RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 8, 105-110 (1994)

Capillary Column Supercritical Fluid Chromatography/Mass Spectrometry of Polycyclic Aromatic Compounds Using Atmospheric Pressure Chemical Ionization' Darren Thomas* and P. Greig Sim National Research Council Canada, Institute for Marine Biosciences, 1411 Oxford Street, Halifax, Nova Scotia, B3H 321, Canada

Frank M. Benoit Environmental Health Centre, Health and Welfare Canada, Tunney's Pasture, Ottawa, Ontario, KIA OL2, Canada

Capillary column supercritical fluid chromatography (SCF) was combined with atmospheric pressure chemical ionization (APCI) mass spectrometry through a heated pneumatic-nebulizer interface, originally developed for liquid chromatographylmass spectrometry (LCIMS), and subsequently modified for use with supercritical fluid chromatography/mass spectrometry (SFC/MS). A high pressure syringe pump was used to pass SFC-grade carbon dioxide, used as the mobile phase, through a capillary SFC column which was contained in an oven, maintained at 100 "C. The eluent passed through the heated pneumatic-nebulizer interface into the ionization region of the mass spectrometer with the aid of the flow of nebulizing gas. The system was optimized using benz[a]anthracene, and then applied to analyze a standard mixture of polycyclic aromatic compounds (PACs) as well as complex mixtures of PACs obtained by the fractionation of a pond sediment contaminated by coke-oven residues. A detection limit of 40 pg was established for chrysene.

The mass spectrometer is the ideal detector for chroma- tography due to its inherent specificity, selectivity, and sensitivity.'-3 The direct interfacing of supercritical fluid chromatography (SFC) with mass spectrometry may provide significant advantages over liquid chromatography/mass spectrometry (LC/MS) because of the relative ease of mobile phase evaporation and removal in SFC. Potential application areas for SFC/MS are in the analysis of high molecular weight compounds that are of limited volatility, and of therm- ally labile compounds for which gas chromatography (GC) is either impossible or impracticable.' The oper- ation of an SFC/MS interface requires that the mobile phase undergo a pressure reduction prior to ionization within the mass spectrometer. For capillary SFCIMS, the sensitivity obtained using electron ionization (EI) is one or two orders of magnitude less than that obtained using chemical ionization (CI);4 in any event, mass spectra that contain ions resulting from both EI and CI are often obtained due to the high C 0 2 pressures in the EI ion s o ~ r c e . ~ Most EI ion sources are relatively gas-tight and are designed to operate at low flow rates or with gases that produce spectra similar to EI spectra (i.e., charge-transfer from He as in GUMS). By using a more open source design to minimize the pressure in the ionization volume, CI spectral contributions in capillary SFC/MS can be rninimi~ed.~ Detection limits of 20 pg have been reported for the capillary SFC/MS analysis of pyrene.'

For the development of an effective SFC/MS inter- face, three requirements have to be fulfilled. First, the gas flow rates generated by volatilization of the mobile phase have to be handled by the interface (this is obviously easier for capillary SFC/MS than for packed

+ NRCC # 34889. * Author for correspondence.

column SFC/MS, due to the lower flow rates associated with the former technique). Secondly, the analyte must be transported into the ion source of the mass spec- trometer without thermal decomposition. Thirdly, the chromatographic integrity must be maintained.

