mass spectrometer interface

In the mid and low mass ranges, similar resolution checks were developed using ions containing a single 13C ion at masses 199 and 70. In each instance...
7 downloads 0 Views 497KB Size
In the mid and low mass ranges, similar resolution checks were developed using ions containing a single 13C ion a t masses 199 and 70. In each instance, the ions are very likely assigned the correct composition and the theoretical percentages may be compared with the experimental. At mass 199, nine measurements gave an average of 8.0% and u = 2.3. The criterion suggested is 7 f 2% which compares with the theoretical value of 6.6%. At mass 70, the theoretical value is 1.1%, but most of the nine measurements gave near zero values for this ion. Perhaps this was caused by very slight changes in the base-line (threshold) adjustments. I t is very difficult to make accurate and precise measurements of relatively non-abundant ions when observing very small amounts (-20 ng) in fast (3-4 sec) spectrometer scans. Therefore, for mass 70, we suggest a nominal criterion of less than 2% of mass 69. This is mainly a check on excessive broadness,or poor peak shape in the low mass region for those data systems that interpret broadness as ion abundance. Because of the probable compositions of the mass 198 (CsFsP+) and mass 69 (CFs+) ions, it is unlikely that mass 197 and mass 68 ions would be present. Indeed repeated measurements have shown that they are not present. Therefore, we suggest that mass 197 should be less than 1% of the base peak, and mass 68 less than 2% of mass 69. Both criteria are checks on excessive broadness and skew as discussed above.

CONCLUSION The set of relative abundance ranges proposed for D F T P P has been very useful in evaluating the performance

of a number of GC/MS systems. These ranges are the basis for the proposed standard ion abundance calibration and provide a reasonable basis for comparing the output from the wide variety of systems in use.

ACKNOWLEDGMENT We express our sincere appreciation to the individuals and laboratories that participated in the interlaboratory study. These included J. Peterson, Fish and Wildlife Service; E. M. Chait, E. I. DuPont de Nemours & Company; J. C. Cook, University of Illinois; J. B. Knight, Finnigan Corporation; and F. Biros, J. Blazevich, H. Boyle, M. Carter, P. Clifford, F. Farrell, G. Muth, H. Rodriguez, D. C. Shew, and A. Wilson of the Environmental Protection Agency.

LITERATURE CITED (1) S. S. Dua. R. C. Edmondson, and H. Gilman. J. Organometal. Chem., 24, 703 (1970). (2) J. W. Eichelberger. L. E. Harris, and W. L. Budde. to be published; presented at the 22nd Annual conference on Mass Spectrometry and Allied Topics, Philadelphia, PA, May 19-24, 1974. (3) American Society for Testing and Materials Committee D-2, 21st Annual Conference on Mass Spectrometry and Allied Topics, San Francisco, CA. May 20-25, 1973. (4) 8.S.Middleditch, Anal. Cbem., 41, 2092 (1969). (5)R. H. Wallick, G. L. Peele, and J. B. Hynes. Anal. Cbem., 41, 388 (1969). (6) D. M. Schoengold and W. H. Stewart, Anal. Cbem., 44, 884 (1972).

RECEIVEDfor review October 29, 1974. Accepted January 24, 1975.

A Liquid Chromatograph/Mass Spectrometer Interface Patrick R. Jones’ and Shen K. Yang Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasadena, CA 9 1125

A silicon rubber membrane molecular separator has been used for the coupling of a mass spectrometer to a liquid chromatograph. The three-stage membrane separator allows liquid flow rates in excess of 2 ml/mln while maintalning Torr In the mass spectrometer when polar solvents are used. The dimethylsiloxane polymer membrane transmits nonpolar molecules and rejects polar solvents with high efficiency. Nanogram levels of suitably volatile compounds may be detected In the liquid chromatograph effluent.

Liquid chromatography has become increasingly important for the separation and detection of polyfunctional and thermally sensitive compounds. In many cases, the limitations on the LC technique are related to finding a suitable detector. The common detectors of highest sensitivities are the light absorption, and the fluorescence detectors ( I ) . These, however, are far from universal in their application. Mass spectrometry affords a more nearly universal detector, providing an appropriate interface is available. Present address, Department o f Chemistry, U n i v e r s i t y of t h e Pacific, Stockton, C A 95204.

