Mass analyzers for inductively coupled plasma time-of-flight mass spectrometry

Abstract

The use of a time-of-flight mass analyzer for inductively coupled plasma mass spectrometry is examined, with emphasis placed on the attributes of the different instrument geometries that have been employed. The common and contrasting design concepts, operating principles and experimental performance of both orthogonal-acceleration and axial time-of-flight inductively coupled plasma mass spectrometers are examined and contrasted.

---

Introduction

Inductively coupled plasma mass spectrometry (ICP-MS) has become an indispensable tool for the elemental analysis of a variety of sample types in a range of diverse applications. The significant attributes of ICP-MS include the following: high elemental sensitivity and low background noise that combine to yield low limits of detection for a good portion of the periodic table; the availability of isotopic information in a relatively simple spectrum; a utility with varied matrices, sample types and sample-introduction techniques; and a large and growing knowledge database of associated techniques. Despite these attributes, efforts aimed at addressing the remaining limitations and extending the strengths of ICP-MS continue by exploring alternative or modified sample introduction techniques, varied plasma architectures, and alternative mass analyzers, among which the time-of-flight mass spectrometer (TOFMS) is numbered.

Like other simultaneous mass analyzers, a TOFMS can collect complete mass spectra with the same precision, sensitivity, and resolution regardless of the number of m/z being investigated. This is in contrast to sequentially scanning ICP-MS instruments (such as the quadrupole or double-sector mass spectrometers) that are capable of observing only a single m/z at any moment. As a consequence, the measurement fidelity available over a given time with scanning instruments is degraded with the observation of additional m/z values. In some instances, measurement fidelity can be maintained simply by increasing the measurement time; however in other cases, such as when transient samples are measured, it is not recoverable.

The relatively narrow mass range required for elemental determinations permits ICP-TOFMS instruments to attain high repetition rates, often tens-of-thousands of complete mass spectra per second. In conjunction with the complete multielemental and isotopic coverage of such instruments, this allows the comprehensive investigation of single ionization events or transient signals with high temporal resolution. Further, as with all mass analyzers capable of the simultaneous extraction of all masses of interest, these determinations are accomplished without spectral skew.¹ Spectral skew errors can occur within sequential mass spectrometers as a consequence of the changing sample concentration during a spectral scan; the result is a misrepresentation of the relative abundance of different m/z due to the order in which they are observed. Because all m/z are extracted into a TOFMS at the same instant in time, this error is avoided.

Not surprisingly, the ICP-TOFMS has been coupled to a number of transient sample introduction sources, such as laser ablation (LA),^2–7 flow injection (FIA),^8–10 electrothermal vaporization (ETV),^11,12 and hydride generation (HG),^10,13 as well as capillary electrophoretic (CE)¹⁴ and gas chromatographic (GC)^15,16 separations. One advantage of such couplings stems from the ability to examine single transient signals, since the S/N available from each transient increases as its temporal duration is reduced. This is in contradistinction to strategies that employ the dilution of sequential transients as a means of generating a pseudo steady-state signal in order to accommodate the scanning of a mass spectrometer. Chromatographic separations benefit from a tolerance for non-ideal separations and the lack of spectral skew, which often permit such separations to be time-compressed. In some instances, the combination of such transient sample introduction systems with an analyzer capable of complete mass coverage and high temporal resolution yields unusual benefits. For example, Mahoney et al.¹¹ coupled an ETV sample introduction accessory to an ICP-TOFMS and reported the resolution of specific isobaric overlaps based upon differences in the vaporization characteristics of the interferents. Leach and Hieftje⁴ reported the use of LA with ICP-TOFMS to accomplish single-shot semi-quantitative solids analysis. In part, overcoming the limiting shot-to-shot imprecision is made possible by the ability of the mass analyzer to examine and integrate the signals from all constituent analytes during a single laser ablation pulse.

Simultaneous-sampling mass spectrometers such as a TOFMS are capable of improved measurement precision by employing ratioing schemes, such as internal standardization or isotope dilution, to correct for multiplicative noise sources. In contrast, scanning instruments that are unable to investigate all m/z values of interest at the same moment in time are much more susceptible to low-frequency noise and drift. Thus, the ratio precision obtainable from scanning ICP-MS instruments improves as the time between successive m/z observations is reduced; because this quantity is fundamentally restricted, the precision or mass range is similarly limited. This is not the case when one employs ICP-TOFMS. Several investigators have reported achieving high ratio precision with ICP-TOFMS instruments, often limited either by counting statistics in the case of ion counting detection,¹⁷ or by electronic limitations in the case of analog detection.^9,18–20

The lack of restrictive architecture within ICP-TOFMS instruments offers greater transmission efficiency than that obtained with most other mass spectrometers, and thus promises greater sensitivity. Additionally, in these instruments, every m/z within each extracted ion packet has the possibility of being detected. In concert, these attributes make ICP-TOFMS potentially capable of greater ion utilization efficiency than scanning ICP-MS instruments. In the present manuscript, several TOFMS instrument geometries are reviewed, the strengths and weaknesses of each are examined, and experimental results obtained in the authors' laboratory are offered in support. The interested reader should also consult a recent review by Guilhaus on the subject.²¹

TOFMS geometries for plasma source MS

Most ionization sources utilized for plasma source MS operate continuously, so the design of a TOFMS for use with continuous ion beams is of particular interest. The goals are to make the most efficient use of the ions created by the source, to realize adequate resolving power, and to retain the speed, mass range and simplicity for which TOFMS is known.^17,21 The most common TOFMS design in use is the orthogonal acceleration or right-angle geometry, a simple diagram of which is shown in Fig. 1.

Fig. 1 Schematic diagram of an orthogonal-acceleration TOFMS. R, repeller electrode; G₁, first acceleration electrode; a₁, first acceleration region (extraction region); a₂, second acceleration region; L, distance to primary space-focus plane; V_R, repeller (extraction) potential; d-, extraction region width; V_y, mean ion beam velocity; ΔV_y, beam velocity distribution; β, ion beam divergence along x-axis. Three open circles represent isomass ions with different initial positions in the direction of the flight tube.

In the right-angle design, the incoming primary ion beam continuously traverses an extraction region (a₁), defined between the repeller plate (R) and extraction grid (G₁) in Fig. 1. At an appropriate repetition frequency, an electrostatic field is created within this region by application of a voltage pulse (V_R), simultaneously injecting all of the ions present in a₁ into the acceleration region (a₂), where the final drift energy is gained. The ability to sample the incoming beam fairly efficiently is among the many recognized strengths of the right-angle design.

