![[logo] US EPA](https://webarchive.library.unt.edu/eot2008/20090508233339im_/http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Ion Composition Elucidation (ICE)
Evolution of ICE
The Past
We first monitored multiple m/z ratios across individual profiles to investigate why expected ion ratios between profiles from the molecular ion group for polychlorinated dioxins were in error by more than the 15% criterion specified by dioxin methods. Were profiles for other ions overlapping the profile of one of the analyte ions? Not in our case. Lengthening the transition time between monitoring of the heaviest ion for one congener and the lightest ion for the next higher congener from the default time of 25 msec to 30 msec solved the problem. This was an obvious experiment to perform and at least one other group performed similar experiments and published their results.[1]
Rather than monitor multiple m/z ratios across a profile, two other groups scanned narrow mass ranges across the analyte profiles to check for interferences.[2, 3] All three groups were searching for interferences or documenting that none were present using polychlorinated furans and dioxin analytes. They also obtained exact masses from scanned profiles.
In our first paper published in Rapid Communications in Mass Spectrometry in 1992,[4] we also utilized the new Selected Ion Recording (SIR) technique to measure mass resolution as a compound eluted from a gas chromatograph into the ion source of the mass spectrometer.
The paper in Analytical Chemistry in 1994 [5] used exact masses determined from profiles plotted from SIR data to identify several compounds, thereby elevating the new technique from a novel curiosity to a valuble new analytical tool. Determination of the exact mass of Acid Orange 8 illustrated that ICE can be used with liquid-phase sample introduction, as well as with gas-phase introduction from a gas chromatograph or insertion probe.
The second Analytical Chemistry article in 1996 [6] made use of the +1 and +2 profiles to extend four-fold the mass limit for which unique ion compositions could be determined. Partial profiles were now used to monitor three analyte profiles simultaneously, in addition to the lock mass ion profile. The exact masses were calculated as the weighted averages of the top several chromatographic peak areas used to plot the profiles and relative abundances were simply the ratio of the sums of these areas x 100%. Unpublished calculations using curve fitting to a Gaussian distribution provided no more than 0.0003 amu differences with the masses calculated from weighted averages. The weighted average calculation is still used, because it would be much easier to explain in legal proceedings.
One of the first compounds investigated provided a molecular ion with the composition C28H41NS. At 20,000 resolution the M+2 profile displays a valley between two maxima as shown in the figure. When the blue portion of the profile was monitored, the data was confusing without this figure for guidance. Was the maximum missed and was there a low mass interference? Not at all. It was this strange partial profile that made obvious the need for a profile generation model to provide such figures. Only then could interferences be distinguished from natural profile broadening that occurs at high mass resolution, especially when S atoms are present.
A Profile Generation Model (PGM) was written in QuickBASIC to construct +1 and +2 profiles from the contributions made by all ions containing an atom of a +1 isotope (+1 profile) or ions containing an atom of a +2 isotope, two atoms of the same +1 isotope, or one atom each of two different +1 isotopes (+2 profile). This model was described in detail in the Journal of the American Society of Mass Spectrometry in 1997[7]. Simulations were performed to show that if 15 determinations were made of the three exact masses and two relative abundances entered into the PGM, the low error limits for the averaged measurements would provide a unique composition for ions with m/z ratios up to 600 that contain C, H, N, O, P, or S atoms.
The Present
Early publications [6-9] emphasized the utility of the shape of the +2 profile for distinguishing between compositions. However, comparison of the shape of a profile delineated by only 10 to 15 points to the shape of calculated profiles is subjective. More compelling evidence for an ion's composition is compiled by comparing measured values with values calculated as sums of atomic masses and isotopic abundances, which are fundamental physical properties. The apparent resolution originally included in the PGM has also been de-emphasized, since explaining this concept in a legal proceeding would be more difficult than explaining simple addition of physical properties. Hence, the PGM has been fully automated to calculate exact masses and relative abundances and to compare them to the measured values determined from profiles plotted from SIR data. Little ability to distinguish among compositions was sacrificed, since substantial profile broadening of the +2 profile only occurs when an S or Si atom is present. The exact mass of the +2 profile also indicates one of these atoms is present. Hence, the shape and apparent resolution of the +2 profile are usually redundant for the purpose of rejecting compositions.
Although 15 determinations were made in a reasonable time using probe introduction to estimate error limits at both 10,000 and 20,000 resolution for several standards [6, 7], it is impractical to make 15 determinations for analytes injected onto a GC column prior to a 30 min temperature program. Generally, current practice is to make only one determination of the exact masses and relative abundances of an analyte's partial profiles. If multiple compositions remain viable, at least one prominent fragment ion is investigated. More is learned about the analyte's structure by investigating fragment ions than by repeating data acquisitions for the apparent molecular ion. Usually, only one composition is possible for a fragment ion and the exact mass difference between the molecular and fragment ions corresponds to a single composite neutral loss. This is illustrated in an article in Rapid Communications in Mass Spectrometry [10] in which determining the compositions of 10 fragment ions and the corresponding composite neutral losses greatly reduced the number of possible compound identities, for a series of isomers found in a well water extract. A quick search of the chemical literature soon revealed the source of the compounds, a polymerization process that yielded 2:1 acrylonitrile:styrene byproducts. Reference 11 also provides examples of fragmentation schemes buttressed by knowledge of fragment ion and composite neutral loss compositions.