1000

ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975

Numerous GC/MS coupling systems have been described (2-5) and the characteristics of various separators have been summarized by Milne (6). There are only a few LC/MS systems. Lovins has used a semiautomatic, mechanical, solid-probe inlet for injection of portions of the LC effluent that have been adsorbed on a carrier support ( 7 ) . A chemi-ionization scheme using a split stream and the solvent from the LC as the CI gas has been described by McLafferty (8, 9). A third system using an atomospheric pressure ionizer and a 25-micron hole to sample the ionization chamber by MS has been discussed by Horning as a viable interface (10, 11). In both of the latter systems, the solute/solvent ratio is unchanged. A separator that would efficiently reject the solvent would be highly desirable in terms of maintaining a low pressure in the MS. Furthermore, it is desirable to enrich the sample in the fluid mixture that is mass analyzed by selectively rejecting solvent molecules. The silicon rubber membrane separator designed by Llewellyn and Littlejohn (12, 13) for use in gas chromatography has been successfully used as an LC/MS interface. The operation of the interface is described in this paper. For LC/MS systems, a molecular separator that discriminates against the solvent is very important, and the membrane separator is, therefore, the interface of choice in

SAMPLE

k

1

1

7

FLASH

EVAPORATOR

'

1 I50

I

I

200

250

MEMBRANE TEMPERATURE, "C

Flgure 2. Data for the transmission (in arbitrary units) of sample through the membrane interface to the mass spectrometer are given as a function of the membrane temperature for samples of 3methylcholanthrenein methanol QUADRUPOLE RECORDER

COMPUTER SPECTROMETER MAGNET IC

Figure 1. A schematic diagram of the components used in a liquid chromatograph/mass spectrometer system is shown in conjunction with the three-stage membrane molecular separator that provides the interface

terms of the enrichment factor of solute to solvent in the portion transmitted to the MS (6). The disadvantages of the silicone rubber membrane separator lie in the selectivity of the membrane against polar samples and in the limitation to a maximum working temperature of about 250 "C. For compounds that are sufficiently volatile a t 250 "C and that are relatively nonpolar, the silicone rubber membrane separator works quite well.

EXPERIMENTAL A Dupont Model 830 liquid chromatograph with a gradient elution accessory, and a 254-nm UV absorption detector was used without modification. A one-meter Permaphase ODs, analytical column was used for all compounds discussed. The LC effluent was conducted via a 0.025-cm i.d. stainless steel tube to the inlet of a flash evaporator, which consisted of a 15-cm section of 0.025-cm i.d. stainless steel tubing, heated by a 6-cm diameter aluminum block with four 75-watt cartridge heaters and a proportional heat controller (RFL Industries). The vaporized stream expanded into a 4-cm long section of 0.16-cm i.d. tubing, and passed over the first membrane surface. The vapors not transmitted by the membrane were pumped through a cold trap at reduced pressure (500 Torr) using a rotary vane vacuum pump (Gast Corp.). Each of the membranes was a circular disc with an area of 2.8 cm2 exposed to the incoming vapors. The vapor stream entered a t the central portion of a membrane and travelled radially through a gap of about 0.025 cm between the stainless steel support and the membrane surface before being pumped away as waste material. The transmitted portion of each membrane was directed toward the center of the next membrane, a t which point the above separation process was repeated a t lower pressure. A total of three stages was utilized. The required pressure differential across each membrane was adjusted such that there was a step gradient throughout the device. T h e actual performance of the separation was dependent not only on the pressure gradient but also on the quality of each individual membrane. Optimum compound transmission was obtained by trial and error, not by reproducing interstage pressures. The interstage pressures were regulated with precision metering valves (Whitey Corp.); and the pumping was via a mechanical oil pump (Welch) with a 13 X molecular sieve trap to protect against backstreaming of pump oil. An EA1 QUAD 300D quadrupole mass spectrometer was used for the mass analysis of the solutes transmitted by the membrane. T h e MS was controlled by an SCC 4700 computer, with a provision for simultaneous spectral display and storage of the data on magnetic tape. A dual filament, electron-impact ionizer was used in these experiments. The normal operating parameters of the MS were the following: ionization energy, 40 eV; ion energy, 10 eV; emission current, 100 wA; pressure to lo-; Torr. The membrane material used was 0.0254-mm thick polymeric dimethylsiloxane that is manufactured by General Electric Co.

