Anal. Chem. 1990, 62, 1696-1700
1696
Nonpolar Solvents for Normal-Phase Liquid Chromatography and Postcolumn Extraction in Thermospray Liquid ChromatographyIMass Spectrometry Damii Barcel6* and Gael Durand Environmental Chemistry Department, CID-CSIC, Jordi Girona 18-26, 08034 Barcelona, Spain
Robert J. Vreeken, Gerhardus J. de Jong, and Udo A. Th. Brinkman Department of Analytical Chemistry, Free University, De Boelelaan 1083, 1081 HV Amsterdam, T h e Netherlands
The use of n-hexane, cyclohexane, and dichloromethane as liquid chromatographlc eluents has been explored In thermospray llquld chromatography/mass spectrometry (TSP LC/MS). By use of poSnlve and negatlve Ion modes (PI and NI, respectlvely), and the fllament-on mode, three dlfferent groups of model compounds have been studled: organophosphorus pestlcldes, chlorophenols, and chlorlnated phenoxy acid herbkldes. I n the P I mode, [M HI+ was the base peak for the organophosphorus pestlcldes In all solvents, and [MI", [M - HI+, or [M - CI]' were the base peaks for the other compounds depending on the solvent used. I n the NI mode, the functlonal group fragment [FGI-, [M - R]- (R being a]-,and [M - HCII- were the base methyl or ethyl), [M peaks for the organophosphorus pesticides. Depending on the compound and the solvent used, [M - HCII-, [M - HI-, [M HI-, and [M CIT were the base peaks for the chlorophenols and the chlorlnated phenoxy aclds. The formation of a chloCII-, was more important when rlde attachment ion, [M cyclohexane or dlchloromethane were used as eluents. By use of P I and N I modes under full scan condltlons, detection limits In the low nanogram range were obtalned. The sensltlvlty In the P I mode Improved 1 order of magnltude compared wlth previous studies uslng reversed-phase eluents In TSP LCIMS. Two applkatlons are reported: (1) the analysis of organophosphorus pestlcides, uslng n-hexane-dlchloromethane (5050) as LC eluent, and (2) the use of a postcolumn extractlon system wlth cyclohexane-dlchloromethane1-butanol (45:45:10) as extractlon solvent after the Ion-suppressed LC separation of Chlorinated phenoxy aclds uslng acetonltrlle-water contalnlng a nonvolatlle buffer.
+
+
+
+
+
INTRODUCTION Thermospray liquid chromatography/mass spectrometry (TSP LC/MS) is a well-established technique with either acetonitrile-water or methanol-water mixtures as eluents (1-3). Regarding the effect of the eluent composition, it has been demonstrated both in filament-on and filament-off TSP LC/MS that the sensitivity increases at higher water percentages ( 1 , 2 , 4 , 5 ) . Most of the work carried out with TSP LC/MS uses reversed-phase LC with ammonium acetate as the ionizing additive; the addition of this salt is necessary for the TSP ionization of neutral compounds (6). It is generally recognized that, with neutral compounds, a complex ionization process takes place with the various reactant ions generated from the LC eluent used (7-10). In previous papers we have demonstrated that the addition of ammonium formate (3) or chloroacetonitrile (11) can be efficiently used for the char-
* To whom correspondence should be addressed.
acterization of a variety of chlorinated pollutants. The aims of this paper are to explore (1)the influence of the normal-phase LC solvents n-hexane, cyclohexane, and dichloromethane on the chemical ionization process during the generation of TSP mass spectra of organophosphorus pesticides, chlorophenols and chlorinated phenoxy acids and (2) the use of a postcolumn extraction system using a nonpolar solvent for the analysis of chlorinated phenoxy acid herbicides with a mobile phase containing a nonvolatile buffer. All experiments have been performed by using the filament-on mode of operation. The results are compared with those obtained in reversed-phase thermospray LC/MS (3,4,11,12) and direct liquid introduction (DLI) LC/MS (13-16).
