Jump to main content.


Ion Composition Elucidation (ICE)

Determination of Ion and Neutral Loss Compositions and Deconvolution of Product Ion Mass Spectra Using an Orthogonal Acceleration, Time-of-Flight Mass Spectrometer and an Ion Correlation Program

Andrew H. Grange,1 Michael C. Zumwalt,2 and G. Wayne Sovocool1

1Environmental Sciences Division, NERL, U.S. EPA, P.O. Box 93478, Las Vegas, NV 89193-3478
2Agilent Technologies, PO Box 4026, Englewood, CO 80155-4026

[Note: minor content and formatting differences may exist between this web version and the published version.]

ABSTRACT

Exact masses of monoisotopic ions and the relative isotopic abundances (RIAs) of ions greater in mass by 1 and 2 Da than the monoisotopic ion are independent and complementary physical properties useful for distinguishing among ion compositions possible for a given nominal mass.  Using these properties to determine compositions of product ions and neutral losses increases the mass of precursor ions for which unique compositions can be determined.  Precursor ion, product ion, and neutral loss compositions aid mass spectral interpretation and guide modest chemical literature searches for candidate standards to be obtained for confirmation of tentative compound identifications.  This approach is essential for compound characterization or identification due to the absence of commercial libraries of ESI and APCI product ion spectra.  For a series of 34 exact mass measurements, an orthogonal acceleration time-of-flight mass spectrometer provided 34 and 29 values accurate to within 2 and 1 mDa, respectively, for ions from eight simulated unknowns with [M+H]+ ion masses between 166 and 319 Da.  Of 36 RIA measurements for +1 Da or +2 Da ions, 35 were accurate to within 20% of their predicted values (or to within 0.4 RIA % when the RIA value was less than 1%) in the absence of obvious interferences, when the monoisotopic ion peak areas were at least 1.7 x 105 counts, and the ion masses exceeded 141 Da.  An Ion Correlation Program (ICP) provided the unique and correct compositions for all but three of the 34 ions studied.  Manual inspection of the data eliminated the incorrect compositions.  To test the utility of the ICP for deconvoluting composite product ion spectra, all 34 ions were tested for correlation.  Six of eight precursor ions were identified as such, while two were compositional subsets of others and were not properly identified.  The six precursor ion compositions were still found by the ICP even though ions with masses less than 158 Da were not considered since they could no longer be correlated with a single precursor ion.  Finally, two unidentified analytes were characterized, based on data published by others and using the ICP together with mass spectral interpretation.

INTRODUCTION

Most mass spectrometric methods applied to environmental sample extracts quantitate a short list of target analytes using relatively inexpensive instruments.  Relatively few of the 2800 high-volume production chemicals (annual production of at least 106 lbs)(1) and 87,000 commercially produced chemicals,(2) along with their synthetic precursors, byproducts, transformation products, and metabolites, are target compounds.  For GC/MS analyses a small minority of these compounds can be tentatively identified based on the best mass spectral library match (if any matches are found by the data system), but additional attempts to identify unknowns are rare.  For LC/MS methods, ESI and APCI mass spectral libraries are instrument-specific and are not commercially available.  Assessments of risks posed by Superfund sites and water discharges to ecosystems and human health would be more thorough if non-target compounds were identified and their toxicological properties were considered.

Atomic masses and isotopic abundances are two independent physical properties useful for determining elemental compositions of ions, henceforth referred to as "ion compositions".  To exploit these properties to identify compounds in complex extracts, data providing accurate exact masses of ions and accurate relative isotopic abundances (RIAs) must be acquired on-the-fly as chromatographically separated analytes elute into a mass spectrometer.  For the past decade this laboratory has employed an Ion Composition Elucidation (ICE) approach to determine compositions of ions observed in mass spectra, by acquiring selected-ion-recording data to delineate mass peak profiles of monoisotopic ions and their accompanying ions greater in mass by 1 Da and 2 Da that arise from the presence of atoms of heavier isotopes such as 13C, 2H, 15N, 17O, 18O, 33S, 34S, 37Cl, and 81Br.  In this paper, RIAs are normalized to the monoisotopic ion, which has an RIA of 100%.  Using ICE, compounds from Superfund sites,(3) municipal wells,(4) lakes,(5) streams,(6) and sewage treatment effluents(7) have been characterized or identified.  Entering measured exact masses and RIAs into an in-house Profile Generation Model(8) provided unique and correct ion compositions for the apparent molecular ions and prominent product ions observed in mass spectra of unidentified compounds.  This information often led to compound identification based on mass spectral interpretation and modest library searches.  However, the complexity of ICE, the time and effort required to use it, the expense, size and maintenance of a double focusing mass spectrometer, and the customized software that must be developed for each data system, discouraged adoption of ICE by other laboratories.

Triple-quadrupole mass spectrometers capable of accurate mass measurements (AM3QMS, Thermo Finnigan), and orthogonal acceleration time-of-flight mass spectrometers (oa-TOFMS), that provide combinations of scan speed, mass accuracy, linear dynamic range, and resolving power adequate to measure exact masses and RIAs with accuracies sufficient to provide ion compositions(9), are now available.  An AM3QMS was used to determine unique and correct ion compositions for precursor ions, product ions, and neutral losses formed through APCI from nine simulated unknowns introduced by HPLC.(10)  Data acquisition using the AM3QMS data system software was much simpler than when using a double focusing mass spectrometer, because no custom software was necessary.  After entering exact masses accurate to within 5-10 mDa and RIAs accurate to within 5-10% of their values measured for multiple ions from individual analytes, an in-house Ion Correlation Program (ICP) provided the unique compositions.(10)

A quadrupole orthogonal acceleration time-of-flight mass spectrometer (QqTOFMS) has also been used to measure exact masses to within 7 mDa and RIAs to within 10-20%.  Ib摙ez et al.(11) compiled a list of possible compositions for each precursor ion based on these error limits.  When multiple compositions remained, product ions were also considered.  Even so, multiple precursor ion compositions usually remained before these authors examined the chemical literature for candidate compounds.  The authors identified two of five compounds that were found in environmental samples from the ion compositions they determined.

Both double focusing mass spectrometers and the AM3QMS retain the disadvantage of high capital cost.  Compound identification strategies thought to require multiple mass spectrometers(12,13) are also expensive and time consuming.  Here, the utility of a less expensive, single-MS-stage oa-TOFMS for determining ion compositions of eight simulated unknowns is demonstrated.  To cope with potential coelution, the ability of the ICP to deconvolute composite mass spectra is demonstrated for lists of ions obtained for multiple analytes.  Finally the ICP is applied to published data(11) acquired for two real-world unknowns.

