Anal. Chem. 1998, 70, 2660-2665
Laser Mass Spectrometric Analysis of Polycyclic Aromatic Hydrocarbons with Wide Wavelength Range Laser Multiphoton Ionization Spectroscopy Olivier P. Haefliger and Renato Zenobi*
Department of Chemistry, Swiss Federal Institute of Technology (ETH), CH-8092 Zu¨ rich, Switzerland
In many analytical techniques, 1+1 resonance-enhanced multiphoton ionization (1+1 REMPI) is used because it is an efficient and optically selective soft ionization method. While 1+1 REMPI of jet-cooled molecules has been extensively studied, little has been reported so far about this mechanism as it is used in analytical techniques, that is, in the cases where the molecules are not jet-cooled and where widely varying ionization wavelengths are employed. We used two-step laser mass spectrometry (L2MS) to study the wavelength (238-310 nm) dependence and the laser pulse energy dependence of the ion yield for 17 polycyclic aromatic hydrocarbons (PAHs). We discuss how these data allow prediction of the efficiency of 1+1 REMPI for a given compound. These advances open new perspectives for better understanding the L2MS spectra obtained directly from complex mixtures such as environmental samples. Polycyclic aromatic hydrocarbons (PAHs), a family of molecules built up from fused benzene rings, make up a major class of environmental pollutants. They are emitted as a result of incomplete combustion of organic material. Major sources are traffic, industrial processes, power and heat generation, and tobacco smoke.1,2 Typical air concentrations are in the range of 0.1-10 ng/m3 for individual compounds.3 Strong variations have been observed during the day and between different locations.4 It is now well established that most PAHs have carcinogenic, mutagenic, teratogenic, or immunotoxic effects on animals and humans. The reaction products of PAHs with other atmospheric pollutants can be even more toxic than the original PAH.2,3 A reliable, sensitive, and selective method for the identification and quantitation of PAHs in environmental samples is, therefore, essential to estimate the potential health hazards. The most widely used method is gas chromatography (GC),5-12 with either a flame ionization (GC-FID) or a mass spectrometric (GC/MS) detector. Boesl and co-workers described GC separation of PAHs
with detection using a supersonic beam expansion, 1+1 resonanceenhanced multiphoton ionization (REMPI), and time-of-flight mass spectrometry.13 Several other analytical methods have been reported, such as high-performance liquid chromatography (HPLC),14,15 on-line liquid chromatography coupled with gas chromatography (LC-GC),16,17 laser-induced time-resolved fluorescence,18 visible and ultraviolet spectroscopy,13,19 and mass spectrometry.13,19,20 The limitations of GC are well known: a tedious sample preparation, usually comprising a Soxhlet extraction followed by a purification and a concentration step, is necessary; at least about 0.2 ng of a given PAH must be injected into the column to allow its detection; a single measurement can last for a few hours; due to their insufficient volatility, PAHs heavier than 300 mass units do not pass through standard columns. Two-step laser mass spectrometry (L2MS) offers several advantages over conventional methods.4,21,22 Little or no sample preparation is necessary, and the direct chemical analysis of complex environmental samples is possible. Furthermore, a measurement is performed within a few minutes. A detection limit in the low attomole range, much lower than for most conventional methods, has been demonstrated several times.23-25 In the first
* Address correspondence to this author at Department of Chemistry, ETHZentrum, Universita¨tstr. 16, CH-8092 Zu ¨ rich, Switzerland. Tel.: +41-1-632 4376. Fax: +41-1-632 1292. E-mail:
[email protected]. (1) Lee, M. L.; Novotny, M. V.; Bartle, K. D. Analytical Chemistry of Polycyclic Aromatic Compounds; Academic Press: New York, 1981. (2) Finlayson-Pitts, B. J.; Pitts, J. N. Science 1997, 276, 1045. (3) Atkinson, R.; Arey, J. Environ. Health Perspect. 1988, 102, 117-126. (4) Zhan, Q.; Voumard, P.; Zenobi, R. Rapid Commun. Mass Spectrom. 1995, 9, 119-127.
