Kinetically Labile Equilibrium Shifts Induced by the Electrospray

The complexation reactions between the alkaline earth metal ions and EDTA were studied by electrospray mass spectrometry to measure the change in ...
1 downloads 0 Views 72KB Size
Download PDF
Anal. Chem. 1999, 71, 4166-4172

Kinetically Labile Equilibrium Shifts Induced by the Electrospray Process Hongjun Wang† and George R. Agnes*

Department of Chemistry, Simon Fraser University, Burnaby, British Columbia, V5A 1S6 Canada

The complexation reactions between the alkaline earth metal ions and EDTA were studied by electrospray mass spectrometry to measure the change in concentration of the metal ion-EDTA complex (MY2-) in the gas phase relative to the solution-phase equilibrium concentration. This work focused on the solution pH range from 4 to 7 where there exists free metal ions in solution at equilibrium. The equilibrium shift, measured through quantitation of the increased abundance of the MY2- species in the gas phase, was largest for barium and smallest for magnesium. The cause of the net equilibrium shift of the MY2- species is the combined effect of an electrolytic increase in pH within the capillary plus an additional shift within the evaporating droplets. In a thin diffusion-limited layer created by the products of electrolysis mixing with the bulk solution at the ES capillary tip, the labile species reequilibrate at a new, higher pH. In the evaporating droplets, the formation of new labile species due to increased solute concentrations is kinetically controlled because the ion residence time in the droplet prior to desorption is only ∼5 µs. These results are briefly discussed with respect to the potential for utilizing electrospray mass spectrometry for kinetically labile equilibrium studies. The motivation for our obtaining detailed information about the electrospray (ES) process is that ES mass spectrometry (MS) could prove invaluable as a tool in the study of kinetically labile species in complex sample types. The potential for ES-MS in this application is that multiple solution species can be monitored simultaneously with high selectivity. The purpose of this work was to study well-understood kinetically labile systems by ESMS to learn how idealized systems are perturbed by the ES process. The reactions studied are the complexation reactions between EDTA and each of the alkaline earth metal ions. This investigation was necessary because it was not clear from the accepted mechanistic aspects of the ES source how labile equilibria would be perturbed in passage through an ES source. * Corresponding author: (e-mail) [email protected]; (fax) 604 291-3765; (phone) 604 291-4387. † Present address: W. M. Keck Foundation Biotechnology Resource Laboratory, Yale University, 295 Congress Av., BCMM 302, New Haven, CT 065360812.

4166 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

The present state of understanding how ions in solution are transferred to the gas phase in the ES process has been reviewed.1 The ES process is the sum of all steps involved in the transfer of ions from solution to the gas phase: electrical atomization to form isolated droplets that possess a net charge, shrinkage of these droplets by evaporation of solvent, and fragmentation (Coulomb fission) of the droplets due to the increased repulsive electric field, ultimately leading to very small droplets from which the gas-phase ions are generated. Though the major steps in this process are known, the mechanistic details of several of the steps in the process remain unknown or unresolved. For example, the increase in solute concentration in the evaporating droplets has not been experimentally measured. The difficulty in making such a measurement is that several Coulomb fission events are believed to have occurred prior to ion release. Kebarle has estimated that the ratio of the initial to final droplet volume (Vi/Vf) is ∼50.1 This could be a crucial parameter in the study of labile equilibria that are perturbed in response to a decreased droplet volume. The question of whether ions are locked on the droplet surface by the electric field2,3 or are able to exchange with solutes from the bulk of the droplet prior to desorption4,5 has yet to be conclusively answered. The ion generation step itself, where an ion in solution enters the gas phase, has yet to be directly investigated. Consequently, the experimental data necessary to reach a clear consensus regarding the two proposed mechanisms of ion desorption have yet to be reached.1,2 The observation that protonated amino acids6 or multiprotonated peptide and protein ions7 can be produced from basic solutions has yet to be fully explained, though the electrolytic nature of the electrospray source could offer an explanation for these observations.8-11 With these well-recognized caveats in mind, numerous groups have investigated the degree to which solution-phase equilibria (1) Kebarle, P.; Ho, Y. In Electrospray Ionization Mass Spectrometry: fundamentals, instrumentation, and applications; Cole, R. B., Ed.; John Wiley and Sons: New York, 1997; pp 3-63. (2) Fenn, J. B.; Rosell, J.; Meng, C. K. J. Am. Soc. Mass Spectrom. 1997, 8, 1147-1157. (3) Fenn, J. B. J. Am. Soc. Mass Spectrom. 1993, 4, 524-535. (4) Tang, L.; Kebarle, P. Anal. Chem. 1993, 65, 3654-3668. (5) Enke, C. G. Anal. Chem. 1997, 69, 4885-4893. (6) Mansoori, B. A.; Volmer, D. A.; Boyd, R. K. Rapid Commun. Mass Spectrom. 1997, 11, 1120- 1130. (7) Kelly, M. A.; Vestling, M. M.; Fenselau, C. C.; Smith, P. B. Org. Mass Spectrom. 1992, 27, 1143-1147. (8) Van Berkel, G. J.; McLuckey, S. A.; Glish, G. L. Anal. Chem. 1992, 64, 1586-1593. (9) Van Berkel, G. J.; Zhou, F. Anal. Chem. 1995, 67, 2916-2923. (10) Van Berkel, G. J.; Zhou, F.; Aronson, J. T. Int. J. Mass Spectrom. Ion Processes 1997, 162, 55- 67. 10.1021/ac981375u CCC: $18.00

