Anal. Chem. 1998, 70, 4111-4119
An Ion Funnel Interface for Improved Ion Focusing and Sensitivity Using Electrospray Ionization Mass Spectrometry Scott A. Shaffer, David C. Prior, Gordon A. Anderson, Harold R. Udseth, and Richard D. Smith*
Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352
To improve upon the already impressive sensitivity achievable with electrospray ionization sources, a novel electrohydrodynamic ion funnel interface has been developed and implemented with a triple-quadrupole mass spectrometer. The ion funnel interface effectively consists of a series of ring electrodes of increasingly small internal diameters to which rf and dc electric potentials are coapplied. In the 1-10-Torr pressure range, the electric fields cause the collisionally damped ions to be more effectively focused and transmitted as a collimated ion beam. This paper describes the ion funnel design and presents an evaluation of its performance using a triplequadrupole mass spectrometer. Ion transmission and m/z discriminating parameters (resulting in both effective low- and high-m/z cutoffs) are presented based upon both ion current measurements and mass spectra. Electrospray ionization mass spectra of selected protein solutions demonstrated well over 1 order of magnitude increase in signal relative to that of the instrument operated in its standard (inlet capillary-skimmer) configuration under similar conditions. The present results suggest that it will be feasible to realize close to 100% ion transmission efficiency for analytically relevant ions through the electrospray ionization interface and into the mass analyzer. Electrospray ionization-mass spectrometry (ESI-MS) has become an important tool for a wide variety of applications, including the analysis of drugs and biopolymers.1-4 The useful applications of ESI-MS are often defined by the overall sensitivity of the technique and present ion sources and interfaces have improved over the last 10 years to the point where detection limits are generally in the femtomole to attomole range.5-10 While the (1) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Science 1989, 246, 64-71. (2) Fenn, J. B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse, C. M. Mass Spectrom. Rev. 1990, 9, 37-70. (3) Smith, R. D.; Loo, J. A.; Edmonds, C. G.; Baringa, C. J.; Udseth, H. R. Anal. Chem. 1990, 62, 882-899. (4) Smith, R. D.; Loo, J. A.; Loo, R. R. O.; Busman, M.; Udseth, H. R. Mass Spectrom. Rev. 1991, 10, 359-451. (5) Wahl, J. H.; Goodlet, D. R.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1992, 64, 3194-3196. (6) Wahl, J. H.; Goodlet, D. R.; Udseth, H. R.; Smith, R. D. J. Am. Chem. Soc. 1993, 115, 803-804. (7) Andren, P. E.; Emmett, M. R.; Caprioli, R. M. J. Am. Soc. Mass Spectrom. 1994, 5, 867-869. S0003-2700(98)00217-0 CCC: $15.00 Published on Web 08/29/1998
© 1998 American Chemical Society
formation of ions at atmospheric pressure by ESI has been established to be very efficient, particularly for very low sample flow rates, only a small fraction of the ions created are subsequently transmitted through the mass spectrometer and ultimately detected. Most of the ion losses occur where ions must traverse high-pressure regions in order to reach lower pressure regions, namely, both before and following the sample inlet (regions at approximately atmospheric pressure and ∼0.5-10 Torr, respectively).11,12 Generally several regions of differential pumping are needed to accommodate the dual demands imposed by the gas load resulting from the sample or ion inlet aperture and the vacuum requirements of the mass analyzer. At higher pressure, electrostatic fields (as well as magnetic fields) are insufficient to reverse the Coulombic ion cloud expansion and disruptive effects due to gas dynamics. To date, ESI source designs have advanced to the point where design modifications produce only minor gains in ion transmission in the higher pressure regions of ESI sources. Consequently it is of interest to develop approaches to more effectively focus and transmit ions and hence increase the fraction of the analytically useful ion current transmitted to the mass spectrometer’s mass analyzer (e.g., a quadrupole mass filter). Our laboratory has recently developed an ESI-MS “ion funnel” interface based upon a series of cylindrical ring electrodes of increasingly smaller internal diameter for the purpose of collisional focusing at elevated pressures.13 The ion funnel uses a coapplication of rf (opposite polarity on adjacent electrodes) and dc electric fields to efficiently focus and transmit ions from regions of higher pressure to regions at lower pressure. It is well established that rf quadrupole devices14-18 and segmented rf (8) Valaskovic, G. A.; Kelleher, N. L.; Little, D. P.; Aaserud, D. J.; McLafferty, F. W. Anal. Chem. 1995, 67, 3802-3805. (9) Valaskovic, G. A.; Kelleher, N. L.; McLafferty, F. W. Science 1996, 273, 1199-1202. (10) Figeys, D.; Vanoostveen, I.; Ducret, A., Aebersold, R. Anal. Chem. 1996, 68, 1822-1828. (11) Busman, M.; Sunner, J.; Vogel, C. R. J. Am. Soc. Mass Spectrom. 1991, 2, 1-10. (12) Lin, B. W.; Sunner, J. J. Am. Soc. Mass Spectrom. 1994, 5, 873-885. (13) Shaffer, S. A.; Tang, K.; Anderson, G. A.; Prior, D. C.; Udseth, H. R.; Smith, R. D. Rapid Commun. Mass Spectrom. 1997, 11, 1813-1817. (14) Douglas, D. J.; French, J. B. J. Am. Soc. Mass Spectrom. 1992, 3, 398-408. (15) Chen, Y.-L.; Collings, B. A.; Douglas, D. J. J. Am. Soc. Mass Spectrom. 1997, 8, 681-687. (16) Tolmachev, A. V.; Chernushevich, I. V.; Dodonov, A. F.; Standing, K. G. Nucl. Instrum. Methods Phys. Res. B 1997, 124, 112-119.
