MALDI-TOF Mass Spectrometry of Insoluble Giant Polycyclic Aromatic

Sep 6, 2000 - ... prevents their characterization by conventional analytical methods, which require a ..... On the other hand, the problem of finding ...
0 downloads 0 Views 117KB Size
Download PDF
Anal. Chem. 2000, 72, 4591-4597

MALDI-TOF Mass Spectrometry of Insoluble Giant Polycyclic Aromatic Hydrocarbons by a New Method of Sample Preparation Laurence Przybilla, Johann-Diedrich Brand, Kimihiro Yoshimura, Hans Joachim Ra1 der,*,† and Klaus Mu 1 llen*

Max-Planck-Institut fu¨r Polymerforschung, Ackermannweg 10, D-55128 Mainz, Germany

The insolubility of giant polycyclic aromatic hydrocarbons (PAHs) prevents their characterization by conventional analytical methods, which require a solubilization of the analyte. Laser desorption mass spectrometry may be used to analyze insoluble samples but is limited to relatively low molecular weights (∼2000), in the case of PAHs. To overcome this limitation, we applied MALDI-TOF mass spectrometry. Since MALDI sample preparation also requires solubility of analyte and matrix molecules, the sample preparation needed modification. The giant PAHs (>2000 Da) were investigated after using a new sample preparation, consisting of mechanically mixing analyte and matrix without any solubilization procedures. This solvent-free process allows insoluble compounds to be characterized. Furthermore, new organic molecules can be used as a matrix. Indeed, 7,7,8,8-tetracyanoquinodimethane, a new matrix with promising properties, has proven to be particularly suitable for the measurement of PAHs. Thanks to the successful characterization with MALDI-TOF mass spectrometry, the chemical design of giant PAHs, which was hindered until now for a lack of analytical methods, can now continue to develop. The disk-shaped polycyclic aromatic hydrocarbons (PAHs) and their supramolecular motif are the object of intensive research.1 A systematic alteration of the shape and size of these extended aromatic systems allows for investigation of their supramolecular ordering and for establishing a correlation with their electronic properties. In particular, these compounds may help in discovering the missing link between molecularly undefined, macroscopic graphite and large, well-defined polyaromatic hydrocarbons. In contrast to the common building up of large PAHs by thermal treatments,2-5 which usually form a wide variety of products, we have developed a synthetic route allowing the design of welldefined large PAHs.6-12 This route involves first the construction † (E-mail) [email protected] (fax) +49 6131/679 100. (1) Zander, M. Aromatische Kohlenwasserstoffe-Kohlenwasserstoffe und Fullerene; B. G. Teubner: Stuttgart, 1995. (2) Fetzer, J. C.; Biggs, J. C. Polycyclic Aromat. Compd. 1994, 4, 3. (3) Fetzer, J. C. Adv. Chem. Ser. 1988, No. 217, 309. (4) Zander, M.; Friedrichsen, W. Chem. Z. 1991, 115, 360-361. (5) Scott, L. T.; Bratcher, M. S.; Hagen, S. J. Am. Chem. Soc. 1996, 118, 87438744. (6) Mu ¨ ller, M.; Mauermann-Du ¨ ll, H.; Wagner, M.; Enkelmann, V.; Mu ¨ llen, K. Angew. Chem., Int. Ed. Engl. 1995, 34, 1583.

