Environ. Sci. Technol. 1990, 24, 1856-1860
(21) Zhang, P.; Sparks, D. L. Soil Sci. SOC.Am. J., in press. (22) Atkinson, R. J.; Posner, A. M.; Quirk, J. P. J . Phys. Chem. 1967, 71, 550-558. (23) Carter, D. L.; Mortland, M. M.; Kemper, W. D. In Methods of Soil Analysis, 2nd ed.; Klute, A., Ed.; Soil Science Society of America: Madison, WI, 1986; Part I, pp 413-423. (24) Westall, J. C. FITEQL;Oregon State University: Corvallis, OR, 1982. (25) Eigen, M.; De Maeyer, L. In Techniques in Organic Chemistry, 2nd ed.;Weissberger, A., Ed.; Wiley Interscience: New York, 1963; Vol. 8, Part 2. (26) Patel, R. C.; Taylor, R. S. J . Phys. Chem. 1973, 77, 2318-2323.
(27) Patel, R. C.; Atkinson, G.; Boe, R. J. Chem. Instrum. (N.Y.) 1974,5, 243-255. (28) Bernasconi, C. F. Relaxation Kinetics; Academic Press: New York, 1976; Chapter 1. (29) Rand, R. H. Computer Algebra in Applied Mathematics: An Introduction to MACSYMA; Pitman Advanced Publishing Program, London, 1984. (30) NAG Fortran Library Manual Mark 11; Numerical Algorithms Group Ltd., Downers Grove, IL, 1984.
Received for review April 19,1990. Revised manuscript received July 16, 2990. Accepted July 24, 1990.
Choosing between High-Resolution Mass Spectrometry and Mass Spectrometry/Mass Spectrometry: Environmental Applications M. Judith Charles"
Department of Environmental Sciences and Engineering, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-7400 Yves tondeur
Triangle Laboratories, Inc., 801-10 Capitola Drive, P.O. Box 13485, Research Triangle Park, North Carolina 27709 Selectivity in environmental analyses requires the use of fractionation techniques and HRMS or MS/MS t o eliminate specific and nonspecific interferences. In the analysis of TCDDs and TCDFs, HRMS is the method of choice when specific interferences arising from compounds with molecular or fragment ions can be separated from TCDD and TCDF ions at a resolving power of 10000. In cases where HRMS does not provide adequate selectivity at this resolving power, MS/MS is needed. Analyses on a pulp and paper effluent extract show that MS/MS was able to substantially eliminate interferences due to the presence of methyl and ethyl tetrachlorinated dibenzofurans that were not removed by HRMS at resolving powers of loo00 and 18o00. Nonspecific interferences may also be present due to coelution of compounds that cause changes in the response of the mass spectrometer and are best eliminated by fractionation techniques or by altering conditions of analyses.
Introduction Questions and problems that an analyst encounters in choosing between high-resolution mass spectrometry (HRMS) and mass spectrometry/mass spectrometry (MS/MS) are exemplified in the analysis of tetrachlorinated dibenzo-p-dioxins (TCDDs) and tetrachlorinated dibenzofurans (TCDFs). TCDDs and TCDFs exist in environmental matrices as components of complex mixtures at trace levels (ppt, ppq) compared to the other environmental contaminants. Detection and quantification at these low levels require sensitivity and selectivity. Selectivity is achieved by fractionation techniques in combination with high-resolution gas chromatography and HRMS or MS/MS. Fractionation procedures using acid/ base-treated silica gel, alumina, and carbon columns separate polychorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs) from the bulk material ( I ) , leaving behind coextractables that can act as specific and nonspecific interferences ( 2 ) . These interferences affect instrument sensitivity and stability (1, 2). Specific interferences are due to compounds remaining in sample extracts that exhibit molecular or fragment ions 1856
Environ. Sci. Technol., Vol. 24, No. 12, 1990