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4111
Figure 1. The prototype ion funnel interface: (A) stainless steel pumping port; (B) PEEK heated capillary block holder; (C) stainless steel capillary tubing; (D) aluminum heated capillary block; (E) O-ring (O-V020) groove; (F) O-ring (O-V028) groove; (G) stainless steel front flange of the vacuum house; (H) O-ring (O-V154) groove; (I) tapped hole for 8-32 screw; (J) stainless steel vacuum house; (K) PEEK ring (ion funnel holder); (L) tapped hole for 4-40 screw; (M) tapped hole for 8-32 screw; (N) O-ring (O-V152) groove; (O) O-ring (O-V142) groove; (P) ceramic insulating washer; (Q) nickel-coated brass ion funnel electrode; (R) ceramic rod; (S) tapped hole for 2-56 screw; (T) O-ring (O-V125) groove; (U) tapped hole for 0-80 screw; (V) O-ring (O-V014) groove; (W) nickel-coated brass final orifice electrode (conductance limit); (X) final PEEK ring (ion funnel holder); (Y) stainless steel mounting peg.
quadrupole devices19,20 can be used for collisional focusing at pressures of 10-4 up to ∼1 Torr. However, rf quadrupoles and other rf-multipoles have limits in their utility for focusing more diffuse ion sources because of their relatively small effective acceptance apertures; i.e., the ion acceptance area is closely related to the ion emittance area.21-24 The pioneering work of Gerlich and co-workers25,26 demonstrated that ions could be effectively contained and/or transmitted through a series of stacked cylindrical ring electrodes of fixed ring diameter using rf electric fields of opposite polarity applied on adjacent electrodes, an arrangement that creates a “pseudopotential” that corresponds to a steep potential gradient formed near the surface of the electrodes and a near field-free region over most of the internal volume. In this arrangement, the electrodes can effectively perform as an ion guide or ion trap (the latter when “exit” is blocked by application of appropriate potentials) with ion confinement under vacuum being dependent on the ion’s initial velocity, ion m/z, pressure, rf frequency, and amplitude. This and other work with rf multipoles (17) Tolmachev, A. V.; Chernushevich, I. V.; Standing, K. G. Proceedings of the 45th ASMS Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, 1997; p 476. (18) Douglas, D. J. J. Am. Soc. Mass Spectrom. 1998, 9, 101-113. (19) Dodonov, A.; Kozlovsky, V.; Loboda, A.; Raznikov, V.; Sulimenkov, I.; Tolmachev, A.; Kraft, A.; Wollnik, H. Rapid Commun. Mass Spectrom. 1997, 11, 1649-1656. (20) Javahery, G.; Thomson, B. J. Am. Soc. Mass Spectrom. 1997, 8, 697-702. (21) Dawson, P. H. Quadrupole Mass Spectrometry and its Applications; Elsevier Scientific: Amsterdam, 1976. (22) Dawson, P. H. Mass Spectrom. Rev. 1986, 5, 1-37. (23) Miller, P. E.; Denton, M. B. Int. J. Mass Spectrom. Ion Processes 1986, 72, 223-238. (24) Tosi, P.; Fontana, G.; Longano, S.; Bassi, D. Int. J. Mass Spectrom. and Ion Processes 1989, 93, 95-105. (25) Bahr, R.; Gerlich, D.; Teloy, E. E. Verhandl. DPG (VI) 1969, 4, 343. (26) Gerlich, D. In State Selected and State-to State Ion-Molecule Reaction Dynamics. Part 1. Experiment; Ng, C. Y., Baer, M., Eds.; Wiley: New York, 1992; Vol. LXXXII, pp 1-176.
4112 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
at low pressures have shown that collisions with neutrals can lead to damping of ion motion and effective confinement within such ring electrode stacks or linear rf multipoles. A crucial attribute of the ion funnel concept we describe is that the ion acceptance characteristics of the device are effectively decoupled from the ion emittance, and in principle, arbitrarily large ion clouds can be effectively focused by reversing the Coulombically driven ion cloud expansion. In short, a diffuse ion cloud (i.e., from a plume of expanding gas and ions exiting from a heated inlet capillary into the first differentially pumped region of a mass spectrometer following electrospray ionization) can be focused and transmitted through a relatively small exit aperture. The small exit aperture feasible with the ion funnel is compatible with the acceptance aperture of an rf multipole operating at lower pressure in an adjacent differentially pumped region, thus providing efficient ion transport to the mass analyzer. Here we present the design of an ion funnel prototype and discuss ion current measurements and mass spectra obtained by operating this prototype with a commercial triple-quadrupole mass spectrometer. The results unambiguously support the ion funnel concept and indicate the basis for obtaining significant improvement in the already impressive sensitivity obtainable with ESI-MS. EXPERIMENTAL SECTION All experiments were performed using a Finnigan TSQ 7000 triple-quadrupole mass spectrometer (Finnigan MAT, San Jose, CA) either modified with an ion funnel interface or using the standard ESI ion source, as indicated. The prototype ion funnel design (Figure 1) consists of a 28element stack of 1.59-mm-thick nickel-coated brass ring electrodes (38 mm o.d.) that begins with an initial inner diameter of 22.15 mm and decreases parabolically to a final electrode inner diameter of 1.00 mm. The inner dimensions of all the electrodes are listed
Table 1. Ion Funnel Electrode Inner Diameters electrode no.
i.d. (mm)
electrode no.
i.d. (mm)
electrode no.
