Cationization of organic molecules using fast atom bombardment

Frank J Novotny , Jane M Eischen , James A Rice. Analytica Chimica Acta 1999 392 .... Lisa D. Detter , Owen W. Hand , R. Graham Cooks , Richard A. Wal...
0 downloads 0 Views 1MB Size
Anal. Chem. 1985, 57, 2027-2034

dimethyl ether may also become useful reagent gases. Although detailed experiments have not been done, it is unlikely that the chemical nature of X will significantly affect the spectra. To establish the general utility of these ternary mixtures will require additional experiments to determine how sensitive the spectra and selectivity of the ternary reagents are to the composition of the ternary mixture. Registry No. CH4,74-82-8; (CH3)&3i,75-76-3;HzO,7732-18-5; CH3COCH3,67-64-1; CH30CH3,115-10-6;2-ethyl-1-butanol,9795-0; 2-phenyl-1-ethanol,60-12-8; cyclohexanone, 108-94-1;hexanal, 66-25-1;tetrahydrofuran, 109-99-9;tripropylamine, 102-69-2; diallylamine, 124-02-7;2-ethyl-1,3-hexanediol,94-96-2;isopentyl benzyl ether, 122-73-6;ethyl acetate, 141-78-6;2-octanone, 11113-7; 1-phenyl-1-ethanol,98-85-1; 1,7-heptanediol,629-30-1.

LITERATURE CITED (1) Clemens, D.; Munson, B. Org. M8SS Spectrom., In press. (2) Krause, J. R.; Potzinger, P. Int. J . Mass Spectrom. Ion Phys. 1975, 18, 303-316. (3) Klevan, L.; Munson, B. Int. J. Mass Spectrom. Ion Phys. 1974, 1 3 , 26 I -268. (4) Odiorne, T. J.; Harvey, D. J.; Vouros, P. J . Phys. Chem. 1972, 6 , 3217-3220. ( 5 ) Odiorne, T. J.; Harvey, D. J.; Vouros, P. J . Org. Chem. 1973, 38, 4274-4278. (6) Vouros, P.; Harvey, D. J.; Odiorne, T. J. Spectrosc. Left. 1973, 6, 603-615. (7) Stillwell, R. N.; Carroll, D. I.; Nowlin, J. G.; Horning, E. C. Anal. Chem. 1983, 55, 1313-1318. (8) Blair, 1. A.; Phllllpou, G.; Bowle, J. H. Aust. J . Chem. 1979, 32, 59-64.

2027

(9) Blair, I. A.; Bowie, J. H. Aust. J . Chem. 1979, 32, 1389-1393. (IO) Trenerry, V. C.; Bowie, J. H.; Blalr, I. A. J . Chem. Soc., Perkin Trans. 2 1979, 1640-1643. (11) Trenerry, V. C.; Blair, I. A,; Bowle, J. H. Aust. J . Chem. 1980, 33, 1143-1146. (12) Blair, I. A,; Bowie, J. H.; Trenerry, V. C. J . Chem. SOC., Chem. Commun. 1979, 230-231. (13) Trenerry, V. C.; Klass, 0.; Bowie, J. H; Blair, I . A. J . Chem. Res., synop. 1980, 368-387. (14) Hendewerk, M. L.; Well, D. A,; Stone, T. L.; Ellenberger, M. R.; Farneth, W. E.; Dixon, D. A. J . Am. Chem. Soc. 1982, 104, 1794-1799. (15) Ellenberger, M. R.; Hendewerk, M. C.; Weil, D. A.; Farneth, W. E.; Dixon, D. A. Anal. Chem. 1982, 54, 1309-1313. (16) Spreen, R. Ph.D. Thesis, University of Delaware, June 1983. (17) Lias, S. 0.; Liebman, J. F.; Levin, R. D. J. Phys. Chem. Ref. Data 1984, 13, 695-808, (18) Hendewerk, M. L.; Frey, R.; Dixon, D. A. J . Phys. Chem. 1983, 87, 2026-2032. (19) Pitt, C. 0.; Bursey, M. M.; Chatfield, D. A. J . Chem. Soc., Perkin Trans 2 1976, 434-438. (20) Pedley, J. B.; Iseard, B. S. CATCH Tables, Silicon Compounds, University of Sussex, 1972. (21) Szepes, L.; Baer, T. J . Am. Chem. SOC. 1984, 706, 273-278. (22) Hunt, D. F. Prog. Anal. Chem. 1973, 6 , 359-376. (23) Meot-Ner, M.; Sieck, L. W. J. Am. Chem. SOC. 1983, 105, 2956-2961. (24) Rudewicz, P.; Munson, B. Presented at 32nd Annual Conference on Mass Spectrometry and Allied Topics, San Antonio, TX, May 27-June

I, 1984. (25) Shea, K. J.; Gobeille, R.; Bramblett, J.; Thompson, E. J . Am. Chem. SOC. 1978, 100, 1611-1613.

