Ambient Gas Influence on Electrospray Potential As Revealed by

Mar 27, 2007 - The effect of ambient gas on potentials inside the electrospray (ES) capillary ... readily adsorbed from the surrounding ambient gas in...
0 downloads 0 Views 217KB Size
Download PDF
Anal. Chem. 2007, 79, 3383-3391

Ambient Gas Influence on Electrospray Potential As Revealed by Potential Mapping within the Electrospray Capillary Boguslaw P. Pozniak and Richard B. Cole*

Department of Chemistry, University of New Orleans, 2000 Lakeshore Drive, New Orleans, Louisiana 70148

The effect of ambient gas on potentials inside the electrospray (ES) capillary was investigated. Potential measurements and differential electrospray emitter potential (DEEP) maps were obtained with the help of a small, movable, disklike platinum wire electrode inserted into the ES capillary. Typical solvents used for electrospray mass spectrometry such as methanol and mixtures of methanol-water and chloroform/methanol have been tested. It was found that oxygen is readily adsorbed from the surrounding ambient gas into the spraying liquid. Following adsorption, it resides in, or near to, the Taylor cone, thereby affecting the electrochemical potential near the ES capillary exit, as well as the character of the inherent electrochemical reactions occurring during the ES process. The potentials measured in an air environment with reactive oxygen present are contrasted against those obtained in an inert nitrogen environment. The kinetics of oxygen admission have been found to be quite fast, i.e., occurring in a matter of seconds, but it takes far longer to purge the system of oxygen by changing the ambient atmosphere to nitrogen. The oxygen effect is present in negative and positive ion modes of ES, but the total ES current is not affected by the change of ambient gas. The magnitude of the oxygen effect owing to ambient air was compared to the effect caused by initially dissolving oxygen in the solution prior to the start of ES; it was found that the presence of oxygen in the ambient gas has a far greater consequence. These results indicate that the presence of reactive gases, such as molecular oxygen, in the region of the ES emitter may have unintended secondary effects on the ES process prior to mass spectrometric analysis. In our investigations of electrochemical properties of electrospray (ES) operation, our attention was drawn to the fact that the ambient gas atmosphere must have an influence on electrochemistry inside the ES capillary. In every place where electrochemical reactions are present, such as in corrosion and passivation of metals, ambient gases do participate, but the importance of such participation has not been elucidated for the electrochemical processes inherent to ES. Electrochemical experiments are customarily run under inert gas atmospheres using carefully deaerated solutions to prohibit * To whom correspondence should be addressed. E-mail: [email protected]. 10.1021/ac062329u CCC: $37.00 Published on Web 03/27/2007

© 2007 American Chemical Society

oxygen from being involved in electrode processes. By contrast, in a regular electrospray mass spectrometry experiment, neither of the above-mentioned procedures is routinely performed. In pneumatically assisted electrospray,1 however, nitrogen, argon, or even air is added to aid in droplet shearing. As will be shown, the injudicious addition of gases in the region of the ES emitter may have unintended effects. Oxygen is both ubiquitous in the environment and highly reactive; i.e., it readily enters many electrochemical processes.2 It constitutes about a 21% mole fraction of dry air; the remainder is nitrogen and small amounts of even less active noble gases. Being chemically inert, nitrogen is often used to create a protective atmosphere. Only under rare and extreme conditions can electrochemical oxidation of nitrogen or noble gases occur.3 One can safely assume that, in dry clean air, oxygen is the only electrochemically reactive gas. Previously, we presented systematic studies of the effects of different solution parameters on potential and current maps in the ES capillary operating in the positive4 and negative ion5 modes. This presentation aims at showing the influence of the ambient atmosphere on the distribution of electrochemical potential in the ES capillary. We also investigate the kinetics of the change of measured potential (variation as a function of time) in response to a change in ambient gas. To examine these effects, we have performed ES experiments on a series of commonly used electrolytes in a controlled ambient environment. The initial goal pursued here is to document phenomena attributable to the ambient atmosphere through mapping of potential gradients while contrasting ES operational characteristics observed under neutral gas versus those of operation in air. In a later phase, we plan to couple the device to a mass spectrometer to allow monitoring of the actual products. Theoretical Background of the Method. The electrospray phenomenon is achieved by placing a liquid-containing capillary in a high electric field.6 The field is produced by setting an electric potential difference between the capillary and a second electrode (1) Bailey, A. G. Electrostatic Spraying of Liquids; John Wiley & Sons: New York, 1988; pp 60-89. (2) Sawyer, D. T.; Sobkowiak, A.; Roberts, J. L. Electrochemistry for Chemists, 2nd ed.; John Wiley & Sons: New York, 1995; pp 358-401. (3) Dibble, T.; Bandyopadhay, S.; Ghoroghchian, J.; Smith, J. J.; Sarfarazi, F.; Fleischmann, M.; Pons, S. J. Phys. Chem. 1986, 90, 5275-5277. (4) Li, Y.; Pozniak, B. P.; Cole, R. B. Anal. Chem. 2003, 75, 6987-6994. (5) Pozniak, B. P.; Cole, R. B. J. Am. Soc. Mass Spectrom. 2004, 15, 17371747. (6) Bailey, A. G. Electrostatic Spraying of Liquids; John Wiley & Sons: New York, 1988; pp 10-41.