i.d. (mm)
1 2 3 4 5 6 7 8 9 10
22.15 20.61 19.13 17.71 16.34 15.04 13.79 12.60 11.47 10.40
11 12 13 14 15 16 17 18 19
9.38 8.42 7.52 6.68 5.90 5.17 4.51 3.90 3.35
20 21 22 23 24 25 26 27 28
2.85 2.42 2.04 1.72 1.46 1.26 1.11 1.02 1.00
in Table 1. The electrodes have a rounded and polished inner surface and are equally spaced from each other using 1.59-mmthick ceramic insulating washers. The electrodes and washers are mounted on four 107-mm-long (3.18-mm-diameter) ceramic rods using four tapped holes (equally spaced on d ) 31.75 mm) on each electrode. Additionally, each electrode has four slots (8.9 mm wide, 5.1 mm deep, all equally spaced) to facilitate connection of electrical components in the relatively tight enclosure of the vacuum housing. The entire electrode assembly is mounted on a polyether ether ketone (PEEK) ring (86.1-mm o.d., 25.4-mm i.d., 6.35 mm thick, and mounted adjacent to the largest inner diameter ion funnel electrode) with 4 holes to fit the ceramic mounting rods, 12 holes (5.1-mm diameter all equally spaced on d ) 47.0 mm) to facilitate electrical connections, and 6 additional holes to mount the ion funnel (by 4-40 screws) to the inside of the vacuum housing. The electrode assembly in turn mounts onto a final PEEK ring (49.5-mm o.d., 3.8 mm thick) following the final electrode of the ion funnel which has four equally spaced holes (3.18-mm diameter equally spaced on d ) 31.75 mm) 2.54 mm deep in which the ceramic mounting rods make a “press” fit. The final PEEK ring has a centered 25.4-mm-diameter, 3.18-mm-deep hole to mount a nickel-coated brass final orifice electrode (25.4mm o.d., 1.0-mm i.d., 1.6 mm thick) by six equally spaced 0-80 screws. The final PEEK ring further extends on its other side an additional 4.6 mm with an outer diameter of 30.5 mm and an inner diameter of 10.16 mm. This allows a secure fit into the vacuum housing as depicted in Figure 1. A voltage divider was used to provide a linear dc voltage gradient between electrodes 1 and 25 and consisted of one 1/4W, 22-MΩ ((10%) carbon resistor (Allen-Bradley, Bellevue, WA) soldered between each adjacent electrode. Additionally, a 22-MΩ resistor was soldered to electrodes 1 and 25 through which the initial and final potentials from the dc power supply were connected, respectively. These two leads allowed independent control of the initial and final potentials of the dc gradient. The final three electrodes (i.e., electrodes 26-28) and the final orifice electrode were independently connected without a resistive load to separate outputs of the dc power supply. All dc potentials to the ion funnel originated from a high-voltage mainframe dc power supply (model 1454, LeCroy, Chestnut Ridge, NY). Rf voltages of equal amplitude but opposite phase were applied between adjacent electrodes. Capacitors were utilized to decouple the rf and dc power sources. Further, since the capacitance between adjacent electrodes increases as the internal diameter of the electrodes decreases, a large relative value for the capacitors
Figure 2. Rf circuits for the ion funnel: (A) adjustable rf/dc coupler applied to electrodes 26-28; (B) high-Q-head. Full details of the electronics are given in the text.
was chosen to avoid a capacitive gradient. The capacitors were attached by soldering one 680-pF ceramic capacitor (3 kV dc maximum; Sprague-Goodman, Westbury, NY) to each electrode but alternating the position of attachment to opposite sides of the electrode assembly between adjacent electrodes. By the latter arrangement, a bus bar (tinned copper) was soldered to each of the two rows of capacitors and thus provided the two leads for rf voltage of equal amplitude but opposite phase. Capacitors were pressed tightly into the areas formed by the slots on each electrode (see electrode description) and pieces of 0.5-mm-thick Teflon sheeting (Laird Plastics, West Palm Beach, FL) were placed between the capacitors and the electrodes to prevent electrical discharge. In the cases where a variable rf amplitude was applied on electrodes 26-28 (as compared to the nominal rf amplitude set on electrodes 1-25) the 680 pF capacitors were removed and both the rf and dc potentials were coapplied externally to the ion funnel inside a shielded (aluminum) box (i.e., to prevent rf emissions) using an adjustable rf/dc coupler shown in Figure 2A. The circuit consists of four (C1-C4) 9-110-pF air variable capacitors (4 kV dc maximum; Surplus Sales of Nebraska, Omaha, NE), four (C5C8) 1-nF ceramic capacitors (3 kV dc maximum; SpragueGoodman), and eight (R1-R8) 2-W, 10-MΩ carbon resistors (Allen-Bradley). In short, lowering the value of the variable capacitors reduces the rf amplitude on the ion funnel electrodes. High-value resistors allow coupling of the rf and dc potentials external to the ion funnel; this coupling was needed only because of a limited number of electric feedthroughs. The rf signal originated from a wave form generator (model 33120A, Hewlett-Packard, Palo Alto, CA), was amplified using a 150-W broad-band rf amplifier (model 2100L, ENI, Rochester, NY), and passed through an in-house-built high-Q-head. The high-Qhead (Figure 2B) converts the unbalanced output from the rf amplifier into a balanced output (i.e., signals of equal amplitude and 180° out of phase with each other) for the ion funnel using a Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4113