EXPERIMENTAL

Standards

Nine standards simulated unidentified compounds:  4-aminobiphenyl (97%), phenazine (98%), 2-(methylthio)benzothiazole (97%), N,N-diethyl-3-methylbenzamide (97%), N-butylbenzenesulfonamide (99%), n-(3-chloro)-2,2'-iminodiethanol (98%), tris(2-chloroethyl)phosphate (97%), and chlorpromazine hydrochloride (98%) were from Aldrich (Milwaukee, WI), and pseudoephedrine was extracted from a cold remedy tablet.  The two internal mass calibrants were provided by a reference compound solution from Agilent Technologies (Santa Clara, CA), which contained 0.6 ng/無 of purine (121.05087 Da) and 1.2 ng/無 of HP-0921 (922.00980 Da), a proprietary polyfluorinated compound, in acetonitrile:water (95:5).

HPLC

An Agilent 1100 Series HPLC was used with a Zorbax SB-C18, 2.1 x 50 mm, 3.5-痠 particle size HPLC column (both Agilent, Santa Clara, CA).  An Agilent well-plate autosampler (WPS) made 10-無 injections of a 1:1 water:methanol solution containing 1% acetic acid (EMD, Glacial, Reagent, ACS, Darmstadt, Germany) and 1 ng/無 of each analyte.  The two solvents, A and B, were water (HPLC Grade, ACS, Fisher Chemicals, Fair Lawn, NJ ) and methanol (Burdick & Jackson, HPLC Grade, ACS, Muskegon, MI) both with 0.2% acetic acid.  The linear solvent gradient was:  2% B to 98% B in 17 min with a 2-min post gradient hold.  The flow was 200 無/min.  No effort was made to optimize chromatographic separation of analytes because obtaining exact masses and RIAs for a predetermined list of ions was the sole goal of data acquisition.  The mass calibrant solution was infused at 100 無/min through a separate nebulizer needle.

Mass Spectrometry

The HPLC system was interfaced to an Agilent G1969A LC/MSD TOF oa-TOFMS with an Agilent G3251A Dual ESI source.  The drying gas (N2) flow was 10 L/min and was heated to 325oC.  The nebulizer pressure was 30 psi and the capillary voltage was 4 kV.  The range recorded was m/z 50 - 950 with 10,000 transients/spectrum and 0.88 spectra/s.  Fragmentor voltages of 100, 150, 200, and 250 V were applied.  For mass peaks the FWHM was 0.05 Da, half that provided by the AM3QMS.(10)  Hence, at m/z 300 the FWHM resolving power was 6000.

RESULTS AND DISCUSSION

Fuller Utilization of Information in Mass Spectra

Product ion mass spectra acquired with multiple CID voltages provide a precursor ion and usually, one or more product ions.  (A lack of product ions suggests a high degree of aromaticity and a lack of unsaturated side chains.)  The number of possible compositions increases exponentially with mass and is highly dependent on the non-integer part of the mass (non-integer mass) as illustrated in Figure 1.

Figure 1 - For further information contact grange.andrew@epa.gov

Figure 1.  (a) A plot demonstrating the exponential increase in the number of possible compositions with increasing ion mass for a mass error limit of 2 mDa.  The non-integer mass was 0.0000 and the mass increment was 20 Da between bars.  The numbers of elemental compositions were calculated considering the elements C, H, Cl, N, O, P, and S with between 0 to the integer of M/AM atoms of each element, where M was the mass of an ion and AM was the atomic mass of an element.  (b) The number of possible compositions for an ion of nominal mass 319 Da calculated for different non-integer masses at 0.05 Da mass increments assuming a mass error limit of 2 mDa and the same possible elements.  (c) A plot of the possible compositions calculated for the measured exact masses of a precursor ion and its seven product ions listed in Table 1 and (d) the seven corresponding neutral losses.  The mass error limit is 2 mDa for the ions and 2.83 mDa (2^.5 x 2 mDa) for the neutral losses, which are determined as the difference in measured exact masses for the precursor and product ions.

Of the eight ions produced from the last analyte in Table 1, the precursor ion (m/z 319 in Figure 1c) has the highest mass and the largest number of possible compositions for a given mass error limit.

Table 1 - For further information contact grange.andrew@epa.gov






Consequently, determining the unique correct compositions of any product ion and its corresponding neutral loss is more probable than determining the composition of the higher-mass precursor ion.  The product ion and neutral loss compositions sum to that of the precursor ion.  One can resort to consideration of product ions only after a unique composition is not found for the precursor ion,(11) or one can routinely use calculations to correlate possible compositions of the precursor ion, product ions, and neutral losses.(10)  For cases where data for the precursor ion alone would suffice to reveal its unique composition, this strategy provides more confidence that it is related to the product ions observed in its product ion spectra.  In addition, each sum of product ion and corresponding neutral loss compositions provides confirmatory evidence for the precursor ion composition based on additional measurements.

Using Exact Masses and Relative Isotopic Abundances (RIAs) to Reject Possible Compositions

For the measured mass of 319.1039 Da in Table 1, 47 compositions are possible based on the assumptions that at least 1/3 of the ion mass is from C atoms, that the ion contains up to 26 C, 316 H, 22 N, 19 O, 10 P, 9 S, and 9 Cl atoms, and that the mass error is 2 mDa or less.  When the exact masses of the precursor ion and of its seven product ions in Table 1 and the corresponding neutral losses were also considered by the ICP, only 17 possible compositions remained.  Consideration of exact masses alone with a 2 mDa mass error limit was inadequate for determining unique ion compositions.  After changing the error limit to only 0.319 mDa (1 ppm), seven compositions were still possible for the precursor ion’s calculated mass of 319.10302 Da.  An independent means for distinguishing among compositions is needed to determine the unique compositions of precursor ions.

RIAs meet this requirement.  The numbers of Cl or Br atoms, and estimates of the number of S atoms in an ion, are often deduced from the relative abundance of the (A+2) ion (the %2 RIA).  The %1 RIA provides a range for the number of C atoms.  Unusually small RIAs suggest the presence of monoisotopic elements other than the usually considered P, such as an I or As atom, or multiple F atoms.  However, systematic consideration of RIAs by comparing measured values with values calculated for compositions that are possible based on measured exact masses is a more effective and convenient means of utilizing the discriminating power of RIAs.  This point is illustrated in Figure 2.