EXPERIMENTAL SECTION Chemicals. HPLC grade cyclohexane, dichloromethane, nhexane and 1-butanol (SDS, Peypin, France) were passed through a 0.45pm filter (Scharlau, Barcelona, Spain) before use. Analytical reagent grade fonofos, trichlorfon, parathion-ethyl, fensulfothion, phosmet, chlorpyrifos, ronnel, (2,4-dichlorophenoxy)aceticacid (2,4-D),(2,4,5-trichlorophenoxy)aceticacid (2,4,5-T),and (2,4,5trich1orophenoxy)propionic acid (Silvex) were purchased from Polyscience (Niles, IL), 2,4-dichlorophenol from Merck (Darmstadt, FRG), 2,4,5-trichlorophenol from Fluka (Buchs, Switzerland), and pentachlorophenol from Supelco (Bellefonte, PA). The chemical structures of all these chemicals used as model compounds are given in Figure 1. The sodium phosphate and the phosphoric acid were purchased from Merck (Darmstadt, FRG). Chromatographic and Postcolumn Extraction Conditions. Eluent delivery was provided by a high-pressure pump (Model 510, Waters Chromatography Division, Millipore, MA). Injection was carried out with a Model 7125 injection valve with a 20-pL loop from Rheodyne (Cotati, CA). A 25 X 0.40 cm i.d. stainless-steel column packed with 7-pm Nucleosil silica material from Macherey-Nagel (Dueren, FRG) was used. Four different LC mobile phases were used during flow injection analysis or chromatographic analysis of the test compounds, namely, cyclohexane, n-hexane, dichloromethane, dichloromethane-n-hexane (5050), all at a flow rate of 0.7 mllmin. The postcolumn extraction system consisted of a home-made sandwich phase separator equipped with two stainless steel blocks with and without a groove and a PTFE disk with a groove (27). The extraction solvent was delivered by a second high-pressure pump (Model 510, Waters Chromatography Division, Millipore, MA). A schematic diagram of the experimental setup is given in Figure 2. Samples were introduced via a Rheodyne (Cotati, CA) injection valve equipped with a 20-wL loop. The organic normal phase of cyclohexane-dichloromethane-1-butanol (45:45:10) was added to the aqueous stream via a Valco T-piece (Houston, TX) with a 0.25-mm bore. The extraction took place in a 1.5 m X 0.8 mm i.d. stainless steel capillary (helix diameter, 40 mm). After separation of the phases, the analytes together with the organic normal phase were introduced into the TSP LC/MS. With this phase separator,a purely organic normal phase can be obtained. The organic flow through the detector was regulated by means of a PTFE capillary equipped with a restrictor.
0003-2700/90/0362-1696$02.50/00 1990 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 62, NO. 15, AUGUST 1, 1990
1697
a
9 0
40
80
120 Mass/ Charge
160
@
-.
ICI1' c111
WQSMEl
200
4.OE4 120
m
P n5
149
CHLORPYRlFOS
*
o 40
Figure 3. (A) Reagent Figure 1.
Chemical structures of model compounds used.
~
1
.--JTSP
1 Column
u
LC/MS INTERFACE
Figure 2. Schematic dhgram of the postcolumn extraction setup used.