10.1021/ac000372q CCC: $19.00 Published on Web 09/06/2000

© 2000 American Chemical Society

of soluble oligophenylene precursors of well-defined structure with a close spatial arrangement of the phenyl rings similar to the framework of the target molecule. In the second step, these precursors are planarized by intramolecular cyclodehydrogenation, yielding the PAHs. However, the extended polybenzenoid disks that are obtained have a very low solubility, a crucial weakness for their characterization, since the conventional analytical methods require sample preparation from solution. We looked for a way around this problem by applying laser mass spectrometry. Taking advantage of their UV absorption, insoluble PAHs with a limited molecular weight can be characterized by laser desorption (LD) mass spectrometry. LD mass spectrometry,13-16 which was originally developed for surface analysis, can be applied to characterizing insoluble substances, absorbing at the laser wavelength (337 nm in the case of a N2 laser). However, the packing of the PAHs in the crystalline state17,18 is expected to induce a high lattice energy of the sample. For LD measurements of larger PAHs, a higher laser fluence is required for the analysis and generates fragmentation in the mass spectrometer. In a previous publication, we described a sample preparation developed for the measurement of PAHs with LD,19 which reduces the intermolecular interaction by breaking the crystal lattice. It allows one to use lower laser fluence for desorption of PAHs and thereby to avoid fragmentation of the analyte. Nevertheless, LD experiments on organic ions reveal an upper limit to the size of molecules that (7) Stabel, A.; Herwig, P.; Mu ¨ llen, K.; Rabe, J. P. Angew. Chem., Int. Ed. Engl. 1995, 34, 1609. (8) Morgenroth, F.; Reuther, E.; Mu ¨ llen, K. Angew. Chem., Int. Ed. Engl. 1997, 36, 631-634. (9) Morgenroth, F.; Ku ¨ bel, C.; Mu ¨ ller, M.; Wiesler, U. M.; Berresheim, A. J.; Wagner, M.; Mu ¨ llen, K. Carbon 1998, 36, 833-837. (10) Iyer, V. S.; Wehmeier, M.; Brand, J. D.; Keegstra, M. A.; Mu ¨ llen, K. Angew. Chem., Int. Ed. Engl. 1997, 36, 1604-1607. (11) Mu ¨ ller, M.; Iyer, V. S.; Ku ¨ bel, C.; Enkelmann, V.; Mu ¨ llen, K. Angew. Chem. Int. Ed. Engl. 1997, 109, 1607-1610. (12) Mu ¨ ller, M.; Ku ¨ bel, C.; Mu ¨ llen, K. Chem. Eur. J. 1998, 4, 2099-2109. (13) Cotter, R. J. Anal. Chim. Acta 1987, 195, 45-59. (14) Constantin, F.; Schnell, A. Mass Spectrometry; Ellis Horwood Ltd.: Chichester, U.K., 1991. (15) Nuwaysir, L.; Wilkins, C. L. Anal. Chem. 1988, 60, 279-282. (16) Nuwaysir, L.; Wilkins, C. L.; Simonsick, W. J., Jr. J. Am. Soc. Mass Spectrom. 1990, 1, 66-71. (17) Gavezzotti, A.; Desiraju, G. R. Acta Crystallogr. 1988, B44, 427-434. (18) Desiraju, G. R.; Gavezzotti, A. Acta Crystallogr. 1989, B45, 473-482. (19) Yoshimura, K.; Przybilla, L. M.; Brand, J.-D.; Wehmeier, M.; Ito, S.; Ra¨der, H.-J.; Mu ¨ llen, K. Macromol. Chem. Phys., in press.

