763
Anal. Chem. 1985, 57,763-765
Analysis of Polynuclear Aromatic Mixtures by Liquid Chromatography/Mass Spectrometry Kenneth J. Krost
U.S. Environmental Protection Agency, Environmental Monitoring Systems Laboratory, Research Triangle Park, North Carolina 27711 LCEFFLUENT
The use of hlghperformance liquid chromatography combined wlth mass spectrometry Is discussed for the analysis of polynuclear aromatlc hydrocarbons. Parameters were determined as a functlon of eluent composltlon and flow. Preclslon and sensltlvlty were evaluated, along wlth the effect of the aforementloned parameters on overall performance.
LCELUENT STRIPPER
1
(INFRA-RED HEATER
SPLIT OPTION LOADED WHEEL
WHEELS TOVACUUM P m i s
High-performance liquid chromatography (HPLC) has proven to be an invaluable analytical tool in the separation of organic compounds. A basic incompatibility however exists with mass spectrometry. Up to 1000 cm3/min of vapor can be generated from the eluent solvent during normal operation. Under these conditions, maintaining the low-mass analyzer pressure necessary for electron ionization mass spectrometry is all but impossible. Three general approaches have been taken to circumvent this incompatibility. First is direct liquid injection, which involves vaporizing the eluent with the analyte (1, 2). The vapor is then directed as an aerosol to the ion source, serving as both an analyte carrier and a reagent gas for chemical ionization. Pressures of torr me observed in the ionization chamber. Furthermore, solvent flow is limited to 10-15 pL/min. When conventional mass spectrometry is used, informational content is limited to chemical ionization spectra. Scott described a second general technique used in HPLC/MS analysis, in which the analyte was deposited onto a moving wire (3,4). This was followed sequentially by solvent evaporation and thermal desorption. Unfortunately, the capacity of the transport wire was restricted to around 1% of a standard HPLC column flow (1.0 cm3/min), with a concurrent loss in instrument sensitivity. A third popular technique involves analyte deposition onto a moving belt, followed by thermal eluent stripping (5-9). This technique suffers from the previously discussed incompatibility. Necessity dictates the use of a sample splitter, with a loss in sensitivity of up to 95 70. A fourth popular technique is that described by Vestal (IO). This technique involves thermospraying effluents directly into the ion source and pumping away excess vapor. A conventional electron beam is used to provide gas-phase reagent ions for chemical ionization of solute molecules.
Flgure 1. Schematic of Finnigan belt-driven LC interface. 100K.H~O
500 400
'O0I 200
/ -*__.
E.I.
I I 100% HEXANE 30
+
SAMPLE VOLATILIZATION Conditions necessary for sample volatilization were established by using a standard test mixture of anthracene (12.4 ng/pL), pyrene (13.2 ng/pL), chrysene (10.2 ng/pL), and benzo[a]pyrene (10.2 ng/pL). A sample loop of 10 pL was used for injection. During eluent stripping, one attempts to preserve the analyte while removing excess eluent. Figure 3 depicts this situation. Analyte losses, incurred during eluent stripping, accounts for a significantly lowered molar response for pyrene and anthracene. The recoveries of benzo[a]pyrene and chrysene, however, vary from essentially zero a t temperatures of 100 O C to semiquantitative a t temperatures of 200 O C . In the lowered response observed for pyrene and anthracene, one is dealing with a loss in sensitivity due to prevolatilization prior to the ionization source. In the case This artlcle not subject to US. Copyright. Published 1985 by the Amerlcan Chemical Society
-
764
ANALYTICAL CHEMISTRY, VOL. 57,
NO. 3,
MARCH 1985 V A P O R I Z E R . 155 OC CLEAN-UP. 50 OC I O N S O U R C E . 100 OC
V A P O R I Z E R . 1 0 0 OC C L E A N 4 P 50 OC I O N S O U R C E . 100 OC
-
I
I
I
MASS ION 202 COUNTS = 7560
ElCP
h
>
c
*----I
I
h
MASS ION 178 COUNTS = 3384
v1
I
c yi
E
MASS ION 228
COUNTS = 10,032
r
1
1
VAPORIZER -200 OC CLEAN-UP. 50 OC I O N S O U R C E . 100 OC
I
I
c R
ElCP COUNTS = 3300
A
COUNTS MASSIoN = 214,304 02
MASS ION 202 COUNTS = 7032 -m
yi I
'
VAPORIZER -200 OC
I
MASS I ON 178 COUNTS = 3500
~
l.*cIIII-c-cIh---J CLEAN-UP. 1 1 0 OC IONSOURCE-200°C
I
I
I
1
1
A
MASSION228 COUNTS = 21,504
L
MASS ION 252 COUNTS = 20,512
0
100
A
MASS ION 228 COUNTS = 20,128
A
MASS ION 252 COUNTS = 22.784
200 300 SCAN NUMBER
400
500
0
200 300 SCAN NUMBER
100
400
500
Figure 3. Ion intensity as a function of volatilization temperature: (a) vaporizer 100 "C, cleanup 50 OC, ion source 100 "C; (b) vaporizer 155 "C, cleanup 50 OC, ion source 100 "C; (c) vaporizer 200 "C, cleanup 50 "C, ion source 100 OC; (d) vaporizer 200 OC, cleanup 110 OC, ion source 200 "C.
