Design of a Time-of-Flight Mass Spectrometer as a Detector for

Sensar Corporation, 1652 West 820 North, Provo, Utah 84601. High-efficiency separation methods such as capillary electrophoresis (CE) interfaced with ...
0 downloads 0 Views 331KB Size
Anal. Chem. 1997, 69, 3205-3211

Design of a Time-of-Flight Mass Spectrometer as a Detector for Capillary Electrophoresis Iulia M. Lazar, Baomin Xin, and Milton L. Lee*

Department of Chemistry and Biochemistry, Brigham Young University, Provo, Utah 84602 Edgar D. Lee, Alan L. Rockwood, Jacqueline C. Fabbi, and Harold G. Lee

Sensar Corporation, 1652 West 820 North, Provo, Utah 84601

High-efficiency separation methods such as capillary electrophoresis (CE) interfaced with structural information-producing detection systems such as mass spectrometry (MS) represent invaluable tools in analytical chemistry. Most CE applications to date have been reported using UV absorption detection. To extend the analytical power of CE, new detectors which demonstrate increased speed, sensitivity, and specificity must be developed. In this paper, we describe the design of an electrospray ionization time-of-flight mass spectrometer and its performance as a detector for CE. Emphasis is placed on fulfillment of the speed and sensitivity requirements. Low attomole and low femtomole detection limits are demonstrated for continuous infusion and CE-MS operation, respectively. While capillary electrophoresis (CE) was initially considered to be a technique for the analysis of ionic species, it and its derivative forms can now be applied to an extremely wide range of compounds. This includes nonionic (polar and nonpolar) as well as ionic compounds with molecular weights extending into the particle range. The coupling of CE to mass spectrometry (MS) using quadrupole, ion trap, or Fourier transform ion cyclotron resonance instruments has been described in numerous papers; a recent review was written by Cai and Henion.1 The main limitation of these mass spectrometers is the effective time they take to produce a full spectrum. The current trend in separation science is the achievement of the most efficient separation in the shortest possible time. The time-of-flight mass spectrometer (TOFMS) is the only MS which can meet the detector speed requirement without compromising the CE separation efficiency. While a quadrupole MS takes 0.1-1 s, an ion trap takes approximately 0.2-1 s, and a Fourier transform ion cyclotron resonance (FTICR) mass spectrometer takes 2 s at the best to produce a full spectrum (at the expense of resolution), the TOFMS needs only 100-200 µs for the same purpose.2-4 The record for the fastest and most efficient separations was established using CE. Even though, with the presently available instrumentation, high-speed separations can be achieved in 1-2 (1) Cai, J.; Henion, J. J. Chromatogr. A 1995, 703, 667. (2) Smith, R. D.; Wahl, J. H.; Godlett, D. R.; Hofstadler, S. A. Anal. Chem. 1993, 65, 574A. (3) Hofstadler, S. A.; Severs, J. C.; Smith, R. D.; Swanek, F. D.; Ewing, A. G. Symposium Book, Eighteenth International Symposium on Capillary Chromatography, Riva del Garda, Italy, 1996; p 233. (4) Sin, C. H.; Lee, E. D.; Lee, M. L. Anal. Chem. 1991, 63, 2897. S0003-2700(97)00028-0 CCC: $14.00

