730
Energy & Fuels 1990,4, 730-736
Fast Atom Bombardment Ionization of Porphyrins: Studies of Reduction Processes and Use in Coupling of Thin-Layer Chromatography with Mass Spectrometry Helen H. Schurz and Kenneth L. Busch* School of Chemistry and Biochemistry, Georgia Institute of Technology, Atlanta, Georgia 30332 Received May 10, 1990. Revised Manuscript Received August 6, 1990
The positive-ion fast atom bombardment mass spectra of porphyrins usually contain substantial signals corresponding to ions from the reduced forms of the protonated molecule, specifically those ions that incorporate extra hydrogen atoms. Occurrence of such reduction reactions is not unexpected since the same processes have been throughly characterized in ammonia chemical ionization. Several authors have also noted the presence of these ions in fast atom bombardment mass spectra. However, this work documents the effect of several experimental parameters on the extent of these reactions. The extent of reduction is measured with precision for a number of different porphyrins from two different solvent matrices commonly used in fast atom bombardment ionization. Time dependences of the signal are estalished for the lifetime of the solvent in the mass spectrometer source. For several porphyrins, the extent of reduction is great, but constant, suggesting a steady-state reduction process at the surface of the bombarded liquid. An analogy to electrochemically induced reduction is drawn, and a correlation to electrochemical reduction potentials is firmly established. Finally, the use of fast atom bombardment mass spectrometry (under conditions in which a steady ion signal can be obtained) for the measurement of porphyrin mass spectra directly from thin-layer chromatography plates, and with an intermediate extraction device, is described. Full mass spectra are obtained for about 50 ng of porphyrin within the chromatogram.
In trod uction Porphyrin structural analysis by mass spectrometry has been pursued since about 1960, when the first generalpurpose organic mass spectrometers became available. Mass spectrometry of geoporphyrins, in particular, has been recently reviewed,' and there is no need to reiterate that material here. A signature aspect of the applications of mass spectrometry to problems in porphyrin analysis is continual expansion to include each new method of ionization available as it is brought from the research laboratory into the marketplace. There is, of course, considerable overlap in the applications of ionization methods such as electron ionization, chemical ionization, fast atom bombardment mass spectrometry, and now electrospray,*but there is also considerable diversity of the structures of geological or synthetic porphyrins that are to be examined. Masses of the common porphyrin molecules are rapidly increased by inclusion of large substituents on the pyrrole rings, as well as the presence of multimer forms of the basic structural units. The smaller porphyrins can be volatilized within the source of the mass spectrometer through usual methods, standard derivatization reactions, or with simple modifications of the direct insertion probe. However, more nonvolatile porphyrin samples require the use of ionization methods with wider applicability to sample molecules that cannot be volatilized without degradation. Such methods include field desorption,3 nebulization ionization methods such as the (1)Gallegos, E. J.; Sundaraman, P. Mass Spectrom. Reu. 1985, 4 ,
ss-85. ..
(2)Smith, R. D.;Loo, J. A.; Edmonds, C. G.;Barinaga, C. J.;Udseth, H. R. Anal. Chem. 1990,62, 882-899. (3) Prokai, L.Field Desorption Mass Spectrometry; Brame, E. F., Jr., Ed.; Practical Spectroscopy Series; Marcel Dekker: New York, 1990; Vol. 9.