(5) Giger, W.; Schaffner, C. Anal. Chem. 1978, 50, 243-249. (6) McDow, S. R.; Giger, W.; Burtscher, H.; Schmidt-Ott, A.; Siegmann, H. C. Atmos. Environ. 1990, 24A, 2911-2916. (7) Grimmer, G.; Jacob, J.; Dettbarn, G.; Naujack, K.-W. Fresenius’ J. Anal. Chem. 1985, 322, 595-602. (8) Leuenberger, C.; Czuczwa, J.; Heyerdahl, E.; Giger, W. Atmos. Environ. 1988, 22, 695-705. (9) Canton, L.; Grimalt, J. O. J. Chromatogr. 1992, 607, 279-286. (10) Oehme, M. Anal. Chem. 1983, 55, 2290-2295. (11) Wilkerson, C. W., Jr.; Colby, S. M.; Reilly, J. P. Anal. Chem. 1989, 61, 2669-2673. (12) McClennen, W. H.; Arnold, N. S.; Roberts, K. A.; Meuzelaar, H. L. C.; Lighty, J. S.; Lindgren, E. R. Combust. Sci. Technol. 1990, 74, 297-309. (13) Zimmermann, R.; Lermer, C.; Schramm, A.; Kettrup, A.; Boesl, U. Eur. Mass Spectrom. 1995, 1, 405-412. (14) Go ¨tze, J.-J.; Schneider, J.; Herzog, H.-G. Fresenius’ J. Anal. Chem. 1991, 340, 27-30. (15) Pace, C. M.; Betowski, L. D. J. Am. Soc. Mass Spectrom. 1995, 6, 597-607. (16) O ¨ stman, C.; Bemgard, A.; Colmsjo¨, A. J. High Resolut. Chromatogr. 1992, 15, 437-443. (17) Kelly, G. W.; Bartl, K. D.; Clifford, A. A. J. Chromatogr. Sci. 1993, 31, 7376. (18) Niessner, R.; Robers, W.; Krupp, A. Fresenius’ J. Anal. Chem. 1991, 341, 207-213. (19) Giger, W.; Blumer, M. Anal. Chem. 1974, 46, 1663-1671. (20) Dass, C. J. Am. Soc. Mass Spectrom. 1990, 1, 405-412. (21) Zare, R. N.; Hahn, J. H.; Zenobi, R. Bull. Chem. Soc. Jpn. 1988, 61, 87-92. (22) Zenobi, R. Chimia 1994, 48, 64-71.
2660 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
S0003-2700(97)01264-X CCC: $15.00
© 1998 American Chemical Society Published on Web 05/21/1998
related techniques, such as 1+1 REMPI coupled with TOF-MS, for the on-line analysis of room-temperature gases.11,37-39 The ultimate goal was to select suitable ionization wavelengths for selective isomer ionization. It turned out that this goal could only be partly achieved; we will discuss here the reasons for this finding and the possibilities that our results, nevertheless, offer. In particular, a much better interpretation of L2MS spectra obtained directly from more complex mixtures such as environmental samples is now possible.
Figure 1. Typical L2MS spectrum. PVC membrane with eight PAHs. About 100 fmol of each PAH is ablated by a single laser shot. Ionization at 250 nm. Mass resolution: routinely about 1200. No lowmass fragments are observed.