Analytical Chemistry, Vol. 79, No. 9, May 1, 2007 3383

positioned a few millimeters in front of it. Typically, sprayed solvents of moderate polarity can be considered as being partially conductive (leaking) dielectrics.7 The electric field inside such a liquid is a superposition of the external field(s) with contributions from all charged species and dipoles in solution. Relevant equations of electrodynamics and hydrodynamics, solved within appropriate boundary conditions, can provide a detailed description. However, in practice, such ab initio treatment remains quite challenging.8-10 Immediately upon imposition of the high electric field used to begin the electrospray process, the initial distribution of charges in the liquid starts to shift.11,12 After a certain time period, a steady state of charge production/consumption is reached that comprehends coexisting electric potential gradients, chemical potential gradients, and liquid flow patterns. In that complex environment, ions will move by migration, diffusion, and bulk flow transport. The combined effect of all forces will lead to a distribution of concentrations of molecules in the capillary and will establish potentials at the electrode-solution interfaces.13,14 An electrode placed in a liquid will sense the presence of species in its vicinity through processes of adsorption, inductive charging and charge transfer at the surface15 to determine the electrode potential. The electrochemical potential of the electrode should not be confused with the electric potential gradient (field) outside of (or inside) the electrolyte. Electrospray operates as a controlled-current electrochemical cell,16 implying that the current of the cell remains constant regardless of (small) changes in the composition of electrolyte between the electrodes. Certain changes, such as depletion of one species and accumulation of others, may occur as a direct result of cell operation. In contrast to a conventional electrochemical cell, however, in an electrospray source, the total current is not directly imposed by the experimentalist, but rather, it is a response of the system to a variety of experimental parameters and sprayed liquid properties such as electrical conductivity, surface tension, viscosity, flow rate, and so on. The portrayal of the electrospray device as an electrochemical cell implies that the total ES current flows through the sprayed liquid to the metal capillary walls via a variety of paths. The (7) Saville, D. A. Annu. Rev. Fluid Mech. 1997, 29, 27-64. (8) West, A. C.; Newman, J. Determination of Current Distributions Governed by Laplace’s Equation. In Modern Aspects of Electrochemistry; Conway, B. E., Bockris, J. O’M., White, R. E., Eds.; Plenum Press: New York, 1992; Vol. 23, pp 101-148. (9) Ibl, N. Current Distribution. In Comprehensive Treatise of Electrochemistry; Bockris, J. O’M, Conway, B, E, Yeager, E., Eds.; Plenum Press: New York, 1980; Vol. 6, pp 239-313. (10) Van Berkel, G. J.; Giles, G. E.; Bullock, J. S., IV; Gray, L. J. Anal. Chem. 1999, 71, 5288-5296. (11) Blades, A. T.; Ikonomou, M. G.; Kebarle, P. Anal. Chem. 1991, 63, 21092114. (12) Van Berkel, G. J.; Asano, K. G.; Schnier, P. D. J. Am. Soc. Mass Spectrom. 2001, 12, 853-862. (13) Ibl, N. Fundamentals of Transport Phenomena in Electrolytic Systems. In Comprehensive Treatise of Electrochemistry; Bockris, J. O’M., Conway, B. E., Yeager, E., Eds.; Plenum Press: New York, 1980; Vol. 6, pp 1-63. (14) Levich, V. Physicochemical Hydrodynamics; Prentice Hall: Englewood Cliffs, NJ, 1962; pp 231-371. (15) Trasatti, S. The Electrode Potential. In Comprehensive Treatise of Electrochemistry; Bockris, J. O’M., Conway, B. E., Yeager, E., Eds.; Plenum Press: New York, 1980, Vol. 1, pp 45-82. (16) Van Berkel, G. J. The Electrolytic Nature of Electrospray. In Electrospray Ionization Mass Spectrometry; Cole, R. B., Ed.; John Wiley and Sons, Inc.: New York, 1997; pp 65-105.

3384 Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

Figure 1. Schematic of experimental setup. Capillary diameter is exaggerated for visibility. (a) gas inlet; (b) solution inlet; (c) fusedsilica shield surrounding wire probe electrode; (d) mechanical movement of wire probe electrode; (e) connecting cross; (f) electrochemical workstation; (g) high-voltage supply; (h) wire attached to reference electrode lead; (i) wire attached to working electrode lead; (j) highvoltage counterelectrode; (k) Taylor cone “skin”, where electrochemically active gas is adsorbed; (l) disk-shaped end of wire probe electrode in contact with sprayed solution; (m) Faraday cage.

