A "screened" electrostatic ion trap for enhanced mass resolution, mass

Shrink-wrapping an ion cloud for high-performance Fourier transform ion cyclotron ..... Michael L. Easterling , Cynthia C. Pitsenberger , Shubhada S. ...
1 downloads 0 Views 1MB Size
Download PDF
Anal.

1288

Chem. 1989, 61. 1288-1293

A "Screened" Electrostatic Ion Trap for Enhanced Mass Resolution, Mass Accuracy, Reproducibility, and Upper Mass Limit in Fourier Transform Ion Cyclotron Resonance Mass Spectrometry Mingda Wang and Alan G. Marshall*J

Department of Chemistry, The Ohio State Uniuersity, 120 West 18th Auenue, Columbus, Ohio 43210

Until now, H was thought that the optimal static eiectromagnetic ion trap lor Fourier translorm ion cyclotron resonance (FT-ICR) mass spectrometry should be designed to produce a quadrupolar electrical potential, lor which the ion cyclotron frequency is independent of the ion's preexcitation location within the trap. However, a quadrupoiar potential resuits in a transverse (to the magnetic field) electric field that Increases linearly with distance from the center of the trap. That radially linear electric field shifts the observed ICR frequency, increases the I C R orbHal radius, and ultimately limns the highest mass-to-charge ratio ion that can be contained within the trap. I n this paper, we propose a new static electromagnetic ion "trap" in which grounded screens placed just inside the usual "trapping" plates produce a good approximation to a "particle-in-a-box" potential (rather than the quadrupoiar "harmonic oscillator" potential). SIMION calcuialions conlirm that the electric potential of the screened trap is near zero almost everywhere within the trap. For our screened orthorhombic (2.5 in. X 2 in. X 2 in.) trap, the experimental ICR frequency shin due to trapping voltage is 100, and the experimental variation reduced by a factor 01 of ICR frequency with ICR radius is reduced by a factor of -10 compared to a conventional (unscreened) 2-in. cubic ion trap. The new "screened" trap therefore offers higher mass resolution during defecllon(due to more uniform electric lieid) and thus higher mass measurement preclsion and accuracy: higher m a s resolution during excllafion (since ICR frequency no longer varies significantly with ICR orbital radius), for improved MSlMS and greater ease in tune-up and reproducibiiny: and potentially much higher upper mass limit (since the radial electric field has been eliminated) than obtained with prior electromagnetic ion traps.

-

INTRODUCTION In Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry (1-161, ions are trapped hy static magnetic and electric fields produced by any of several possible geometrical arrangements (see Figure 1). In the limit that the electric field is zero everywhere inside the trap and the magnetic field is spatially uniform, the trajectory of an ion in the x-y plane perpendicular to the applied magnetic field direction (see Figure la) follows a circular orbit with a natural "cyclotron frequency" w. =

qBdm

(1)

in which w, is the cyclotron frequency (radians per second), Bo is the applied magnetic field strength (tesla), q is ionic

*Towhom correspondence should he addressed. 'Also a member of the

Department of Biochemistry. 0003-2700/89/0361-1200$01.50/0

Figure 1. Static electromagnetic ion traps for FT-ICR mass spectrometry. Bodenotes the (static)magnetic fieid direction. The excitation, detection, trap, and screen electrodes are designated by E. D, T, and S, respectively. The three traps shown are (a) cubic trap, (b) hyperbolic trap. and (c) screened trap.