RECEIVED for review March 12,1985. Accepted May 10,1985. This research was supported in part by a grant from the National Science Foundation, CHE-8312954.

Cationization of Organic Molecules Using Fast Atom Bombardment Mass Spectrometry T. Keough T h e Procter and Gamble Go., Miami Valley Laboratories, P.O. Box 39175, Cincinnati, Ohio 45247

Cationization of organic molecules has been studied using fast atom bombardment mass spectrometry and MS/MS techniques. The purpose of this study was to develop methods to faciiltate the formation of catlonlzed molecules for analytical applications. The Influence of Ion-permanent dipole Interactions on alkali ion affinities and on the structures of some cationlzed species Is elucidated with MS/MS. Different alkali cation donors produce comparable yields of cationlzed analytes. A precharged anaiyte (sodium octanoate) gives a significantly more abundant MNa+ adduct ion than a structurally slmiiar neutral molecule (octanoic acid) which has comparable surface activity. Furthermore, the choice of the FAB matrix, glycerol vs. thiogiyceroi, may slgnificantly aff ect the yield of catlonized analytes. Finally, the utility of MS/MS for studying gas-phase desolvation processes is Illustrated.

Ion/molecule reactions have been used extensively, in analytical applications of mass spectrometry, for the purpose of determining the molecular weights of organic molecules that are unstable with respect to electron ionization (1). Some commonly used ion/molecule reactions include protonation (eq l),electrophilic attachment (eq 2), and cationization (eq 3) with metal ions. Exothermic gas-phase proton transfer 0003-2700/85/0357-2027$01.50/0

M

+ NH4+A MNH4+ N M + C+ MCt

-

reactions (eq l),between reactant ions (RH+) and organic molecules (M), occur collisionally and are among the fastest known elementary processes ( 2 ) . Thus, chemical ionization (CI) via eq 1provides a sensitive method for producing stable MH+ adduct ions, and the technique has been studied extensively (3). Electrophilic attachment (eq 2) of a basic reactant ion such as NH4+or C5H6N+can also be used for chemical ionization applications, particularly for volatile materials that are unstable toward protonation ( 4 , 5 ) . However, these reactions occur with much lower rates (sensitivity) than exothermic proton transfer reactions and, at least in the case of NH, CI (6),are only applicable to organic molecules possessing a relatively narrow range of proton affinities around that of the reactant ion. Cationization (eq 3) of organic molecules with metal ions (C+)is a process that has also been used to characterize materials that are unstable toward protonation. The approach is most often applied to nonvolatile, thermally labile materials using various desorption ionization methods such as field desorption (FD) (7, 8),plasma desorption (PD) (91,laser desorption (LD) (10-12),secondary 0 I985 American Chemical Society

