Article pubs.acs.org/ac
Cite This: Anal. Chem. XXXX, XXX, XXX−XXX
High-Capacity Electrostatic Ion Trap with Mass Resolving Power Boosted by High-Order Harmonics Li Ding* and Aleksandr Rusinov
Downloaded by BOSTON UNIV at 21:15:51:745 on May 29, 2019 from https://pubs.acs.org/doi/10.1021/acs.analchem.9b00206.
Shimadzu Research Laboratory (Europe) Limited, Wharfside, Trafford Wharf Road, Manchester, M17 1GP, U.K.
ABSTRACT: A form of electrostatic ion trap mass analyzer, named the orbital frequency analyzer (OFA), has been developed. The ions in the analyzer are trapped around the middle plane and orbit around the central axis perpendicular to the middle plane with high-ellipticity and precessing trajectories. The orbital frequency of the ions in this device has been optimized to be independent of ions’ energy so that the image charge signal picked up by some of the field-forming circular/ring electrodes can be used to produce a mass spectrum after Fourier transform data processing. Spectra acquired by the OFA are rich of high-order harmonics, which offer higher mass resolving power than that for fundamental frequency components. The experiment shows that the resolving power is proportional to the harmonic order and exceeds 150 k for mass-to-charge ratio (m/z) of 526 Th and the transient length of 500 ms. Using high-order harmonics, an isotopic cluster of a heavy protein was resolved with a shorter transient length. The transient signals from different pick-up electrodes give different waveform shapes, and therefore, their harmonic peak distributions in a frequency spectrum are different, thus allowing the removal of unwanted harmonic peaks. The preliminary results also show a wide dynamic range of the analyzer.
F
motion which can be converted to the mass spectrum of the trapped ions. The mass resolving power (or resolution) of these FTMS devices depends on the frequency of ion oscillations and transient length of the image charge signal.5 When there is a need for high scan rate, the transient length has to be short and resolving power is limited. The mass resolving power per acquisition time for an electrostatic trap is directly proportional to the constant A in eq 1, before it reaches the physical limits of the device (field accuracy, power supply stability, etc.). On this specification, FTMS in any form does not really compete with the time-of-flight mass analyzers (such as in a Q-TOF) even if considering the need of tens of TOF scan averages to match the peak statistics. A compact trap size usually produces a higher oscillation frequency for an ion at a given acceleration voltage
ourier transform mass spectrometry (FTMS), based on electrostatic ion traps (ESTs), has been developed rapidly over the last 2 decades, symbolized by the success of the commercialization of the Orbitrap.1 Other types of ESTs have also been developed, such as the electrostatic linear ion trap (ELIT),2,3 Cassinian ion trap,4 the planar electrostatic ion trap (PEIT),6 as well as the ion trap using a hybrid magnetic−electric field.7 The electric field in any of these traps is designed to keep ions flying in the trap and make them oscillate harmonically or at least energy-isochronously along one of the motion directions. An image charge signal picked up by one or more of the electrodes bears information on the frequency of ion oscillation, which relates to the mass-to-charge ratio (m/z) of the ion, as imy f = Ajjj zzz kz{
−1/2
(1) Received: January 11, 2019 Accepted: May 15, 2019
where A is the constant of proportionality. Fourier transform of the image charge signal gives a frequency spectrum of ion © XXXX American Chemical Society
A
DOI: 10.1021/acs.analchem.9b00206 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 1. Schematic diagram of the orbital frequency analyzer: (A) section view by removing the front set of electrodes; (B) section view by cutting the analyzer along a diameter.