distribution of the total current into local current paths will be governed in such a way so as to minimize the overall dissipation (loss) of energy. In simple words, the current will flow where the resistance is the lowest. The places of highest (faradaic) current densities at the electrode (ES capillary) surface will be at points that allow the lowest resistance path to the tip of the Taylor cone. Locations further away from the tip, sitting deeper inside the capillary, will also pass a certain amount of current. This stems from the fact that, at a certain level of current density, the added potential necessary to achieve an even higher current density is more than the extra potential necessary to overcome additional solution resistance encountered when electrons are exchanged further inside the ES capillary. Such behavior will create gradients of interfacial potential and current density along the ES capillarysolution interface. Maps of such gradients inside the ES capillary were presented in previous publications,4,5,17 and experimental limitations relating to such measurements were discussed. Of course, local interfacial potential and local current densities are not independent variables;13,14 thus, the potential maps presented in this paper also reflect the distribution of current in the ES capillary. However, for a more complete picture of surface processes, it is useful to independently map the current because current and potential measurements are subject to different limitations and errors.17 EXPERIMENTAL SECTION The instrument (Figure 1) was described previously.4,5 It consists of a standard electrospray system equipped with a platinum (ES) capillary (i.d. ) 0.5 mm, o.d. ) 0.6 mm, 99.95% Pt; Goodfellow Cambridge Ltd., Huntington, England). A platinum wire probe (diameter 0.127 mm, 99.9% Pt, Alfa Aesar) is inserted into the capillary. The wire is sealed in a fused-silica tube (o.d. ) (17) Pozniak, B. P.; Cole, R. B. J. Am. Soc. Mass Spectrom. 2007, in press.

0.32 mm; SGE Inc., Austin, TX) that is approximately concentric with the surrounding ES capillary. The silica tube and the epoxy resin sealant isolate the wire from the liquid everywhere except for the extreme end where only the disk-shaped exposed surface of the wire makes contact with the electrolyte. The wire tip was flame annealed to smooth the exposed disklike surface prior to sealing, as opposed to the previously employed method of polishing with an abrasive medium. The wire probe electrode enters the ES capillary through a mixing “cross”; another arm of the cross is used to deliver the liquid to the ES capillary by syringe pump (Orion 341B, Sage Instruments, Boston, MA). The design of the potentiostat (660A electrochemical work station, CH Instruments, Austin, TX) is such that the working electrode port is hardwired to a common ground that we assign as “zero potential”. The ES capillary is grounded through an ammeter, thus enabling measurement of the total electrospray current. The wire probe electrode is connected to the “reference electrode” port. The potentiostat displays the potential of the working electrode with respect to the reference electrode. The electrospray high voltage (typically 3300-3600 V, power supply by Glassman, High Bridge, NJ) is applied to a brass plate counterelectrode located at 8 mm from the capillary tip. The highvoltage hazard to the operator is minimized by the placement of a cylindrical Faraday cage around the entire emitter and counterelectrode assembly (Figure 1). Chemicals were used as supplied (without further purification) by the following manufacturers: MeOH (HPLC grade) by EMD Chemicals (Gibbstown, NJ); chloroform (chromatography grade) by J. T. Baker Inc. (Philipsburg, NJ); and KCl by Mallinckrodt (St. Louis, MO). Deionized water (18.2 MΩ) was obtained using a Milli-Q water system (Millipore, Billerica, MA). Gases (air and nitrogen) were from Nordan-Smith (Hattiesburg, MS). Solutions were not deaerated except when specified. All experiments were done with a low gas flow rate (∼80-100 mL/min) to minimize the problem of a drying effect caused by flowing gas. Our coordinate system employed for all differential electrospray emitter potential (DEEP) maps is such that position “zero” of the wire probe electrode corresponds to the point where the probe tip is flush with the plane of the ES capillary exit as determined by visual observation through a microscope; all other positions are deduced from the velocity of the mechanical positioner, which was 10 µm/s. Positive values mean that the wire electrode tip is outside the capillary; i.e., it has entered into the Taylor cone; negative values imply that the wire probe electrode is inside the ES capillary. The error in assigning the zero position is less than 0.05 mm, and any further error introduced by the mechanical positioner is negligible. RESULTS AND DISCUSSION Method and Associated Errors. Two kinds of experiments are presented. The first, potential maps (DEEP maps) were acquired by a previously described method,4,5 i.e., recording the potential difference between a wire probe electrode and the surrounding (grounded) ES capillary, as the wire probe is continuously moved inside the ES capillary. An open circuit potential reading (zero current flow) is taken every 0.1 s. Displayed potential maps have been smoothed with a 30 or 60 adjacent point averaging routine. In the second type of experimentskinetics of ambient gas changesthe wire probe electrode was placed in a