© 1997 American Chemical Society

s (half-height peak width less than 100 ms), as reported by Monnig and Jorgenson,5 such separations are not currently practical. However, fast separations which occur in less than 10 min (halfheight peak width of 1-2 s) are becoming common practice.6-8 Obviously, appropriate detection tools which can respond fast enough to quantify the narrow peaks which elute from the capillary column must be developed. TOFMS was first introduced as a new idea by Stephens in 1946,9 but it has enjoyed rapid growth only since the 1980s10-12 after efficient and practical solutions were developed to improve the mass resolution and adequate highspeed data acquisition systems evolved to record the fast events occurring in the TOFMS. Along with speed, other attributes of TOFMS, such as unlimited mass range, high ion transmission efficiency, high duty cycle, sensitivity, and simplicity lead to predictions that TOFMS may become the detection tool of choice for CE.13 The interfacing of CE to TOFMS has recently been reported, but only in a few papers.14-19 On-line CE-MS has been performed using electrospray and continuous-low fast atom bombardment ionization sources, while on-line CE-TOFMS has been reported only with the electrospray ionization source. At first glance, the electrospray appears to be an ideal source for CE, producing mainly the protonated molecular ion for low-molecular-weight compounds and multiply charged ions for high-molecular-weight compounds, thus making possible the MS analysis of analytes with molecular weights in the range of millions. However, the direct coupling of CE to ESI is not always straightforward, since the (5) Monnig, C. A.; Jorgenson, J. W. Anal. Chem. 1991, 63, 802. (6) Weinberger, R. Practical Capillary Electrophoresis; Academic Press, Inc.: Boston, MA, 1993; p 8. (7) Cai, J.; El Rassi, Z. J. Liq. Chromatogr. 1992, 15, 1193. (8) Kleibohmer, W.; Cammann, K.; Robert, J.; Mussenbrock, E. J. Chromatogr. 1993, 638, 349. (9) Stephens, W. E. Phys. Rev. 1946, 69, 691. (10) Wiley, W. C.; McLaren, I. H. Rev. Sci. Instrum. 1955, 26, 1150. (11) Mamyrin, B. A.; Karatajev, V. J.; Shmikk, D. V.; Zagulin, V. A. Sov. Phys. JETP 1973, 37, 45. (12) Lincoln, K. A.; Bechtel, R. D.; Mateos, M. A. Int. J. Mass Spectrom. Ion Processes 1990, 99, 41. (13) Price, D.; Milnes, G. J. Int. J. Mass Spectrom. Ion Processes 1990, 99, 1. (14) Fang, L.; Zhang, R.; Williams, E. R.; Zare, R. N. Anal. Chem. 1994, 66, 3696. (15) Muddiman, D. C.; Rockwood, A. L.; Gao, Q.; Severs, J. C.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1995, 67, 4371. (16) Li, X.; Hongwei, L.; Lubman D. Proceedings of the 43rd ASMS Conference on Mass Spectrometry and Allied Topics, Atlanta, GA, May 21-26, 1995; TPB089. (17) Banks, J. F.; Dresch, T. Anal. Chem. 1996, 68, 1480. (18) Wu, J.-T.; Qian, M. G.; Li, M. X.; Liu, L.; Lubman, D. M. Anal. Chem. 1996, 68, 3388. (19) Verentchikov, A.; Hsieh, F.; Gabeler, S.; Martin, S.; Vestal, M. Proceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR, 1996; May12-16, p 285.

Analytical Chemistry, Vol. 69, No. 16, August 15, 1997 3205

high-concentration buffer systems characteristic of CE are not compatible with the optimal performance of the electrospray source. In this study, we report the design and construction of an electrospray ionization TOFMS and its preliminary evaluation as a detector for CE. Emphasis was placed on optimizing the system for minimum detectable quantity. EXPERIMENTAL SECTION Reagents. Standard solutions were prepared in HPLC grade solvents. Methanol and water were purchased from Mallinckrodt (Chesterfield, MO). Glacial acetic acid was obtained from EM Science (Gibbstown, NJ). All peptides were purchased from Sigma (St. Louis, MO). Instrumentation. Capillary Electrophoresis. The research was carried out using a Crystal CE 300 system (ATI, Madison, WI). The UV detector was an Applied Biosystems (Foster City, CA) Model 785A detector. Uncoated fused silica capillaries, 50 µm i.d., 190 µm o.d. (Polymicro Technologies, Phoenix, AZ), were used. Electrospray Interface. A home-built microelectrospray device was used to perform the interfacing. Since plugging of the capillary tips (3-5 µm) typically used in the micro-/nanoelectrospray source may occur frequently, we conducted several studies with capillaries drawn out to 10-20 µm i.d. and 40-60 µm o.d. The ESI tips were prepared by drawing out the fused silica tubing between two electrodes, using a Model M100 manipulator (Polymicro Technologies, Phoenix, AZ). The sharp ESI tip was cut to 1-1.5 cm and connected to the separation capillary by pushing the capillary and the tip against each other inside a metal union and cementing the fused silica to the metal with epoxy glue. The electrospray voltage was applied to the metal union. The metal union consisted of a 1 cm long, 27 gauge stainless steel needle tube (Hamilton, Reno, NV). A small eluent leak through the junction proved to be sufficient to ensure electrical contact. Additional details concerning the construction of the electrospray setup will be given in a future paper.20 Continuous infusion of the analyte solutions was obtained with a Harvard 22 syringe pump (South Natick, MA). Time-of-Flight Mass Spectrometry. The main objective in the design of the present TOFMS was to produce an instrument which allowed for maximum flexibility in construction and operation. The ion optics can be easily converted from one setup to another, and the individual components and their physical arrangement inside the mass spectrometer can be changed without perturbing the main layout of the instrument. Optimization studies required to achieve the desired performance could, therefore, be easily conducted. The TOFMS was built as a stand-alone instrument, with a vertical flight tube configuration, on its own four-wheel support. The main features of the MS design include a nozzle/ skimmer setup with supersonic expansion for sampling from the electrospray ion source, radio frequency (rf)-only quadrupole ion optics, orthogonal extraction, two-stage acceleration, first-order space focusing Wiley-McLaren pulsing arrangement, two-stage ion mirror, and a high-speed data acquisition system. The flight tube and the reflectron were purchased from R. M. Jordan (Grass Valley, CA). A schematic representation of the newly constructed ESI-TOFMS system is given in Figure 1. (20) Lazar, I. M.; Lee, E. D.; Rockwood, A. L.; Lee, M. L. J. Chromatogr., in press.

