Anal. Chem. 2006, 78, 2113-2120

Accelerated Articles

Performance Evaluation of a Hybrid Linear Trap/

Alexander Makarov,* Eduard Denisov, Alexander Kholomeev, Wilko Balschun, Oliver Lange, Kerstin Strupat, and Stevan Horning

Thermo Electron (Bremen) GmbH, Hanna-Kunath-Strasse 11, Bremen 28199 Germany

Design and performance of a novel hybrid mass spec- A typical example of such a mass spectrometer is the quad- trometer is described. It couples a linear mass rupole/time-of-flight (TOF) hybrid,1,2 wherein precursor are spectrometer to an orbitrap mass analyzer via an rf-only selected by a quadrupole mass filter and accurate mass determi- trapping quadrupole with a curved axis. The latter injects nation (including analysis of fragment ions) is carried out in an pulsed ion beams into a rapidly changing electric field in orthogonal-acceleration TOF. Though very successful for a range the orbitrap wherein they are trapped at high kinetic of applications, such hybrids nevertheless suffer from low ion energies around an inner electrode. Image current detec- transmission (resulting in poor MS/MS sensitivity and detection tion is subsequently performed after a stable electrostatic limits) and a limited intensity range over which accurate mass field is achieved. Fourier transformation of the acquired data can be acquired. transient allows wide mass range detection with high These shortcomings were overcome in a next-generation resolving power, mass accuracy, and dynamic range. The hybrid instrument, a linear ion trap/Fourier transform ion cyclo- 3 entire instrument operates in LC/MS mode (1 spectrum/ tron resonance (FTICR) mass spectrometer. Like earlier FTICR 4,5 s) with nominal mass resolving power of 60 000 and uses hybrids, this mass spectrometer accumulates ions externally to automatic gain control to provide high-accuracy mass a superconducting magnet, but in addition, it combines high n measurements, within 2 ppm using internal standards trapping capacity, MS capabilities, and automatic gain control 6 and within 5 ppm with external calibration. The maximum (AGC) of linear ion trap mass with the unsurpassed resolving power exceeds 100 000 (fwhm). Rapid, auto- mass accuracy, dynamic range, and resolving power of FTICR mated data-dependent capabilities enable real-time ac- mass spectrometers. The unprecedented specificity and quality of data from this combination necessitates the added complexity quisition of up to three high-mass accuracy MS/MS of the superconducting magnet. This inspired the quest for a mag- spectra per second. netic field-free analyzer of comparable performance, which would be more compatible with the capacities of a typical laboratory.

(1) Chernushevich, I. V.; Loboda, A. V.; Thomson, B. A. J. Mass Spectrom. Increasing speed of chromatographic separation and complex- 2001, 36, 849-865. ity of analyzed mixtures provides a continuous impetus for the (2) Morris, H. R.; Paxton, T.; Dell, A.; Langhorne, J.; Berg, M.; Bordoli, R. S.; Rapid Commun. Mass Spectrom. 10 - development of faster and simultaneously more intelligent and Hoyes, J.; Bateman, R. H. 1996, , 889 896. robust mass spectrometric detectors. Reliable separation and (3) Syka, J. E. P.; Marto, J. A.; Bai, D. L.; Horning, S.; Senko, M. W.; Schwartz, reliable identification of complex mixtures with multiple coeluting J. C.; Ueberheide, B.; Garcia, B.; Busby, S.; Muratore, T.; Shabanowitz, J.; Hunt, D. F. J. Proteome Res. 2004, 3, 621-626. compounds necessarily require higher resolving power, while (4) Senko, M. W.; Hendrickson, C. L.; Emmett, M. R.; Shi, S. D.-H.; Marshall, structural analysis by MS/MS and accurate mass determination A. G. J. Am. Soc. Mass Spectrom. 1997, 8, 970-976. become complementary confirmation tools. (5) Belov, M. E.; Nikolaev, E. N.; Anderson, G. A.; Udseth, H. R.; Conrads, T. P.; Veenstra, T. D.; Masselon, C. D.; Gorshkov, M. V.; Smith, R. D. Anal. Chem. 2001, 73, 253-261. * Corresponding author. Tel: +49(0)421-5493-410. Fax: +49(0)421-5493-426. (6) Schwartz, J. C.; Senko, M. W.; Syka, J. E. P. J. Am. Soc. Mass Spectrom. E-mail: [email protected]. 2002, 13, 659-669.

