Design and Implementation of a New Electrodynamic Ion Funnel

Apr 7, 2000 - ... model TSQ 7000 (ThermoQuest Corp., San Jose, CA), was modified to .... These results show that the rf electric fields in the ion fun...
4 downloads 0 Views 177KB Size
Anal. Chem. 2000, 72, 2247-2255

Design and Implementation of a New Electrodynamic Ion Funnel Taeman Kim, Aleksey V. Tolmachev, Richard Harkewicz, David C. Prior, Gordon Anderson, Harold R. Udseth, and Richard D. Smith*

Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352 Thomas H. Bailey, Sergey Rakov,‡ and Jean H. Futrell‡

Department of Chemistry and Biochemistry, University of Delaware, Newark, Delaware 19716

A new electrodynamic (rf) ion funnel has been developed and evaluated for use in the interface regions (at ∼1-10 Torr) of atmospheric pressure ion sources (e.g., electrospray ionization (ESI) for mass spectrometry). The ion funnel consists of a ring electrode ion guide with decreasing i.d. and with a superimposed dc potential gradient along the ring stack. The thicknesses of the ring electrodes and the spacings between them were reduced to 0.5 mm from 1.59 mm compared to those used for previous designs. The new ion funnel displays a significant improvement in low-mass transmission (m/z >200) and sensitivity compared to previous designs. The transmission efficiencies for electrosprayed peptides and proteins (ranging in mass from 200 to 17 000 Da) were typically 50-60% of total incoming currents from a heated capillary inlet. The transmitted ion currents were a factor of 30-56 greater than those of the standard interface for peptide samples and a factor of 18-22 greater than those for protein samples. The sensitivity gains realized at the MS detector were somewhat lower, possibly due to space charge effects in the octapole ion beam guide following the ion funnel. The improved ion transmission properties result primarily from the use of reduced spacings between ring electrodes. We also show that the ion funnel can be operated in two different modes, one using low-rfamplitude scans, allowing fragile noncovalent complexes (as well as generally undesired adducts) to be transmitted, and the other using high-rf-amplitude scans, providing greater collisional activation and more effective adduct removal (or the dissociation of lower m/z species). Atmospheric pressure ion (API) sources, including inductively coupled plasma, corona discharges, various laser desorption/ ionization sources, and particularly the electrospray ion source, are now widely used in mass spectrometry. Since ions are created at nearly atmospheric pressure, they must then be transferred to a mass analyzer that is generally located in a high vacuum. A differential pumping system, involving several stages of progres* Corresponding author. ‡ Current address: Pacific Northwest National Laboratory. 10.1021/ac991412x CCC: $19.00 Published on Web 04/07/2000

© 2000 American Chemical Society

sively lower pressure, is commonly used to achieve high overall transmission efficiencies. A major issue in such designs is the effectiveness of ion focusing through conductance-limiting apertures located in the higher pressure regions. Buffer gas cooling in rf ion guides has been previously used to improve ion transmission in differentially pumped stages at pressures ranging from 10-4 to 0.3 Torr.1,2 An axial field rf quadrupole with segmented rods also has been described for use in the first vacuum stage at pressures up to 3 Torr3 and at lower pressures in the collisional cell of a triplequadrupole mass spectrometer.4 However, such quadrupole ion guides have significant limitations for capturing ions that have divergent trajectories. The acceptance of a quadrupole ion guide is determined by the physical size of the quadrupole, which is limited in practice by the onset of electric breakdown phenomena, since the applied potential scales with the square of the rod diameter. The electric breakdown problem is most significant at pressures of ∼1 Torr, which is also a typical operating pressure in the interface regions of API-MS instrumentation. An electrodynamic ion funnel was previously developed in our laboratory to improve ion acceptance and to overcome the electrical breakdown problem. The earlier ion funnel designs demonstrated significantly improved transmission efficiencies for higher m/z ions but also some discrimination against low-m/z ions.5-7 This report describes developments resulting in significantly improved transmission for low-m/z ions. We also show that the new ion funnel can be used to transmit adduct ions at very low rf amplitudes and also effectively remove adducting species by collisional activation using higher rf amplitudes. (1) Douglas, D. J.; French, J. B. J. Am. Soc. Mass Spectrom. 1992, 3, 398-408. (2) Tolmachev, A. V.; Chernushevich, I. V.; Dodonov, A. F.; Standing, K. G. Nucl. Instrum. Methods Phys. Res., Sect. B 1997, 124, 112-119. (3) Dodonov, A.; Kozlovsky, V.; Loboda, A.; Raznikov, V.; Sulimenkov, I.; Tolmachev, A.; Kraft. A.; Wollnik, H. Rapid. Commun. Mass Spectrom. 1997, 11, 1649-1656. (4) Javahery, G.; Thomson, B. J. Am. Soc. Mass Spectrom. 1997, 8, 697-702. (5) Shaffer, S. A.; Tang, K.; Anderson, G.; Prior, D. C.; Udseth, H. R.; Smith, R. D. Rapid Commun. Mass Spectrom. 1997, 11, 1813-1817. (6) Shaffer, S. A.; Prior, D. C.; Anderson, G.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1998, 70, 4111-4119. (7) Shaffer, S. A.; Tolmachev, A. V.; Prior, D. C.; Anderson, G. A.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1999, 71, 2957-2964.

