Ion Composition Elucidation (ICE)
Identification of Pollutants in a Municipal Well Using High Resolution Mass Spectrometry
Andrew H. Grange,1* G. Wayne Sovocool,1
Joseph R. Donnelly,2
Floyd A. Genicola3 and Donald F. Gurka1
1USEPA, NERL, Environmental Sciences Division,
P.O. Box 9348, Las Vegas, NV 89193-3478, USA
2Lockheed Martin Environmental Services,
980 Kelly Johnson Drive, Las Vegas, NV 89119, USA
3New Jersey Dept. of Environmental Protection, Office of Quality Assurance,
9 Ewing Street, Trenton, NJ 08625, USA
*Grange currently holds a National Research Council/NERL-ESD-LV Senior Research Associateship.
SPONSOR REFEREE: Dr. Ron Hass, Triangle Laboratories, Durham, N.C., USA
An elevated incidence of childhood cancer was observed near a contaminated site. Trace amounts of several isomeric compounds were detected by gas chromatography/mass spectrometry (GC/MS) in a concentrated extract of municipal well water. No matching library mass spectra were found and Fourier transform IR and NMR analyses were not feasible due to the low concentration of the compounds. Mass peak profiling from selected-ion-recording data (MPPSIRD) provided the sensitivity and scan speed necessary to acquire mass peak profiles at mass resolutions of 10,000 to 20,000 for the molecular ion (M+) and 10 fragment ions as capillary GC peaks eluted. Using a profile generation model (PGM), the elemental composition of the molecular ion was determined from the exact masses and abundances of the M, M+1 and M+2 profiles. Fragment ion compositions were determined from their exact masses based on the elements in the molecular ion. Exact mass differences between the molecular and fragment ions corresponded to unique combinations of atoms for the neutral losses. Consequent reduction of the number of possible structures for the fragment ions simplified mass spectral interpretation. After inspecting library mass spectra for smaller molecules, isomeric structures were hypothesized with cyano and alkylcyano groups attached to tetralin. A literature search found such isomers produced by an industrial polymer synthesis. Three isomers in a standard from polymerization of styrene and acrylonitrile provided the same mass spectra and GC retention times as isomers in the extract.
An elevated incidence of childhood cancers near Toms River, NJ,(1) raised concern among citizens over environmental contaminants. The nearby Reich Farm Superfund Site is a relatively flat, sandy area of approximately three acres, surrounded by commercial establishments and woodland. In 1971, an independent waste hauler leased the site and dumped drums containing solvents, still bottoms and residues from organic chemical manufacturing. A well 1 mile from the site served 50,000 people.(2)
When analyzing chemicals dumped at or migrating from a Superfund site located within 100 miles of numerous industries that generate waste, one does not know a priori which elements are contained in the compounds. For industrially generated organic compounds, C, H, N, O, P, S, Si, F, CI and Br should all be considered.
In legal proceedings, identification of contaminants is required to assign responsibility for dumping of wastes. High resolution mass spectrometry (HRMS) can provide the elemental composition of a compound, which limits its identity to a number of structural isomers. Knowing the elemental compositions of fragment ions and of neutral losses from the molecular ion can greatly reduce the number of possible isomers. Further structural details can be provided by FTIR and NMR when the concentration of the compound is much greater than that needed to perform high resolution mass spectrometric (HRMS) studies. When concentration is limited, however, a search can be made of chemical inventories and for industrial processes that might generate one or more of the isomers consistent with HRMS data. Finally, mass spectra and chromatographic retention times of standards can be compared with those of environmental contaminants to confirm the identity of released chemicals.
To warrant investigation of a corporations's proprietary activities, strong evidence suggesting culpability is required. Determination of elemental compositions from fundamental physical properties of atoms, e.g. by addition of atomic masses and isotopic abundances, provides such strong evidence. Such data establishes the elemental compositions of contaminants, while eliminating all other possible compositions.
EPA analytical methods based on GC/MS are used to analyze many compounds dumped into the environment. Each method targets a list of particular compounds, and their mass spectra reside in mass spectral libraries included on many mass spectrometer data systems. Compounds not targeted are reported either as "tentatively identified compounds", when similar mass spectra are found in a library, or as "unknown" when similar spectra are not found. Compounds are not identified when low ion abundances or partial coelution of more abundant compounds yield distorted mass spectra, even after background subtraction. In addition, industrial chemical manufacturing processes create a myriad of waste compounds,3 most of which are not found in mass spectral libraries.