charge (coulombs), and m is ionic mass (kilograms). Although a static magnetic field applied along the z direction constrains ion motion in the x-y plane, ions are free to move in the magnetic field (z axis) direction. Therefore, an additional electrostatic "trapping" potential, produced by applying a static voltage (-0.5 V/cm) to the "end cap" ("trap") plates in Figure 1, is needed to provide a restoring z force to prevent ions from escaping along the z direction. Unfortunately, Gauss' law requires that there he a radially outward electric field to balance the "trapping" electric z field, in order that no net charge he contained in the trap in the absence of ions. For the ion traps shown in Figure la,h, the radial electric field opposes the inward-directed Lorentz force (9v X B, in which v is ion velocity and B is the magnetic field). The radial electric field thus has the same effect as a decrease in magnetic field strength, thereby decreasing the ICR frequency of eq 1 ( I 7, 18). Moreover, if the radial electric field varies nonlinearly with radial distance from the center of the trap, then the ICR frequency will vary with the preexcitation position of the ion in the trap (19-23), thereby rendering tune-up more difficult and limiting the ultimate mass resolution available during the excitation event, particularly in the first stage of MS/MS experiments (10). Finally, even for a perfectly quadrupolar field (i.e., electric field varies linearly with x, y . or z distance from the center of the trap), the radial electric field acts to limit the highest mi9 ion that can he held in the trap (24). The first FT-ICR experiment (25)was done in a cubic trap (26), and the method was subsequently extended to several other ion trap geometries (orthorhombic (29, cylindrical (281, hyperbolic (29-311, multiannulus trapping plates (32,331and multiple-section (24) traps), in attempts to reduce the above-listed undesirable effects of the trapping z potential. 0 1989 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 61, NO. 11, JUNE 1, 1989

Figure l a shows a cubic trap consisting of three orthogonal pairs of flat electrodes (26). The two end-cap (“trap”) plates are orthogonal to the magnetic field ( z axis) direction. The static electric potential in the cell has D& symmetry. A typical orthorhombic trap also has a square cross section but is elongated along the z axis. Although cubic (or orthorhombic) traps have the advantage of conceptual simplicity and ease of construction, their electrostatic field description is mathematically quite complicated. The electric field in such a trap has D4h symmetry and is represented by an infinite series solution of Laplace’s equation. Provided that the origin of the Cartesian coordinate system is chosen to lie at the center of the cell, expansion of the series in the spatial region near the center of the cell yields an approximately quadrupolar electric potential (17) V(X,Y,Z) =

3 2 + v*(-.(

;y .( ;y + ;y 2 4

-

-

.)

(2)

in which a is the distance between the two end-cap (“trap”) electrodes, V , is the trapping voltage applied to the trap electrodes, and a and y are simple functions of the cell dimensions. The equation of motion of an ion of mass m and charge q in a static electromagnetic ion trap (see Figure 1)is given by dv

m- = q(E + v dt

X

B)

(3)

in which the (uniform) magnetic field direction is assigned to the z axis. In the quadrupolar approximation, ion motion in the z direction is separable from motion in the x-y plane in eq 3. There are two eigenfrequencies for x-y motion: the so-called “cyclotron” and “magnetron” frequencies, w, and w, (17, 27, 34-37) a , = -I( a,+

2

(

w,2--

(4)

ma2 Equation 5 shows that the observed ICR frequency, w,, varies with trapping potential, VT In order to reduce the ICR frequency shift induced by the trapping potential, Hunter et al. (27)introduced an elongated (along the z axis) rectangular trap in order to reduce a (at least near the center of the trap) in eq 5 . However, for the elongated trap, the quadrupolar approximation breaks down except near the center of the trap, so that ICR frequency still varies with preexcitation position of the ion within the trap. Lee et al. (38) designed a “cylindrical” trap that has cylabout the z axis and thus has the indrical symmetry (Dmh) same symmetry as a quadrupolar potential in the radial direction. However, the use of flat trapping plates again leads to a spatially inhomogeneous electrostatic field. The quadrupolar electric potential is again approached only near the center of the cylindrical trap (28). Therefore, ion cyclotron frequency again varies with ion preexcitation position in the trap. The ion trap geometry that most closely approximates a quadrupolar electrostatic potential is the “hyperbolic” trap shown in Figure lb. Two trap electrodes are separated by a ring electrode which is cut lengthwise into quadrants (31). Each of the three electrode surfaces has the shape of a hyperboloid of revolution. The hyperbolic trap produces a near-perfect quadrupolar electric potential within the trap