A 10 cm X 0.4 cm i.d. stainless-steel column packed with 5-rm Lichrosorb RP 18 material (Merck, Darmstadt, FRG) was used. The flows of the aqueous phase (water-acetonitrile, 5050 + 0.1 M phosphate buffer (pH = 2.5)) and organic phase (cyclohexane-dichloromethane-n-butanol, 45:45:10) were both 1 mL/min; the flow to the mass spectrometer was adjusted to 0.8 mL/min. Mass Spectrometric Analysis. A Hewlett-Packard (Palo Alto, CA) 5988A TSP LC/MS quadrupole mass spectrometer and a Hewlett-Packard Model 59970C computer for data acquisition and processing were employed. The TSP temperatures were as follows: stem, 86 "C; tip, 176 "C; vapor, 216 'C, and ion source, 240 "C. The filament-on mode was used in all cases. Due to the relative high volatility of the solvents used, the variation of the relative abundance of the total ion current with the stem temperature is not the same as when reversed-phase LC eluents are used (3). Therefore, the optimum stem temperature was set at 86 "C, i.e. 14 "C lower than when using methanol-water or acetonitrile-water mixtures. As in earlier experiments with nonpolar solvents (9),an optimum flow rate of 0.7-0.8 mL/min was used, which is considerably lower than in previous reports, where nhexane was used at a flow rate of 3 mL-min with a Vestec TSP LC/MS interface (2). The model compounds (ca. 500 ng each) were injected into the TSP-MS system via a flow injection system.
RESULTS AND DISCUSSION Reagent Gas. Depending on the solvent used different ions appeared in the reagent gas spectrum. In the PI mode, when cyclohexane was used, the main ions had mlz values of 83 and 93 with relative abundances of 100% and 90%, respectively; for n-hexane the main ions had m / z values of 105 and 125 with relative abundances of 100% and 8070,respectively. In
80
120 Mass/Charge
160
200
gas spectrum of n-hexane-dichloromethane
(5050)under PI mode TSP MS. (B) Reagent gas spectrum of cyclohexane-dichloromethane-1-butanol (4545:10) in N I TSP MS.
the case of dichloromethane the main reagent gas ions had mlz values of 97 and 129 with relative abundances of 100% and 7070, respectively. In the NI mode, for cyclohexane and n-hexane the main ions had mlz values of 84 and 105 with relative abundances of 75% and loo%, respectively. For dichloromethane the main ions were the C1- and the CH2C13-ions at m / z 35 and 119 with relative abundances of 75% and 100%, respectively. Some of these reagent gas ions have also been detected in DLI LC/MS when using n-hexane or cyclohexane (13). Further experiments were carried out in the P I mode with n-hexane-dichloromethane (5k50) which was used for separation of the organophosphorus pesticides and in the NI mode with cyclohexane-dichloromethane-1-butanol (45:45:10) as extraction solvent used in the postcolumn extraction system (18, 19). The complete spectra of these reagent gases are shown in Figure 3. For the first mixture the reagent gas spectrum is very complex and not yet fully understood. In the case of the postcolumn extraction a signal appeared at m/z 109 (chloride attachment to 1-butanol) with a relative abundance of 10070. Furthermore there were signals a t m / z 35 and 119 from dichloromethane. No signal from the cyclohexane was found, because of the higher electron-capture ability of the other two solvents in the mixture. From the above it is obvious that in most cases the scan range has to be restricted to m / z values larger than 140 in order to avoid interferences from the reagent gas ions. Mass Spectra in PI Mode. For all analytes studied Table I shows the ions, their relative abundances and their proposed identities with n-hexane, cyclohexane and dichloromethane as carrier streams. The organophosphorus pesticides fonofos, fensulfothion, parathion-ethyl, trichlorfon, chlorpyrifos, phosmet, and ronnel exhibited an [M + H]+ ion as base peak in all three solvents, as in DLI LC/MS with acetonitrilewater mixtures (16). With the chlorophenols studied, only 2,4-dichlorophenol and 2,4,5-trichlorophenol gave a response in the P I mode, with [M - H]+ as base peak when using cyclohexane or dichloromethane as carrier stream. When n-hexane was used, no signal was obtained for the chlorophenols. For pentachlorophenol the formation of positive ions does not take place in the normal-phase eluents, because of the higher electronegativity of this compound in comparison with the lower chlorinated chlorophenols. When reversed-phase eluents were used in TSP LC/MS, none of the chlorophenols gave a response (12).