fixed position (for example, x ) 0) and the system was allowed time to stabilize. Next, the gas supply was manually switched by closing and opening clamps on flexible tubing leading to the ES compartment, and the evolution of the measured potential was then monitored. Open circuit potential (OCP) measurements reflect the concentration of species in the electrode vicinity that are capable of adsorbing or transferring charge to the electrode. There are two main sources of error in the performed OCP measurements: (1) the absence of a true reference electrode; and (2) the unknown ohmic resistance of the solution. Without a proper reference electrode, the potential difference is being measured between the exposed end of the platinum wire, where redox couples are not necessarily reversible, and the surrounding (grounded) platinum capillary. Further complicating matters is the fact that one electrode (e.g., an upstream point of the ES capillary) may produce electrochemical products that alter the conditions at the other electrode (wire probe). This may lead to an alternation or a slow drift in measured potential values. In principle, no current should flow between electrodes in OCP measurements, but, in practice, a potentiostat requires a very small current, typically in the picoampere range; this current flow slightly alters the read potential. The error is often imperceptible in electrochemistry of conductive solutions, but it can be substantial for liquids of low conductivity (e.g,, neat organic solvents). The current drawn in OCP measurements is inversely proportional to the sum of cell resistance and potentiostat internal resistance; if the cell resistance is very high and it approaches the potentiostat internal resistance, the error in potential measurements becomes significant. The measured value for the electrochemical cell potential is thus always an underestimation of the true value. It should also be noted that when cell geometry changes during the course of wire probe electrode movement, displacement of solution will cause current paths to vary, and total resistance will also change, even though the shape of the Taylor cone is not visibly altered. These effects, which are more significant for measurements made with the wire probe inside the Taylor cone, alter the measured values obtained when the probe is moved in either direction, shifting them closer or further from the actual values. For example, when the wire probe electrode is located further into the Taylor cone, it has a longer resistance path to the ES capillary, but a shorter resistance path to the end of the Taylor cone. The situation is further complicated by the fact that measurements are made in a flowing solution that is not well mixed; i.e., specific conductivity may vary significantly from one location to another. In that case, potential readings would fluctuate even when the electrode geometry remained fixed. Total Current. Figure 2 presents a measurement of total electrospray current as the ambient gas is changed. The ES current corresponds to positive ion mode operation; thus, according to convention, it is negative or anodic (oxidative). The total current value is steady and altogether indifferent to the changes in ambient gas. Under the employed low current operating conditions (below the threshold for corona discharge), all ES current must flow across the ES capillary-solution interface, then through the solution, and finally through the Taylor cone filament. The current Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

3385

Figure 2. Total current to the capillary vs time as ambient gas is varied. Positive ion mode, 100% methanol with 10 µM KCl supporting electrolyte. Due to convention, the oxidative current has a negative sign.

controlling feature can be considered18 to be located at the end of the Taylor cone, where a filament of liquid is formed. In that place, the balance of forces (e.g., electrostatic interactions of free charges and physicochemical properties of the liquid) allows for only a certain amount of current to be passed. One of the solution parameters that has a large influence on the total current is solution conductivity. Molecular nitrogen or oxygen when dissolved in a liquid is not expected to change its overall conductivity, and in all experiments performed, using different combinations of water, methanol, and chloroform with various supporting electrolytes, no change of total current was ever observed as a result of changing the ambient gas from nitrogen to oxygen. Small oscillations of the current appearing in Figure 2 are commonly observed.17 These peak-to-peak oscillations constitute ∼2% of the total current, and they may be a consequence of ripple in the liquid supply line. Notably, a change in the ambient gas causes no phase shift or change in the amplitude of oscillations. In considering ES as a constant-current electrochemical cell, it may be concluded that the current regulator feature of ES is not influenced by the presence of electroactive oxygen, but rather, the electrochemistry at the ES capillary adjusts itself to feed the current demand of the system and not vice versa. Negative Ion Mode. The change in wire electrode potential resulting from a change in ambient gas atmosphere for sprayed methanol is shown in Figure 3a. The position of the wire probe is fixed at +0.3 mm; i.e., it is inside the Taylor cone and is well exposed to the ambient environment. The difference of measured potential in air versus nitrogen environments (∆E) is ∼150 mV for the stationary wire probe. This ∆E value also corresponds to that shown for the Figure 3b DEEP maps. This consistency is evidence of good reproducibility in data, and it confirms that the potentials plotted in the DEEP map have reached stable levels. Figure 3a also shows that changes in measured potential caused by the presence of a reactive ambient gas are reversible; i.e., upon a switch back to the initial gas, the potential returns to the initial level with no permanent “poisoning” of the electrode. The shapes of the DEEP maps shown in Figure 3b are typical. Potential maps acquired in the presence of nitrogen are always

smooth and monotonically rising with increases in x value; they consistently undergo a steep rise as the Taylor cone tip is approached. On the other hand, DEEP maps obtained in the presence of oxygen often show unexpected “dips” and deviations from monotonicity at wire probe positions near the ES capillary exit. The origin of these dips will be discussed in the conclusion. In the negative ion mode of operation, the electrochemical process responsible for charge production and the flow of current is reduction. A variety of chemical entities that are available at the electrode may accept electrons (i.e., be reduced), not just the major solution component(s). These species may include unintentionally added trace contaminants, residual water, or dissolved gases. The solution under investigation in the negative ion mode in Figure 3 is methanol. Electrochemical reduction of methanol typically proceeds to hydrocarbons and hydroxide anion;19 oxygen is not required in this reaction. When dissolved in solution, however, molecular oxygen is readily reduced to peroxide. Generally speaking, the reduction of molecular oxygen has a rich

(18) de la Mora, J. F.; Loscertales, I. G. J. Fluid Mech. 1994, 260, 155-184.

(19) Popp, F. D.; Schultz, H. P. Chem. Rev. 1962, 62, 19-40.

3386 Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

Figure 3. (a) Negative ion mode measured potential vs time as a function of ambient gas using a solution of 100% methanol with no added supporting electrolyte. The total electrospray current is 13 nA, and the wire electrode is fixed at position x ) + 0.3 mm. (b) Negative ion mode DEEP maps obtained in nitrogen and air atmospheres using the same solution as in (a). Arrows in (b) indicate the position of the wire electrode used to generate (a), and the ∆E values obtained in the two experiments correlate well with one another.