1:1 impedance balun transformer (T1) consisting of a Toroidaltype core (Amidon, Santa Ana, CA) wound with 14 turns of 14gauge Formvar magnet wire with bifilar windings (Amidon). The circuit is housed in a shielded (steel) box and the combination of the 50-µH inductors (L1, L2; wound on Toroidal-type cores with 31 turns of 14-gauge Formvar magnet wire), the 30-300-pF air variable capacitor (C1; Surplus Sales of Nebraska), and the capacitance of the ion funnel produce a series resonant circuit. The Q or quality factor of the circuit is largely determined by the 50-W, 25-Ω noninductive power resistors (R1, R2; Cesiwid, Niagara Falls, NY) and is ∼10 (i.e., output voltage ) 10× input voltage) when operating at 1 MHz. The variable capacitor serves to finetune the amplitudes of the two rf outputs. The resonant frequency for the ion funnel using the high-Q-head was ∼700 kHz and was thus the operating frequency for the majority of the work reported in this study. However, when the adjustable rf/dc coupler was employed to lower the rf levels on electrodes 26-28, the resonant frequency shifted to ∼825 kHz and thus defined the operating frequency used for those studies. The front flange of the stainless steel vacuum house (Figure 1) was fitted with a 18.4-mm-i.d. elbow pumping port, a 62.7-mmlong aluminum (7000 series) block for heating the inlet capillary, eight welded electric feedthroughs providing rf and dc potentials to the ion funnel, and eight clearance holes (equally spaced on d ) 106.7 mm) to mount the flange to the vacuum housing using 8-32 screws. The front end of the aluminum block was threaded to fit a 76-mm-long, 1.6-mm-o.d., 0.51-mm-i.d. stainless steel capillary (Alltech, Deerfield, IL) held in place by a Swagelock (Solon, OH) fitting. Three 3.2-mm-diameter, 41-mm-deep holes (equally spaced on d ) 12.3 mm) were drilled in the front of the aluminum block to house two 3.18-mm-diameter stainless steel cartridge heaters (100 W, 120 V; Omega, Stamford, CT) and a Teflon-insulated thermocouple wire (Type K, Omega). The thermocouple wire was inserted into a hollow ceramic rod (3.1mm o.d., 1.6-mm i.d., 45 mm long) containing vacuum grease (Dow Corning, Midland, MI) to make good thermal contact with both the wire and the block. The temperature was regulated using a 110-V variable AC transformer (Staco, Dayton, OH) coupled to a programmable temperature controller (model CN 9000A, Omega). The TSQ 7000’s standard (1.0-mm-i.d.) skimmer and octapole ion guide (117.5-mm-long, 2.0-mm-diameter rods equally spaced on d ) 6.0 mm) were removed and a new octapole, made longer to fill the space created by removing the skimmer, was implemented (139 mm long, same rod size and spacing). In this arrangement, the conductance limit from the first stage pumping to the octapole ion guide is set by the final orifice electrode of the ion funnel. The ion funnel assembly was mounted into the stainless steel vacuum house (lined with 0.5-mm-thick Teflon sheeting to prevent electrical discharge) and the assembly fit into a modified ion source block on the TSQ 7000 mass spectrometer. Two stainless steel (2.4-mm diameter, 6.6 mm long) pegs on the vacuum housing inserted into holes drilled inside the ion source block fixing the exit to the ion funnel directly in front of the octapole entrance on the mass spectrometer. Additionally, eight 8-32 screws mount the vacuum housing directly to the source block. Vacuum seals were provided by Viton (DuPont Dow Elastomers, Wilmington, DE) O-rings as indicated in Figure 1. 4114 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
Initial electrospray ion current measurements were measured on the final orifice electrode tightly covered with aluminum foil. The measurements were made at ground potential on the foil using a Keithley (model 480, Cleveland, OH) picoammeter. The ion funnel region was pumped via the pumping port on the front flange of the vacuum housing utilizing a Leybold (Export, PA) mechanical pump (267 L/min). The pressure was measured by a convection gauge mounted just outside the vacuum housing which read ∼1.6 Torr (the actual pressure in the ion funnel due to displacement of the gauge is estimated to be a factor of ∼2-3 higher). The dc gradient on the ion funnel was as follows: initial gradient potential (electrode 1), 300 V; final gradient potential (electrode 25), 100 V; electrode 26, 95 V; electrode 27, 85 V; electrode 28, 50 V. Experiments at increased pressure were achieved by partially closing a block valve (Kurt Lesker, Clairton, PA) located between the ion funnel and the first-stage mechanical pump. For the remaining ion current measurements and for the acquisition of mass spectra, the ion funnel utilized a Leybold mechanical pump with a pumping speed of 600 L/min. The other mechanical pump (267 L/min) was connected to the standard pumping port of the TSQ ion source block and pumped the region of the octapole ion guide through two 12-mm-wide semicylindrical pumping channels cut in the ion source block directly between the vacuum housing and the block. The pumping channels are a nonoptimum design and resulted from the previous ion funnel design in which a skimmer was utilized between the funnel and the octapole.13 With this arrangement, the pressure in the ion funnel was ∼1.3 Torr (as read off the convection gauge) and ∼(2-3) × 10-6 Torr in the mass analyzer chamber. The applied dc potentials for these studies were as follows: initial gradient potential (electrode 1), 225 V; final gradient potential (electrode 25), 80 V; electrode 26, 70 V; electrode 27, 50 V; electrode 28, 25 V; final orifice electrode, 10 V. The current