1698
ANALYTICAL CHEMISTRY, VOL. 62,NO. 15, AUGUST 1, 1990
Table I. Mass Fragments a n d Relative Intensities of t h e Model Compounds i n Flow Injection TSP M S Using n -Hexane, Cyclohexane, or Dichloromethane a s t h e Carrier Stream"
mol wt 246 256
291
compound
fonofos trichlorfon
parathion-ethyl
308
fensulfothion
317
phosmet
320
ronnel
349
chlorpyrifos
162
196
264 220
254
268
2,4-dichlorophenol
2,4,5-trichlorophenol
pentachlorophenol 2,4-D
2,4,5-T
Silvex
m/z
fragment
169 247 241 257 291 512 262 291 292 325 169 279 309 157 318 141 305 321 169 212
[FGI[M + H]+ [M-R][M + H]+ [M + Cl1[2M]+ [M - R][MI*[M + H]+ [M - H + C1][FGI[M - R][M + H]+ [FGI[M + H]+ [FGI[M - R][M + H]+ [FGIIM - (C&O)zPo]-IM - "211iM]'. [M + H]+ [M - HI+/ [M - HI[M + C11[2M - H)[2M + C1][M - HI+/ [M - HI[MI'[M + C11[2M - 2C1][2M - C1][ 2M - HI[M - HI-
313 349 350 161 197 323 359 195 196 231 323 357 391 263
220 [Ml'+/[Ml'219 [M - HI+/ [M - HI255 [M + C11218 [M - HCl]? 219 [M - C1]+ 253 [M - H]+/ [M - HI254 [MI'+ 289 [M + C11508 PMI232 [M - HCl]267 [M - HI268 [MI'+ 269 [M + HI303 [M + Cll-
Table 11. Detection limits (Full Scan a n d Filament-on Mode) for t h e Organophosphorus Pesticides, Chlorophenols, a n d t h e chlorinated Phenoxy Acid Herbicides Obtained with Different TSP LC/MS Techniques.
relative intensity, 70 PI NI compounds
100 100
organophosphorus pesticides chlorophenols phenoxy acid herbicides
20 100
100 20
detection limits, ng PI NI NP" RP NP" RP 1-10 2-206 2-20
20-50 >200 >200
10-100 1-10 1-10
70-175 1-10 1-10
" N P = normal phase (this paper). R P = reversed phase (from refs 3, 4, 11, and 12). *Except pentachlorophenol (see text).
100 10
100 153 100 80 100
100 100 70 100 100 40 60 100
15*s3 100 1002~3 1001 100 152f3 30
1002~3 io01 203 1002~3 252 252 302,1003 100
1003
1001~2
1001~2
203 701, 1002~3 1001~2, 103
1001*2 202 202~3 1003 102,
1003
103 50 50 100 100 701, 10023
" Relative intensities in n-hexane, cyclohexane, and dichloromethane are indicated by the superscripts 1, 2, and 3, respectively. No superscripts are used when the relative intensities were the same in all cases. FG = specific group fragment of the organophosphorus pesticide, R = methyl or ethyl, depending on the substituent. In other words, the use of normal-phase eluents offers distinct advantages. Remarkable mutual differences in signal intensities and fragmentation patterns were observed for the phenoxy acid herbicides 2,4-D, 2,4,5-T, and Silvex using the present solvents. Good sensitivity for these compounds is achieved here for the first time in T S P PI LC/MS. The base peaks correspond to