electrospray method mentioned above, thermospray,* and other liquid introduction methods, and also methods based on energetic particle bombardment, including laser desorption, plasma desorption, and fast atom bombardment (FAB) mass spectrometry. Of the particle-bombardment methods of ionization, FAB is the most widely disseminated and perhaps the simplest experimentally. Numerous reports of FAB mass spectra of porphyrins have appeared in the literature. For the most part, FAB mass spectra of porphyrins contain the protonated molecule (M + H ) ' , where M is the mass of the uncharged porphyrin molecule. In some cases, the odd-electron molecular ion is particularly stable, and the M'+ can be observed in the mass spectrum. The generalized fragmentation processes for porphyrins and metalloporphyrins have been de~cribed.~BThere are also adduct ions formed between porphyrin molecules and molecules of the solvent in which the sample is dispersed for FAB analy~is.~ Several authors5y8have also noted the occurrence of additional ions in the vicinity of the molecular ion or the protonated molecule of the porphyrin. C a ~ t r ofor , ~ example, mentions the presence of (M-H)+, (M + 2H)+,and (M + 3H)+ in the positive-ion FAB mass spectra of several geoporphyrins. Musselmans studied the reduction of three etioporphyrins and presented evidence for reduction up to (M + 5H)+ in the case of the anthracenyl-substituted etioporphyrin. Musselman suggested Vestal, M. L. Mass Spectrom. Reu. 1983,2,447-480. (5)Castro, A. J.;Van Berkel, G.J.; Doolittle, F. G.;Filby, R. H. Org. Geochem. 1989, 14, 193-202. (6)Zhang, M.-Y.; Liang, X.-Y.; Chem, Y.-Y.; Liang, X.-G. Anal. Chem. 1984,56, 2288-2290. (7)Campana, J. E.Org. Geochem. 1989, 14, 171-181. (8) Musselman, B. D.; Watson, 3. T.; Chang, C. K. Org. Mass Spectrom. 1986,21, 215-219. (4)
0 1990 American Chemical Society
FAB Ionization of Porphyrins Table I. Porphyrins Studied (2,3,7,8,12,13,17,18-octaethyl-21~,23H-porphinato)iron(111) chloride 5,10,15,20-tetrakis(4-methoxyphenyl)-2lH~23H-porphine [5,10,15,20-tetrakis(4-methoxyphenyl)-21H,23H-porphinato]cobalt(II) [ 5,10,15,20-tetrakis(4-methoxyphenyl)-2lH,2lH-porphinato]iron( 111)
chloride (5,10,15,20-tetraphenyl-21H,23H-porphinato)vanadium~IV~ oxide (5,10,15,20-tetraphenyl-21H,23H-porphinato)zinc (5,10,15,20-tetraphenyl-2lH,23~-porphinato)cobalt(II) (5,10,15,20-tetraphenyl-2lH,23H-porphinato)copper(II) (5,10,15,20-tetraphenyl-21~,23~-porphinato)manganese(III) chloride (5,10,15,20-tetraphenyl-2lH,23~-porphinato~iron~III~ chloride 5,10,15,20-tetraphenyl-21H,23H-porphine
that the choice of liquid solvent used for FAB should influence the extent of reduction (as had been noted in the reduction processes observed in other FAB mass spectra) and further noted that his work presented no evidence for changes in the relative abundances of these ions with extended time of irradiation by the particle beam. The purpose of this work is to investigate the extent of reduction observed for a number of porphyrins, including metalloporphyrins, in several different liquid solvents used for FAB. The work explicitly establishes the precise extent of reduction and the time dependence of the ion signals. In general, no clear time dependence can be established, forcing specific conclusions about the details of the mechanisms for reduction. There are several implicit assumptions in this work. The first is that the porphyrins themselves are pure and do not contain reduced forms of the sample prior to measurement of the FAB mass spectrum. A second assumption is that there is no complicating reduction chemistry that occurs in the absence of the solvent and that there is no photochemical contribution to reduction. These issues have been shown to be important in understanding the reduction chemistry of some other systemsg that may be predisposed to such complicating reactions. A third assumption is that the liquid solvent itself does not independently contribute to the reduction of the porphyrin molecule, although there is some precedence in the general literature for reductions initiated by specific solvents. A final assumption is that the reduction exemplified by the addition of hydrogen atoms to the intact molecule can be considered independently of the redox chemistry of the metalloporphyrin that involves the central metal.lOJ1 It is perhaps this last assumption that is weakest, since the redox chemistry involving the central metal or the surrounding molecular structure is substantial and complex. Central to any understanding of the reduction processes is an accurate measurement of the extent of hydrogen reduction, which is completed here. In work that requires high sensitivity, knowledge of the exact molecular ion in which the ion current is concentrated, whether it be the protonated molecule or a more reduced form, is a requisite for selected ion monitoring. A second aspect of this work is the use of FAB mass spectrometry and, in particular, continuous-flow FAB mass spectrometry, in conjunction with the determination of porphyrins separated by thin-layer chromatography. Spatial analysis of the ion image requires that the relevant masses in the molecular ion region be established ahead of time, as does maximum sensitivity analysis using the continuous-flow FAB device. Examples in both areas are given. (9) Bentz, B. L.; Gale, P. J. Int. J . Mass Spectrom. Ion Phys. 1987,
5. .78. - , 11 - - -.
(10)Lexa, D.; Momenteau, M.; Mispelter, J. Biochim. Biophys. Acta
1974, 338, 151.
(11) Felton, R. H.; Linschitz, H. J. Am. Chem. SOC.1966, 88, 113.