transmitted to the octapole ion guide was measured by tightly covering the entrance to the octapole with aluminum foil and then measuring the current with a Keithley (model 617) picoammeter. Ion current entering the mass spectrometer was measured using the picoammeter via a nickel-coated brass plate (38-mm-o.d.) located ∼5 mm beyond the exit of the heated inlet capillary. Electrospray emitter “tips” were made by pulling 0.185-mmo.d., 0.050-mm-i.d. fused-silica capillary tubing (Polymicro Technologies, Phoenix, AZ). The electrospray voltage was 2.0 kV and the inlet capillary was biased at 500 V (ion funnel interface only) using dc power supplies (models 305 and 303, respectively, Bertan, Hicksville, NY). Mass spectra and ion current measurements were obtained at an ESI flow rate of either 200 or 400 nL/min using a Harvard syringe pump (South Natick, MA). The heated inlet capillary was maintained at a temperature between 170 and 215 °C. The ion funnel was operated at a frequency of 700 kHz or as otherwise indicated. For comparison, mass spectra were acquired using the standard TSQ 7000 ESI ion source equipped with a 114-mm-long and 0.4-mm-i.d. heated inlet capillary using operating and tuning conditions similar to that used with the ion funnel. The mass spectra obtained with the standard ESI ion source were measured with three different Finnigan inlet capillaries (identical dimen-
sions) for the data presented (e.g., reconstructed ion currents). In either the case of the ion funnel or standard ESI ion source, the mass spectrometer was tuned to maximize ion transmission and obtain identical resolution for selected peaks from a 2.9 µM solution of horse heart myoglobin or a mixture containing 2.9 µM horse heart myoglobin and 20 µM synthetic Phe-Met-Arg-Phe amide, depending on the required mass range. Conditions such as electrospray voltage (2.0 kV), inlet capillary temperature (200 °C), electron multiplier voltage (1200 or 1400 V), sample flow rate (200 or 400 nL/min), acquisition scan rate (typically m/z 2002500 in 3 s), and total acquisition time (1- or 2-min averages) were held constant when spectra from the two designs were directly compared. The ion source block was pumped by an Edwards (Wilmington, MA) mechanical pump (549 L/min). The pressure measured in ion source block (i.e., between the inlet capillary and the skimmer) was 870-915 mTorr and in the region of the mass analyzer was ∼(2-4) × 10-6 Torr. All of the data presented was reproduced at least twice. Myoglobin (horse heart), cytochrome c (horse heart), ubiquitin (bovine red blood cell), gramicidin S (Bacillus brevis, hydrochloride salt), Phe-Met-Arg-Phe amide (synthetic), poly(ethylene glycol) (average MW, 8000), methanol, and glacial acetic acid were purchased from Sigma (St. Louis, MO). Standard solutions were prepared in methanol/deionized water/acetic acid (50:50:1%) except for poly(ethylene glycol), which was prepared in methanol/ deionized water (50:50). Solutions were kept refrigerated and were prepared from the corresponding standard material biweekly or as needed. RESULTS AND DISCUSSION The purpose of the ion funnel interface is to realize improved sensitivity by more efficient transmission of the electrospray ion current to the mass analyzer. The ion funnel’s ability to do this rests upon three aspects of operation: (a) efficient capture of the electrospray ion plume emanating from the heated capillary, (b) effective collisional focusing of the ions in the ion funnel through the use of rf fields, and (c) the imposed drift of the ions toward the “bottom”/exit of the funnel due to the dc potential gradient. Results in this paper support these basic premises, and we report the operating conditions and performance of this interface in terms of both ion current measurements and mass spectra. Ion Current Measurements. Initial experiments involved measuring ESI current collected on a plate at ground immediately following the final electrode of the ion funnel. Figure 3A shows a plot of detected current measured for the 100-400 Vpp rf amplitude range from ESI of a 58 µM bovine ubiquitin solution. Beginning at 15 pA, corresponding to a “dc-only” mode of operation, the detected ion current increases as the rf amplitude is increased to a maximum exceeding 1800 pA. This 2 orders of magnitude increase in detected current demonstrates that the presence of rf fields with this device clearly results in improved ion focusing. The effects of rf fields at the bottom of the funnel were explored in particular because it is a region where space charge and other effects are likely to be most problematic. Using the adjustable rf/dc coupler (see Experimental Section), the rf amplitude on electrodes 26-28 was reduced relative to the nominal rf amplitude on electrodes 1-25. Note that the change in operation frequency from 700 to 825 kHz reflects the change in resonating frequency of the series circuit (i.e., the adjustable
Figure 3. Ion current measured on the final orifice electrode from a bovine ubiquitin solution with the following concentration and rf operating conditions: (A) 58 µM, full rf (700 kHz) on all ion funnel electrodes; (B) 58 µM, full rf (825 kHz) on electrodes 1-25 and 80% of the nominal rf on electrodes 26-28; (C) 58 µM, full rf (825 kHz) on electrodes 1-25 and electrodes 26-28 operated in the dc-only mode; (D) 5.8 µM, full rf (700 kHz) on all ion funnel electrodes; (E) 0.58 mM, full rf (700 kHz) on all ion funnel electrodes. Inlet capillary temperature, 170 °C.