and mass-to-charge ratio,8 which offers higher mass resolution per transient time. However, the smaller trapping volume results in higher space charge density, leading to space charge related deteriorations of performance, such as mass inaccuracies, nonlinear spectral intensity responses, and coalescences.9,10 There have been attempts to utilize the frequency components that are the multiples of the ions’ rotation or oscillation frequency in the ion trap. In the ion cyclotron resonance cell,11,12 frequency multiples at 2×, 3×, etc. were created by segmenting the detector electrodes. In the ELIT,2,13,14 the relative smaller size of the pick-up electrode compared to the length of the trap resulted in an image charge signal in pulsed waveform which contains abundant higher harmonics. Both technologies gave alternative ways to boost the resolving power per transient time without fly path length reduction. In an electrostatic ion trap, of any of the above-mentioned formats, ion oscillatory motion is reflected at the two ends of the trap. There, the ions’ motion becomes the slowest (stop and return), and the density of the ion population is the highest, resulting in the strongest space charge interaction between ions.21 In the configuration of the ELIT, ion trajectories are gathered around the central axis, so there is not much room near the reflecting ends, leading to strong space charge interaction in the ion cloud. In the Orbitrap and the Cassinian trap, the ions’ rotational motion or other unharmonized motions extend the ion beam to form a cylindrical or rectangular sheet, thus reducing the space charge density in the reflecting ends. In the configuration of PEIT which we previously reported,6 ion trajectories are gathered around the middle plane. The ion cloud for one m/z takes a ring shape, and it is reflected only at the perimeter of the trap where it has a maximum possible expansion and therefore lowest density. Due to the existence of the rotational energy, each ion trajectory forms a precessing orbit around the central axis which is orthogonal to the middle plane. Such a rotational precession, initially proposed for a TOF analyzer,15 has an advantage to average out any minor mechanical defects of ring electrodes of a trap.16 As ions are orbiting around the central axis they do not pass the center of the trap, thus avoiding high space charge density there, as in the case of diametrical oscillations. After the working principle was proved by the simulation work, we designed and constructed the EST for ultrahigh-resolution mass analysis. On the basis of the shape of ion trajectories in the trap, and its purpose of mass
analysis, we named the device as the orbital frequency analyzer (OFA), and hereafter we present its preliminary performance.
■
WORKING PRINCIPLE As described in earlier publications,6,17 the OFA is constructed by two sets of coaxially placed ring electrodes. The sets are opposed to each other, leaving a space in between for trapping the ions. Two section views of the OFA system are shown in Figure 1, where Figure 1A also shows a curved injection ion guide (CIG) outside the perimeter of the OFA, for ion transfer and injection. Each circular and ring electrode of the OFA is supplied with a dc voltage, designated as Vn (n = 1, 2, ..., 10 as shown in Figure 1B). Ions from the external ion source are precooled (0.02−0.04 eV) and guided through the thin quadrupole, which limits the gas load, and are continually guided into the CIG surrounding the OFA which is kept in an ultrahigh-vacuum (UHV) environment. The CIG is basically a half-circular quadrupole ion guide, driven by high-frequency square wave voltage, and its function has been previously explained6 in detail. Without being stopped and trapped in the CIG, the ions are pushed into the OFA by switching the quadrupole square wave to a dipole pulse (in a radial direction) on the electrodes of the CIG. The largest ring electrode of the OFA, which is next to the CIG, is supplied with a switchable voltage V10 and works as the injection gate as well as the reflector electrode. The ions are pushed in during the injection by lowering the V10 voltage for a short time, from 3 to 10 μs, depending on the mass range, and then the voltage is quickly restored so that a potential barrier is established to reflect the ions. The voltage V10 has the highest value among all OFA electrodes so that ions are reflected every time they reach the edge of the trap. Due to the initial tangential energy of ions (energy along the ion transfer direction) in the curved ion guide, trapped ions will never pass the central axis of the trap. As shown by the spirograph line in Figure 1, the motion of the trapped ion follows a high-ellipticity, precessing orbit around the central axis of the OFA. Ions injected from a different part of the CIG (with a different azimuthal angle) quickly form a disc-shaped ion cloud occupying the whole middle plane of the analyzer. In general, any field-forming electrode in the OFA can be used as an image charge pick-up electrode. In practice, it is ideal to use the central circular electrode and some inner ring electrodes to pick up the image charge signal because they bear lower or zero B
DOI: 10.1021/acs.analchem.9b00206 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 2. Aberration of an test ion trajectory (dashed) relative to a reference trajectory (solid), studied between a starting plane and a detection plane each located at a perihelion point: a is the velocity in the z direction, and b is the velocity in the x direction on the respective plane.