RESULTS AND DISCUSSION The experimental arrangement shown schematically in Figure 1 was evaluated in terms of its applicability to the analysis of polycyclic aromatic hydrocarbons, and metabolites of such compounds. The membrane separator was first tested for the transmission of polycyclic aromatics from methanol solutions. The transmission of a given compound is a function of the membrane temperature, the interstage pressures, and the rate of flow of vapor over the top membrane. If the flow rate is too high, membrane contact with the vapor is inadequate, and compound transmission is low. The interstage pressures have to be adjusted for maximum vapor contact with the surface, and for a pressure differential across each membrane. Most of the solvent is rejected a t the first membrane surface and is collected in the trap shown in Figure 1. The interstage pressures between the first and second, and the second and third membranes are adjusted with precision metering valves which are located between the body of the separator and the interstage pumps. Figure 2 provides an evaluation of the quantity of a given compound that is transmitted to the MS a t various membrane device operating temperatures. The quantity of sample transmitted a t any given temperature is given in arbitrary units by the area under the ion current vs. time curve. The data shown in Figure 2 were obtained by repetitive injections of 50-ng samples of 3-methylcholanthrene (3-MC) in methanol a t a variety of separator temperatures. The molecular ion mle 268 of 3-MC was monitored. For 3-MC, the integrated ion current, shown as compound transmission in Figure 2, maximized a t approximately 200 "C. In many analytical applications, the major problems in MS analyses arise from the difficulty of observing the sample peaks above the background. In these cases, the quantity of sample transmitted may be less important than the rate of compound transmission. Figure 3 shows the intensity of the molecular ion of 3-MC as a function of time a t a series of membrane separator temperatures. Again, the data were obtained using repetitive injections of 50-ng samples of 3-MC in methanol and monitoring the molecular ion. The ion intensity was corrected by subtraction of background. Although the integrated ion current or total quantity of 3-MC transmitted to the MS is maximized a t 200 "C, the intensity of the molecular ion of 3-MC is seen in Figure 3 to increase as the separator temperature is increased to 250 "C. Higher temperatures do not increase the intensity of the M + ion current vs. time curves. These results indicate that the highest signal-to-noise ratio will be obtained a t temperatures for which the rate of transmission through the system is maximized, even though the absolute quantity transmitted may have diminished a t that temperature. As a general procedure, the separator temperature and ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE 1975

1001

$2 bl

I

I

1

1

20

40

60

00

I

TIME, sec

Flgure 3. The ion intensity of background corrected d e 268 is plotted as a function of time for a series of membrane device temperatures Each curve represents a 50-ng injection of 3-methylcholanthrene, with the membrane temperature shown adjacent the curve in degrees Centigrade

interstage pressures were adjusted for maximum signal-tonoise using a sipping technique. A solution of approximately 5 ng of a given compound per microliter of solvent was slowly drawn out (0.1 ml/min) of an open vessel and over the first membrane surface using the rotary vane pump. This procedure provided a constant stream of compound for separator and MS tuning and calibration. Subsequently, fine tuning could be accomplished using repetitive injections through the LC. Unlike the procedure used for drug analysis from body fluids utilizing a membrane separator MS system (14, 15), the use of a membrane device as a chromatograph interface requires careful attention to dead volume in the evaporator and in the first membrane regions. Dead volume was minimized to avoid remixing of chromatographically separated compounds. I t is also experimentally important to avoid cold spots in the connecting tubing to the mass spectrometer. For reasonably volatile compounds, little tailing of the ion current peak was found, even though the membrane separator has been criticized for this property ( 1 6 ) . Figure 4 compares the UV-absorption chromatogram of three polycyclic aromatic hydrocarbons with the ion chromatograms for the molecular ions of each component. The mixture consisted of 82 ng of anthracene (A), 46 ng of benzo[a]anthracene (BA), and 45 ng of benzo[a]pyrene (BP); and was separated by reverse phase chromatography using a one-meter Permaphase ODS column a t 50 "C with a pressure of 500 psi, and a water/methanol gradient which changed at a rate of 5% per minute. The LC flow rate was 0.5 ml per minute; and the separator was operated a t a constant temperature of 250 "C. The degree to which tailing occurs may be readily ascertained from the data in Figure 4. Repetitive injections of varying sample loads indicate that the limit for sample detection lies in the region of a few nanograms, although sub-nanogram quantities of 3-MC could be seen by single ion monitoring. As a general rule, however, using a water/methanol gradient a t 500 psi and a flow rate of 0.5 ml/min, the minimum usable sample was about 25 ng. That corresponds, for example, to an overall concentration of benzo[a]pyrene (BP) in the effluent of about 10-TM for a peak that elutes in one minute. Metabolic products of benzo[a]pyrene were obtained by in vitro incubation with rat liver microsomal enzymes (17). Hydroxylated metabolites were separated from each other by LC and were identified by mass fragmentography and 1002