© 1999 American Chemical Society Published on Web 08/31/1999

do or do not correlate with ion currents by ES-MS.6,10,12-18 Ayed et al. have reported binding equilibrium constants for proteinprotein and protein-ligand reactions,19 though solute clustering, known to be appreciable at millimolar concentration levels20,21 has also been observed at micromolar concentration levels.22,23 Other groups have reported differences in “desorption efficiency” for various species in thermodynamic equilibrium24 or in competition for surface charge site occupancy.4 In this work, the pH of an aliquot of a stock solution was adjusted using microliter volumes of 10 mM NaOH for the purpose of varying the degree that the alkaline earth metal ion was complexed by EDTA in solution. In solutions where there existed free metal ion in solution, the gas-phase concentration of the metal ion-EDTA complex (MY2-), as measured by the mass spectrometer, was found to be greater than the solution-phase concentration of MY2-. These results are analyzed with respect to understanding the path through the ES source taken by the ions detected in the mass spectrometer. EXPERIMENTAL SECTION The solution equilibrium between a metal ion and EDTA is accurately calculated in 100% aqueous solutions.25 For this reason, organic solvents were avoided in this work. All salts used in the preparation of the calibration solution sets were ACS grade (Sigma). For each alkaline earth metal ion examined, two solution sets were prepared, one called the kinetically inert set and the other the kinetically labile set. The instrument response for MY2was calibrated using the kinetically inert calibration set. All solutions in this set were at constant high pH to ensure that the complexation reaction between the alkaline earth metal ion and EDTA was complete (>99% of the metal ion in solution is complexed). As such, there were no kinetically labile species in these calibration sets. Within the kinetically labile calibration set, the solution pH of each individual standard was varied by adding microliter volumes of 10 mM NaOH, but the total concentration of metal ion was maintained constant. The free metal ion present (11) Van Berkel, G. J.; Giles, G. E.; Bullock, J. S.; Wendel, M. W.; Gray, L. J. Modeling the Electrolytic Processes within a metal Electrospray Emitter Proceedings of the 46th ASMS Conference on Mass Spectrometry and Allied Topics, Orlando, FL, May 31-June 4, 1998. (12) Guevremont, R.; Siu, K. W. M.; Le Blanc, J. C. Y.; Berman, S. S. J. Am. Soc. Mass Spectrom. 1992, 3, 216-224. (13) Le Blanc, J. C. Y.; Guevremont, R.; Siu, K. W. M. Int. J. Mass Spectrom. Ion Processes 1993, 125, 145-153. (14) Chillier, X. F. D.; Monnier, A.; Bill, H.; Gulacar, F. O.; Buchs, A.; McLuckey, S. A.; Van Berkel, G. J. Rapid Commun. Mass Spectrom. 1996, 10, 299304. (15) Mirza, U. A.; Chait, B. T. Int. J. Mass Spectrom. Ion Processes 1997, 162, 173-181. (16) Ding, J.; Anderegg, R. J. J. Am. Soc. Mass Spectrom. 1994, 6, 159-164. (17) Hiraoka, K.; Murata, K.; Kudaka, I. J. Mass Spectrom. Soc. Jpn. 1995, 43, 127-138. (18) Cunniff, J. B.; Vouros, P. J. Am. Soc. Mass Spectrom. 1995, 6, 437-447. (19) Ayed, A.; Krutchinsky, A. N.; Ens, W.; Standing, K. G.; Duckworth, H. W. Rapid Commun. Mass Spectrom. 1998, 12, 339-344. (20) Busman, M.; Knapp, D. R.; Schey, K. L. Rapid Commun. Mass Spectrom. 1994, 8, 211-216. (21) Meng, C. K.; Fenn, J. B. Org. Mass Spectrom. 1991, 26, 542-549. (22) Thomson, B. A. J. Am. Soc. Mass Spectrom. 1997, 8, 1053-1058. (23) Zhan, D.; Rosell, J.; Fenn, J. B. J. Am. Soc. Mass Spectrom. 1998, 9, 12411254. (24) Leize, E.; Jaffrezic, A.; Dorsseler, A. V. J. Mass Spectrom. 1996, 31, 537544. (25) Harris, D. C. Quantitative Chemical Analysis, 5th ed.; W. H. Freeman and Co.: New York, 1999.