Analytical Chemistry, Vol. 72, No. 10, May 15, 2000 2247

EXPERIMENTAL SECTION Ion Funnel Interface. To increase the m/z transmission window and reduce low-m/z discrimination issues associated with the previous ion funnel design, the center to center spacing of the electrodes (d) was reduced to 1.0 mm from 3.18 mm7 while the length of the tapered section was extended until the last electrode was of 1.5 mm i.d. compared to 2.04 mm used for the previous prototype. Therefore, the ratio of the last ring radius to ring spacing (F/d) was increased to 0.75 from 0.32. The increased F/d ratio reduces the axial rf pseudopotential well depth at the bottom of the ion funnel (see Results and Discussion). The new ion funnel design also incorporated an additional constant-i.d. ring electrode “drift” section at the funnel entrance. That is, the funnel interface has two major parts: (a) a front section that consists of 55 constant 25.4 mm i.d. rings and (b) a rear section that has 45 ring electrodes with diameters uniformly decreasing from 25.4 to 1.5 mm. The front section reduces the gas dynamic effects upon ion confinement, allows improved conductance for pumping, reduces the gas load downstream of the ion funnel, and provides an extended ion residence time for enhanced desolvation of charged clusters or droplets. Since the new ion funnel had thinner electrodes, it required more precise mechanical alignment than the previous designs. The accurate fabrication of parts and alignment was facilitated by the following procedure: Brass plates, 0.5 mm thick, with three preliminary aligning holes for mounting and with electric connectors were fabricated by metal stamping (Blazak Mfg., Newark, NJ). The plates had square shapes and tabs to mount electrical connectors. The plates were stacked and aligned using the three holes in each element through which three alumina rods passed. The surfaces of central holes defined the ion funnel electrodes and were machined simultaneously by boring a stack of electrode plates with a tapered boring tool. The three alignment holes were also drilled in the stack of plates to ensure proper alignment of the central holes. All odd (or even) numbered plates in the stack were collected and used in assembly of an ion funnel (i.e., two ion funnels were assembled from each stack of lenses). The edges of the central holes were rounded using abrasive paper (600 grit) to reduce the likelihood of electrical breakdown between adjacent ring electrodes. The gaps between electrodes, made by removing the even (or odd) numbered electrodes, were filled by insulating spacers made from 0.5 mm thick Teflon sheets. Electric connections were made to each ring electrode by using two custom-made zero-insertion-force (ZIF) sockets (Tactic Mfg., Dallas, TX). Electric components were mounted on circuit boards that were soldered onto the ZIF sockets. Driving potentials were applied to connectors at the ends of each circuit board. A resistor chain on one circuit board works as a dc voltage divider, providing the desired dc bias to each ring electrode. The ring electrodes have a continuously decreasing dc bias (for positive ions) toward the bottom of the ion funnel electrode stack. Typical dc biases applied to the ion funnel were 160-300 V for first ring electrode, 0-10 V for last ring electrode (giving an axial electric field strength of 15-26 V/cm), and 0 V for the “dc-only” lens after the ion funnel. The other ZIF socket mounts two capacitor networks, one providing rf potentials to odd-numbered ring electrodes and the other providing rf potentials 180° out of phase to the evennumbered ring electrodes. The rf signal was generated by an rf 2248 Analytical Chemistry, Vol. 72, No. 10, May 15, 2000

Figure 1. Photograph of the assembled ion funnel.