1289

in which 2ro and 22, are the radial and z dimensions of the trap. The ICR frequency for such a trap is given by

m(rO2+ 2zO2) Although ICR frequency is invariant with ion position in a perfect quadrupolar potential, eq 7 shows that ICR frequency still varies with trapping voltage. However, because the actual hyperbolic electrodes are not infinite in extent, the actual hyperbolic trap of Figure l b does not generate a purely quadrupolar electrostatic potential, and the observed ICR frequency still varies somewhat with ICR orbital radius (31). In attempts to minimize the ICR frequency shift and sidebands resulting mainly from nonzero x-y components of the inhomogeneous electrostatic field, two groups of investigators have proposed “compensated trap electrodes divided into annular segments held at different potential. In the Yang et al. design (32),the segments are coplanar, whereas Naito et al. (33)separated the segments along the z axis for ease in construction. For either trap, the radial component of the electric field is reduced without loss in ion trapping efficiency. In addition, Naito et al. have shown that the magnetron frequency shift is also reduced by a factor of - 5 . In yet another approach, Grosshans et al. (24)proposed and tested an elongated ( 6 1 aspect ratio) three-section cell. If +1 V is applied to each of the two end cap (“trap”)electrodes, the electric potential drops to less than 10 pV in the center of such a trap. Thus, radial electric field-induced ICR frequency shifts can be reduced accordingly if ICR detection is limited to the central electric field-free region of the threesection trap. Mass resolution is also improved by a factor of -2. However, ions nevertheless oscillate back and forth along the z axis and in fact spend most of their time in the two end sections, in which the electric field again has major radial components (leading to radial loss of high-mass ions). Furthermore, as in the other above-mentioned trap designs, if ions are distributed nonuniformly in the trap before excitation, then ions of different z amplitude will dephase with respect to each other, leading to inhomogeneous line-broadening. In this paper, we propose and demonstrate the performance of a conventional orthorhombic trap, in which a grounded screen is placed just inside each end cap (“trap’’) electrode in an otherwise conventional orthrhombic trap, as shown in Figure IC. We then compute and map the electrostatic potential, which closely approaches a “particle-in-a-box”along the z axis. We show that the electrostatic potential is near zero until ions approach to within about one mesh width of the screen. Finally, we report direct experimental FT-ICR comparisons of the variation of ICR frequency with either trapping voltage or ICR orbital radius between the “screened” and cubic traps. Cyclotron and magnetron frequency shifts are largely eliminated (see below).

THEORY Electrostatic Potential in an Ion Trap. If the “space charge” generated by the ion ensemble itself is neglected, then the electrostatic potential, V(x,y,z),within a static electromagnetic ion trap may be determined by solving Laplace’s equation v2

V(x,y,z) = 0

(8)

subject to the boundary conditions corresponding to the voltages applied to electrodes arranged as shown in Figure 1. Laplace’s equation may be solved numerically by converting

1290

ANALYTICAL CHEMISTRY, VOL. 61, NO. 11, JUNE 1, 1989

1v I

I

I

1

4

I

h ,

I

I

ov

I

Figure 2. Arrangement of grid points for evaluation of the electrostatic potential from the finite difference form of Laplace's differential equation in two dimensions, for a Cartesian coordinate frame. The potential at grid point zero is computed from the arithmetic average of the potential values of grid points 1, 2, 3, and 4 (see text). the original differential equation to a set of finite difference equations (39). For example, in order to solve Laplace's equation (in Cartesian coordinates) in two dimensions, the two-dimensional surface is divided into a grid with spatial period, h, as shown in Figure 2. If the electrostatic potentials a t points 0, 1, 2, 3, 4,... are V,, VI, V,, V3, V4, ..., and the grid period, h, is sufficiently small, then difference equations (eq 9) provide a good approximation to the continuous first derivatives of the potential a t the point, 0

av

- = lim (;(VI ax

ax

h-O

= lim h-0

- Vo)) (approached from t h e right)