An extract of the municipal well water was prepared by the NJ Department of Environmental Protection. Similar mass spectra for isomers corresponding to four chromatographic peaks in the total ion chromatogram could not be matched with entries in either the National Institute of Standards and Technology library of mass spectra (62,235 entries)(4) or the Wiley library (262,000 entries).(5) The compounds in the extract could not be identified by conventional mass spectrometric techniques and their low concentrations precluded use of FTIR or NMR. Several laboratories, including the U.S. EPA's Environmental Sciences Division (ESD) in Las Vegas, received portions with a request that the compounds be identified. This report describes the methodology developed at the ESD to identify the major components in the sample.
Mass peak profiling from selected-ion-recording data (MPPSIRD), an HRMS technique developed by the Environmental Sciences Division of the EPA, provides over 100 times greater sensitivity and, at 20,000 resolution, a 6 times faster cycle time than conventional mass spectrometric scans.(6, 7) Consequently, exact masses and relative abundances are determined for ions from compounds that enter the ion source as chromatographic peaks. In addition, routine use of up to 20,000 resolution provides better discrimination against interferences than that provided by 10,000 resolution, which is specified for the EPA's HRMS dioxin methods.(8, 9)
Previously, a profile generation model (PGM)(10) was used to plan data acquisition and to interpret MPPSIRD data obtained at 20,000 resolution for the highest mass ion [M]+ containing only the most abundant isotope of each element in mass spectra of compounds corresponding to 47 chromatographic peaks in a different Superfund site sample.(3) Elemental compositions were determined for these 47 ions based on three exact mass and two relative abundance criteria for the M, M+1, and M+2 profiles. Most of the compounds were benzothiazoles, which confirmed that their source was a dye producer in the area, and identification of most of the compounds was not necessary.
For the well contaminants, MPPSIRD and the PGM were also used to determine compositions of numerous fragment ions and neutral losses produced from a group of isomers to aid in their identification. These compositions simplified interpretation of the low resolution mass spectra and reduced the number of possible structures for the isomers.
Well water extract
Water taken from the well was transported at 4 oC in a 1 L, amber-glass bottle with a TeflonTM-lined cap. The sample was preserved in the field by adding 6 N HCl to lower the pH to 2. EPA Method 525.2 was used to extract 1 L of the well water. The C18 extraction disk was eluted with 5 mL of ethyl acetate and then with 5 mL of methylene chloride. The combined extract was evaporated to 1 mL and provided a 1000-fold concentration of the analytes.
On-column injections of 0.5-2 mL of the well-water extract, depending on the ion abundance of the target ion, were made onto a 30 m, 0.25 mm I.D., 0.25 æ coated, Restek Rt x-1701 column (Bellefonte, PA, USA), which separated the target compounds from other components by at least 1 min. A Hewlett-Packard model HP 5890 gas chromatograph (Wilmington, DE, USA) was interfaced to a VG70-250SE double focusing mass spectrometer (Danvers, MA, USA). The temperature program was 170 oC for 1 min followed by heating at 4 oC/min to 280 oC. The analytes eluted between 232 oC and 237 oC.
Low resolution (1000) mass spectra were acquired using 70 eV electron impact ionization, a 500 mA filament current, an ion source temperature of 250 oC, and a photomultiplier voltage of 350 V. The photomultiplier voltage was increased to 390 V (to provide the maximum S/N ratio11) and to 500 V (to provide the maximum signal) when resolutions of 10,000 and 20,000 were used, respectively.
In Figure 1(a) are shown ion chromatograms for three m/z ratios that include the exact mass of an analyte ion (top three traces) and the m/z ratio for the exact mass of a calibrant ion (lock mass) produced from perfluorokerosene (PFK).
Figure 1. (a) Ion chromatograms acquired using 20,000 resolution (+ to -10%) for three m/z ratios near the maximum in the presumed molecular ion's profile (top 3 traces) and for the calculated exact mass of the calibration ion from PFK (bottom trace), (b) a full profile plotted from the chromatographic peak areas under 17 such ion chromatograms, and (c) partial profiles plotted from six ion chromatograms acquired at 10,000 resolution for m/z ratios across 60% of the mass range for the (M)+, (M+1)+ and (M+2)+ profiles of C14H14N2.
