Anal. Chem. 2003, 75, 4624-4630
Mechanistic Studies of Photochemical Reactions with Millisecond Time Resolution by Electrospray Ionization Mass Spectrometry Wei Ding,† Keith A. Johnson,‡ Charles Kutal,† and I. Jonathan Amster*,†
Department of Chemistry, University of Georgia, Athens, Georgia 30602, and Wyeth BioPharm, Andover, Massachusetts 01810
Electrospray ionization mass spectrometry is used as an on-line technique to investigate the solution photochemistry of [CpFebz]+ (Cp, η5-cyclopentadienyl; bz, η6-benzene). Direct irradiation of samples in the optically transparent tip of the electrospray source allows the detection of products with lifetimes down to the millisecond regime. Photolysis of [CpFebz]+ in acetonitrile (AN) yields the short-lived half-sandwich complex, [CpFe(AN)3]+. The suspected parent-offspring relationship between this species and the fully ring-deligated product, [Fe(AN)6]2+, has been confirmed. In solutions containing cyclohexene oxide, [CpFebz]+ generates several photoproducts capable of initiating the cationic polymerization of the epoxide monomer. The initiation process and the first few steps in the subsequent growth of the polymer chain have been observed. Electrospray ionization mass spectrometry (ESI-MS) is a convenient technique for transferring ions from the liquid phase to the gas phase for mass analysis.1,2 This transfer occurs under relatively mild conditions, which minimize the fragmentation of analyte ions and simplify the interpretation of the mass spectrum. The technique has proven to be useful for the analysis of thermally sensitive and nonvolatile substances and, for this reason, has been employed widely for the characterization of biomolecules,3 synthetic oligomers and polymers,4-7 and inorganic and organometallic complexes.8-11 ESI-MS also has been used for the on-line * Corresponding author: (e-mail)
[email protected]; (Fax) (706) 542-9454. † University of Georgia. ‡ Wyeth BioPharm. (1) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451-4459. (2) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4671-4675. (3) Griffiths, W. J.; Jonnson, A. P.; Liu, S.; Rai, D. K.; Wang Y. Biochem. J. 2001, 355, 545-561. (4) Maziarz, E. P., III; Baker, G. A.; Mure, J. V.; Wood, T. D. Int. J. Mass Spectrom. 2000, 202, 241-250. (5) McEwen, C. N.; Simonsick, W. J., Jr.; Laesen, B. S. Am. Soc. Mass Spectrom. 1995, 6, 906-911. (6) Koster, S.; Duuesma, M. C.; Boon, J. J.; Heeren, R. M. A. Am. Soc. Mass Spectrom. 2000, 11, 536-543. (7) Shan, L.; Murgasova, R.; Hercules, D. M.; Houalla, M. J. Mass Spectrom. 2001, 36, 140-144. (8) Løver, T.; Henderson, W.; Bowmaker, G. A.; Seakins, J. M.; Cooney, R. P. J. Mater. Chem. 1997, 7, 1553-1558. (9) Kane-Maguire, L. A. P.; Kanitz, R.; Sheil, M. M. J. Organomet. Chem. 1995, 486, 243-248.
