A Paul Trap Mass Analyzer Consisting of Two Microfabricated Electrode Plates

We report the design and performance of a novel radiofrequency (RF) ion trap mass analyzer, the planar Paul trap, in which a quadrupolar potential distribution is made between two electrode plates. Each plate consists of a series of concentric, lithographically deposited 100-micrometer-wide metal rings, overlaid with a thin resistive layer. To each ring is applied a different RF amplitude, such that the trapping field produced is similar to that of the conventional Paul trap. The accuracy and shape of the electric fields in this trap are not limited by electrode geometry nor machining precision, as is the case in traps made with metal electrodes. The use of two microfabricated plates for ion trap construction presents a lower-cost alternative to conventional ion traps, with additional advantages in electrode alignment, electric field optimization, and ion trap miniaturization. Experiments demonstrate the effects of ion ejection mode and scan rate on mass resolution for several small organic compounds. The current instrument has a mass range up to ~180 Thompsons (Th), with better than unit mass resolution over the whole range. INTRODUCTION Since the invention of the radiofrequency (RF) quadrupole ion trap by Wolfgang Paul et al. in 1953, quadrupole ion trap mass analyzers have played an increasingly important role in chemical and biological analyses. In addition to high sensitivity and specificity, ion traps combine reasonable simplicity of operation with complex functions such as multi-stage tandem mass analysis using a single analyzer. However, the hyperboloidal electrode shape of the original Paul trap is difficult to machine, especially on the miniaturized scale. As a result, significant effort has been spent on the development of alternative ion trap structures. In 1998, Wells et al. demonstrated a mass-selective instability scan on an ion trap with cylindrical geometry. The cylindrical ion trap, which had been introduced previously, 4 simplified the hyperbolic ring electrode and end-cap electrodes with a cylindrical electrode and planar end-caps. This simplified trap geometry has facilitated and been the basis for most miniaturized ion trap systems. The performance of any ion trap depends to a large extent on the quality of the trapping electric field. In conventional mass analyzers, the electric field is determined by the shape and arrangement of a set of metal electrodes. Although curved (hyperboloidal) surfaces produce the most accurate electric fields, they are more difficult to fabricate accurately for miniaturized ion traps. Planar metal electrodes can be machined more easily, but even multiple planar electrodes, such as those used in rectilinear traps, must be accurately positioned and mounted. High performance in miniaturized ion traps requires accurate electric fields produced by geometrically simple electrode structures. Recently, a novel ion trap mass analyzer was presented by Austin el al., which was based on a toroidal (circular) trapping geometry and microfabrication technology. The device, called the Halo ion trap, consisted of a pair of planar ceramic plates mounted in parallel, in which the facing surfaces were lithographically imprinted with sets of concentric ring electrodes, then covered with a layer of resistive germanium. The electric fields, established by applying different RF potentials to each ring, produced the same field shape as that in the toroidal ion trap. Although this type of mass analyzer is of promise due to its high ion storage capacity, sensitivity, and ease of fabrication and miniaturization, its performance (e.g., resolution and mass range) as presented was not optimal. In the present work, the electrode approach of the Halo ion trap has been used to produce a mass analyzer of the Paul trap geometry. Whereas the electric fields of the Halo ion trap mimicked those of the toroidal trap, including a toroidal trapping volume, the electric fields in the present trap follow the design of the conventional Paul trap. Instead of the toroidal trapping volume of the Halo trap, ions in the present trap—the planar Paul trap—are confined to a small spherical volume at the device center. Although the larger trapping volume of the toroidal geometry is lost, the equations of ion motion are better understood in the Paul geometry. In particular, ion ejection is more straightforward. Construction of an ion trap mass analyzer using two microfabricated plates provides several important advantages. For instance, two pieces can be mechanically aligned more easily than a larger number of electrode pieces. Polished flat plates have a smoother surface than traps made using other methods. Hence surface roughness, which has been identified as an issue for miniaturized traps, is less of a problem. Microfabricated plates can be produced in quantity less expensively and more accurately than machined electrodes. The space between the plates provides convenient access for ionization