2028

ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985

ion mass spectrometry (SIMS) (13-15), and fast atom bombardment (FAB) (16). Cationization reactions are ubiquitous in FD mass spectrometry, where traces of alkali salts are often sufficient to produce intense cationized species. Furthermore, these reactions have been extensively studied (13-15) using static SIMS, a technique similar to FAB but which does not employ a viscous liquid matrix. Given the widespread application of cationization reactions and the rapid growth in the use of the FAB technique, it seems appropriate to investigate some cationization processes that occur under FAB conditions. The purpose of this study is not to try to rigorously define the ion chemistry responsible for the production of cationized molecules. Rather, it is hoped that this work will facilitate optimization of FAB yields of cationized species for analytical applications. EXPERIMENTAL SECTION Instrumentation. All spectra were obtained on a Vacuum Generators ZAB-2F reverse-geometry, double-focusing mass spectrometer operating at an ion accelerating voltage of 8 kV and a mass resolution of approximately lo00 (10% valley definition). Fast atom bombardment was accomplished with Xe (99.995%, Matheson Gas Products, Inc., Joliet, IL) as the primary particle, with a modified saddle-field ion source (Ion Tech, Ltd., Teddington, UK). Typically, the tube current was 1 mA and the energy 7.5 kV. The mass spectra were acquired and processed with a Vacuum Generators 2350 data system. The MS/MS spectra (second field-freeregion) were output to a UV recorder (Honeywell Model SE-6150). MS/MS spectra were obtained either without collision gas (analyzer pressure -lo4 torr) or with collision gas, such that the pressure in the analyzer region increased to -lo4 torr. Reagents. Commercially available glycerol (Fisher Scientific, Fairlawn, NJ, 99.5%) and thioglycerol (Aldrich, Milwaukee, WI, 95%) were used as FAB matrices without further purification. The alkali halides and organic analytes were obtained from various commercial sources and used without further purification. Procedure. Various solutions of known concentration (see text) were prepared for analysis by FAB. The more concentrated alkali halide solutions (>0.1 M) were heated to aid in solubility. Solutions containing labile analytes, such as sucrose, were prepared without heating to avoid thermal decomposition. The glycerol surface tension measurements were conducted at 21 O C using the ring method (17) with a 6 cm Pt ring. RESULTS AND DISCUSSION The mechanism(s) of production of cationized molecules under FAB conditions is not fully understood. However, Cooks and Busch (18)have put forth a very useful desorption ionization model that describes qualitatively the features of ion emission in a number of particle-induced emission techniques such as FAB, SIMS, LD, and PD. This model provides a framework for interpreting much of the data presented here. The model suggests that there is a delay between energization and desorption (this has been measured under LD conditions to be greater than several microseconds (19)). Preformed ions and neutral molecules are desorbed in high yield. The majority of these preformed ions traverse the selvedge region (high pressure region just above the surface) and are observed as intense ions in the mass spectrum. Desorbed neutral molecules, on the other hand, must be converted to ions by fast ion/molecule reactions in the selvedge region before they can be detected. Some reactions such as (eq 3) and (eq 4) M GlyNa+ MNa+ + Gly (4)

+

-

might enhance the formation of the cationized molecules of interest while processes such as cation transfer from the anal@, to other molecules (M') present in a mixture (eq 5 ) ,will MNa+ + M'

-

M

+ M'Na+

(5)

suppress the detection of the desired cationized species. Beyond the selvedge region, no further ionization takes place.

0 01

0.0

-3

00

-2

0

-10

log c

-5

0

log c

Figure 1. Variation in relative abundance of some glycerol ions as a function of the concentration (C) of added NaI.

However, unimolecular fragmentation of the cationized molecule or desolvation of large cluster ions (eq 6), formed MNaGly,+

-

MNa+

+ nGly

(6)

by direct emission from the surface, can either suppress or enhance the desired cationized species. This model stresses the importance of chemical factors in determining the nature of secondary ion yields. Obviously, the presence of the liquid matrix is a major difference between experiments conducted under FAB vs. static SIMS conditions. In FAB, desorbed neutral matrix molecules may be the most abundant species in the high pressure region just above the liquid surface, particularly for analytes that are not good surfactants. Thus, the ultimate yield of cationized molecules will depend on how well the analyte competes, with the matrix, for available alkali cations in surface ionization processes and processes such as eq 3-6. The relative contributions of ion/molecule reactions (eq 3-5), gas-phase desolvation processes (eq 6), and ionization on the liquid surface to the yield of cationized molecules is unknown and cannot be determined from the experiments reported here. However, it has been previously suggested (20), based on classical dynamic models (21), that the proportion of species formed in the gas phase by association processes (eq 3-5) is probably small, relative to the condensed phase. Cationization of Glycerol. Before considering the cationization of analyte molecules, we investigated the cationization of glycerol, the most commonly used FAB matrix. Figure 1shows the variation in abundance of some of the more intense glycerol ions as a function of the concentration of added NaI. Similar results were obtained by doping glycerol with NaC1. At a NaI concentration of -0.01 M, we begin to observe a decrease in protonated glycerol ions, such as GlyH+ and Gly2H+,and an increase in abundance of GlyNa' and Gly,Na+. The GlyNa+ ion is presumably formed by desolvation of desorbed cluster ions (eq 6) and possibly by the association reaction (eq 3). A "switching mechanism" (eq 7 ) GlyH+