4624 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
monitoring of thermal, electrochemical, and photochemical reactions in real time.12-24 This application has yielded valuable information about reaction kinetics, reaction stoichiometry, and the identities of intermediates. Typically, the ESI source is coupled directly to a reactor, from which the various species produced are transported to the electrospray tip end for spraying. To be detectable, however, a species must complete this journey to the tip end in a time, denoted here as the transit time, tr, that is not significantly longer than its solution lifetime, τ. For the majority of reported studies, transit times of several seconds to a few minutes have been achieved. For example, Arakawa et al. studied the photosubstitution and photooxidation reactions of Ru(II) complexes19-22 by directly irradiating a sample solution passing through a quartz photolysis cell located in the middle of the ESI spray tip and successfully detected intermediates with lifetimes of more than a few minutes. Similar time resolution was obtained by Brum and Dell’Orco, who monitored the photolysis of idoxifene in a jacketed reactor by ESI-MS.23 Volmer investigated the photochemical behavior of pesticides in a reactor coil coupled to an ESI-MS system and could detect species with lifetimes on the order of 30 s.24 We report here an on-line ESI-MS technique for studying photochemical reactions that greatly reduces the transit time of (10) Traeger, J. C. Int. J. Mass Spectrom. 2000, 200, 387-401. (11) Henderson, W.; Nicholson, B. K.; McCaffrey L. J. Polyhedron 1998, 17, 4291-4313. (12) Espenson, H. E.; Tan, H.; Mollah, S.; Houk, R. S.; Eager, M. D. Inorg. Chem. 1998, 37, 4621-4624. (13) Bakhtiar, R.; Hop, C. E. C. A. J. Phys. Org. Chem. 1999, 12, 511-527. (14) Arakawa R.; Liu, J.; Mizuno, K.; Inoue, H.; Doe H.; Matsuo, T. Int. J. Mass Spectrom. 1997, 160, 371-376. (15) Aliprantis, A. O.; Canary, J. W. J. Am. Chem. Soc. 1994, 116, 6985-6986. (16) Brum, J.; Dell’Orco, P.; Lapka, S.; Muske, K.; Sisko, J. Rapid Commun. Mass Spectrom. 2001, 15, 1548-1553. (17) Zechel, D. L.; Konermann, L.; Withers, S. G.; Dounglas, D. J. Biochemistry 1998, 37, 7664-7669. (18) Xu, X.; Lu, W.; Cole, R. B. Anal. Chem. 1996, 68, 4244-4253. (19) Arakawa, R.; Jian, L.; Yoshimura, A.; Nozaki, K.; Ohno, T.; Doe, H.; Matsuo, T. Inorg. Chem. 1995, 34, 3874-3878. (20) Arakawa, R.; Tachiyashiki, S.; Matsuo, T. Anal. Chem. 1995, 67, 41334138. (21) Arakawa, R.; Mimura, S.; Mastsubayashi, G.; Matsuo, T. Inorg. Chem. 1996, 35, 5725-5729. (22) Arakawa, R.; Matsuda, F.; Mastsubayashi, G. Am. Soc. Mass Spectrom. 1997, 8, 713-717. (23) Brum, J.; Dell’Orco, P. Rapid Commun. Mass Spectrom. 1998, 12, 741745. (24) Volmer, D. A. J. Chromatogr., A 1998, 794, 129-146. 10.1021/ac0264269 CCC: $25.00
© 2003 American Chemical Society Published on Web 07/30/2003
room temperature and rapidly decomposes to produce the fully ring-deligated metal ion and ferrocene (eq 2). Much less certain is the identity of the photogenerated species actually responsible for initiating polymerization in systems containing a monomer. Most previous studies assign this role to [CpFe(monomer)3]+, but neither it nor any polymeric products containing the [CpFe]+ unit have been directly observed. Accordingly, we undertook the present study with the goal of developing an analytical methodology that would allow us to do the following: (1) completely characterize the solution photochemistry of [CpFebz]+ (bz, η6-benzene; structure shown below), Figure 1. Schematic diagram of the nanospray tip of the ESI source. The distance, D, is measured from the midpoint of the irradiated zone to the tip end.