sources, optics, pressure measurement, or other peripheral components. Finally, the use of an array of microfabricated electrode rings underneath a resistive layer allows the electric fields within the trap to be modified in a way that is not possible using machined electrodes. Although the microfabricated plates themselves are fairly complex in both design and fabrication, other advantages make this approach potentially valuable. EXPERIMENTAL SECTION Plate Fabrication Figure 1 shows the planar Paul ion trap implemented in this study. Each of the two trapping plates started as an aluminum oxide ceramic plate (99.6 % purity, Hybrid-Tek, Clarksburg, NJ) with dimensions of 46.95 × 36.20 × 0.635 mm. In each plate a central hole, lasercut to 1 mm diameter, was used for ion ejection. Holes for electrical connections between the front and back sides of the plate (vias) were laser drilled to 127 μm diameter, arranged in a spiral pattern, each at an increasing distance from the central hole. Due to constraints in via hole drilling near an edge in the ceramic plates, the width of the first ring was 1.30 mm. From the second ring to the 24th ring, the width was 0.10 mm, as indicated in Table 1. Additional holes were cut to fit positioning rods and screws used for trap assembly and alignment. After laser cutting, the via holes were filled with a gold-tungsten alloy, and both sides of the alumina substrates were polished to a surface roughness of better than 1 μm. The active trapping area of each plate was evaporatively coated with a 100-nm layer of germanium, which prevented unwanted charge build-up and established a continuous, well-defined electric potential surface over the network of underlying rings. After deposition of germanium, the electrical resistance between adjacent rings was on the order of 10-100 MΩ. The other fabrication procedures were the same as that of the Halo ion trap, and a detailed description was given in a recent publication. Experimental Setup Figure 1(b) shows the instrument setup for the experiments, including the electron gun assembly, trapping region, and the detector assembly. Behind each of the two ceramic plates comprising the trapping region was a printed circuit board (PCB) with a capacitor network. The capacitor network was used to establish the voltages on each of the ring electrodes under RF excitation. Springloaded pins were soldered to the PCB boards in order to make electrical contact with the back sides of the trapping plates. A 6-mm stainless steel spacer was mounted between the trapping plates. Holes in the spacer admitted the electron beam, sample vapor, helium gas, and a Teflon tube leading to a pirani gauge (Kurt J. Lesker, Clairton, CA). An RF signal with a frequency of 1.26 MHz and variable amplitude up to 738 V0-p (PSRF-100, Ardara Technologies, North Huntingdon, PA) was applied to the capacitor network on the PCBs, and the spacer was grounded during ion ejection. In addition, a supplementary low-voltage ac signal, generated using two 30 MHz synthesized function generators (DS345, Stanford Research Systems, Sunnyvale, CA) with 180 phase difference, and amplified to 3.5 V0-p by a custom-made amplifier, was applied between the trapping plates to provide a dipole field for resonant ion ejection during the RF scan. The amplified supplementary ac signals were applied to the innermost ring on each plate, using a simple filter circuit to isolate the supplementary ac from the main RF signals. The applied frequency of the ac signal was 290 kHz, and βz was approximately 0.46. Other values of βz up to ~1 were also tested, with comparable mass resolution but reduced peak intensity. Operational details of the planar Paul ion trap are given in Figure 2, which shows the time intervals and sequence for ionization, RF trapping, and ejection. First, the RF voltage was turned off to clear previously-trapped ions out of the trap. Then the RF was turned back on along with the electron gun, allowing sample to be ionized in the trapping volume. The electron gun was then turned off, allowing the ionized and trapped sample to collisionally cool. The ejection ac was then turned on, and a voltage sweep of the drive RF was initiated. As the RF amplitude reached a level at which the secular frequency of any ion matched the applied supplementary ac frequency, that ion was resonantly ejected from the trap. Because ejection voltage was ramped from lower to higher voltages, ions were ejected in order of increasing m/z out of the trap. Once an ion was ejected through the hole in the trapping plates, it continued toward the detector. Ejected ions were detected using an ETP electron multiplier detector (SGE Analytical Science, Austin, TX), with a conversion dynode operated at -4000 V. The signal was amplified (427 Current Amplifer, Keithley Instruments, Cleveland, OH) and recorded using a digital oscilloscope (WaveRunner 6000A, LeCroy, Chestnut Ridge, NY). In the experiments reported herein, helium was used as the buffer gas at an indicated pressure of 5.34 × 10 