Because the energy of the ions in the primary beam (1–100 eV) is typically much lower than the acceleration potentials that are employed in the direction of the flight tube (1–10 keV), the primary ion beam slowly refills the extraction region during the time the ions extracted in the previous cycle are being mass analyzed. This results in a fairly high ion-utilization efficiency, quantitated in terms of the duty cycle of the instrument:

Here, V_y represents the mean velocity of the primary ion beam, d is the length of the extracted ion packets (equal to the width of the extraction zone in the implementation shown in Fig. 1), and f is the repetition frequency of the mass analyzer. When coupled with atmospheric-pressure ionization sources, duty cycles of 1–10% are typically achieved. It is noteworthy that the energy of the incoming ion beam greatly affects the efficiency with which particular ionization sources can be coupled to TOFMS, and that a sampling mass bias can arise if different masses possess greatly different velocities.

The velocity of a particular ion within the drift region (flight tube) is determined solely by its m/z and energy. However, the distribution of flight times of a population of isomass ions, and thus the resolution of a TOFMS, is greatly affected by the initial temporal, spatial and velocity distributions of the ions within the extraction region. Flight-time error resulting from the finite width of the ion packets along the axis of the flight tube is commonly minimized by incorporating the two-stage acceleration geometry and space-focusing principles introduced by Wiley and McLaren.²² By application of appropriate accelerating fields within the extraction and acceleration regions, a distance (L) along the drift length can be found at which the isomass ion packets will attain the smallest spatial distribution, effectively collapsing the width of the extracted ion packets. By placing a detector at this position, or by reflecting this plane onto a detector surface by use of an ion mirror,²³ the minimal spatially related flight-time error is realized. Space-focusing techniques can be augmented in right-angle instruments simply by limiting the width of the ion beam in this extraction region. Myers et al.²⁴ used a dc quadrupole lens stack to transform the original circular cross section of the incoming ion beam into a slit image, reducing the width of the sampled ion packets in the direction of the flight tube without unduly sacrificing beam flux.

Another strength of the right-angle geometry is that resolving power is relatively insensitive to the distribution of ion velocities in the primary ion beam. Ion packets are extracted in a direction orthogonal to their original motion, along an axis in which their velocity distribution is low. The use of an ion mirror and collimating ion optics enhances resolution. The perpendicular orientation of the flight axis and the incoming ion beam has the added advantage of limiting the access of neutral species and continuum ions to the flight tube, partially suppressing background noise levels.

Not surprisingly, the right-angle design also has its shortcomings. When ions are extracted into the acceleration region, they are subjected to an electrostatic field that is oriented perpendicular to the direction of their initial motion. Therefore, this original motion remains unaffected by the field and extracted ion packets attain some angle (α) from perpendicular in the flight tube. This angle depends upon the initial energy of the primary ion beam and is calculated as:

where V_y represents the velocity of an ion within the primary ion beam and V_x represents the velocity of the ion within the flight tube. When coupled to ionization sources that produce monoenergetic ion beams, all m/z within an extracted ion packet will attain the same angle in the drift tube. However, when such instruments sample an isokinetic ion beam in which all m/z move with the same velocity (and therefore possess different energies), the different m/z within the extracted ion packet necessarily follow different angles in the flight tube. Further, because there exists a distribution of energies of isomass ions in the primary ion beam (ΔV_y) there exists also a distribution of trajectories in the flight tube.

These effects are illustrated in Figs. 2A and B by use of an ion trajectory simulation program,²⁵ and are based upon an orthogonal acceleration ICP-TOFMS similar in dimensions and field distribution to that described by Myers et al.²⁶ Several investigators have shown that the energy of a particular m/z within an ion beam produced by an ICP to be (roughly) the sum of the energy provided by the supersonic expansion sampling process (isokinetic) and that caused by the plasma offset potential (isoenergetic).²⁷ Fig. 2A depicts ion trajectories for a group of elements based upon experimentally measured mean ion energies (shown in parentheses) for one such ICP-TOFMS. These mass-dependent energies span a range from 1–10 eV, and are similar to those reported by other investigators.²⁷ From an identical starting position, the atomic mass range from 0–238 u is spread across a distance of approximately 9.2 cm on the detector surface. Because the detectors used for TOFMS are often of limited size (microchannel plates (MCPs) two to four centimeters in diameter), all masses from a given extraction point cannot be recorded under a single set of operating conditions. In Fig. 2B, the angular divergence of an isomass ion packet is illustrated by two copper ions of identical m/z and initial starting position, but with an energy distribution (FWHM = 1.6 eV) measured experimentally. This distribution results in an angular divergence of 2 degrees and a spread of 3 cm at the detector surface. It is clear that the mass-dependent angular distribution can limit the multielemental performance of ICP-TOFMS instruments, and that the angular divergence of isomass ion packets can seriously limit transmission efficiency and thereby sensitivity.

Fig. 2 Ion trajectory simulations depicting mass-dependent trajectory and angular divergence in an orthogonal-acceleration TOFMS; A, mass-dependent trajectory of several isotopes of identical initial starting position; and B, angular divergence (α) of two copper ions of identical initial starting position but with differing energy, shown in parentheses; a₁, first acceleration (extraction) region; a₂, second acceleration region. See text for details.

Several strategies have been suggested to circumvent this potential difficulty. Often, steering plates are placed within the flight tube in order to direct the extracted ion packets onto the detector; however, there are several disadvantages to this approach. The sampling of atmospheric-pressure ionization sources creates an isokinetic ion beam, so ions of different m/z require a mass-dependent steering potential. Myers et al.²⁸ and others^29,30 have used a voltage ramp applied to the steering plates as a means of providing the correct steering potential for different masses. However, the linear ramp is only an approximation of the quadratic potential gradient that is truly required and does nothing to address the loss in transmission efficiency by angular divergence. Moreover, Guilhaus³¹ has shown that the use of such steering plates can lead to a loss in resolution due to temporal defocusing effects of the inhomogeneous electrostatic fields. The use of steering plates is often further complicated by the large cross-sectional area of the extracted ion beam required to realize an adequate duty cycle.