(

i ( V o - V 3 ) )(approached from t h e left) (9b)

av = (:(V2 - Vo)) h ay

lirn h-O

(approached from t h e top) (9c)

dV = lim (i ( V o- V4)) (approached from t h e bottom) ay h-O (9d) Similarly, the difference equations corresponding to the second derivatives of the potential at point, 0, become

- Vo)- (V, - V,)] -

V,) - (V, - V,)]

Substitution of eq 10 into eq 8, followed by some rearrangement, gives

1 v, = -(VI + v, + v3 + V4) 4

(11)

In other words, the electrostatic potential a t any given point of the grid is simply the arithmetic average of the potentials a t the nearest-neighbor grid points: the smaller the grid period, h, the more precise is eq 11. The SIMON program (40) used in this work computes the potential a t specified grid points (by the finite difference method outlined above) iteratively until a self-consistent array of grid point potentials is found (Figure 3). A major limitation of SIMION is that although it can determine the potential at each of the points of a three-dimensional spatial grid, the potential boundary conditions may be specified in only two perpendicular directions. Thus, SIMION is quantitatively accurate only for

ov

1v

ov Figure 3. Two-dimensional contours of constant electrostatic potential for (top)orthorhombic trap and (bottom)screened orthorhombic trap of the same dimensions. The contours were computed from SIMON, for which the spacing between one pair of side plates must be infinite (see text). Note that the electrostatic potential drops rapidly to near zero just inside the screens. cylindrically symmetric electrode arrangements or for geometries in which the electrodes separated along one axis are infinitely far apart. Nevertheless, SIMION is suitable for qualitative and/or semiquantitative analysis, as in the present work.

EXPERIMENTAL SECTION A screened orthorhombic ion trap, 2.5 in. (screen-to-screen separation) x 2.0 in. X 2.0 in., was constructed from flat solid oxygen-free hard copper electrodes separated by Macor spacers. Each screen electrode was constructed from 0.0015 in. diameter tungsten 50 X 50 mesh (Unique Wire Weaving Co., Inc., Hillside, NJ); two out of every three wires were then removed to give a final mesh spacing of 16 per inch. Each trap electrode was placed 1/4 in. outside its adjoining screen electrode. The mesh was held in place by spot-welds to a 314 stainless steel frame. FT-ICR mass spectra were produced with a Nicolet Fl'MS-1000 instrument operating at a magnetic field strength of 3.058 T (1 in. cubic and 2 in. screened traps) and a Nicolet FTMS-2000 instrument at 3.003 T (2 in. cubic trap). Benzene was introduced through a Varian No. 951-5100 leak valve to a pressure of (1.2-1.5) X Torr. C6H6+ions were produced by impact of an electron beam (50 V for 20-40 ms at an emission currsnt of 60-100 nA measured at a collector located outside one of the trapping plates). For the mass calibration data in Table I, spectral peak frequencies were determined by parabolic three-point approximation and fitted to a mass calibration equation (34) of the form, m = ( A / v )+ ( B / v 2 ) ,with v2 and relative peak height weighting. The calculated mass for each singly charged positive ion was computed by subtracting the mass of an electron from the combined mass of the most abundant isotopes. Ions were produced by electron ionization (50 ms at 50 eV at a detected emission current of 50 nA). The spectrum was excited by frequency-sweep from dc to 715 kHz (corresponding to m/z = 68.79) in 3.58 ms at a radio frequency amplitude of 30 V (peak to peak), and acquired in heterodyne-mode with a reference frequency of 705.96 kHz for 53.47 ms to give 64K time-domain data points, to which another 64K zeros were added before discrete Fourier transformation. No apodization (windowing)was applied. For determination of ICR frequency shift as a function of trapping voltage (Figure 4), each time domain signal was produced