10.1021/ac0518811 CCC: $33.50 © 2006 American Chemical Society Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 2113 Published on Web 02/18/2006 From its first appearance as a proof-of-principle device,7 the ejected to a pair of secondary electron multipliers.6 An important orbitrap mass analyzer was considered as a potential alternative feature of the instrument is the procedure of AGC,12 wherein a for FTICR because of its high resolution and good mass accuracy. short prescan is used to determine the ion current within the mass Using the principle of orbital trapping in electrostatic fields,8 the range of interest, hence enabling storage of a defined number of orbitrap consists of an inner and an outer electrode, which are ions (“AGC target value”) in the subsequent analytical scan. As shaped to create a quadro-logarithmic electrostatic potential. Ions in the LTQ FT mass spectrometer, all modes of LTQ operation rotate about the inner electrode and oscillate harmonically along remain available, but they are complemented by the ability to its axis (the z-direction) with a frequency characteristic of their analyze ions in an additional “mass detector” (FTICR or, as in m/z values. An image current transient of these oscillations is this work, the orbitrap). converted to a frequency spectrum using a Fourier transform In the LTQ orbitrap, a transfer octapole (300 mm long, 400 V similar to the approach used in FTICR. For a given spread of p-p rf, 5.7 mm inscribed diameter) delivers ions into a curved initial parameters of ions, the orbitrap provides the best perfor- rf-only quadrupole whose central axis follows a C-shaped arc mance within the smallest dimensions, relative to other types of (hence the name C-trap). The C-trap uses rods with hyperbolic electrostatic traps,8 thus making it the most suitable candidate surfaces and is enclosed by two flat lenses with apertures for ion for image current detection and Fourier transform mass spec- transport through them. The plate between the octapole and C-trap trometry. forms the gate electrode and the other plate forms the trap Initially, high resolving power, internal mass accuracy, and electrode. The enclosed volume is filled with nitrogen bath gas high space charge capacity were demonstrated with the orbitrap at ∼1 mTorr via an open-split interface connected to the nitrogen mass analyzer using pulsed ion sources. Because of high ion line of electrospray . Nitrogen has been chosen as a velocities and the absence of any collisional cooling inside the bath gas in favor of due to better collisional damping and orbitrap, injection must occur in a short burst, on a time scale lower gas carryover toward the orbitrap. The C-trap assembly is well below 1 ms (actually, only a few microseconds if coherence pumped by the split-flow turbopump of the LTQ mass spectrom- of ion packets is to be provided by the injection, thereby eter, as indicated in Figure 1. eliminating the need for additional excitation9). For continuous Ions start from the back section of the linear trap held at a dc ion sources, this is possible only when additional ion storage is offset of 6-10 V (here and below voltages are given for positive introduced to store ions and then inject them rapidly into the ions). After acceleration into the transfer octapole by a 2-10 V orbitrap. Such implementation for injection of electrosprayed ions potential difference, ions are transferred into the C-trap held at a into the orbitrap was initially based on axial ejection of ions from dc offset of 0 V with an applied rf voltage of 500-2000 V p-p and a linear rf-only quadrupole.9-10 As the space charge capacity of are reflected by the trap electrode (12-15 V). The gate electrode such a trap is compromised by the need to extract ions within is always kept at a positive offset of 3-6 V. On entry into the hundreds of nanoseconds, use of large ion numbers resulted in C-trap, ions lose energy in collisions with nitrogen bath gas, these great variation of ion kinetic energies, together with angular and collisions being mild enough to avoid any fragmentation. Due to spatial spreads that limited mass range, transmission, dynamic the relatively low gas pressure and short length of the C-trap, ions range, and mass accuracy over a wide mass range. need more than one pass through the entire system to be trapped. This work involves a new approach to ion storage that is based Nevertheless, ions finally come to rest at the place where gas on orthogonal, rather than axial, ion ejection from an rf-only collisions occur with the lowest dc offset along their path, i.e., in quadrupole.11 This implementation alleviates the above problems the C-trap. and paves the way for a hybrid mass spectrometer combining the As the result of collisional cooling, the ions form a thin, long tandem capability of the linear ion trap mass thread along the curved axis of the C-trap. This thread is spectrometer with the high resolution and mass accuracy capabil- compressed axially by applying 200 V to both the gate and the ity of the orbitrap. This combination allows high-quality accurate trap electrodes. After that, the rf voltage on the quadrupolar mass MSn spectra to be acquired using brief ion accumulation electrodes of the C-trap is rapidly ramped down (over 100-200 periods and relatively small ion populations. Here we characterize ns) and dc pulses are applied to the electrodes as follows: 1200 the analytical parameters and features of this hybrid instrument. V to the push-out electrode (i.e., the electrode furthest from the center of C-trap curvature), 1000 V to the pull-out electrode (the EXPERIMENTAL SECTION electrode closest to the center of curvature), and 1100 V to both Figure 1 presents a schematic diagram of the LTQ orbitrap the upper and lower electrodes. This voltage distribution forces hybrid mass spectrometer (Thermo Electron, Bremen, Germany). ions orthogonally to the axis of the C-trap (center of curvature of The design, operation, and control of the front-end LTQ mass the C-trap) where they leave via a slot in the pull-out electrode. spectrometer has been described elsewhere.6 Briefly, ions from Unlike axial ejection,9,10 fast and uniform extraction is provided the electrospray ion source are admitted via rf-only multipoles into for large ion populations. the linear trap of the LTQ, wherein ions are analyzed and radially After leaving the C-trap, the ions pass through appropriately curved ion optics, are accelerated to high kinetic energies, and (7) Makarov, A. A., Anal. Chem. 2000, 72, 1156-1162. (8) Kingdon, K. H. Phys. Rev. 1923, 21, 408-418. converge into a tight cloud, which is able to pass through a small (9) Hardman, M. E.; Makarov, A. A. Anal. Chem. 2003, 75, 1699-1705. entrance aperture and enter the orbitrap tangentially. On their (10) Hu, Q.; Noll, R.; Li, H.; Makarov, A. A.; Hardman, M. E.; Cooks, R. G. J. way from the C-trap, ions pass through three stages of differential Mass Spectrom. 2005, 40, 430-443. (11) Makarov, A. A.; Denisov, E.; Lange, O.; Kholomeev, A.; Horning, S. Proc. pumping until they reach the ultrahigh vacuum compartment of 53rd Conf. Am. Soc. Mass Spectrom., San Antonio, TX, June 5-9, 2005; Poster 1885. (12) Schwartz, J. C.; Zhou, X. G.; Bier, M. E. U.S. Patent 5,572,022.