In the earliest SFClMS experiment, Randall and Wahrhaftig6 interfaced a packed column SFC system to an EI mass spectrometer using a nozzle/skimmer/ collimator arrangement with multiple-stage pumping. This system provided EI-like spectra, but was compli- cated by the formation of solvent clusters which varied as a function of the interface temperature in the zone immediately preceding the expansion region, by poor detection limits, and by plugging of the nozzle. Guow et al. pursued a different approach, whereby a packed- column SFC system was connected to a mass spectrom- eter and eluent flow into the ion source was restricted by crimping the capillary at its inlet rather than at its outlet. Packed column SFC/MS was also achieved using a moving-belt interface originally designed for use in LC/MS.* This interface has allowed for super- critical mobile-phase flow rates of up to 4 mL/min even for mobile phases that contain high quantities (30%) of organic modifier. True EI and CI mass spectra are obtainable. However, care must be exercised when thermally labile compounds are analyzed using this interface, as these compounds can decompose both when they are deposited onto the belt and when they are desorbed from the belt. Packed-column SFC has also been used in conjunction with thermospray mass spectrometry,' but the thermospray source had to be modified to accommodate the high pressures associated with SFC. The vaporizer line had to be crimped in order to maintain supercritical conditions until the eluent left the vaporizer, and also to enable the mass spectrometer's vacuum system to handle the gas flow and thus avoid reproducibility problems. When CO,

0 Crown copyright 1994 Government of Canada. Received 28 October I993 Accepted (revised) 26 November I993

106 SFC/MS OF PACs USING APCI

with polar modifiers was used as the mobile phase, solvent CI mass spectra were produced but, when CO, alone was used, EI-type mass spectra could be obtained.' Recently, packed-column SFC/MS was achieved using a particle beam interface" which allowed EI mass spectra to be obtained, but this arrangement suffered from detection problems and highly non-linear calibration curves.

Capillary SFC can be directly coupled to a mass spectrometer by utilizing commercially available inter- faces. In these, the restrictor passes through a probe, at the end of which is positioned a heating element used to heat directly (up to 450 "C) the end of the restrictor and prevent solute precipitation and freezing of the mobile phase in the restrictor during decompression. Such SFC interfaces, which pass directly into the ion source of the mass spectrometer, suffer from sensitivity problems, especially in the EI mode.4

The various problems associated with SFC/MS" could, in principle, be removed effectively if the ioniza- tion of the compounds were to occur at atmospheric pressure. SFC/MS using atmosperic pressure chemical ionization (APCI) has been demonstrated for packed column SFC.'*. l3 The objective of the present experi- ments was to evaluate the potential of coupling capil- lary SFC with APCI for the analysis of polycyclic aromatic compounds (PACs), both as standard solu- tions and as components of complex 'real-world' sam- ples, and to determine the detection limits obtainable.

EXPERIMENTAL The PAC standards used in these experiments were provided by Supelco (Supelco Inc., Bellefonte, PA, USA), and were dissolved (5 mg/mL) in 50: 50 aceto- nitrile + dichloromethane. The contaminated pond sediment originated from the Sydney Tar Ponds, Sydney, NS, Canada. The sediment was fra~tionated,'~ and the fractions produced (containing different compound types) were also dissolved in acetonitrile + dichloromethane (50 : 50). The capillary SFC and capil-

Table 1. Density conditions employed for the separation of PACs by capillary SFC-FID and SFC/APCI-MS

Initial density Final density Ramp rate Time W m L ) W m L ) (g/mL/min) (min) 0.200 0.200 O.Oo0 5 0.200 0.780 0.029 20 0.780 0.780 O.Oo0 5

lary SFC/MS experiments were performed using an Isco SFC-500 micro flow pump (Isco Inc., Lincoln, NE, USA) which was controlled by custom-made software (Isco) installed on a microcompuer. The SFC pump was used in conjunction with a Hewlett-Packard 5890 gas chromatographic oven (Hewelett-Packard Ltd., Avondale, PA, USA) and equipped with a flame ioni- zation detector (FID) which was maintained at 350 "C for the off-line studies. Sample injection was performed using a Rheodyne 7520 internal loop injector fitted with a 0.5 pL loop (Rheodyne Inc., Cotati, CA, USA) in the dynamic split mode. A linear quartz capillary was used as the split restrictor (10 cm in length, 20 pm i.d., and 142 pm 0.d. , obtained from Polymicro Technologies Inc., Phoenix, AZ, USA). SFC-grade CO, with helium head space was used as the mobile phase (Air Products, Allentown, PA, USA) and the column oven was main- tained at 100 "C. For the off-line experiments, the GC oven was controlled, and data collected and processed, by Hewlett-Packard 3365 Chemstation software.