Figure 2 - For further information contact grange.andrew@epa.gov

Figure 2.  Partial tables of possible compositions:  (a) for an m/z 319.1039 ion based on an exact mass error limit of 2 mDa, (b) for a nominal mass 319 ion based on RIA error limits of 20%, and (c) on an m/z 319.1039 ion based on both of these error limits.  The measured exact mass and measured RIAs from Table 1 were entered into a composition generator (PGM)(8) and C, H, Cl, N, O, P, and S were the elements considered.

Figure 2a is a partial list of the 47 compositions that are possible based on the assumptions stated in the previous paragraph.  Figure 2b is a partial list of the 402 compositions that are possible for a nominal mass of 319 Da with RIAs within 20% of 19.00% and 36.78%, the measured %1 RIA and %2 RIA values for this precursor ion.  As for exact masses, considering RIAs alone did not yield a unique precursor ion composition.  Figure 2c lists the five compositions that are common to full listings of possible compositions some of which are shown in the tables in Figures 2a and 2b.

The numbers of possible compositions in the tables based on exact masses and RIAs depend on the error limits.  For Figure 1 in Reference (10), a mass error of 7 mDa assumed for an exact mass measured by an accurate-mass triple-quadrupole mass spectrometer provided 172 possible compositions, while a smaller RIA error limit of 10% provided only 154 possible compositions for the measured RIAs.  Seven compositions were common to both tables.  Only the correct composition, C17H20N2SCl, appeared in the final table for both sets of measurements and error limits.(10)

Ion Correlation Program (ICP)

An ICP was written in QuickBASIC version 4.5 (Microsoft Corp., Bellevue, WA, USA) to (1) compile lists of possible compositions for the precursor ion, up to seven product ions, and their corresponding neutral losses, (2) determine which compositions are consistent with one another, and (3) discard those that are not.  A four-step process is executed, as described in reference (10):

      1.  All possible compositions having at least -0.5 rings and double bonds (RDB), that are consistent with the elemental limits considered, exact masses and mass error limit, and RIA and RIA tolerance set by the user, are calculated for the precursor and product ions and stored for further processing.  The precursor ion compositions are calculated first to establish upper elemental limits for the subsequent calculation of the possible product ion compositions.

      2.  All possible neutral loss compositions are calculated based on the mass differences between the precursor ion and all product ions.  The formulas characterized by an RDB value of no less than -2.0 are saved and stored for further processing.

      3.  Those precursor ion compositions are rejected which cannot be derived by the summation of the number of atoms of each element in a product ion-neutral loss pair.  This formula discrimination criterion is applied for each product ion exact mass.

      4.  Product ion compositions that do not provide a remaining possible precursor ion when summed with a corresponding neutral loss composition are rejected, as are neutral loss compositions that do not provide a remaining possible precursor ion when summed with a corresponding product ion composition.

A saturated [M+H]+ ion has RDB = -0.5.  Three successive losses of H2O would yield H6O3, a total neutral loss for which RDB = -2.0 would be calculated.  The possibility of higher valences for N, P, S, and Se atoms is also considered as described in Reference (10).

The RIA error limit entered by the user applies to all RIA values of 1% or more calculated for possible compositions.  The error limit is automatically set to 0.02 times the entered error limit (e.g., 0.4 RIA % units for a 20% error limit) for RIA values less than 1%.  To illustrate this point, for m/z 166 the %1 RIA calculated for C10H16NO+ is 11.67%, 0.1% less than the measured value in Table 1 of 11.68%, while the calculated %2 RIA of 0.82% was 0.38 RIA % less (but it is less by 46.3% on a relative scale) than the measured value of 1.20%.  This approach provides a wider error acceptance for RIAs less than 1%, for which low ion abundances and interferences are most detrimental to accurate measurement.

The narrower mass error limits provided by the oa-TOFMS relative to the AM3QMS provided fewer possible compositions based on exact masses before RIAs were considered.  Hence memory overloads,(10) necessitating elimination of an element from consideration or limitation of the number of atoms of an element, were required only for the highest mass compound (m/z 399) in the examples discussed herein.

To be considered, an ion's exact mass must be entered into the ICP.  In addition, the exact masses of the +1 ion or +2 ion, the %1 RIA, or the %2 RIA can, but need not, be entered.  Consideration of the RIAs rejects more compositions than consideration of the exact masses of the +1 Da and +2 Da ions.(10)  Examples of ICP input and output are shown in Figure 3.

Figure 3 - For further information contact grange.andrew@epa.gov

Figure 3.  (a) ICP display of the user inputs (yellow text) including exact masses for the precursor and four product ions and of the RIAs for the precursor ion and (b) ICP output display of the possible ion and neutral loss compositions.  When elements alone are entered for the upper elemental limits, the ICP calculates the number of atoms for each element as the integer of M/AM, where M is the mass of the ion and AM is each atomic mass.

Measured exact masses for a precursor ion and four of its product ions from Table 1 were entered along with the RIAs for the precursor ion.  These inputs provided a unique composition for the precursor ion, all four product ions, and their corresponding neutral losses.

Entering the RIAs for the m/z 198 ion rather than for the m/z 216 ion also provided unique compositions for all ions and neutral losses.  Correlation of related ions extends the reduction of possible compositions realized by utilizing RIAs for one ion to other ions.  This is fortuitous, since RIA measurements are much more susceptible to interferences than are exact mass measurements.  A discussion of exact mass and RIA errors, and a more detailed example of the synergy realized by entering only a limited number of RIAs into the ICP, are presented elsewhere.(10)

Product Ion Spectra Generated by In-Source CID

The full scan, background-subtracted, product ion spectra in Figure 4 were produced at the largest fragmentor voltage used (250 V) for eight of the analytes as they eluted from an HPLC into the ion source of the oa-TOFMS.

Figure 4 - For further information contact grange.andrew@epa.gov


Figure 4.  Full scan, background subtracted, ion product spectra for eight simulated unknowns acquired with a fragmentor voltage of 250 V.  The exact masses measured for this study are labeled.  To provide a higher ion abundance for the m/z 166 ion and to avoid approaching saturation for the m/z 119 ion, these two exact masses were obtained with a fragmentor voltage of 200 V.