that are not resolved from ions monitored for TCDDs and TCDFs at low-resolution MS. These interferences lead to either false positive or negative results. They can be alleviated by optimization of sample fractionation techniques and/or use of HRMS or MS/MS. Examples of specific interferences that can be eliminated by HRMS a t a resolving power of 10000 by selected ion monitoring (SIM) techniques, where ions of a given elemental composition are measured ( 2 , 3 ) ,are nonachlorobiphenyl, DDE, tetrachlorinated methoxybiphenyl, and tetrachlorinated benzyl phenyl ether (2). Pentachlorinated benzyl phenyl ethers, tetrachlorinated xanthenes, and tetrachlorinated methyland ethyldibenzofurans are specific interferences that require HRMS analysis at resolving powers greater than 18000 (3, 9). The presence of polychorinated biphenyls (PCBs) is exacerbated if M - COCl confirming ions of TCDD are monitored. Interferences in M - COCl channels due to the loss of three chlorines from hexachlorobiphenyl and four chlorines from heptachlorobiphenyl require resolving powers on the order of 300000 and 48000, respectively ( 2 , 3 ) .Enhanced selectivity needed to separate these ions from TCDD ions requires the use of MS/MS by selected reaction monitoring (SRM), where the parent ion is transmitted through the first mass analyzer (MS1) to a collision cell area and collisionally induced daughter ions characteristic to the and@ of interest are formed and detected on a second mass analyzer (MS2) (5). Since daughter ion formation is affected by the collision energy, nature of the collision gas, and collision gas pressure, optimization of an MS/MS method requires considering the effect of these parameters on collision-induced dissociations (CID) of the analyte. Nonspecific interferences originate from substances generally of unknown composition and structure that remain after sample cleanup. These interferences coelute within the TCDD/TCDF GC retention time window and result in changes on measurement and detection of TCDDs and TCDFs. In this study, we present results of experiments conducted to optimize MS/MS for the analysis of PCDDs and PCDFs and examples of how specific and nonspecific interferences affect identification and quantification of
0013-936X/90/0924-1856$02.50/0
0 1990 American Chemical Society
TCDDs and TCDFs to illustrate the utility and problems encountered in HRMS and MS/MS analyses. Our objective is not an attempt to demonstrate the general superiority of one approach over another but to examine specific environmental applications showing the advantages of what is increasingly becoming the two most common options offered to analysts for the determination of trace components.
Table I. Measured vs Pressure of Argon in the Collision Cell
Experimental Section Standard Preparation. Standard mixtures were prepared using pure standards obtained from Cambridge Isotope Laboratories, Inc. One mixture used in experiments evaluating optimal collision energy and argon gas pressure containing 20 ng/pL each of 2,3,7,8-tetrachlorodibenzofuran (TCDF), 1,2,3,6,7,8-hexachlorodibenzodioxin (HxCDD), octachlorodibenzodioxin (OCDD), and octachlorodibenzofuran (OCDF) was prepared in toluene. The second mixture in these experiments containing 8.7 pg/pL 2,3,7,8-TCDDand 11.6 pg/pL 2,3,7,8-TCDFwas prepared in dodecane. The third mixture used to measure the detection sensitivity of MS/MS was prepared in toluene and contained 2,3,7,8-TCDD, [13C12]-2,3,7,8-TCDDand ['3C1J-2,3,7,8-TCDF. Sample Preparation. The pulp effluent sample was fortified with an acetone solution containing the labeled internal standards and filtered through glass-fiber filters. The filtrate was solvent extracted with methylene chloride and the filter was Soxhlet extracted with toluene for 16 h. The two extracts were combined, concentrated, and solvent exchanged to hexane before undergoing acid/base partitioning and cleanup involving acid/base-modified silica gel, basic alumina, and AX-21/Celite 545 carbon columns. Instrumental Methods. (a) High-Resolution Gas Chromatography. Experiments determining optimal collision energy and argon pressure conditions on formation of the (M - COCl)'+ daughter ions was conducted using an OV-1, 25-m-length, 0.25-mm-i.d. fused-silica capillary column. The GC oven was held a t an initial temperature of 100 "C for 1 min and then temperature programmed at a rate of 15 "C/min to 200 "C and 10 "C/min to 300 "C. The experiment optimizing collision gas pressures was conducted with a DB-5,60-m-length, 0.25mm-i.d. fused-silica capillary column and a temperature program of 100 "C for 1 min followed by a