system (r−z) and the aberration of the test ion trajectory relative to the reference ion trajectory is studied between the two connected perihelion points. The reference ion (trajectory shown in the right figure and by the solid line in the left figure) starts from the first perihelion at radius rmin, with z = 0 and zero axial velocity so that its trajectory always lies in the middle (z = 0) plane. When it completes the half cycle of the orbital motion at the second perihelion, it returns to rmin and z = 0. The starting and detection planes are defined as the planes at the two perihelion points perpendicular to the reference trajectory in the left figure, and the ion oscillatory motion in the trap can be considered as repetitive flights from the starting plane to the detection plane. The position and velocity parameters of the ions on the planes are defined in 2D Cartesian coordinates, in order to agree with the definition in the classic aberration theory. The test ion (shown by the dashed line) starting with parameters z0, a0, x0, b0 at the starting plane will arrive at the detection plane with z1, a1, x1, b1 parameters, and this will take half of the orbital period time t1 − t0. Here, a and b parameters stand for the velocity components in axial (z) and drift (x) directions and δ stands for the relative total energy variation. Considering the symmetry of the system and omitting terms of third order and above, we can write aberration expansion20 for the time of flight between starting and detection planes as
voltage and have relatively lower capacitance between each other. In this work, both the central circular electrode and first ring electrode are used for image charge detection, and the capacitance between them is less than 5 pF. Every time an ion cloud passes a pick-up electrode, an image charge pulse is induced in the electrode and the amplifier connected to the electrode will amplify and send it to an analog-to-digital conversion device for data recording. The periodically pulsed signal contains the information on mass-to-charge ratio, and a mass spectrum can be obtained after applying the fast Fourier transform and subsequent deconvolution of a frequency spectrum. The ion motion must be isochronous for the OFA to be a mass analyzer, which means the orbital frequency is to be independent of the ions’ injection positions and energy, and only dependent on their mass-to-charge ratio. This was achieved through the design and optimization using an ion optical simulation tool, such as our in-house developed package AXSIM or its later variation SIMSOL.
■
DESIGN OF THE OFA The field design and optimization relied on intensive ion optical simulation, and it was previously reported for the tangential energy Et of up to 1.75 eV.6 When Et → 0, ion motion trajectories are within a diametric plane (a plane passing through the central z-axis), and the optimization of the field can follow the classic planar 2D theory19 where at least second-order energy focusing can be achieved. However, zero or very low Et in the CIG at the injection moment will result in ion oscillations with trajectories of very low precession speed. As shown by our previous work, 16 this operation mode is vulnerable to mechanical errors and causes ion segregation in different sectors of the trapping volume. Ions oscillation frequencies may be different for different trapping sectors and this caused peak split. Simulations proved that, if ions have an initial Et of 4 eV, it will speed up precession and help to overcome circular unevenness up to 60 μm. With mechanical error in combination of circularity and concentricity less than 60 μm, ions with this initial tangential energy can quickly spread out in the whole trapping disc plane. This not only avoids possible peak splitting but also makes full use of the large trapping volume of the OFA, thus increasing the charge capacity of the analyzer. However, time aberration in 3D space should be considered for field optimization when ions are orbiting around the z axis and Et is as high as 4 eV; this optimization becomes very complicated. As shown in Figure 2, the trajectories of a test ion and a reference ion are placed in the cylindrical coordinate
t1 = t0 + (t |δ)δ0 + (t |zz)z 0 2 + (t |za)z 0a0 + (t |aa)a0 2 + (t |δδ)δ0 2 + (t |x)x0 + (t |xx)x0 2 + (t |bb)b0 2 + (t |xδ)x0δ0
(2)
Note that the terms in the second line are due to deviations in drift direction, which result from the initial spread of Et. As reported before6 there is a suitable V2 that enables (t|x) = 0 for an ion with initial tangential energy Et. Thus, in the optimization procedure we first find such V2 value and determine rmin for the reference ion trajectory, as shown in Figure 2, right. To avoid complication, we used a quasi-3D approach. In this approach the test ion is also launched with x0 = 0 in the coordinate system of the reference ion trajectory. Then, the 2D optimization algorithm is applied in which an ion having a0 and z0 deviations at the starting plane is launched in simulation and its a1, z1 deviations are measured at the detection plane. No deviations of x and b are considered (x0 = 0, b0 = 0), and it is assumed that there are no deviations in x1 and b1 parameters at the detection plane, which in general is not valid. Nevertheless, due to the small values of the latter deviations the approach showed successful results which were proved by a full ray tracing 3D simulation of a “real” ion cloud, with actual r, z, Vr, Vz, Et distributions at the injection moment. Using our C
DOI: 10.1021/acs.analchem.9b00206 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Table 1. One Set of Optimized Voltages (V) of Electrodes in the OFA and Their Respective Injection Energy Er (eV) V1
V2
V3, V4, V5
V6
V7
V8
V9
V10
Er
0
290
0
−500
−5600
4100
3120
4400
3800
where voltages V7 and V8 (in opposite signs) are applied is the strongest, being responsible for z focusing of the ion cloud. The gaps between these electrode pairs are narrower than those of all other electrodes, and their profiles are also specially designed in order to reduce the required voltage.17 If these electrodes had flat tops similar to other electrodes, their optimized voltages would be much higher than those listed in Table 1, and this could result in discharge between these electrodes.