+ NaCl

-

GlyNa+

+ HC1

(7)

is also thermodynamically allowed so long as the proton affinity of glycerol is less than 236 kcal/mol. We have shown (see below) that the proton affinity of glycerol is less than that of 1,lO-decanediol. Furthermore, the proton affinity of 1,lO-decanediolcan be estimated in a number of ways. Based on the known proton affinities of 1-alcohols (22,23)and on the enthalpy change gccompanying the gas-phase solvation of H30+ by H 2 0 (AH= -35 kcal/mol) (24),we obtain a value of -225 kcal/mol. This estimate represents an upper limit since it ignores any weakening of the intramolecular hydrogen bond due to ring strain effects (25). Alternatively, the proton affinity of 1-decanol can be increased by the difference in proton affinities between ethers and polyethers (-30 kcal/ mol) (26) or by the difference in proton affinities between amines and diamines (18-26 kcal/mol) (27). These latter procedures yield estimates ranging between 220 and -208

-

ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985

kcal/mol, respectively. The calculation of the enthalpy change accompanying eq 7 also assumes the following thermochemical data (22): sodium ion affinity of glycerol -35 kcal/mol (see below); AHf(Na+)= 143 kcal/mol; M f ( H + )= 365 kcal/mol; AHf(HC1)= -22 kcal/mol; and AHf(NaC1)= -43 kcal/mol. Finally, as the NaI concentration is increased above 0.01 M, we observe (Figure 1)an increass in relative abundance of Na+, Na21+,and (Gly - H 2Na)+. The increased yield of Na+ might arise because of processes such as eq 8. The formation

+

GlyH+

+ NaI

-

Gly

+ Na+ + HI

(8)

of small cluster ions such as Nazi+ may result from ion/ molecule reactions in the selvedge. This is the case, at least, under LD (28)and SIMS (29)conditions. In the SIMS study (29),it was shown that the gas-phase association reaction (eq 3, M = NaC1, C+ = Na+) was responsible for the production of these types of cluster ions. The important point to note, from an analytical viewpoint, is the fact that Nazi+ and (Gly - H 2Na)+ are observed in yields comparable to those obtained for many analytes under FAB conditions. For example, the ratio of intensity of (Gly - H + 2Na)+/(GlyNa)+is -0.1 while that of Na21+/(GlyNa)+is -0.03 at the highest NaI concentrations reported in Figure 1. Since ion/molecule reactions can occur on the surface and in the high pressure region just above the glycerol surface, these results suggest that the Na+ affinities of NaI and (Gly - H + Na) are greater than that of glycerol (see below) and, therefore, greater than that of many organic analytes. High concentrations of NaI and/or (Gly - H + Na) could suppress the detection of cationized analytes via eq 5 (M' = NaI or (Gly - H + Na)). It seems reasonable, for analytical applications of cationization, to dope glycerol solutions with alkali halides at concentrations 10.1 M and to remove any large excesses of alkali salts contained in samples isolated from complex matrices such as biological fluids. The alkali ion affinities of glycerol have not yet been reported. In fact, little work has been directed toward construction of an alkali ion affinity scale comparable to the proton affinity scale that has been previously developed (27). Staley and Beauchamp (30)have determined the Li+ affinities of 30 n and ?r donor bases by use of trapped ICR techniques. Davidson and Kebarle (31) have determined the alkali ion affinities of acetonitrile and the K+ affinities of several nitrogen and oxygen bases using high pressure mass spectrometry (32). Cooks and co-workers (11) have used MS/MS to order the relative Ag+ affinities of several simple alcohols. Based on Kebarle's work (31),we would expect that the K+ affinity of glycerol would be comparable to or somewhat greater than that of ethylenediamine (25.7 kcal/mol). The Na+ affinity would be greater than the K+ affinity by - 5 to 10 kcal/mol. Clearly, the alkali ion affinity of glycerol is higher than would be expected for simple organic molecules containing a single functional group. Thus, competition between glycerol and monofunctional analytes for available alkali cations will, in general, inhibit the formation of this type of cationized analyte. We have attempted (see Figure 2) to order the Na+ affinities of Gly, (Gly - H + Na), and NaI using the MS/MS approach (11). This approach utilizes the gas-phase fragmentation behavior of some unsymmetrical cluster ions, LIMLz+,where L, and Lz are the species whose relative metal ion affinities are to be determined. In this study, it was assumed that the species with the higher Na+ affinity yields the more abundant fragment ion (LNa+) in the MS/MS spectrum. It must be emphasized that the MS/MS method has been previously utilized to order the proton and Ag+ ion affinities of closely related series of structurally similar materials. The use of the method to order the Na+ affinities of structurally dissimilar materials such as Gly, NaI, and (Gly - H + Na) is untested

