High-Speed Digital Frequency Scanning Ion Trap Mass Spectrometry

Article pubs.acs.org/ac

High-Speed Digital Frequency Scanning Ion Trap Mass Spectrometry Di Wang,† Friso H. W. van Amerom,‡ and Theresa Evans-Nguyen*,§ †

Department of Pharmacology, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205, United States Mini-Mass Consulting, Inc. St. Pete Beach, Florida 33706, United States § Draper Laboratory, Tampa, Florida 33612, United States ‡

ABSTRACT: A novel means of simultaneous ion injection and mass scanning has been studied. Classic injection methods for ion trap mass spectrometry proceed by loading the trap and scanning out in mutually exclusive time segments. This mode implicitly imparts a duty cycle and acquisition rate limit to ion trap mass analysis. Using digital frequency scanning methods without the use of discrete injection phases, we demonstrate continuous injection and acquisition rates up to 1000 Hz (averaging 400 000 Th/s) in a 3D ion trap configuration showing an alternatively faster control of ions over the classic voltage ramping mass-selective instability scanning method. The digital frequency scanning method may accommodate current advances in high-speed separation [e.g., ultraperformance liquid chromatography (UPLC), two-dimensional gas chromatography (GC×GC), ion mobility spectrometry (IMS), and capillary electrophoresis (CE)] and ion transfer efficiency (e.g., ion funnels). These ion sources may comprise high ion currents with compositions that change quickly and require high-speed mass analysis. Though resolution is compromised with higher acquisition rates, the frequency scanning mode permits a more flexible platform to accomplish the optimal trade-off of speed and mass spectral quality.

R

radio frequency (rf) fundamental voltage, r and z are the dimensions defined by the electrodes’ hyperbolic surfaces, Ω is the rf fundamental frequency, and qz is the parameter at the boundary edge of the Mathieu stability conditions. Thus, to produce the mass/charge spectrum with all other parameters being fixed, the rf fundamental voltage, V, is raised over the course of another ∼200 ms, which imparts instability to successively higher and higher mass ions for ultimate ejection and detection. Further, prescans that permit automatic gain control (AGC) methods require additional time. Therefore, a complete cycle may realistically span ∼500 ms, equating to an acquisition rate of 2 Hz, though single MS scans have been demonstrated with voltage ramp mass scanning in so-called turboscans and microscans over 50−150 ms.6 Notably, in commercial hybrid linear trapping instruments, MS/MS acquisitions are achievable at rates upward of 12 Hz7 but again with voltage ramp mass scanning. Over the past decade, digital ion trap frequency scanning methods have been implemented for the impressive analysis of high mass proteins8 and even ultrahigh mass nanoparticles.9−11 A handful of groups have investigated frequency scanning digital modes of quadrupole and ion trap mass analyzers.8,12−21 In these methods, the fundamental rf waveform applied to the ring electrode is provided by high-speed, high-voltage switches,

ecently, high-throughput chemical analysis capability has grown with biomedical applications in metabolomics,1 proteomics,2 and clinical diagnostics.3,4 Mass spectrometry has played a central role as a tool with great chemical specificity that had already achieved respectably high-speed mass spectral acquisition rates typically in under a second. However, with the increased demands of high-throughput pan-omics applications, pressure has built to go beyond even these speeds. Due to the operational characteristics of different types of mass spectrometers, some are more suitable for high-speed analysis than others. Whereas time-of-flight (ToF) instruments are capable of kilohertz mass spectral acquisition rates, ion traps are typically low-speed mass analyzers limited by duty cycle restrictions but afford tandem mass spectrometry (MS/MS) capability. This is easily shown by the very nature of the ion trap, employing an alternating electric field to trap ions in circuitous paths according to the Mathieu stability parameters. The elegant application of various waveforms to manipulate ion motion, reaction, and ultimate ejection makes the use of ion traps something of an art. Herein, we propose that modern digital frequency waveform applications enhance ion trap performance to accommodate the demands for high-throughput analyses. Traditionally, externally injected ions are first trapped in an oscillating electric field for a distinct time (10 ms−200 ms) and often allowed to cool their trajectories (∼50 ms) toward the center of the trap.5 Mass scanning classically proceeds according to the governing equation m = 8 eV/[qz(2z2 + r2)Ω2], where m is the mass of the ion, e is the charge, V is the © XXXX American Chemical Society