3206 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

Figure 1. Schematic diagram of the TOFMS system. (1) Electrospray needle, (2) interface plate, (3) nozzle, (4) skimmer, (5) and (7) rf-only quadrupoles, (6) interquad lens, (8) slit/lens, (9) pulser, (10) grids, (11) flight tube, (12) deflection plates, (13) reflectron, (14) electron multiplier, (15) ion source, (16) body I, (17) body II, (18) base cube, (19) fourth stage housing, (20) and (21) turbo pumps, (22) rotary pump, and (23) vacuum gauge.

The TOFMS consists of an ion source and four differentially pumped vacuum stages. A countercurrent of hot nitrogen gas allows for the drying of the electrosprayed ions, protects the nozzle, and maintains a clean atmosphere in the source. The transparency of the ion source window ensures visual ability to position the sprayer tip and to monitor the spray stability. A microscope (Edmund Scientific, Barrington, NJ) positioned in front of and at an oblique angle above the source permitted close evaluation of the spray characteristics. Magnification settings of 12× or 25× on the microscope proved to be sufficient. A fiberoptic illuminator (Edmund Scientific) was used to shine light over the spray. A rod heater, Model BXD-04B-35-K (ARI Industries, Addison, IL), placed between the interface plate and nozzle allowed the heating of the source up to 90-100 °C. Ions produced in the electrospray source are pulled into the first vacuum stage through a 250 µm orifice nozzle (3, Figure 1). The first and second vacuum stages are separated by a skimmer (4). The ion beam entering the second stage is subsequently passed through the first set of rf-only quadrupoles (5). Much improved ion transmission and focusing efficiencies, as high as 80-90%, compared to static lens configurations, are obtained when these rf quadrupoles are operated at relatively high pressures, (10-1-10-3 Torr).21,22 The joint action of the rf field and of velocity dampening induced by the collision of ions with the buffer gas (21) Douglas, D. J.; French, J. B. J. Am. Soc. Mass Spectrom. 1992, 3, 398-408.

confines the ions close to the axis of the quadrupole. Both radial and axial velocities, and their distributions as well, are significantly reduced. The association of the high-efficiency focusing capabilities of the quadrupole with the high-speed characteristics of the TOF analyzer can lead to the ideal, sensitive mass spectrometer detector, essential for capillary electrophoresis. By applying an appropriate extraction potential on the exit lens (interquad lens, 6), ions are pulled out from the first set of rods and introduced into the third vacuum stage, which houses the second set of rf-only quadrupole rods (7). The third vacuum stage was necessary since the direct passage from the high-pressure region (10-1-10-3 Torr) of the second stage to the desired lowpressure region of the fourth stage (10-7 Torr) was not possible with the current pumping system. Even though there is no collisional focusing at the low pressures in this stage, the second quadrupole is necessary for efficient transport of ions into the fourth stage. Up to this point, electrically insulating spacers ensure the proper alignment of the ion optics elements inside the mass spectrometer housing. By varying the thickness of the spacers, the proper distances between ion optics elements were easy to optimize. The ion beam is further introduced through a slit/lens (8) into the pulsing region, where it is repelled in a perpendicular direction to the incident beam by a fast 200 V pulse. The resultant ion packets enter the flight tube (11) and strike the electron multiplier detector (14), either at the top of the flight tube when the system is operating in the linear mode or at the bottom of the flight tube when the system is operating in the reflecting mode. A GRX1.5K-E pulser (Directed Energy, Fort Collins, CO) and an AF860H electron multiplier (ETP Scientific, Auburn, MA) were used. The pulsing region, defined by the repeller plate (pulser, 9) and a series of grids (10), contains two accelerating regions. The appropriate spacing between these elements was calculated according to previous modeling work reported by Wiley and McLaren10 and Pollard et al.23 An appropriate housing (19) was built around the pulsing region to separate the third from the fourth vacuum stages. At the time of these experiments, we were not yet in possession of a second electron multiplier, and the performance of the instrument in the reflecting mode could not be tested. The pumping system consists of an E2M18 rotary pump (Edwards, Wilmington, MA) for the first vacuum stage, a twostage TMH-520-020 turbo pump (Balzers-Pfeiffer, Hudson, NH) of 20 and 520 L s-1 for the second and third vacuum stages, respectively, and a TMH-260 Balzers turbo pump, of 260 L s-1, for the fourth vacuum stage. The two turbo pumps are backed up by an Edwards E2M2 rotary pump. Since the Jordan flight tube and the TOF base cube were standard configurations, special adaptors were constructed to connect the two Balzers pumps to the system. A Balzers EVA 040 S spindle valve was mounted on the differential stage of the TMH-520-020 turbo pump, which pumps the second vacuum stage or first quadrupole housing. Opening or closing the valve resulted in various pressures in the second stage, without significant alteration of the vacuum in the third stage. Under each turbo pump, there is a fan that provides for air cooling of the pumps. (22) Krutchinsky, A. N.; Chernushevich, I. V.; Spicer, V.; Ens, W.; Standing, K. G. Presented at the 43rd ASMS Conference on Mass Spectrometry and Allied Topics, Atlanta, GA, May 21-26, 1995; Poster MPB 085. (23) Pollard, J. E.; Lichtin, D. A.; Janson, S. W.; Cohen, R. B. Rev. Sci. Instrum. 1989, 60, 3171.