2114 Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 Figure 1. (A) Schematic layout of the LTQ orbitrap mass spectrometer: (a) Transfer octapole; (b) curved rf-only quadrupole (C-trap); (c) gate electrode; (d) trap electrode; (e) ion optics; (f) inner orbitrap electrode; (g) outer orbitrap electrodes. (B) Simplest operation sequence of the LTQ orbitrap mass spectrometer (not shown are the following: optional additional injection of internal calibrant; additional MS or MSn scans of linear trap during the orbitrap detection) the orbitrap sustained at ∼2 × 10-10 mbar. The ion optics assembly Ion packets are injected into the orbitrap with a duration much is pumped by two 70 L/s TMH-71P turbopumps and the orbitrap shorter than one axial oscillation, ensuring coherent motion. A by a 250 L/s TMU-262 turbopump (Pfeiffer Vakuum GmbH, deflector electrode near the entrance slit of the orbitrap is used Asslar, Germany). To avoid direct carryover of gas from the C-trap to compensate fringing fields. Detection of image current from to the orbitrap, ions are displaced in the vertical direction using coherent ion packets takes place after the voltage on the inner a dual electrostatic deflector.9 The short transfer distance reduces electrode has been stabilized at ∼3.5 kV. Signals from each of deleterious time-of-flight separation and thus minimizes differences the outer electrodes are amplified by a differential amplifier and in intensity distributions between mass spectra acquired with the transformed into a frequency spectrum by fast Fourier transforma- linear trap and orbitrap mass analyzers. tion. These frequencies relate to axial oscillations of ions along Ions are captured in the orbitrap by rapidly increasing the the orbitrap, which are independent of the energy and spatial electric field7,9,10 and gradually spread into rotating thin rings spread of the ions. The typical frequency for m/z ) 524 is ∼300 oscillating axially along the inner electrode. The inner diameter kHz. Single zero-filling and Kaiser-Bessel apodization13 is used of the orbitrap outer electrodes is 30 mm, and the maximum outer to improve the peak shape. The frequency spectrum is converted diameter of the inner electrode is 12 mm. Axial oscillations are into a using a two-point calibration and processed initiated by injecting ions at an offset of 7.5 mm relative to the (13) Goodner, K. L.; Milgram, K. E.; Williams, K. R.; Watson, C. H.; Eyler, J. R. equator of the orbitrap, eliminating the necessity of any excitation. J. Am. Soc. Mass Spectrom. 1998, 9, 1204-1212.

Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 2115 approximately a 0.25-0.3-s scan cycle time, wherein ion filling of the linear trap and ion transfer take a significant share of this time. Measured values of resolving power are always higher than the nominal value. Typical AGC target values N are 5000-30 000 for the linear trap detector and 20 000-2 000 000 for the orbitrap detector. With the use of an electrometer, it was determined that the number N is actually quite close (within factor of 2) to the real number of ions stored in the linear trap. Mass calibration coefficients are determined for four different AGC target values and interpolated for intermediate values. No magnitude-dependent corrections of m/z are made for data processing. Initial calibration of the instrument is performed using the standard LTQ calibration mixture with caffeine, the peptide MRFA, and Ultramark 1600 dissolved in 50:50 v/v. water/acetonitrile solution. Bradykinin, bovine serum albumin, horse heart apomyo- globin, horse cytochrome c, and carbonic anhydrase proteins were obtained from Sigma-Aldrich Chemie GmbH (Munich, Germany) and used without further purification. A Finnigan Surveyor Plus liquid chromatograph (Thermo Electron, San Jose, CA) fitted with a Hypersil C18 column (2.1-mm i.d., 5-µm particles, Thermo Electron) was used for LC/MS runs. Nanospray emitters from Proxeon (Odense, Denmark) were used for static nanospray experiments. PicoFrit columns (75-µm i.d., New Objective, Woburn, MA) were used for nano-LC/MS. During experiments, special care was taken handling caffeine, TFA, and acetonitrile according to the manufacturer’s guidelines.

RESULTS AND DISCUSSION Resolving Power. Figure 2 demonstrates wide-mass range, single-scan mass spectra acquired with different resolving powers. Intensity distributions are similar at the two resolving powers (R ) 7500 and R ) 100 000) and absolute abundances differ by <20%. This reflects the fact that resolving power is determined by the acquisition time and not by orbitrap imperfections such as inaccuracy of manufacturing or insufficient vacuum. It should be noted that the resolving power is also unaffected by the AGC target value. Because trapping in the orbitrap is electrostatic (i.e., no magnetic field and no rf potentials), the frequency of the axial oscillations is inversely proportional to the square root of m/z,7 in contrast to the cyclotron frequency in FTICR, which is inversely proportional to m/z. As a result, for a fixed acquisition time, the resolving power of the orbitrap mass analyzer diminishes as the square root of m/z, i.e., slower than in FTICR (though it should Figure 2. Wide-mass range, single-scan mass spectra of calibration be noted that the resolving power of the orbitrap remains lower mixture in the linear trap (a) and in the orbitrap at nominal resolving ∼ ∼ powers of (b) 7500 and (c) 100 000 at AGC target value N ) 100 000 than that of 7-T FTICR until m/z 900 and to m/z 2500 of 12-T with external calibration. Inset shows a minor doublet of two isoto- FTICR). 34 13 9 pomers ( S and C2) of MRFA peptide separated by a theoretical As previously observed, the resolving power of the orbitrap mass difference 0.0109 Da. diminishes with an increase of ion mass, even if m/z is unchanged. This effect has been attributed to collisions with background gas with Xcalibur software. Mass spectral data can be stored in full- that lead to fragmentation of ions and formation of noncoherent profile or reduced-profile format. In the latter, data below the packs of fragment ions. For a given m/z, the center-of-mass threshold of detection (i.e., close to the thermal noise of the collision energy with residual gas remains the same while the preamplifier) are removed to reduce the size of the dataset. collision cross section increases with mass, thus leading to faster The resolving power of the orbitrap (full width half-height) is scattering, fragmentation, and transient decay. This process is switched in discrete steps between the following nominal values more pronounced in the orbitrap than in FTICR because the ion at m/z ) 400: 100 000 (1.9-s scan cycle time), 60 000 (1-s scan energy is independent of m/z in the orbitrap, while in FTICR it cycle time), 30 000, 15 000 and 7500. The latter corresponds to decreases as (m/z)-1. Thus, ultrahigh vacuum is critical for high-