Received: July 31, 2013 Accepted: October 8, 2013

A

dx.doi.org/10.1021/ac402403h | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

(r0 = 0.707 cm, z0 = 0.783 cm) were fitted into a custom-built vacuum chamber. Perfluorotributylamine (PFTBA) was procured from Sigma−Aldrich and introduced directly into the ion source volume with a precision leak valve to a partial pressure in the range of ∼2 × 10−5 Torr according to the chamber pressure gauge. Helium buffer gas was directly leaked into the trap with nominal chamber pressure readings ∼1 × 10−3 Torr, unless otherwise stated, to facilitate ion trapping. Pressure was read by a Hornet IGM401 ion gauge calibrated for helium. Filament current was raised from 1.0 to 1.28 A according to the subsequently faster scan and cycle rates employed. The ion source and lenses employed approximately the same dc voltage potentials of the commercial implementation of the GCQ but without an automatic gain control mechanism. The end-cap electrodes in all experiments were grounded to restrict analysis to the mass-selective instability mode. A high-voltage digital switch was commissioned to be capable of 1500 Vpp, with up to 1 MHz pulse repetition frequency. The Bournlea model 2610 received inputs of a low-level precise digital waveform from a Stanford Research DS345 multifunction generator and a high-voltage dc potential from a Glassman EK series high-voltage power supply. The ions ejected through the rear end-cap electrode were detected by a Photonis Megaspiraltron. After amplification, the analog signal was captured by a LeCroy model 7200A oscilloscope either in single scans or over an average of several sweeps as specified in each experiment. The oscilloscope and multifunction generator were synchronized with a Stanford Research DG535 delay pulse generator to trigger spectral acquisition. Mass calibrations for spectra were based on baseline-subtracted and smoothed spectra, and the mass scale was exponentially fit with standard PFTBA peaks: 69, 131, 264, 414, and 502, according to the NIST chemistry webbook.25 Injection Modes. Figure 2 depicts both the pulsed and continuous operational modes for injection. In the traditional

much like those employed by high-voltage pulsing ToF instruments. In the simplest operation employing mass instability scanning, rather than scanning the trapping voltage as is done classically, the frequency of the ring electrode, Ω, is scanned. This mode was applied in this work to the study of small molecules ( 0.712 (for square waveforms). For both single-shot and 10cycle average spectra, higher pressure data showed higher

Figure 6. (Top left) Single mass scan of PFTBA at partial pressure of 2 × 10−5 Torr at 2 × 10−4 Torr helium buffer gas. (Top right) Single mass scan of PFTBA at partial pressure of 2 × 10−5 Torr at 1 × 10−3 Torr helium buffer gas. (Bottom) Averages of 10 spectra for both pressures.

signal/noise ∼50%. A possible reason that continuous injection modes have not been previously extended may be due to the historical convention of internal ionization schemes using electron beams collinear with the ion trap axis. This would have resulted in significant chemical noise (such as helium ions produced along the route of the beam, during the ejection scan) that was most conveniently mitigated by distinct trapping and injection phases. The collision cross section of the electrons in the ionization beam being so much smaller than any ion meant that even higher pressures could not mitigate the strength of the ionization beam noise. However, when external injection was employed in this case, higher pressures provided modest repression of collinear chemical noise. While chemical D

dx.doi.org/10.1021/ac402403h | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

noise would be the same whether the fill−pause−scan or the continuous injection method was used, signal at fast scan rates for the two modes would differ. Continuous injection effectively harvests ions that the fill−pause−scan method rejects during its scanning phase. Therefore, one might expect that for traps using higher pressures and orthogonal detectors (or nonlinear ion guides), signal/noise enhancement may be achieved via the continuous injection method. In Figure 7, a plot of peak fwhm versus cycle scan of the 30 Hz scan rates in Figure 4 portrays a dependence of resolution

Figure 8. Full width at half-maximum (fwhm) of mass 69, 131, and 264 m/z for scan rates 100, 200, 500, 750, and 1000 Hz, determined from Figure 5.