Analytical Chemistry, Vol. 72, No. 19, October 1, 2000 4591

can be desorbed as intact ions.20 This limit depends on molecular structure. For instance, biopolymers have a limit of 1000 Da20 and PAHs have a limit of ∼2000 Da.19 Above this limit, the laser fluence required for the desorption of the molecule is sufficient to induce fragmentation of the analyte, complicating or even hindering the interpretation of the mass spectrum. For the characterization of very large PAHs, even LD with an adapted sample preparation can no longer be applied, thereby leaving these molecules without any analytical method and blocking the development of their chemistry. A solution could be found by applying the matrix-assisted laser desorption/ionization (MALDI) method to these insoluble PAHs. In the case of soluble organic compounds, MALDI has been introduced to circumvent the fragmentation problems induced by LD.21,22 It requires high dilution and distribution of analyte molecules in an absorbing matrix.23 Until recently, this sample preparation was performed exclusively in solution. However, a mechanical distribution of analyte in an excess of matrix without using any solvent might be sufficient to achieve the MALDI process and would enable the characterization of insoluble samples. In this paper, we present, for the first time, a dry sample preparation for MALDI-TOF experiments and a solution to the analytical problem of insoluble PAHs. EXPERIMENTAL SECTION Materials. 1,8,9-Trihydroxyanthracene (dithranol), 3,β-indoleacrylic acid (IAA), R-cyano-4-hydroxycinnamic acid (CCA), 9-nitroanthracene, 2,5-dihydroxybenzoic acid (DHB), and 5-chlorosalicylic acid (CSA) were obtained from Aldrich (Steinheim, Germany). 7,7,8,8-Tetracyanoquinodimethane (TCNQ) was obtained from Merck (Darmstadt, Germany). all-trans-Retinoic acid and 2,3,5,6-tetrafluorotetracyanoquinodimethane (TCNQF4) were purchased from Fluka (Buchs, Switzerland). The measured PAHs were synthesized in our group and the methods published elsewhere.10,11 MALDI Sample Preparation. Mixing of the analyte and matrix powders was performed using the ball mill MM2000 from F. Kurt Retsch GmbH & Co. KG (Haan, Germany). Successful sample preparation was achieved with 0.5 mg of analyte. Analyte and matrix were first mixed in a 1:50 molar ratio and ground for 10 min. To this mixture, an excess of matrix was then added, bringing the final molar ratio to 1:500. The mixture was further milled for 10 min and then crushed on the sample holder. In the case of dithranol, a mixture with a molar ratio 1:5000 was also prepared. Especially for the preparation with dithranol, the grinding was carried out during cooling with liquid nitrogen, to achieve a better homogenization. To avoid pollution of the ion source due to a powder excess, the mixture can also be deposited on the sample holder as a suspension. The mixture obtained after grinding was suspended in a nonsolvent (e.g., water or cyclohexane in the case of the TCNQ matrix) and sonicated for 5 min in the ultrasonic homogenizer HD2070 from Bandelin Electronic GmbH & Co. KG prior to deposition of 1 µL of the suspension on (20) Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Anal. Chem. 1991, 63, 1193A-1203A. (21) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. (22) Tanaka, K.; Waki, H.; Ido, Y.; Akita, S.; Yoshida, Y.; Yoshida, T. Rapid Commun. Mass Spectrom. 1988, 2, 151-153. (23) Horneffer, V.; Dreisewerd, K.; Lu ¨ demann, H.-C.; Hillenkamp, F.; La¨ge, M.; Strupat, K. Int. J. Mass Spectrom. 1999, 185/186/187, 859-870.

4592

Analytical Chemistry, Vol. 72, No. 19, October 1, 2000

the sample holder. The drop was dried and the sample holder was introduced into the ion source of the mass spectrometer. MALDI-TOF Mass Spectrometer. LD-TOF and MALDI-TOF mass spectra were recorded using a Bruker Reflex II MALDITOF mass spectrometer (Bremen, Germany) equipped with a N2 laser (λ ) 337 nm) operating at a pulse rate of 3 Hz. The ions were accelerated with pulsed ion extraction (PIE design from Bruker) after a delay of 50 ns by a voltage of 28.5 kV. The analyzer was operated in reflectron mode, and the ions were detected using a microchannel plate detector. The mass spectrometer was calibrated prior to measurement with a polystyrene standard of appropriate molecular mass. RESULTS AND DISCUSSION Characterization of PAHs with LD and Its Limitations. Extended PAHs are difficult to characterize because of their insolubility. Laser desorption mass spectrometry, which is compatible with insoluble analytes, can be used because the large polybenzenoid disks absorb at 337 nm, the wavelength of a N2 laser, which is commonly used for MALDI-mass spectrometers. PAHs with a limited molecular weight (below ∼2000) can be indeed characterized with LD.19 Figure 1 shows, for example, LDTOF mass spectra of PAHs produced in high purity, which have been synthesized in our group from the oligophenylene precursors depicted in Scheme 1.7,8,11,24 With increasing molecular weight, the LD mass spectrum shows additional side peaks at lower massto-charge ratios than the expected signal (see Figure 1d of PAH 4). These can be attributed to molecules having one, five, or nine aromatic rings less than PAH 4, which are formed either during the synthesis or by fragmentation in the mass spectrometer. However, the second possibility seems to be more probable since LD reveals an upper limit to the size of molecules that could be desorbed as intact ions.20 Above this limit, the laser fluence required for the desorption of the analyte unavoidably causes its decomposition. This phenomenon can be illustrated by measuring PAH 311 with increasing laser fluence (see Figure 2). Figure 2a shows the LD-TOF mass spectrum of 3 obtained upon irradiation at near-threshold laser fluence. In this condition, an efficient and controllable energy transfer to the sample is achieved, which allows the desorption of the intact ion and avoids its decomposition. However, with increasing laser fluence (Figure 2b and c), the desorption of the analyte is accompanied by fragmentation and coalescence (recombination of fragments with fragments or intact molecules in the ion source), leading to signals that are detected by mass-to-charge ratios below and above the analyte mass (962 g/mol). The mass spectra of 3 at high laser fluence (Figure 2b and c) can be considered as a simulation of the behavior of larger PAHs with a molecular weight out of the application range of the LD method. For such a large PAH, a laser fluence exceeding the fragmentation threshold would be required for its desorption. Although we know that LD experiments are limited by the molecular weight of the analyte, it is nevertheless difficult to predict, in the case of PAHs, at which structuredependent molecular weight limit this appears. Within the scope of designing larger extended PAHs, synthetic experiments have been carried out in order to planarize the (24) Iyer, V. S.; Yoshimura, K.; Enkelmann, V.; Epsch, R.; Rabe, J. P.; Mu ¨ llen, K. Angew. Chem., Int. Ed. Engl. 1998, 37, 2696-2699.