100

60 40 80

20

~

2029

A '

(Gly-H+ZNa)+ (Gly-HtNa). NaN . .a'l N , aJ

I

0d

+

80

40

-

20 204 0

GlyNa'

\

I

I

Flgure 2. Collisionactivated MS/MS spectra of some unsymmetrical cluster ions generated by FAB.

at this time; however, the internally consistent set of results shown in Figure 2 suggest that the order of Na+ affinities is Gly < NaI < (Gly - H Na). A relatively high Na+ affinity for NaI is not surprising considering the electrostatic nature of the bond formed between metal ions and molecules. Kebarle (32)has emphasized the importance of ion-permanent dipole interactions in stabilizing these bonds. The total ion-permanent dipole interaction was shown to be proportional to CQi/Riwhere Qi is the point charge on the ith atom of the molecule and Riis the distance between the center of the metal ion and the ith atom. There is a much larger electron population in the I atom of NaI (permanent dipole moment -9 D) than on the basic atoms of most organic compounds which have substantially lower permanent dipole moments (2-3 D) (33). Glish and Todd have proposed the following structure (a),

+

+ Ho..'\ Li "'0 H / CHrCH-CH,

a

based on MS/MS results, for the (Gly - H + 2Li)+ cation generated by FAB (34). The MS/MS spectrum of (Gly - H + 2Na)+ exhibits NazOH+as the largest fragment ion and is also consistent with a structure similar to a (35). It is not surprising, then, that (Gly - H + Na) also has a high Na+ affinity since it has a large permanent dipole moment (along the -0-Na+ bond) and because the attached metal ions can be chelated by the adjacent hydroxyl groups. Alkali Cation Effects. We have studied the influence of the alkali cation on the yield of cationized molecules by using LiI, NaI, KI, and CsI as the alkali cation donors. The distributions of solvated metal ions, Gly,M+ (n = 0-4), obtained from glycerol solutions containing 0.1 M MI (with no analyte) are summarized in Table I. The most abundant metal-containing ion is, in all cases, GlyM+. The relative abundance of the larger cluster ions, Gly,M+, decreases quickly with size because the free energy change accompanying successive additions of solvent molecules rapidly approaches zero with

2030

ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985

Table I. Distribution of Solvated Metal Ions Obtained from FAB of Glycerol Containing 0.1 M MI Ii+/zIi+a

MI

M'

GlyM'

Gly2M'

GlyaM'

Gly4Mt

LiI NaI KI CsI

0.00

0.61 0.62 0.54

0.35 0.24

0.03 0.03 0.02 0.03

0.01 0.01 0.01

" CIi'

0.10 0.32 0.37 E

M'

0.49

0.11 0.10

0.01

+ GlyM' + Gly2Mt + GlySM' + GlydM'.

Table 11. Abundance Ratio of Some Cationized Molecules (MC') to Protonated Molecules (MH') Produced by FAB