step of L2MS, an infrared laser pulse desorbs intact neutral molecules from the sample surface or, as used in this work, ablates intact neutral molecules from a thin polymer membrane containing the analytes. In the second step, a pulse from a tunable ultraviolet laser is used for resonance-enhanced two-photon ionization (1+1 REMPI) of the desorbed species;26 this soft ionization scheme prevents fragmentation. Mass analysis is then performed in a reflectron time-of-flight mass spectrometer (TOF-MS). The mass spectra are dominated by intact parent ions of those mixture components that strongly absorb the selected ionization laser wavelength (Figure 1). Two remaining problems have prevented L2MS from finding broader fields of application until now: quantitative analysis in complex mixtures was difficult or impossible and has only been reported with artificially made-up mixtures,23 and mass isomers could not be separated in a satisfactory way. We have recently addressed the first limitation by a novel sample preparation method.27 To overcome the second limitation, we used standards to measure optical 1+1 REMPI spectra for 17 PAHs at ionization wavelengths covering the complete range between 238 and 310 nm. This broad wavelength range has recently become available using optical parametric oscillator (OPO) lasers. The spectra we present below can be very useful to assess the ionization efficiency if using fixed-wavelength lasers or dye lasers in L2MS28-36 or (23) Hahn, J. H.; Zenobi, R.; Zare, R. N. J. Am. Chem. Soc. 1987, 109, 28422843. (24) Voumard, P.; Zhan, Q.; Zenobi, R. Chem. Phys. Lett. 1995, 239, 89-94. (25) Voumard, P.; Zhan, Q.; Zenobi, R. Rev. Sci. Instrum. 1993, 25, 3393-3402. (26) Grotemeyer, J.; Schlag, E. W. Angew. Chem. 1988, 100, 461-474. (27) Haefliger, O. P.; Zenobi, R. Rev. Sci. Instrum. 1998, 69, 1828-1832. (28) McKay, D. S.; Gibson, E. K., Jr.; Thomas-Keprta, K. L.; Vali, H.; Romanek, C. S.; Clemett, S. J.; Chillier, X. D. F.; Maechling, C. R.; Zare, R. N. Science 1996, 273, 924-930. (29) Clemett, S. J.; Maechling, C. R.; Zare, R. N.; Swan, P. D.; Walker, R. M. Science 1993, 262, 721. (30) Dale, M. J.; Jones, A. C.; Pollard, S. J. T.; Langridge-Smith, P. R. R.; Rowley, A. G. Environ. Sci. Technol. 1993, 27, 1693-1695. (31) Dale, M. J.; Jones, A. C.; Pollard, S. J. T.; Langridge-Smith, P. R. R. Analyst 1994, 119, 517-578. (32) Kornienko, O.; Ada, E. T.; Hanley, L. Anal. Chem. 1997, 69, 1536-1542. (33) Hankin, S. M.; John, P.; Smith, G. P. Anal. Chem. 1997, 69, 2927-2930. (34) Hankin, S. M.; John, P.; Simpson, A. W.; Smith, G. P. Anal. Chem. 1996, 68, 3238-3243. (35) Belov, M. E.; Alimpiev, S. S.; Mlynsky, V. V.; Nikiforov, S. M.; Derrick, P. J. Rapid Commun. Mass Spectrom. 1995, 9, 1431-1436.
EXPERIMENTAL SECTION A. Chemicals and Sample Preparation. All of the PAHs were purchased from either Chem Service Inc. (West Chester, PA) or Aldrich (Buchs, Switzerland) and had a purity of at least 95%. Stock solutions were prepared in methylene chloride (J. T. Baker, Deventer, Holland) with concentrations around 20 mg/10 mL. A procedure described elsewhere was used to prepare thin poly(vinyl chloride) (PVC, high molecular weight, Fluka, Buchs, Switzerland) membranes containing the analytes that could be used to perform laser ablation.27 This sample preparation procedure allows a homogeneous spreading of the analytes and an accurate control of the ablated quantity and ensures that the PAHs do not sublime in the vacuum chamber despite their high vapor pressures. It takes about 30 min to produce a membrane that is ready to be measured. B. L2MS System. A home-built L2MS system was used that has already been described in detail elsewhere.25 Only the major characteristics are given here. The output of the CO2 laser used for ablation (Alltech 853 MS, Lu¨beck, Germany, 0.6 J/cm2, 100 ns) was directed into the system by a set of mirrors. An iris and a ZnSe lens (f ) 50 mm) mounted on a micrometer screw allowed control of the energy density of the beam and the size of the ablation spot. The settings used for this project corresponded to about 6 J/cm2 on an elliptic spot of 135 µm × 180 µm. This fluence generates a cylindrical hole that completely penetrates the 50-µm-thick membrane. The ionizing laser radiation, delayed by 10 µs with respect to the desorption laser, was produced by an optical parametric oscillator laser (MOPO-730D20, Spectra Physics Lasers Inc., Mountain View, CA, 10 ns, 0.001-nm line width) pumped by the third harmonic of a pulsed Nd:YAG laser (GCR-230, Spectra Physics Lasers). Wavelengths between 238 and 310 nm were used for this project, which correspond to photon energies between 5.21 and 4.00 eV; shorter wavelengths were not available, and longer wavelengths did not yield efficient enough ionization of the selected PAHs to make this range interesting for L2MS applications. The UV beam passed through the plume of ablated analytes 2.5 mm above the sample surface after being gently focused by a cylindrical lens (f ) 250 mm). The UV energy density was adjusted to the desired value with a remote-controlled motorized polarizer. A pyroelectric detector (Gentec ED-100) was placed after the ionization source to measure the energy of each (36) Alimpiev, S. S.; Mlynski, V. V.; Belov, M. E.; Nikiforov, S. S. Anal. Chem. 1995, 67, 181-186. (37) Zimmermann, R.; Heger, H. J.; Kettrup, A.; Boesl, U. Rapid Commun. Mass. Spectrom. 1997, 11, 1095-1102. (38) Gittins, C. M.; Castaldi, M. J.; Senkan, S. M.; Rohlfing, E. A. Anal. Chem. 1997, 69, 286-293. (39) Rhodes, G.; Opsal, R. B.; Meek, J. T.; Reilly, J. T. Anal. Chem. 1983, 55, 280-286.
Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2661
UV pulse; the recorded values are necessary for normalization of the mass spectra (cf. Results section). Pieces of membrane with a diameter of 12 mm were mounted onto the tip of a sample holder prior to their introduction into the vacuum chamber. A remote-controlled motorized stage allowed rotation of the sample in order to expose a new spot for each ablation laser shot. Mass spectrometric analysis was performed using a reflectron time-of-flight instrument (R. M. Jordan Co., Grass Valley, CA). A mass resolution of 1200 was routinely achieved (Figure 1). RESULTS AND DISCUSSION A. UV Energy Dependence of the Signals. The energy of the UV laser pulses is known to greatly influence the intensity of the signals in REMPI experiments, so an optimum energy setting had to be found. In addition, pulse-to-pulse energy fluctuations of the OPO laser made it necessary to record the energy of each ionization pulse to normalize the mass spectra. The energy dependence discussed below follows the treatment by Johnson and Otis.40 If an unfocused UV beam is used, and in the absence of saturation, a quadratic dependence of the signal intensity (S) on the UV energy (E) is expected for a two-photon process with identical absorption cross sections for the ground state (σ1) and for the excited state (σ2): S ∝ E2. If, however, one of the two cross sections is much larger than the other one, the signal simply becomes linearly proportional to the energy: S ∝ E. The general case is S ∝ E2-x. To ensure good mass resolution, we had to use a slightly focused laser beam in our experiment. In this case, saturation of the ionization in the focus region may occur at elevated energies. If this happens, an increase of the laser intensity creates no more ions in the Rayleigh length. Rather, the ionization region begins to move out of the laser focus region. It can be shown that S ∝ E3/2, independent of how many photons are involved in the ionization. Spectroscopic studies require one to work below the saturation threshold, as was done for all of the spectroscopic data presented below. The UV energy dependence of the L2MS signals was measured at an ionization wavelength of 255 nm for seven PAHs of various sizes (Figure 2). Two cases were observed. (a) The two smallest analytes, naphthalene and acenaphthene, were observed to exhibit a behavior close to S ∝ E2; this probably means that σ1 ≈ σ2. (b) For the other PAHs (phenanthrene, fluoranthene, chrysene, benzo[a]pyrene, dibenz[a,h]anthracene), a linear behavior was found. At the laser energies used, no saturation was observed. Therefore, we interpret this finding as indicative for either σ1 . σ2 or σ1 , σ2. Another finding was that the signal intensities decreased when pulse energies greater than about 32 µJ were applied; fragment ions that were not present at lower fluences simultaneously appeared in the mass spectra. This means that a third photon was absorbed, either by neutral molecules that have already absorbed two photons or by newly formed ions, and that this additional energy induces fragmentation. It is only possible to speculate why two different patterns of behavior are observed. The number of conjugated rings as well as differences in the electronic states accessed may play a role. Therefore, all of our spectroscopic measurements were performed using pulse energies of 25-30 µJ. The recorded mass (40) Johnson, P. M.; Otis, C. E. Annu. Rev. Phys. Chem. 1981, 32, 139-157.