temperature rise of 20 "C to 190 "C and 3 "C/min to 300 "C. Sample extracts were analyzed under these conditions and a temperature program of 100 "C for 1 min and a rise of 8 "C/min to 300 "C. (b) HRMS Selected Ion Monitoring. Experiments were performed on VG70S and VG70-250SEQ mass spectrometers operating in the selected ion recording mode with resolving powers of 10000 and 18000 (10% valley definition). A high-sensitivity electron ionization source, as supplied by the manufacturer at a source temperature of 250 "C, electron energy of 34 eV, and a filament emission current of 0.5 mA was used throughout the study. TCDD ions monitored were m / z 319.8965 (M)'+ and 321.8936 (M + 2)'+; I3C labeled TCDD ions monitored were m / z 331.9368 and 333.9334 (M + 2)'+. A PFK lock mass ion m / z 318.9792 was monitored a t a resolving power of 10000, and mlz 316.9824 was monitored a t a resolving power of 18000. Dwell times ranged from 40 to 60 ms. (c) MS/MS Selected Reaction Monitoring. A VG250SEQ (EBqQ configuration) hybrid mass spectrometer was used to conduct the MS/MS analyses. Parent ions of PCDDs and l3CI2-labeledTCDD ( m / z 332 and 334) were transmitted into the collision region by using MS1 in the
Pressures of argon inside the collision cell are approximately 127 times those indicated on the ion gauge
(df'+
measd pressure on ion guage, mbar 3x 1x 3x 1x
pressure in collision cell,' mbar 4 x 10-4 1 x 10-3 4 x 10-4 1 x 10-2
10+ 10-5 10-5 10-4
Table 11. Ions and Transitions Monitored in Collision Energy and Collision Gas Experiments analyte 2,3,7,8-TCDF
ion
M M+2 2,3,7,8-TCDD M M+2 M 13C-labeled2,3,7,8 M+2 TCDD 1,2,3,6,7,8-H~CDD M + 2 M+4 M t 2 OCDF M+4 M+4 OCDD
transition monitored 304' 306' 320' 322' 332' 334' 390' 392' 442' 444+ 460'
------
241' + C035Cl 343+ t c035c1 257' + C035Cl 259' + C036Cl 268' + 13C036Cl 270' + 13C035C1 327' + C035Cl 329+ + C035Cl 379+ + c035c1 444+ + c035c1 460' + C035Cl
selected ion monitoring mode a t a resolution of -700. Selected reaction monitoring of the (M - COCl)*+daughter ions formed during collision-induced dissociations was completed on MS2 set a t unit mass resolution. The first optimization experiments were conducted a t argon pressures of 0.5 X lo4 to 1 X mbar by varying collision energies across the peak in 10-eV increments. Readings on the gauge are calibrated relative to nitrogen, requiring that a correction factor be used to calculate the pressure inside the collision cell based on conductance and pumping speed. Ion gauge readings vs pressures within the collision cell are presented in Table I. The pressures as read on the ion gauge are used in presentation of the data. The second optimization experiment was conducted a t argon pressures of 1 X and 1 X lod4mbar a t a 3X collision energy of 20 eV. The ions and transitions monitored in HRGC/MSMS analyses are shown in Table 11.
Results and Discussion Optimization of HRGC/MS/MS Conditions. Initial experiments were conducted to determine the optimal collision energy and collision gas pressure on CID for PCDDs and PCDFs. Argon was chosen as the target gas because it has a relatively high ionization potential and does not readily promote ion/molecule reactions. The results are presented in Figure 1 as plots of response vs collision energy at specific measured pressures. For all the compounds examined, the formation of the (M - COCl)*+ daughter ions maximized at the highest pressure examined, 1 X loT5mbar, and at collision energies of 20-40 eV. At collision energies greater than 60 eV a decrease in response was observed, most likely due to more extensive fragmentation of the parent ion. The results also show that under unimolecular dissociation conditions the (M COCl)*+daughter ion is not the predominant fragment ion. A second experiment was performed at higher pressures, 1x and 1 x IO4 mbar, at a collision energy 3x of 20 eV to determine optimum pressure conditions for the formation of M - COCl daughter ions of 2,3,7,8-TCDF and 2,3,7,8-TCDD. The results are presented in Table 111. The data show that the formation of the (M - COCl)'+ Environ. Sci. Technol., Vol. 24, No. 12, 1990
1857
m
3500
400 -
?
d
N
m
.