Alternatively, the steering plates can be omitted from the TOFMS and the ions allowed to follow their original angular trajectory, eliminating the deleterious effects of the steering plates and mass-dependent steering. Guilhaus and coworkers^21,32–36 have described such a "spontaneous drift" design and have successfully coupled the instruments to a number of ionization sources. The spontaneous drift geometry is particularly effective when employed with monoenergetic ion beams, since all masses attain the same angle within the flight tube and strike the detector surface at the same point. The use of such a design with isokinetic ion sources, however, can result in complications due to the resulting mass-dependent spatial offset on the detector surface, which may necessitate the use of a large extraction region or a detector of larger surface area and expense. This spatial offset does not, however, affect the flight time of the ions; as the author notes, “It matters when the ions strike the detector, not where”.³⁴

Recently, Bandura and Peile³⁷ and Davis et al.³⁸ reported a modification of the spontaneous drift geometry that greatly minimizes the aforementioned mass bias and angular divergence complications when isokinetic sources are coupled with orthogonal acceleration TOFMS. The strategy involves tailoring the width and spatial position of the extraction region to the energy distribution of the source of interest. To a first approximation, the distance (D) that a particular ion will travel along the y-axis (parallel to the detector plane in Fig. 3) can be calculated based on the ion mass and initial velocity oriented perpendicular to the flight tube (V_y) as:

where a is the slope of the mass calibration curve for a particular TOFMS (time-of-flight = ). In order for isomass ions of different initial velocity (V_y) to strike the detector plane at the same point, ions with greater initial velocities must be extracted into the flight tube at points farther removed from the detector surface along the y-axis. Conversely, ions of lower velocity must be extracted from points nearer the detector; or, more generally, ions must be spatially arranged along the extraction region linearly according to their incoming velocity. It also follows from eqn. (3) that, for an isokinetic ion beam (identical V_y for all m/z), ions must be arranged according to the square-root of their m/z within the extraction region in order to strike the detector plane at the same point.

Fig. 3 Orthogonal acceleration spontaneous drift TOFMS geometry with a large extraction region. A, sampling position of lowest ion energy; and B, sampling position of greatest ion energy; d, extraction region width; d_detector, detector width; a₁, first acceleration (extraction) region; a₂, second acceleration region; V_y, mean ion beam velocity; ΔV_y, beam velocity distribution.

These aims can be accomplished by increasing the size of the extracted ion packet well beyond the proportions of the detector surface. Effectively, this represents extracting each ion of a particular V_y or m/z from the appropriate position within the extraction region such that it will finally strike the detector surface. Although only a small portion of the total number of ions within the extraction region will finally strike the detector, those having the correct trajectory to do so should provide an accurate representation of the homogeneous ion beam located within the extraction region.

Viewed differently, the design of the extraction region in this manner represents a means of initial ion energy bandpass selection. Eqn. (3) can be rearranged to give:

where E_y is the energy in electron volts of the ion oriented perpendicular to the flight tube, N_A is Avogadro's number, and q is the fundamental charge. As eqn. (4) states, the offset distance (D) is independent of m/z and is dependent solely on the portion of the ion energy oriented perpendicular to the flight tube. This fact can be exploited in order to select a range of ion energies to be detected. From Fig. 3, it is evident that the minimum ion energy to be sampled is established by those ions which, when extracted from the end of the extraction region closest the detector along the y-axis (point A in Fig. 3), will strike the edge of the detector nearest the extraction region. Similarly, the greatest ion energy to be sampled is restricted by the energy of an ion beginning at the point within the extraction region farthest removed from the detector (point B in Fig. 3) that is still capable of striking the edge of the detector farthest from the extraction region. Ion energies that fall within this envelope will be extracted and strike the detector surface, regardless of their m/z.

For a known energy distribution such as produced by the ICP (or any other source), the extraction region length (d) can be tailored to include the ion population within this envelope. Fig. 3 depicts one possible TOFMS geometry, based loosely on a modification of that in Fig. 2, designed for use with an ICP source. The lowest ion energy to be sampled is limited by geometric concerns to ions extracted at point A in Fig. 3. They must have a minimum of 0.02 eV to strike the detector. Similarly, ions extracted from the far end of the extraction region must possess a maximum of 15 eV so they do not cover a distance longer than the 12.5 cm along the y-axis to strike the detector. Fig. 4 compares this ion energy envelope to the ion energies experimentally measured from an ICP source, where the error bars represent the ion energy distribution as measured at FWHM. It is evident from Fig. 4 that all ions should fall well within the ion energy limits of this hypothetical ICP-TOFMS, and thus should provide an accurate representation of the population of ions within the extraction region.

Fig. 4 Comparison of sampled ion energy envelope and the ion energy distribution produced by an ICP.

This approach has several significant advantages, most importantly that the angular divergence and mass-bias limitations noted previously are overcome, and without the use of a detector of large area and expense. Provided the ion kinetic energy distribution within the incoming primary ion beam is included within the sampled energy range, mass analyzer response will be independent of m/z across the mass range. Further, changes in these ion energies that might occur over the course of an experiment due to space charge, changes in plasma temperature, or sample introduction parameters will not affect instrument response, as long as these energies do not fall outside the sampled ion energy envelope.

When the duty cycle of such a geometry is calculated, the size of the detector (d_detector in Fig. 3) should be substituted for the size of the extraction region (d) in eqn. (1). This substitution is appropriate because only a small fraction of those ions that are extracted into the flight region will finally be detected. At any given point within the extraction region, ions of many different m/z and energies will be present. Only a portion of those ions at any particular point will possess the correct energies to finally strike the detector; all others will be lost. While this might seem to be wasteful, it must be recalled that other ICP-TOFMS instruments often extract less than 10% of the ions presented to the mass analyzer. This ICP-TOFMS utilizes ions that would otherwise be wasted anyway. For this reason, the sensitivity of such an instrument will be comparable to the other geometries described here and will exhibit a similar mass bias due to differences in the velocities (V_y) of different m/z according to eqn. (1). Further, because many more ions are extracted into the flight tube than will finally strike the detector surface, background noise due to scattering or charge-exchange reactions might be greater than in other instruments.

Because the extraction-region width (d) in this geometry is comparable to the distance traveled by the primary ion beam during the time required for mass analysis, consideration must be given to the maximum repetition frequency. The maximum repetition rate depends on both the flight time of the greatest m/z and the time required for the primary ion beam to refill the extraction region, or more accurately, on the time required for the ion of greatest mass and lowest energy to traverse the extraction region. In the simplest case of a completely isokinetic ion beam, the maximum rate is simply the ratio of the ion velocity of the incoming beam and the extraction region width (V_y/d). In the instrument pictured in Fig. 3, the maximum repetition rate for a beam moving at 2300 m s⁻¹ (5000 K ICP gas temperature) would be 25.5 kHz. It is equivalent to consider the maximum mass range, which is a consequence of the constant velocity of all masses in this example (energy proportional to m/z) and the maximum sampled ion energy set by the TOFMS conditions. For the instrument pictured in Fig. 3, the mass range would extend from 1–560 u under the same isokinetic conditions.

The extraction of such a large section of the ion beam might, in this geometry, adversely affect internal-standardization precision. In order to compensate for multiplicative noise, the various m/z that are ratioed must be observed at the same instant in time. In ICP-TOFMS, that requires extraction of all ions from the incoming beam at the same time, or, for an isokinetic ion beam, extracting all m/z from the same section of the incoming ion beam. Because this geometry relies upon extracting different ion energies (and therefore m/z) from different spatial positions within the extraction region, the various masses arise from different time segments and internal standardization precision might therefore suffer.