Table I. Precision Measurements of Standard Mixture
UNCORRECTED RAW DATA N U M B E R OF READINGS
CONC. IN STD. ABSOLUTE A M O U N T PEAK A R E A AVERAGE COUNTS R A N G E OF PEAK A R E A
ANTHRACENE(178) 12
PYRENE(202) 12
CHRYSENE(228) 12
B(a)P(252)
22ng
24ng
18ng
18ng
4763 6400-3888
10,972 13.1 36-8736
20,073 29,248-14.41 6
20,006 24,544-14,768 16.0
12
E R R O R AT 1.0 SD,
percent ABSOLUTE E R R O R AT 2.0 SD STANDARD DEVIATION
16.4
13.0
19.0
22i7ng -3.52ng
2416ng -3.12ng
18+7ng -3.42ng
18+6ng -2.88ng
CORRECTED RAW DATA E R R O R AT 1 .O S D ,
percent ABSOLUTE E R R O R AT 2.0 S D
12.4
10.2
22i5ng
24+5ng
of chrysene and benzo[a]pyrene losses a t reduced temperatures are probably caused by lack of volitilization. The necessity of choosing volatilization temperatures appropriate for the compounds under consideration is apparent.
PRECISION AND SENSITIVITY Table I summarizes the precision and sensitivity for 12
9.1 18f3ng
-
determinations of the compounds in the standard polynuclear aromatic mixture. In the corrected data, the use of an internal standard (chrysene) was used to normalize the response of the three other compounds. In the uncorrected data, no internal standard was used to normalize the response of the three compounds under evaluation. This experiment was designed
Anal. Chem. 1985, 57,765-768
to evaluate the effect of an internal standard to compensate for instrumental variations. The variation for benzo[a]pyrene was somewhat less than either the corrected or uncorrected values of other compounds. This phenomenon probably reflects the fact that less sample loss occurred during solvent stripping. It evidences the need for accurate control of LC interface setting. The use of internal standards will not adequately compensate for day-today variation in stripper and volatilization temperatures.
CONCLUSIONS PNA sensitivity is, to an extent, a direct function of compound volatility. Losses incurred during eluent stripping appear to be a significant factor in the case of lower molecular weight compound sensitivity. Volatilization temperatures are critical for the recovery of high molecular weight compounds. The lower limit of detectability for polynuclear aromatics in the ion-selective mode is approximately 10 ng. The precision observed on corrected raw data appears somewhat better at higher PNA molecular weights. This again probably reflects losses during solvent stripping and prevolatilization.