The PFK continually entered the ion source from a heated reservoir to provide a constant signal. The baseline excursions in the ion chromatogram for the calibrant were induced by partially closing and opening the valve between the source and analyzer regions of the mass spectrometer to block the ion beam. Automated procedures determined the area between the baseline excursions, which defined a simulated chromatographic peak for the ever-present calibration ion.
In Figure 1(b) is shown a full mass peak profile plotted from the areas under the chromatographic peaks in 17 ion chromatograms for m/z ratios that included 10 points across the profile. For molecular or fragment ions of any mass, the mass increment is 5 ppm (1 mmu for m/z 210) at 20,000 resolution, 10 ppm at 10,000 resolution and 20 ppm when 5000 resolution is used. Also plotted in Figure 1(b) from simulated chromatographic peak areas for 5 m/z ratios is the partial profile for the calibrant ion. The error in the exact mass determined for the calibrant ion is subtracted from the exact mass determined for the analyte. The initial calibration of a mass range that includes the analyte and calibration ions is based on PFK ions and is performed by the data system in the normal fashion.
Shown in Figure 1(c) are three partial profiles, each plotted from 6 m/z ratios that span 60% of the mass range of each profile. Four m/z ratios also monitored the top of the calibration ion's profile (not shown). Partial rather than full profiles were used to simultaneously monitor three analyte profiles because no more than 25 m/z ratios can be monitored by a SIR descriptor. For both full and partial profile data acquisitions, each of 22 m/z ratios was monitored for 30 ms and the total scan cycle was 0.8 s, rapid enough to provide 10 points across a chromatographic peak 8 s wide. The weighted average of several points across the top of each profile provided its exact mass. Exact masses and abundances of the M+1 and M+2 profiles relative to the M profile were determined from partial profiles. Relative abundances were determined as ratios of the sums of the six areas used to construct each partial profile.
Automated procedures run on a personal computer (PC) interfaced to the data system through an RS-232 cable prepare the SIR descriptors, process the data, and are as simple to use as data acquisition methods provided by the manufacturer. DOSTM (v3.1) batch files executed on the PC invoke sequentially programs in VG (vB2.2), the data system software; Lotus 123TM (v2.2), a spreadsheet; ReflectionTM (v4.2) for terminal emulation; and WordPerfectTM (v5.1) to plot the profiles.(12) No hardware modifications were required.
Low analyte concentration
Ion abundance decreases as the mass resolution is increased. So little of the target compounds was present that 20,000 resolution was used only to obtain an average exact mass for the presumed molecular ion at m/z 210 ion from full profiles. A resolution of 10,000 was used to acquire the M, M+1 and M+2 partial profiles from which the exact mass and relative abundance of the M+1 partial profile were obtained. Full profiles for 10 fragment ions were also plotted from data acquired with 10,000 resolution, since several of these ions were barely visible in the low resolution mass spectrum. To obtain a strong signal, the exact mass and relative abundance of the M+2 partial profile were determined from partial profiles acquired at 5000 resolution.
For each nominal mass, an estimate of the exact mass was first obtained by examining a mass range of 1600 ppm using a mass increment of 100 ppm with a resolution of 3000. Three points were acquired on each profile. A SIR descriptor was then prepared to acquire data at 10,000 resolution after a second injection. A third injection was made when 20,000 resolution was used. Two additional injections were made for triplicate determinations. Thus, from two to five injections were required for each ion. Baselines were induced using the valve between the source and analyzer before and after the first two isomers eluted and after the last two isomers eluted. In this way, two estimates of the exact masses and relative abundances were obtained for each injection.
RESULTS AND DISCUSSION
Determining the elemental composition of the presumed molecular ion
Background subtracted, low resolution (~1000) mass spectra were printed out for the four chromatographic peaks observed in the total ion chromatogram shown in Figure 2(a).
Figure 2. (a) The total ion chromatogram acquired at 1000 resolution in the full scan mode for the well-water extract and (b) a background subtracted mass spectrum for the fourth isomer.