The error limits in the table from Reference 7 were established using probe introduction of standards into a VG -70SE double focusing mass spectrometer and a single calibration mass.
Our new instrument, a Finnigan MAT 900S mass spectrometer, provides 31 m/z ratios (rather than the 25 m/z ratios provided by the VG instrument) for SIR data acquisition and provides a second calibration mass. The analyte masses are bracketed by the two calibration masses and more accurate exact masses are obtained. The error limits used in the PGM are likely a factor of 2 higher than those actually observed for pure standards. The higher error limits were kept to allow for distortions in measured values caused by low-level interferences. For any legal proceedings, error limits should be established for a standard under the experimental conditions for which analyte data is obtained.
Partial profiles are monitored routinely because they provide three exact masses and two relative abundances for a single data acquisition. Articles in Rapid Communications in Mass Spectrometry [12] and the International Journal of Environmental Forensics [13] utilized pairs of full mass peak profiles to determine exact masses and more accurate relative abundances. Using full profiles eliminated three sources of error associated with partial profiles, which are examined in Reference 7. The remaining errors due to isotopic abundance variations and instrumental precision would be simpler to explain to a judge or jury. In the earlier paper, a novel concept was demonstrated: determination of ion compositions without using mass calibrants. Determination of relative abundances and exact mass differences between fragment ions and observation of the shape of +1 profiles provided the necessary data. This capability may prove useful for identifying compounds using electrospray ionization (ESI). Ideal calibrants such as perfluorokerosene are not available for ESI and when calibrants are used, their ions are a major fraction of the ions produced. Their production increases the detection limit for analytes, which are usually much higher than the detection limits observed with GC or probe introduction. The later paper illustrates the forensic applications of ICE.
ICE was adapted to accurately measure the mass of a hemoglobin adduct with electrospray ionization[14]. The sample was infused and multiply charged alpha- and beta-chains served as the calibrants to determine the mass of a glycate adduct. Masses accurate to within 0.2 amu were obtained. Conventional measurements provide errors of 1.5 to 2.3 amu.
The Future
The drafts of publications and posters on this website have established that ICE is a very useful analytical technique for identifying compounds. Our highest priority is to transfer this technology to other laboratories. This website is a part of that effort. It provides easy access to all previous applications and provides a repository for future advances. A narrated, animated, PowerPoint video presentation, "ICE is Nice," available from the author, explains the science behind ICE and illustrates three analytical applications. A second PowerPoint video, "ICE is Easy", serves as a "how-to" manual for a Finnigan MAT 900 or MAT 95 mass spectrometer. All user inputs and video screens are illustrated as the compositions of three ions are determined.
The MPPSIRD code is written in Lotus 123 version 9.0, which is commercially available as part of Lotus Smart Suite. The PGM is written in QuickBASIC version 4.5, which can still be found for sale at various web sites (search: QuickBASIC 4.5). After installing ICE on other instruments, other researchers applying this new technology to their problems should author numerous publications. Hopefully, these successes will further demonstrate the enhanced analytical capability ICE brings to double focusing mass spectrometers and its ease of use. In time, a market for ICE sufficient to encourage manufacturers to include ICE on data systems of new double focusing mass spectrometers should develop.
Determination of the exact mass of Acid Orange 8 [5] illustrated that ICE can be used with liquid phase introduction. ICE will be used to determine compositions of ions produced by liquid sample introduction followed by electrospray or atmospheric pressure chemical ionization.
"Mystery" analytes encountered by EPA regional labs and others will continue to be investigated. We encourage those with compound identification problems to contact us for possible collaborations.
-
A.R. Allan and J. Roboz; Rapid Commun. Mass Spectrom. 1998, 2, 246-249.
-
Tong, H. Y.; Giblin, D. E.; Lapp, R. L.; Monson, S. J.; Gross, M. L. Anal. Chem., 63, 1772-1780.
-
D. J. Harvan, J. R. Hass, J. L. Schroeder and B. J. Corbett, Anal. Chem. 1755 (1981).
-
Grange, A.H.; Brumley, W.C. Rapid Comm. in Mass Spectrom. 92, 6, 68-70.
-
Grange, A.H.; Donnelly, J.R.; Brumley, W.C.; Sovocool, G.W. Anal. Chem., 1996, 68, 553-560.
-
Grange, A.H.; Brumley, W.C. J. Amer. Soc. for Mass Spectrom., 1997, 8, 170-182.
-
Grange, A.H.; Brumley, W.C. Trends in Analytical Chemistry, 1996, 15(1), 12-17.
-
Grange, A.H.; Brumley, W.C. Environmental Testing & Analysis, 1996, March/April, 22-26.
-
Grange, A.H. and Sovocool, G.W. J. AOAC Internat., 1999, 82, 1443-1457.
-
Grange, A.H.; Sovocool, G.W. Rapid Commun. Mass Spectrom. 1999, 13, 673-686.