Energy & Fuels, Vol. 4, No. 6,1990 731 Flow FAB Solvent
n
/
(
w
-Silicon
Septum
I
d
Empore Gel Extraction Device
I
To Flow FAB Probe
Figure 1. Device for pressurized extraction of samples from TLC plates and transfer to a continuous-flow fast atom bombardment source.
Experimental Section Positive-ion FAB mass spectra were recorded for the 11 porphyrins and metalloporphyrins listed in Table I. The porphyrins were used as received from Aldrich Chemical Company. The liquid solvents used for measurement of the FAB mass spectra were m-nitrobenzyl alcohol (mNBA) and a 3:l mixture of dithioerythritol and dithiothreitol (DTE/D'IT), with a small amount (about 5 pL, regardless of the amount of D T E or DTT) of methanol cosolvent. The DTE/DTT mixture with methanol was liquified by heating. The 3:l ratio of DTE/DTT with methanol was used instead of the more familiar 5:l ratio without methanol because of the inability of the 5:l mixture to remain a liquid. These two liquid solvents provided an extended sample lifetime in the source, allowing the time dependence of the ion signals to be established. The DTE/D'IT solvent provided a shorter sample lifetime (20 min) than did the mNBA solvent (30 min). The samples were mixed into these matrices at a level of about 1pg/pL, and approximately 3 p L of sample solution was placed on the stainless steel sample platform. The sample solution was then immediately analyzed. The FAB source was operated with argon a t a source voltage of 8 kV. The angle of incidence to the plane of the sample platform was 45O. A steady particle flux was maintained, and this is estimated a t about 1014particles/(cm2 s), estimated from previous direct measurements of ion current and ion neutral ratios. The ion current was maintained a t 1 pA for all of the measurements. Data were recorded directly on an xy recorder; no background subtraction or signal averaging was used to process the ion signals. MS/MS data were recorded using linked scans on a VG70S forward geometry mass spectrometer, using a linked scan at a constant ratio of B 2 / Eto record parent ion mass spectra from reactions occurring in the first field-free region of the instrument. Helium was used as the collision gas to attenuate the intensity of the ion beam from the source by 50%. Standard solvent systems were used to separate mixtures of porphyrins and metalloporphyrins on Empore silica gel plates purchased from Analytichem. These plates consist of a standard silica gel separation medium supported on a flexible polymer support. T h e sample spots were visually located, excised with a circular punch, and the planchet so obtained placed in the extraction apparatus shown in Figure 1. The two gaskets hold the disk in place while pressure is applied by tightening the torque nuts. An aliquot of solvent is then added from the inlet side; the outlet side is led through a reducer directly to a 75-pm inner diameter capillary line that transfers the solvent to a continuous-flow FAB probe of our own construction. The surface of the flow probe is a rounded sphere of copper centrally drilled to accept the flow capillary. The solvent is drawn into the mass spec-
Schurz a n d Busch
732 Energy & Fuels, Vol. 4, No. 6, 1990 ComDlete FAB Mass SDectrum of TPP Mn(lllK1 in m-NBA
154
mlz 63
mlz 270
b P
c d
o V
O )
V
N )
V
( )
O Y
P )
V
1 )
C V
trometer as a result of the pressure differential; the flow probe acts as a continuous leak into the vacuum system. The pumping capacity of the vacuum chamber is high enough to accommodate continuous solvent flows of up to about 10 rL/min of all but the most volatile solvents. Operating pressure in the chamber (without Torr. Details of this and other differential pumping) is 5 x interface devices for the coupling of thin-layer chromatography with mass spectrometry can be found in other recent papers.I2
u )
Y
I )
Jlme DeDV -
of TPP W l l r C l in m-NBA
Results and Discussion Time Dependence of Ion Signals. Figure 2 is the full mass range positive-ion FAB mass spectrum of TPPMnrn C1 recorded in an mNBA liquid solvent. The spectrum is not corrected for background. The protonated molecule Scan of Molecular Ion Region (m/z 660-674) is seen a t m / z 668, as well as a series of fragment ions in over 8 period of ten minutes. accordance with the general mechanisms established earlier. The ion at m / z 154 is from the solvent itself. The v of TPP -1 in DTELRII persistence of the ion signal from the molecular ion of the porphyrin samples dissolved in mNBA allows FAB mass spectra to be recorded for an extended period of time, allowing exact mass measurements or MS/MS experiments. Figure 3 illustrates typical behavior; here, repetitive scans over the molecular ion region are shown for (5,10,15,20-tetraphenyl-21H,23H-porphinato)manganese(111) chloride in mNBA (Figure 3, top) and DTE/DTT (Figure 3, b o t t o m ) matrices. The longer sample lifetime for the signal in the mNBA matrix is apparent. That there is reduction of the porphyrin to produce (M + 2H)+, (M + 3H)+, and (M + 4H)+ will become apparent in later discussion. Here we wish to establish the precision with Scan of Molecular Ion Region (m/z 660-674) which the ion abundances in the isotopic envelope can be over a period of ten minutes. measured. For a signal that does not degrade with time (as in the mNBA matrix), the measurement is straightFigure 3. Time dependences of the absolute intensity of the molecular ion of (TPP)Mn"'Cl in mNBA (top) and D T E / D T T forward. For the DTE/DTT matrix, each set of absolute (bottom) liquid solvents. signal intensities must be renormalized to establish if there are systematic changes in the abundances of the various ion signals within the envelope. Figure 4 shows the normalized relative abundances for the various forms of the molecular ion for the same tetraphenylporphine in the two (12) Brown, S. M.; Schurz, H.H.; Busch, K. L. J . Planar Chromatogr. liquid solvents. The relative abundance values established 1990, 3, 222.