765
LITERATURE CITED (1) Tal'rose, V. L.; Karpov, G. V.; Gordoershll, I. G.; Skurat, V. E. Russ. J. Phys. Chem. 1968, 42, 1658. (2) Tal'rose, V. L.; Grlshln, V. D.; Skurat, V. E.; Tantsyrev, G. D. "Recent Developments In Mass Spectrometry"; Ogta, K., Hagakawa, T., Eds.; University Park Press: Baltimore, MD, 1970. (3) Scott, R. P. W.; Scott, G. C.: Munroe, M.; Hess, J. J. Chromatogr. 1974, 99,395. (4) Scott, R. P. W. J . Chromatogr. Llbr. 1977, 11. (5) McFadden, W. H.; Schwartz, H. L.; Evans, S. J. Chromatogr. 1976, 122, 389. (6) McFadden, W. H.; Bradford, D. C.; Games, E. E.;Gower, J. L. Am. Lab (Fairfieid, Conn.) 1977, 9 , 55. (7) Dark, W. A.; McFadden, W. H.; Bradford, D. C. J. Chromatogr. Sci. 1977, 15,454. (8) Dark, W. A.; McFadden, W. H. J. Chromatogr. Sci. 1976, 16, 289. (9) McFadden, W. H.; Bradfor, D. C.; Egllnton, G.; Haglbrahim, S.; Dark, W. A.; Nlcolaides, N., 26th Annual Conference of Mass Spectrometry and Allied Topics; May 1978, St. Louis, MO.
RECEIVED for review October 5,1984. Accepted December 3, 1984. This article has not been subjected to Agency review and does not necessarily reflect the views of the Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.
CORRESPONDENCE Ionization and Mass Analysis of Nonvolatile Compounds by Particle Bombardment Tandem-Quadrupole Fourier Transform Mass Spectrometry Sir: Recent advances in the development of new ionization techniques have now made it possible to generate (M + H )', (M - H)-, and (M + cation)' ions as well as structurally significant fragment ions from many polar biological molecules having molecular weights up to as high as 13000 (1-3). Techniques such as fast atom bombardment (FAB) or liquid secondary ion mass spectrometry (L-SIMS) in which sample ions are ejected from a liquid matrix such as glycerol under bombardment with kilovolt atoms or ions, respectively, are presently in widespread use on both commercial quadrupole and magnetic sector instruments. Unfortunately, advances in the design of instrumentation for mass analysis and structural characterization of the ejected ions have not kept pace with developments in new ionization methodology. Large magnetic sector instruments operating a t full accelerating potential provide molecular weight information up to 12 000 for samples in the 0.05-5 nmol range. Ion current in the mass range greater than 3000 daltons is usually too weak, however, to allow determination of molecular structure from fragmentation patterns. Collision-activated dissociation provides a mechanism for enhancing the structural information obtained in particle bombardment experiments, but kinetic energy released during the fragmentation process limits the resolution achieved by using tandem four sector instruments on samples at the 50 pmol level to ca. 2000. Considerably higher resolution can be achieved if the quantity of sample is increased by a factor of 10-100. Above mass 2500-3000, the combination of weak sample ion current and low ion transmission at unit resolution following the collision cell severely restricts the analytical utility of tandem sector instruments. Presently available tandem quadrupole instru-
ments provide structurally rich collision-activated dissociation (CAD) spectra a t unit resolution (4, 5 ) , but inefficient ion transmission above mass 1000 makes these instruments unsuitable for trace level analysis of high molecular weight biological materials. Fourier transform mass spectrometry (FT-MS) offers an attractive alternative to the above instrumental methods (6-10). It can provide a mass range in excess of 10000 daltons in principle and can function as an ion storage device to accumulate ions produced in low abundance. Mass analysis is accomplished in Fourier transform mass spectrometry without destruction of the ions, and mixture components can be separated by using the double resonance technique. Consecutive collision-activated dissociation (11, 12) and photodissociation (13-16) experiments can be performed on the same ion population. Long ion storage times and ultrahigh mass resolution have been demonstrated, but only when tlie pressure in the analyzer cell is below lo-* torr. Glycerol and other liquid matrices employed in the particle bombardment ionization methods all have vapor pressures in excess of lo4 torr and are therefore difficult to use in conjunction with Fourier transform mass spectrometry. One way to overcome the problem associated with high gas flow during the ionization step is to use a differentially pumped, two compartment cell within a conventional Fourier transform mass spectrometer. An instrument containing such a cell has recently become available from Nicolet Instruments. Another approach utilizes a tandem-quadrupole Fourier transform mass spectrometer (QFT-MS) in which sample introduction and ionization are carried out in a differentially pumped quadrupole ion source and only the ions of interest
0003-2700/85/0357-0765$01.50/00 1985 American Chemical Society