Power supplies were built in-house using power modules purchased from Bertan (Hicksville, NY) and Matsusada Precision (Mountain View, CA). The quadrupole system was driven by a Model 4017, 10 MHz sweep/function generator (B&K Precision, Chicago, IL), a 240L broadband power amplifier (ENI, Rochester, NY), and a step-up transformer provided upon request by Larson‚Davis (Provo, UT). The signal preamplifier was a Model 310 Sonoma Instrument device (Santa Rosa, CA). The high-speed data acquisition system was developed by Sensar (Provo, UT). RESULTS AND DISCUSSION Evaluation of Ion Optics and Optimization of Operational Parameters. A series of optimization studies were conducted to improve the performance of the TOF instrument as a CE detector. These included evaluation of the nozzle/skimmer distance, rod diameter and length, pulser arrangement, and the orifice sizes in the nozzle, skimmer, and lenses. Our first results are presented in this paper. Different pumping configurations were tested to achieve the required high pressure for optimal operation of the rf quads (10-2-10-3 Torr) in the second vacuum stage, without disturbing the necessary low pressure in the fourth vacuum stage, which houses the pulser, the flight tube, and the electron multiplier (10-7 Torr). The evaluation of the mass spectrometer started with a 0.25 mm sampling nozzle, a 0.80 mm skimmer, a 0.75 mm interquad lens, and a 1 mm slit/lens between the third and fourth vacuum stages. A spacer of adequate thickness placed the skimmer 1.5-2 mm downstream from the nozzle at about twothirds the distance from the nozzle to the Mach disk onset. This arrangement resulted in the following pressures in the mass spectrometer: 3 Torr in the first stage, 1.5 × 10-1 Torr in the second stage, 7 × 10-6 Torr in the third stage, and (3-4) × 10-7 Torr in the fourth stage. Douglas and French21 demonstrated a maximum ion transmission of rf-only quadrupoles at (6-8) × 10-3 Torr. It was found that collisional focusing at higher pressures increased with the mass of the ion (not with the mass-to-charge ratio). The largest analyte they studied was myoglobin, with a mass of 16 950 Da. The ion trajectories inside a quadrupole are described by Mathieu type equations.24 A stability diagram is defined in the a and q coordinates, where a ) 4eU/mω2r02; q ) 2eV/mω2r0; ω ) 2πf; r ) 1.148r0; r is the radius of the rods, mm; r0 is the field radius, mm, which is the circle radius on which the rods are placed; e is the charge, C; U is the dc voltage applied between opposite rods, V; V is the zero-to-peak rf voltage applied between opposite rods, V; ω is the angular frequency; f is the frequency applied to the rods, Hz; and m is the mass of the ion, kg. There are several regions of the diagram which allow for stable ion trajectories inside the quadrupole, the most common being the first stability region (used by commercial rf/dc quadrupole instruments), defined between the following values: a ) 0-0.236 and q ) 0-0.908. In rf-only operation, the dc component is set to zero, and, as a consequence, stable ion trajectories are defined for a ) 0, solely on the q axis, for 0 < q < 0.908. Ions with different q values can be transmitted more or less efficiently, depending on the operating parameters, V, ω, and r0. Several features must be taken into consideration:25,26 (a) Low frequencies and small-diameter rods favor high-mass operation, while large frequencies and large-diameter rods favor (24) Dawson, H. P., Ed. Quadrupole Mass Spectrometry and Its Applications; Elsevier Scientific: Amsterdam, 1976.

Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

3207

Figure 2. Current measurements for gramicidin S, 10 and 100 µM solutions. Abbreviations: IP, interface plate; N, nozzle; S, skimmer; IQL, interquad lens; L, lens; BP, back plate.

low-mass operation. (b) An increase in power, V, to the rods, increases the transmitted mass range, shifting it toward higher values. (c) An increase in rod diameter yields better sensitivity since transmission is proportional roughly to the square of the radius of the rods; however, a decrease in diameter increases the transmitted mass range (an increase in rod length increases the ability to handle high-energy ions). (d) As a rule of thumb, if one seeks for extended mass range capability, then a reduction in rod diameter, a reduction in operating frequency, and an increase in power to the quadrupoles would be effective, while if one seeks for maximum sensitivity within a limited mass range, then the largest rod diameter and the highest frequency consistent with the mass range of interest must be chosen. The transmission and focusing properties of the 0.5 in. quadrupole rods in our instrument were initially studied at f ) 1.96 MHz and V0-peak ) 500-600 V, at 1.5 × 10-1 Torr pressure. Both quadrupoles were operated under identical conditions. A solution of gramicidin S in CH3OH/H2O/CH3COOH (70:30: 0.1 v/v) was electrosprayed using the microelectrospray source with the spray tip placed at 9-12 mm from the sampling nozzle. The optimum N2 curtain gas flow was 1200 mL min-1. The results from this experiment are shown in Figure 2. The ion optic elements where current measurements were made are (from left to right) the interface plate, nozzle, skimmer, interquad lens, slit/ lens, and back plate placed behind the pulser. Measurements were performed by disconnecting the power supply to an element and measuring the current that reached it. Since the optimized arrangement is perturbed under these conditions, the actual currents, under real, working circumstances, might be higher. A description of the requirements and limitations of such a setup, and considerations regarding ion energies which result in maximum ion transmission and focusing, were explained in detail by Douglas and French.21 Briefly, to achieve collisional focusing, the injection energy of the ions must be low (less than 30 eV), and there is an optimum pressure range where the collisional focusing reaches its maximum effect. It should always be remembered that collisional focusing will work better for higher mass ions. From the measurements presented in Figure 2, it can be observed that, from the initial current measured on the interface (25) Pedder, R. E.; Schaeffer, R. A. Presented at the 43rd ASMS Conference on Mass Spectrometry and Allied Topics, Atlanta, GA, May 21-26, 1995; Posters ThPB 047 and ThPB 048. (26) Product Bulletin, Extrel Corp., Pittsburgh, PA.

3208

Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

plate, 0.05-0.075% reached the back plate. The current measurements on the interquad lens indicate the degree of focusing achieved under these conditions. If the offset voltage on the second set of rods was set to zero, the ion beam passed through the 0.75 mm orifice of the interquad lens and was detected partially on the slit and partially on the back plate. If the offset voltage on the second set of rods was increased to 20-30 V, the ion beam was stopped and the current added up on the interquad lens. We observed that, for the 10 µM solution, (126-74)/126 × 100% ) 41%, and for the 100 µM solution, (160-87)/160 × 100% ) 45% of the beam was focused within the 0.75 mm diameter orifice. From the transmitted ion beam, 19/(19 + 43) × 100% ) 30% for the 10 µM solution and 50/(46 + 50) × 100% ) 52% for the 100 µM solution were measured on the back plate. Losses to the second set of rods were not considered. From these results, a preferential focusing of the higher mass gramicidin ion (1141 amu) is inferred. We also monitored the signal intensity as a function of the voltage applied to each ion optics element. A very strong dependence was found for the skimmer, rods offset, and interquad lens voltage. These elements control the ion energy in the highpressure collisional focusing region of the first set of rods and have a crucial influence on the focusing properties of the quadrupole. An expected effect is the mass discrimination introduced by these elements. Maximum intensity signals are observed for each compound at different voltage settings. We compared the gramicidin S signal to the paraquat signal (186 amu for the free ion). Gramicidin S gave mainly the doubly charged species at 571 m/z ratio, while a doubly charged paraquat ion would have given a signal at 93 m/z. However, the paraquat spectrum was mainly composed of a series of ion clusters. The signal intensity at 93 m/z was extremely weak and dependent on the first quadrupole offset voltage. Initially, we attributed this effect to the high pressure in the quadrupole region and low transmission efficiency for low-mass ions. Later experiments have proven that the nozzle/skimmer distance plays an important role in the efficient sampling of the paraquat ion as well. By lowering the pressure in the first quadrupole region to 3 × 10-2 Torr, increasing the nozzle/skimmer distance to 6-7 mm, and positioning the electrospray tip very close to the nozzle (4-5 mm), we increased the signal intensity for the doubly charged gramicidin S ion approximately 3 times. The signal intensity for the paraquat doubly charged ion increased from a practically nonexistent peak, to a level comparable to that of the gramicidin S signal. The spectrum became very clean, displaying mainly the m/z ) 93 ion. However, the absolute peak intensity still remained 5-7 times lower than that of gramicidin S (Figure 3). Therefore, we focused our attention on the optimization of the first quadrupole, which operates under collisional focusing conditions. We did not study closely the transmission efficiency of the second quadrupole. However, after disconnecting the rf power to the second quadrupole, the signal was almost completely lost. Data Acquisition System. An extremely important concern when interfacing CE to MS is the effective speed of the data acquisition system. Millions of plates have been demonstrated in capillary electrophoresis with half-height peak widths on the order of milliseconds. If we consider a CE peak of 1 s in width, to quantify the amount represented by the peak, at least 10 points must be measured across the peak width. It is also desirable to average as many spectra as possible to increase the signal-to-noise ratio. If we acquire 10 spectra across a 1 s peak, and the MS