ANALYTICAL CHEMISTRY, VOL. 61, NO. 11, JUNE 1, 1989

Table I. Fourier Transform ICR Mass Calibration over Approximately 1 Decade in Mass for Perfluorotri-n -butylamine (9 X 1O-O Torr)O ICR frequenion

cy,

Hz

CF3+

680 642.69 394 651.03 358496.37 C,F9+ 214438.89 C6FloNt 177881.48 C8F16Nt 113427.71 C9FzoN+ 93 542.06

C85+ C85+

true mass, p

measured mass, p

68.994 66 118.99147 130.99147 218.98508 263.986 56 413.97698 501.97060

68.994 67 118.99146 130.99144 218.98490 263.986 60 413.977 24 501.971 12

error, ppm +0.1 -0.1 -0.2

-0.8 +0.2 +0.6

+1.0

Produced by a Nicolet FTMS-1000 instrument operated at 3.058 T with a screened orthorhombic (2.5 in. X 2.0 in. square cross-section)ion traD. ICR Frequency Shift OF c6H6+ (kHz)

1291

sidered to be infinitely far apart in this computation. The electrostatic potential varies approximately quadratically with z distance (at zero x distance) from the center of the orthorhombic trap. For the screened trap, Figure 3 clearly shows that the z-axis trapping potential drops rapidly to near zero as one moves away from the screen toward the center of the trap. Beyond about one mesh diameter away from the screen, the electrostatic potential in the screened trap is about 15-20 times smaller than at the corresponding z position in the unscreened orthorhombic trap. Similar results (not shown) apply for a cylindrical trap. ICR Frequency Variation with Trapping Voltage. The dependence of observed ICR frequency on trapping voltage was given in eq 4, which may be approximated as follows: w o = - 1( w c +

(

w,2-- 8 y a y 2 )

2 2qavT =q--

ma2 2av, =%--

2" Cubic Trap

-3 0

2

4

6

8

10

12

Trapping Voltage (V) Figure 4. Experimental ICR frequency shift (of C,H6+ at 3 T) with trapping voltage in 1 in. cubic, 2 in. cubic, and 2.5 in. X 2 X 2 screened orthorhombic ion traps. The grounded screens reduce the frequency shift by a factor of 100.

-

by single-frequency on-resonance excitation of 30 V (peak-to-peak) amplitude at 601.703 kHz for 250 ps (screened trap) or 60 ps (1 in. cubic trap), and detected in heterodyne mode at a bandwidth of 62.992 kHz (16K time-domain data, padded with another 16K of zeros before discrete FT).Trapping dc potential was varied from 0.5 to 10 V. For determination of ICR frequency shift as a function of ICR orbital radius (Figure 5), each time-domain signal was again excited by single-frequency on-resonance excitation for 310 bs (2 in. screened trap) or 97 ps (1 in. cubic trap) and detected in heterodyne mode at a bandwidth of 17.582 kHz (16K time-domain data padded with another 16K zeros before discrete FT). Trapping voltage was maintained at 1V dc and radio frequency excitation magnitude adjusted from 0.2 to 30 V (peak to peak) by variable attenuation. An ultrahigh resolution (m/Am= 6.08 X lo6) mass spectrum Torr, of HzO+was obtained at a neutral pressure of 1.2 X by means of single-frequencyon-resonance excitation (30 V (peak to peak) at 2.607 MHz for 260 ps), and detected in heterodyne mode at a bandwidth of 1.689 kHz (16K time-domain data padded with 48K zeros) before discrete FT).

RESULTS AND DISCUSSION Electrostatic Potential Maps. Figure 3 presents SIMION (40) contour maps of the two-dimensional x-z electrostatic potential of orthorhombic and screened ion traps of the same dimensions. As noted above, the two y electrodes are con-

a2Bo

in which VT is the trapping voltage (see below). From the derivative of eq 12 with respect to VT, we obtain the variation of ICR frequency with trapping voltage, VT