FAB Ionization
of
Porphyrins
Energy & Fuels, Vol. 4, No. 6,1990 733
Table 11. Extent of Hydrogen-Addition Reduction in Positive-Ion FAB Mass Spectrum of Porphyrins and Metalloporphyrinsa mNBA solvent DTE/DTT solvent (M+ nH)+,n = (M + nH)+, n= sample 2 3 4 5 6 2 3 4 (0EP)Fe"'Cl 7.9 2.6 3.7 6.6 3.8 4.2 4-methoxy(TPP) 2.2 15.5 9.1 2.3 21.6 36.9 15.8 [4-methoxy(TPP)]Co 23.7 5.1 4.3 1.8 [4-methoxy(TPP)]Fe1"C1 2.5 1.9 4.6 1.6 35.6 10.0 7.8 (TPP)VIVO 31.2 6.9 10.2 7.4 8.9 no molecular ion 17.8 8.1 36.5 (TPP)Zn (TPP)Co" 30.4 1.3 2.5 (TPP)Cu" 20.7 no molecular ion (TPP)Mnil'CI 5.5 3.9 2.6 6.2 5.3 4.0 (TPP)Fe"'CI 26.6 11.7 12.6 23.0 13.7 10.1 TPP 1.7 11.2 18.7 7.8
5
3.4 11.8
-
1.3
-
-
Values are expressed as a percentage of the most abundant peak in the isotopic envelope; all values are corrected for the natural isotropic distributions. A dash indicates that no reduction was observed.
in 10 separate measurements are connected by the best-fit line in each case. The fact that the lines are parallel is compelling evidence that there is no accumulation of the reduction products with time. Since the sample concentration is rising significantly as the liquid solvent evaporates when DTE/DTT is used, we can further conclude that changes in concentration of the porphyrin likewise do not affect the amount of reduction observed. This time independence is generally true for the entire set of model porphyrins used in this study. The extent of reduction will change with the choice of the liquid solvent but does not vary with sample concentration across the range studied, nor does it increase with time, as might be expected if the products of reduction were somehow accumulated within the solvent matrix. The implications for the mechanisms involved with the reduction are explored in a later section. Finally, if the ion signal in the molecular ion region does not change with time, repetitive scans provide an excellent estimate of the precision with which the relative ion abundances can be measured. Statistical evaluation of all of the ion signals measured show that, within the isotopic envelope, ion signals at about 50% of the base peak intensity are measured with a 95% confidence limit interval of f3%, an interval of *2% a t the 20% relative abundance level, and an interval of 1% at the 5% relative abundance level. This level of precision is sufficient to compare the observed isotopic envelope against that predicted by theory and to thereby assign the extent of reduction. This is just what has been accomplished in this study. In some instances, additional data points are required to unambiguously assign levels of reduction; alternatively, if enough signal can be obtained, exact mass measurements may provide evidence of reduction. Extent of Molecular Ion Reduction. Table I1 summarizes the results obtained in this study for specific study of the extent of hydrogen-addition reduction processes for this group of 11porphyrins. The data are arranged in the same order of porphyrins as presented in Table I. In each case, the extent of reduction is stated as a percentage of the base peak in the isotopic envelope. All stated values are statistically significant, but of course, those values very near the level of statistical significance should be interpreted with caution. In two cases for the DTE/DTT solvent system shown in Table 11, no molecular ion at all is observed. The sample must be soluble in the solvent for a good-quality mass spectrum to be observed, and in these particular instances, the sample porphyrin did not appear to go into solution nor were good-quality mass spectral data obtained. Problems with solubility have been observed by previous workers in this field; proper selection
TPP Mn(l1l)CI In m-NBA
n=l n=2 n=3 1
n=4
O1 2 0
Tlnu
Figure 4. Time dependences of the relative intensity of the ion signals associated with the molecular ion for (TPP)MnttlC1in mNBA (top) and DTE/DTT (bottom) liquid solvents. The vertical axes are on a logarithmic scale for illustrative purposes only.
of liquid solvent continues to be a difficult problem in fast atom bombardment mass spectrometry. In this work, no molecular isons could be observed for (TPP)VIVO or (TPP)Cu from the DTE/DTTmatrix. Approximately the same m a t r i x was used by Castro et al.5 to record the positive-ion FAB mass spectrum of the same compound; we have no explanation for this discrepancy. A different composition of dithioerythritol and dithiothreitol may alter the solubility (or the surface activity) of this particular compound, and we are investigating this possibility. No reduction whatsoever is noted in the FAB mass spectrum for [Cmethoxy(TPP)]Co and (TPP)Co" in the DTE/DTT solvent mixture, although these same compounds did show evidence of significant hydrogen-addition
734 Energy & Fuels, Vol. 4, No. 6, 1990
reduction in the mNBA solvent. In reviewing literature on FAB mass spectral analysis of porphyrins, or in assessing the FAB mass spectrum of a new compound, the possibility of reduction must always be considered. Even if reduction in one solvent system is not observed, a small change in the composition of the solvent, such as might be produced in efforts to increase solubility or sensitivity, may produce a shift in the extent of reduction. In some of the published FAB mass spectra of porphyrins, reduction occurs even when it is not explicitly noted in text or tables. There is no better argument for the publication of mass spectra rather than tabular mass spectral data. Consideration of the data shown in Table I1 shows that when hydrogen-addition reduction occurs in both solvent systems, it generally occurs to an equal or greater extent in the DTE/DTT solvent mixture than in mNBA. Note the large degree of reduction (35% enhancement in signal intensity) noted for compounds such as [4-methoxy(TPP)]FelllCIand (TPP)Zn. The disparity in the extent of reduction in these two solvent systems is most clear for the former case of [4-methoxy(TPP)]Fe"'C1, which is reduced only slightly in the mNBA matrix but provides evidence of a high degree of reduction in the DTE/DTT liquid solvent. With precise values for the degree of reduction for these 11 model compounds, the next step is to correlate the extent of reduction by hydrogen addition in FAB mass spectrometry with the relative ease of reduction as determined independently and, in particular, with the reduction potentials established by electrochemistry. Such correlations have been suggested in the case of reduction of dye molecules in FAB mass ~pectr0metry.l~In fact, the bombarded surface of the liquid solvent can function as a miniature electrochemical cell, as suggested explicitly in recent work on the reduction of doubly charged ammonium salts in FAB.14 Electrochemicallybased reactions to increase sensitivity in FAB have been explored by Phillips and Bartmess.15J6 The extensive body of work on the electrochemistry of porphyrins is an inviting pool from which to draw structure/function correlations and to relate the extent of hydrogen-addition reduction in FAB to independently measured properties. However, electrochemical reduction potentials are measured in solvent systems that are far removed from the support matrices used in FAB and in the presence of supporting electrolytes not used in FAB. An observed correlation would certainly reflect a complicated interdependence rather than a direct relationship between electrochemical and mass spectrometric behavior. Subject to those constraints, we present here a limited discussion of the reductions of TPP metalloporphyrins, for which we have established the extent of hydrogen-addition reduction as described above. These compounds have been extensively studied by polarography and cyclic voltammetry in a number of nonaqueous solvents. Up to six electrons may be added into the iron porphyrin complex, for example, without disruption of the conjugated ring system. Two steps in a sequence of reduction are typically written as transitions at the central metal ion, viz, Fe(II1) Fe(I1) Fe(1). One notes that, in the FAB mass spectrum, the oxidation state of the metal is not known. The metal ion itself does not usually appear in the positive-ion FAB mass spectrum; if it does, the metal
- -
(13) Pelzer, H.; De Pauw, E.; Marien, J. J.Phys. Chem. 1984,88,5065. (14) Collins, M. W. M.S. Thesis, Indiana University, 1989. (15) Bartmess, J. E.; Phillips, L. R. Anal. Chem. 1987, 59, 2012. (16) Phillips, L. R.; Bartmess, J. E. Org. Mass Spectrom. 1989, 24, 855-856.