2662 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
Figure 2. UV energy dependence of the L2MS signal of selected analytes. Ionization at 255 nm. (a) Naphthalene. Fit: S ∝ E2-x, x ) 0.06. (b) Benzo[a]pyrene. Linear fit using points up to 35 µJ.
spectra were normalized by dividing them by the energy of the UV pulses. This is completely justified for PAHs showing a linear energy dependence of the signal, i.e., for most of them. It is considered a good enough approximation for naphthalene and acenaphthene as long as the energies are kept in the well-defined window between 25 and 30 µJ, a range that allows sensitive detection of PAHs without leading to fragmentation. A strict data analysis would require the measurement of the power dependence at each wavelength. Although this may be of spectroscopic interest, it is definitely not an important step toward our goal of chemically analyzing complex mixtures. B. 1+1 REMPI Spectra. The 1+1 REMPI spectra were measured for all 17 PAHs using two mixture solutions, one containing eight analytes and the other one 10 analytes (Figure 3). Isomers were not present in the same solution. Coronene was present in both solutions. The data acquisition procedure was the following one: (i) Measure eight single shot mass spectra at each wavelength. (ii) Normalize the single mass spectra by the energies of the corresponding UV pulse as described above. (iii) For each wavelength, average the peak heights of the eight mass spectra. (iv) Normalize the 1+1 REMPI spectra obtained for each compound by the number of moles calculated to be present in an ablation crater. This final step allows the direct comparison of the spectra obtained for different analytes. The units used on the ordinate of the 1+1 REMPI spectra are mV/ (pmol‚µJ). Precision. Errors bars were not drawn on the graphs of Figure 3 in order not to overload them. Typical 95% confidence intervals are in the range of (15%, less in the case of strong signals and more in the case of weak signals. The scattering of the points is
Figure 3. 1+1 REMPI spectra for 17 PAHs. Each point is the average of eight measurements. The normalized units on the ordinates (signal/ energy‚ablated amount) are mV/(µJ‚pmol). For example, 500 mV, 25 µJ, and 0.2 pmol yields 100 normalized units. Coronene was measured twice.
mainly due to uncompensated pulse-to-pulse fluctuations of the IR laser. Another reason is a slight shift of the UV beam position when the wavelength is changed. The membrane is considered homogeneous enough not to cause scattering.27 These experimental uncertainties should be kept in mind when interpreting the spectra. Any vibronic structure of the bands can probably not be studied. Line Width. The 1+1 REMPI spectra exhibit widths of a few tens of nanometers. Therefore, the wavelength of the maxima is often not clearly defined. Rather, a range a few nanometers wide where efficient ionization takes place can be given. The width is a consequence of the fact that the ablated analyte molecules are not cold. Hager and Wallace reported bands with 0.03-nm halfwidths by cooling the gaseous molecules in a supersonic jet.41 However, this causes significant losses due to dilution. Their detection limit was 0.5 pg; this is much higher than the one reached by L2MS.23,25 Selective Ionization of Isomers. A claim that is often made by users of resonance ionization methods is that the great selectivity (41) Hager, J. W.; Wallace, S. Anal. Chem. 1988, 60, 5-10.
inherent to the process allows to optically pick out single compounds in a complex mixture. While this may be true for compounds with sharp optical lines, e.g., atoms or jet-cooled molecules, the validity of this claim is questionable for REMPI of room-temperature gas-phase molecules. One question we want to address here is to what extent REMPI can be used for selective ionization of isomers. Two cases can be distinguished. (i) Take, for example, the isomer pairs phenanthrene and anthracene (m/z ) 178) and fluoranthene and pyrene (m/z ) 202). One of the two isomers is ionized much more efficiently than the other one, independent of the wavelength; only the ratio of the signals changes. Unambiguous identification and quantitation of such isomers is not really possible. Zare’s group has noted this difficulty before in the case of phenanthrene/anthracene. They simply assigned the peaks at mass 178 in their L2MS spectra to the isomer with the highest ionization efficiency, phenanthrene.23,28,42 (ii) Consider the isomer pairs benz[a]anthracene and chrysene (m/z ) 228) and benzo(42) Zenobi, R.; Zare, R. N. In Advances in Multiphoton Spectroscopy and Processes; Lin, S. H., Ed.; World Scientific: Singapore, 1991; Vol. 7, pp 1-144.
Analytical Chemistry, Vol. 70, No. 13, July 1, 1998
2663
Table 1. Comparison of the Features of the Absorption Spectra in Solution and of the Experimental 1+1 REMPI Spectraa analyte
IP/2 (nm)b
τS in jet (ns)
τS in sol (ns)f
λmax(abs) (nm)g
λmax(L2MS) (nm)h
(λmax)g
276