Torr (uncorrected, 1 Torr = 133 Pa) as read from a pirani gauge (Kurt J. Lesker, Clairton, CA). Headspace vapor of the organic compounds of interest, without further purification, was leaked into the vacuum through two Swagelok leak valves (Swagelok, Solon, OH) to maintain a nominal pressure of 1.0-8.0 × 10 Torr. In situ electron ionization was achieved using a custom-built electron gun comprising an iridium-tungsten filament, lens, gate, and a 1.6-amp power supply. Optimization of the electric field. As shown in Figure 3(a), the planar Paul ion trap consists of two parallel ceramic plates with facing surfaces imprinted with concentric metal rings, overlaid with germanium. The metal rings superimpose a potential function on the germanium layer, which in turn establishes the three-dimensional potential distribution of the trapping region. This method of producing the trapping field is distinctive from the method used in conventional ion traps— both those made using hyperboloidal or curved electrodes (e.g., Paul trap, quadrupole mass filter and linear ion trap) and from traps made using planar metal electrodes (e.g., cylindrical trap, rectilinear trap). The electric field within the planar Paul trap is a function of the potentials applied to each ring, as well as the spatial arrangement of the rings and plates. As the RF potential on each ring is independently adjustable, there is a great deal of flexibility in constraining and optimizing the trapping field. As with any other ion trap, the shape of the electric field inside the trap plays an important role in determining the performance of the ion trap as a mass analyzer. The ion motion for the present trap is governed by the RF electric field and by the auxiliary ac signal applied to the plates. Optimal electric fields for several trapping geometries have been reported by Ouyang el al. In general, the performance of any ion trap is influenced by components of the electric field that are a higher order than quadrupolar (i.e., octopole). After investigation of the electric field for the cylindrical ion trap with different dimensions, Wu et al. concluded that the increase of spectral resolution can be realized by appropriate compensation for high-order, nonlinear field components, particularly octopolar and dodecapolar fields. This approach was also used in the original Finnigan Ion Trap Detector, and was accomplished by increasing the spacing between electrodes. Lammert and co-workers reported that a certain amount of positive octopole contributed to the increase of resolution of the toroidal ion trap. For the planar Paul trap, the electric fields within the trapping volume were calculated, and voltages on individual rings optimized, using SIMION 7. The nonlinear components of the axial electric field (along r=0) in the planar Paul trap were selected so as to be similar to those used in the asymmetric toroidal trap and cylindrical ion trap. During the course of optimization of electric fields, several sets of potentials were identified as feasible. The one chosen for this study, as given in Table 1, gave the greatest linear axial field of those examined. Surprisingly, the potentials on rings 1, 3, and 5 are all zero, resulting in an unusual feature in the electric field near the plate surfaces at these radii (observable in Figure 3(b)). It is not clear at this point why these values produced the best field among those examined, or whether there might be better sets of potentials possible. The permutation of possible values is large, and as yet no algorithm for complete optimization exists. Figures 3(c) and (d) show the axial electric field (Ez) and non-linear contribution to the axial electric field (∆Ez) along the z-axis by subtracting a linear extrapolation of a narrow region of the derived electric field near the center of the trap, respectively. Similar to the electric fields in the traps mentioned above, the potential distribution used in the planar Paul trap included a small positive compensation of higher-order components, and is expected to improve the mass resolution of this novel trap. In contrast to methods used with other ion traps, however, the higher-order components were not added into the planar Paul trap by modifying the shape or arrangement of the electrodes, but rather by choosing the appropriate potential function that was applied to the set of rings. Changing the electric fields within the planar Paul trap is done by changing the values of the capacitors on the PCBs. With the current plate spacing, only the first 11 rings had a noticeable effect on the electric fields in the trapping region. In order to save on cost, the plates were fabricated with additional rings, intended to be used in other experiments. In the present work, rings beyond ring 11 were shorted to ring 11. In conventional three--dimensional ion traps, ion behavior is understood and predicted by reference to stability parameters in the Mathieu equation. The commonly-given form of the qz stability parameter from the Mathieu equation is: