Anal. Chem. 1989* 61 775-777 775

peaks in the F2 axis are suppressed, but in the F1 axis the correlation between the Hd protons at 1.05 ppm together with the set of He and H,‘ protons at 0.32, 0.50, and 0.59 ppm of the oxazoline compound is clearly seen. This example does not show a fully optimized correlation but it indicates that it is possible to perform two-dimensional assignment tech- niques in on-line HPLC/NMR coupling.

Further optimization of these techniques will enhance the attractiveness of HPLC/NMR coupling in the field of hy- phenated methods.

ACKNOWLEDGMENT The authors gratefully acknowledge the help of Peter

Dvertsak in recording the COSY 45 experiment. Registry No. 2-[ (Dicyclopropylmethyl)amino]-A2-oxazoline

dihydrogenophosphate, 85409-38-7; dicyclopropyl ketone, 1121- 37-5.

LITERATURE CITED (1) Watanabe, N.; Niki, E. Proc. Jpn. Acad., Ser. 8 1978, 54, 194. (2) Bayer, E.; Albert, K.; Nieder, M.; Grom, E.; Keller, T. J. Chromafogr.

1979, 186, 497. (3) Bayer, E.; Albert, K.; Nieder, M.; Grom, E.; Zhu An fresenlus’ 2.

Anal. Chem. 1980, 304, 111. (4) Bayer, E.; Albert, K.; Nieder, M.; Grom, E.; Wolff, G.; Rindiisbacher, M.

Anal. Chem. 1982, 54, 1747. (5) Dorn, H. C. Anal. Chem. 1984, 56, 747A. (6) Albert, K.; Nieder, M.; Bayer, E.; Spraul, M. J. Chromafogr. 1985,

346, 17. (7) Laude, D. A., Jr.; Lee, R. W. K.; Wilkins, C. L. Anal. Chem. 1985, 57,

1464.

(6) Laude, D. A., Jr.; Wilkins, C. L. Anal. Chem. 1987, 59, 546. (9) Allen, L. A.; Glass, T. E.; Dorn, H. C. Anal. Chem. 1988. 60, 390.

(10) Albert, K.; Kunst, M.; Bayer, E.; Spraul, M.; Bermel, W. J. Chroma- togr., in press.

(11) Hore, P. J. J. Magn. Reson. 1983. 55, 283. (12) Bauer, C.; Freeman, R.; Frenkiel, T.; Keeler, J.; Shaka, A. J. Magn.

Reson. 19849 58, 442. (13) Kessler, H.; Oschkinat, H.; Griesinger, C.; Bermel, W. J. Magn. Re-

son. 1988, 70, 106.

Klaus Albert Manuel Kunst Ernst Bayer*

Institut fur Organische Chemie Auf der Morgenstelle 18 D-7400 Tubingen, FRG

Hendrik Jan de Jong Patrick Genissel

Institut De Recherches Internationales Servier 27, rue du Pont 92202 Neuilly-sur-Seine Cedex, France

Manfred Spraul Wolfgang Bermel

Bruker Analytische Messtechnik GmbH Silberstreifen D-7512 Rheinstetten-Fo, FRG

RECEIVED for review June 16, 1988. Resubmitted December 6, 1988. Accepted December 14, 1988.

Supercritical Fluid Injection of Polysiloxane Solutions for Calibration and Tuning in Supercritical Fluid Chromatography/Mass Spectrometry

Sir: Our laboratory has been involved in the development and application of supercritical fluid chromatography/mass spectrometry (SFC/MS) over the past 3 years. We have been especially interested in the use of “high mass” quadrupole instruments (i.e., of mass range greater than 2000 Da) during the past year (1). For tuning, for calibration, and especially for comparing the performance of various instruments in the SFC/MS mode, we needed a method that would provide a steady ion current over a wide mass range a t high mass in an ionization mode similar to that of SFC/MS.

A possible solution was the direct injection into the ion source of an appropriate solution a t or near supercritical conditions, or supercritical fluid injection/mass spectrometry (SFI/MS). It has been used for other purposes (2, 3) , pri- marily in combination with supercritical fluid extraction ( 4 ) for solutes of molecular weight below 1000. Poly(dimethy1- siloxanes) have been proposed as reference standards for exact-mass measurement in chemical ionization mass spec- trometry (5). Indeed, SFI/MS of solutions of polysiloxanes provides a steady ion current over a wide mass range, fulfilling our initial needs. Here we discuss the development of SFI/MS for tuning, mass calibration, and instrument performance evaluation prior to SFC/MS. Recent results obtained by using a particularly well-suited fluorinated polysiloxane are pres- ented.

EXPERIMENTAL SECTION A 3% solution (v/v) of poly(methyl-3,3,3-trifluoropropyl-

siloxane) (structure I, 300 cs, Petrarch Systems, Inc., Bristol, PA) in SFC-Grade C 0 2 with 5% (w/w) 2-propanol (Scott Specialty Gases, Plumsteadville, PA) was prepared for SFI/MS. The po- lysiloxane was added to a l-L, double-ended, stainless steel sample cylinder (304L-HDF4-1000, Nupro, Willoughby, OH). One end

FF3 ,CH2

CH3 -Si-0-[ Si- 01 - Si-CH3 I I n~ CH, CH, CH3

T I of the cylinder was fitted with a 0-20.7 MPa (0-3000 psi) pressure gauge (Matheson, East Rutherford, NJ). The other was fitted with a shut-off valve which incorporates a safety rupture disk for overpressure protection (SS-16DKMCF4-A-1, Whitey, Highland Heights, OH). A rupture disk burst pressure range of 13.1-13.7 MPa (1900-2OOO psi) was chosen. The sample cylinder was then connected to a tank of SFC-grade COz with 5% 2-propanol equipped with a dip tube, flushed to remove air in the sample cylinder, and filled. No cooling of the sample cylinder was necessary. The mass of C02/2-propanol mixture transferred was determined by weighing the sample cylinder before and after filling.

After agitation, the approximate solution concentration of polysiloxane in C02/2-propanol was checked in the following manner. First, a 5-mL container was filled with the mixture. Second, the C02/2-propanol was slowly vented from the container. Third, the polysiloxane left in the container was washed out with methylene chloride, dried under a stream of dry nitrogen, and weighed. A reducing union and 1.6 mm (0.0625 in.) 0.d. stainless steel tubing were used to link the shut-off valve on the sample cylinder to an in-line filter (ZUFR1, Valco Instruments Co., Houston, TX) with a 2-pm filter screen (2SR2, Valco). Similar tubing was used to connect the filter to the heated SFC/MS interface probe.

0003-2700/89/0361-0775$01.50/0 0 1989 American Chemical Society

776 ANALYTICAL CHEMISTRY, VOL. 61, NO. 7, APRIL 1, 1989

.% 60 1 c

Flgure 1. Ammonia C I SFIIMS spectrum of 3 % poiy(methyl-3,3,3- trifiuoropropylsiloxane) in COP modified with 5 % 2-propanol.

The interface probe used in this work differs from the previous description ( I ) only in the design of the probe tip. The tip was machined to mate with the ion source of a Finnigan TSQ-70 triple-quadrupole mass spectrometer (Finnigan-MAT, San Jose, CA). The probe stem was held at 100 "C while the tip was maintained at 300-350 "C. The heart of the probe is a robot- pulled, fused-silica-capillary flow restrictor. The restrictor used in this work was typical of those commonly used in our laboratory for SFC and was prepared as previously described (6). It tapered from the 25 pm i.d. to an aperture of approximately 10 pm over 3 cm. The restrictor aperture was positioned just inside the probe tip.

The interface probe was inserted through the direct probe inlet until the probe tip butted with the ion source. The cylinder valve was then opened and the solution was bled continuously into the ion source while tuning and evaluating the instrument's per- formance.

The mass spectrometer was an unmodified Finnigan TSQ-70 triple quadrupole. The first and third quadrupoles (Q1 and Q3, respectively) were operated in the "high mass range mode" (m/ z 10-4000) with Q1 and Q2 transmitting all ions. Q3 was used as the mass filter. Automatic tuning software was used to optimize all lens voltages at clusters throughout the mass range. Resolution and Q3 offset were adjusted m a n d y via the instrument trackball to provide optimal signal and resolution for each cluster. Mass calibration was also performed manually. The ion source tem- perature was held at 240 "C. A chemical ionization (CI) ion volume was used. The instrument was operated in the NH3 CI mode with 99.99% anhydrous NH3 (AGA Specialty Gas, Maumee, OH). Electron energy was 88 eV and emission current was 200 MA. Only positive ions were investigated. The electron multiplier was operated at -1300 V with the conversion dynode at -4800 V. The electrometer gain was set at lo-' A/V.

RESULTS AND DISCUSSION Figure 1 is a typical spectrum acquired during SFI/MS of

the poly(methyl-3,3,3-trifluoropropylsiloxane). This spectrum was collected during a single, 10-s scan from 1500 to 4000 Da. Ammonia CI yields a simple spectrum. The major ions are ammonium adducts of the polysiloxane oligomers. A second series of protonated oligomers is also clearly visible. These results are in keeping with the previously reported behavior of polysiloxanes under ammonia CI conditions ( I ) . The spectrometer was calibrated at high mass by using the cal- culated masses of the ammonium adducts. This particular sample of the polysiloxane (300 cs) is well suited for studying instrument performance as it clearly covers the mass range of interest.