ANALYTICAL CHEMISTRY, VOL. 47, NO. 7, JUNE

1975

s

1

2

3

4

5

6

TIME, min

Figure 4. A liquid chromatogram and three mass chromatograms are plotted on the same time scale Shown are the UV absorbances at 254 nm due to anthracene (A), benzo[a]anthracene (BA), and benzo[a]pyrene (BP), as well as the ion currents for the respective molecular ions: m/e 178, 228, and 252

by comparison with known synthetic compounds (17). However, BP-metabolites were not sufficiently volatile for adequate transmission through the membrane device, even a t 260 "C. Methoxy derivatives of the hydroxylated-BP compounds were formed by reacting the metabolites with trimethylanilinium hydroxide, "methelute" (Pierce Chemical Co.), in methanol. Direct injection of the derivatized metabolites into the membrane device allowed mass spectral analysis of the products. The derivatives, however, hydrolyzed rapidly in aqueous media. A tetrahydrofuran/methanol gradient elution scheme was attempted for the LC separation of the derivatized metabolites. Although tetrahydrofuran is a good solvent for use with the membrane device, the nonaqueous method did not give adequate chromatographic resolution. This result typifies the trade-off between LC separation without derivatization, and mass spectral identification that may require derivatization to obtain adequate volatility for higher molecular weight or more polar compounds. Any LC/MS interfacing device is, of course, limited in applicability by the necessity that the solutes in the effluent have an appreciable vapor pressure a t the operating temperature of the interface. Whereas the membrane separator is currently limited to about 250 "C by virtue of the thermal instability of the silicone rubber membrane, the device has the advantage of very efficiently protecting the MS from being overloaded with solvent vapors, a problem which mechanical/diffusion molecular separators have to overcome by splitting the effluent stream and, consequently, discarding much of the sample with the excess solvent. Thus, although other interfaces may operate a t higher temperatures, the membrane device has the capability of partially offsetting the temperature limitation by sampling the entire effluent stream. A polymeric material of greater thermal stability than dimethylsiloxane silicone rubber should improve the performance of the interface considerably. The use of a mass spectrometer as a detector for liquid chromatography has several advantages aside from the availability of structural chemical information. The analysis of metabolites of naphthalene illustrated the advantage that exists in the case of inadequate chromatographic resolution. Metabolites of naphthalene were obtained by in vitro incubation with rat liver microsomal enzymes ( 1 7 ) ,

but the LC water/methanol gradient technique did not separate the parent compound from the major metabolic product, hydroxy-naphthalene. Both compounds were, however, readily identified in the mass spectra corresponding to a single LC peak. Furthermore, an LC/MS combination obviates the need for the high purity, optically transparent solvents used with UV-detection systems. The use of the membrane device, however, limits the solvents to readily volatile, polar compounds. An LC/MS interface can be, therefore, of great utility for the identification of compounds in the LC effluent, even though the physical requirements of the MS limit the general applicability. Important considerations in the selection of an LC/MS interface include: 1)the volatility of the compounds of interest in relation to the operating temperature of the interface; 2) the solvent rejection of the device with respect to the LC flow rate and MS pumping speed; and 3) the minimum quantities of material available for detection. Insofar as volatility and polarity of a compound are concerned, the membrane device is limited to relatively volatile and relatively nonpolar compounds. This selectivity, however, is used to great advantage for the rejection of polar solvents. The consequently high enrichment factor in the transmitted vapors is the most important aspect of the membrane device as an LC/MS interface. When tuned for optimum transmission of one of the polycyclic aromatic hydrocarbons, a pressure of Torr could easily be maintained in the mass spectrometer with effluent flow rates of greater than 2 ml/min of water or methanol or tetrahydrofuran. Thus, in applications such as the analysis of hydrocarbons in aqueous systems, the analysis of volatiles in body fluids, or volatile solutes in other complex sample mixtures, the LC/membrane device/MS system has great promise as an analytical technique which requires a minimum of sample manipulation.