An example is shown in Figure 2(b). The m/z 210 ion was the largest mass ion not associated with column bleed that contained only the most abundant isotope of each element. An average exact mass of 210.1160 u was obtained for this presumed molecular ion from full profiles acquired with 20,000 (+ to -10%) resolution for triplicate injections. One profile is shown in Figure 1(b). In Table 1, the possible compositions based on C, H, N, O, P, S, Si, and F atoms and the error limits for each resolution and number of determinations are listed for the three mass resolutions used in this study.
Table 1. Elemental compositions and quantities useful for distinguishing among them at three mass resolutions
aRings and double bonds.
bCalculated mass defect.
cBased on partial profiles centered about the calculated mass for the composition.
dBased on partial profiles centered about the calculated mass of the hypothetical composition, accounting for errors of up to + to -1 mass increment at + to -10% of resolution, and for isotopic abundance errors. An "X" indicates application of this criterion will reject this composition if the hypothetical composition is correct.
Chlorine, bromine and other elements having prominent isotopes were not considered, because no isotopic abundance patterns arising from the high relative abundances of 37Cl, 81Br, or other isotopes were observed in the low resolution mass spectra. Even at 20,000 resolution with an error limit of + to -2.5 ppm for triplicate determinations, five compositions were possible; the correct composition could not be determined from the exact mass of the molecular ion alone.
The four additional quantities listed by the PGM in Table 1 were used to identify the correct composition: 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. An "X" next to a calculated quantity indicated the experimental value should fall outside the maximum error limits for that quantity. Exact mass error limits decrease for higher mass resolution and greater numbers of determinations. Ranges of %M+1 and %M+2 were provided based on maximum possible errors from several sources that are discussed in Reference 10. Relative abundance error limits for the correct composition are independent of the resolution when interferences are absent. Table 1 was provided by the model after the user selected C14H14N2 as the hypothetical composition, which was proven correct. A hypothetical composition is necessary to calculate the m/z ratios monitored for the partial profiles.
The average exact mass of 211.1187 u and average relative abundance of 16.3% for the M+1 partial profile determined from triplicate injections with 10,000 resolution were consistent with C14H14N2 in Table 1. Equally important, the %M+1 value of 16.3% exceeded the %M+1 range for the other eight possible compositions in the list for 10,000 resolution. Thus, the composition of the m/z 210 ion was C14H14N2 and could be no other composition, based on the 8 elements considered by the model.
The exact mass and relative abundance of the M+2 partial
profile were also determined from a single injection to provide additional
evidence for this composition. The %M+2 value of 1.3% was within
%M+2 range listed in Table 1 for C14H14N2 at 5000 resolution and outside of the ranges for 18 of the 19 other possible compositions, including the 4 in the list for 20,000 resolution. The exact mass was 212.1208 u. The mass error of -5.2 ppm was less than the + to -12 ppm error limit for a single determination at 5000 resolution.
Determining compositions of fragment ions and combinations of atoms for neutral losses
Exact masses were determined from single injections at 10,000 resolution for the 10 fragment ions labeled in Figure 2(b). The maxima in the ion chromatograms of the m/z ratios used to construct the profiles of the fragment ions coincided with the maxima observed for the presumed molecular ion. Ions that resulted from column bleed as indicated by a steady increase in ion abundance as the temperature in the GC oven increased were not investigated.
In the upper portion of Table 2 are listed the experimental exact masses of the fragment ions, the calculated exact masses for the compositions closest in mass, and the observed mass error in mmu and ppm.
Table 2. Exact masses determined for 13 ions produced from the isomers and the corresponding neutral losses
aRings and double bonds.
bNumber of possible compositions based on the nominal mass + to -0.5 u.
cNumber of possible compositions based on the maximum error limits for the exprimental mass.
dExact mass and relative abundance criteria eliminated all other possible compositions.
eNumber of compositions in parentheses that are possible based on the presence of only C, H, and N atoms in the molecular ion.
fMaximum error expected based on the sum of the maximum errors for determination of the molecular and fragment ion exact masses.