On-axis geometry

An alternative to the right-angle geometry is to place the incoming ion beam on the same axis as the flight tube in an “in-line” or “on-axis” geometry. Historically, such in-line designs have been based upon a beam-sweeping method in which a continuous ion beam is first accelerated, then allowed to pass between a set of steering plates.^39–45 Application of a voltage pulse to the plates deflects the ion beam laterally across the flight tube, and ultimately across a slit located at the end of the drift region. Those ions with the correct trajectory to pass through the slit are thus extracted from the continuous ion beam, and strike a detector located behind the slit. This design, first described by Bakker,^39,40 has found limited use because of the low duty cycles it offers, typically a factor of 1000 less than the right-angle approach.

Li and Hieftje⁴⁶ recently described a novel in-line TOFMS designed for use in ICP-MS. A diagram of this system, given in Fig. 5, differs from the beam-sweeping arrangement in that the ion packet is selected before acceleration. Consequently, it can incorporate the two-stage acceleration geometry and space-focusing principles of Wiley and McLaren,²² providing a much higher duty cycle and resolution, but with an in-line geometry. Conceptually, then, this instrument is more closely related to the right-angle geometry than to an in-line beam-sweeping design.

Fig. 5 Axial ICP-TOFMS instrument diagram.

In this in-line configuration, a continuous incoming ion beam is focused and transported by a set of extraction optics into a beam-modulation region. In this modulation region the continuous beam is parsed into discrete ion packets, which are then allowed to travel by means of their original motion into the extraction region. At a time delayed from the beam modulation step and at the moment when the ion packet fills the extraction region, a high-voltage pulse is applied to a repeller grid, driving the ions within the extraction region into the acceleration region, and from there into the flight tube for mass analysis.

With the right-angle configuration, the processes of beam modulation and extraction occur simultaneously. A section of the ion beam is removed from the continuum, injected into the mass analyzer, and given the spatially dependent energy required for space focusing all at once. In contrast, in the in-line geometry, the primary beam and flight axis are co-linear; a separate mechanism must be employed to separate the ions that have been selected for mass analysis from the remainder of the beam. This separation has been accomplished by two alternative means.

First, the in-line geometry can operate as just described, parsing the continuous ion beam into packets before the extraction event. Fig. 6 depicts the three steps of one such arrangement by diagram, field distribution and timing. In a three-grid geometry, two gridded ring electrodes are placed immediately prior to the repeller grid. During mass analysis, the continuum ion beam is prevented from entering the extraction region (defined between the repeller grid (R) and the first acceleration electrode (G₁)) by application of a small positive potential to the modulation grid (M). The magnitude of the potential (V_M) is chosen to be greater than that possessed by the bulk ion beam. At a time prior to the extraction event (t₀), the potential applied to the modulation grid is pulsed to ground, allowing the continuous beam to pass the modulation region and enter the extraction region. The modulation potential remains at ground for a period of time required to create a packet capable of filling the extraction region and, upon reapplication of the modulation bias at time t_M, an ion packet is formed. This packet travels by means of its original velocity into the extraction region where it is extracted for mass analysis by application of the repeller pulse (V_R). Optimum voltages and time delays experimentally determined for an in-line ICP-TOFMS are included in parentheses in Fig. 6.

Fig. 6 Three-step sequence of pre-extraction beam modulation: Ext, extraction optics; G_A, shield grid; M, modulation grid; R-repeller grid, G₁-first acceleration grid. Frame A depicts potential fields within each region during each step of the sequence. Frame B depicts the pulse timing sequence: V_M, modulation potential; V_R, extraction (repeller) potential; t₀, initiation of pulse sequence; t_M, end of modulation pulse; t_R, initiation of repeller pulse; t_R′, end of repeller pulse. Quantities in parentheses denote typical values utilized within an in-line ICP-TOFMS.

Pre-extraction ion beam modulation has the advantage of limiting access to the drift tube to only those ions that will finally result in a signal, thereby yielding significantly lower background levels. However, the success of the method depends heavily on the energy range of the incoming ion beam. Accurate and reproducible sampling of the bulk ion beam requires that the relative concentrations of different-mass ions within the extraction region reflect those in the original ion beam. This, in turn, requires that all masses within the beam move with the same velocity, so the fidelity of the ion packet is maintained during transport into the extraction region.

Unfortunately, the incoming ions do not travel with exactly the same velocity. The effect of the modulation pulse width (t₀–t_M) on the signal from several different masses for an in-line ICP-TOFMS is shown in Fig. 7. As the modulation pulse width is increased (here t_M = t_R), the signal rises progressively for increasing m/z values. Lithium (m/z = 7) reaches a limiting value at a pulse width of 750 ns, whereas bismuth (m/z = 209) requires 3.3 µs to do the same. The mass dependence of the pulse width is a result of the non-isokinetic nature of the sampled ion beam, and is consistent with an approximately 2 eV ICP offset potential. Here, lithium possesses greater than twice the velocity of bismuth and therefore travels the 3 mm distance from the modulation region to the extraction region in less than half the time. The sloping response of the signal from each element is a result of the same differences in velocity and velocity distribution of each m/z, as the extraction region is being filled more rapidly by the faster-moving, lighter ions. This behavior illustrates that both the in-line and orthogonal-acceleration TOFMS instruments respond to the concentration of each particular m/z in the section of the ion beam within the extraction region. As quantified by eqn. (1), the duty cycles of different m/z will differ if V_y is mass-dependent. This is in contrast to scanned mass analyzers wherein response is related to the observation time at a particular m/z and is therefore, to a first approximation, independent of small differences in velocity.

Fig. 7 Dependence of signal on modulation delay (t₀–t_m) for the in-line geometry shown in Fig 6. (●) ⁷Li, (○) ²⁷Al, (✦) ⁶³Cu, (◊) ⁸⁸Sr, (■) ¹¹⁵In, (□) ¹⁶⁵Ho, (▲) ²⁰⁹Bi.

Experimentally, such problems can be overcome by reducing the physical distance between the modulation and extraction regions, and by overfilling the extraction region (i.e., by operating at a modulation pulse width long enough to realize maximum signal levels for all m/z of interest (4 µs in Fig. 7). Doing so requires that a significant fraction of lighter-mass ions be allowed to pass through the extraction region, and thus necessitates the use of noise attenuation grids, as described by Mahoney et al.⁴⁷ Such energy discrimination grids are capable of separating ions based upon differences in their energy and are described in more detail in the following section. Other disadvantages of this form of pre-extraction modulation include the possibility of spectral artifacts caused by ions in the immediate vicinity of modulation grids at the time of a field change, and the loss in transmission efficiency that accompanies the use of gridded electrodes.