Schurz and Busch Table 111. Electrochemical Reduction Potentials for the TPP Metalloporphyrins in Several Different Solvent Systems, Chosen as the First Ring Reduction Potential, and Measured Extent of Hydrogen-Addition Reductions, n = 2, for the Samples in a mNBA Liquid Solventa samvle measured E , ,. V extent of reduction (TPP)V'"O -0.94 31.2 (TPP)Fe"'C1 -1.12 26.6 (TPP)Cu" -1.20 20.7 (TPP)Zn -1.35 17.8 (TPP)Mn"'C1 -1.52 5.5 (TPP)Co" -1.87 30.4 Electrochemical reduction potentials are measured versus the saturated calomel electrode. Reductions were measured in dimethylformamide or dimethyl sulfoxide, except for Mn(III), for which the solvent was not specified. Values are taken from: Felton, R. H. Primary Redox Reactions of Metalloporphyrins; Dolphin, D., Ed.; The Porphyrins, Vol. 6; Academic Press: New York, 1978. TPP CORRELATION
I7 30
-
20
-
Correlation coefficient = 0.953
.1 5
.1 4
.1 3
-1 2
-1 1
.1 0
-0
9
Porphyrln ring reduction potentlal VI SCE
Figure 5. Correlation of electrochemical ring reduction potentials for TPP metalloporphyrins with extent of hydrogen-addition reduction (n = 2).
ion would appear (in the case of the iron porphyrins) as Fe+. This does not mean that the iron is reduced to that oxidation state, as has been amply demonstrated in previous FAB work. In a hydrogen-addition reduction, the position(s) of the added hydrogen(s) idare) similarly not known. Further work may ultimately correlate the daughter ion MS/MS spectra of the various reduced forms of the molecular ion in FAB with those measured from reduced forms of the molecular ion generated in ammonia chemical ionization,"J8 but such work has not yet been completed. Finally, one can envision many paths to reduced forms of the molecular ion. Among such paths are multiple additions of protons to the molecule, followed by sequential electron addition to recreate the singly charged ion. Alternatively, hydrogen radicals (also present at the surface in great abundance) may add to the protonated molecule. Finally, the porphyrin molecule may first be reduced to the anion form by electron reduction, followed by additions of protons to form the singly charged ion. Table I11 summarizes reduction potentials established for the TPP metalloporphyrins. The reduction potentials (17)Van Berkel, G. J.; Glish, G. L.; McCluckey, S. A. Org. Geochem.
1989, 14, 203-212.
(18)Van Berkel, G. J.; Glish, G. L.; McCluckey,S. A.; Tuinman, A. A. J. Am. Chem. SOC.1989, 111, 6027.
FAB Ionization of Porphyrins
Energy & Fuels, Vol. 4, No. 6,1990 735
listed are those measured for the first ring reduction, corresponding to the addition of a single electron to form the anion radical. Figure 5 illustrates the excellent correlation achieved between this electrochemical parameter and the extent of hydrogen addition to form (M 2H)+ in the mNBA solvent for five metalloporphyrins (Mn, Zn, Cu, Fe, VO). Those compounds that are more easily reduced electrochemically are also those that exhibit a greater degree of hydrogen-addition reduction. Of the alternative mechanisms provided above, this correlation suggests that addition of an electron to the porphyrin is followed by addition of a proton to balance the charge; radical hydrogen atom addition seems less likely. The degree of correlation is gratifying, but the anomalous behavior of (TPP)Co" was at first disconcerting. The extent of hydrogen-addition reduction is large, but electrochemically, the compound is the most difficult to reduce among the TPP metalloporphyrins investigated. We suggest that the (TPP)Coprophyrin coordinates with solvent molecules in mNBA and that the relevant electrochemical reduction potential is that of the solvated form. Bottomley et al.l* provide evidence that the electrochemical reduction potential of such a solvated form of the (TPP)Co porphyrin molecule exhibits a ring reduction potential close in value (-1.87 V) to that predicted from an examination of the plot in Figure 5. Clearly, more work is needed to establish the validity of this explanation. The Sputtered Surface as an Electrochemical Cell. Many organic solvents used as FAB matrices are bulk insulators but do not accumulate a charge under bombardment with either a fast primary ion or atom beam. Charge that may build up on the surface is dissipated through loss of ions into the vacuum (in reality, a minor process) and a more substantial flow of electrons across the surface of the liquid solvent to the sample support and hence to instrument ground. The magnitude of that electron flow is related to the magnitude of the charged component of the primary particle beam. At each instant of bombardment, a miniature electrochemical cell is set up between the point of primary particle impact and adjacent areas on the surface and between the surface and the bulk liquid solvent. One notes that the current flow induced by the bombardment is greater than that imposed externally and that electrochemicalderivatization methods may meet with only sporadic success unless the desired cell current can be superimposed over that resulting from the bombardment. The influence of the surface electron current, or the presence of low-energy electrons in