The ninth analyte was not observed, and might not have eluted during the HPLC solvent program.  The precursor and product ion exact masses entered into the ICP are labeled.  These product ions were studied previously(10) using an AM3QMS, and were the focus of the present study in order to directly compare the utility of the two instruments for determining ion compositions.  Exact masses were obtained from these ion product spectra using data processing procedures provided by the oa-TOFMS data system.  For background subtraction, product ion spectra were averaged over a 0.5 min range preceding elution of each chromatographic peak.  Exact masses were obtained from analyte ion chromatograms integrated over a 0.2-Da mass window centered about the approximate exact masses observed in the background-subtracted ion product spectra.  Average exact masses were obtained from the points between the two 50%-of-maximum points of the extracted ion chromatographic peak.

Each ion abundance observed in the ion chromatogram was the sum of 10,000 TOF transients.  Since the detector sampling rate was 1 GHz, each spectrum consisted of 1 billion data points across the mass range.  Each data point was summed in an 8-bit register so that the maximum signal the TOF instrument could measure was 10,000 x 255 (8-bit), or 2.55 x 106 counts.  Skewing of mass peaks, resulting in larger mass errors, can occur for ion abundances exceeding 50% of this maximum.  This level was approached only for the m/z 119 product ion observed with a fragmentor voltage of 250 V.  Therefore, the exact mass for the m/z 119 ion was obtained from the product ion spectrum generated using a fragmentor voltage of 200 V.

The RIAs were obtained as ratios of areas under mass peak profiles observed in background-subtracted, full-scan, product ion spectra normalized to 100%.  Because accurate masses and RIAs were obtained for ions across the entire mass range (50-950 Da) during full scanning, no selected ion detection modes requiring menu preparation before acquiring data were necessary.  Hence, the data in Table 1 were obtained from only two data acquisitions.  Compared to a double focusing mass spectrometer(14) or an AM3QMS,(10) exact masses and RIAs were measured with much less effort in much less time.  The exact masses and RIAs for ions from eight simulated unknowns are listed in Table 1.

RIA and Exact Mass Accuracies

Interferences with RIA measurements were a greater problem at masses below 142 Da than when a resolving power of 10,000 (10% valley) was used with a double focusing mass spectrometer or when the two-MS-stage AM3QMS was used to select ions across a mass range of 10 Da by Q1 for collision in Q2 and analysis by Q3.(10)  Above m/z 141 and for monoisotopic ion peak areas greater than 1.7 x 105 counts, 35 of 38 measured RIAs (92%) were accurate to within 16% or 0.38 RIA % if the RIA was less than 1%.  The average RIA error for 32 RIAs greater than 1% was -7.8 %, suggesting a bias.  If a similar bias is observed in future work, a bias correction will be applied; but for this work, no correction was made.  These RIA errors are consistent with the 10%-20% RIA error limits observed by Ib摙ez(11) using a QqTOFMS.  The RIAs for the +1 Da and +2 Da ions for m/z 152 were high.  Figure 4b revealed that both m/z 152 and 153 ions were formed from the precursor ion at m/z 170.(10)  The m/z 153 ion and its +1 Da ions severely distorted the RIAs but not the measured exact mass of the m/z 152 ion.

Ions formed from an analyte within 2 Da of a monoisotopic product ion are mass interferences that can sometimes be reduced by changing the CID voltage.  These "inherent interferences" are discussed elsewhere.(10)  A two-MS-stage instrument allows one to observe the inherent interferences by acquiring monoisotopic product ion spectra, providing the first MS stage has sufficient resolving power to isolate individual precursor ions.  The monoisotopic precursor ion is fragmented to produce only monoisotopic product ions containing no atoms of heavier isotopes.  Examining these spectra allows the analyst to avoid entering RIAs likely to be distorted into the ICP.  But such spectra cannot be obtained with a single-MS-stage instrument, and only major interferences such as m/z 153 with m/z 152 are easily observed.  Fortunately, it was observed that product ions produced by simple bond cleavage at low CID voltage generally did not yield ions within 2 Da of the monoisotopic product ion.(10)  A single trustworthy pair of RIAs often provides unique compositions for the precursor ion, product ions, and neutral losses.

Measured exact masses of monoisotopic ions are much less susceptible to mass interferences.(10)  The monoisotopic precursor and product ion abundances are usually large relative to interferences of the same nominal mass, and their exact masses are measured accurately.  For the 34 measured exact masses for monoisotopic ions in Table 1, the mass errors were less than 1 mDa for 29 measurements (85%) and less than 2 mDa for all 34 measurements (100%).

Assumptions Made When Using the ICP

The goal of using the ICP is to provide a list of unique and correct compositions for the precursor ion, product ions, and neutral losses.  A trial and error process was employed.  Error limits that are too liberal can provide multiple compositions for one or more ions or neutral losses, while overly restrictive error limits will generally provide no compositions, if even one measured exact mass or measured RIA falls outside of the values calculated for the correct composition plus the error limits entered.  Failure to include an element found in an analyte ion would also provide no compositions.  Throughout this study unless stated otherwise, the assumptions entered into the ICP were as follows:

        (i)  The relative abundances associated with the precursor ion were entered if the
                monoisotopic ion peak area was 1.7 x 105 counts or greater.  This was true in
                all cases reported here.

        (ii)  Since RIAs were more susceptible to interferences than exact masses, the
                smallest number of RIAs needed to find unique compositions was used.

        (iii)  A mass error limit of 3 mDa and an RIA error limit of 20% were used.

        (iv)  No RIAs were entered for masses of 141 Da or less.

        (v)  The elements considered were C, H, N, O, P, S, Cl, and Br.

        (vi)  The precursor ion was not assumed to be an [M+H]+ ion.

The monoisotopic ion was chosen for the minimum ion peak area assumption (i) rather than the isotopic ions(11) for two reasons.  First, for a high monoisotopic ion abundance, the signals for the analyte monoisotopic, +1, and +2 ions are less likely to be inflated by signals from interferences.  Second, a large isotopic ion abundance might result primarily from an interference, especially for a %2 RIA as is frequently seen in Table 1, for compositions lacking S and Cl atoms.  A very low ion abundance for the +2 Da ion indicates both a lack of S, Cl, and Br atoms and a lack of interferences.  A very low +1 Da ion abundance would indicate the presence of one or more atoms of monoisotopic elements and a lack of interferences.  Important information would be lost if a small +1 Da or +2 Da ion abundance was discarded for failing to meet a minimum abundance criterion.  In fact, establishing a threshold for the isotopic ion abundances discriminates in favor of RIAs inflated by interference contributions.