optimization code, V7, V8, V9, and the total energy E0 can be scanned to achieve the aberration coefficients in the first line in the eq 2 to be zero. With the newly optimized potentials, we then check the time focusing to tangential energy Et, because (t|x) is not zero anymore with the new set of electrode potentials. Necessary adjustments on V2 are made to minimize the (t|x) coefficient in the second line. With the new set of voltages Vn, the above process is repeated, and finally, optimal voltages Vn can be obtained by such iterations to get second-order energy, secondorder spatial, and first-order Et focusing. There are other considerations that need to be taken, such as that the optimization should work for zero space charge condition as well as some light space charge condition. While the coefficient (t|δ) and (t|δδ) reached zero through the above optimization, the remaining (t|δδδ) may be positive or negative. It has been found that a small positive (t|δδδ) may be helpful in the situation of space charge interaction, in which case, the leading ions gain kinetic energy from the following ions via Coulomb repulsion. Higher energy results in a higher orbital period, so the leading ions swap the position with the following ions, just like the migrating birds taking turns to lead the flock. It worth mentioning that tuning the high-order aberration coefficient (t|δδδ) to positive is different from the “selfbunching” mode which was used in ELIT (or electrostatic ion beam trap, EIBT).21 In self-bunching, where (t|δ) > 0, the original energy spread of ions causes immediate energy swap and cloud bunching via space charge interaction. While this stops the ion cloud of same m/z spreading out, it also causes coalescence between ion clouds having small difference in m/z. When ion abundance is low and space charge interaction is weak, selfbunching disappears and a positive (t|δ) leads to a large time defocusing, so we must avoid it. We have found16 that the geometry of the focusing ring electrodes (with voltages V7 and V8) may be varied to control the sign of remaining term (t|δδδ) so that the optimized geometry and voltages suit both non-space charge and slight space charge conditions. One of the optimized voltages set for Et = 4 eV is listed in Table 1. With the voltage setting above, an ion (assumed charge +e) injected from the CIG will gain 3800 eV of kinetic energy when it reaches the center of the analyzer. When it bounces back to the original injection side, the voltage V10 has restored to 4400 V; thus, it is entrapped. As has been presented previously,16 the optimized OFA can achieve a time aberration less than 50 ps for a half-turn time of 4.4 μs (m/z = 526 Th) and a radial energy spread of 100 eV. This is equivalent to a frequency spread about 2.5 Hz at frequency of 250 kHz. In fact, our simulation showed that the radial energy spread of ions is less than 50 eV after being injected with 300 V(p−p) dipole voltage on the CIG.6 All ring electrodes and the central circular electrode are mounted on a base made from alumina ceramic. The nonflat profile of the ceramic bases (with deep circular grooves and steps) gives longer tracking distances between adjacent electrodes having large voltage differences. The two electrode arrays are assembled coaxially with the common central axis z and are axially offset from each other, defining a trapping space between them. The gaps between the opposing ring electrodes are 5−9 mm wide depending on the field strength required for the respective regions. The field near the electrodes 7 and 8
■
EXPERIMENTAL SECTION A Shimadzu LCMS 8040 was modified to provide the mass spectrometer front end of the experimental system. The quadrupole analyzer, Q3 in the original system, was removed, and a rectangular cross section ion guide was used to bridge the collision cell to the OFA, installed in a separate chamber and pumped by an additional turbomolecular pump with pumping speed of 250 L/s. The bridging ion guide includes a storage segment at the high-pressure end and a thinner quadruple segment (as seen in Figure 1) with inscribed radius of 0.5 mm, for limiting the gas flow to the UHV chamber. It gives about 20% ion transmission, and the OFA chamber can reach 5 × 10−8 Pa pressure after 48 h of baking. At the end of 180° arc of the CIG, a channeltron particle detector was installed. This detector can either receive ions that have traveled through the CIG, or ions that were injected into the OFA and then released at the opposite side of the CIG. The former measurement is useful for tuning the incoming ion beam and optimization of the tangential energy of ions in the CIG. The latter measurement is also called the TOF test, activated by lowering down the voltage V10 on gate/reflecting electrode once again after several turns of flight of ions in the analyzer. Using the time-of-flight