Separations were performed using a J&W DB 225 capillary column which was 10m in length and had 0.05 pm film thickness (J&W Scientific, Folsom, CA, USA). Pressure restriction for the SFC/FID study was accomplished using a frit restrictor (Dionex Corp., Salt Lake City, UT, USA) which had an i.d. of 50 pm and from which the exiting mobile phase had an average velocity of 1.5 cm/sec. The SFC/MS experiments util- ized an integral restrictor (Suprex Corp., Pittsburgh, PA, USA) which had a mobile phase velocity of 1 mL/ min to maintain supercritical conditions. Separation of

Electrical Connection

Figure 1. Schematic diagram of the SFC/APCI-MS interface.

SFCIMS O F PACs USING APCI 107

4

6 3

2

1

Q

m .- c) - 40 2

20

5

0 1 . . . , , . . * . , . . . . , . . . . , . . . . , . . . . , 0 5 10 15 20 25 30

Time (minutes)

Figure 2. Total-ion chromatogram obtained for (1) naphthalene, (2) phenanthrene, (3) pyrene, (4) benz[a]anthracene, (5) perylene, and (6) benzo[ghi]perylene by capillary SFCIAPCI-MS (5 ng of each injected on-column).

1 Napthalene 128 IMI' 14000 1 (a)

1 Phenanthrene 179 +HI+

17R IMI 1 203 [M+H]+

Pyrene

c 0 5 10 15 25 30

Time (minutes)

Figure4. Chromatograms obtained for fraction 1 of the Sydney Tar Ponds sediment by (a) capillary SFC-FID and (b) capillary SFCIAPCI-MS (TIC of mlz 150-650).

ndz Figure 3. Mass spectra obtained for naphthalene, phenanthrene and pyrene, by capillary SFCIAPCI-MS (5 ng of each injected on- column).

108 SFClMS OF PACs USING APCI

I . . . .

n i l

230 '@I 1 I

22s 250

0 4 . , . , . , . , . , . , 0 5 20 25 30 15 10

Time (minutes)

Figure 5. Chromatograms obtained for fraction 8 of the Sydney Tar Ponds sediment by (a) capillary SFC-FID and (b) capillary SFC/APCI-MS (TIC of mlz 150-650). Inset shows the mass spec- trum obtained for the indicated peak, which confirms the compound to be a PANH of MW 229.

the polycyclic aromatic compounds (PACs) in both modes was achieved using a linear density ramp as described in Table 1.

Capillary SFC/MS studies were undertaken using a SCIEX API 111 triple quadrupole mass spectrometer (SCIEX, Thornhill, ON, Canada) equipped with an APCI source. The heated nebulizer interface used for the SFC/MS studies (see Fig. 1) was a modification12 of a conventional LC/MS interface. Here, the capillary column was connected to the transfer line and restrictor through a zero-dead-volume union mounted between the oven and the probe handle of the interface. High-purity air was introduced as the nebulizing gas (0.5-1.5 L/min) and flowed co-axially around the re- strictor to the tip of the interface. The end of the restictor was flush with that of the probe tip, and the design of the tip was such that good thermal contact was maintained between the tip and the end of the restrictor at the exit, after which the supercritical fluid expanded into the APCI source. The final centimetre of the probe tip was heated by a coil of resistance wire using a constant current supply and the temperature (250-300 "C) was monitored by a thermocouple." Mass spectrometer control, data acquisition, and data pro- cessing were accomplished using a Macintosh Quadra 950 microcomputer. The corona-discharge needle was maintained at a current-controlled 3 yA discharge. The quadrupole mass spectrometer was operated using dwell times of 2 and 5ms per m/z for the full-scan studies of the standard solutions and the complex solutions, respectively, and 200 ms per channel for the selected-ion monitoring (SIM) experiments.