ACKNOWLEDGMENT The authors gratefully acknowledge helpful discussions with John D. Baldeschwieler, California Institute of Tech-

nology, Pasadena, CA. The willingness of James Arnold, Seth Abbot (Varian Associates), and Robert Wilcox (Dynaspec) to share their technical experience using the membrane device is also gratefully acknowledged. Jonathan Burke provided invaluable assistance with the computer control of the mass spectrometer.

LITERATURE CITED (1)P. R. Brown, "High Pressure Liquid Chromatogrgphy", Academic Press, New York, 1973. (2)G. A. Junk, lnt. J. Mass Spectrom. /on Pbys., 8, 1 (1972). (3)R. Ryhage and S. Wikstrom in "Mass Spectrometry", G. W. A. Milne, Ed., Wiley-lnterscience, New York. 1971,p 91. (4)D. I. Rees. Talanfa, 16, 903 (1969). (5)R . A. Flath. "Guide to Modern Methods of Instrumental Analysis", T. H. Gouw, Ed., Wiley-Interscience. New York, 1972,p 323. (6)G. W. A. Milne, "Mass Spectrometry". Wiley-lnterscience, New York, 1971,p 96. (7)R. E. Lovins, S. R. Ellis, G. D. Tolbert, and G. R. McKinney. Anal. Chem., 45, 1553 (1973). (8) P. Arpino, M. A. Baidwin, and F. W. McLafferty, Biomed. Mass Spectrom., 1, l(1974). (9)M. A. Baldwin and F. W. McLafferty, Org. Mass Spectrom.. 7, 1 1 1 1

11973). ~, (10)D. I. Carroll, I. Dzidic, R. N. Stillwell, M. G. Horning. and E. C. Horning, Anal. Chem., 46, 706 (1974). (11) E. C. Horning. M. G. Horning, D. I. Carroll, I. Dzidic. and R. N. Stillwell, Anal. Chem., 45, 936 (1973). (12)P. M. Llewellyn and D. P. Littlejohn, U.S. Patent 3,429,105, February 1969. (13)P. M. Llewellyn and D. P. Littlejohn, "Abstracts, 1 l t h Annual Conference I

~

of Analytical Chemistry and Applied Spectroscopy", Pittsburgh, Pa., February 1966. (14)U. Boerner and S. Abbott, Experientia, 29, 180 (1973). (15)U. Boerner. S. Abbott, J. C. Eidson. C. E. Becker, H. T. Horio, and K. Loeffler, Clin. Chim. Acta, 49, 445 (1973). (16)D. R. Black, R. A. Flath, and R. Teranishi, J. Chromatogr. Sci., 7, 284

(1969). (17)J. K. Selkirt, R. G. Croy, P. P. Roller, and H. V. Gelboin, Cancer Res., 34, 3474 (1974).

RECEIVEDfor review October 14, 1974. Accepted February 3, 1975. This work was supported by the National Science Foundation under Grant No. Gp-38855X-2 and the National Institutes of Health under Grant No. GM-21111-02. Contribution No. 4982 from the Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasedena, CA 91125.

Separation and Detection of Low Concentrations of Polythionates by High Speed Anion Exchange Liquid Chromatography Aaron W. Wolkoff and Richard H. Larose Analytical Methods Research Section, Canada Centre for Inland Waters, P.O.Box 5050, Burlington, Ontario, Canada L 7R 4A6

A method for the separation of thiosulfate and polythionates by high speed anion exchange chromatography is given. Detection of these anions at very low concentrations is possible using a cerium( IV)-fluorescence detection system. A technique for the determination of these sulfur-oxygen anions in mining wastewater and environmental samples is also presented. Detection limits of about 0.3 part per million are obtainable.

The decomposition of thiosulfate and polythionates

(S,ORZ-,n = 3-6), present in tailings pond effluents from the mining and smelting industry, is believed ( I , 2 ) to lead

to acid pollution of rivers surrounding these industries. I t has been found that the pH of the effluents varies from pH 9 to 6, but regardless of the effluent pH, acidic conditions ( p H 2.5 to 3.5) are produced in the receiving stream within a distance of thirteen miles. Obviously, this would create quite adverse conditions for the aquatic life in these streams. It was initially postulated (3) that the acidic conditions in the receiving streams were due to two factors: (a) the precipitation of calcium carbonate, lime being used in the mill for p H control, and (b) the oxidation of iron(I1) by the bacterium T. ferrooxidans. I t is now believed ( 2 ) that, since these bacteria can also utilize thiosulfate, thionates, ANALYTICAL CHEMISTRY, VOL. 47, NO. 7 , JUNE 1975

1003