signal for various CIG potentials, preliminary tuning of the OFA can be achieved. An in-house designed image charge amplifier was used for signal pick-up and amplification, where the first-stage preamplifiers were constructed by J-FET low-noise transistors, and their UHV-compatible printed circuit boards (supplied by Rogers Corporation) were placed very close to the central circular and first ring electrodes inside the vacuum. After a subsequent amplification by an external amplifier, the two channel signals were recorded by a Picoscope oscilloscope (Pico Technology, U.K.) and transferred to a PC for Fourier transform processing. In order to obtain live frequency spectrum display, in-house written software for data acquisition and conversion uses the GPU card installed in the PC to perform the Fourier transformation. All spectra presented in this paper were obtained by means of fast Fourier transform and are shown in magnitude mode, with Hann window function, and zero padded to 4× the original transient length. A syringe pump operating at 20 μL/min was used for feeding the sample into the electrospray ionization (ESI) source. Samples of MRFA (Met-Arg-Phe-Ala acetate salt), cytochrome c, and buspirone were purchased from Sigma-Aldrich, and solutions at the concentration of 1 nM to 1 μM, in H2O and CH3CN, were prepared via standard chemical lab procedures.
■
RESULTS AND DISCUSSION Voltages from simulation (as shown in Table 1) provided good guidance for setting up the operation conditions of the OFA. D
DOI: 10.1021/acs.analchem.9b00206 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 3. Image charge signal transient and its FFT for the isolated MRFA ions. Signal was averaged 20 times. The insets for the second, fourth, and sixth harmonic peaks show the isotope fine structure being resolved.
Figure 4. Spectrum with multiple charge states and harmonic orders from protein sample bovine cytochrome c. The insets using the m/z scale give the zoom-in details for the first and the third harmonics of isotope cluster groups of [M + 15H]15+.
The real operation voltages may be 10% higher or lower than the list values while their ratios are kept. Fine tuning by measuring the TOF signal and image charge signal is necessary to reach good signal intensity as well as time focusing.
Figure 3 shows a 500 ms transient of an image charge signal acquired by the central pick-up electrode and its FFT spectrum (in frequency scale) for a 100 nM MRFA sample. The first and the second isotope ions (m + 1, m + 2) were selected by the E
DOI: 10.1021/acs.analchem.9b00206 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 5. (A) The 99 ms image charge transient for selected [M + 15H]16+ of cytochrome c, and (B) the zoom-in mass spectrum using FFT with a Hann window.
quadrupole mass filter in front of the OFA, so neither monoisotopic MRFA ions nor other impurity ions contribute to the spectrum. High-harmonic peaks up to eighth order are shown, although their intensities decrease with the harmonic order. The mass resolutions were calculated using the relation f M = 2Δf , where the Δf is the full width at half-maximum of the ΔM
gradually spread out over the whole trapping region, making the image charge signal undetectable. At the time of the second beat, one isotope ion catches up its 1 Da heavier isotope by a full lap, so all ions in the isotope cluster become in phase again, thus making the useful signal to be maximum. On the basis of this understanding, the time between the two beats can be calculated as22
frequency peaks, and it almost linearly increases with the harmonic order N. The proportionality constant A in eq 1 is given by calibration between frequency and mass: A = 4.79 × 106N (Hz Th )
tb =
2M f1
(4)
It is about 147 ms for a protein isotope cluster with M ≈ 12 327 Da at charge of 15. The transient time for the above experiment is only about the 2 times the beat interval, so for a normal FTMS with a Hann window it is just about to resolve the isotope cluster, and this is exactly the case for peaks at H1. However, using higher harmonics from H3, we see that the isotope peak can be baseline-resolved. When the transient length falls short of the beat interval, the conventional FTMS will not be able to resolve the isotope structure and the resolution of mass peaks will be the same as though the transient length just covers the initial beat. This is not the case for the OFA where high harmonics are utilized. The advantage is illustrated by using a transient length of 99 ms, which is shorter than one beat interval or before the first lapping point (ion catches the 1 Da heavier isotope by a full lap). In this experiment, using the same cytochrome c sample, the isotope distribution for [M + 15H]16+ was selected by the upper-stream quadrupole mass filter. The fundamental frequencies of this isotope cluster are around 172 kHz. The conventional beat interval tb is about 143 ms, so the second major beat is not reached at 99 ms transient time as shown in Figure 5A. FFT with Hann was used in data processing, and the zoom-in mass spectra are shown in Figure 5B. The H1 and H2 groups do not give enough resolving power to reveal the isotope structure of the protein. However, at H3 the isotope peaks just start to separate, whereas at H5 they are totally separated. The resolving power at
