chemfinder) http://www.chemfinder.camsoft.com/(for coumarin 1).
(Coburn, 1997) Coburn, T., Dumanoski, D. and Myers, J.P. 1997. Our Stolen Future: Are We Threatening Our Fertility, Intelligence, and Survival? New York: Plume/Penguin.
CRC Handbook of Chemistry and Physics 74th ed., Section 1, pp. 10-11, Lide, D.R. editor, Boca Raton, FL, CRC Press.
Daughton, C.G.; Ternes, T.A. "Pharmaceuticals and Personal Care Products in the Environment: Overarching Issues and Overview," in Pharmaceuticals and Personal Care Products in the Environment: Scientific and Regulatory Issues; Daughton, C.G. and Jones-Lepp, T., Eds.; American Chemical Society: Washington, DC (in press 2001).
(Grange et al., 1994) Grange, A.H.; Donnelly, J.R.; Brumley, W.C.; Billets, S.; Sovocool, G.W. 1994, "Mass Measurements by an Accurate and Sensitive Selected Ion Recording Technique" Anal. Chem., 66, 4416-4421.
(Grange et al., 1996a) Grange, A.H.; Donnelly, J.R.; Sovocool, G.W.; Brumley, W.C. 1996a "Determination of Elemental Compositions from Mass Peak Profiles of the Molecular Ion (M) and the M+1 and M+2 Ions" Anal. Chem. 68, 553-560.
Grange, A.H. And Brumley, W.C. 1996b. "Identification of Ions Produced from Components in a Complex Mixture Using Mass Peak Profiling" LC GC, 14, 978-986.
Grange, A.H., Brumley, W.C. 1997. "A Mass Peak Profile Generation Model to Facilitate Determination of Elemental Compositions of Ions Based on Exact Masses and Isotopic Abundances" J. Amer. Soc. Mass Spectrom. 8, 170-182.
Grange, A.H., Sovocool, G.W. , Donnelly, J.R., Genicola, F.A., And Gurka, D.F. 1998. "Identification of Pollutants in a Municipal Well Using High Resolution Mass Spectrometry" Rapid Commun. Mass Spectrom., 12, 1161-1169.
Grange, A.H. And Sovocool, G.W. 1999. "Mass Peak Profiling from Selected Ion Recording Data as a Tool for Regulatory Analysis" J. AOAC Internat., 82, 1443-1457.
Merck Index 11th edition, Budavaari, S.; editor, 1989, Merck, Rahway, NJ.
NIST/EPA/NIH Mass Spectral Library, Version 1.6d, Standard Reference Data Program, National Institute of Standards and Technology, US Dept. of Commerce, NIST Mass Spectrometry Data Center, Gaithersburg, MD 20899, 1998.
(Snyder et al., 2001) Snyder, S.A., Kelly, K.L., Grange, A.H., Sovocool, G.W., Snyder, E.M. And Giesy, JP 2001, "Pharmaceuticals and Personal Care Products in the Waters of Lake Mead, Nevada" in Pharmaceuticals and Personal Care Products in the Environment: Scientific and Regulatory Issues; Daughton, C.G. and Jones-Lepp, T., Eds.; Symposium Series 791; American Chemical Society: Washington, DC, Ch. 7, 116-141.
Valente, C.M. and Valente, W.D. 1995 Introduction to Environmental Law and Policy St. Paul, MN, West Publishing Co.
- Whittsitt II, R.E., Kelly, K.A., LaMont, M.D., Oliva, R.A., Hoffman, R.I., Zastrocky, M., Reed, J.F. And Battaglia, S. 1981. Calculator Decision-making Sourcebook, 2nd Edn. Texas Instruments, Inc.
- Wiley Registry of Mass Spectra, 6th ed. Palisade
Corp., Newfield, NY. (wiley)
Notice: The US 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.