photogenerated species. As shown in Figure 1, sample solutions are irradiated directly in the optically transparent nanospray tip of the ESI source. Subsequent thermal reactions of the primary photoproducts take place in the region between the photolysis zone and the tip end. The transit time of a photoproduct depends on the volumetric flow rate of the sample, the inner diameter of the tip, and D, the distance between the midpoint of the irradiated zone and the tip end. For example, with D ) 0.84 mm, a tip diameter of 40 µm, and a flow rate of 40 µL/h, products require 95 ms to arrive at the tip end for spraying. All chemical reactions are quickly quenched (µs time scale) once the sample solution leaves the tip owing to rapid desolvation of the solutes that occurs during the electrospraying process. Consequently, ionic species with solution lifetimes in the millisecond range or longer should be detectable by this technique. In this report, the new ESI-MS technique is used to characterize the solution photochemistry of a member of the [CpFe(η6-arene)]+ family (Cp, η5- cyclopentadienyl). Practical interest in these mixed-ring sandwich complexes arises from their application as panchromatic photoinitiators for the polymerization of epoxides25-27 and other monomers.28,29 Seminal studies by Mann et al.30-32 and by Schuster et al.27,33 have established that irradiation of the complexes in the wavelength region of their ligand field absorption bands induces loss of arene with the concomitant formation of [CpFe(L)3]+ (eq 1), where L can be solvent or any hν
CpFe(η6-arene)+ 9 8 CpFe(L)3+ + arene L
(1)
2CpFe(L)3+ f Fe(L)62+ + FeCp2
(2)
other potential ligand present in solution. When L is weakly coordinating, the half-sandwich complex is thermally unstable at (25) Roloff, A.; Meier, K.; Reideker, M. Pure Appl. Chem. 1986, 58, 1267-1272. (26) Lohse, F.; Zweifel, H. Adv. Polym. Sci. 1986, 78, 61-80. (27) Park, K. M.; Schuster, G. B. J. Organomet. Chem. 1991, 402, 355-362. (28) Rabek, J. F.; Lucki, J.; Zuber, M.; Qu, B. J.; Shi, W. F. J. Macromol. Sci. Appl. Chem. 1992, A29, 297-303. (29) Kotch, T. G.; Lees, A. J.; Fuerniss, S. J.; Papathomas, K. I. Chem. Mater. 1995, 7, 801-805. (30) Gill, T. P.; Mann, K. R. Inorg. Chem. 1983, 22, 1986-1991. (31) McNair, A. M.; Schrenk, J. L.; Mann, K. R. Inorg. Chem. 1984, 23, 26332640. (32) Boyd, D. C.; Bohling, D. A.; Mann K. R. J. Am. Chem. Soc. 1985, 107, 1641-1644. (33) Chrisope, D. R.; Park, K. M.; Schuster, G. B. J. Am. Chem. Soc. 1989, 111, 6195-6201.
including the unambiguous identification of short-lived intermediates; (2) assess the role, if any, of [CpFe(monomer)3]+ in initiating the cationic polymerization of an epoxide; (3) monitor the first few propagation steps in the polymerization process. Some of our initial results have been reported in a recent communication.34 EXPERIMENTAL SECTION Reagents. Literature procedures were followed in the preparation of [CpFebz]PF6.35-37 The crude product was purified by column chromatography on silica gel, with acetonitrile/dichloromethane solvent mixtures (0:100-20:80 v/v) as the eluting media. Anal. Calcd for C11H11F6P1Fe: C, 38.40; H, 3.22. Found: C, 38.11; H, 3.24. 1H NMR [CD3C(O)CD3]: δ 6.22 (s, 6H, C6H6), 5.02 (s, 5H, C5H5). Complexes containing deuterated Cp or bz ligands were prepared by the same procedures using the appropriately labeled starting material (Fe(C5D5)2 or C6D6, respectively). Acetonitrile (HPLC grade from Baker) and dichloroethane (ACS grade from Fisher) were dried over CaH2 and then fractionally distilled. Cyclohexene oxide (98% from Aldrich) was distilled twice under reduced pressure through a bead-packed fractionating column. Instrumentation. Electrospray ionization mass spectra were recorded in the positive ion mode using a Mariner Biospectrometry Workstation (Applied Biosystems), an apparatus that combines a time-of-flight mass spectrometer with an orthogonal electrospray ion source. Typical operating conditions for these experiments were as follows: spray tip potential, 1.9 kV; nozzle potential, 60 V; skimmer 1 potential, 11.5 V; nozzle temperature, 150 °C; quadrupole temperature, 90-150 °C. The sample solution emerged from a nanospray tip (Figure 1) fabricated in-house from fused-silica capillary tubing (Supelco, 100-µm i. d.). The tip was (34) Ding, W.; Johnson, K. A.; Amster, I. J.; Kutal, C. Inorg. Chem. 2001, 40, 6865-6866. (35) Dabirmanesh, Q.; Roberts, R. M. G. J. Organomet. Chem. 1993, 460, C28C29. (36) Dabirmanesh, Q.; Fernando, S. I. S.; Roberts, R. M. G. J. Chem. Soc., Perkin Trans. 1 1995, 743-749. (37) King, R. B. Organomet. Synth. 1965, 1, 138-139.