We have used traditional CI tuning and calibration methods for SFC/MS at lower mass. Fast atom bombardment (FAB) desorption-ionization of a solution of CsI in glycerol may have been adequate for mass-axis calibration for our "high mass" SFC/MS work, yet the TSQ-70 used for SFC/MS is not equipped for FAB. We tried introducing poly(methy1-3,3,3- trifluoropropylsiloxane) and other polysiloxane mixtures to the CI source by heated direct insertion probe, but this was not satisfactory. The spectrum produced never extended

3450 3454 3458 3462 3466 3470

m/z Flgure 2. Observed and calculated isotope cluster for the ammonlum adduct of the n = 21 oligomer.

above the adduct of oligomer number 17, changed with time, and was not as long-lived as the spectrum produced by SFI/MS. Moreover, we were particularly interested in com- paring the performance of various instruments a t high mass in an ionization mode similar to that encountered in SFC/MS.

In earlier experiments (I) we used poly(dimethylsi1oxane) in unmodified COZ as a tuning mix and as a means of studying instrument performance. The sample cylinder was often cooled in ice water or dry ice while filling with C02 since the particular C02 supply tank used at the time had no "dip tube". This mixture proved unsatisfactory since the poly(di- methylsiloxane) began to precipitate as the mixture warmed to room temperature. The spacing of the major ions in the spectrum of this mixture was also closer than desired (74 Da) for a tuning mix. These problems have been alleviated by using the fluorinated polysiloxane and C02 modified with 5% 2-propanol. The fluorinated polymer was used to increase the spacing between the major ions in the spectrum (156 Da). We expected the trifluoropropyl groups to impart some degree of polarity to the polysiloxane. Polar "modifiers" have been used to increase the solubility and selectivity of solutes in supercritical COz in both SFC (7) and supercritical fluid ex- traction (8). We therefore used the 2-propanol-modified COz in hopes of enhancing the solubility of the fluorinated poly- siloxane. Since using this mixture, we have not encountered the solubility problems described above. However, the solu- bility of fluorinated alkanes in supercritical C02 is known to be higher than that of alkanes (9)) and the Kamlet-Taft ?r*

solvent polarity values for supercritical COz are similar to those of the perfluoroalkanes and freons (IO). Thus the fluorinated polysiloxane may inherently be more soluble in C02 than the poly(dimethylsi1oxane).

Figure 2 is a visual comparison of measured vs calculated isotope clusters centered about m/z 3460. This is the am- monium adduct of the oligomer with an "n" value of 21. The measured spectrum was acquired in the "profile" mode and is an average of four scans from m/z 3450 to m/z 3470 in 1 s. The mass resolution (m/ Am where Am is full width at half peak height) estimated for m/z 3460 in the spectrum is 2.7 X lo3. Higher values of resolving power could be obtained, but only at the expense of signal strength. This spectrum was a good compromise, with the valleys between the major peaks being at roughly 70% of the peak height. The spectrum in Figure 1 was acquired by using these settings. Table I com- pares the calculated and measured (simple peak heights) relative intensities of the major peaks in the cluster. Most

Anal. Chem. 1989, 61, 777-779 777

Table I. Calculated and Measured’ Relative Intensities for the n = 21 Ammonium Adduct Isotope Multiplet

relative intensity mlz calcd, % measd; % re1 deviation, %

3457 3458 3459 3460 3461 3462 3463 3464 3465 3466

28.4 63.0 92.1

100.0 89.5 68.4 46.1 27.9 15.4 7.8

35.5 68.6 92.4

100.0 89.5 70.3 47.1 30.2 15.7 7.6

22.2 8.5 0.0

0.0 2.7 2.1 7.9 1.9 2.6

From peak heights.

of the calculated and observed relative intensities deviate from one another by less than 3 % . The median deviation is 2.6 % . The peaks tail to the low mass side, with peak height mea- surements becoming less accurate in this direction. We ob- tained these reasonable isotope ratios after a minimum amount of tuning.