at equilibrium in these solutions was kinetically labile. In the kinetically labile calibration sets for the Mg2+, Ca2+, and Ba2+ + EDTA systems, the mole ratio of the analytical EDTA concentration to the total alkaline earth metal ion concentration (100 µM) was fixed at 1.5:1. This ratio was 3.6:1 in the Sr2+ + EDTA kinetically labile calibration set, with the total strontium ion concentration constant at 42 µM. The method of internal standards was used throughout.26 The internal standard was NiY2- for the Sr2+ + EDTA study. I- was used as the internal standard for the Mg2+, Ca2+, and Ba2+ + EDTA studies. In addition to correcting for the possible differences in ion desorption, the internal standard minimized potential changes in the ES process due to small variations in ionic strength over the entire solution set. Though the instrument parameters for ion sampling were set to minimize water clusters, the possibility of a spectral overlap of CaY2-(H2O)1 on NiY2- existed. Rather than search for another suitable metal ion-EDTA complex, the use of I- as an internal standard was examined. Using either NiY2- or I- as the internal standard, identical results for the Ba2+ + EDTA system were obtained. The Mg2+ + EDTA system presented another potential spectral overlap. HNaY2-, abundant at pH >7, coincides with MgY2-. For this reason, K2EDTA and KOH were used in the Mg2+ + EDTA studies. The formation constants for the alkaline earth metal ions with EDTA are KBaY2- ) 6.03 × 107 M-1, KSrY2- ) 6.31 × 108 M-1, KCaY2) 1 × 1011 M-1, and KMgY2- ) 1.32 × 109 M-1.27 The complexation rate of the alkaline earth metal ions by EDTA is rate limited by the dissociation of a water molecule from the solvation sphere of the metal ion. The first-order rate constants for water exchange on the metal ions are kBa(II) ∼2 × 109 s-1, kSr(II) ∼4 × 108 s-1, kCa(II) ∼2 × 108 s-1, and kMg(II) ∼1 × 105 s-1.28 The off-rate for the MY2- complexes is e1 s-1. Instrumentation. The in-house-assembled pneumatically assisted ES source consisted of a sample delivery capillary (stainless steel, 0.1 mm i.d. × 0.21 mm o.d.) that protruded 0.3 mm from a second stainless steel tube (0.28 mm i.d. × 0.51 mm o.d.). Sample solution was pumped into the capillary at 20 µL/min. A N2 flow rate of 0.7 L/min. in the interspace between the capillary and tube assisted the electrical dispersion of the sample solution into droplets. With 100% aqueous solvents, electrical discharging will preclude analytical utility of the ES source for small inorganic ions with a net charge of >(1 unless steps are taken to minimize this problem. To enable stable ion currents for small inorganic ions in solution, an O2 bath gas at 0.7 L/min. was directed at the tip of the pneumatically assisted ES source using a 5-mm-i.d. glass tube. This arrangement permitted reproducible ion currents from 100% aqueous samples.26 The nebulizer gas flow of N2 and the discharge suppressing gas flow of O2 were at ambient laboratory temperature (293 K). The triple-quadrupole mass spectrometer (AROMIC 9100, SCIEX) was used as received from the manufacturer, except the corona discharge needle was replaced with an ES source. The ES source was biased to -3400 V to generate droplets with a net negative charge. Some of these droplets passed through a 3-mm(26) Wang, H. J.; Agnes, G. R. Anal. Chem. 1999, 71, 3785-3792. (27) Kotrly, S.; Sucha, L. Handbook of Chemical Equilibria in Analytical Chemistry; J. Wiley & Sons: Toronto, 1985. (28) Huheey, J. E. Inorganic Chemistry: Principles of Structure and Reactivity; Harper & Row: New York, 1972; pp 429-431.

Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

4167

diameter hole in the counter electrode (-600 V) and drifted toward the ion sampling orifice (-70 V) to the mass spectrometer. The 250-µm-diameter orifice, laser drilled into a 50-µm-thick stainless steel foil (Harvard Apparatus, Saint-Laurent, Canada), separated atmospheric pressure from the vacuum chamber in a single-stage expansion. The foil was seated in a stainless steel mount and sealed on both sides using aluminum O-rings. A curtain gas of N2 at 0.8 L/min, also at ambient laboratory temperature, flowed into the region between the orifice and the counter electrode to promote solvent evaporation from the droplets and to minimize solvent clustering onto the ions sampled into the vacuum chamber. q0 (rod offset -45 V) collected the ions sampled from atmospheric pressure and transferred them to the first mass analyzing quadrupole, Q1. The vacuum chamber was cryopumped to an operating pressure of 1.2 × 10-5 Torr with a cryoshell surrounding each of the two wire-fabricated rf-only quadrupoles, q0 and the collision cell, q2. No gas was introduced into q2 in this work. The difference in applied voltage bias between the orifice foil and q0 was chosen to maintain the ion count abundance of hydrated ions to less than 1% the ion count abundance of the bare ions. The current removed from the ES capillary by droplets was measured at the counter electrode. The voltage bias of this electrode was removed, and the current arriving at this plate was monitored using an in-house-constructed current-to-voltage amplifier. The average of 20 voltage readings were taken for each solution examined. For the entire range of solutions employed in this work, the measured current was 30 ( 10 nA. The reported gas-phase concentrations of MgY2-, CaY2-, and BaY2- in the solutions from the kinetically labile calibration sets were determined by summing the ion signal intensity ratios (analyte/internal standard) from 20 replicate 20-ms integration periods. The reported confidence level, plus/minus one standard deviation, was based on the linear regression of the instrument response calibration data acquired with the solutions from the kinetically inert calibration sets using identical experimental conditions. For the SrY2- gas-phase concentration measurements using the kinetically labile solution sets, the mean and standard deviation reported was based on five within-run replicates, each based on the sum of the ion signal intensity ratio (analyte/internal standard) from 20 replicate 20-ms integration periods. RESULTS AND DISCUSSION The measured concentrations of MY2- in the gas phase from the kinetically labile calibration solution sets are plotted as a function of pH in Figure 1 along with the calculated solution-phase equilibrium concentrations. The abundance of the MY2- species measured by the mass spectrometer represents the integral of MY2- released from all droplets, at all stages of droplet desolvation between the ES capillary tip and the sampling orifice. In all solutions that contained free alkaline earth metal ions at equilibrium with excess EDTA present, there was a greater abundance of MY2- in the gas phase. In comparing the data for the alkaline earth metal ion + EDTA systems plotted in Figure 1, the reaction for Mg2+ + EDTA was shifted the least, and Ba2+ + EDTA was shifted the most, in general agreement with the forward rate of these complexation reactions. For example, the test sample that had 21% of the total Ba2+ complexed at equilibrium in the solution phase yielded 63% complexed in the gas phase after having passed 4168 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

Figure 1. Comparison of gas-phase versus solution-phase concentrations of MY2-. The metal ions were (a) Mg2+, (b) Ca2+, (c) Sr2+, and (d) Ba2+; the calculated solution-phase concentrations of MY2(solid triangles, dotted line) and the measured concentration of MY2in the gas phase (error bars, solid line). The open circles represent the equilibrium concentration of MY2- in the capillary surface layer, calculated to be ∼400 µm long × 2 µm deep, created by the products of electrolysis diffusing into the sample solution at the end of the ES capillary.

through the ES source. By comparison, the test sample with 19% of the Mg2+ complexed in the solution phase yielded only 30% of the Mg2+ complexed by EDTA in the gas phase. Complexation Rate. The lifetime of a droplet in the ES source is finite, and the extent of new complexation for the alkaline earth metal ions with EDTA would be expected to differ according to the complexation rate of each reaction. The complexation reaction of the alkaline earth metal ions by EDTA can be generalized. Chelation of the metal ion by the HY3- and H2Y2- species, followed by rapid loss of H+, is slower than the reaction with the Y4species.29 In addition, the formation constant for complexation of a metal ion by the HY3- species is too small to be of importance.30 Hence, only the Y4- species is important in the complexation reaction, but because the EDTA species rapidly interchange form by proton loss or gain, the total EDTA concentration is considered in the complexation rate. The complexation rate of eq 1 is rate limited by water exchange on the alkaline earth metal ion, but solution concentrations of the metal ion and EDTA employed in this work were similar, necessitating the use of a second-order rate equation (eq 2). In eq 2, cA° and cB° are the available metal (29) Laitinen, H. A.; Harris, W. E. Chemical Analysis, 2nd ed.; McGraw-Hill: New York, 1975. (30) Kula, R. J.; Reed, G. H. Anal. Chem. 1966, 38, 697-701.

ion and EDTA concentrations, respectively, when t ) 0. At time t ) t, cA and cB are the available metal ion and EDTA concentrations, respectively. k is the water dissociation rate constant for the alkaline earth metal ion that rate limits the chelation reaction.