Pre-extraction modulation of the incoming ion beam can be omitted in favor of a second approach that relies upon separating continuum ions from selected ions on the basis of energy. Figs. 8A and B depict such an energy discrimination (ED) strategy. In this scheme, a set of three gridded electrodes is placed immediately ahead of the detector surface, the outer two of which are held at flight-tube potential (V_FT) while the center ED electrode is held at a potential greater than the maximum energy of ions within the incoming beam (V_ED). During mass analysis (repeller OFF condition in Fig. 8A), the continuous incoming ion beam is allowed to traverse the extraction region of the instrument and follow the same path as intentionally extracted ions: through the acceleration region and into the flight tube. Ions entering the flight tube in this manner attain the energy dictated by the acceleration potential, so upon reaching the ED grids they do not have enough energy to pass the potential barrier (V_ion < V_ED), and are reflected back into the drift region. In contrast, those ions that are within the extraction region at the moment of an extraction event (Fig. 8B) gain an additional 0–500 eV of energy due to the linear extraction field and therefore gain enough energy to traverse the ED barrier (V_ion > V_ED), strike the detector, and be detected. Instrument dimensions and typical operating parameters for a typical in-line ICP-TOFMS are included in parentheses in Fig. 8. In this diagram the ion reflectron has been omitted for simplicity.

Fig. 8 Energy-discrimination system for beam modulation and background reduction: E. O., extraction optics; Ext, extraction region; Accel, acceleration region; Drift, field free drift region (flight tube); ED, energy discrimination region; DET, detector distance; V_DET, detector bias (MCP); V_FT, acceleration bias; V_R, repeller potential. Quantities in parentheses denote typical values utilized in an in-line ICP-TOFMS. A, illustrates fields and ion behavior during periods when no extraction event is underway (Repeller OFF condition); and B, illustrates the same during an extraction event (Repeller ON condition).

Fig. 9 reveals how signal and background levels depend upon the energy-discrimination potential. The background signal was determined approximately 1 u (80 ns) removed from the peak of interest where no other peak was known to exist, and thus represents a continuum level. These background levels drop quickly upon application of less than +50 V to the ED grid and remain unaffected by greater potentials, betraying a monoenergetic composition. The signal produced by a 1 ppb solution of uranium (monitored at m/z = 238) undergoes the same initial steep decline upon application of the ED potential. However, increasing ED potentials over the range +50 V to +500 V results in only a gradual decrease in signal levels, until the signal finally reaches the background level. The linear signal reduction over this range is a consequence of the linear acceleration field within the extraction region and indicates that the region is completely filled and that all ions within it are transported to the detector. It is also noteworthy that the limiting continuum background levels are not at zero, thereby compromising the utility of the technique.

Fig. 9 Effect of energy discrimination potential on background and signal. (○) Continuum background count rates (50 ns window) and (●) ²³⁸U signal (1 ng mL⁻¹).

This approach is essentially the same as that utilized by Mahoney et al.⁴⁷ for noise attenuation in a orthogonal-acceleration ICP-TOFMS; however, here it is utilized as a means of beam modulation. As a means of modulation, the energy-discrimination technique has the advantage of being relatively unaffected by the initial energy of the incoming ion beam, overcoming the major limitation of pre-extraction modulation. The major disadvantage of the approach lies in the elevated background noise levels cited above.

Mahoney et al.⁴⁷ found that the limiting background in a right-angle ICP-TOFMS was due to neutral species which, they hypothesized, resulted from charge-exchange collisions between accelerated ions and neutral background gas species. In the present system, extracted ion packets traversing the last leg of the drift length between the reflectron and detector presumably undergo charge-exchange collisions with background neutral species, forming fast-moving neutral atoms and ions at nearly thermal translational energies. The resultant neutral species retain most of their original mass-dependent velocity and trajectory, striking the detector at nearly the same moment as their charged counterparts. However, ions also enter the drift tube in the absence of an extraction pulse and undergo similar neutralization; they appear as a continuum background noise spread across the entire mass spectrum. The original authors⁴⁷ demonstrated a direct dependence between pressure within the flight tube of the TOFMS and the ratio of charged-to-neutral species, as would be expected from simple two-body collision theory. Identical results were obtained when the same experiments were performed using an in-line ICP-TOFMS. While the fundamental origin of this noise is the same in both instruments, it is of much greater concern with the in-line geometry because of the large ion currents that continually enter the drift region. In either instrument this source of noise is minimized by reducing the pressure within the third vacuum region, or by allowing only those ions introduced into the flight region by an extraction pulse to traverse the flight path.

Comparison of orthogonal-extraction and on-axis TOFMS geometries

The simple shift in orientation from perpendicular to coaxial alignment of the flight axis and the incoming continuous ion beam creates significant differences in design and performance features, several of which are summarized in Table 1. First, the coaxial orientation of flight axis and incoming ion beam effectively eliminates the difficulties with angular divergence and mass-dependent steering mentioned in the previous section. Although the major axis of velocity distribution within the primary ion beam lies along the direction of extraction, it constitutes only a small portion of the overall ion energy, typically 1–10 eV of 2000–2500 eV. Consequently, these mass-dependent energies have a negligible effect upon the steering of different masses, and steering can be accomplished independently of m/z.

Table 1 Comparison of orthogonal-acceleration and on-axis ICP-TOFMS geometries^a

	Orthogonal-acceleration TOFMS	On-axis TOFMS
a Figures of merit reflect instruments constructed in the authors' laboratory, described in ref. 24, 26, 28 and 49, and herein. b Determined at peak FWHM. c Based upon 10 s integration time.
Mass-dependent spatial offset at detector plane	Large, and dependent upon initial beam energy	Small
Angular divergence	Large (>20%)	Small
Initial spatial distribution in flight-tube direction	Small; can be minimized by extraction optics	Large
Initial energy distribution in flight-tube direction	Small	Large (1–10 eV); dependent upon initial beam energy distribution
Ion beam modulation/ion packet extraction	Simultaneous, single step	Sequential, 2-step
Ion beam cross section	Large; duty-cycle dependent	Small; duty-cycle independent
Resolving power^b	2000	1400
Dynamic range	10^6	10^6
Detection limit^c	0.3–10 ppt	0.9–10 ppt
Isotope ratio precision	0.056% RSD	0.02% RSD
Repetition rate	17 000 Spectra s^−1	25 000 spectra s^−1
Abundance sensitivity	>10^6	>10^5
Duty cycle	≈10%	≈6%

---

Such an effect is illustrated in Fig. 10, where simulations identical to those in Fig. 2 are performed but with an in-line TOFMS geometry. The in-line ICP-TOFMS described by Li and Hieftje⁴⁶ utilizes steering plates and also the angle between the reflectron and the flight axis (θ) as means of directing ions at a spatially offset detector. In Fig. 10, only the reflectron angle is utilized and ions of identical energies and mass to those utilized in Fig. 2 are arrayed across the entire extraction region, a distance of 1 cm. When acceleration and extraction potentials similar to those in Fig. 2 are utilized, the ions attain a distribution of acceleration energies dictated by their initial positions. In addition to having different starting positions, these ions also possess different m/z values and therefore different flight times. They have therefore been arrayed spatially in this simulation so as to produce the greatest possible beam divergence caused by the initial energy offset. For example, the heaviest mass of common interest in ICP-TOFMS (uranium m/z = 238) is positioned closest to the repeller electrode, so it will possess the greatest possible energy upon acceleration and will attain the greatest offset distance along the y axis. Conversely, the lighter masses that possess less total energy will travel shorter distances. This simulation, therefore, reflects a “worst case” scenario. Even so, an ion beam with an initial diameter of 1 cm is observed to increase in width to only approximately 1.5 cm, a spread much less than the diameter of the detector. Thus, a single set of operating conditions can be utilized under which all masses of interest can be detected simultaneously, overcoming the complications of mass-dependent steering encountered with orthogonal acceleration.