Figure 3. Comparison of signal intensities of the paraquat and gramicidin S doubly charged ions in the TOF mass spectrum. Conditions: solution of 50 µM paraquat and 10 µM gramicidin S in CH3OH/H2O/CH3COOH (70:30:0.1 v/v), 0.3 µL min -1; microelectrospray, 2100 V; 40 °C ion source temperature, 5000 Hz pulsing rate, 5000 spectra averaged. TOFMS voltages were optimized for the paraquat ion and are slightly different from the optimum values for the gramicidin S ion. Flight time range corresponds to 0-1545 m/z range.

Figure 5. Effect of signal averaging on signal/noise ratio. Conditions: 10 µM solution of gramicidin S in CH3OH/H2O/CH3COOH (70: 30:0.1 v/v), 0.3 µL min-1.

Figure 6. Continuous infusion of gramicidin S. Conditions: 0.1 µM solution of gramicidin S in CH3OH/H2O/CH3COOH (70:30:0.3 v/v), 0.2 µL min -1. (A) 5,000 Hz pulsing rate, 5000 spectra averaged; (B) 5000 Hz pulsing rate, 500 spectra averaged; (C) 5000 Hz pulsing rate, 100 spectra averaged; (D) 10 000 Hz pulsing rate, 200 spectra averaged; (E) 10 000 Hz pulsing rate, 100 spectra averaged. Lower mass ions represent background clusters. The spectra correspond to 0-1545 m/z range.

Figure 4. Schematic function of the high-speed data acquisition system.

accumulates spectra at 5000 Hz frequency, 500 averages per spectrum will be possible, and 100 ms is required in order to perform this operation (Figure 4). To date, the high-speed data acquisition system produces sufficient noise to require a relatively high number of spectra to be averaged to obtain a clean spectrum. A total number of 1500 averaged spectra can be stored, which translates into a 25 min CE-MS run if one averaged spectrum per second is acquired, or into a 5 min CE-MS run if five averaged spectra per second are acquired. The high-speed data acquisition system collects 32 000 points after each pulse. Depending on the desired acquisition rate or resolution, the data points can be collected at a 2, 4, or 8 ns time difference. The result is that a full spectrum is produced in 64, 128, or 256 µs. Obviously, if the choice is 8 ns resolution and consequently the spectrum acquisition time is 256 µs, the pulsing

rate should be less than 5000 Hz when a maximum of only 200 µs is allowed for a spectrum to be collected. The starting point for spectrum acquisition can be shifted wherever desired across the 200 µs time scale, provided the full acquisition time is completely included within the available 200 µs window. A visualization of what signal averaging can provide is given in Figure 5. For 10 and 100 averaged spectra, the singly charged gramicidin S ion is lost in the noise. For practical purposes, 5001000 averages proved to be sufficient for spectral acquisition even at very low analyte concentration. Continuous Infusion Experiments. Excellent detection levels were established from continuous infusion experiments. Figure 6 represents full-scan mass spectra of gramicidin S acquired from a 0.1 µM solution at different pulsing rates and signal averaging. The spectral acquisition time decreased from 1 to 0.01 s by decreasing the number of averaged spectra. The signal-tonoise ratio decreased accordingly. However, a distinct signal can still be observed in Figure 6E, which was produced from 3 amol of consumed analyte (S/N ≈ 3) and is the result of 100 averaged spectra at a pulsing rate of 10 000 Hz. The total acquisition time Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