M2+(H2O)6 + Y4- ) MY2- + 6H2O

(1)

ln(cA/cB) ) (cA° - cB°)kt + ln(cA°/cB°)

(2)

Using the data points at pH 6.1 plotted in Figure 1d, the time required for the Ba2+ + EDTA system to increase the [BaY2-] from 21 to 63 µM in solution would be ∼7 µs. For the Ba2+ + EDTA system, the concentration difference between the ES-MS measured concentration of 63 µM and the calculated solution concentration of 21 µM could be explained by an equilibrium shift in desolvating droplets because the time required to achieve this concentration change is only 7 µs. This is much shorter than the droplet lifetime estimated to be ∼1 ms,31 and also on measured droplet velocities,32 so the assumption that the Ba2+ + EDTA equilibrium is able to keep pace with the droplet desolvation is valid, especially considering the ∼100-µs delay between successive Coulomb fission events that has been estimated by Kebarle.33 Either the extent of droplet desolvation shifting the position of the equilibrium reaction between Ba2+ and EDTA was very small prior to ion desorption or the residence time of the ions detected by the mass spectrometer in the evaporating droplet was very short. In marked contrast, the much slower reacting Mg2+ + EDTA system requires ∼12 ms to increase the [MgY2-] from 19 µM in the solution phase to 30 µM in the gas phase (Figure 1a, data points at pH 5.3). With an estimated droplet lifetime of 1 ms, ∼11 ms of reaction time in the chelation of Mg2+ by EDTA is left unaccounted by a droplet desolvation-induced path shifting eq 1 to the right. Increase of MY2- by Electrolysis in the ES Capillary. The electrolytic nature of the ES process, investigated in detail by Van Berkel, is another path by which eq 1 can be shifted to the right. Van Berkel has been shown that the position of the conductor for applying the dc potential need not be at the spray tip.34 Stable ion currents were obtained with the position of electrical connection separated from the spray tip by a 20-cm-long fused-silica capillary. With this configuration, the products of the electrolysis had sufficient time to diffuse throughout the bulk liquid in the capillary during transport (∼70 s) to the tip, changing the pH of the bulk solution. Following on the work of Van Berkel,10 the increase of [MY2-] in the gas phase due to the electrolytic nature of the ES source was attempted because the additional 11 µM concentration of MgY2- measured in the gas phase could be accounted for by a pH increase of ∼0.2. With respect to an electrolytic mechanism as the source of the charge imbalance on a droplet, four reduction reactions that (31) Ikonomou, M. G.; Blades, A. T.; Kebarle, P. Anal. Chem. 1990, 62, 957967. (32) Olumee, Z.; Callahan, J. H.; Vertes, A. Large Droplets and Velocity Compression in High Flow Rate Nanospray of Methanol-Water Mixtures. Proceedings of the 46th ASMS Conference on Mass Spectrometry and Allied Topics, May 31-June 4, Orlando, FL, 1998. (33) Kebarle, P.; Tang, L. Anal. Chem. 1993, 65, 972A-986A. (34) Van Berkel, G. J. In Electrospray Ionization Mass Spectrometry; fundamentals, instrumentation, and applications; Cole, R. B., Ed.; John Wiley and Sons: New York, 1997; pp 65-105.

Table 1. Reduction Reactions that Might Take Place in Negative Mode ES10 reaction

Ered (V vs SHE)

pH change

eq

O2(g) + 4H+ + 4e- ) 2H2O O2(g) + 2H2O + 4e- ) 4OH2H+ + 2e- ) H2 2H2O + 2e- ) H2 (g) + 2OH-