general, can be deduced from the formation of Me ions for organic compounds with suitable high electron affinities, from the general reduction of doubly charged inorganic ions such as Ba2+and Ca2+ to their singly charged analogues, and from the dominant reduction of doubly charged organic compounds such as the bis quaternary ammonium salts to form the singly charged ammonium salts in the positive-ion FAB mass spectra.lg It is not therefore surprising that reactions similar to those occurring in electrochemistry are observed in the FAB mass spectra of the porphyrins. Whether the reactions are the same in fact as those independently established is open to question. Certainly, the composition just at the surface will have an effect on the potential electrochemical reactions occurring there, just as it does in solution. A mixture of two solvents, or a pure solvent picking up water from the atmosphere but later releasing that water into the vacuum of the mass spectrometer, will
+
(19) Bottomley, L. A.; Gorce, J.-N.; Davis, W. M. J. Electroanal. Chem. 1986, 202, 111-124.
produce a surface solvent composition somewhat different from the bulk composition. That the latter issue is not important with the present work is shown by the invariance in the spectra with time. However, with the more widely used glycerol solvent, the amount of water in the glycerol solvent is seldom monitored, but certainly significant. Thin-Layer Chromatography Coupled with Mass Spectrometry. Previous workm deonstrated that porphyrins could be sputtered directly from the surfaces of thin-layer chromatography (TLC) plates with the use of a suitable solvent that served to extract the sample from the silica gel and support the persistence of the sputtering of characteristic ions. The limits of detection in this work remained a t a high level of several hundred nanograms since the solvent chosen had to satisfy a number of criteria simultaneously. Such a solvent had to be vacuum compatible, since the entire TLC plate with solvent was placed in the vacuum chamber of the mass spectrometer.21*22At the sPme time, the solvent had to possess the particular property of supporting a persistent ion emission from the surface upon bombardment by the primary ion beam without allowing excess diffusion of the sample spot within the silica gel itself. Additionally, the chosen solvent had to be able to extract efficiently the sample molecules from within the chromatographic matrix. Porphyrins, in particular, were found to be more difficult to extract than many of the other classes of compounds investigated by the combined TLC/mass spectrometer, which has been demonstrated successfully with a number of other system~.~~-~~ Recently we have developed several devices that are used when the full spatial resolution capabilities of the TLC/ MS instrument are not required. In such cases, the sample spot on a TLC plate can be readily identified, as in the location of the colored spots of the porphyrins against the white silica gel. What is required is simply the rapid mass spectral identification of the sample molecules present in the spot. A recent paper describes several interface devices developed for use in such situations.12 For porphyrins separated on TLC plates, we have used the simple device shown in Figure 1 to quickly extract sample porphyrin molecules from within a TLC plate with transfer to the mass spectrometer and a flow FAB ionization method. Instead of extraction of sample over a long time period, or the use of higher temperatures, pressure can be used to force solvents to percolate through a planchet excised from a TLC plate, and the sample is extracted into a solution introduced continuously into the source of the mass spectrometer. The solvent is introduced from a syringe and is introduced onto the planchet; pressure is then exerted by tightening the external torque nut. Moderately high pressures a t the plate are achieved, and the flow of solvent through the coating is rapid, as in overpressure thin-layer chromtography. The sole exit from the pressurized chamber is the capillary line leading to the source of the mass spectrometer. In the experiments described here, the capillary transfer line terminates in a flow FAB source, but other ionization methods (even chemical ion(20) Ryan, T. M.; Day, R. J.; Cooks,R. G. Anal. Chem. 1980,52,2379. (21) Dean, L. K. L.: Busch, K. L. Adu. Mass Snectrom. 1989, IIB, 1646-1647. (22) Fiola. J. W.: DiDonato. G. C.: Busch. K. L. Reu. Sci. Instrum. 1986,57, 2294. (23) Duffin, K. L.; Flurer, R. A.; Busch, K. L.; Sexton, L. A.; Dorsett, J. W. Reu. Sci. Instrum. 1989, 60, 1071. (24) Stanley, M. S.; Duffin, K. L.; Doherty, S. J.; Busch, K. L. Anal. Chim. Acta 1987, 200, 447. (25) Dunphy, J. C.; Busch, K. L. Biomed. Enuiron. Mass Spectrom. 1988, 17, 405. '
Schurz and Busch
736 Energy & Fuels, Vol. 4, No. 6, 1990
Ewu-
'Tpp FeC'
of 20
P 9S01
I
Flow FAB Matrix: 10% m-NBA in CH2 C12
_,-r7
?r
-r-r,7--T-r
150
_I
0
7
190
5?@
630
r
T
1
-
T
'10
r-T_-r---7
-
8 0
'330
1110
m/z
Figure 6. Continuous-flow FAB mass spectrum obtained from (TPP)Fe"'Cl extracted directly from an Empore silica gel thinlayer chromatography plate.