The lower part of Table 2 provides the same information for the neutral losses, except that an error range in mmu is substituted for the observed error in ppm. The exact masses of neutral losses were determined by subtracting the experimental exact masses of the fragment ions from the experimental exact mass of the m/z 210 ion. The error ranges are the sums of the maximum mass error for triplicate exact mass determinations at 20,000 resolution for the m/z 210 ion (2.5 ppm or 0.5 mmu) and for a single exact mass determination of the fragment ion at 10,000 resolution (6 ppm or 0.7 to 1.2 mmu). The neutral losses listed are combinations of atoms that could result from single or multiple losses.
Also listed in Table 2 are the total numbers of possible compositions calculated by the PGM for each nominal mass based on C, H, N, O, P, S, Si and F atoms. For this demonstration, the fragment ions were assumed to have 0 or more rings-and-double-bonds. The number of possible combinations increased rapidly with mass and for the larger fragment ions, computer memory was exhausted before reaching this number. If the composition of the molecular ion were not known and only nominal masses were available for the ion and neutral fragments, all of these compositions would be arithmetically possible and mass spectral interpretation would be highly uncertain.
No more than 18 possible compositions were calculated for each fragment using the maximum error limits associated with each exact mass determination or pair of determinations for the neutral losses. For the m/z 129 and m/z 156 fragment ions (F), the %F+1 values that were provided by partial profiles acquired at 10,000 resolution eliminated all other compositions. For all of the fragments, only 1 composition was possible based on the atoms in the presumed molecular ion (14 C, 14 H, and 2 N), the exact masses observed, and the maximum error limits associated with MPPSIRD.
The additional combinations of atoms in Table 2 for eight of the neutral losses were consistent with the composition of the m/z 210 ion. For MS/MS experiments that provided only nominal masses for the fragment ions, uncertainty about which fragments did or did not contain N atoms would occur. For example, the nominal mass loss of 54 Da could correspond to C4H6, C3H4N, or C2H2N2 leaving 2, 1, or 0 N atoms in the fragment ion with m/z 156. Identification of the neutral loss combinations and fragment ion compositions reduced the number of candidate structures for the fragment ions.
The correlation between the fragment ions and the neutral losses in Table 2 provided strong evidence for the supposition that the fragments listed were formed by logical neutral losses from the m/z 210 ion and provided a basis for proposing possible structures of that ion. No inexplicable neutral losses suggesting the presence of a larger, unobserved ion were found. Methane, isobutane and acetonitrile chemical ionization (CI) mass spectra obtained in other laboratories all suggested that the m/z 210 ion was the molecular ion.
However, the exact mass table and the CI mass spectra cannot rule out the existence of a higher mass, odd electron, molecular ion that provides the m/z 210 odd electron ion after one or more neutral losses equivalent to one or more stable molecules. The extract was examined by GC/atomic emission detection (AED) and no additional elements were observed, which could be a part of such a neutral loss fragment. Even if such a higher mass molecular ion existed, isomeric structures for the m/z 210 ion would provide an important portion of the molecule for which a search of chemical inventories and waste streams would be justified.
Eliminating functional groups from consideration
The nominal mass of the molecular ion indicates whether it contains an even or odd number of N atoms, but provides no other information about the functional groups in a molecule. The elemental composition of the m/z 210 ion determined from high mass resolution data alone excluded alcohols, carboxylic acids, aldehydes, ketones, ethers, esters, peroxides, nitro-, nitroso-, or other oxygen, phosphorus, sulfur, halogen or metal-containing compounds. Neither FTIR nor NMR data were required to eliminate these functional groups. When the formula, C14H14N2, was searched rather than the molecular weight of 210 Da, the number of mass spectra selected was reduced from 1285 to 33 for the Wiley library and from 342 to 13 for the NIST library.
The compositions for [M]+ listed in Table 1 were possible based on the sums of masses of atoms and the valences of those atoms from which the number of rings-and-double-bonds were calculated. Obviously, many of the possible compositions in Table 1 are unlikely. However, no chemical arguments were required to eliminate these alternative compositions. Simple addition of masses and isotopic relative abundances provided a unique composition. Partially automated data collection and interpretation provided [C14H14N2]+ and the fragment compositions arithmetically based on physical properties of atoms. It is only at this point that chemical knowledge must be invoked to construct plausible structures for the isomers using limited mass spectral data bases and literature searches.