Figure 4. Negative ion mode DEEP maps obtained using methanol with 10 µM KCl or a methanol-chloroform (90:10) mixture with 10 µM KCl. The two solutions were each tested in the presence of air and in the presence of nitrogen.

chemistry with many branching pathways; solvent and local surface characteristics will dictate which mechanism(s) are favored.2,20 Most electrode processes are preceded by some form of adsorption at the electrode surface. Adsorption may also lead to electrochemically induced dissociation, and the whole process may involve many elementary steps. While it is not possible to state the exact details of the electrochemical reactions involving oxygen in our system, it is likely that molecular oxygen is reduced to some anionic or radical form, which in turn reacts with methanol. The presence of chemical reduction products of oxygen that may react with methanol does not change the balance of charges in the system; i.e., the number and rate of production of excess negative charges does not change relative to the solution devoid of oxygen. When oxygen is present, however, the metalsolution interface potential must correspond to the energy characteristic of oxygen reduction and not of methanol reduction. For an electrochemical reaction involving oxygen at the ES capillary, there are two possible sources of molecular oxygen: oxygen adsorbed from ambient air through the Taylor cone skin and oxygen initially dissolved in the liquid. Molecular oxygen can react with water as well as with methanol. An additional source of oxygen atoms is from water, which is a common contaminant in all organic solvents unless they are dried under the strictest conditions. Admittedly, in our “pure” methanol, water is present in up to millimolar amounts. In our experimental setup, ES current is typically in the range of 10-150 nA. In the examples presented in Figures 3 and 4, total ES current was only 14 nA, i.e., 14 nC/s. This value multiplied by the Faraday constant reveals that only 0.14 pmol (∼84 billion molecules) per second needs to accept one electron. With liquid flow containing on the order of 1.5 mmol of methanol or water per second, it can be seen that only a very tiny fraction of all molecules in the liquid must participate in electrochemical reactions in order to supply the required amount of current. Contaminants present at millimolar or even micromolar levels are abundant enough to be the sole electron donors or acceptors or to act as intermediates in electrochemical reactions. (20) Vetter, K. J. Electrochemical Kinetics; Academic Press: New York, 1967; pp 632-633.

Actual contaminant participation may be restricted if hydrodynamic conditions limit their availability at the electrode. Detailed studies of tubular flow cells have shown that not every available molecule will undergo electrochemical reaction.21-23 Nonetheless, if a dissolved contaminant is highly electroactive, then it is likely to undergo reaction to furnish ES current, regardless of the solvent that it is dissolved in. Thus, even in a protective atmosphere of neutral gas, it is not always the major component of the liquid that undergoes electrochemical transformation. Molecular oxygen, when present, competes not only with the solvent but also with contaminants whose presence may not even be anticipated. In general, reduction of molecular oxygen, regardless of the specific reaction pathway, is always favored over reduction of methanol to hydrocarbons.2,19 Indeed, our DEEP maps in Figure 3b reveal lower measured potential values at all wire probe positions when oxygen is present. The changes to the potential map acquired in the oxygen-containing atmosphere thus clearly reflect the presence of electroactive species which are absent under protective nitrogen. Figure 4 shows a more complex problem of potential maps for two solutions: pure methanol and a mixture of methanol and chloroform. Both solutions contain 10 µM KCl supporting electrolyte that serves to do the following: (1) increase liquid conductivity, which always leads to a reduction of the interface potential at the tip of the ES capillary (x ) 0); and (2) increase the total ES current. The potential maps in Figure 4 illustrate several trends. In comparing the two DEEP maps obtained in the presence of nitrogen, it can be seen that addition of chloroform moves the potential map upward, toward more extreme potentials at all positions, with both curves rising monotonically. When comparing the two Figure 4 maps obtained in the presence of oxygen, it is noted that they have remarkably similar shapes and that the addition of chloroform again pushes the entire DEEP map upward. Thus, when the methanol data are compared with that of the methanol-chloroform mixture (in either air or nitrogen), chloroform addition causes a notable upward shift in the DEEP map at all points. Normally chloroform is more easily reduced than methanol, but the presence of Cl- disfavors the reduction.24 Thus, the supporting electrolyte, potassium chloride, appears to participate in establishing the reaction equilibrium. The combination of the above effects creates a situation whereby, although chloroform is normally more easily reduced than methanol, the reduction is hindered. We have observed distinct temporal behaviors of the electrode potential in methanol in the presence and absence of KCl,5 which further supports the argument that that the presence of electrolyte is affecting the preferred reduction pathways. Oxygen and nitrogen tracks in Figure 4 are distinctly different. For both tested solutions, the presence of molecular oxygen causes a “kink” to develop in the DEEP map near the ES capillary exit. Moreover, DEEP maps for each individual solution, obtained in the absence and presence of oxygen, cross at a certain point indicating a coincidental identical measured potential for two different electrode reactions at that crossing point. In switching (21) Alkire, R.; Mirarefi, A. A. J. Electrochem. Soc. 1973, 120, 1507-1515. (22) Alkire, R.; Mirarefi, A. A. J. Electrochem. Soc. 1977, 124, 1043-1049. (23) Van Berkel, G. J. J. Am. Soc. Mass Spectrom. 2000, 11, 951-960. (24) Fry, A. J. Synthetic Organic Electrochemistry; Harper & Row Publishers: New York, 1972; pp 170-188.

Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

3387

Figure 6. Positive ion mode DEEP maps obtained in the presence and absence of oxygen using two methanolic solutions containing either 10 or 100 µM supporting electrolyte (KCl). Inset: changes of measured wire electrode probe potential with changes in ambient gas atmosphere.

Figure 5. Negative ion mode measured potential vs time as a function of ambient gas using solutions of: (a) 90:10 methanolchloroform with 10 µM KCl and a fixed wire electrode position (x ) 0). The value of ∆E corresponds to the difference in the DEEP map curves (Figure 4) of the same solution obtained in the presence and absence of oxygen. (b) 50:50 methanol-chloroform mixture (no added salt); position of the wire electrode is x ) 0.4 mm; (c) 50:50 methanolchloroform mixture with 100 µM KCl. Position of wire electrode is x ) 0.0 mm.

the ambient gas from nitrogen to oxygen, both employed solutions yield higher potentials at the tip of the ES capillary (x ) 0), but the measured potential goes down more quickly (has a steeper slope) as the wire probe is moved inside of the ES capillary. The higher potential at x ) 0 for the methanol-chloroform mixture in the presence of ambient oxygen appears to pose an inconsis3388 Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

tency. With oxygen being reduced more easily than methanol, one might always expect a lowering of measured potentials in its presence. In an oxygen environment, it is possible that a different redox couple, with higher exchange current, takes over the electrode kinetics, and a depletion of a reduced form of that couple shifts the measured potential to more positive values. To verify that there was no error in measurement, one may compare the Figure 4 DEEP maps with data shown in Figure 5. Figure 5 illustrates the change of potential caused by a change of ambient gas atmosphere with the wire probe fixed at one position. The solution in Figure 5a is the same methanol-chloroform mixture used to generate two DEEP maps in Figure 4. Immediately upon changing the gas atmosphere from air to nitrogen, the measured potential jumps to a higher value, and then it enters a period of slow descent and stabilization. The initial extremes are not accompanied by any change in shape or position of the Taylor cone, nor by a change in the total current to the ES capillary. They evidence an abrupt change of chemistry in the presence and absence of oxygen. The values of ∆E in Figure 5a agree well with potential maps (compare to difference in x ) 0 values for Figure 4 MeOH/CHCl3 curves). Figure 5b shows the same phenomenon, but with an even higher fraction of chloroform and a more exposed position (more positive x value) of the wire probe. Under these conditions, the oxygen influence sets in faster. On the other hand, an increase in conductivity caused by an increase in concentration of KCl supporting electrolyte to 100 µM reduces the magnitude of the effect (Figure 5c). The difference in potential (∆E) in the absence and presence of molecular oxygen is now only 15 mV. In addition, oscillations of measured potential are now large compared to ∆E, but the response of the cell to the gas change becomes smoother, without any unusual kinks. The above is consistent with our previous observations4,5,17 that an increase in conductivity of the liquid leads to a flattening (reduction of slope) of potential (and current) maps because the distribution of the total current along the ES capillary walls in more conductive solutions becomes more even. Positive Ion Mode. Oxidation is the dominant electrochemical reaction responsible for current generation at the electrode in

Figure 7. Positive ion mode DEEP maps for a methanol-water (50:50) solution with 100 µM KCl. There are three solutions: (1) carefully deaerated (red), (2) prepared and stored without any special precautions in the laboratory atmosphere (green), and (3) bubbled with air for 1 h in order to saturate with oxygen (blue). Each solution was run in an ambient atmosphere of nitrogen and again under oxygen (air). Note that the bottom three curves near position x ) 0 become the top three curves at positions more negative than x ) -1 mm.

positive ion mode electrospray operation. The very name of the process reminds one that the presence or absence of oxygen may be a critical factor. Oxygen participation in electrochemistry is clearly visible in Figure 6. Four DEEP maps are shown corresponding to two ambient gas conditions and to two concentrations of supporting electrolyte (KCl). The inset shows the kinetics of gas change for 10 µM KCl in methanol at a fixed wire probe position of x ) 0. The shift in measured potential as a function of ambient gas (∆E in the inset) is in excellent agreement with the DEEP map data for this solution at x ) 0. The effect of increased solution conductivity on the DEEP maps obtained under a protective atmosphere of nitrogen agrees well with previous observations;4,5 i.e., the solution with higher conductivity gives a map with higher potentials at all positions, but the slope of the plot is not as steep in the vicinity of x ) 0 as that of the lowconductivity solution. For both solutions, the presence of oxygen shifts the potential maps downward. It also changes the reaction at the very tip, producing a local minimum in potential, a kind of “dip” that was previously observed in the negative ion mode that will be further discussed in the conclusion. To investigate whether a distinction exists between the effects of oxygen dissolved in the initial solution and oxygen entering into solution via adsorption near the ES capillary exit, six DEEP maps were obtained as shown in Figure 7. Oxygen dissolves in most common liquids and its concentration may reach 2 mM for water and alcohols.25 The employed water-methanol solution is either untreated, deaerated, or oxygenated; each of the three solutions were run in nitrogen and oxygen atmospheres. Oxidation of methanol on noble metal surfaces is a thoroughly studied reaction, and many details are known.26 Regardless of the source of oxygen, the reaction products are the same: carbon dioxide and hydronium ions. Moreover, the reaction is irreversible, pH (25) Sawyer, D. T.; Sobkowiak, A.; Roberts, J. L. Electrochemistry for Chemists, 2nd ed.; John Wiley & Sons: New York, 1995; p 359.