[MI'+ for 2,4-D, 2,4,5-T, and Silvex when using dichloromethane as carrier stream. However, when using cyclohexane or n-hexane, the base peaks correspond to [M - H]+ and [M - C1]+ for 2,4-D and 2,4,5-T, respectively. For Silvex, [MI'+ was the base peak in all three solvents. In DLI with reversed-phase LC, [MI'+ and [M - H]+ have also been observed for 2,4-D and 2,4,5-T, with relative intensities below 11% (20). When dichloroethane was used in a postcolumn extraction setup, [M + H]+ had a relative abundance of 90% for 2,4-D (13). Mass Spectra in NI Mode. For all test compounds Table I also shows the ions, their relative abundances, and their proposed identities in the NI mode for the same three solvents as above. The NI T S P mass spectra of the organophosphorus pesticides fonofos, fensulfothion, and phosmet are characterized by their functional group ion, [FGI-, as base peak. [M - R]was the base peak for parathion-ethyl and ronnel whereas chlorpyrifos gave [M - Hell- as base peak. The fragmentation that is observed for some of the organophosphorus pesticides is similar to that previously observed when using reversedphase eluents in DLI LC/MS (16), in TSP LC/MS, with reversed-phase eluents ( 4 , 5 ) ,and in negative chemical ionization GC/MS (21). For trichlorfon, [M + C1]- was obtained as the base peak. This indicates a high degree of chloride attachment that can possibly be attributed to the availability of chlorine from pyrolysis of the analyte, which was previously observed with reversed-phase eluents in TSP LC/MS ( 4 ) . 2,4-Dichlorophenol and 2,4,5-trichlorophenol exhibit [M HI- as base peak in n-hexane, and [M + C1]- in cyclohexane and dichloromethane. Pentachlorophenol has [M - HI- as base peak in all three solvents. The [M - HI- ion has previously been reported using reversed-phase LC eluents in DLI (14) and TSP (12)LC/MS. In the latter case even an ion at m / z [M + HI- was obtained with a high relative abundance. The formation of [M + C1]- as base peak is particularly noteworthy because this behavior is the same as when chloroacetonitrile was added to a reversed-phase LC eluent in TSP LC/MS (11). In DLI LC/MS the [M + C1]- ion was less abundant than the [M - HI-ion, which was the base peak after the addition of 1%chloroacetonitrile to the LC eluent (14). Major peaks in NI for 2,4-D correspond to ions produced by the resonance electron capture and the chloride attachment process, i.e. [MI'- and [M + el]-, respectively. When dichloromethane was used, only the [M + C1]- ion was observed. For 2,4,5-T, the base peak corresponded to [M + el]-, with dichloromethane as carrier stream, while [M - HCl]- was the base peak with the other two solvents. For Silvex the main ions were [M HI- and [M + C1]-. Most of the ions tabulated in Table I for the chlorinated phenoxy acids in NI have previously been reported in DLI LC/MS (14,20). Even the unusual formation of [M + HI- for Silvex has been observed with acetonitrile-water mixtures in T S P LC/MS (11).
+
ANALYTICAL CHEMISTRY, VOL. 62, NO. 15, AUGUST 1, 1990
2
100
?-
>
t In
i
1
z w
1
100,
3
1699
I
I
I F1
I12
I Ill i
c
z w
-5 !-
*4
0
5
10
TIME (rnin.)