Equation 13 predicts that the frequency shift, av,/aVT, in a 1-in. cubic trap is -223 Hz/V, in excellent agreement with the experimental result of 231 Hz/V obtained from a plot of ICR frequency of c6H6+at 3.0 T vs trapping voltage shown in Figure 4. Thus, the electrostatic potential in the cubic trap is quadrupolar to a good approximation. Similarly, the %in. cross-section cubic trap gives a smaller (but still substantial) experimental ICR frequency shift of 64.6 Hz/V, again in reasonable agreement with the value of 56.8 Hz/V computed from eq 13 at the slightly lower magnetic field strength (3.003 T) for those measurements. In contrast, the 2-in. cross-section screened trap gives a frequency shift of only about 0.67 Hz/V, or about 100 times smaller than that for an unscreened trap of the same approximate dimensions. The screens therefore effectively reduce the static electric field in the trap by a very large factor, except at z positions within about one mesh diameter of either screen. Since the main source of deviation of ICR frequency from the cyclotron frequency of eq 1is the shift induced by trapping potential, the screened trap should improve the accuracy of mass measurement in FT-ICR mass spectrometry, because the mass calibration equation will require smaller correction terms. For example, the mass calibration obtained with the 2-in. cross-section orthorhombic screened trap (Table I) is better (by a factor of 2 or more) than we have obtained for any trap size or shape. (It is difficult to compare different traps quantitatively, since the optimal operating parameters are generally different for different traps.) ICR Frequency vs ICR Orbital Radius. In a cubic or cylindrical trap, the quadrupolar electric potential approximation is accurate only near the center of the trap. As a result, the detected ICR frequency varies with ICR orbital radius. Because the addition of grounded screens reduces the electric field magnitude everywhere in the trap (except in the immediate vicinity of the screens), the radial component of electric field in particular is also reduced. Because the screens

1292

ANALYTICAL CHEMISTRY, VOL. 61, NO. 11, JUNE 1, 1989

ICR Frequency of C&

(KHz)

+

601.8,

I

601.7

The first term in the denominator of eq 14 is generally negligible. For example, if we seek the highest ion mass for which 95 % of room-temperature ( T = 300 K) ions are ejected before the excitation event, the first term of the denominator of eq 14 contributes only -0.0053 eV. Thus, to a good approximation, the upper mass limit becomes

2” Screened Trap

I

i

601.6

601.5

{

jy-? I

1” Cubic Trap

601.4 1

0

1000 2000 3000 4000 5000

Figure 5. 1CR frequency of &He+ as a function of V,7/a1 in which V , is the trapping voltage, 7 is excitation period (microseconds), and

a is the separation between the two excitation plates (inches). In each case, ICR orbital radius is directly proportional to V , r / a , so that the graph may be viewed as a plot of ICR frequency versus ICR orbital radius. reduce the average electric field magnitude by a factor of -15-20 (see Figure 3), the variation of ICR frequency with ICR orbital radius is also reduced by a factor of 10-20 (see Figure 5). In fact, the variation of ICR frequency with ICR orbital radius is smaller for the screened orthorhombic trap than for a “hyperbolic” trap of similar dimensions (31) presumably because the finite extent of the electrodes of the hyperbolic trap distorts its potential from a purely quadrupolar shape. The screened trap performance shown in Figure 5 is important for two reasons. First, mass accuracy (e.g., Table I) should now be more reproducible, since a variation of ion orbital radius (arising from, e.g., variation in ICR excitation conditions) from one experiment to another should have much less effect on measured ICR frequencies. Second, FT-ICR mass resolution during the detection period has until now been much higher than mass resolution during the excitation period, because ICR detection is conducted at a fixed ICR orbital radius, whereas ICR orbital radius must necessarily increase during excitation. Thus, any variation in ICR frequency with ICR orbital radius must reduce the maximal mass resolution during excitation. Stored waveform inverse Fourier transform (SWIFT) excitation (reviewed in ref 10) offers optimal mass selectivity for MS/MS and other single- or multiple-ion excitation experiments but is ultimately resolution-limited by any ICR frequency shift during excitation. With the screened trap, FT-ICR mass resolution during excitation (e.g., for the first stage of an MS/MS experiment) should now approach the spectacularly high mass resolution already demonstrated during ICR detection (e.g., for the second stage of an MS/MS experiment). Upper Mass Limit. For the last several years, analysis of large (especially biological) molecules by FT-ICR mass spectrometry (12) has been a prime experimental goal. For an ion ensemble described by a Boltzmann distribution, the upper mass limit above which a fraction, K , of ions has ICR orbital radius larger than the trap dimension is given by (24)