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4625
Figure 2. Schematic diagram of the ESI-MS apparatus used for the on-line photochemical experiments.
drawn to a fine point at the spray delivery end by heating the tubing with a microtorch (Microflame, Inc.) while applying a constant force. The polymer coating on the tubing was burned away with the flame, providing an optically transparent region for irradiation that was ∼20 mm in length. Sample solutions were delivered to the other end of the capillary tubing via a 50-µL syringe and a syringe pump (model 55-111, Harvard Apparatus). Flow rates of 10-80 µL/h were required to maintain a stable spray from the tip. Cross-sectional images of each working nanospray tip and the original capillary tubing were taken with a Hitachi KP-D50 digital camera mounted on a Jenavert optical microscope. Feature sizes were calibrated against the manufacturer-quoted outer diameter (300 ( 15 µm) of the capillary tubing. Using this approach, we determined that the inner diameters of the nanospray tips used in our experiments ranged from 20 to 50 µm. Lateral images of the tips revealed that the diameter varied less than 10% along the ∼10-mm irradiation zone (defined by D in Figure 1). Figure 2 schematically illustrates the ESI-MS apparatus used for the on-line photochemical experiments. The 488-nm output of an Innova 70A argon ion laser (Coherent Radiation) was directed onto the nanospray tip via a 0.3-mm-diameter optical fiber (quartz core, Liteway, Inc.). The diameter of the light beam striking the tip was 0.6 mm. Laser power was varied from 160 to 400 mW; the observation of the same photoproducts at different laser powers establishes that multiphoton excitation and secondary photolysis processes were unimportant. The distance D (Figure 1) was varied by adjusting the position of the optical fiber with a precision x,y,z translation stage. Transit times calculated for various combinations of D and volumetric flow rates for a 40-µm tip are compiled in Table 1. These results indicate that the detection of ionic photoproducts with solution lifetimes down to 10-20 ms should be feasible with this ESI-MS technique. 4626
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
Table 1. Estimated Transit Times (in Milliseconds) of Photoproducts in a 40-µm-i.d. Nanospray Tip as a Function of the Sample Solution Flow Rate (ν) and the Distance from the Midpoint of the Irradiated Zone to the Tip End (D) D (mm)
ν ) 10 µL/h
ν ) 20 µL/h
ν ) 40 µL/h
ν ) 80 µL/h
0.5 1.0 2.0 4.0 8.0
226 452 900 1800 3600
113 226 452 900 1800
56 113 226 452 900
28 56 113 226 452
RESULTS AND DISCUSSION Optimizing ESI Conditions. Transition metal complexes can undergo several reactions during the electrospray process that complicate the interpretation of the resulting mass spectrum. Examples include metal-ligand bond cleavage, oxidation, and adduct formation with protons or other cations present in solution.11,20,38,39 To minimize the occurrence of these processes for [CpFebz]+, we examined the effect of various instrumental parameters on the parent ion signal. Figure 3 shows the effects of varying the nozzle and skimmer 1 potentials. Setting these potentials at 120 and 12.2 V, respectively, results in a clean mass spectrum of the complex (spectrum a). Higher potentials, particularly at skimmer 1, facilitate cleavage of the Fe-benzene bond to yield an abundant [FeCp]+ peak (spectra b-d). We attribute the formation of this highly coordinatively unsaturated fragment to collision-induced dissociation (CID) of the parent complex in the quadrupole region located downstream of skimmer 1.40 For (38) Kane-Maguire, L. A. P.; Kanitz, R.; Sheil, M. M. Inorg. Chim. Acta 1996, 245, 209-214. (39) Berkel, G. J. V.; Giles, G. E.; Bullock, J. S.; Gray, L. J. Anal. Chem. 1999, 71, 5288-5296.