generator (model 33120A, Hewlett-Packard, Palo Alto, CA), amplified by an rf amplifier (model 2100L, ENI, Rochester, NY), and split into two (out of phase) rf signals using a center taped transformer in an in-house-constructed wide-band tuning circuit. The assembled ion funnel with its electrical connections is shown in Figure 1. A triple-quadrupole mass spectrometer, model TSQ 7000 (ThermoQuest Corp., San Jose, CA), was modified to incorporate an ion funnel interface (see Figure 2). The 2 mm i.d. dc-only lens after the ion funnel was inserted to prevent rf cross-talk between the ion funnel and the octapole ion guide and was separated by a Teflon ring from last funnel electrode. Gas conductance, however, was mainly limited by the 1.5 mm diameter last ring electrode of the ion funnel. Ion transmission and mass spectra were studied for various peptide and protein samples over a range of pressures, rf amplitudes, and dc potentials. Samples and Electrospray Ionization Conditions. Four peptide and two protein samples were used to study the ion transmission properties of the funnel over a wide range of m/z values. Gly-Gln hydrate, Val-Gly-Asp-Glu (synthetic), leucine enkephalin (synthetic, acetate salt), gramicidin S (bacillus brevis, hydrochloride salt), cytochrome c (bovine heart), myoglobin (horse heart), methanol, and glacial acetic acid were purchased from Sigma (St. Louis, MO) and used in the sample preparation. The sample peptide and protein solutions were prepared at the indicated concentrations in 50:50:1 water:methanol:acetic acid solvent. The sample concentrations were roughly in the low millimolar range for peptides and in the few tens of micromolar range for proteins. These relatively high concentrations were chosen so that we could measure the analyte ion transmission efficiencies through the ion funnel with minimal influence from solvent-related ion currents. The same concentrations were used in the sensitivity gain measurements to compare these results with the ion transmission measurements. As is evident from our results,

Figure 2. Schematic diagram of the ion funnel interface in the context of the overall instrumentation.

it is likely that significantly greater overall ion transmission efficiency would be realized with the use of lower concentration samples (where effects due to space charge would be minimized). Solutions were infused by a syringe pump at flow rates of 0.20.5 µL/min, through fused-silica capillary “tips” (the ESI emitters). The tips were made by pulling 190 µm o.d. × 50 µm i.d. capillaries (Polymicro Technologies, Phoenix, AZ) while parts of the capillaries were heated. The 2-4 cm long electrospray tips were each connected to the syringe through a union (P720 MicroTight union, Upchurch, Oak Harbor, WA). Biases of 1800-2600 V were applied to the syringe needle. The distances between the ends of the electrospray tips and the heated metal inlet capillary was ∼1 mm. Heated Metal Inlet. A heated metal inlet capillary was used for the introduction of electrosprayed ions and charged droplets at atmospheric pressure. A 0.51 mm i.d. × 76 mm length stainless steel tube was inserted and silver-soldered into a central hole of a cylindrical heated stainless steel block. The metal block was heated using two cartridge heaters (60 W, Ogden, Arlington Heights, IL) inserted into two off-centered holes in the cylindrical block. The temperature of the capillary was monitored by a thermocouple inserted into another off-centered hole and maintained for all studies by a controller to temperatures of 190-210 °C. Pumping System. The ion funnel interface region was pumped by a Roots blower (84 L/s, model WSU251, Leybold, Koln, Germany). The pressure of the ion funnel chamber was monitored by a capacitance manometer (model CMLA-11-001, 10 Torr maximum, Varian, Lexington, MA) and was controlled in some experiments by adjusting a butterfly valve installed in the pumping line. The chamber containing the ion funnel operates at a pressure similar to, or slightly higher than, the pressure of the first vacuum stage of standard ESI interfaces (i.e., above 1 Torr). Since the conductance of the last ring electrode (1.5 mm i.d.) of the ion funnel was larger than that of the 1.0 mm i.d. skimmer of the standard interface, the gas load downstream was somewhat greater than that with the standard interface. To operate the mass spectrometer at an acceptable pressure, an additional turbo pump (345 L/s, model Turbovac 361, Leybold) was connected to the chamber of the octapole ion guide. Current Measurements. The incoming ion current (i.e., from the heated capillary inlet) transmitted to the ion funnel was measured by summing the currents flowing to the ion funnel and octapole ion guide. The transmitted current was determined using the current flowing to the octapole ion guide after the ion funnel