Fig. 10 Ion-trajectory simulations for an in-line ICP-TOFMS depicting mass-dependent ion trajectory of several isotopes of differing initial starting position: a₁, first acceleration region (extraction region); a₂, second acceleration region. See text for details.

The in-line TOFMS geometry also generates a well-defined, nearly collimated ion beam of small cross section. Because the distribution of ion energies perpendicular to the flight tube of the in-line TOFMS is small and isotropic, and because the narrow primary beam allows effective use of focusing along the drift length, angular divergence can be minimized and transmission efficiency increased. The smaller cross section of the ion beam also offers two additional advantages: use of smaller detectors and higher efficiency of the ion optics. The duty cycle of a TOFMS is dictated, in part, by the length of the extraction region and by the resulting physical size of the extracted ion packets. A right-angle design necessitates that the extracted packets be of increasingly larger area to raise duty cycle, requiring in turn larger (and more expensive) detectors and larger ion optics. In contrast, the duty cycle (ion packet width) of the in-line geometry is independent of the cross section of the beam width in the flight tube. Consequently, large duty factors are possible without requiring large ion optics or detectors. As a result, both secondary-electron multipliers (SEM) and microchannel plate (MCP) detectors can be utilized.

Unfortunately, the in-line TOFMS geometry also sacrifices several of the advantages of the orthogonal acceleration approach. Most important is the orientation of the major axis of velocity distribution. With the in-line geometry, the axis of greatest velocity dispersion is oriented in the direction of the flight tube. Thus, it can contribute directly to a spread in the observed flight times of a particular m/z, compromising resolving power. Fortunately, this complication can be largely avoided by use of a properly designed ion mirror.

The in-line orientation presents another challenge in attaining adequate resolving power, related to the initial spatial distribution of ions in the extraction region. As was mentioned previously, an orthogonal-acceleration TOFMS can make use of appropriate ion-optical elements in order to limit the initial spatial distribution of ions in the direction of the flight tube. Such strategies do not compromise duty cycle, which is directly proportional to the length of the extracted packet. If a large extraction region is completely filled with ions in order to obtain the greatest possible duty factor in the on-axis geometry, much greater demands are placed on space focusing. This realization has led to the development of improved space focusing strategies.⁴⁶

There are several consequences of these two factors. First, operation of an in-line TOFMS in a linear mode (i.e., without a reflectron) yields poor performance. Because the ionization source and detector in a linear instrument lie on the same axis, neutrals and photons have a direct path to the detector, yielding high noise. Additionally, the relatively large axial energy distribution and initial spatial spread can seriously compromise resolution. Realistically, then, an in-line TOFMS will operate exclusively with a reflectron. The ion mirror acts not only as a means of separating neutrals from ions, but also to compensate for the energy dispersion of the incoming ion beam.

The need for an additional step of beam modulation is another comparative disadvantage of the in-line geometry. Although the orthogonal-acceleration geometry can directly select a packet of the incoming ion beam for mass analysis, the in-line geometry must utilize a sequential two-step process. This extra complication might lead to a sampling mass bias, more difficult optimization, or spectral artifacts, although they have not been serious in our experience.

In summary, then, both the in-line and orthogonal-acceleration TOFMS geometries possess significant, particular strengths. The in-line TOFMS offers advantages in mass-independent steering, low angular divergence, small ion-beam dimensions and high transmission efficiency. The orthogonal-acceleration geometry holds potential advantages in resolution and simplicity of beam modulation. In practice, however, both geometries can yield similar performance and possess the attributes for which TOFMS is known.

Experimental performance

The first coupling of an ICP to a TOFMS was reported by Myers and Hieftje,⁴⁸ and a detailed description of their right-angle ICP-TOFMS was the subject of a subsequent three-paper series.^24,26,28 Because this material has been recently reviewed,¹⁷ no attempt will be made to provide a comprehensive treatment here.

As the previous discussion indicated, the greatest potential weakness of the in-line geometry compared to the right-angle design is in attaining adequate resolution. Of course, high resolution is not required for the separation of simple atomic species; however, it is valuable in resolving isobaric overlaps and in providing adequate abundance sensitivity. An example of a complete atomic mass spectrum produced by an in-line ICP-TOFMS instrument and obtained by the ultrasonic nebulization (USN) of a multielemental standard solution is shown in Fig. 11A. An expanded section of the same spectrum, in Fig. 11B, shows good separation of the isotopes of lead and thallium, each present at 0.5 ng mL⁻¹. The ²⁰⁸Pb isotope exhibits a temporal width of 12 ns (FWHM), corresponding to a resolving power of 1380. Although this resolution is lower than that typically demonstrated with the right-angle ICP-TOFMS instrument of Myers et al.⁴⁹ (R_FWHM = 1800–2000), it is adequate for routine elemental analysis and is significant for two other reasons. First, achieving such a resolution indicates there is adequate compensation for the initial ion velocity distribution of the entering ion beam in the direction of the flight tube. This distribution was measured to be 1–8 eV (FWHM) in this instrument, a range similar to that reported by other investigators.²⁷ Second, this resolving power is noteworthy because this in-line ICP-TOFMS employs a drift length that is only about half that of the aforementioned right-angle instrument,²⁶ and, under similar accelerating conditions, is capable of a greater repetition frequency and correspondingly higher duty factor.

Fig. 11 A, Multielemental ICP-TOFMS spectrum obtained from an on-axis ICP-TOFMS instrument. Solution is 0.5 ng mL⁻¹ in labeled elments. B, Expanded section of multielemental ICP-TOFMS spectrum (Fig. 9A) showing resolution of Pb and Tl isotopes. Inset shows background noise in a 50 ns window.