0 TIME (min.) Figure 4. Reconstructed ion current chromatogram in PI-mode TSP LClMS of (1)ronnel, (2) chlorpyrifos,and (3) parathionethyl: stationary phase, 7 pm Nucleosii silica material; mobile phase, n-hexane-dichloromethane (5050);flow rate, 0.7 mL/min; amount injected, 500
ng. Sensitivity a n d Chromatographic Analysis. Table I1 summarizes the detection limits of the three different groups of analytes in the P I and NI mode obtained by use of normal-phase LC eluents discussed in this paper compared with the detection limits in the PI and NI mode obtained by using reversed-phase eluents reported in the literature. The P I mode T S P LC/MS chromatogram of a mixture of the pesticides ronnel, chlorpyrifos, and parathion-ethyl obtained by using a silica column and n-hexane-dichloromethane (50:50) as eluent-at 0.7 mL/min because of the limited MS pumping capacity-is shown in Figure 4. For the organophosphorus pesticides, when using a normal-phase eluent, the P I mode of operation yields a detection limit that is 10 times lower than that obtained in the NI mode. The detection limit for parathion-ethyl, in the PI mode, under full scan conditions was calculated to be 5 ng. Determination of detection limit was carried on by injecting different amounts (n = 6), 500,50, and 5 ng, of parathion-ethyl, thus giving in each case abundances of the [M + H]+ in the spectra of parathion-ethyl of 80000, 7500, and 800 arbitrary units, respectively. Considering that the noise level in the TSP spectrum is 150 arbitrary units, and that 5 ng gave a signal of 800 arbitrary units with 25% relative standard deviation, a variation in signal-to-nose ratio between 4 and 6 is obtained. This 5-ng detection limit is 1 order of magnitude better than that compared with reversed-phase eluents (4). In the NI mode there is only a 2-fold increase in sensitivity when a normal-phase eluent is used instead of a reversed-phase eluent. For the chlorophenols and the chlorinated phenoxy acids, the NI mode of operation was a 2-fold more sensitive than the PI mode. In all cases the detection limits for the NI mode of operation were between 1 and 10 ng under full scan conditions, which is of the same order as previously reported using reversed-phase eluents in TSP LC/MS (3,11,12). The detection limits for 2,4,5-trichlorophenol and 2,4-D as model compounds have been calculated in a similar way as for parathion-ethyl, e.g., considering always the base peak of the spectrum. For 2,4-dichlorophenol, 2,4,5-trichlorophenol, and the chlorinated phenoxy acids, a clear gain in sensitivity is obtained in the PI mode when normal-phase eluents were used instead of reversed-phase eluents. In the latter case even 200 ng could not be detected (3, 12). In the NI mode there were no differences between the detection limits with normal-phase or reversed-phase eluents.
-
Figure 5. Reconstructed ion chromatogram obtained with N I TSP LC/MS of a water sample from Barcelona Harbor spiked with 0.1 ppm of (1) 2,4-D, (2) 2,4,5-T, and (3) Silvex: stationary phase, 5 pm Lichrosorb RP-18; mobile phase, acetonitrile-water (5050)+ 0.1 M phosphate buffer (pH = 2.5); flow rate, 1 mL/min; extraction solvent, cyclohexane-dichloromethane-1-butanol (4545: 10) at 1 mL/min; flow rate into the mass spectrometer, 0.8 mL/min.
On-Line Postcolumn Extraction w i t h TSP LC/MS. The possibility to use nonpolar solvents in TSP LC/MS has been applied to the postcolumn extraction of the chlorinated phenoxy acids from an aqueous eluent following the ionsuppressed reversed-phase separation. Using acetonitrileaqueous 0.1 M phosphate (pH = 2.5) buffer (50:50) a t 1 mL/min, and a postcolumn extraction mixture of dichloromethane-cyclohexane-1-butanol (45:45:10) as mobile phase at 1mL/min, the extraction efficiencies for 2,4-D, 2,4,5-T, and Silvex were loo%, 70%, and 66%, respectively. Because only volatile buffers such as ammonium acetate or ammonium formate have been used in T S P LC-MS so far (1-3), the use of an LC eluent containing a nonvolatile acidic buffer is a novelty. This buffer suppresses the ionization of the three chlorinated phenoxy acids. After the chromatographic separation is completed, it can be coupled on-line with extraction into a nonpolar solvent mixture and