RESULTS AND DISCUSSION

Optimization of system using PAC standards

The optimum conditions for the mass spectrometer were determined by monitoring the protonated mole- cule (mlz 229) of benz[a]anthracene (1 mg/mL) while varying the temperature of the nebulizer interface, the fore-pressure of the nebulizer gas, and the position of the corona-discharge needle. As a consequence of the particular design of the SFC interface used in these experiments, when the temperature of the interface was raised to 300°C the polyimide coating of the restrictor burned off along a 10 cm length which left it very brittle and, in some cases, caused the restictor to break while inside the interface. Therefore, it was not possible to determine the optimum temperature set- ting for the mass spectrometer and the SFC/MS experi- ments had to be conducted using an interface tempera- ture of 250°C. This was the highest possible temperature that could be used for nebulization while maintaining the restrictor coating. Variation of the nebulizing gas fore-pressure (and thus flow rate) led to an increase in signal as the pressure was increased from 12 psi to a maximum of 20 psi, but the signal decreased as the pressure was raised above this value. Finally, the optimal corona-discharge-needle position was located when the point of the needle was roughly mid-way between the end of the nebulizer assembly and the orifice of the mass spectrometer, and just below the exit aperture of the APCI source, so that the eluent from the restrictor flowed over the needle. The optimum SFC conditions were established using a standard solu- tion (5 mg/mL) containing the following PACs: naph- thalene, MW 128; phenanthrene, MW 178; pyrene, MW 202; benz[a]anthracene, MW 228; perylene, MW 252; and benzo[ghi]perylene, MW 276. This test mix- ture was analyzed using the above optimal interface conditions and the SFC density programme described in Table 1. The total-ion chromatogram (TIC) obtained for the PAC standards is shown in Fig. 2. Clearly this technique produced excellent chromatographic efficiency, with all of the compounds well separated from one other in order of increasing molecuar weight. One of the inherent draw-backs of this technique is the very high background signal (Fig. 2) produced as a result of the burning off and subsequent ionization of the polyimide material of the restrictor (uide supra). However, this effect was seen to diminish after the restrictor had been in use for some time.

When the APCI-MS spectra of the lower-molecular- weight compounds (naphthalene and to a lesser extent phenanthrene) are compared with those of the higher- molecular-weight PACs (Fig. 3), the response is seen to be qualitatively different. The mass spectra of naptha- lene and phenanthrene are characterized by molecular ion species [MI" as well as protonated molecules [M + HI', reflecting the relatively low proton affinities of these compounds. This resulted in the ionization of these two compounds being accomplished by both charge-transfer and proton-transfer, whereas the larger PACs, of MW 200 or more, are ionized almost entirely by protonation (Fig. 3). Charge-transfer mass spec- trometry is much less sensitive than proton-transfer mass spectrometry" for PACs under the present condi- tions. This effect can be understood by consideration of the fundamental gas-phase chemistry of atmospheric-

SFClMS OF PACs USING APCI 109

pressure ionization. l6 In air, the positive primary ions N l * , O:., H20+ ' , and NO+ are formed initially, but then undergo a complex reaction sequence which pro- duces, at equilibrium, the proton hydrates (H,O+[H,O],) as the domnant reaction ions.I6

The reactant ions (R) can react with the analyte molecules (T) by charge-transfer (Reaction 1):

R+ + T -.T+ + R and/or by simple proton-transfer (Reaction 2):

RH+ + T +TH+ + R In an APCI source, proton transfer to the analyte from a proton hydrate can also result in a hydrated TH+ ion formed (Reaction 3) within the APCI source.