3209

Figure 7. CE-TOF and CE-UV separation of enkephalins. Conditions: 85 cm × 50 µm i.d. uncoated fused silica capillary, CH3CN/ H2O/CH3COOH (50:50:0.3 v/v), 30 kV, 10 mbar. (A) CE-TOF electropherogram. Conditions: 10 mbar × 0.03 min injection (3.34.6 fmol), 1600 V microelectrospray, 80 °C ion source temperature, 5000 Hz pulsing rate, 1000 spectra averaged per data point, 5 data points s-1. (B) CE-TOF electropherogram. Conditions same as in (B), except 20 mbar × 0.04 min injection (7-10 fmol). (C) CE-UV electropherogram. Conditions: 20 mbar × 0.1 min injection (88-122 fmol), detection at 210 nm after 70 cm column length. Peak identifications: (1) des-Tyr1,[D-Ala2,D-Leu5]-enkephalin, (2) leucine enkephalin, and (3) methionine enkephalin.

to produce this spectrum was only 0.01 s. The smaller mass ions represent background clusters, and they are particularly intense due to the very close positioning of the microspray tip to the sampling nozzle (∼3 mm). For comparison, we are aware of only one report27 claiming lower detection (neurotensin, 35 zmol consumed analyte). However, to produce the spectrum, an acquisition time of 8 s was necessary. CE-TOFMS. The interfacing of the CE system to the TOFMS was evaluated using a mixture of enkephalins (Figure 7). An untreated fused silica capillary column was rinsed with 1 N sodium hydroxide and HPLC water prior to analysis. The physical arrangement of the CE and the TOFMS system did not allow for capillary columns shorter than 85 cm to be used for the separation. To reduce the migration time and to improve the spray performance, a mixture of CH3CN/H2O/CH3COOH (50:50: 0.3 v/v) was used as running buffer. The presence of acetonitrile reduced the wall adsorption of the analytes, improved the peak shapes, and allowed for fast and efficient separations on bare fused silica capillaries. The EOF under these conditions was approximately 280 nL min-1. The sample (14 µM in des-Tyr1,[DAla2,D-Leu5]-enkephalin acetate, 10 µM in leucine enkephalin acetate, and 10 µM in methionine enkephalin acetate) was prepared in CH3OH/H2O/CH3COOH (50:50:0.03 v/v). A pressure of 10 mbar was applied on the inlet buffer vial with the purpose of improving the electrospray stability. Nonaqueous CE has significant potential for future CE-MS interfacing. First, it extends the applicability of CE to hydrophobic compounds, which are not easily solubilized in water, or which have close electrophoretic mobilities in water. Second, it offers a larger range of polarity, density, viscosity, dielectric constant, and acid-base equilibria which can be manipulated for a given CE separation.28 Third, if used with low-concentration volatile (27) Andren, P. E.; Emmett, M. R.; Caprioli, R. M. J. Am. Soc. Mass Spectrom. 1994, 5, 867.

3210 Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

Figure 8. TOF spectrum of leucine enkephalin (MH+ ) 556) from a CE injection. Conditions same as in Figure 7A (∼1 fmol injected from a 3.3 µM solution). Lower mass ions represent background electrolyte clusters.

buffers, it can be easily interfaced to MS using a microelectrospray ionization source. The spraying efficiency of organic/aqueous mixtures is superior to that of purely aqueous solutions, which ultimately results in improved sensitivities. Fourth, organic modifiers should improve the resolution of small molecules,29 and if they also reduce the viscosity of the running buffer (as acetonitrile does), they lead to increased analyte mobilities, faster separations, improved peak shapes, and, consequently, to even better sensitivities. Nonaqueous CE-MS was reported previously, but only in conjunction with a liquid sheath electrospray source.29,30 The TOF electropherogram in Figure 7A was produced from approximately 3.3-4.6 fmol of injected analytes (signal-to-noise ratio of ∼35). The spectrum of leucine enkephalin produced from ∼1 fmol of injected analyte (3.3 µM solution) and acquired at the CE peak maximum, is given in Figure 8. The protonated molecular ion dominates the spectrum. The lower mass ions in the spectrum represent background electrolyte clusters. Similar spectra were obtained for the other two enkephalins as well. Minimum detectable quantities in the mid-attomole region should be possible. The signal-to-noise ratio was calculated by dividing the analyte signal intensity at the peak maximum by the standard deviation of the background noise measured at the flight time of the given analyte. To be certain that contamination did not occur during this experiment, blank samples, or samples containing only the leucine enkephalin component, were alternately injected with the sample of interest. An acquisition rate of 5 data points s-1 (5000 Hz pulsing rate, 1000 averaged spectra per data point) proved to be sufficient for accurate monitoring of this CE separation. The three components eluted within a 17-20 s time window, having approximately 2-3 s peak widths at half-height. Each peak was defined by 20-25 data points. To reduce the amount of stored data and the data work-up time, data collection usually started 30-40 s prior to elution of the analytes. Figure 7 represents only the time window of interest for the elution of the three analytes. A representation of the entire length of the electropherogram would have resulted in undistinguishable peaks. The electropherogram was reconstructed by integrating each mass spectrum within the mass range of interest. Loss of efficiency and resolution can be observed in the CE-TOF (28) Sahota, R. S.; Khaledi, M. G. Anal. Chem. 1994, 66, 1141. (29) Tomlinson, A. J.; Benson, L. M.; Naylor, S. LC-GC 1994, 12, 122. (30) Tomlinson, A. J.; Benson, L. M.; Naylor, S. J. High Resolut. Chromatogr. 1994, 17, 175.