Figure 3. ESI mass spectrum of a 21 µM solution of [CpFebz]+ in acetonitrile obtained at a spray tip potential of 1900 V and various nozzle and skimmer 1 potentials: (a) nozzle, 120 V; skimmer 1, 12.2 V; (b) nozzle, 250 V; skimmer 1, 18 V; (c) nozzle, 250 V; skimmer 1, 35 V; (d) nozzle, 260 V; skimmer 1, 45 V.
the experiments reported in this study, the potentials at the nanospray tip, nozzle, and skimmer 1 were set at levels (see Experimental Section) that minimize possible thermal and electrochemical reactions of [CpFebz]+ during the electrospray process. We also examined the influence of the inner diameter of the nanospray tip on the volumetric flow rate of the sample solution. For unirradiated samples, the measured flow rate agreed closely with the value programmed into the syringe pump over the entire range of tip diameters (20-50 µm) tested in this study. The situation was more complex, however, for irradiated samples. For tip diameters below ∼30 µm, the measured flow rate was less than the programmed value, suggesting that some clogging of the tip, presumably by some insoluble photoproduct, had occurred. This problem could be avoided by using tips with an inner diameter of g34 µm, for which good agreement once again was obtained between measured and programmed flow rates. All experiments involving the determination of the solution lifetimes of photogenerated species were conducted with these larger diameter tips. Solution Photochemistry of [CpFebz]+. Previous work has shown that the photolysis of [CpFe(η6-arene)]+ in acetonitrile (AN) at -40 °C results in the release of arene accompanied by formation of the half-sandwich complex [CpFe(AN)3]+ (eq 1), a purple intermediate stable for several hours at that temperature. The intermediate was characterized at low temperature by 1H NMR, electronic spectroscopy, and cyclic voltammetry.30,32 When the photolyzed solution is warmed to 0 °C, the intermediate decom(40) Reference 20 provides another example of CID loss of AN from a dipositive transition metal ion.
Figure 4. ESI mass spectrum of the products obtained upon 488nm photolysis of 70 µM [CpFebz]+ in acetonitrile; ttr ) 90 ms.
poses to yield ferrocene and [Fe(AN)6]2+ (eq 2). The same reaction sequence should occur at room temperature, but with [CpFe(AN)3]+ exhibiting a much shorter lifetime. Our ESI-MS technique has allowed us to investigate these processes in real time and search directly for any short-lived intermediates. Irradiation of an acetonitrile solution of [CpFebz]+ in the nanospray tip of the ESI source yields two major series of ionic products: [CpFe(AN)2-3]+ and [Fe(AN)3-6]2+ (Figure 4). The first series results from the photoinduced release of benzene from the parent complex (eq 1), while the second series reflects the subsequent thermal decomposition of the half-sandwich product Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4627
Figure 5. Relative abundance versus ttr for two series of ions in a photolyzed solution of 70 µM [CpFebz]+ in acetonitrile: (b) ∑{I [CpFe(AN)m]+}/[∑{I [Fe(AN)n]2+} + ∑{I [CpFe(AN)m]+}]; (2) ∑{I [Fe(AN)n]2+}/ [∑{I [Fe(AN)n]2+} + ∑{I [CpFe(AN)m]+}]. The symbol I represents the abundance of an ionic species as measured by the intensity of its mass spectral peak, and the summation extends over all members of a series. The relationship between ttr and D can be seen by comparison of the two horizontal axes.
(eq 2). It is important to note the absence of species containing more than six direct metal-ligand bonds (Cp is typically a tridentate ligand). This result suggests that the products are “normal” complexes of divalent iron, which typically favors a coordination number of six or less. Products containing less than six metal-ligand bonds (e.g., five-coordinate [CpFe(AN)2]+ and three-coordinate [Fe(AN)3]2+) result from collision-induced dissociation of acetonitrile from [CpFe(AN)3]+ and [Fe(AN)6]2+. In support of this explanation, we observe that increasing the kinetic energy of the electrosprayed ions by raising the skimmer 1 voltage shifts the distribution within each product series toward species containing fewer coordinated AN molecules. The parent-offspring relationship between [CpFe(AN)3]+ and [Fe(AN)6]2+ (eqs 1 and 2) is clearly established by the effect of the transit time, ttr, on the relative abundance of the two product series. When ttr, ) 90 ms, a significant amount of the former complex survives passage to the tip end, where it is sprayed and subsequently detected by mass spectrometry (Figure 4). Increasing ttr to 230 ms allows for nearly complete thermal decomposition of [CpFe(AN)3]+, and only signals from members of the [Fe(AN)6]2+ series are evident in the mass spectrum. Figure 5 depicts in graphical form the results of a detailed study of the relationship between ttr and the relative abundance of the parent and offspring ions. The disappearance of the [CpFe(AN)m]+ series can be fit by a simple exponential decay, consistent with first-order kinetics for the reaction.30 From these data, we calculate a lifetime (reciprocal of the first-order rate constant derived from the decay plot in Figure 5) of 95 ms for [CpFe(AN)3]+ in room-temperature acetonitrile. Photoinitiated Polymerization Studies. Photolyzing [CpFe(η6-arene)]+ complexes in the presence of an epoxide monomer generates one or more species capable of initiating cationic polymerization.25-27 Carrying out this photoinitiated chemistry in the nanospray tip of our ESI-MS apparatus allows us to investigate 4628 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
Figure 6. ESI mass spectrum of the products obtained upon 488nm photolysis of a 1,2-dichloroethane solution containing 100 µM [CpFebz]+ and 40 mM CHO. ttr ) 50 ms. Representative members of several different product series are labeled.