All measured exact masses and RIAs entered into the ICP, and all compositions found by the ICP as described below, are listed in Table 1.  The treatment of the data for each of the precursor ions is described in what follows.

m/z 166.  The three measured exact masses and two RIAs for the m/z 166 ion were entered into the ICP.  Unique compositions were obtained for the three ions and two neutral losses.  For this low mass, large-mass-defect analyte, the same result was obtained by entering only the three exact masses.

m/z 170.  Entering the two measured exact masses provided seven compositions for each of the two ions and one composition for the neutral loss.  Further, entering the two RIAs for the m/z 170 ion provided unique compositions for the two ions and the neutral loss.

m/z 182.  The three measured exact masses and the RIAs for the m/z 182 ion were entered into the ICP.  Unique compositions were obtained for the three ions and two neutral losses.

m/z 192.  Entering the three measured exact masses and RIAs for the precursor ion provided unique compositions for all three ions and the two neutral losses.

m/z 214.  The four measured exact masses and RIAs for the m/z 214 ion were entered into the ICP.  Two compositions were found for the three largest-mass ions and for two neutral losses.  The same result was obtained when the RIAs for the m/z 158 ion were also entered.  Instead, the mass error limit was reduced to 2 mDa, and unique compositions were found for all three ions and both neutral losses.

m/z 216.  As illustrated in Figure 3, the five measured exact masses and RIAs for the m/z 216 precursor ion were entered into the ICP.  Unique compositions were obtained for all five ions and the four neutral losses.

m/z 285.  Entering the six exact masses for the m/z 285 ion and its product ions, and RIAs for the m/z 285 ion, yielded between three and six possible compositions for all ions and neutral losses.  Also entering the RIAs for the m/z 223 and 161 ions provided unique compositions for the m/z 285, 223, and 161 ions and their corresponding neutral losses, but two compositions for the m/z 187, 125, and 99 ions and for their neutral losses.  The mass error limit was reduced to 2 mDa, but the same list of compositions for the six ions and five neutral losses was obtained.

Note that a %2 RIA value need not be accurate to within 20% to reveal the number of Cl atoms present (32%/Cl atom).  For m/z 186, the two possible compositions contained either 1 or 2 Cl atoms (C4H9ClO4P+ or C5H9Cl2O3+).  The observed %2 RIA was 25.09%.  This value is 23.7% low for the composition containing a single Cl atom and 61.2% low for the composition with 2 Cl atoms; hence, the composition containing 1 Cl atom is taken as correct.  For the m/z 125 ion, either 0 or 1 Cl atom (C2H6O4P+ or C3H6ClO3+) was present.  The measured %2 RIA was only 1.40%.  The composition with no Cl atoms must be the correct one.  Finally, for the m/z 99 ion, compositions with 0 or 1 Cl were possible (H4O4P+ or CH4ClO3+); the measured %2 RIA of 0.44% clearly indicated no Cl atoms were present in the ion.  The corresponding neutral losses must then have included 2, 3, and 3 Cl atoms, respectively.  The charge on all product ions was associated with the highly stable PO4 moiety.

Alternatively, the more reliably measured exact mass differences among product ions can be used to reach these same conclusions.  Only one composition was possible for the m/z 161 ion, C2H7ClO4P+.  The exact mass differences between the m/z 187 and 161 ions (26.0163 Da) and m/z 161 and 125 ions (35.9769 Da) provided unique compositions of C2H2 and HCl, respectively.  Hence, the m/z 187 ion was C4H9ClO4P+ (C2H7ClO4P+ + C2H2) and the m/z 125 ion was C2H6O4P+ (C2H7ClO4P+ - HCl).

m/z 319.  The measured exact masses for the eight ions (the maximum number permitted by a memory restriction) and RIAs for the m/z 319 precursor ion were entered into the ICP.  Between one and six possible compositions were found for all ions and neutral losses.  After also entering the RIAs for the m/z 274 ion or for both the m/z 274 and 246 ions, unique compositions were found for all but the m/z 214 ion and its corresponding neutral loss.  Also, entering the RIAs for the m/z 239 ion provided no compositions due to the large error in its %2 RIA.  Reduction of the mass error limit to 2 mDa, and entering the m/z 319 and 274 RIAs along with the eight exact masses, provided unique compositions for all eight ions and seven neutral losses.

The dependencies of the number of possible compositions on mass (Figure 1a) and non-integer mass (Figure 1b) explain why determining unique compositions for the m/z 285 precursor ion and its product ions presented the most difficulty.  In Figure 1b for an m/z 319 ion and for a similar bar chart for the m/z 285 ion (not shown), the greatest number of possible compositions correspond to a non-integer mass of 0.95, which is close to the non-integer masses measured for the m/z 284.9628 precursor ion and its product ions.  Six of the other seven precursor ions had measured non-integer masses between 0.0790 and 0.1385, for which many fewer compositions are possible.  Based on exact mass alone, a mass error limit of 2 mDa, and consideration of C, H, Cl, N, O, P and S, the m/z 182.0096 ion has only 19 possible compositions due to its low mass, as expected from Figure 1a.  Consistent with Figure 1b, the high-mass, large non-integer mass m/z 319.1039 ion has 82 possible compositions.  However, despite its lower mass, the m/z 284.9620 ion has 140 possible compositions due to its smaller non-integer mass.

Selecting Candidate Standards for Analyte Identification

The oa-TOFMS provided unique and correct compositions for the precursor ions, product ions, and neutral losses for the eight analytes.  The lack of commercial libraries of ESI and APCI mass spectra and tandem mass spectra necessitated hypothesizing possible compound identities based on these compositions.  This task was simplified by noting the strong correlation among compounds used commercially, compounds found in the environment, the number of literature references for compounds, compounds listed in chemical catalogs, and compounds in chemical inventories of laboratories.  SciFinder(15) was used to survey the chemical literature for the known chemical structures of isomers M with compositions corresponding to those deduced for the [M+H]+ ions.  Listed under each structure was the number of literature references associated with it.  This feature made selection of candidate isomers easy and rapid.  Generally, one or a few structures had many more references than all others.  These structures were examined to determine if simple bond cleavages could produce any of the observed product ions and neutral losses, or if the fragmentations could not be explained by the structure.  This approach left one to three candidate compounds that would have been purchased, if commercially available, to identify each analyte through comparison of MS spectra at different CID voltages and retention times, had the simulated unknowns in this study been analytes in environmental extracts.  This strategy might be less successful for production byproducts, transformation products, or emerging contaminants.(16)  The structures of the product ions and the selection of candidate compounds based on structures provided by SciFinder for all eight analytes were provided in reference (10).