Figs. 11A and B also illustrate the typical sensitivity of the in-line instrument. Sensitivities, background levels and limits of detection determined simultaneously from a multielemental standard solution presented to the in-line ICP-TOFMS constructed within the authors' laboratory are listed in Table 2. A 10 s integration period was used. Given similar sample-introduction schemes and interface construction, the sensitivity of a TOFMS should be a function of duty cycle and transmission efficiency. The duty cycle of the right angle ICP-TOFMS²⁶ exceeds that of the in-line ICP-TOFMS by 1.7 times; however, the sensitivities reported here for the in-line ICP-TOFMS are 1.2–1.6 times greater than those reported for the right-angle ICP-TOFMS by Mahoney et al.⁴⁷ Since both instruments utilized similar sample-introduction and interface systems, this disparity suggests higher transmission efficiency for the in-line ICP-TOFMS, probably a result of the narrower ion beam. Myers et al.²⁶ calculated the transmission efficiency of the right-angle ICP-TOFMS to be no greater than 20% based upon angular beam divergence and geometric considerations, and measured a transmission efficiency of 18%. Although the transmission efficiency of our in-line ICP-TOFMS has not yet been determined, it would appear to be between 2 and 3 times greater, or between 36% and 54%.

Table 2 Limits of detection, sensitivity, and background measurements for an in-line ICP-TOFMS

m/z and element	Mean signal/counts s^−1^a	Mean background/counts s^−1	STD background/counts s^−1	Limit of detection/ppt^b	Sensitivity/counts s^−1 per ppb^c
a Background-subtracted average of 10 replicate peak-area measurements of a 0.1 ppb standard solution; 10 s integration time. b Determined at 3σ; 10 s integration time. c Based upon least squares fit of a standard curve of 5 standard solutions. In all cases r^2 >0.98.
^7Li	923	85.3	15.8	5.2	9148.1
^24Mg	1074	122.2	10.4	2.6	12913.2
^60Ni	568	86.7	13.4	2.2	17302.1
^115In	2472	45.5	9.1	1.1	24601.5
^105Ag	1133	39.0	9.6	1.4	21409.8
^138Ba	2156	95.4	12.1	1.3	28250.6
^208Pb	1673	64.1	13.4	1.3	31094.3
^209Bi	3135	44.9	10.1	0.98	32621.0

---

Detection limits depend upon both signal strength and background noise level; although the sensitivity of the in-line ICP-TOFMS is typically 10⁷ counts s⁻¹ per ppm, detection limits are compromised by elevated background noise. As Table 2 indicates, continuum noise levels can reach 50–100 counts s⁻¹ within a 50 ns section of the mass spectrum, and will thus contribute 7–10 counts s⁻¹ of background noise to any particular m/z peak. Because of this heightened noise, the detection limits reported here for the in-line ICP-TOFMS developed within the authors' laboratory are similar to those reported by Mahoney et al.⁴⁷ for a right-angle ICP-TOFMS, despite greater sensitivity. For both ICP-TOFMS designs, detection limits are predominantly background-noise limited and are roughly one order of magnitude greater (worse) than those reported for commercial quadrupole systems, when a single mass is monitored under similar conditions. This discrepancy has also been noted by Tian et al.⁵⁰ with a similar commercial in-line ICP-TOFMS, as well as by Sturgeon et al.⁹ with a commercial orthogonal-acceleration geometry ICP-TOFMS. This discrepancy is a result, in part, of the difference in duty cycles between the current ICP-TOFMS and quadrupole instruments when a single m/z is measured. However, it must be emphasized that these limits of detection can be achieved for all masses simultaneously with an ICP-TOFMS.

It was suggested earlier that an in-line TOFMS geometry should not exhibit strongly mass-dependent steering potentials. The right-angle ICP-TOFMS described by Myers et al.²⁶ utilizes steering plates in order to compensate for the portion of the ion velocity that is oriented perpendicular to the flight-tube axis. This in-line ICP-TOFMS, however, utilizes a combination of reflectron angle and steering plates to give the ions the correct trajectory to strike the detector, which is spatially offset from the primary drift axis (see Fig. 3). As this angle is small (approximately 1.5°), a weak electrostatic field is sufficient and the deleterious effects such plates have upon resolution becomes negligible. As Fig. 12 illustrates, elements across the atomic mass range exhibit approximately the same steering-potential dependence, so a compromise voltage can easily be found that enables all masses to be observed simultaneously.

Fig. 12 Effect of steering plate potential on signal level of several different masses for an in-line ICP-TOFMS. (▼) ⁷Li, (■) ²⁷Al, (⊞) ⁶³Cu, (△) ⁸⁸Sr, (●) ¹³³Cs, (+) ¹⁶⁵Ho, (○) ²⁰⁹Bi.

Conclusions

The adaptation of TOFMS for use in ICP-MS has progressed to the point where such instruments provide analytical performance approaching that of their scanning counterparts; thus they have been recently offered as commercial instruments (GBC Scientific Equipment, Dandenong, Victoria, Australia and LECO Corporation, St. Joseph, MI, USA). It is clear that TOFMS has demonstrated a number of valuable advantages and capabilities for elemental analysis. In particular, the ICP-TOFMS has a talent for transient signal analysis (often extending the capabilities of established elemental methods), has achieved improved measurement fidelity through the use of ratioing techniques, and offers the complete simultaneous multielemental mass coverage required with a potentially multielemental technique such as ICP-MS. It is also clear from the preceding discussion that those TOFMS geometries developed for continuous ionization sources (and the ICP in particular), although possessing individual and specific strengths and weaknesses, retain these advantages.

Although approaching competitive performance, ICP-TOFMS instruments still lag behind their scanning counterparts in several areas, all of which are the focus of current research activity. The sensitivity of ICP-TOFMS instruments is as much as an order of magnitude poorer than comparable quadrupole systems, and the detection limits correspondingly worse by a similar margin. This is in spite of the fact that a theoretical comparison concludes that the instruments should offer similar performance; while a scanning system such as a quadrupole holds an advantage in duty cycle at a single mass (100% versus about 10% in ICP-TOFMS instruments), this advantage should be offset by higher transmission efficiency of TOFMS instruments.⁴⁸ For this reason, the improvement of duty cycle and transmission efficiency of ICP-TOFMS instruments is ongoing. Further, observation of the entire atomic mass range with each extraction event places great demands on the dynamic range of ICP-TOFMS instruments. In instances in which peaks occur with greater than a 10⁶ disparity in abundance within a single spectrum, detector saturation can influence the remainder of the spectrum. Currently, there exists no effective means of limiting the intensity of any single m/z within the mass spectrum while leaving other m/z unaffected. By default, an instrument capable of observing a single m/z by the exclusion of all others does not possess this limitation.

Acknowledgements

This research was supported in part by the US Department of Energy through grant DE-FG02-98ER14890 and by ICI Technologies. The authors wish to acknowledge the continuing support of the Mechanical Instrument Services and Electronic Instrument Services at Indiana University, as well as CETAC Technologies Inc. for the loan of sample introduction equipment and SPEX CentriPrep Inc. for the donation of standard solutions.