analyte sucrose 1-hydroxyethane-1,l-diphosphonic acid

1,lO-phenanthroline p-aminobenzoic acid

(MCt)/(MHt)p' Li Na K Cs 1.56 2.19 0.53 b 0.28 0.14 0.02 0.07

E

Y x

E x

2.02 0.60

1.54 0.46 0.08 0.06 0.07 0.18

"MC' MC' + C,MC(Gly),'; MH' MHt + znMH(Gly),'. *Not determined since MH' overlaDs with Glv,Na' at m l z 207. increasing n (31). There is also a dramatic difference in the amount of nonsolvated metal ion obtained from each of these solutions. The abundance of Li+, normalized to the sum of the cluster ion abundances, is 0.0 while that of Cs+ is 0.37. It is interesting to note that, under these conditions, virtually all of the Li+ ions are attached to glycerol. Consequently, little free Li' is available for association with desorbed neutral analytes (eq 3) in the selvedge region just above the liquid surface. This latter process is considered to be a major mechanism of cationization in SIMS conducted without the liquid matrix (18). If this process is also the dominant mode of cationization in FAB, then one might expect that the yield of cationized analytes would decrease significantly when using LiI as the alkali cation donor. Results obtained from several glycerol solutions containing 0.05 M analyte and 0.1 M MI are summarized in Table 11. The intensities of the cationized analytes (MC+) have been normalized relative to the protonated molecular adduct ions (MH+)to facilitate comparison. With the possible exception of p-aminobenzoic acid, a significant decrease in the relative abundance of MLi+ compared to the other MC+ cations is not observed. In the case of 1,lO-phenanthroline,the relative abundance of the cationized molecule is enhanced using LiI rather than Na, K, or CsI as the alkali cation donor. Similarly, with the possible exception of p-aminobenzoic acid, increasing the selvedge concentration of free alkali cations also does not produce a significant effect on the yield of cationized molecules. The ratio of MCs+/MH+ is generally not substantially greater than the ratio of MC+/MH+, obtained with the other alkali cations, even though the nonsolvated metal ions are expected to be more reactive than solvated species such as GlyC+ (by the binding energy of the solvent). This observation may be explained in part by the fact that the alkali ion affinities of organic molecules decrease with increasing alkali ion size; however, the Cs+ affinities of many organic molecules are expected to be appreciable. For example, the Cs+ affinity of H 2 0 is 14 kcal/mol while that of acetonitrile is -19 kcal/mol (31). These results indicate that the association of free alkali cations with desorbed neutral molecules (eq 3) is not the dominant mode of cationization in FAB. Processes such as cation exchange in the selvedge region (eq 4) and direct sputtering of cationized species that exist in solution (eq 6) are implicated. On the basis of OUT own limited study and the work of others ( 3 4 ,it is apparent that the choice of cation does not produce dramatic effects on the yield of cationized molecules or on the

-

-4

-3

-2 log c

-1

0

Figure 3. Variation in glycerol surface tension as a function of the bulk concentration (C) of added surfactant. nature of the fragmentation pattern (13). However, the use of cations such as Ag+, with distinctive isotope patterns, aids in the identification of cationized molecules (16,36). Finally, it has also been suggested that the cation may influence the nature of fragment ions produced by MS/MS of cationized species (34). Cationization of Preformed Ions. It is well-known that salts produce the most abundant signals with various desorption ionization techniques (37,38). This fact previously led us to develop methods for significantly enhancing the FDMS sensitivity for various zwitterionic compounds (37). Sensitivity enhancement was achieved by simply converting the neutral zwitterion to a quaternary ammonium salt prior to MS analysis. Subsequently, it was shown that this "reverse derivatization" approach was generally applicable to other desorption ionization methods (LD, SIMS) (38),including FAB (39),and that the approach could be easily extended to other classes of organic compounds. Octanoic acid and sodium octanoate have been employed as model compounds to compare the extent of cationization resulting from FAB of a neutral molecule and a structurally similar preformed ion. We expect that the glycerol surface concentration of each material will be comparable in solutions of identical bulk concentration and that any differences observed in secondary ion yields will not reflect differences in the surface activities of these materials. This conclusion is based on the surface tension measurements summarized in Figure 3. The results suggest that octanoic acid is a slightly better surfactant than sodium octanoate at all bulk concentrations examined within the range of to 10-1M. The surface excess concentration, calculated using the Gibbs expression (40),is 4.1 x W0 mol/cm2 for octanoic acid and 1.9 x mol/cm2 for sodium octanoate. Thus, if the cationization efficiencies of the neutral and the preformed ion are identical, we would expect to observe, within about a factor of 2, comparably abundant cationized molecules in the FAB spectra of these materials. The results obtained from analysis of octanoic acid and sodium octanoate (both 0.1 M) from

ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985

9

2031

Table 111. Collision-Activated MS/MS Spectra of Some Unsymmetrical Naf Bound Cluster Ions (SNaGly+) 8 O i I/GlyNa'

SNa+/ GlyNa'

Solvent (S) 1,2,6-hexanetriol glycerol

1,3-butanediol 1,3-propanediol 3-mercapto-1,2-propanediol

ethylene glycol P

3.20 1.00 0.51 0.37 0.30 0.18

ilr;

I 120

160

MNaGly'

I

200 MI2

240

280

In solution, glycerol is known to function as a bidentate ligand with alkali cations. Chelation apparently involves the hydroxyl groups on adjacent carbon atoms while the third hydroxyl group is prevented from participating because of steric factors (46).The fact that the observed Na+ affinities of the trihydroxy compounds are greater than those of the dihydroxy species suggests that, in the gas phase, the third hydroxyl group is involved in some way in stabilizing the adduct ion. This could result by intramolecular hydrogen bonding (one isomer given in b). 4-

Figure 4. FAB mass spectra of octanoic acid plus N a I (upper) and sodium octanoate plus N a I (lower).