Hypothesizing structures of the isomers
A combination of manual interpretation, interactive library searching, organic chemical reasoning, and searching of the chemical literature was used to propose isomeric structures consistent with the mass spectral data. As the compositions of the molecular ions, fragment ions, and neutral losses were determined, the number of possible structures diminished.
A search of the formula indices of Chemical Abstracts from 1920 to the present found citations for about two hundred compounds with the composition of C14H14N2. Quilliam and Wright (13) used a soft ionization technique to determine by HRMS the composition of an unidentified toxin and also produced fragment ions by MS/MS to narrow their literature search. In this study, exact masses were determined for the fragment ions produced with electron impact ionization, and Table 2 was used to exclude a larger fraction of C14H14N2 isomers from consideration than could be done based on nominal masses of fragment ions and neutral losses. With fewer plausible isomers, both mass spectral data base and literature search times could be reduced.
Although no library matches were found for the low resolution spectra of the unidentified compounds, the relative abundance ratio of the m/z 156 to m/z 129 fragments was similar to that provided by 1,2-benzenediacetonitrile. A benzene ring was suggested by the series of ions at lower masses: m/z 39, m/z 51, m/z 77 and m/z 89. Initially, the loss of 54 Da from the presumed molecular ion ([M]+, m/z 210) was thought possibly to correspond to C4H6 in a retro-Diels-Alder fragmentation, but exact mass differences established that this neutral loss was C3H4N as listed in Table 2. Structures with compositions containing oxygen were also considered initially, since losses of 28 Da and 29 Da suggested possible losses of CO and HCO. The exact mass of 29.0273 u for the second neutral loss eliminated the possibility of oxygen by establishing the loss to be (2H + HCN). These two erroneous hypotheses illustrate that Table 2 should be complied for unidentified compounds before mass spectral interpretation is attempted.
Complete searches of the two libraries for the elemental composition C14H14N2 located mass spectra for 33 compounds. The nitrogen-containing functional groups included azo, amino, substituted amino, nitrile and heterocyclic compounds. The molecular ions in all but two of these 33 mass spectra were either the base peak or present in more than 30% relative abundance. In contrast, the observed relative abundances of the m/z 210 ion for the unknown isomers of only 3-12% were unusually low. The two compounds found in the library with small or unobserved molecular ions in their spectra were compounds containing a cyclopropane ring and had fragment ions inconsistent with the mass spectra of the isomers in the well water. Only library spectra of dinitriles contained fragment ions that were also observed for the unknowns, specifically m/z 156 and m/z 140. These ions had greatly different relative abundances, and many fragment ions in the library spectra were not seen for the unknowns. However, these fragment ions suggested the presence of two cyano groups. Organic chemical considerations of long-term stability in water and solubility were also consistent with this functional group hypothesis.
The composition C10H9 for the base peak at m/z 129 was consistent with the aromatic substructure shown in Figure 3 that originates from non-nitrogen containing compounds such as indanes, indenes and tetralin.(14) Hence, the nitrogen atoms were probably external to the rings.
Figure 3. In the boxes, structures for the isomers with alpha-cyanoethyl and beta-cyanoethyl groups and a plausible fragmentation scheme for an isomer with an alpha-cyanoethyl group for the 10 fragment ions studied. Isomers result from the asymmetric C atoms; the ring attachment positions are the same for each isomer.
At one time, the apparent loss of 19 Da to yield m/z 191 (net loss of CH7, by exact mass difference) caused concern that the m/z 210 ion was not [M]+, but additional accurate mass measurements on the low-abundance ions at m/z 208 and m/z 206 indicated neutral losses of 2H and 4H, respectively. Loss of four hydrogens is consistent with the formation of an aromatic ring from the saturated ring in tetralin and is energetically favorable under mass spectrometric conditions.
A search of the libraries for compounds with m/z 156 ions that lose at least 3 hydrogen atoms found that 1-cyano-tetralin (5,6,7,8-tetrahydro-1-naphthalenecarbonitrile ) lost 1 to 4 hydrogen atoms from the molecular ion (m/z 157) to yield m/z 156, m/z 155, m/z 154 and m/z 153 ions. The library mass spectrum of tetralin (1,2,3,4-tetrahydro-naphthalene) with a molecular ion of m/z 132 showed similar losses of multiple hydrogens to yield the naphthalene radical cation at m/z 128, followed by two additional hydrogen losses. Similar, unusual, multiple hydrogen losses were noted to occur in the unknowns' mass spectra from the m/z 156 fragment ion to yield, perhaps, the cyanonaphthalene radical cation. These observations and the hydrogen losses from the molecular ion suggested that the isomers had a tetralin core with groups attached to the rings at unknown positions. The tetralin core accounted for five of the nine rings-and-double bonds, and two cyano groups were required to account for the other four.