Figure 7. Relative abundance versus ttr for three series of ions in a photolyzed solution of 100 µM [CpFebz]+ and 40 mM CHO in 1,2dichloroethane: (b) ∑{I[∑{I [CpFe(H2O)(CHO)i]2+}/∑{all ions}, where ∑{all ions} ) [∑{I Fe(H2O)n(CHO)j]2+} + ∑{I [CpFe(H2O)(CHO)i]+} + ∑{I [X(CHO)k]+}]; (2) ∑}I [Fe(H2O)n(CHO)j]2+}/∑{all ions}; (9) ∑{I [X(CHO)k]+}/ ∑{all ions}. The symbol I represents the abundance of an ionic species as measured by the intensity of its mass spectral peak, and the summation extends over all members of a series. The relationship between ttr and D can be seen by comparison of the two horizontal axes.
the mechanism of the initiation process in greater detail than was heretofore possible. Irradiating [CpFebz]+ in an AN solution containing up to 400 mM cyclohexene oxide (CHO) yields the same products observed in the pure solvent (see Figure 4). In addition, species containing coordinated CHO, such as [Fe(AN)3(CHO)]2+ and [Fe(AN)4(CHO)]2+, become increasingly abundant at the higher epoxide concentrations. No products containing more than one epoxide molecule are observed, however, even at the highest CHO concentrations used in our experiments. This result indicates that
Figure 8. Two models for representing the structures of iron-containing products detected in photolyzed solutions of [CpFebz]+ and CHO in 1,2-dichloroethane.
CHO cannot compete effectively with AN for coordination sites on the metal center. Another product, X+ (m/z ) 140), has not been identified, but an analysis of isotope patterns indicates that it does not contain iron. We can detect [X(CHO)]+ at high CHO concentrations, but here again, species containing X+ and more than one epoxide molecule are absent. Photolyzing [CpFebz]+ and CHO in the poorly coordinating solvent, 1,2-dichloroethane (DCE), results in a much richer assortment of products. With ttr ) 50 ms, we detect three major series: half-sandwich complexes of general formula [CpFe(H2O)(CHO)0-5]+, fully ring-deligated complexes of general formula [(H2O)Fe(CHO)2-12]2+, and [X(CHO)0-5]+ (Figure 6). The absence of products containing DCE underscores the weak coordinating ability of this solvent. In contrast, even trace amounts of water (most likely introduced to the solvent during sample preparation or present in the ambient atmosphere surrounding the electrospray tip) are capable of binding to the divalent metal center. Peaks arising from Fe-containing products with two, three, and even four water molecules (e.g., [(H2O)2Fe(CHO)11]2+, [(H2O)3Fe(CHO)9]2+, and [(H2O)4Fe(CHO)2]2+) are discernible in the mass spectrum, although with very low intensities. Weak signals corresponding to the water-free products, [Fe(CHO)5-8]2+ and [CpFe(CHO)1-5]+, also can be detected. The parent-offspring relationships involving the three major product series are readily discernible from plots of relative abundance versus the transit time (Figure 7). These data support the conclusion that the primary photochemical reaction is loss of benzene to produce [CpFe(L)3]+ (eq 1). Subsequent thermal decomposition of the half-sandwich complexes yields the fully ring-deligated Fe2+ complexes and X+containing species. From its relative abundance as a function of ttr, we calculate that [CpFe(L)3]+ has a lifetime of 42 ms in the room-temperature DCE/CHO solution. What are the molecular structures of products such as [(H2O)Fe(CHO)10]2+, [Fe(CHO)8]2+, and [CpFe(CHO)5]+? It is unlikely that all of the epoxide molecules are bound directly to iron, since this would result in unusually high coordination numbers for the metal. Instead, we shall consider two alternative models that avoid this difficulty. Common to each model is the assumption that iron forms a maximum of six strong coordinate-covalent bonds to the