N
N
m 2 E
+
N
4 , d
N
. E
N
2000
a . rm
200 -
m
N
N
z
I
0
Collision Energy (eV)
Collision Energy (eV)
Flgure 1. Effect of collision energy and argon pressure on formation of (M - COCI)'+ daughter ions of PCDDs and PCDFs.
Table 111. Response of 2,3,7,8-TCDD and 2,3,7,8-TCDF as a Function of Argon Pressure
pressure, mbar 1 x 10-5
3 x 10-5
mean area f SD (f% variability) 2,3,7,8-TCDD 2,3,7,8-TCDF 3.41 f 0.32 (9.34) n=2 6.60 f 0.22 (3.33)
1 x 10-4
n=4 11.64 f 2.0 (17.18) n=2
5.14 f 0.29 (5.64) n=2 10.27 f 1.08 (10.52) n=4 19.78 f 3.38 (17.08) n=2
daughter ions is enhanced by increasing the collision gas pressure. An increase greater than 3-fold was observed at a pressure of 1 X mbar over the response at 1 X mbar, as one would expect since at higher pressure the probability for an ion to undergo single or multiple collisions increases. Detection Sensitivity of TCDD by HRGC/MS/MS. The instrumental detection sensitivity based on signalto-noise (S:N) measurement on 2,3,7,8-TCDD was calculated a t a collision energy of 20 eV and a collision gas pressure of 1 X lo4 mbar. The masses of the M'+ and (M + 2)*+ ions for 2,3,7,8-TCDD and [l3CI2-2,3,7,8-TCDD, making a total of four ions, were monitored by SRM with a dwell time of 50 ms per ion. A S:N ratio of 7:l was measured on 1 pg of 2,3,7,8-TCDD. Other detection sensitivities reported on MS/MS instruments are summarized in Table IV and are dependent on how the measurement is made (e.g., dividing by the peak-to-peak baseline noise or half the peak-to-peak baseline noise, and the number of ions monitored). An accurate interlaboratory comparison requires that the determination be made in the same fashion. Detection Sensitivity of TCDD by HRGC/HRMS. Typical signal-to-noise ratios obtained by HRGC/HRMS (10000 resolving power) for 2,3,7,8-TCDD are 50:l on 0.5 Pg. 1858
Environ. Sci. Technol., Vol. 24, No. 12, 1990
Table IV. Comparison of Detection Sensitivities for 2,3,7,8-TCDD by HRGC/MS/MS
no. ions monitored
ref
Clement et al. (6) Schellenberg et al. (7) Tondeur et al. (3) Reiner et al. (8) this study Signal height was divided by noise.
11
1
loo
S:N
1OO:l (12 pg) 3:l (450 fg)O 4:1 (2 Pg) 1O:l (500 fg) 4 7:1 (1 p d half the peak-to-peak baseline 6 3
IO,OOO Resolution m/z 319.8965
%
O
8,000Resolution
I
,
1
L
!?
I001
MSlMS
I 0
mlz 257 from m/z 320
1
20 22 24 Time (minutes) Flgure 2. Comparison of HRGClHRMS at resolving powers of 10 000 and 18 000 of (SIM of m l z 319.8965) and HRGUMSlMS (SRM of m l r 257) on a pulp effluent extract. 16
18
Interferences. (a) Specific Interferences. An illustration of this type of interference was encountered in the analysis of TCDD in a pulp effluent extract (see Fig-
-
-INTERFERENCES
1I
loo
'3Cl,-2.3,7,8.TCDD
I (1 /
1,
"iDS5
10.000 Resolution
50-
HRMS 10,OOo R.P.
'%12-2.3,7,bTCW
I
-8 "1
1I = c
2
I
08225
18,000 Resolution
-w 501: 0'
I
1
I
I001
M
b
"1
DE5
MSIMS
mlz 254 from mlz 318
mlz 268 from mlz 332
0
lime
0
16
18
20
22
24
Time (minutes)
Figure 4. Comparison of GC and mass spectrometric resolution of ethyl tetrachlorinated dibenzofwans and ['%,,]-2,3,7,8TCDD on DB-5 and DB-225 GC columns and by HRMS and MSIMS.