separate glycerol solutions also doped with NaI (0.1 M) are given in Figure 4. The spectrum of octanoic acid (upper) is dominated by cationized glycerol while the cationized acid (mlz 167) is barely evident above background. The spectrum of sodium octanoate (lower) exhibits an abundant cationized molecule (mlz 189) in addition to the glycerol-containingions. Clearly, the secondary yield of cationized molecules is significantly greater for the salt than for the free acid. We anticipate that sodium octanoate will have a higher sodium ion affinity than octanoic acid based on the ionpermanent dipole considerations presented earlier. For example, the electron population on each of the oxygen atoms of formic acid has been previously calculated to be 0.3e- (41). In the case of the formate anion, the electron population on each oxygen atom increases to -0.73e- (42). For sodium formate, we would expect that the electron population of the oxygen atoms would be reduced somewhat relative to the formate anion, possibly to -0.6e-,as was found in a theoretical study of calcium binding to the formate anion (43). The increased charge on the oxygen atoms of sodium octanoate, relative to the free acid, favors the ion-permanent dipole interactions which facilitate cationization and mitigates against processes such as eq 5 (M' = Gly). It should also be noted that the mechanisms of cationization of the acid and the salt may differ significantly and that this difference may also contribute to the enhanced cationization yield observed for the salt. It is likely that the cationized salt is formed, at least in part, via decomposition of neutral dimers that preexist on the glycerol surface (eq 9). Processes of this

nature, which do not yield the desired product ion for the free acid, have been previously proposed to occur during thermal desorption of carboxylic acid salts (44, 45). Matrix Effects. We investigated whether it might be possible to increase an analyte's ability to compete with the FAE!matrix, for available alkali cations, by using FAB solvents with alkali ion affinities lower than that of glycerol. Therefore, the Na+ affinities of several potential FAB matrices were ordered, using the MS/MS approach discussed earlier, and the results are summarized in Table 111.

b

The unimolecular MS/MS spectrum of GlyNa+ does exhibit a significant ion corresponding to H 2 0 loss (approximately 20% as large as NaOH loss, the most abundant fragment ion observed). The water-loss fragment ion is produced after an intramolecular proton transfer to the hydroxyl group that is lost as HzO. This proton transfer process most likely involves the hydroxyl hydrogens since they are more acidic than those attached to the carbon atoms. Obviously, complete proton transfer yields an alcoholate anion that further facilitates binding to Na+ (47). Thus, a structure such as b, which is stabilized by intramolecular hydrogen bonding, does not appear to be at all unreasonable. From Table 111,we also note that the Na+ affinities of the 1,3-diols are greater than that of ethylene glycol. Finally, the Na+ affinity of 3-mercapto1,2-propanediol (thioglycerol) is lower than that of glycerol and is comparable to that of ethylene glycol and 1,3propanediol. Thus, the S H group is not as effective as a third -OH group in stabilizing the adduct ion. One wonders if the Na+ affinity of 2-mercapto-l,3-propanediolwould be more nearly comparable to that of glycerol. Unfortunately, our attempts to prepare this compound by catalytic hydrogenation of 2-thioglyceraldehyde were unsuccessful. On the basis of the data in Table 111, it was decided to compare the extent of analyte cationization from the glycerol and thioglycerol matrices. Choice of the FAE! matrix will determine the relative alkali ion affinities of the analyte and the matrix. This choice will also influence other properties such as sample solubility, surface activity (481, matrix volatility (49), rate of sample utilization, and rate of gas-phase desolvation (see below). In this experiment we sought to determine the effect of relative alkali ion affinities on the yield of cationized analytes with minimal contribution due to variations in these other properties. Our approach utilized a single analyte that could be ionized by two competing ionization processes, e.g., protonation and cationization (eq 10). Factors such as sample M

+ H+ AI--MH+

(loa)

M

+ Nat

(lob)

kz

MNa+

solubility and surface activity will determine the quantity of M on the surface available for ionization by one of these routes. However, the relative yield of these distinct product ions

2032

ANALYTICAL CHEMISTRY, VOL. 57, NO. 11, SEPTEMBER 1985

100,

131 197

'OI

175

0

20

60

120 180 240

0

60

120

180

240

Time(sec)

0

100

120

140

160

180

200

220

MI2

Flgure 5. FAB mass spectra of 1,lOdecanedlolplus NaCl in glycerol (upper) and 1,lO-decanediolplus NaCl in thioglycerol (lower).