(3)
where N is the harmonic order. The insets show the local zoomin spectrum regions for the second, fourth, and sixth harmonics (marked as H2, H4, H6). In the H2 inset, both m + 1 (right) and m + 2 (left) peaks are shown, where the two isotope fine structures of 13C2 (526.261 Th) and 34S (526.272 Th) just start to be resolved. They are fully resolved in H4 and H6. The mass resolving power for H6 reaches 155 450, and the small 18O and 15 13 N C peaks in between the two distinct peaks are about to be revealed. Figure 4 gives the FFT spectrum of a 300 ms image charge signal transient in an experiment with electrospray of 1 μM cytochrome c. The charge states from 12+ to 18+ and harmonic orders from H1 to H5 are all shown in the spectrum. The insets show the details for the 15+ charge state, where the isotope cluster structure is hardly seen at the first harmonic but it is almost baseline resolved at the third harmonic. It is worth mentioning that the orbital frequency of the [M + 15H]15+ group is only about 84 kHz, and the H1 frequency f1 is double the ion orbital frequency as ions are passing by the central pick-up electrode twice in one orbital cycle. It is well-established in FTMS that, in order to resolve the isotope cluster of a high-mass protein, the length of a signal transient should cover at least two beats of interfered image charge signal.18 After the start of oscillations, all isotope ions F
DOI: 10.1021/acs.analchem.9b00206 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
indicate that the ion cloud passes the ring electrode twice. This difference in waveforms gives different harmonic distributions which allow us to generate certain coefficients in order to eliminate one unwanted harmonic component. There are other deconvolution methods which can also be used,14,24,25 separately or combined, to remove the overlaps between the interested and the unwanted harmonic peaks. These methods applied to the signal from the OFA will be published in a separate paper in future. Although we have not thoroughly studied the space charge tolerance of the OFA up to date, the analyzer has shown a very wide dynamic range of signal response. It was measured in an experiment with sample injections using an autosampler directly feeding an LC pump running at 200 μL/min without using a separation column. Buspirone solutions at concentrations of 1 nM, 10 nM, 100 nM, 1 μM, and 10 μM were prepared, and the injection volumes of 1, 2, 3, 5 μL were optionally selected in the experiment to form various injection amounts. The front quadrupole worked in broad mass range mode (no resolving dc). Figure 7 gives the measured signal response on a
H3 and H5 in Figure 5B are comparable to that of at H1 and H3 in Figure 4. This proves that the resolution is improved with the increase of the length of signal transient but does not necessarily step up to reveal the isotopic peaks for protein until the neighboring isotopes lapped each other and create the major beat of the image charge signal again. In fact, the transient signals of the ion packets in the protein isotopic cluster do not totally cancel each other through interference in between the major beats. It can be seen in Figure 5A that there are many small “sub-beat” structures before the second major beat which would be at 147 ms, and the FFT can make use of such signal structure to get better resolving power in high harmonics. These “sub-beat” structures are caused by the higher-order frequency components of the image charge signal, because their frequency is 2, 3, or 4 times higher, so the high harmonics experience faster beats than the fundamental frequency. Due to this phenomenon, the measurement time for achieving isotopic resolution of a highly charge protein may be shortened. Highly charged protein ions tend to have very large collision cross section with background residual gases and they will not survive the long flight once a collision with the residual gas molecules occurs. Shortening the measurement time also relaxes the extreme high-vacuum requirement in the analyzer chamber. While a frequency spectrum consisting of many high harmonics can improve mass resolving power, it also complicates the spectrum structure, especially when a wide mass range of ions are measured together. As seen in Figure 4, the H4 group is about to overlap into H5 group, although the m/z range from 12+ ions to 18+ ions is not very large (