H,O+(H,O), + T+ TH+(H20), + (n - m + 1)H,O (3) In practice, Reaction 2 is not observed and Reaction 3 is the dominant protonation mechanism. However, only TH+ ions are transmitted into the mass spec- trometer and subsequently detected, since any clus- tered water molecules are stripped off during passage through the gas curtain and declustering-lens region of the mass spectrometer. Proton affinity is therefore an important factor determining analyte response in APCI mass spectrometry. Reaction 3 will proceed only if the proton affinity of the analyte is greater than that of water clusters." Further, the efficiency of production of TH+ is affected by the presence of other compounds in the source, due to possible competition for proton transfer. The ionization efficiency is thus a function of relative proton affinities of all species present in the source at that time. The relative APCI response of the PACs becomes greater as the proton affinity increases, i.e. generally with increasing size and number of rings."

In general, proton hydrates have lower recombina- tion energies than most compounds and therefore do not ionize many analytes via charge transfer. The reactant ions responsible for charge transfer in air are 0:. and its hydrates and, to some extent, N i ' , HzO+', and NO+. Since these ions react efficiently with water to form the proton hydrates,16 the APCI system under ambient conditions exhibits lower sensitivity for charge-transfer than for proton-transfer.

(1)

(2)

Analysis of complex mixtures In order to evaluate the performance of the interface with respect to real samples, eleven mixtures of PACs, obtained by the fra~tionation'~ of a contaminated pond sediment, were analyzed by capillary SFC-FID and SFC/MS. The results obtained for the first (Fig. 4) and eighth (Fig. 5) fractions (they were analyzed both by FID and MS) show that these two fractions have markedly different compositions. The chromatograms of Fig. 4 (fraction 1) arise entirely from polycyclic aromatic hydrocarbons (PAHs); the FID trace is shown in Fig. 4(a) and the TIC of the mass spectra in Fig. 4(b). The chromatograms of Fig. 5 (fraction 8) arise from polycyclic aromatic nitrogen compounds (PANH) in the FID trace (Fig. 5(a), and from protonated PANH in the TIC trace). The identity of these compounds can be illustrated by considering the inset in Fig. 5(b) where the spectrum extracted at the crest of the indicated chromatographic peak shows that this compound is a

d z 217

250 300 350

mlr

0 5 10 15 20 25 30

Time (minutes)

Figure 6. Reconstructed ion chromatogram for mlz 277 'obtained by capillary SFC/APCI-MS for fraction 1 of the Sydney Tar Ponds sediment. Inset shows the mass spectrum obtained for the indicated peak confirming the compound to be a PAH of MW 276.

PANH of MW 229 (a benzacridine). The full charac- terizations of these fractions will be published ~eparately. '~

The retention times obtained by the FID and MS detection systems for the fraction differ by a few minutes, due to the differing flow characteristics of the restrictors used in each experiment. The construction of the SFC/MS interface, and its positioning with respect to the oven, resulted in a length of 10-15 cm of the column and restrictor being outside both the oven and heated nebulizer interface. However, the chroma- tographic integrity could be maintained without having to change the density programme or the oven tempera- ture. The preservation of chromatographic integrity, and the confirmation of the identities of the eluting compounds as PAHs, is exemplified in Fig. 6 which shows the reconstructed ion chromatogram (RIC) for mlz 277 obtained from the first fraction (the mass spectrometer was scanned from mlz 150 to m/z 650 in steps of one mass unit using a dwell time of 5 ms per step). These ions correspond to protonated molecules of isomers of MW 276 (possibly indeno[l,2,3- cdlpyrene and benzo[ghi]perylene, confirmation being obtained by the spectrum shown in the inset). Comparison of the RIC (Fig. 6) and the FID trace (Fig. 4(a), region 1) shows that the transmission of the compounds through the heated nebulizer interface has been accomplished without any significant loss in reso- lution although the peaks did tail somewhat, probably due to the cold spot between the G C oven and the heated nebulizer interface. However, the information content of the SFC/MS experiment is far greater than that of the SFC-FID analysis because of the infor- mation concerning molecular mass that is provided.