Deconvolution of Product Ion Spectra Using the ICP

A single-stage MS cannot selectively fragment precursor ions; all precursor ions were fragmented by in-source CID.  Hence, ions produced from coeluting compounds, or from the solvent system or column bleed when background subtraction is not employed, can be observed in such product ion mass spectra.  To deconvolute composite product ion spectra, the Ion Correlation Program can be used as an Ion Non-correlation Program.  The premise is that, when data for unrelated ions are entered into the ICP, no compositions will be found unless an unrelated ion is a compositional subset of the atoms in the precursor ion.

Initially, the ICP was used to test pairs of ions for correlation.  The compositions listed in Table 2 illustrate why this approach failed.

Table 2.  Possible compositions for the m/z 319 and 119 ions, 319 and 274 ions, and 319, 274, and 119 ions.  The error limits were 3 mDa and 20% and C, H, Br, Cl, N, O, P, and S atoms were considered.

Table 2 - For further information contact grange.andrew@epa.gov

The ions at m/z 319 and 119 are not related, yet compositions were found for both ions, including the three incorrect precursor ion compositions in the left column that contain an O atom.  The single possible composition for the m/z 119 ion forced selection of these three incorrect precursor ion compositions out of many more that were possible during step 3 of ICP processing.  For the center column in Table 2, the related ions m/z 319 and 274 were tested for correlation, and three compositions were found for both the precursor and product ions, including the correct precursor ion composition.  Neither the single possible neutral loss nor the three possible product ion compositions contained an O atom.  Hence, all precursor ions containing an O atom were rejected in step 3.  Because no precursor ion compositions are common to both columns, entering the data for m/z 319, 274, and 119 ions provided no compositions in the right column.  To avoid such false correlations, all related higher mass ions were included when determining if lower mass ions were related to the precursor ion.

To test the ICP as a means of deconvoluting product ion mass spectra arising from coelution of analytes, and for revealing problems that could arise from considering too many superimposed product ion mass spectra at one time, all of the ions in Table 1 from the eight simulated unknowns were considered as one product ion spectrum.  Although coelution of eight analytes is unlikely, recently developed surface sampling ionization techniques(17,18) could produce ions from eight or more compounds simultaneously.

A manual procedure was used.  Data for the highest mass ion, m/z 319, were entered into the ICP with the data for each lower mass sequentially.  If no correlation was found, the second ion was also assumed to be a precursor ion to be correlated with other ions later.  When a correlation was found, both the m/z 319 ion data and that for the correlated ion were entered to test the next lower mass ion.  This process is illustrated in Table 3.

Table 3 - For further information contact grange.andrew@epa.gov

After all lower mass ions were tested, this process was repeated starting with m/z 285, the next highest-mass precursor ion.  Ions that did not correlate with either the m/z 319 or 285 precursor ions were also later treated as precursor ions.  This process was likewise performed for the m/z 216, 214, 192, and 182 precursor ions.  RIAs for all monoisotopic ions with peak areas of 1.7 x 105 counts or more and with masses exceeding 141 Da were entered to maximize discrimination among ions.

Six of the eight precursor ions were found.  The other two were compositional subsets of larger-mass precursor ions.  The m/z 170 ion was correctly found to be C12H12N+, which is a subset of the m/z 319 ion (C17H20ClN2S+), and the m/z 166 ion was found to be C10H16NO+ which is a subset of the m/z 214 ion (C10H16NO2S+) and of the m/z 192 ion (C12H18NO+).  To determine that m/z 170 and m/z 166 were precursor ions, the m/z 319, 214, and 192 compounds would have to be confirmed with standards and found to lack these two ions in their product ion mass spectra.  Entering data for unrelated ions that are compositional subsets of a precursor ion does not result in false non-correlations.

The first column in Table 4 lists the ions with masses of 158 Da or more.

Table 4 - For further information contact grange.andrew@epa.gov

The ICP found that multiple precursor ions could produce all of the lower-mass ions when all 34 ions were tested for correlations.  This occurred because the ion compositions were subsets of multiple precursor ions or subsets of incorrect precursor ion compositions found by the ICP.  As examples, C7H7+ (m/z 91) is a subset of seven precursor ion compositions, and C12H8+ (m/z 152) is a subset of three.  Entering data for the m/z 285, 223, 187, and 161 ions provided the two possible precursor ion compositions, C5H14N2Cl3OP2+ and C6H13Cl3O4P+.  Also entering the data for m/z 132 provided the single precursor ion composition C5H14N2Cl3OP2+ and the m/z 132 ion composition C5H13N2P+.  This false correlation and false correlations with m/z 140 and 117 ions were not made when the mass error was reduced to 2 mDa, after which no compositions were found.

Obtaining a unique and incorrect precursor ion composition is less likely when testing correlation for higher-mass product ions.  The elemental limits provided by precursor ions provide many more possible compositions for lower-mass ions than for those resulting from a small neutral loss.  A small-mass neutral loss has few possible compositions and restricts the number of possible compositions for a product ion formed from a precursor ion.  As the neutral loss mass increases, the number of compositions possible for the product ion increases, since more neutral loss compositions are available for subtraction from the precursor ion compositions.

Despite the loss of unique analyte assignments for lower-mass ions, unique precursor ion compositions were still determined for the m/z 182, 214, and 319 precursor ions using the same assumptions and data entered before, but only for ions with masses greater than 158 Da.  For m/z 192, only its mass and RIAs were available, and the assumption that the precursor ion was protonated was necessary to obtain its unique composition.  This same assumption was necessary to obtain only the correct composition for the m/z 216 ion when the exact masses for the m/z 140 and 132 ions were not entered.  Without entering the exact masses for the m/z 125 and 99 ions, the unique m/z 285 ion composition was obtained by lowering the mass error limit to 2 mDa.

To simulate coelution or surface sampling of three compounds, three sets of ions were considered based on the observed elution order of the analytes producing the precursor ions:  m/z 166, 285, and 216; 214, 319 and 192; and 170 and 182.  The ions from these sets of analytes are grouped in the second through fourth columns in Table 4.

The deconvolution of the first set of ions was nearly complete.  Of the 14 ions considered for the first group of three analytes listed in the second column of Table 4, three ions were found to be precursor ions, nine product ions were correlated only with their precursor ion, and two product ions were each correlated with two precursor ions of which they were compositional subsets.