Figure 1. LD mass spectra of PAHs: (a) 1; (b) 2; (c) 3; (d) 4.

Scheme 1. Oxidative Cyclodehydrogenation of Oligophenylene Precursors to Well-Defined PAHs 1-4

polyphenylene precursor 5 into the very large polybenzenoid disk 6 (Scheme 2).10 Considering the LD-TOF mass spectrum (Figure 3) of the sample, it is difficult to evaluate which products were actually obtained from the reaction. The mass spectrum shows many signals below and above the molecular weight of the expected product (2708), which hinder a reliable verification that the target molecule has been obtained. Obviously, the mass of

the target molecule 6 is out of the application range of LD and decomposition of the product takes place during its desorption. Moreover, the somewhat regular distance between the numerous peaks acquired above 1500 Da (Figure 3b), which amounts to ∼24 Da, could indicate that fragmentation and coalescence reactions have taken place. Indeed, an analogy can be established with the fragmentation and coalescence reactions of fullerenes generated by LD experiments, which also reveal a distribution of signals separated by 24 Da (C2 unit).25 MALDI: A Solution for Characterizing Huge PAHs?. Hence, the characterization of this sample as well as even bigger PAHs requires a softer desorption process. In the case of soluble organic compounds, MALDI is a mass spectrometric method achieving a very soft desorption process allowing macromolecules to be detected as intact ions up to few hundred thousand daltons. This method involves high dilution and distribution of analyte molecules in a matrix.23 Although this mixture is introduced into the mass spectrometer in the condensed phase, the MALDI sample preparation previously required the solubilization of the analyte, thereby restricting the application of MALDI to soluble samples. This is true for the most widely used techniques: the so-called “dried-droplet method”,21 as well as for surface preparation,26 crushed crystal,27 or electrospray deposition28,29 techniques. (25) Yeretzian, C.; Hansen, K.; Diederich, F.; Whetten, R. L. Nature 1992, 359, 44-47. (26) Vorm, O.; Roepstorff, P.; Mann, M. Anal. Chem. 1994, 66, 3281-3287. (27) Xiang, F.; Beavis, R. C. Rapid Commun. Mass Spectrom. 1994, 8, 199-204. (28) Axelsson, J.; Hoberg, A.-M.; Waterson, C.; Myarr, P.; Shield, G. L.; Varney, J.; Haddleton, D. M.; Derrick, P. J. Rapid Commun. Mass Spectrom. 1997, 11, 209-213. (29) Hensel, R. R.; King, R. C.; Owens, K. G. Rapid Commun. Mass Spectrom. 1997, 11, 1785-1793.

Analytical Chemistry, Vol. 72, No. 19, October 1, 2000

4593

Figure 3. (a) LD mass spectrum of PAH 6; (b) expanded region.

Figure 2. Influence of increasing laser fluence on LD mass spectra of PAH 3: (a) at near-threshold laser fluence (relative unit, 55); (b) slightly higher laser fluence (60); (c) high laser fluence (64).