110 SFClMS OF PACs USING APCI

O J 14 16 18

Time (minutes)

Figure 7. Total-ion chromatogram obtained by capillary SFC/APCI-MS for triplicate injections of 40 pg of chrysene.

The mass spectrometer is a much more selective detec- tor and this gives greater certainty in the compound identification, with consequent advantages in quantifi- cation.

Finally, the detection limits (signal-to-noise ratio = 2 : 1) for pure chrysene were determined using the optimum conditions described above and SIM of the [M + HI+ ion. The SFC system was operated using constant density conditions, and triplicate injections of chrysene solution were performed to check the repro- ducibility over a range of concentrations. The results shown in Fig. 7 correspond to an instrumental detection limit of 40 pg injected onto the column.

CONCLUSION Capillary column SFC can be successfully interfaced, using a heated pneumatic nebulizer, to an APCI mass spectrometer. The results show that the technique is indeed selective and sensitive. The system is not com- pletely optimized as yet, but the results are promising and improvements will be made. Future work will be

concentrated on developing an interface in which no part of the column is outside the oven or interface, and other classes of compounds will be investigated to determine their compatibility with capillary SFC/APCI-MS and their detection limits. Because APCI response is a function of the concentration of co- eluting analytes, internal standards will be needed to measure calibration curves.

REFERENCES 1. R. D. Smith, W. D. Felix, J. C. Fjeldsted and M. L. Lee, Anal.

Chem. 54, 1883 (1982). 2. B. W. Wright, H. T. Kalinoski, H. R. Udseth and R. D. Smith,

J . High Resoln. Chromatogr. IChromatogr. Commun. 9 , 145 (1986).

3. A. J. Berry, D. E. Games, I. C. Mylchreest, J. R. Perkins and S. Pleasance, J. High Resoln. Chromatogr. IChromatogr. Commun. 11, 61 (1988).

4. R. D. Smith, J. C. Fjeldsted and M. L. Lee, J . Chromatogr. 247, 241 (1982).

5. R. D. Smith, H. R. Udseth and H. T. Kalinoski, Anal. Chem. 56,2971 (1984).

6. L. G. Randall and A. L. Wahrhaftig, Anal. Chem. 50, 1703 (1978).

7. T. H. Guow, R. E. Jentoft and E. J. Gallegos, J. High Press. Sci. Technol. 583 (1979).

8. A. J. Berry, D. E. Games and J. R. Perkins, J . Chromatogr. 363, 147 (1986).

9. A. J . Berry, D. E. Games, I. C. Mylchreest, J. R. Perkins and S. Pleasance, Biomed. Environ. Mass Spec. 15, 105 (1988).

10. J. Beaman, D. E. Games, P. J. Jackson and D. Thomas, Adv. Mass Spectrom. 12 (1992).

11. P. J. Aprino, F. Sadoun and H. Virelizier, Chromatographia 36, 283 (1993).

12. J. F. Anacleto, L. Ramaley, R. K. Boyd, S. Pleasance, M. A. Quilliam, P. G. Sim and F. M. Benoit, Rapid Commun. Mass Spectrom. 5 , 149 (1991).

13. K. Matsumoto, S. Nagata, H. Hattori and S. Tsuge, J. Chromutogr. 605, 87 (1992).

14. H. Perreault, L. Ramaley, F. M. Benoit, D. Thomas, S. Crain, G. K. McCully and P. G. Sim (in preparation).

15. J. F. Anacleto, Ph.D. Thesis, Dalhousie University, Halifax, 1993.

16. The API Book, SCIEX, Thornhill, Ontario, NS, Canada, 1992,

17. Introduction to Mass Spec trometry: Biomedical, Environmental, and Forensic Applications, ed. by J. T. Watson, p. 114, Raven Press, 1976.

p. 15.

18. M. Mautner, J . Phys. Chem. 84, 2716 (1980).