Flgure 3. Comparison of HRGCIHRMS at resolving powers of 10 000 and 18 000 (SIM of m l z 331.9368) and HRGCIMSIMS (SRM of m l z 268) on a pulp effluent extract.
ures 2 and 3) on a 60-m DB-5 fused-silica capillary GC column. Interferences caused by methyl and ethyl tetrachlorinated dibenzofurans appear in ion channels used to monitor the molecular ion of TCDD, m/z 319.8965 (Figure 2), and the isotopically labeled standard of 2,3,7,8-TCDD, m/z 331.9368 (Figure 3). In Figure 3, [13C12]-1,2,3,4-TCDD was spiked into the sample as a recovery standard and elutes prior [l3CI2]-2,3,7,8-TCDD.Since GC conditions were not optimized for isomer specificity these isomers are not well resolved. A resolving power of 18OOO is required to separate the (M + 2)'+ ion of methyl tetrachlorinated dibenzofurans ( m / z 319.9143) from the molecular ion of TCDD and a resolving power greater than 85000 is required to resolve the ethyl tetrachlorinated dibenzofurans from the molecular ion of [13C12]TCDD(9). Increasing the mass spectrometer resolving power from 10000 to 18000 decreases sensitivity and the effect of the interferences in both samples, although as expected, interferences due to the presence of ethyl tetrachlorinated dibenzofurans remain. Here, HRGC/MS/MS is the only technique that can remove these specific interferences without a significant loss in sensitivity. Methyl and ethyl tetrachlorinated dibenzofurans do not coelute with 2,3,7,8-TCDD (Figure 2) or [13C,2]-2,3,7,8TCDF (Figure 4) on a DB-5 60-m GC column. Ethyl tetrachlorinated dibenzofurans, however, do coelute with TCDFs on a DB-225 column (Figure 4) required for identification and quantification of the 2,3,7,8-TCDF isomer. In the analysis of the pulp effluent extract at a resolving power of 5000, a 38% deviation from the theoretical isotope ratio between ions mlz 315.9419 and 317.9389 was observed. This deviation was reduced to 18% at a resolving power of loo00 but still falls outside the limit of a 15% deviation acceptable by EPA in method 8290. The analysis of TCDFs in the presence of methyl and ethyl tetrachlorinated dibenzofurans is thus a case where MS/MS is needed. (b) Nonspecific Interferences. Nonspecific interferences of (1,3) TCDD/TCDFs, as before, coelute within the TCDD/TCDF GC retention time window. In this case the effect is an ephemeral alteration (Figure 5B) in the response of the mass spectrometer during the time period in which the extraneous substance($ is(are) present in the
-
.-J
n
l
W
L
0 '
Time
Figure 5. Mass chromatogram of PFK lock-mass ion in the absence (A) and presence (B) of nonspecific interference(s).
ion source. This change can be a source of poor recoveries if it occurs during elution of either the recovery or internal standard. If these alterations happen at any other time they will remain unobserved unless mass chromatograms for nonanalyte ion(s) from a reference compound such as PFK (2, 4 ) are monitored. In HRMS analyses, a small amount of a reference compound is constantly bled into the ion source via the septum inlet to provide reference ions to compensate for mass drifts during the analysis. Experimentally, mass drifts can be observed by monitoring the ion signal of the reference compound. In the example presented in Figure 5 , a constant amount PFK was bled into the ion source and the ion signal at mlz 331.9792 was monitored over the time period coincident with elution of TCDDs. The ion signal remains steady with time (Figure 5A) in the absence nonspecific interferences. In the presence of these compounds, deviations in the form of negative deflections (Figure 5B) are observed that also occur while other lock masses or other congener groups are being monitored. Two explanations are possible for nonspecific interferences. The first is that the reduction or loss of the lockmass ion results from ion source detuning, possibly due to space-chargeeffects that occur when large (e.g., 50 ng/s at the apex of the peak) quantities of materials enter the ion source. In this situation, the problem may be solved by additional sample fractionation techniques. In Figure 6, the SRM profile of a sample containg trace levels of 2,3,7,8-TCDD in the presence of high levels of PCBs is shown. The negative deflections at the PFK lock mass demonstrate that specific interference may also act as nonspecific interferences. In this case, neither HRMS nor Environ. Sci. Technol., Vol. 24, No. 12, 1990
1859
A
-
m / r 320 -mi2
2 5 7 (Unlabeled T C D D )
385
i . 1
7
B
-
m l r 322
C
-
m l z 331
D
-
m/z 332
m l z 2 5 9 (Unlabeled TCOOl
-
m l z 243 ( P F I O
9818
m l r 2 6 8 (Labeled TCOD's)
-
500
~J,~.~-TCDO
E
-
m/z 334
-
m l z 2.70 ( L a b e l e d TCOD's)
522
T I M E Iminutrs)
Fwre 6. Selected reaction monitoring profiles from HRGC/MS/MS of 2,3,7,&TCDD, ['3C12]-2,3,7,8-TCDD,and [13C12]-1,2,3,4TCDDin the presence of-Aroclor 1260.