For the second group of three analytes, multiple correlations occurred for five of 15 ions that were compositional subsets of the correlated precursor ions.  The five smallest-mass ions (less than 120 Da) provided no discrimination among these three compounds.  A unique precursor ion was found for both the m/z 319 and 192 ions without considering these low-mass ions.  Reducing the mass error limit to 2 mDa provided a unique precursor ion composition for the m/z 214 ion as well.  For the group of ions in the last column of Table 4 from the two remaining analytes, all five ions were correlated correctly.

Characterization of Two Unidentified Analytes from Published Data

Ib摙ez et al.(11) published the %1 RIA value and exact mass for the precursor ion, and exact masses for two product ions, from the positive ion fragment spectrum of an unidentified analyte.  The authors determined that the precursor ion was C13H13N2O2+.  The three exact masses and RIA value were entered into the ICP using error limits of 7 mDa and 10%,(11) F was added to the list of elements considered, and the molecule was assumed to be protonated.  The unique compositions obtained were:  C13H13N2O2+, C12H11N2+, and C12H10O+ for the ions, and CH2O2 and CH3N2O for the neutral losses.  The CH2O2 loss was not due to a carboxylic acid group, because the negative full scan spectrum did not reveal an [M-H] ion.(11)  A mass spectral interpretation specialist hypothesized that the lowest-mass ion was the diphenylether ion and that the total neutral loss of CH2O2 resulted from loss of CO from the diphenylether portion of the ion(19) and of H2O from a side chain.  The CH3N2O neutral loss could result from a single bond cleavage and a hydrogen shift.  Figure 5 shows two possible structures.

Figure 5 - For further information contact grange.andrew@epa.gov

Figure 5.  Two possible structures postulated for the analyte whose [M+H]+ ion loses CH2O2 and CH3N2O.

SciFinder was searched for the composition of the molecule, and the structure in Figure 5a returned 18 references while 5 references were associated with the other structure.  Eighteen other structures had between 20 and 88 references cited, but no apparent fragmentation scheme explained the observed ions or neutral losses.  One of the cited references for the structure in Figure 5a listed 4-phenoxyphenylurea as a pesticide.(20)  This tentative identification is by no means certain.  However, it is sufficiently plausible to justify purchase of this compound to compare its retention time and product ion mass spectra to those of the analyte.

Ib摙ez et al.(11) also published exact masses for eight ions formed from an unidentified analyte in a negative product ion spectrum for which they determined the precursor ion composition ([M-H]) to be either C12H6Cl3O5S2 or C10H9Cl3F2O2PS2.  Also provided were the precursor ion RIAs.  The two RIAs and seven largest exact masses were entered into the ICP using the assumptions that at least 10 C and 3 Cl atoms were present based on the %1 RIA of 15.1% and %2 RIA of 106.5%, and that the ion was deprotonated.  The upper elemental limits were 50 C, 100 H, 20 N, 20 O, 4 S, and 3 Cl atoms, and the RIA error limit was 10%.(11)  A 7 mDa error limit(11)  was first used and was reduced in 1 mDa increments.  At 6 mDa, only the correct precursor ion composition was found, and at 3 mDa, no compositions were found.  For a 4 mDa error limit, 1, 1, 2, 1, 1, 2, and 2 possible compositions were found for each of the seven ions in order of decreasing mass.  The deduced composition of the molecule was searched in SciFinder and only one structure was provided, a diphenylsulphone to which three Cl atoms and a sulfonate group were attached at unknown positions.  (No structures were found for C10H10Cl3F2O2PS2).  The eight structures hypothesized for the ions by a mass spectral interpretation specialist based on this information are shown in Figure 6.

Figure 6 - For further information contact grange.andrew@epa.gov

Figure 6.  Eight structures postulated for the precursor ion and seven product ions from an unidentified analyte studied by Ib摙ez et al.(11)  Ring attachment positions are uncertain.  One of several possible isomers is shown for each structure.  The mass errors in mDa are within the parentheses.

Only for the precursor ion and highest-mass product ion did the hypothesized structures agree with the possible composition for each found by the ICP.  The compositions for the other product ions did not contain six or 12 C atoms, or contained multiple O atoms but no S atoms.  No plausible fragmentation schemes could account for these compositions.  If the hypothesized structures were correct, the mass error for the m/z 191 ion was unacceptably large.  Entering the other seven ion exact masses with an error limit of 4 mDa provided 1, 1, 2, 1, 2, 2, and 1 possible compositions for the ions in order of decreasing mass.  Again, only for the precursor ion and largest-mass product ion were the expected compositions found.  The mass errors provided in Figure 6 suggest that the error limit was lowered excessively (four are greater than 4 mDa), and that the expected result of obtaining no compositions as a result did not occur.  For the simulated unknowns, none of the 34 exact measurements had a mass error greater than 2 mDa, which explains why no such rejections of correct compositions occurred.

This unacceptable outcome suggested that a modified approach should be developed.  In Table 5, compositions are listed for the precursor ion and six product ions (all except m/z 191) for an error limit of 6 mDa, the largest mass error limit that provided a single precursor ion.

Table 5 - For further information contact grange.andrew@epa.gov

More compositions are possible for some ions than for the 4 mDa error limit, but now included for each ion is the composition corresponding to a structure in Figure 6.  The exact mass differences between pairs of ions adjacent in mass, and the possible compositions for those masses, are also included in Table 5.  Of the three possible compositions for the m/z 224 ion, only one corresponds to the composition of the m/z 335 ion minus one of the two possible neutral loss compositions for the mass difference between these product ions.  For the lower-mass ion pairs, only one neutral loss composition corresponds to the mass differences and the correct compositions are immediately obtained by subtracting each neutral loss composition from the composition of the higher-mass ion.  The unique neutral losses corresponding to each product ion are then obtained by subtracting the product ion compositions from that of the precursor ion.  This mass difference strategy could also be applied using a lower mass product ion with a single composition to eliminate compositions for a higher mass product ion.

The mass difference between the m/z 399 precursor ion and the m/z 224 product ion was larger than each mass difference between the m/z 399 and 335 ions and between m/z 335 and 224 ions.  Hence, for an error limit of 9.9 mDa (2^.5 x 7 mDa, since a mass difference between two measurements is at issue) only two compositions were possible for each mass difference of 63.9645 (SO2 and CHClO) or 110.9965 Da (C6H4Cl and C5H3OS), and six were possible for the total mass difference of 174.691 Da, including C3H5Cl2O4, C6H4ClO2S, and C9ClO2, the three neutral losses for this exact mass found by the ICP (not shown).  Of these three neutral losses, only C6H4ClO2S is a sum of m/z 64 and m/z 111 neutral losses.  This example demonstrates that considering the mass differences between product ions can further eliminate incorrect compositions.