rf/dc circuit, high-Q-head, and ion funnel). ESI of the same ubiquitin solution and operating at 80 and 0% of the nominal rf amplitude applied to electrodes 1-25 yields a maximum ion current of 1.1 and 0.5 nA, respectively (Figure 3B and C). The overall shapes of these two curves are similar, but the overall amount of detected ion current is reduced to less than half by operating ion funnel electrodes 26-28 in the dc-only mode. Interestingly, the shape of the curve at 700 kHz is markedly different and shows a much sharper transmission maximum than observed at 825 kHz. Currently, work in our laboratory is aimed at elucidating the effects of frequency using an improved rf circuit that allows operation of the ion funnel over a broad frequency range. Thus, the data show that the rf fields clearly mediate the ion current focused through the interface and that the presence of rf fields in the bottom of the funnel can strongly affect ion transmission through the ion funnel device. To accomplish effective capture of the expanding ion plume, first the exit of the heated capillary was positioned to be both flush with the opening of the first electrode and aligned with the central axis of the funnel. This choice was based in part on results that indicated maximum ion currents (58 µM ubiquitin solution) detected when the heated capillary was flush with the opening of the first electrode. Second, the heated inlet capillary was maintained at a higher relative potential to that of electrode 1 to assist ion transport into the entrance of the ion funnel. For example, for positive ions, with the initial potential of the dc gradient on electrode 1 set at 300 V, ion transmission (same ubiquitin solution) was consistent for a heated inlet capillary potential in the 300-500-V range. However, if the capillary potential was lowered to 200 V, then the observed transmission in ion current decreased to ∼70% of the values observed for a capillary voltage in the 300-500-V range. The latter observation corresponds to a fraction of the ions electrostatically rejected from entering the funnel. Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4115
Ion currents were also measured as a function of concentration for ubiquitin solutions ranging from 0.58 to 58 µM (Figure 3A, D, and E). The detected current increases (although not linearly) with the concentration of the analyte. This indicates that the majority of the detected ion current for higher concentrations is lower m/z related and not solvent related ions and/or charged droplets, both of which would be inefficiently transmitted by the ion funnel. A particularly interesting aspect of these measurements is the roll off of transmitted ion current above rf amplitudes of ∼170 Vpp. This is tentatively attributed to the expected lowm/z cutoff due to the instability of low-m/z ions at higher rf amplitudes, an attribute that is similar to that found with rf multipoles.21,22 The results also indicate a maximum ion transmission for the 5.8 and 0.58 µM ubiquitin solutions to be at ∼170 Vpp, a lower rf amplitude than for the 58 µM solution, which has a maximum at ∼220 Vpp. We suspect this occurs because at high analyte concentrations the average distribution of charge states shifts to higher m/z values (which are most efficiently focused at higher rf levels). These data are consistent with mass spectra (not shown) taken with the standard ESI ion source which yield significantly higher relative abundances of high-m/z charge states for the 58 µM solution than for either the 5.8 or 0.58 µM solution. The increase in abundances of higher m/z (i.e., lower charge states) with increasing concentration is well established for ESI of biopolymers.3,27 The effects of pressure were explored by partially closing a valve located between the ion funnel and the first-stage mechanical pump. As the pressure in the ion funnel is raised, a higher rf amplitude is required to achieve similar ion transmission than when measured at lower relative pressure (Figure 4A). For the 1-10-Torr range, as measured using the convection gauge, maximum ion currents were achieved for the 1-5-Torr range but above this the required rf amplitude needed to maximize ion transmission was above the rf breakdown threshold (i.e., 400500 Vpp) of the ion funnel. Increasing the size of the inlet capillary from a 510- to 760-µm inner diameter will let in more ions, but consequently results in a higher operating pressure (7.1 Torr, unless other aspects of the design are changed) and thus results in a larger rf requirement to focus the available ions. Note that the appearance of this curve (Figure 4B) is similar to the curve measured at 7.8 Torr with the 510-µm-i.d. capillary (Figure 4A). Therefore, there exists a useful operating pressure range for the ion funnel at a given rf frequency, and this operating range is in practice determined on the low end by the size of the inlet capillary and the pumping speed applied to the ion funnel region and on the high end by the onset of rf electrical breakdown between ion funnel components. A similar operating pressure has been noted for a novel ion conduit which operates in part on the basis of rf electric fields.28 Ion current transmitted to the octapole ion guide was measured for 29 and 2.9 µM solutions of horse heart myoglobin for the 0-350 Vpp rf amplitude range (Figure 4C). Similar to the results obtained with ubiquitin, the maximum ion current displays a 2 order of magnitude increase compared to the ion funnel operating in the dc-only mode. An important figure of merit for the ion (27) Winger, B.; Light-Wahl, K. J.; Loo, R. R. O.; Udseth, H. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 1993, 4, 536-545. (28) Hars, G.; Meuzelaar, H. LC. Rev. Sci. Instrum. 1997, 68, 3351-3356.
4116 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
Figure 4. (A) Ion current measured on the final orifice electrode using full rf (825 kHz) on electrodes 1-25 and 80% of the nominal rf on electrodes 26-28 from a 58 µM bovine ubiquitin solution using a 510-µm-i.d. inlet capillary with first-stage pumping in the ion funnel regulated to six selected pressures. Inlet capillary temperature, 170 °C. (B) Ion current measured on the final orifice electrode using full rf (825 kHz) on electrodes 1-25 and 80% of the nominal rf on electrodes 26-28 from a 58 µM bovine ubiquitin solution using a 760µm-i.d. inlet capillary (7.1 Torr). Inlet capillary temperature, 170 °C. (C) Ion current measured on the octapole ion guide using full rf (700 kHz) on all electrodes for a horse heart myoglobin solution with a concentration of 29 and 2.9 µM. Inlet capillary temperature, 215 °C.