should be independent of these factors. The ratio of MNa+/MH+ observed in the spectra will reflect the yields obtained in eq 10. For this study, l,lO-decanediol(l,lO-DD) appears to be an ideal analyte. MS/MS spectra of the unsymmetrical proton bound cluster ions (c, C+ = H+) formed 1, I O - D D - C + - - S C

between 1,lO-decanedioland the matrix molecule (S = glycerol or thioglycerol) reveal that the order of increasing proton affinity is thioglycerol < glycerol < 1,lO-decanediol. Thus, either protonated matrix molecule would be expected to protonate the analyte via eq 11. Furthermore, back reaction

SH++ M -,S + MH+

(11)

of MH+, as a result of collision with the matrix molecule in the selvedge, is not favored. On the other hand, the MS/MS spectra of the corresponding Na+ bound cluster ions (c, C+ = Na+) reveal that the Na+ affinity order does not follow the proton affinity order. The order of increasing Na+ affinities is thioglycerol < 1,lO-decanedioldglycerol. Cation transfer from the matrix molecule to 1,lO-decanediolshould proceed readily with thioglycerol as the matrix. From the glycerol matrix, the process is expected to be slightly endothermic. Thus, thermodynamic effects upon the extent of cationization should be prominent with this particular analyte. Some results obtained with 0.1 M 1,lO-decanediolfrom NaCl(O.1 MI-doped glycerol and thioglycerol are shown in Figure 5. In line with our expectations, the ratio of MNa+/MH+ increases substantially (a factor of 13) when the species is analyzed from thioglycerol. At lower analyte concentrations (0.01 M), the difference is even more pronounced (a factor of -80). Another experiment was conducted to further probe the utility of thioglycerol as a cationization matrix. In this experiment, the absolute intensities of the MNa+ cations produced by FAB of separate solutions of either glucose or sucrose (0.1 M) plus sodium chloride (0.2 M) in either glycerol or thioglycerol were compared, Figure 6. These results were obtained by tuning the magnet to transmit either the MNa+ cation of glucose ( m / z 203) or sucrose ( m / z 365) to the detector. The abundance of these ions, obtained with as nearly identical mass spectrometry conditions as possible, was monitored as a function of irradiation time. The absolute abundance of the MNa+ ion of either analyte, obtained from the glycerol matrix, remained essentially constant over the 4-min analysis time. The MNa+ signal initially observed for glucose, from the thioglycerol matrix, is -2 times greater than that obtained from the glycerol matrix. For sucrose, the initial

Flgure 6. Variatlon in abundance of cationized glucose and sucrose from the glycerol and thioglycerol matrices as a function of FAB irradiation time.

MNa+ signal from the thioglycerol matrixis -4 times that obtained from the glycerol matrix. In both experiments involving thioglycerol, a steady increase is observed in the absolute intensity of the cationized molecule, with irradiation time, until the matrix has been depleted. A much slower increase in abundance was observed for the MNa+ ions that were generated from the glycerol matrix. This latter result is similar in nature to some previously observed results on the cationization of stachyose from glycerol (50). The much more rapid increase in MNa+ yield from the thioglycerol matrix reflects the high volatility of thioglycerol relative to glycerol. In the few systems that we have studied, more abundant cationized molecules have been obtained using thioglycerol rather than glycerol as the matrix. The practice in this laboratory is to use thioglycerol, so long as the analyte exhibits suitable solubility in that matrix. Gas-Phase Desolvation Studies. Cooks and Busch (18) have suggested that the observation of large cluster ions in the desorption ionization mass spectra of salts and organic molecules could be most readily accommodated as the result of the direct emission of molecular clusters. Furthermore, Williams has suggested that solvated ions that preexist in polar solutions can be sputtered into the gas-phase without loss of their charge (51). The solvent shell may actually aid t L 3 process by screening that charge. It was further suggested that the internal energy imparted to these ions during sputtering is beneficial, facilitating the splitting off of weakly bound soIvent molecules and the production of relatively "clean" cationized species (51). These studies suggest that gas-phase desolvation of desorbed cluster ions (eq 6) may be a major route of formation of the cationized analytes that are observed in FAB. There is theoretical and experimental evidence indicating that gas-phase cluster ions are highly unstable and that extensive unimolecular desolvation will occur within the microsecond time frame (ion source residence time). For example, Sunner and Kebarle (52) studied the desolvation of K(H20),+ cluster ions possessing thermal internal energy distributions. They calculated that the degree of fragmentation increases with the size of the cluster ion. The fraction of K(H20)3+ions that dissociate within 2 ps is