Scheme 2. Synthesis of Giant PAH 6 via Cyclodehydrogenation Reaction by Copper(II) Triflate/Aluminum Chloride in Carbon Disulfide

However, a mechanical homogenization of the sample in an excess of matrix, excluding the solubilization of any components of the mixture, could be an alternative for the sample preparation. We composed such a mixture in a ball grinder, reaching a final particle size of roughly 1 µm, according to the specifications of the manufacturer. We tested this new sample preparation for insoluble compounds first with dithranol, which is a versatile matrix for the analysis of synthetic polymers. The mixture was obtained by shaking the analyte and matrix powders for altogether 30 min in a molar ratio of 1:5000. After homogenization, the analyte/matrix 4594 Analytical Chemistry, Vol. 72, No. 19, October 1, 2000

mixture still has a macroscopic powder aspect. We assume that the milling consists of a simple mixture of separate particles rather than a melting of matrix and/or analyte. A small amount of this mixture was then crushed on the target, and Figure 4 shows the corresponding MALDI-TOF mass spectrum. In contrast to the LDTOF mass spectrum (Figure 3), fragmentation and coalescence have been avoided during desorption. A weak signal corresponding to the intact ion of the expected molecule 6 is now detected for the first time as well as signals at higher mass-to-charge ratios, which could be attributed to side products of the reaction (partially cyclized products). To further characterize the sample, the quality (intensity, resolution) of the MALDI mass spectrum must be improved. This should be achieved with a better distribution of the analyte in the matrix powder, and for this reason, we prepared an analogous mixture by cooling during the grinding process with liquid nitrogen. The obtained mixture led to a mass spectrum (see Figure 5) of much higher intensity, even though the resolution is still low. Obviously the homogeneity of the matrix/analyte mixture was increased by cooling with liquid nitrogen during the milling process. It seems to confirm our assumption about the powder nature of the analyte/matrix mixture and excludes melting of any of the components in the ball mill. The mass spectrum obtained demonstrates that matrix assistance has undeniably taken place. The PAH 6 analyzed with MALDI is detected as a radical cation, similarly to smaller PAHs measured with LD. This ionization mode in a MALDI process is observed for analytes absorbing at the laser wavelength30-32 whereas cation attachment or protonation (30) Juhasz, P.; Costello, C. E. Rapid Commun. Mass Spectrom. 1993, 7, 343351. (31) Ra¨der, H. J.; Spickermann, J.; Kreyenschmidt, M.; Mu ¨ llen, K. Macromol. Chem. Phys. 1996, 197, 3285-3296.

Figure 4. (a) MALDI mass spectrum of PAH 6 obtained by applying the dry sample preparation with dithranol as matrix; (b) expanded region.

is typically involved in the ionization of nonabsorbing analytes. According to our interpretation, the transfer of intact analyte molecules into the gas phase is indeed achieved by matrix assistance, but the ionization involves photoionization.31,32 Since the dry sample preparation allows MALDI to be applied without preparing solutions prior to measurement, the solubility of the matrix is also no longer a restriction for the method and a multitude of new organic compounds can be used instead of the standard matrixes. This allows for a significant simplification of sample preparation since it is not necessary to consider compatible matrix/analyte/solvent systems which build up homogeneous solutions in the liquid and finally in the solid state on the sample holder. To improve the MALDI process in the case of PAHs, we aimed at promoting the formation of radical cations by using a matrix having additionally the properties of an electron acceptor. Well-known in the field of organic metals, TCNQ is a strong electron acceptor having an absorption at 337 nm, thereby representing a potential matrix for our problematic molecules. Figure 6 shows the MALDI-TOF mass spectrum obtained by using TCNQ as matrix for our dry sample preparation (without cooling). This spectrum has by far the best quality compared with the mass spectra obtained with usual matrixes. The signal-to-noise ratio is much higher, and isotopic resolution and good spot-to-spot reproducibility are obtained. A detailed interpretation of the spectrum is now possible: the signal of the expected PAH 6 is in agreement with the theoretical isotopic distribution and additional signals above this mass-to-charge ratio can be attributed to partially cyclized and/or chlorinated products which were obtained during the reaction. It is now obvious that the additional signals in the (32) Remmers, M.; Mu ¨ller, B.; Martin, K.; Ra¨der, H.-J.; Ko ¨hler, W. Macromolecules 1999, 32, 1073-1079.