funnel is the fraction of total current entering the interface that is effectively transmitted. The ion current entering the vacuum chamber and directed toward the entrance to the ion funnel was measured immediately following the exit of the inlet capillary (∼5 mm). Table 2 gives the currents measured for myoglobin, cytochrome c, and gramicidin S solutions. These values allow an overall minimum transmission efficiency estimate for the ion funnel of approximately 21-25% for the protein solutions. In fact, the actual transmission of the ion funnel is certainly higher since the total current measured at the entrance includes both low-m/z (solvent related) and high-m/z droplet (e.g. residue) components. The low-m/z ions will not be transmitted (due to instabilities in the applied rf fields) while the high-m/z ions will not be focused at the applied rf amplitude and will be transmitted with very low efficiency. Previous measurements based upon ion mobility studies suggest that these contributions account for ∼50% or more of the total current for the present solutions.4 Thus, the overall efficiency of protein ion transmission through the ion funnel for the analytically significant portions of the ion current transmitted
Table 2. Ion Current Measured on Octapole Ion Guide Using the Standard Ion Source (A), Ion Funnel (B),a Ion Current Measured Entering the Ion Funnel (C), Ratio of B/A, and Ratio B/C sample (µm) myoglobin 29 2.9 cytochrome c 40 4.0 gramicidin S 3.0
A (pA)
B (nA)
C (nA)
B/A
B/C (%, ×100)
77 18
1.5 0.75
6.0 3.2
19 42
25 23
57 20
1.4 0.84
5.8 4.0
25 42
24 21
15
0.13
2.7
9
5
a Measured at 700 kHz with 98 V except gramicidin S, which used pp 75 Vpp.
through the inlet capillary is likely 50% or greater. The transmission efficiency for the peptide, however, is lower by a factor of ∼5. This stems from the fact that there is an apparent low-m/z cutoff for the ion funnel, i.e., a low-mass limit to which ions are not efficiently transmitted through the interface. In our laboratory, current research is aimed at a complete understanding of the transmission properties and improving the m/z transmission range of the interface. Ion current transmitted to the octapole ion guide was measured using the standard Finnigan ESI ion source for selected concentrations of myoglobin, cytochrome c, and gramicidin S (Table 2). The ratio of the ion current measured with the ion funnel over the ion current measured with the standard ESI ion source can be used to estimate an upper limit for the effectiveness or overall sensitivity gain using the present ion funnel design. For the proteins studied, the ratios indicate that the ion funnel delivers a 20-40 times greater ion current to the octapole ion guide (and eventually the mass analyzer) than the standard ESI ion source. The peptide gave a ratio of only 9 times the ion current over that of the standard ESI ion source and can be directly attributed its lower transmission efficiency through the ion funnel, as noted. Mass Spectra. Mass spectra for selected protein and peptide solutions were acquired with the prototype ion funnel mounted directly in front of the octapole ion guide using a Finnigan TSQ 7000 triple-quadrupole mass spectrometer. The relative ion current (RIC), detected by the mass spectrometer, was then compared to the RIC obtained with the standard ESI ion source under identical multiplier and other operating conditions. An example of such a comparison for a 4.0 µM solution of horse heart cytochrome c is shown (Figure 5A and B). The spectrum obtained using the ion funnel displays 10 times the RIC and over 20 times the base peak intensity compared to the spectrum with the standard ESI source. Interfacing the ion funnel directly to the octapole ion guide eliminates the use of a skimmer. In fact, replacement of the skimmer by a simple conductance limiting aperture (i.e., final orifice electrode with a similar internal diameter of 1.0 mm) led to a factor of 2-3 increase in the RIC measured for all of the protein solutions studied. Hence, in this new design, the ions are more efficiently transmitted to the octapole ion guide, which enables a lower potential gradient to be used between the final orifice electrode and octapole ion guide. This characteristic is generally desirable since it minimizes the likelihood of undesired
collisional activation in this region, which may induce dissociation or preclude detection of noncovalent complexes. Ratios of relative ion current were derived from mass spectra for solutions of myoglobin, cytochrome c, and gramicidin S (Table 3). When comparing the RIC measured using the ion funnel to the standard ESI ion source, the ion funnel yielded a 12-14 times improvement over the standard ESI ion source for the proteins. The measurements for the standard ion source were obtained with three different inlet capillaries, all equivalent in dimensions but which differed in performance. For this reason, the results for the least sensitive capillary were dropped while the results for the two most sensitive capillaries were averaged, the latter being in good agreement. The RIC ratios derived from the mass spectra are more consistent and are significantly lower than the ratios derived from ion current measured on the octapole given in Table 2. The result in Table 3 for gramicidin S, however, displays a significantly smaller gain of only 3 times the peak intensity based on its 2+ charge state, the dominant ion in its spectrum under acidic conditions. This observation is in line with the low-mass cutoff of the prototype interface, i.e., a lower limit in m/z for which ions are not efficiently transmitted through the device. Work with other singularly charged peptides indicates a nominal cutoff at m/z ∼500 for the present design and operating conditions. This cutoff, and indeed the entire transmission window, can be illustrated by comparing the spectrum of poly(ethylene glycol) (average molecular weight, 8000) obtained with both ESI interfaces (Figure 5C and D). The spectrum taken with the ion funnel yields a transmission window of ∼2 (i.e., high m/z/low m/z) or less than 1000 m/z units at the rf amplitude used for these examples. As expected, the rf amplitude has a direct effect on the lowm/z cutoff of the interface and the transmission window. This effect is illustrated with mass spectra obtained using a 29 µM solution of horse heart myoglobin (Figure 6). At first, as the rf amplitude is increased, the signal intensity for all of the charge states (i.e., 26+ to 12+) increase until the ions of low m/z (i.e., the high-charge states) are unstable by the imposed rf fields and are therefore unable to be transmitted through the ion funnel. Continuing to increase the rf amplitude increasingly shifts the lowm/z cutoff to higher m/z values. As the low-m/z ions are lost, the higher m/z ions are more effectively focused through the ion funnel. This effect is shown in Figure 7, which plots the RIC and selected peak intensities of individual charge states for the same myoglobin solution. The 19+ charge state (m/z 893.1), typically the base peak in the ESI mass spectrum for denatured myoglobin obtained with a conventional ion source, is the base peak in the spectrum for an rf amplitude of up to ∼100 Vpp after which its intensity is sharply reduced due to its instability in the higher rf fields. As the rf amplitude is increased, the lower charge states (e.g., 12+, 10+, and 7+ shown) sequentially increase in relative abundance. The expected linear relationship is evident by plotting m/z versus the rf amplitude needed to maximize the peak intensity for a given charge state (Figure 8A). Increasing the rf amplitude increases the RIC of the myoglobin spectra to 150 Vpp where the overall RIC begins to decline (Figure 7). Operating the ion funnel at 150 Vpp rf (700 kHz) results in an increase in RIC by over 50 times compared to the ion funnel Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4117
Figure 5. Mass spectra of a 4.0 µM horse heart cytochrome c solution taken with (A) the ion funnel or (B) the standard ESI ion source. Mass spectra of a 0.25 mg/mL poly(ethylene glycol) (average MW 8000) solution taken with (C) the ion funnel or (D) the standard ESI ion source. Inlet capillary temperature, 200 °C. Table 3. Ratio of Relative Ion Current Obtained from Mass Spectra Measured with the Ion Funnel Prototype Divided by That Measured with the Standard Ion Sourcea ratio 29 µM myoglobin 2.9 µM myoglobin 40 µM cytochrome c 4.0 µM cytochrome c 3.0 µM gramicidin S
12 12 12 14 3
a Ion funnel operated at 700 kHz (98 V , except for gramicidin S, pp which used 75 Vpp). Ratios based on RIC for the proteins and peak intensity for the 2+ charge state (m/z 572) for gramicidin S.