Figure 7. Full width at half-maximum (fwhm) of masses 69, 131, 264, 415, and 505 m/z for cycle rates at 10, 15, 20, and 25 Hz, determined from Figure 4. Scan rate was kept constant at 30 Hz.

scans and from 100 kTh/s up to 1.2 MTh/s at the end of the 1000 Hz scans. Also, our analog detection system has a limited response time on the order of 10 μs. The detector must keep up with scans at an average of 400 000 Th/s (0.45 mass unit/ μs) at 1000 Hz. The resolution degradation at faster scan speeds was also observed by Ding et al.,8 where at 40 000 Th/s it was not possible to resolve the isotope peaks of horse heart myoglobin ions. The limited performance of the detector amplifier may be avoided by pulse counting, although pulse pileup is much more likely at higher scan speeds.

on the trapping time parameter. Mass m/z 264 displayed lower resolution than both higher and lower masses. At first glance, one may suspect resolution differences to be attributed to mass scan speed variation in which high speeds equate to poor resolution. Recalling that our mass scan speeds are not linear because of our logarithmic frequency sweep, we have referred to the average mass scan speed based on the scan rate and the calibrated mass range. In reality, the mass scan speed varies across a single scan but is nominally fastest at higher masses, such that one would expect the poorest resolution at higher masses compounded at higher scan rates. But as observed in the fwhm comparison, higher scan rates and higher masses actually result in narrower peak widths and suggests that another factor is in play: overfilling of the trap. Overfilling of the trap is normally mitigated by automatic gain control, which is clearly not employed here. But the increased space charge accommodation by wider peak widths at lower scan rates might be understood in the context of the relatively longer pure injection and cooling times in these experiments: 66.7 ms for 10 Hz and 6.7 ms for 25 Hz. Therefore, this figure may be explained by space charge effect resulting in trap overfilling rather than varied mass scanning speeds. We report peak broadening effects as shown in Figure 8, depicting a plot of fwhm for the peaks of mass 69, 131, and 264 during fast scanning up to 1000 Hz for the spectra depicted in Figure 5. Resolution decreases linearly with increased cycle rate and scan rate, which could be attributed to factors such as less time for axial cooling, space charge broadening, and frequency and voltage jitter on the scan waveform. Notably, the scan speed varies from 10 to 120 kTh/s at the end of the 100 Hz

CONCLUSIONS This preliminary work suggests a novel use of digital frequency scanning techniques for dedicated small-molecule analysis, affording a flexible platform to accommodate high-speed applications. By a simple enhancement of signal/noise, the continuous injection method enabled by frequency scanning makes the method even more attractive. Though careful adjustments of the ion currents are necessary and despite the low-resolution drawback at very high speeds, we demonstrate faster mass scan rates than have been previously reported for ion trap mass spectrometry. Fast frequency scanning allowing for an automatic gain control scheme may be as good as or possibly better than current systems because the response time to overloading is much faster. This may prove valuable for now common enhanced separation technologies that currently require ToF instruments to accommodate high-speed peaks with temporal existences of less than a second. Such speeds may even conceivably benefit studies in chemical reaction monitoring. To ultimately determine true mass resolution achievable with this method, several steps have to be taken. First, a matching detector system has to be developed. An automatic gain control system needs to control the number of ions in the trap at all times by manipulation of the ion current or cycle time dynamically (as may be possible with modern field programmable gate array technology) and higher accuracy and precision of both low- and high-voltage electronics to avoid unnecessary jitter in frequency and voltage. Resonance ejection will further improve resolution. Many cases may benefit from frequency scanning modes in which scan speed is more critical than mass

■

E

dx.doi.org/10.1021/ac402403h | Anal. Chem. XXXX, XXX, XXX−XXX

Analytical Chemistry

Article

(20) Wang, L.; Xu, F.; Ding, C.-F. Anal. Chem. 2013, 85, 1271−1275. (21) Bandelow, S.; Marx, G.; Schweikhard, L. Int. J. Mass Spectrom. 2013, 353, 49−53. (22) March, R. E.; Todd, J. F. In Quadrupole Ion Trap Mass Spectrometry; Winefordner, J. W., Ed.; Wiley: Hoboken, NJ, 2005; pp 1−33. (23) Ramanathan, R.; Jemal, M.; Ramagiri, S.; Xia, Y.-Q.; Humpreys, W. G.; Olah, T.; Korfmacher, W. A. J. Mass Spectrom. 2011, 46, 595− 601. (24) Michalski, A.; Cox, J.; Mann, M. J. Proteome Res. 2011, 10, 1785−1793. (25) http://webbook.nist.gov/chemistry/. (26) Song, Q.; Xu, W.; Smith, S. A.; Gao, L.; Chappell, W. J.; Cooks, R. G.; Ouyang, Z. J. Mass Spectrom. 2009, 45, 26−34.