References

1. J. F. Holland, C. G. Enke, J. Allison, J. T. Stults, J. D. Pinkston, B. Newcome and J. T. Watson, Anal. Chem., 1983, 55, 3.
2. P. P. Mahoney, G. Li and G. M. Hieftje, J. Anal. At. Spectrom., 1996, 11, 401.
3. L. Allen, S. Georgitis, D. P. Myers and K. Brushwyler, Phys. Stat. Sol. (A), 1998, 167, 357.
4. A. M. Leach and G. M. Hieftje, J. Anal. At. Spectrom., 2000, 15, 1121.
5. D. Bleiner, K. Hametner and D. Günther, Fresenius' J. Anal. Chem., 2000, 368, 37.
6. D. Bleiner, A. Plotnikov, C. Vogt, K. Wetzig and D. Günther, Fresenius' J. Anal. Chem., 2000, 368, 221.
7. B. Hattendorf, I. Horn and D. Günther, Fresenius' J. Anal. Chem., 2000, 368, 4.
8. K. Benkhedda, H. Goenaga Infante, E. Ivanova and F. C. Adams, J. Anal. At. Spectrom., 2000, 15, 1349.
9. R. E. Sturgeon, J. W. H. Lam and A. Saint, J. Anal. At. Spectrom., 2000, 15, 607.
10. G. Centineo, M. M. Bayon and A. Sanz-Medel, J. Anal. At. Spectrom., 2000, 15, 1357.
11. P. P. Mahoney, S. J. Ray, G. Li and G. M. Hieftje, Anal. Chem., 1999, 71, 1378.
12. M. Guilhaus, D. Selby and V. Mlynski, Mass Spectrom. Rev., 2000, 19, 65.
13. W. C. Wetzel, J. A. C. Broekaert and G. M. Hieftje, J. Anal. At. Spectrom., 2001, 16, 987.
14. J. M. Costa-Fernandez, N. H. Bings, A. M. Leach and G. M. Hieftje, J. Anal. At. Spectrom., 2000, 15, 1063.
15. A. M. Leach, M. Heisterkamp, F. C. Adams and G. M. Hieftje, J. Anal. At. Spectrom., 2000, 15, 151.
16. N. H. Bings, J. M. Costa-Fernandez, J. P. Guzowski Jr., A. M. Leach and G. M. Hieftje, Spectrochim. Acta, Part B, 2000, 55B, 767–778.
17. P. P. Mahoney, S. J. Ray and G. M. Hieftje, Appl. Spectosc., 1997, 51, A16.
18. F. Vanhaecke, L. Moens, R. Dams, L. Allen and S. Georgitis, Anal. Chem., 1999, 71, 3297.
19. X. Tian, H. Emteborg, M. Barbaste and F. C. Adams, J. Anal. At. Spectrom., 2000, 15, 829.
20. H. Emteborg, X. Tian, M. Ostermann, M. Berglund and F. C. Adams, J. Anal. At. Spectrom., 2000, 15, 239.
21. M. Guilhaus, Spectrochim. Acta, Part B, 2000, 55B, 1511.
22. W. C. Wiley and I. H. McLaren, Rev. Sci. Instrum., 1955, 26, 1150.
23. B. A. Mamyrin, D. V. Karataev, D. V. Schmikk and V. A. Zagulin, Sov. Phys.-JETP, 1973, 37, 45.
24. D. P. Myers, G. Li, P. P. Mahoney and G. M. Hieftje, J. Am. Soc. Mass Spectrom., 1995, 6, 400.
25. D. Dahl, Simion 3D, version 6.0, Scientific Instrument Services Inc..
26. D. P. Myers, G. Li, P. Yang and G. M. Hieftje, J. Am. Soc. Mass Spectrom., 1994, 5, 1008.
27. S. D. Tanner, J. Anal. At. Spectrom., 1993, 8, 891.
28. D. P. Myers, G. Li, P. P. Mahoney and G. M. Hieftje, J. Am. Soc. Mass Spectrom., 1995, 6, 411.
29. M. Pellarin, B. Baguenard, M. Broyer, J. Lerme and J. L. Vialle, J. Chem. Phys., 1993, 98, 94.
30. P. Milani and W. A. de Heer, Rev. Sci. Instrum., 1991, 62, 670.
31. M. Guilhaus, J. Am. Soc. Mass Spectrom., 1994, 5, 588.
32. M. Guilhaus, J. Mass Spectrom., 1995, 30, 1519.
33. M. Guilhaus, V. Mlynski and D. Selby, Rapid Commn. Mass Spectrom., 1997, 951.
34. J. Coles and M. Guilhaus, TrAC, Trends Anal. Chem. (Pers. Ed.), 1993, 12, 203.
35. V. Mlynski and M. Guilhaus, Rapid Commun. Mass Spectrom., 1996, 10, 1524.
36. J. H. J. Dawson and M. Guilhaus, Rapid Commun. Mass Spectrom., 1989, 3, 155.
37. D. Bandura and R. Peile, XXX Colloquium Spectroscopicum Internationale, Paper 227, Melbourne, Australia, 1997..
38. S. Davis, D. Bandura, A. D. Hoffman and A. A. Makarov, WIPO Patent WO9840907A1, 1998..
39. J. M. B. Bakker, J. Phys. E.: Sci. Instrum., 1974, 7, 364.
40. J. M. B. Bakker, J. Phys. E.: Sci. Instrum., 1973, 6, 785.
41. G. E. Yefchak, G. A. Schultz, J. Allison, C. G. Enke and J. F. Holland, J. Am. Soc. Mass Spectrom., 1990, 1, 440.
42. C. Ma, S. M. Michael, C. Mingta, J. Zhu and D. M. Lubman, Rev. Sci. Instrum., 1992, 63, 139.
43. J. D. Pinkston, M. Rabb, J. T. Watson and J. Allison, Rev. Sci. Instrum., 1986, 57, 583.
44. B. A. Eckenrode, J. T. Watson, C. G. Enke and J. F. Holland, Int. J. Mass Spectrom. Ion Proc., 1988, 83, 177.
45. G. L. Glish and D. E. Goeringer, Anal. Chem., 1984, 1984, 2291.
46. G. Li and G. M. Hieftje, US Patent #56147111, 1997..
47. P. P. Mahoney, G. Li, S. J. Ray and G. M. Hieftje, J. Am. Soc. Mass Spectrom., 1997, 8, 125.
48. D. P. Myers and G. M. Hieftje, Microchem. J., 1993, 48, 259.
49. D. P. Myers, P. P. Mahoney, G. Li and G. M. Hieftje, J. Am. Soc. Mass Spectrom., 1995, 6, 920.
50. X. Tian, H. Emteborg and F. C. Adams, J. Anal. At. Spectrom., 1999, 14, 1807.

---