operating in the dc-only mode. Operation at fixed rf amplitude yields similar spectra (in terms of ion m/z) that increase in signal intensity until ∼70 Vpp after which the low-m/z cutoff begins to effect the spectrum by progressively removing the highest charge state on the lower m/z end of the spectrum. Since the effect of rf amplitude on the low-m/z cutoff is linear with m/z, this bias can be used to reduce space charge limits (and improve ion focusing through a conductance aperture) and/or remove lowm/z species from contributing to the capacity of ion trapping instruments. Evident from Figure 6 is that, at rf levels above 100 Vpp, there are a multitude of peaks that appear in the region of the low-m/z cutoff. These are products of collisional induced dissociation (CID) and originate from increased translational energy of lowm/z ions near their stability limit in the ion funnel at the given rf amplitude. Contributions from CID, which has great utility in other applications (i.e., MS-MS), can be effectively minimized by scanning the rf amplitude in-link with the m/z scan of the quadrupole mass analyzer. This method of scanning would also bring in the maximum intensity for all of the charge states produced by the ESI process. This advantage can be illustrated by plotting the maximum peak intensities of the given myoglobin 4118 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
Figure 6. Mass spectra of a 29 µM horse heart myoglobin solution acquired using the ion funnel operating with an rf amplitude (Vpp) as shown at 700 kHz. Base peak intensity is given in the upper right corner. Inlet capillary temperature, 215 °C.
charge states and comparing to the charge-state intensities obtained with most sensitive inlet capillary used on the standard ESI ion source (Figure 8B). A secondary benefit is that moderate amounts of collisional activation can be produced in the ion funnel to reduce contributions due to charge-state adduction. Note that
Figure 7. Log plot of RIC and selected charge-state intensities as a function of ion funnel rf amplitude (700 kHz) from mass spectra for a 29 µM horse heart myoglobin solution. Inlet capillary temperature, 215 °C.
in Figure 6 adducts associated with lower charge states are reduced as the rf level is increased. CONCLUSION An approach for focusing ions at high pressure has been developed and demonstrated in the context of an electrospray ionization source. A prototype ESI ion funnel interface has been shown to provide well over 1 order of magnitude increase in ion current and ion signal intensity for the mass spectra of biopolymers as compared to those spectra obtained with a conventional ESI ion source. This dramatic enhancement is attributed to ion funnel’s ability to effectively capture, focus, and transmit ions from the ESI heated inlet capillary region and thus provide a larger fraction of the ion current to the mass analyzer. Current work in our laboratory is aimed at optimization of ion transmission, including ion transmission improvements both at both lower m/z and over broader m/z ranges. We expect that these advances will enable both a significant improvement over the results presented to date using the present ion funnel prototype and that this performance will enable new applications of ESI mass spectrometry (and other high-pressure ion sources in mass spectrometry) where improvements in sensitivity are required. In fact, given the effective ionization feasible with low-flow ESI sources (e.g., “nanospray”) and the potential of the ion funnel for producing very effective transmission
Figure 8. (A) Plot of rf amplitude versus m/z for maximum chargestate intensities from a 29 µM horse heart myoglobin solution using the ion funnel. (B) Maximum charge-state intensities (recorded at multiple rf amplitudes) using the ion funnel versus the charge-state intensities using the standard ESI source for a 29 µM horse heart myoglobin solution.
of the ions produced, it is conceivable that the best detection limits, which are now in the 100 zmol to attomole range, will be able to be extended into the low-zeptomole or even subzeptomole domain. ACKNOWLEDGMENT The authors thank Sergey Rakov and Professor Jean Futrell at the University of Delaware for helpful discussion. This research was supported by the Laboratory Technology Research Program of the Office of Basic Energy Sciences, U.S. Department of Energy. Additional support for this work came under a Cooperative Research Agreement with Finnigan Corp. (San Jose, CA). Pacific Northwest National Laboratory is operated by Battelle Memorial Institute for the United States Department of Energy through Contract DE-ACO6-76RLO 1830. Received for review February 25, 1998. Accepted July 8, 1998. AC9802170
Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4119