The neutral loss of 54.0344 to yield the abundant ion at m/z 156 indicated a loss of C3H4N and 2.5 rings-and-double bonds, corresponding to a radical cleavage to yield an even electron ion at m/z 156, either as a composite loss of (C2H3 + HCN) or of the intact units, CH2CH2CN or CH3CHCN. Logically, this ion then lost HCN to yield the base peak at m/z 129. Now all significant substructural features were recognized and only required correct assembly as illustrated by the approach of Lee et al.(15)
The Chemical Abstracts were then searched for industrial processes that could produce structures of the correct elemental composition with cyano and alpha-cyanoethyl or beta-cyanoethyl groups attached to a tetralin core. Abstracts showing the structures with -CH2CH2CN and -CH(CH3)CN groups in Figure 3 were located (l6-18) and later, the full papers were obtained.(19-21) These compounds were produced by the copolymerization of styrene and acrylonitrile as 1:2 adducts, and their published mass spectra(20) were very similar to those we acquired for the isomers in the well water. A facile benzylic cleavage of the alkylcyano group in the isomers accounted for the low ion abundances of the molecular ions and the neutral loss of 54 Da to yield the fragment ion at m/z 156. The fragmentation scheme in Figure 3 accounts for the fragment ions and neutral losses investigated and listed in Table 2.
Confirming identification with a standard
The source of these compounds was identified as an industrial styrene-acrylonitrile polymerization process performed in the early 1970's. A corporation that was interested in our study graciously provided an authentic standard obtained by distilling the adducts from a styrene:acrylonitrile polymer, which was prepared by another industrial process. Ion chromatograms for the base peak in the mass spectra (m/z 129) for the well-water extract after concentration by a factor of 10-100 by evaporation and for the standard were compared. A fifth isomer present at a lower concentration was found in the extract and indicated that isomers with both -CH2CH2CN and -CH(CH3)CN groups were in the extract. Three of the isomers in the standard had the same retention times as three isomers in the extract. In addition, the low resolution mass spectra were virtually identical. These data provided confirmation that at least three of the six possible isomeric compounds were present in both the extract and standard. Apparently, this different polymerization process did not form all of the isomers detected in the well water.
After this work, a more concentrated extract was prepared and studied in other laboratories. The presence of one or more cyano groups was confirmed using FTIR and the presence of a cyano group alpha to the benzylic carbon was indicated from an NMR spectrum of poor quality due to the presence of multiple components. This evidence was consistent with, but not necessary for our identifications. Although the isomers were well separated from minority components in the extract, longer GC columns with different phases might better separate the isomers and reveal the sixth isomer, if present. Because the isomers contain asymmetric carbons, enantiomeric separations into twelve components might be possible using a chiral column. Most importantly, now that the identity of the contaminants is known, others could isolate larger amounts of these compounds and perform toxicity studies.
Use of MPPSIRD and the PGM was expanded beyond determination of elemental compositions of molecular ions to those of neutral losses and fragment ions with abundances that precluded use of full scan techniques at high mass resolution. Unique fragment ion compositions greatly limited the number of possible isomers, simplified mass spectral interpretation, reduced search times of mass spectral data bases and the chemical literature, and provided tables of evidence for the identity of the compounds.
The ESD identified five trace-level compounds as five of six isomers of 1:2 styrene:acrylonitrile adducts produced in an industrial synthesis using MPPSIRD, the PGM, mass spectral data bases and the chemical literature. The persistence of the compounds would permit their use as marker compounds for exposure at the site, and more generally, as indicators of pollution by styrene-acrylonitrile polymerization processes.