2116 Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 Figure 3. Measured and theoretical isotope distributions of proteins at setting R ) 100 000 (100 scans averaged, external calibration): (a) horse heart apomyoglobin, m ) 16 940.965 Da (monoisotopic, neutral), z )+10; (b) carbonic anhydrase, m ) 29 006.683 Da (monoisotopic, neutral), z )+21. resolution measurements of proteins. Figure 3 presents examples 6. The full-profile orbitrap spectrum shows the noise band of high-resolution spectra of higher-mass proteins acquired at the produced by thermal noise of the image current preamplifier. It same settings as Figure 2b. Although the resolution of these should be emphasized that linear trap detection employs electron proteins is inferior to that for singly charged ions of the same multipliers capable of single-ion detection while image current m/z (Figure 2), it remains sufficient to correctly resolve the detection of the orbitrap has an inherent noise band equivalent isotopic envelope and to provide good mass accuracy for both of to ∼20 ions (for 1-s acquisition length), as estimated using a the proteins (∼2.5-3.5 ppm). previously published method.14 Such a relatively small noise band Mass Accuracy. A two-point calibration is used to calculate was achieved by optimizing the preamplifier design as well as mass in the orbitrap, even though it is only an approximation for minimizing the length of signal wires. As linear trap noise peaks a more complex calibration law. This can be seen in Figure 4, are clearly single ions of chemical background, this could be used where mass traces for different ions (those in Figure 2) are drawn to estimate the total number of ions at m/z ) 524-527 in Figure as a function of AGC target number N with MRFA at m/z ) 524 6a: ∼1000-1500 ions total in 10 scans (or 100-150 ions in a single serving as an internal calibrant. It is clear that despite widely scan). Given that the ion transmission from the axis of the linear varying peak intensities within and between mass spectra, internal trap to its detectors is better than 50%,6 this brings the original calibration mass accuracy stays well within 2 ppm even for number of ions inside the linear trap to ∼200-300 for N ) 200. extremely large numbers of ions (i.e., N > 106). With even larger For the mass peak at mass m/z ) 524, the signal-to-noise ratio ion populations, space charge effects in the C-trap distort energy on the orbitrap is S/N ) 13 for 10 averaged scans, which results and spatial distributions to such an extent that this starts to affect in S/N ) 4 for a single spectrum. The average S/N ) 4inthe ion distribution and movement inside the orbitrap. orbitrap at noise band equivalent of ∼20 ions yields an estimation Figure 5 shows the long-term stability with external calibration, for an average number of detected ions of ∼100-120 (taking into which demonstrates that mass accuracy stays well within 5 ppm account isotopes). This indicates a transfer efficiency from the over 15 h. Variability is due to shot noise and thermal sensitivity linear trap to the C-trap and then to the orbitrap of ∼30-50% (with of the inner electrode voltage, which is greater than mass shifts the first number corresponding to m/z < 200 and the second to due to space charge effects. To optimize external calibration m/z > 1000), which is consistent with independent direct stability, the orbitrap and associated power supplies are thermally electrometer measurements at higher target numbers N (data not regulated. shown). Transmission and Detection Limit. Signals measured with To estimate detection limits, direct measurements of a low- the linear trap and orbitrap detectors were compared in order to concentration solution of bradykinin were carried out in static estimate the transmission. This comparison was carried out at a (14) Limbach, P. A.; Grosshans, P. B.; Marshall, A. G. Anal. Chem. 1993, 65, very low AGC target value N ) 200 using spectra shown in Figure 135-140.

Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 2117 Figure 6. Comparison of 10 averaged spectra for (a) the LTQ and (b) the orbitrap detectors for the same very low AGC target N ) 200 and a narrow mass range containing MRFA peptide (m/z ) 524.264 964) (R ) 60 000 in the orbitrap, full-profile mode, external mass calibration). Figure 4. Mass errors plotted for different m/z as a function of AGC target value N with the mass peak of MRFA peptide (m/z ) 524.264 964) used as an internal mass calibrant at R ) 30 000: (a) m/z ) 195.087 652, (b) m/z ) 1421.977 862, and (c) m/z ) 1721.958 701.