-

Therefore, since the screened trap reduces the effective trapping voltage by more than an order of magnitude, it is reasonable to expect the upper mass limit for a screened trap to increase by a factor of 110 compared to an unscreened trap of the same dimensions. Testing that prediction is nontrivial, since ions in an FT-ICR experiment are not monoergic. The theory and experimental results of such tests will be reported separately (41). It is important to note that even in the limit of zero trapping potential, the upper mass limit will still be finite, because the first term in the denominator of eq 14 takes over when the second term drops below -0.01 eV. Ion Loss and Mass Resolution. In the limit that the time-domain acquisition period is much longer than the damping constant, 7, for exponential decay of the ICR time-domain signal, FT-ICR, mass resolution, m/Am, increases linearly with T (42) m / A m = 1.81uor

(16)

in which Am is defined as the full magnitude-mode peak width at half-maximum peak height and uo = w o / 2 r is the cyclotron frequency in hertz. In a screened trap, it is possible for ions to collide with screen electrodes and to be lost from the cell. Since any loss of ions accelerates time-domain signal decay, such ion loss could in principle degrade mass resolution. However, our screened orthorhombic trap has produced an FT/ICR mass spectrum of HzOf with mass resolution as high as 6.08 X lo6 in heterodyne mode at a bandwidth of 1.689 kHz, acquisition time 4.857 s, and sample pressure 1.2 X lo* Torr. Therefore, ion losses due to ion-screen collisions do not appear to be significant. Removal of Ions after an Experimental Event Sequence. In order to minimize hardware and software modifications to our instrument, we permanently grounded both screen electrodes. As a result, the “quench” event, in which one trap plate is set to +10 V and the other to -10 V, could in principle be less effective for z ejection of ions, because the screens largely shield the ions from the “quench” voltage pulse. However, experimental FT-ICR mass spectra obtained at quench periods up to 10-100 ms were virtually the same as for a (default) quench period of 200 fis. Therefore, the quench process appears successful even in the presence of the grounded screens. If quenching were insufficient, one could gate the voltage to the screens, so that the quench pulse is applied to the screen electrodes rather than to the trap electrodes.

ACKNOWLEDGMENT The authors wish to acknowledge P. B. Grosshans for independent development of a similar idea. In addition, we thank John Lux and Jerry Hoff for their collaboration in the design and construction of the screened trap, P. B. Grosshans for helpful discussions on the upper mass limit, and Dr. A. Appelhans for helping us to make use of the latest SIMION versions. LITERATURE CITED (1) Marshall, A. G. Acc. Cbem. Res. 1985, 18, 316-322. (2) Marshall, A. G. Adv. Mass Spectrom. 1989, 11. 651-669. (3) Nibbering, N. M. M. A&. Mass Spectrom. 1989, 1 1 , 101-125. (4) Hanson. C. D.; Kerley, E. L.; Russell. D. H. I n Treatise on Analythl Chemistry. 2nd ed.; Wiley: New York, 1988; Vol. 11, Chapter 2.