ization) are possible, if the solvent flow into the source is appropriately managed. No pump is needed; the difference in the chamber pressure and the vacuum within the mass spectrometer source ensures a rapid diffusion of sample from the matrix into the source of the mass spectrometer. Sample extraction and transport times are about 30 s. The device is small and simple enough such that it can be manufactured in quantity so that a large number of samples can be prepared in advance for mass spectrometric analysis at a later date. A significant advantage of this device is that the solvent chosen to extract the sample is selected on the basis of extraction efficiency rather than vacuum compatibility. A small percentage (5-10% of a FAB solvent (glycerol, mNBA, or the DTE/DTT mixture) is added to the solvent mixture. As the solution containing the sample molecules is introduced into the source of the mass spectrometer, the more volatile solvents evaporate quickly to leave a small volume of the concentrated sample solution on the probe tip. A recent article reviews the technology of continuous-flow sample introduction for FAB mass spectrometry.26 A simple mixture of synthetic porphyrins was separated on an Empore brand silica gel TLC plate. A portion of the plate containing TPP was excised from the plate and placed in the extraction device described above. A total of 20 Fg was contained in the spot. Figure 6 provides the full range positive-ion FAB mass spectrum of this porphyrin as extracted from the TLC plate and passed into (26) Dunphy, J. C.; Busch, K. L. Talanta 1990,37, 471. Caprioli, R. M. Anal. Chem. 1990,62,47712.
(27)
a continuous-flow FAB probe introduced on a quadrupole-based mass spectrometer usually used for the spatially resolved work. No background subtraction of the data was carried out. The signal-to-noise ratio for the protonated molecule at this level suggests a limit of detection on this instrument of about 100 ng. Exactly the same device was used to transfer the sample to a commercial flow FAB probe on a high-performance sector instrument. A limit of detection of 50 ng was achieved. One notes in these experiments that the signal persists well past the time required to record the mass spectra. A total integration time of 30 s was used to record the mass spectrum shown in Figure 4, for instance, while the ion signal persisted for several minutes. Further, there is no sample cleanup other than that accomplished by the chromatography itself, nor is there any concentration of the sample in the transfer line to the mass spectrometer. We are continuing in work to explore this interface device for the determination of both biological and geoporphyrins separated from complex mixtures by high-performance thin-layer chromatography.
Conclusions This study provides the first comprehensive and precise set of values for the extent of hydrogen-addition reduction processes in 11different model porphyrins. Although the extent of the reduction process can be great, and its occurrence widespread, the mechanism through which such reduction occurs has not been established. It is shown here that the extent of reduction is a function of solvent and porphyrin but is not a function of time of particle bombardment or sample concentration. Although the isotopic envelope of the protonated molecule of the porphyrins is changed by the occurrence of these reduction processes, the signal is reproducible. The extent of reduction for several different TPP metalloporphyrins can be correlated with the electrochemical reduction potential with a high correlation coefficient. With reproducible mass spectra from the porphyrins, the use of mass spectrometry as a detector for porphyrins separated by thin-layer chromatography can be explored. Difficulties have been encountered in efficient extraction of porphyrin molecules from TLC plates with vacuumcompatible solvents that must be used for measurement of spatially resolved mass spectra. An external extractionltransfer device was used to identify porphyrins in TLC spots at a limit of detection down to 50 ng with an analysis time per spot of approximately 30 s. Acknowledgment. We thank L. A. Bottomley and R. H. Felton for helpful discussions with respect to the chemistry of the porphyrins studied. S. M. Brown and K. Howes aided in the acquisition of the flow FAB mass spectra on the VG'IOSEQ instrument.