1.229 0.401 0.0 -0.83

increase increase increase increase

3 4 5 6

could be occurring are presented in Table 1. On the basis of formal reduction potentials only, alteration of the alkaline earth metal ion concentration by electrolytic reduction at the ES capillary in this work was not likely. Assuming that one of the reduction reactions listed in Table 1 was acting, there would have been a net increase in the pH of the solution. It was interesting to note that without the O2 gas flow, or with a flow of N2 used in replacement, electrical discharge always occurred. The O2 will suppress discharging by acting as a gas-phase electron scavenger, though diffusion of the O2 into the sample stream in the capillary, with subsequent participation in the electrolysis could also have been taking place. It is possible that the electrolytic reduction of O2 was a necessary reaction in maintaining the stable ion currents observed throughout this work with the 100% aqueous samples used and that eq 4 was the dominant electrolysis reaction taking place. In our ion source, droplets were generated at the end of a 4.25cm-long stainless steel capillary into which the sample solution was pumped at 20 µL/min. Diffusion coefficients for small ions, such as the H2Y2- species of EDTA that would react with the products of electrolysis (i.e., OH-), are on the order of 1 × 10-9 m2 s-1. Under these conditions, the products of electrolysis could diffuse a maximum of ∼33 µm into the capillary in ∼11 ms, but if the electrolysis products were generated uniformly along the entire length of this capillary, an average diffusion distance of 15 µm into the capillary would better approximate the mixing volume. The volume of bulk solution that the products of electrolysis (30 ( 10 nA) diffused into was estimated to be 170 nL/s. In the positive ion mode, Van Berkel has discussed the relative fraction of electrolysis of solution species versus oxidation of the ES capillary itself. Under favorable conditions (platinum ES capillary) nearly 100% of the current was electrolytic generation of H+.34 The assumption made by us was that 100% of the electrolytic current in our negative mode ES-MS experiments changed the solution pH by forming OH- according to eq 4, because the reduction of the free metal ions in these solutions, Na+ or K+ and an alkaline earth, was not favorable. Using Faraday’s law, the [OH-] injected into a capillary volume flux of 170 nL/s was 1.79 × 10-6 M, leading to a pH increase of only 0.05 because the uncomplexed EDTA buffers the bulk solution. Note that lower pH shifts would have been estimated if the assumption that 100% of the electrolytic current was being used to create OH- was incorrect or if the average diffusion distance of the electrolysis products into the capillary was greater than 15 µm. The assumption that the products of electrolysis were generated uniformly along the length of the conductor does not account for the increased [MgY2-] in the gas phase. The products of electrolysis are generated concurrently with the removal of negative charge from the capillary by droplets. If Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

4169

the electrolytic reactions in the ES source were localized at the capillary tip, then there would be less time for diffusion of the OH- generated by electrolysis to diffuse into the capillary, leading to locally higher pH shifts in a thin surface layer. On the basis of the measured additional complexation in the Mg2+ + EDTA system, 11 ms was set as the time available for mixing of the products of electrolysis with the solution flowing through the capillary. With a solution flow rate of 20 µL/min, only the last 400 µm of the capillary, plus on the blunt end of the capillary tip, is where the electrolysis would have been taking place. The maximum linear distance that small ions can diffuse in 11 ms is ∼ 3.3 µm. On the assumption that the electrolysis was uniformly distributed over the tip of the capillary, an average diffusion distance of 2 µm was used in the subsequent calculations. The volume flux of the region into which the products of the electrolytic current diffuse into and mix with the bulk sample is 290 pL/11 ms. Using Faraday’s law, the [OH-] injected into this mixing volume was 1.3 × 10-5 M, leading to a pH increase of only 0.21 because the uncomplexed EDTA buffers the bulk solution. The electrolytic induced pH increase in a thin surface layer has been calculated and plotted in Figure 1 for each solution tested based on the current measured at the counter electrode. The increased MgY2- concentration in the gas phase was used as the basis for deriving an estimate of the electrolytic pH shift, and clearly, the gas-phase concentration of the faster chelating reactions of Ba2+ and Sr2+ indicates that droplet desolvation also acts to shifts these reactions to the right. There are several sources of potential error in the interpretation of the electrolytic behavior of the ES source put forth in the previous paragraph. If the earlier assumption that 100% of the current generated creates OH- is not correct, then the size of the surface volume will need to be increased, though recent modeling by Van Berkel et al. indicates that the size of the region important for electrolysis is only in the last 150 µm of the ES capillary.11 Alternatively, if the solution flow rate on the blunt tip of the ES capillary is stagnant, then the size of the region on the capillary tip where the electrolysis reaction took place will have been slightly overestimated. Also, if diffusion is able to take place during the time that the ions are in the liquid cone jet, the volume of bulk solution into which the products of electrolysis diffuse into could be overestimated by a factor of e2. An additional source of error in these calculations, which is difficult to quantitate, is the effect of the electric double layer on the measured abundance of MY2- in the gas phase. The ionic strength of the solutions used in this work ranged from 0.9 to 1.4 mM, and under these conditions, the thickness of the diffuse layer at the electrodesolution interface would be expected to be on the order of ∼10 nm. In the diffuse layer covering a negatively biased electrode, the concentration of cations will be in excess of the bulk concentration, and the anionic concentration in this layer will be slightly lower than in the bulk. The extent to which the alkaline earth metal ion-EDTA complexation in the diffuse layer could be different from that in the diffusion layer is difficult to estimate. However, simple comparison of the volume of the diffuse layer to the diffusion layer, ∼1:250 pL, indicates that ions in the diffuse layer could comprise only 0.4% of the ion signal. Realize that, in the estimated 2-µm-thick diffusion layer, there will be no difference in the total metal and EDTA concentrations in this layer versus 4170 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