Figure 5. (a) MALDI mass spectrum of PAH 6 obtained by applying the dry sample preparation under cooling (with liquid nitrogen), using dithranol as matrix; (b) expanded region.

mass spectrum of PAH 6 are due exclusively to side products of the reaction rather than produced during the measurement itself, in contrast to the situation described for the LD measurement of sample 6 in Figure 3. This can be proven by different observations: (i) the mass spectra show no significant change of relative signal intensities over a broad range of laser fluence as is known in usual MALDI-experiments, (ii) the additional signals can be explained by side reactions during synthesis rather than by fragmentation or coalescence phenomena during analysis, and (iii) the additional signals change depending on the chemical reaction condition. Since MALDI-TOF mass spectrometry is now able to analyze these insoluble PAHs, the different reaction conditions (such as choice of the oxidant, Lewis acid, reaction time, and temperature) can be further optimized. The most important aim of the synthetic work is to minimize the amount of side products as there is no means to separate the insoluble target molecule from the side products, which are insoluble as well. It is necessary to recall that no quantification between the amount of the target molecule and that of side products can be guaranteed by MALDI, since their desorption and ionization probabilities may be quite different. To estimate the proportion of side products in the sample, we mixed a sample containing mainly the target molecule 6swhich is until now the sample with the highest degree of purity (see Figure 7a)swith a sample containing only side products (see Figure 7b) in a defined mass ratio. The sample containing only side products was obtained by milder oxidation conditions. Mixing these two samples in a mass ratio 1:1 and applying our dry sample preparation leads to the MALDI-TOF mass spectrum shown in Figure 7c. Despite a mixing ratio 1:1 by weight, the peak area of the target molecule reaches hardly 0.4% of the peak area of the mixture. This demonstrates that under our MALDI conditions side Analytical Chemistry, Vol. 72, No. 19, October 1, 2000

4595

Figure 6. (a) MALDI mass spectrum of PAH 6 obtained by applying the dry sample preparation with TCNQ as matrix; (b) expanded region.

products are overestimated against the target molecule (by a factor of 285!). Considering the purest sample obtained until now (see Figure 7a) and dividing the intensity of the side products by the factor 285 would lead to an amount of side products in the sample of only 1% instead of 63%. This mass spectrometric approach of quantification is, of course, only an estimation for the purity of the sample. From the analytical point of view, it should be emphasized that this synthetic progress is only feasible owing to the dry sample preparation elaborated for MALDI-TOF mass spectrometry, enabling the characterization of these otherwise “uncharacterizable” structures. Comments on the Dry Sample Preparation with TCNQ as Matrix. It should be noted that with this dry sample preparation there is a risk of contaminating the ion source of the mass spectrometer, since the analyte/matrix mixture is only crushed on the target. To avoid this pollution, it is possible to prepare a suspension of this mixture in a nonsolvent and to homogenize it for 5 min with an ultrasonic homogenizer. A small amount (e.g., 1 µL) of this suspension is then deposited on the target, and after drying, a thin layer is obtained with sufficient adhesion at the target surface. For this procedure, it is essential to choose a nonsolvent to obtain a successful preparation. From a suspension in water or in cyclohexane, in which TCNQ is not soluble, the obtained MALDI mass spectra have a quality similar to that from the crushed mixture. In contrast, when THF or acetone (which both solubilize TCNQ) is used, absolutely no signal can be obtained for the analyte. Obviously, the insoluble analyte has been separated from the TCNQ crystals during evaporation of solvent, thereby binding the MALDI process to fail. 4596 Analytical Chemistry, Vol. 72, No. 19, October 1, 2000

Figure 7. MALDI mass spectra obtained by applying the dry sample preparation with TCNQ as matrix to (a) sample containing a significant amount of target molecule 6 (cyclodehydrogenation reaction with copper(II) triflate/aluminum chloride); (b) sample containing only side products (cyclodehydrogenation reaction with iron(III) chloride); (c) mixing of the two preceding samples in a mass ratio 1:1.

This new sample preparation, achieved very easily with TCNQ as a matrix, is certainly a breakthrough in the characterization of huge PAHs, which previously could not be analyzed at all. To compare TCNQ with other standard matrixes, we applied the same sample preparation (without cooling) with all-trans-retinoic acid, IAA, CCA, DHB, and CSA as matrixes. In the case of all-transretinoic acid, IAA, 9-nitroanthracene, and DHB, the MALDI mass spectra obtained were comparable with that generated with dithranol (Figure 5), that is to say, of much poorer quality (intensity, resolution) than with TCNQ. Using CSA as a matrix, the mass spectrum acquired was analogous to the LD mass spectrum, already shown in Figure 3. This result can be explained by the relatively high laser power required for the desorption of CSA, which induces fragmentation and coalescence of the analyte. Finally, among the usual matrixes tested, only CCA gave a MALDI mass spectrum (not shown here for reasons of shortness) with a quality approaching that obtained with TCNQ (even if the intensity and the resolution are slightly lower). Considering its structural formula, we can actually credit CCA to act as an electron acceptor