introduced into the TSP LC-MS system. Figure 5 shows the chromatogram, in full scan mode, obtained for a water sample from the Barcelona Harbor spiked with 0.1 ppm of 2,4-D, 2,4,5-T, and Silvex, which has been analyzed with reversed-phase LC in combination with the postcolumn extraction system and with TSP detection in the NI mode. These results clearly demonstrate the usefulness of this system for LC analyses where nonvolatile buffers are needed. Probably for the analysis of a complex sample, such a postcolumn extraction can also increase the selectivity of the total analytical system. CONCLUSIONS n-Hexane, cyclohexane, and dichloromethane, which are well-known constituents of normal-phase LC eluents, are useful reagent gases in T S P LC/MS for the determination of organophosphorus pesticides, chlorophenols, and chlorinated phenoxy acids. The results obtained here clearly demonstrate the increase in sensitivity of 1order of magnitude obtained for the three groups of test compounds in P I mode T S P LC/MS using a normal-phase eluent instead of a reversed-phase eluent. Complementary structural information is obtained by using both the PI and NI mode of operation. The fragmentation patterns have been compared with previous results obtained by using reversed-phase eluents in TSP LC/MS. The high degree of chloride attachment when using dichloromethane as (part of) the eluent is noteworthy. The use of mobile phases with nonvolatile buffers in combination with a postcolumn extraction module, containing a sandwich phase separator, for extraction by nonpolar solvents
1700
Anal. Chem. 1990, 62, 1700-1705
in combination with TSP LC /MS is reported here for the fiist time. This work is a sequel to previous research of this group in DLI using a membrane phase separator (13)and of Vouros and co-workers (22) who used a gravity-based separator in combination with a transport interface prior to MS detection. Future work will include the investigation of postcolumn ion-pair extraction prior to TSP MS detection.
ACKNOWLEDGMENT The authors are indebted to the late Professor R. W. Frei for initiating this collaborative research project. R. Alonso from the CID-CSIC is thanked for laboratory assistance. LITERATURE CITED (1) Voyksner, R . D.; Haney, C. A. Anal. Chem. 1985, 5 7 , 991-996. (2) Garteir, D. A,; Vestal, M. L. LC, Liq. Chromatogr. HPLC Mag. 1985, 3,334-346. (3) Barcelb, D. Org. Mass Spectrom. 1989, 2 4 , 219-224. (4) BarcelB, D. Biomed. Environ. Mass Spectrom. 1988, 17, 363-369. (5) Barcelb, 0.; Aibaiggs, J. J. Chromatogr. 1989, 474, 163-173. (6) Covey, T. R.; Bruins, A. P.; Henion. J. D. Org. Mass Specfrom. 1988, 23, 178-186. (7) Smith, R. W.; Parker, C. E.; Johnson, D. M.;Bursey, M. M. J. Chromafogr. 1987, 394, 261-270. (8) Alexander, A. J.; Kebarle, P. Anal. Chem. 1986, 58, 471-478. (9) Barcelb, D. Org. Mass Specifom. 1989, 2 4 , 898-902. (10) Parker, C.E.; Smith, R. W.; Gaskell, S. J.; Bursey, M. M. Anal. Chem. 1986, 58, 1661-1664. (11) Vreeken, R . J.; Brinkman, U. A. Th.; De Jong, G. J.; Barcelb, D. Biomed. Environ . Mass Spectrom., in press.
(12) Barcelb, D. Chromafographia 1988, 25, 295-299. (13) Apffel, J. A.; Brinkman. U. A. Th.; Frel, R. W. J. Chromafogr. 1984, 312, 153-164. (14) Geerdink, R . B.; Maris, F. A.; Frei, R. W.: De Jong, G. J.; Brinkman, U. A. Th. J. Chromatogr. 1987, 394, 51-64. (15) Parker, C. E.; Haney, C. A.; Harvan, D. J.; Hass, J. R. J. Chromatogr. 1982, 242, 77-96. (16) Barcelb, D.; Maris, F. A.; Geerdink, R. B.; De Jong, G. J.; Brinkman, U. A. Th. J . ChrOmatOgr. 1987, 394, 65-76. (17) De Ruiter, C.; Wolf, J. H.; Brinkman. U. A. Th.: Frei, R. W. Anal. Chim. Acta 1987, 192, 267-275. (18) De Ruiter, C. Thesis, Pre- and post column fluorescence derivatization in HPLC, Free University, Amsterdam, The Netherlands, 1989; pp 75-88. (19) Schill. G.;Modln, R.;Borg, K. 0.; Persson, B. A. I n Handbook of Derivatives for Chromatography; Blau, K., King, G. s., Eds.; Heyden: London, UK, 1977; Chapter 14. (20) Voyksner, R . D.; Bursey, J. T., Pelllzari, E. D. J. Cbromafogr.1984, 312,221-235. (21) Stan, H. J.; Kellner, G. Biomed. Mass Specfrom. 1982, 9 , 463-492. (22) Vouros, P.; Lankmayr, E. P. Hayes, M. J.; Karger, B. L.: McGuire. J. M. J. Chromatogr. 1982, 251, 175-188.