In the work described here a simple sample cylinder was used to contain the polysiloxane solution instead of a pump. The pressure was fixed at roughly 5.9 MF’a (58 atm) (measured with the uncalibrated gauge described in the Experimental Section), the vapor pressure of the C02 with 5% 2-propanol at room temperature. I t might be argued that the method described is in reality subcritical fluid injection since the critical pressure of the mix is not exceeded. However, the critical pressure is defined (and only has physical significance) at the critical temperature (11). At a temperature only slightly above the critical temperature, the pressure vs density iso- therm shows a dramatic change in density (and solvating strength) about the critical point. This discontinuity in the pressuredensity isotherm becomes less dramatic as the tem- perature of the system increases. The isotherm is nearly linear in this region at a reduced temperature of three for pure COz, for example. At the point of injection into the mass spec- trometer ion source, the polysiloxane solution is clearly far above its critical temperature. Whether the pressure is slightly above or slightly below the “critical” pressure at this tem- perature has little effect on the solvating power of the carrier, on the flow rate of modified COz through the restrictor, and on the pressure within the ion source region. This is borne out by our observations of the ion source pressure during SFI/MS with our sample cylinder and during normal pres- sure-programmed SFC/MS (1). Thus we feel justified in our

use of this simple method to approximate conditions en- countered in SFC/MS and in our use of the term “supercritical” in SFI/MS.

During these early trials, SFI/MS of polysiloxanes dirtied the ion source and analyzer over a 1-day period. In more recent experiments, less concentrated solutions (0.5%) pro- duced strong, stable ion currents over a 5-day period without rapid deterioration in ion source and analyzer conditions.

These results indicate that SFI/MS of polysiloxane solu- tions is a reasonable tuning and calibration tool for SFC/MS. It is also useful in studying instrument performance over a wide mass range. Figure 1 provides a glimpse at the potential power of SFI/MS for quickly characterizing oligomeric mix- tures, especially those that are difficult to analyze by de- sorption/ionization techniques such as fast atom bombard- ment MS or plasma desorption MS.

LITERATURE CITED (1) Pinkston, J. D.; Owens, G. D.; Burkes, L. J.; Delaney, T. E.; Mllllngton,

D. S.; Maltby, D. A. Anal. Chem. 1988, 60, 962-966. (2) Sin, C. H.; Pang, H. M.; Lubman, D. M.; Zorn, J. Anal. Chem. 1986,

58, 487-490. (3) Smith, R. D.; Udseth, H. R. Anal. Chem. 1983, 55, 2266-2272. (4) Smith, R. D.; Udseth, H. R.; Hazlett, R. N. Fuel 1985, 64, 810-815. (5) Bertrand. M. J.; MaRals. L.; Evans, M. J. Anal. Chem. 1987, 59,

194-197. (6) Chester, T. L.; Innis, D. P.; Owens, G. D. Anal. Chem. 1985, 57,

2243-2247. (7) Wright, B. W.; Smith, R. D. J . Chromatogr. 1986, 355, 367-373. (8) Hacker, D. S. in Supercr/f/ca/ Flu/& - Chemical and €ng/nesr/ng Prin-

ciples and Applications; Squires, T. G., Paulaitis, M. E., Eds.; ACS Symposium Series 329; American Chemical Society: Washington, DC, 1987; pp 213-228.

(9) Brady, J. E., University of Pittsburgh, personal communication, 1988. (IO) Yonker, C. R.; Frye, S. L.; Kalkwarf, D. R.; Smith, R. D. J . Phys.

Chem. 1988, 90, 3022-3026. (1 1) CRC Handbook of Chemistry and Physics, 57th ed.; CRC Press:

Cleveland, OH 1976; p F-98.

To whom correspondence should be addressed.

J. David Pinkston* Grover D. Owens

The Procter & Gamble Company Miami Valley Laboratories P.O. Box 398707 Cincinnati, Ohio 45239-8707

The Finnigan-MAT Institute 4450 Carver Woods Drive Cincinnati, Ohio 45242

Ernest J. Petit

RECEIVED for review October 4,1988. Accepted December 27, 1988.

Supersonic Jet Spectroscopy of Nonvolatiles f rorn Pulsed High Pressure Ammonia Expansions

Sir: The sharp spectral features available by using su- personic jet expansions in conjunction with laser spectroscopic methods have been demonstrated as a means of obtaining enhanced selectivity in chemical analysis ( I ) . In the past, this technique has generally been limited to molecules which can be easily volatilized into the gas phase for expansion with a light carrier gas. More recently there have been a number of methods developed for volatilizing nonvolatile and ther- mally labile molecules into supersonic jets (2-18). These

0003-2700/89/036 1-077780 1.50/0

methods have the potential for extending the selectivity of jet spectroscopy to a broad class of molecules of biological and pharmaceutical importance.

Direct heating methods have been used for some time for volatilization of relatively stable large polynuclear aromatic hydrocarbons (PNAHs) with high melting points (>200 OC) into supersonic jet expansions (2-4). In order to study more labile species several groups have developed specialized heating methods in order to minimize thermal decomposition (5,6).

0 1989 American Chemical Society