ligands in its first coordination shell. In the cluster model, depicted schematically in Figure 8, the remaining molecules surround this inner cationic core as a loose solvation shell held by ion-dipole forces. Many examples of such van der Waals clusters have been reported.41 Nevertheless, we discount this model on the grounds that clusters are not observed in samples electrosprayed from pure AN (Figure 4), even though this solvent possesses a dipole moment almost twice that of CHO (3.92 42 versus 2.08 D43). In the polymer model, also shown in Figure 8, a coordinated CHO molecule has undergone nucleophilic attack by the oxygen atom of an uncoordinated CHO. Formation of a C-O bond between these two units opens a strained three-membered ring and creates another cationic site for continued chain growth. We favor this model, which involves a growing polymer chain bound directly to the metal center, because it provides a chemically reasonable description of the structures of [(H2O)Fe(CHO)10]2+, [Fe(CHO)8]2+, and [CpFe(CHO)5]+. Moreover, it is consistent with the demonstrated efficacy of [CpFe(η6-arene)]+ complexes as photoinitiators for the cationic polymerization of epoxides. Because X+ forms products containing several CHO units (Figure 6), we propose that this non-iron-containing species also can initiate the cationic polymerization of epoxides. The identity of X+, however, remains an enigma. In one set of experiments, we photolyzed DCE solutions containing CHO and either [(η5C5D5)Febz]+ or [CpFe(η6-C6D6)]+. The resulting ESI mass spectra revealed that no deuterium had been incorporated into X+, strongly suggesting that the production of this species does not involve the direct participation of the carbocyclic ligands bound to the metal center. Exact mass determination and MS/MS experiments are underway to ascertain the elemental composition and structure of X+. CONCLUDING REMARKS Our results demonstrate that direct irradiation of samples in the nanospray tip of an ESI-MS apparatus can yield valuable kinetic (41) Duncan, M. A. Annu. Rev. Chem. 1997, 48, 69-93. (42) CRC Handbook of Chemistry and Physics, 76th ed,; Lide, D. R., Frederikse, H. P. R., Eds.; CRC Press: Boca Raton, FL, 1995; p 9.46. (43) Vereshchagin, A. N.; Anastaseva, A. P.; Vulfson, S. G. Bull. Acad. Sci. USSR, Div. Chem. Sci. 1969, 499-503.
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
4629
and mechanistic information about photochemical reactions in solution. This technique provides mass-specific characterization of stable photoproducts and reactive intermediates with lifetimes in the millisecond time regime. For the specific case of [CpFebz]+ in acetonitrile, we have detected the short-lived primary photoproduct, [CpFe(AN)3]+ and have monitored its thermal decomposition to [Fe(AN)6]2+. In epoxide-containing solutions, we have observed the first few steps in the photoinitiated polymerization process and have obtained direct evidence that species containing the [CpFe]+ fragment (e.g., [CpFe(epoxide)3]+) participate in initiation. In addition, at least two other species appear to function as active initiators: fully ring-deligated Fe2+ complexes and a noniron-containing species, X+. This finding indicates that the mechanism of cationic photoinitiation by [CpFe(η6-arene)]+ complexes
4630
Analytical Chemistry, Vol. 75, No. 17, September 1, 2003
is more complicated than previously thought. We plan to undertake additional ESI-MS studies of a wide assortment of inorganic and organometallic photochemical reactions, including those that generate species capable of initiating anionic polymerization. ACKNOWLEDGMENT We are grateful for financial support from the National Science Foundation (CHE-9974579) and the University of Georgia Research Foundation. We also thank Dr. Billy Flowers and Dr. Marcus Lay for valuable technical assistance. Received for review December 13, 2002. Accepted June 4, 2003. AC0264269