MS/MS are able to solve the problem encountered. The second explanation is that coelution of nonspecific interferences, although mass spectrometrically resolved from the lock-mass ion, appears in the mass window ke., 200 ppm a t 10000 resolving power) of the lock-mass ion. When this happens the computer calculates the centroid of the lock-mass ion peak incorrectly. The result is a reduction of sensitivity because the side (or outside) of the analyte peak is monitored instead of the apex. This type of interference may be eliminated by monitoring a lockmass ion different from the same reference compound, selection of another reference compound, and/or fractionation techniques.
Conclusions Fractionation techniques must be used in tandem with mass spectrometry in the analysis of trace levels of contaminants present as components of complex mixtures in environmental samples. HRMS and MS/MS are techniques that can be used to achieve the necessary selectivity in these analyses. An intelligent decision between these techniques requires information about the type of interferences. In analyses of TCDDs and TCDFs, nonspecific and specific interferences arise. When specific interferences are present, HRMS can be used to separate interferences that mass resolve from the analyte of interest at resolving powers less than or equal to 10000. Above this resolving power, the decrease in sensitivity that occurs as the resolving power is increased can limit the practicality of HRMS analyses. In these cases, MS/MS can be the mass spectrometric technique of choice. The presence of nonspecific interferences cause detuning of the ion source
1860
Environ. Sci. Technol., Vol. 24, No. 12, 1990
in HRMS or MS/MS or a deviation from the correct lock mass during HRMS analyses. In either case, a loss in instrument sensitivity results that affects analyses by HRMS and MS/MS. These interferences can be removed by changing conditions of analyses or by chromatographic techniques.
Literature Cited Smith, L. M.; Stalling, D. L.; Johnson, J. L. Anal. Chem. 1984,56, 1830-1842. Tondeur, Y; Niederhut, W. E.; Campana, J. E.; Missler, S. R. Biomed. Environ. Mass Spectrom. 1987, 14, 443-441. Tondeur, Y.; Beckert, W. F.; Billets, S.; Mitchum, R. K. Chemosphere 1989, 18, 119-131. Tondeur, Y.; Hass, J. R.; Albro, P. W.; Schroeder, J. L.; Chae, K. Abstract of Papers, 29th Annual Conference on Mass Spectrometry and Allied Topics; Minneapolis, MN, 1981. Charles, M. J.; Green, B.; Tondeur, Y.; Hass, J. R. Chemosphere 1989, 19, 51-57. Clement, R. E.; Bobbie, B.; Tanguchi, V. Chemosphere 1986, 15, 1147-1156. Schellenberg, D. H.; Bobbie, B.; Reiner, E. J.; Taguchi, V. Y. Rapid Commun. Mass Spectrom. 1987, 1 111-113. Reiner, E. J.; Schellenberg, D. H.; Taguchi, V. Y.; McCurvin, D. M.; Clement, R. E. Abstract of Papers, 37th ASMS Conference on Mass Spectrometry and Allied Topics; Miami, FL. Buser, H. R.; Kjeller, L. C.; Swanson, S. E.; Rappe, C. Enuiron. Sci. Technol. 1989, 23, 1130-1137. Received for review March 26, 1990. Revised manuscript received J u n e 25, 1990. Accepted July 2, 1990.