the bulk solution, apart from the increased complexation due to the higher pH within the diffusion layer. This is because the alkaline earth metal ions and EDTA ions were not expected to participate in the electrochemical reduction reactions taking place at the ES capillary tip. Chelation Reactions within the Evaporating Droplets. The extent of MgY2- formation in the ES process can be explained with the electrolytic nature of the ES process being confined to the tip of the ES capillary. Critical to this explanation is that no further mixing of the ions between the electrolysis-modified surface layer and the bulk took place, either in the cone jet or in the evaporating droplets. Had extensive mixing of ions between the surface layer and bulk taken place, much lower increases in the MgY2- would have been measured because the bulk solution inside the droplets contained a lower equilibrium abundance of MgY2-. Are the excess charges rigidly locked onto the surface of the droplets?2,3 Taken to the extreme, this view would describe EDTA ions that occupy a surface charge site on a droplet as being so rigidly held in space that chemical reactivity decreased, and the chelation step of the metal ion by the EDTA would have become the rate-limiting step. Equivalent degrees of MY2formation in the droplets would be expected if this were the case. Similarly, equivalent degrees of clustering of the alkaline earth metal ion onto the EDTA, brought about by desolvation, for the cases of Mg2+, Ca2+, and Ba2+ at 100 µM in the test solutions and a lower degree of clustering for Sr2+ at 42 µM would have been expected. The data do not support either of these two scenarios that would be indicative of anions being locked onto the droplet surface. Rather, the kinetics data presented earlier in discussing the results from Figure 1 indicate that the chelation reactions in the evaporating droplets are under kinetic control and that the time available for reaction within the droplet is very short. After correcting for the within-capillary electrolytic pH-induced shift to the right for the reactions of the alkaline earth metal ions by EDTA described by eq 1, the residual MY2- measured in the gas phase is plotted in Figure 2. The cause of this additional MY2- was attributed entirely to droplet desolvation. A maximum time of 5 µs for desolvation prior to ion desorption is based on the [BaY2-] in the gas phase not accounted for by the electrolytic pH change in the capillary (refer to Figure 1d). The solid curves drawn in Figure 2 represent the thermodynamic equilibrium abundance of MY2- in the droplet, originally of volume Vd that has shrunk by the factor f. The MY2- ions detected by the mass spectrometer imply a surprisingly low degree of droplet desolvation-induced equilibrium shift. If the droplets leaving the ES capillary tip are stable with respect to Coulomb fission, then quite extensive desolvation prior to ion release would be expected in both the primary and progeny droplets leading to much larger increases of [MY2-] in the gas phase, particularly for the Sr2+ and Ba2+ systems, than were measured. Conversely, if the droplets leaving the capillary tip are in excess of the Coulomb limit, then these droplets would fragment immediately without desolvation and the measured MY2gas phase can be rationalized. For example, assuming that 1-µmradius droplets are generated at the capillary tip, then the resulting first generation progeny droplets would be ∼0.1 µm in radius.33 Allowing for desolvation times of ∼5 µs, the volume of an originally

Figure 2. Positive error in the measured gas-phase MY2- concentration relative to the solution-phase equilibrium concentration in the electrolysis-modified surface layer. The metal ions were (a) Mg2+, (b) Ca2+, (c) Sr2+, and (d) Ba2+. Measured data points are indicated with solid circles. The solid lines represent the calculated equilibrium concentration of MY2- in the surface layer on a droplet originally of volume Vd that has decreased through desolvation by the factor f (Vd/f).

0.1-µm-radius progeny droplet would decrease by a factor of only 1.2, according to eq 9b from ref 32. Realize that the value of 1.2 is an average volume decrease factor due to desolvation as monitored by chemical equilibrium shifts and that individual droplets could have desolvated by larger volume decrease factors. If after 5 µs the first-generation progeny droplet now with a radius of 94 nm were at the Coulomb fission limit,35 it would necessarily contain ∼2000 excess elementary charges according to eq 8b from ref 32. The radius of the second generation progeny droplets would be ∼9.4 nm. In the second-generation progeny droplet of radius 9.4 nm, the number of excess charges is estimated to be 15 and the number of bulk analyte ions ∼0.25,32 based on an analyte concentration of 100 µM in the original sample solution. Continued evaporation of the remaining solvent from the secondgeneration progeny droplets would not further shift the reaction written in eq 1 to the right. Extending these calculations backward, the primary droplets leaving the ES capillary would have ∼266 000 excess elementary charges residing on a droplet 1 µm in radius. Such droplets would be in excess of the Coulomb fission limit by (35) Taflin, D. C.; Ward, T. L.; Davis, E. J. Langmuir 1989, 5, 376-384.