RECEIVED for review December 11,1989. Accepted March 23, 1990. Financial support was provided by NATO Research Grant 0059f 88, the Ministerio de Educacion y Ciencia, the Netherlands Foundation for Chemical Research (SON) with financial acid from the Netherlands Organization for the Advancement of Scientific Research (NWO) (Grant No. 700.344.006),and the Commission of the European Communities fellowship (ST2*-0488).
Effects of Peptide Hydrophobicity and Charge State on Molecular Ion Yields in Plasma Desorption Mass Spectrometry Rong Wang, Ling Chen, and Robert J. Cotter* Middle Atlantic Mass Spectrometry Facility, Department of Pharmacology and Molecular Sciences, The Johns Hopkins University School of Medicine, Baltimore, Maryland 21205
Plasma desorption mass spectra were obtained for a series of peptides, grouped in four mass ranges having approximately 9, 20, 30, and 40 amino acid residues. Within each group, the individual peptides differed in hydrophobicity, charge state, and retention time, as measured on a revenred-phasg HPLC column. Comparison of the molecular ion intensities in the positive ion mass spectra of peptides from each group showed a strong dependence upon hydrophobicity and no correlation with charge state. Plasma desorption mass spectra of mixtures of ail the peptides within each mass range generally resulted in the desorption of a single residue and suppression of the ion slgnai from other components. I n most cases, this could be correlated with hydrophobicity, as calculated from the Bull and Breese index; however, a better correlation existed when the results were compared with reversed-phase retention times. I n general the spectra of mixtures were not Influenced by charge state (except in the absence of hydrophobic peptides), as the same component in each peptide mixture produced the most abundant ions in both positive and negative ion spectra.
INTRODUCTION An effective strategy for the analysis of peptides and proteins with known, or expected, amino acid sequences is the
direct mass spectrometric analysis of their unfractionated enzymatic digests using fast atom bombardment ( I , 2) or plasma desorption mass spectrometry (3). These survey spectra provide a series of peaks whose masses can be correlated directly with those calculated from the sequence for peptide fragments resulting from enzymatic cleavage at specific locations on the protein. The masses observed in such spectra can be and have been used to locate posttranslational and chemical modifications, to reveal oxidation of methionine to the sulfoxide ( 4 ) ,to verify the formation and/or location of disulfide bonds (5,6),and to determine sequence variations (7). FAB mass spectral analysis of complex mixtures often results in preferential desorption of the more surface-active (8) or hydrophobic (9) components. Naylor et al. (9) have shown that the selective desorption of peptide fragments in a tryptic digest can be correlated with their average hydrophobicities calculated from the Bull and Breese indexes (10) and noted that the presence of hydrophobic fragments in a tryptic digest effectively suppresses the desorption of the more hydrophilic residues. While the selectivity observed in the FAI3 technique is related to diffusion of sample to the surface of the liquid matrix, Nielsen and Roepstorff (11) have proposed that the PDMS analysis of tryptic digests (desorbed from the solid state on a nitrocellulose-coated foil) is relatively insensitive to hydrophobicity and found no correlation between the selective desorption of peptides in an enzymatic digest with their
0003-2700/90/0362-1700$02.50/00 1990 American Chemical Society