where the lens after the octapole was biased to +200 V to prevent further downstream transmission of ions. The transmitted current with the standard interface was also measured on the octapole ion guide while the common input of the picoammeter was floated to -20 V (to prevent losing ions to the grounded skimmer). During these measurements, the lens after the octapole was also floated to +200 V to again prevent the downstream transmission of ions. Mass Spectral Transmission with the Ion Funnel Interface. The m/z transmission windows for various rf amplitudes and the fragmentation phenomena were evaluated from the mass spectra. The relative ion currents (RICs) and peak intensities for the mass spectra as a function rf amplitude were compared with ion transmission results to determine if the transmitted current arose from analyte ions or from solvent-related ions and charged droplets having very low or high m/z values, respectively. To transmit m/z 200-2000 ions, the rf amplitude of the ion funnel was ramped with the m/z scan of the analyzing quadrupole by using an output of the TSQ 7000 spectrometer to control an rf function generator (HP 33120A, Hewlett-Packard). RESULTS AND DISCUSSION The operating principle of the ion funnel can be summarized as follows. The oscillating rf fields near the ring electrodes serve to push ions to the weaker electric field region in the central region of the ring electrodes. Concurrently, a low-dc electric field pushes the ions toward the electrodes having progressively smaller apertures (i.e., the bottom of the ion funnel) while buffer gas collisions reduce the energy and collisionally damp the motion of the ions. Thus, the effect of the ion funnel is to reduce both the spatial and kinetic energy distributions of ions. Numerical simulations and analytical studies showed that a major limitation of the previous ion funnel design resulted from the axial trapping of ions at the smallest ring electrodes (i.e., at the “bottom” of the ion funnel). The axial-trapping effect becomes significant when the i.d. of the ring electrode approaches the distance between adjacent ring electrodes. Ions can be transmitted after filling the axial-trapping well, but the trapped ions increase space charge effects and extend the ion cloud to larger radii, where ions can be lost or experience higher rf fields, which result in greater collisional activation (and possibly dissociation). The previous ion funnel design also provided an m/z transmission “window” that was narrower than desired. (This issue also applies to quadrupole ion guides but is progressively less problematic Analytical Chemistry, Vol. 72, No. 10, May 15, 2000

2249

Figure 3. Ion transmission efficiencies of the ion funnel as a function of the rf amplitude for various peptide and protein solutions. The ion funnel was operated at 700 kHz, with an axial electric field of 15 V/cm and at a pressure of 1 Torr. The measured currents were normalized to the incoming ion currents.

for increasing order multipole ion guides.8) This property can be beneficial (due to reduced space charge) for scanning instruments (such as quadrupole and magnetic sector mass spectrometers) where the transmission window can be scanned simultaneously with rf amplitude, but it is undesirable for instruments in which a wide m/z range is to be analyzed simultaneously (e.g., FTICR or ion-trap mass spectrometers). The radial trapping field approaches a quadratic potential as the ring i.d. approaches the ring electrode separation. In a quadratic field, ions lower than a certain m/z will be unstable above a specific rf amplitude and lost to the electrodes (leading to the low-m/z cutoff). Additionally, ions of higher m/z will be less effectively focused, leading to low transmission efficiency (i.e., a high-mass limit). These are the major factors defining the m/z transmission widow of the ion funnel. Ion Transmission. Figure 3 shows the rf amplitude dependence of ion transmission for six peptides and proteins having molecular masses from 204 to 16 969 Da and producing ions across the m/z 200-2000 range. The maximum charge transmission efficiency through the ion funnel was found to be 50-60% of the incoming current. The increase in charge transmission starts at a significantly lower rf amplitude (about 20 Vpp, 700 kHz) than that for the previous ion funnel design7 and remains high for a broad range of rf amplitudes for all samples. The reduced rf amplitude for maximum transmission results from the reduced electrode thicknesses and reduced spacings between the electrodes (the electrode thicknesses and spacings have been reduced ∼3 times compared to those of the earlier design). Thus the same

electric field can be produced with the new design with 9 times lower rf amplitude. When we consider the contribution to the incoming ion current due to solvent-related low-m/z ions and both droplet- and residuerelated high-m/z species, the transmission efficiencies for the analytically relevant ions are obviously higher than those based upon detected currents. An early ion mobility study with electrospray showed that at least 25-50% of the total ion current was contributed by the solvent-related low-m/z ions and that a much smaller amount of the total ion current (∼