resolution. For example, fast screening for nominal identification of specific small molecules in relatively simple mixtures as in security scenarios embodies one application. Yet these results, combined with previously demonstrated fast precursor ion selection capability through waveform duty cycle control by Brancia et al.15 and digital high resolution at high mass capability by Ding et al.,8 suggest that digital frequency scanning methods surely pose an attractive option for highthroughput biomedical applications. In the future, an investigation exploring supplemental frequency scanning will be reported that achieves a continuous storage method with simultaneous resonance ejection scanning that may be useful for low-power portable devices.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]; phone 813-465-5464. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by Grant HDTRA-11-01-0012 from the Defense Threat Reduction Agency Basic Research Program in Nuclear Forensics. We are grateful to Professor Peter Reilly for insightful discussions and R. Timothy Short for helpful review of the manuscript. We dedicate this work to our beloved mentor and friend, Robert J. Cotter.

■

REFERENCES

(1) Patti, G. J.; Yanes, O.; Siuzdak, G. Nat. Rev. Mol. Cell. Biol. 2012, 13, 263−269. (2) Nilsson, T.; Mann, M.; Aebersold, R.; Yates, I.; Bairoch, A.; Bergeron, J. J. M. Nat. Methods 2010, 7, 681. (3) Shushan, B. Mass Spectrom. Rev. 2010, 29, 930−944. (4) Stanczyk, F. Z.; Clarke, N. J. J. Steroid Biochem. Mol. Biol. 2010, 121, 491−495. (5) McLuckey, S. A.; Van Berkel, G. J.; Goeringer, D. E.; Glish, G. L. Anal. Chem. 1994, 66, 689A−696A. (6) Yang, C. G.; Bier, M. E. Anal. Chem. 2005, 77, 1663−1671. (7) Michalski, A.; Damoc, E.; Lange, O.; Denisov, E.; Nolting, D.; Müller, M.; Viner, R.; Schwartz, J.; Remes, P.; Belford, M.; Dunyach, J.-J.; Cox, J.; Horning, S.; Mann, M.; Makarov, A. Mol. Cell. Proteomics 2012, 11, No. O111.013698. (8) Ding, L.; Sudakov, M.; Brancia, F. L.; Giles, R.; Kumashiro, S. J. Mass. Spectrom. 2004, 39, 471−484. (9) Koizumi, H.; Whitten, W. B.; Reilly, P. T. J. Am. Soc. Mass Spectrom. 2008, 19, 1942−1947. (10) Wang, S.; Johnston, M. V. Int. J. Mass Spectrom. 2006, 258, 50− 57. (11) Wang, X.; Chen, H.; Lee, J.; Reilly, P. T. A. Int. J. Mass Spectrom. 2012, 328−329, 28−35. (12) Ding, L.; Brancia, F. L. Anal. Chem. 2006, 78, 1995−2000. (13) Ding, L.; Kumashiro, S. Rapid Commun. Mass. Spectrom. 2006, 20, 3−8. (14) Berton, A.; Traldi, P.; Ding, L.; Brancia, F. L. J. Am. Soc. Mass Spectrom. 2008, 19, 620−625. (15) Brancia, F. L.; McCullough, B.; Entwistle, A.; Grossmann, J. G.; Ding, L. J. Am. Soc. Mass. Spectrom. 2010, 21, 1530−1533. (16) Koizumi, H.; Jatko, B.; Andrews, W. H., Jr.; Whitten, W. B.; Reilly, P. T. A. Int. J. Mass Spectrom. 2010, 292, 23−31. (17) Koizumi, H.; Whitten, W. B.; Reilly, P. T. A.; Koizumi, E. Int. J. Mass Spectrom. 2009, 286, 64−69. (18) Lee, J.; Marino, M. A.; Koizumi, H.; Reilly, P. T. A. Int. J. Mass Spectrom. 2011, 304, 36−40. (19) Bandelow, S.; Marx, G.; Schweikhard, L. Int. J. Mass Spectrom. 2013, 336, 47−52. F

dx.doi.org/10.1021/ac402403h | Anal. Chem. XXXX, XXX, XXX−XXX