Figure 7. Mass spectrum of low-concentration bradykinin (charge state +2, theoretical monoisotopic m/z ) 530.787 98) on the LTQ Figure 5. Long-term mass stability for MRFA (m/z ) 524.264 964) orbitrap (3 amol consumed, 3 nM sample concentration, R ) 60 000, using external mass calibration (R ) 60 000, N ) 1 000 000, single N ) 5000, 1.16-s injection time and 2-s total scan cycle time, external scan acquired every 6 s over 15 h). mass calibration). nanospray mode using a nanospray ion source at 30 ( 10 nL/ life applications much more than other parameters (e.g., resolving min flow rate. Freshly prepared stock solutions of bradykinin at power). While detailed consideration is to be published elsewhere, 100 fmol/µL were serially diluted to a concentration 3 fmol/µL. Figure 8 provides an example of the dynamic range of mass Figure 7 shows a spectrum with ∼3 amol of bradykinin consumed accuracy of the LTQ orbitrap for real-life HPLC-MS analysis and over the duration of a 2-s scan (R ) 60 000). shows that mass accuracy is achieved in a single scan even for Dynamic Range. The dynamic range over which accurate mass peaks with up to 5000 times difference in abundances (peak mass measurements can be made is a key figure of merit as it at m/z ) 260 vs peak at m/z ) 610). The spectrum in Figure 8 actually determines the utility of accurate-mass capability for real- was acquired in the reduced profile mode.

2118 Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 Figure 8. Illustration of dynamic range of mass accuracy of the LTQ orbitrap (siloxane impurities in propanolol sample) in a single 1-s scan (R ) 60 000, N ) 2 × 106, external mass calibration, reduced profile mode).

Figure 9. Example of data-dependent acquisition with external mass calibration for a sample containing small molecules, with one high- resolution mass spectrum recorded of the precursors at R ) 60 000 and N ) 500 000 (a) followed by three data-dependent MS/MS spectra at R ) 7500, N ) 30 000, (b) for precursor at m/z ) 260, (c) for precursor at m/z ) 310, and (d) for precursor at m/z ) 386.

Data-Dependent Acquisition. Figure 9 is an example of rapid accuracy is not sacrificed so elemental composition can be data-dependent acquisition for a sample containing small mol- unambiguously determined in all cases. ecules using a fast gradient (3 min from 100% A to 100% B, where A is water with 0.1% TFA and B is acetonitrile with 0.1% TFA) LC CONCLUSION separation on a Hypersil C18 column (2.1-mm i.d., 5-µm particles). The main performance parameters of a novel linear ion trap/ A mass spectrum at nominal resolving power of 60 000 (scan cycle orbitrap hybrid mass spectrometer have been characterized. A time 1 s) is followed by three rapid data-dependent MS/MS novel approach to ion storage and injection into the orbitrap allows acquisitions at nominal R ) 7500 at a rate of 2.5 spectra/s. Fast high resolving power, mass accuracy, and transmission over a acquisition of MS/MS spectra is possible because of the high wide dynamic range and forms the basis for a hybrid mass transmission between the linear trap and the orbitrap, which spectrometer combining these analytical parameters with the MSn allows shorter ion accumulation times than in FTICR and TOF. capability of the linear ion trap mass spectrometer. Utilizing short Despite lower resolving power settings in MS/MS spectra, mass fill times of the linear trap and relatively low number of ions for

Analytical Chemistry, Vol. 78, No. 7, April 1, 2006 2119 analysis in the orbitrap, this hybrid is capable of providing rapid, (Bremen) GmbH, Dr. Mike Senko, Mark Hardman, Mike Anton- accurate-mass MSn analysis of complex mixtures with external czak, Dr. Eric Hemenway of Thermo (San Jose), Dr. Steve Davis, calibration. Robert Lawther, and Andrew Hoffmann of HD Technologies Ltd. and later Thermo Masslab Ltd. The authors thank Dr. Mike Senko ACKNOWLEDGMENT for valuable comments on the manuscript. The authors express their deep gratitude to colleagues who made invaluable contribution to the development of this instru- ment: Dr. Reinhold Pesch, Dr. Gerhard Jung, Frank Czemper, Received for review October 20, 2005. Accepted February Oliver Hengelbrock, Silke Strube, Dr. Hans Pfaff, Dr. Torsten 1, 2006. Ueckert, Ralf-Achim Purrmann, Juergen Srega of Thermo Electron AC0518811

2120 Analytical Chemistry, Vol. 78, No. 7, April 1, 2006