Product ions adjacent in mass do not always provide mass differences corresponding to one or more atoms.  For the chlorpromazine product ions in Table 6, two different fragmentation paths (319 --> 274 --> 246 --> 214 and 319 --> 274 --> 239 --> 211) resulted in every other ion below m/z 246 having mass differences corresponding to one or more atoms.

Table 6 - For further information contact grange.andrew@epa.gov

Even so, the composition of the m/z 246 ion and the composition corresponding to its mass difference with the m/z 214 ion provide the unique composition for the m/z 214 ion.

Further Work

The ICP is already a powerful aid for identifying compounds, but several improvements are possible.  Starting with an ASCII file containing lists of exact masses, monoisotopic ion peak areas, and retention times, the deconvolution approach should be automated to correlate related ions and, where possible, to provide unique compositions for all ions and neutral losses from one or more precursor ions associated with each chromatographic peak.  Exact mass differences between product ions should be calculated automatically to find unique compositional differences for the purpose of eliminating remaining incorrect compositions as illustrated in Table 5.  The assumptions listed earlier must be codified and mass error adjustments based on the number of compositions obtained must be made automatically.  Finally, the presence of sodiated molecules in Figures 4d and 4f suggests they must be identified as such before the protonated molecule is correctly chosen as the precursor ion.  Inclusion of such an improved ICP in the data system of a single-MS-stage oa-TOFMS would make it nearly as powerful as a more expensive two-MS-stage instrument for deconvoluting product ion mass spectra from a limited number analytes, when each analyte contains a different set of heteroatoms.  In most cases, for ions greater in mass than 120 Da, the ICP software would distinguish among precursor ion sources for ions computationally, rather than through sequential isolation and fragmentation of each precursor ion.  Used together with a two-stage MS/MS instrument, the ICP could often make it unnecessary to perform subsequent experiments involving isolation of multiple precursor ions.

Conceivably, such software could automatically provide a list of unique compositions for the precursor ion, product ions, and neutral losses for the major chromatographic peaks in a total ion chromatogram.  (Limiting HPLC/MS studies to a small minority of compounds that display toxicity21 could be obviated).  HPLC/MS analyses and literature searches would be less arduous than collecting HPLC fractions and then testing their toxicity prior to selection of analytes for HPLC/MS investigation.  Toxicity testing would only be required for unidentified analytes or those of unknown toxicity.  In addition, more complete characterization of environmental extracts would result, and aid in determining pollution sources.

CONCLUSIONS

The orthogonal acceleration time-of-flight mass spectrometer (oa-TOFMS) used here measured exact masses and relative isotopic abundances (RIAs) with sufficient accuracy to determine elemental compositions of protonated molecules in less time and with less effort than are required when using a double-focusing or accurate-mass triple-quadrupole mass spectrometer.  All necessary data were acquired from two full scan data acquisitions; no ion-specific menu preparation was required prior to unnecessary, subsequent data acquisitions.

Mass measurements for 34 monoisotopic ions formed from eight simulated unknown analytes were accurate to within 1 mDa and 2 mDa, for 29 and all 34 measurements, respectively.  Of 36 RIA measurements for +1 Da or +2 Da ions, 35 were accurate to within 16% of their values (or to within 0.38 RIA % for values less than 1%) in the absence of interferences, when the monoisotopic ion peak area was at least 1.7 x 105 counts and for ion masses greater than 141 Da.

Exact masses were measured more reliably than RIAs, and consequently all measured exact masses and only one to three pairs of RIAs were entered into an Ion Correlation Program (ICP) to determine ion and neutral loss compositions.  Generally, a mass error limit of 3 mDa and an RIA error limit of 20% were entered, the elements C, H, N, O, S, P, Cl, and Br were considered, and the precursor ion was not assumed to be a protonated molecule.  In some cases the mass error limit was reduced to 2 mDa, or the precursor ion was assumed to be an [M+H]+ or [M-H] ion, to obtain unique compositions for all ions and neutral losses.

The ICP was also used as an ion non-correlation program to deconvolute simulated composite product ion mass spectra.  When correlations were attempted for all 34 ions, six of the eight precursor ions were identified as such and correct correlations to a single analyte were made for all ions with masses greater than 157 Da, except for the other two precursor ions which were compositional subsets of larger-mass precursor ions.  When the ions from sets of three, three, or two analytes were grouped to simulate composite mass spectra, all eight precursor ions were identified, and their unique compositions were found.  These results indicate that the ICP can compensate for the lack of a second MS stage when coelution of other analytes, solvent system components, or column bleed occurs.

Published exact mass and RIA data were used to characterize two unidentified analytes using the ICP, mass spectral interpretation, and SciFinder.  One compound was hypothesized to be a trichloro sulfonated diphenylsulphone, while the other might be 4-phenoxyphenylurea.  These examples illustrated that mass spectral interpretation skills and a comprehensive data base of the chemical literature can be necessary to make informed hypotheses, even when the compositions of multiple ions and neutral losses from an analyte are known.

When a tentative identification of an analyte is confirmed by comparison of its retention time and ion product mass spectra, acquired at different collision energies, with those of a standard, the number and validity of assumptions made and the specificity of the data acquired upon which the tentative identification was based is unimportant.  But for many analytes, a standard will not be available.  In these instances, the analyte is more thoroughly characterized and the hypothesis that it is one isomer, or one of a limited number of isomers, rests on stronger evidence when unique compositions have been found for the precursor ion, product ions, and the corresponding neutral losses.

Based on the accuracy of the data acquired for this study and the relative ease with which it was obtained, a single-stage oa-TOFMS is the most cost effective instrument for characterizing or identifying compounds separated by HPLC and ionized by ESI or APCI.  Success hinged upon use of an ICP to determine compositions of precursor ions, product ions, and neutral losses.  The ICP also correlated ions from the same analyte, while demonstrating non-correlation for unrelated analytes.  Further development of the ICP might make the single-MS-stage, oa-TOFMS nearly as effective as more expensive, two-stage MS/MS instruments for identifying compounds in environmental extracts.

Acknowledgment.  The authors are grateful to Patrick L. Ferguson for providing access to SciFinder.

REFERENCES

Analytical Environmental Chemistry
ICE Home Page

Environmental Sciences | Office of Research & Development
National Exposure Research Laboratory
Author: Andrew Grange
Email: grange.andrew@epa.gov


Local Navigation


Jump to main content.