and promote the formation of the analyte radical cation. In fact, the cyano and carboxy substituents of CCA are electron-withdrawing groups which should stabilize a radical anion formed by reduction of CCA. Considering the influence of the electronacceptor property of the matrix on the ionization of PAHs, we tested TCNQF4, which is an even stronger electron acceptor than TCNQ, as a matrix. Unexpectedly, the signal of the target molecule had a lower intensity than that measured with TCNQ (mass spectrum with TCNQF4 not shown here for reasons of shortness). Aware that by changing a matrix other influential parameters may be simultaneously modified, it is difficult to evaluate the importance of the electron-acceptor property of the matrix on the measurement of the PAHs. Nevertheless, for the characterization of PAHs with MALDI mass spectrometry, TCNQ has proven to be the most suitable one, among all the tested matrixes. Indeed, this new matrix exhibits some interesting properties that could be profitable for the characterization of other compounds, which tends to fragmentation during the MALDI process. TCNQ has a low desorption threshold, corresponding to that of all-trans-retinoic acid, which has the lowest desorption threshold of all our standard matrixes.33 Moreover, at nearthreshold laser fluence, the number and the intensity of the matrix peaks (peaks in the low-mass region due to ions derived from the matrix) are low; therefore, the ion cloud in the source of the mass spectrometer should have a low density. Hence, the ion collisions in the source should be reduced and the probability for fragmentation should decrease. Furthermore, owing to its electron-acceptor property, we expect TCNQ to promote the formation of radical cations, hence being well adapted for the characterization of analytes absorbing at 337 nm (N2 laser wavelength) and forming radical cations as PAHs do. CONCLUSION In this article, we presented a new sample preparation allowing large, insoluble PAHs to be characterized with MALDI-TOF mass spectrometry for the first time. It consists of mechanically mixing the analyte and matrix powders, thereby avoiding any solubilization procedure. Using TCNQ as a new matrix, we obtained MALDI-TOF mass spectra of extended PAHs with very good (33) Przybilla, L. M., Ph.D. Thesis, Johannes Gutenberg-University, Mainz, 2000.

signal-to-noise ratio, resolution, and spot-to-spot reproducibility. According to a semiquantitative approach developed to estimate the purity of PAH 6, the signals of side products seem to be extremely overestimated in comparison to the signal of the target molecule. Consequently, the obtained product is expected to have a much higher degree of purity than implied by the uncorrected signal intensities in the MALDI mass spectra. Thanks to this new sample preparation, the chemical design of very large PAHs, which was previously hindered for lack of characterization methods, can continue to develop. This study proves that the solubility of the analyte as well as that of the matrix is no longer a limiting factor for MALDI measurements. This means, on one hand, that insoluble compounds also can be now characterized in principle with MALDI mass spectrometry. On the other hand, the problem of finding a suitable sovent system for both matrix and analyte is also eliminated for soluble samples. Hence, there are less limitations for the choice of appropriate matrixes. We introduced, for instance, TCNQ as a new matrix, which shows excellent results for the analysis of PAHs. We speculate that this success may be in part due to the fact that TCNQ not only assists the desorption of the intact PAHs but also supports their ionization by promoting the formation of radical cations. Owing to its electron-acceptor property as well as to its low desorption threshold, TCNQ is promising also for other classes of compounds that particularly require a soft MALDI process. This can be expected for analytes having a strong absorption at 337 nm (laser wavelength of commercial MALDI mass spectrometers), which form radical cations and have a tendency to fragment. ACKNOWLEDGMENT Financial support of the Bundesministerium fu¨r Bildung und Forschung (BMBF Project 03N6010A) is gratefully acknowledged. The authors also acknowledge Sabine Keune (Th. Goldschmidt AG) for encouraging them in applying a dry sample preparation for MALDI mass spectrometry.

Received for review March 29, 2000. Accepted July 5, 2000. AC000372Q

Analytical Chemistry, Vol. 72, No. 19, October 1, 2000

4597