Use of MPPSIRD and the PGM led to identification of municipal well contaminants in an extract at concentrations too low for FTIR or NMR to be useful and where the exact mass of the molecular ion alone could not provide a unique elemental composition. The data was acquired with a cycle time that permitted delineation of chromatographic peaks. In this case and others the advantages of speed, sensitivity and selectivity outweigh the two disadvantages of these tools: use of a relatively expensive double-focusing mass spectrometer and the need for multiple injections. These are the best tools available to help identify low levels of compounds in mixtures for which no information is available. The automated procedures for MPPSIRD and the PGM are available from the ESD, without charge, to others interested in applying them to their analytical problems.
Acknowledgements. The authors wish to express their appreciation to a number of individuals and groups who provided relevant data before we identified these compounds. Dr. Thomas D. Behymer and Dr. Jody A. Shoemaker of the EPA, ORD, NERL Human Exposure Research Division, Chemical Exposure Research Branch, Cincinnati, OH, and Mr. Robert Brittain of Varian Chromatography Systems, Palo Alto, CA, for isobutane and methane chemical ionization (CI) and methane and acetonitrile CI, respectively. Dr. Shoemaker initially provided us with a well water extract. Dr. Fu-Grand Lin and Mr. John Jencks, NJ DEP, and Mr. Julian Trexler, New Jersey Department of Health Labs for much of the early analytical work and for the preparation of the water extracts. Mr. John Gorin, EPA Region II, Remedial Program Manager, Southern New Jersey Remediation Section, for providing this opportunity through Mr. Kenneth W. Brown of the Technology Support Center, ESD, Las Vegas. Dr. William C. Brumley of EPA, ORD, NERL-ESD, Las Vegas for many helpful discussions. We also acknowledge the later FTIR work of Tim Collette of EPA, ORD, NERL-ERD, Athens and the NMR work of Dr. Paul Kiefer from Varian. Finally, we are grateful to the two individuals at the corporation with an interest in this study who supplied us with the standard from the styrene:acrylonitrile polymerization.
Notice: 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.
NJ Dept. of Health and Senior Services, News Release, 18 April, 1997.
EPA Region 2 Report, Reich Farm, NJ, EPA ID# NJD980529713.
NIST Standard Reference Database lA, NIST Mass Spectral Search Program and NIST/EPA/NIH. Mass Spectral Database, Ver. 1.0, For Use with Microsoft Windows, U.S. Dept. of Commerce, Technology Administration, National Institute of Standards and Technology, Standards Reference Data Program, The NIST Mass Spectrometry Data Center, Gaithersburg, MD 20899, January, 1995.
Benchtop PBM, Version 3.1 and the Wiley Registry of Mass Spectral Data, 6th ed. with Structures, Palisade Corp., 31 Decker Rd., Newfield, NY (1995).
Method 8290, Test Methods for Evaluating Solid Wastes, 1B#2, SW-846, US EPA, Office of Solid Waste and Emergency Response, Washington, DC, 1986.
Method 1613, Nat'l. Center. For Environmental Publications and Information, Cincinnati, OH (1994).
V. L. Maggio, L. R. Alexander, V. E. Green, J. B. Gill, D. G. Patterson, Jr., B. N. Green and A. J. Seed, Toxicological Environ. Chem. 28, 143-154 (1990).
A. H. Grange, Proc. 40th ASMS Conf Mass Spectrom. Allied Topics, Washington, D. C., 1143-1144 (1992).
M. A. Quilliam and J. L. C. Wright, Anal. Chem. 61, 1053A-1060A (1989).
F. W. McLafferty and R. Venkataraghavan, Mass Spectral Correlations, 2nd ed., Advances in Chemistry Series 40, American Chemical Society, Washington, D.C. (1982).
M. S. Lee, S. E. Klohr, E. H. Kerns, K. J. Volk, J. E. Leet, D. R. Schroeder and I. E. Rosenberg, J. Mass Spectrom.31, 1253-60 (1996).
Chem. Abstr., 89: 130461z (1978).
Chem. Abstr., 85: 160596g (1974).
Chem. Abstr., 84: 60059v (1973).
Y. Kudo, K. Okawa, S. Ito and Y. Ishida, Japan. Kokai 78 52, 560, 13 May 1978.
K. Kirchner and H. Schlapkohl, Makromol. Chem. 177, 2031-2042 (1976).
K. Kirchner and H. Schlapkohl, Naturwissenschaften 62, 529 (1975).