affects the kinetics, and molecular (dissolved) oxygen is a better supplier of oxygen than water, which donates oxygen through oxidation of hydroxyl ions either prior to or in concert with the oxidation itself. It is seen in the Figure 7 DEEP maps that the presence of initially dissolved oxygen makes only a moderate, although noticeable difference. The solution with the highest amount of dissolved oxygen, i.e., the one through which air has been bubbled, only slightly enhances the oxygen effect when oxygen is available in the atmosphere. Moreover, in nitrogen atmosphere, dissolved oxygen produces only a slight dip near x ) -0.3 mm in the DEEP map. That dip is clearly minor compared with any run where oxygen is present in the ambient gas; thus, it is the presence of atmospheric oxygen that is the dominating factor in creating the observed dip. Molecular oxygen’s adsorption by the exposed part of liquid and its subsequent transport toward the inside of the ES capillary where it becomes involved in electrode processes results in a thorough change of reaction pathways and, potentially, of reaction products. Judging from the relative magnitude of oxygen-related dips in the Figure 7 DEEP maps, one may conclude that deareating solutions prior to ES may not always have a large effect on inherent ES electrochemistry, but the control of the ambient atmosphere is a larger factor. These results may also be contrasted with those plotted in Figure 6, where oxidation was conducted in HPLC-grade methanol. Although, commercial methanol contains millimolar amounts of water, it is not clear to what extent small quantities can participate in the reaction, for example, owing to restricted transport to the electrode surface. Millimolar amounts of contaminants, as discussed above, are sufficient to provide all the necessary current. The potential map in Figure 6 for methanol with 100 µM KCl in air is, in all places, higher as compared to all three Figure 7 100 µM KCl-in-air plots. This comparison suggests that the small amount of water undoubtedly present in solutions used to generate Figure 6 DEEP maps, although theoretically sufficient to provide 100% of ES current, is actually only partially participating in the electrode reactions. Kinetics of Cell Response. The response time required for stabilization of the measured potential upon a change in ambient gas was investigated, and Figure 8 shows three examples of the kinetics of the cell response in switching back and forth between nitrogen and oxygen environments. In this figure, as well as in all examples presented above, the time for the potential to reach a steady value is always longer when the gas is changed from oxygen to nitrogen than in the reverse instance. If the wire probe electrode is exposed to the atmosphere (i.e., at positive x values), the response time for an oxygen to nitrogen change is about 100-200 s; the reverse, nitrogen to oxygen, requires less than 10 s. When the electrode is withdrawn deeper into the ES capillary, the response time becomes longer; i.e., it grows to 550 s for the oxygen to nitrogen change, and 50 s for nitrogen to oxygen. The last panel (Figure 8c) shows the kinetics of gas change for wire position -1.2 mm (deep inside the ES capillary). There, the electrode response time for the oxygen to nitrogen change is 400 s; the reverse change takes ∼60 s. Interestingly, for this wire position, and in agreement with the (26) Beden, B.; Leger, J.-M.; Lamy, C. In Modern Aspects of Electrochemistry; Conway, B. E., Bockris, J. O’M., White, R. E., Eds.; Plenum Press: New York, 1992; Vol. 22, p 97.

Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

3389

that oxygen, which is adsorbed and active, immediately takes part in the reaction. When the supply is cut off, lingering oxygen, which has already been adsorbed in the system, starts to deplete. Only after all the oxygen has been consumed does the electrospray system return to oxygen-free operation, with the electrode potential adjusting itself to the value that corresponds to the nitrogen atmosphere.

Figure 8. Kinetics of change of measured wire probe potential in response to changes in ambient gas for different positions of the wire probe electrode. All data were acquired in the positive ion mode using 50:50 methanol-water mixture (deaerated) with 100 µM KCl. (a) x ) 0.3 mm, (b) x ) -0.2 mm, and (c) x ) -1.2 mm.