electropherogram when compared with the CE-UV electropherogram (Figure 7C), and this is mainly due to the nonperfect alignment of the microspray tip with the CE separation capillary inside the metal union. A similar electropherogram, produced with a different setup in which an ideal fit was achieved, shows no loss of efficiency (Figure 7B). Figure 7C is a representation of the CE-UV electropherogram of the same mixture. Even though we used the same capillary length for separation, UV detection occurred at 70 cm after injection, and consequently the peaks eluted earlier than in the TOF electropherogram. To obtain a strong signal, a much larger injection quantity was required in the CE-UV experiment (88-122 fmol) than in the CE-TOF experiment. The sample had a concentration of 56 µM in desTyr1,[D-Ala2,D-Leu5]-enkephalin acetate, 40 µM in leucine enkephalin acetate, and 40 µM in methionine enkephalin acetate. Both CE-UV and CE-MS data could not be collected simultaneously from the same run, since the incorporation of the UV detector while performing the CE-TOFMS experiment would have resulted in the need for very long capillaries. A pressure of 10 mbar on the inlet buffer vial was applied in the case of CE-UV runs as well. Similar results have been recently published by Wu et al.18 The mass spectrometer used was an ion trap TOFMS, for which initial trapping of ions prior to ejection into the TOF analyzer resulted in a significant increase in sensitivity. Detection limits for CE-MS interfacing in the low attomole, or mid-attomole range have been reported by several groups using selected ion monitoring (SIM) with quadrupole machines,31 CE performed on 5 µm capillaries,32 or nanospray sources used in conjunction with ion trap instruments.33 The improved sensitivity of SIM quadrupole MS detection over scanning operation is obtained at the expense of spectral information, 5 µm capillaries with nanospray sources can become very troublesome to operate due to frequent plugging, and ion trap MS detection is limited when fast and efficient separations are performed. Subattomole detection levels for (31) Smith, R. D.; Wahl, J. H.; Goodlett, D. R.; Hofstadler, S. A. Anal. Chem. 1993, 65, 574 A. (32) Wahl, J. H.; Goodlett, D. R.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1992, 64, 3194. (33) Ramsey, R. S.; McLuckey, S. A. J. Microcolumn Sep. 1995, 7, 461. (34) Valaskovic, G. A.; Kelleher, N. L.; McLafferty, F. W. Proceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR, May 12-16, 1996; p 1286.

ubiquitin and cytochrome c were also demonstrated using 5 µm capillaries and picospray FTICR mass spectrometry;34 however, 45 s of ion trapping was required in order to produce the mass spectra, and the electropherograms were generated by ion current measurements only. Experiments which demonstrate ultralow detection limits are important to indicate the future potential of CE-MS. However, it must be kept in mind that CE-MS, which eventually can be used for routine analysis, must be robust, flexible, easy to operate, and not limiting to the performance of the CE technique, in addition to having good sensitivity. TOFMS offers the potential for this realization. CONCLUSIONS A TOFMS instrument has been built and evaluated as a detector for fast capillary electrophoresis. Excellent detection in the very low femtomole range (CE-TOFMS) and attomole range (continuous infusion TOFMS) have been achieved. High-speed data acquisition allows for high-fidelity monitoring of fast separations. Nonaqueous CE microspray MS appears to be the best approach for CE-MS. ACKNOWLEDGMENT We thank Larry Davis from Larson‚Davis and Keith Kling, Lon Que Adams, and Robert M. Perry from the BYU Instrument and Research Shops for fabrication of the electronics and hardware for the TOFMS system. We thank the research group of Dr. Kenneth G. Standing (University of Manitoba, Canada) for providing useful advice for the design of the rf-only quadrupole electronics. The U.S. Environmental Protection Agency (EPA), through its Office of Research and Development (ORD), partially funded the research described under assistance agreement (Agreement No. CR 824316-01-0) to Brigham Young University. This manuscript has been subjected to the EPA’s peer and administrative review and has been approved by EPA for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation by EPA for use. Received for review January 8, 1997. Accepted May 21, 1997.X AC9700282 X

Abstract published in Advance ACS Abstracts, July 1, 1997.

Analytical Chemistry, Vol. 69, No. 16, August 15, 1997

3211