∼140%. It is not the primary droplets that require desolvation, rather it is the first- and second-generation progeny droplets that need to be desolvated in order to promote the release of ions from the droplets. Also, if these sample calculations are indicative of the true situation, then the ion evaporation theory would not be as good a model for the ion desorption step as is the single ion in a droplet theory.1 Within the primary droplets, the thickness of the electrolysismodified surface layer is an estimated 120 nm, based on the ratio of the volumes of the 400-µm-long, 2-µm-thick electrolysis-modified surface layer to the bulk solution in the capillary. An ion with a diffusion coefficient of 1 × 10-9 m2/s will travel ∼70 nm in 5 µs. Diffusion of an ion from the bulk through the 120-nm-thick surface layer and then desorbing within ∼5 µs is not highly probable, and EDTA or MY2- ions originating in the bulk solution would not be expected to contribute significantly to the ion signal detected by the mass spectrometer. These calculations imply that mixing of solutes between the surface layer and bulk could be taking place in the evaporating droplet. However, because the residence time in the droplet for the ions that desorb and are detected by the mass spectrometer is short, the effective concentration of the solutes in the electrolytic-modified surface layer was not measurably affected by diffusion. An ion with a larger diffusion coefficient, originally in the bulk solution in a droplet, could contribute to the ion signal observed by the mass spectrometer. Implications for Speciation Applications. In light of the results presented in this report, the alteration of kinetically labile solution species by the ES process can be predicted. ES-MS could therefore be applied to the difficult task of determining kinetically labile solution species. With the electrolysis localized at the tip of the ES capillary, a thin layer of the bulk solution reequilibrates at a new pH. In this sense, then, yes, the ion abundance in the mass spectrum does correlate with the pH of the droplet surface layer, but not the bulk. The electrolytic pH shift measured in this work can be significantly reduced by separating the metallic conductor from the spray tip with a fused-silica capillary that is long enough to allow time for bulk mixing by diffusion of the products of electrolysis, as has been demonstrated by Van Berkel.8,9 With the arrangement used by Van Berkel, the electrolytic pH shift would be very small for samples that possess buffering capacity, such as water samples that contain dissolved organic carbon materials. Only for chemical species that react fast, and measurably alter the species distribution during the ∼5-µs residence time in an evaporating droplet, will the solution-phase species distribution be challenging to determine. Even in this extreme, the amount that a kinetically labile solution species will change concentration in response to droplet desolvation will be less than 1 order of magnitude. The complexation of most metal ions in aqueous solution will be rate limited by predissociation of a water molecule, hence only the kinetically labile alkali metals, Hg2+, and from this work, Sr2+ and Ba2+ will cause quantitation challenges. Not included in this list would be the environmentally important metals ions Cu2+ and Cd2+ that have water exchange rates that are close to that for Ca2+, which did not undergo extensive change in species distribution in the evaporating droplet prior to desorption. Hence, the labile solution-phase species distributions for Cu2+ and Cd2+ might be accurately determined using ES-MS. Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

4171

CONCLUSIONS Shifts in the equilibrium position for the reaction between an alkaline earth metal ion being chelated by EDTA have been measured by ES-MS. The shift in these chelation reactions are all toward increased complexation, and the degree to which each equilibrium shifted was in general agreement with the rate of water exchange with the free metal ion. For example, the reaction between Ba2+ + EDTA shifted the most, and Mg2+ + EDTA shifted the least. Two steps in the ES process, distinguished in this work, cause the equilibrium shift. First, there is an electrolytic pH increase (negative ion mode) of a thin layer of solution at the tip of the ES capillary. The electrolytic increase in pH of only 0.21 in the surface layer (∼400 µm long × 2 µm deep) was small because the excess of EDTA from the chelation reaction buffered these solutions. Within the surface layer, the chelation reaction of an alkaline earth metal ion by EDTA reequilibrated in the ∼11ms delay required to pump this mixing volume of solution out of the ES capillary. A second step in the ES process to cause the chelation reaction to favor complexation is droplet desolvation. As solvent evaporates from these droplets, the increase in solute concentration favors higher degrees of complexation. The equi-

4172 Analytical Chemistry, Vol. 71, No. 19, October 1, 1999

librium shift in the droplets is kinetically controlled because the ions detected by the mass spectrometer remain in the surface layer of the droplets for ∼5 µs. Hence, only species distributions that can be measurably altered within this time period will be difficult to quantitate accurately. Also a result of the ∼5-µs period of time that ion desorption from the droplets takes place is the vast majority of the ions detected by the mass spectrometer were derived from the thin electrolysis-modified layer of solution originally formed within the capillary that was partitioned onto the surface of the droplets leaving the ES capillary tip. ACKNOWLEDGMENT Funding for this work was provided by NSERC and Simon Fraser University. The generous donation of the AROMIC instrument from SCIEX made this work possible.

Received for review December 11, 1998. Accepted July 19, 1999. AC981375U