100 µM KCl potential maps in Figure 7, the wire probe potential in the presence of ambient oxygen is higher than it is in nitrogen; a reverse situation compared to higher values of x in both figures. In either positive or negative ion modes of ES operation, the cell response to a change of ambient gas is asymmetrical. Always when the gas is switched from nitrogen to oxygen (air), the potential shift is fast, whereas in the reverse instance, when nitrogen replaces oxygen, the response is far slower. This means 3390 Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

CONCLUSIONS An ambient gas can enter the system only in places where it is in contact with the solution undergoing electrospray. The exposed portion is the surface or “skin” of the Taylor cone. This skin is also in contact with the front metallic surface of the ES capillary that has a substantial wall thickness relative to the capillary diameter, and the front surface is the portion nearest to the counterelectrode. In terms of solution resistance, it is also nearest to the Taylor cone filament, i.e., the place where the excess charges are removed from the liquid. Consequently, the ES capillary front surface is the place where the electrospray current density reaches its highest values and where the greatest variety of electrochemical processes may transpire. For this reason, the changes in chemical composition of the solution around that small, but very active region are of the utmost importance in determining the exact processes that take place. At the surface of the Taylor cone, i.e., at the boundary between liquid and gas phases, rapid adsorption of ambient gases and evaporation of solution constituents are occurring. The Taylor cone surface also has a net charge27-29 that results in a tangential electric field gradient on the surface (skin) causing charges to flow in the direction of the filament. This electrical migration creates a shearing stress on the bulk liquid beneath, and turbulence within the Taylor cone is an unavoidable consequence of its axial symmetry under the stress of surface shear.30 The linear velocities of molecules on the Taylor cone surface are likely to be close to the velocities of the molecules inside the filament at the Taylor cone tip.18 Based on its diameter and the volume of sprayed liquid produced per unit time, the filament linear velocity is estimated to be higher than 10 cm/s. This implies that the residence time of a molecule on the Taylor cone surface is only a few milliseconds before irreversibly entering the filament. This high linear velocity on the skin suggests that only a small fraction of the molecular oxygen adsorbed at the surface finds its way to the bulk because there is not enough time for a molecule to travel to the inside of the Taylor cone simply by diffusion. The small percentage of oxygen molecules that participate in electrochemistry at the ES capillary must be hydrodynamically driven into the Taylor cone and back toward the ES capillary. The “dips” consistently observed in DEEP maps obtained in the presence of oxygen (Figures 3b, 4, 6, and 7) are evidence that the Taylor cone, beneath its skin, is a turbulent reactor. In this region close to the ES capillary exit, particles move back and forth in irregular paths.30,31 Our DEEP maps show that the (27) Hayati, I.; Bailey, A. I.; Tadros, T. F. Nature 1986, 319, 41-43. (28) Hayati, I.; Bailey, A. I.; Tadros, T. F. J. Colloid Interface Sci. 1987, 117, 205-221. (29) Hayati, I.; Bailey, A. I.; Tadros, T. F. J. Colloid Interface Sci. 1987, 117, 222-230. (30) Barrero, A.; Gan ˜a´n-Calvo, A. M.; Davila, J.; Palacio, A.; Gomez-Gonzalez, E. Phys. Rev. E 1998, 58, 7309-7314. (31) Eggers, J. Rev. Mod. Phys. 1997, 69, 865-929.

turbulent region and the range of oxygen influence extend a bit into the capillary cavity (∼0.3-0.5 mm). In small amounts, oxygen may also diffuse along the capillary walls in a stagnant layer reaching points even deeper inside the ES capillary. After a short time (a few seconds), a steady state is reached where the rate of oxygen adsorption is balanced by the rate of oxygen removal (i.e., consumption plus removal by hydrodynamic transport). When the air supply is turned off, oxygen is slowly depleted and only after it is completely gone does the cell go back to oxygen-free operation. This evidence indicates that the adsorption rate is faster than the electrochemically induced consumption of oxygen, so the gas switch from nitrogen to oxygen produces a fast jump in potential, whereas oxygen depletion takes longer. Once molecular oxygen has infiltrated the inside of the diffusion layer near the electrode surface, it substantially alters the nature of the electrochemical activity taking place. This may occur either through direct oxygen involvement in bimolecular or higher order redox reactions or, alternatively, through bulk chemical reactions of molecular oxygen with electrochemically generated redox products. The electrochemical (including oxygenated) products may then find their way to the skin of the Taylor cone where they may again encounter oxygen, and a few of these may again be cycled back into the bulk. Thus, relative to upstream ES capillary positions, a separate chemical-electrochemical reaction zone is created at the ES exit where distinct electrochemistry processes and follow-up reactions can occur. Such oxygenated

products generated by electrochemical (or follow-up chemical) reaction of molecular oxygen adsorbed at the Taylor cone may lead to unexpected signals in the mass spectrometer. Both positive and negative ion modes are susceptible to the influence of ambient gases, but the total electrospray current is not affected, due to the self-regulating nature of the ES device working as a controlled-current electrolytic cell. If present in the ambient gaseous environment, active gases other than molecular oxygen could also enter into electrochemical reactions during ES operation. In addition to demonstrating that gas adsorption can change the composition of reaction products, resulting in the establishment of different electrode potentials, our study has shown that residual gases initially dissolved in the ES solvent also enter electrochemical reactions, but they play a minor role as compared to adsorption at the Taylor cone. ACKNOWLEDGMENT Financial support for this research was provided by the National Science Foundation through CHE-0518288. We are also grateful to ESA Biosciences, Inc. for additional financial support for this project.

Received for review December 8, 2006. Accepted February 16, 2007. AC062329U

Analytical Chemistry, Vol. 79, No. 9, May 1, 2007

3391