Anal. Chem. 1989, 61, 1293-1295 (5) Freiser, B. S. I n Techniques for the Study of Ion Molecule Reactions; Farrar, J. M., Saunders. W., Jr., Eds.; Wiley: New York, 1988. (6) Cody, R. B.; Bjarnason, A,; Weil, D. A. In Lasers in Mass Spectrometry; Lubman, D. A.. Ed.; Oxford University Press: New York, 1989; Chapter 14. (7) Nibbering, N. M. M. Adv. Phys. Org. Chem. 1988, 24, 1-55. (8) Asamoto. B. Spectroscopy 1988. 3 , 38-46. (9) Buchanan, M. V.; Comisarow, M. B. I n Fourier Transform Mass Spectrometry: Evolution, Innovation, and Applications ; ACS Symposium Series 359; Buchanan, M. V., Ed.; American Chemical Society: Washington, DC. 1987; pp 1-20. (10) Marshall, A. G.; Wang, T.-C. L.; Chen. L.; Ricca. T. L., In Fourier Transform Mass Spectrometry : Evolution, Innovation, and Applica tions; ACS Symposium Series 359; Buchanan. M. V., Ed.; American Chemical Society: Washington, DC, 1987; pp 21-33. (11) Laude, D. A., Jr.; Johlman, C. L.; Brown, R . S.; Weil, D. A,; Wilkins, C. L. Mass Spectrom. Rev. 1988, 5. 107-166. (12) Russell, D. H. Mass Spectrom. Rev. 1986. 5 , 167-189. (13) Comisarow, M. B. Anal. Chim. Acta 1985, 178, 1-15. (14) Freiser. B. S. Talanta 1985, 3 2 , 697-708. (15) Gross, M. L.: Rempel, D. L. Science 1984, 226, 261-268. (16) Wanczek, K.-P. Int. J. Mass Spectrom. Ion Processes 1984, 6 0 , 11-60. (17) Sharp, T. E.; Eyler, J. R.; Li. E. Int. J. Mass Spectrom. Ion Phys. 1972, 9 , 421-439. (18) Dunbar, R. C.; Chen, J. H.; Hays, J. D. Int. J. Mass Spectrom. Ion Processes 1984, 57,39-56. (19) Woods, I . 8.; Riggin, M.; Knott, T. F.; Bloom, M. Int. J. Mass Spectrom. Ion Phys. 1973, 12, 341-346. (20) Knott, T. F.; Bloom, M. Can. J. Phys. 1974, 52,426-435. (21) Bloom, M.; Riggin, M. Can. J. Phys. 1974, 5 2 , 436-455. (22) Riggin, M.; Woods, I.B. Can. J. Phys. 1974, 52,456-469. (23) Hartmann, H.; Chung, K. M.; Baykut, G.; Wanczek. K.-P. J. Chem. Phys. 1983, 78,424-431. (24) Grosshans, P. B.; Wang, M.; Marshall, A. G. 36th Am. SOC. Mass Spectrom. Annu. Conf. Mass Spectrom. Allied Topics 1988, 592-593. (25) Comisarow, M. B.; Marshall, A. G. Chem. Phys. Lett. 1974, 25. 282-283.

-

1293

(26) Comisarow, M. B. Int. J. Mass Spectrom. Ion Phys. 1981, 37, 251-257. (27) Hunter, R. L.; Sherman, M. G.; McIver, R. T., Jr. Int. J. Mass Spectrom. Ion Phys . 1983, 5 0 , 259-274. (28) Kofel, P.; Allemann, M.; Kellerhals, Hp.; Wanczek, K.-P. Int. J. Mass Spectrom. Ion Proc. 1986, 74,1-12. (29) Van Dyck, R. S.,Jr.; Schwinberg, P. B. Phys. Rev. Lett. 1981, 47, 395-398. (30) Rempel. D. L.; Ledford, E. B., Jr.; Huang, S. K.; Gross, M. L. Anal. Chem. 1987, 59, 2527-2532. (31) Wang, M.; Ledford, E. B., Jr.; Marshall, A. G. FACSS XIV Annual Mtg. (Detroit, MI, Oct., 1987), Abstract #43. (32) Yang, S. S.; Rempel, D. L.; Gross, M. L. 36th Am. SOC.Mass Spectrom. Annu. Conf. Mass Spectrom. Allied Topics 1988, 586-587. (33) Naito, Y.; Inoue. M. 36th Am. SOC. Mass Spectrom . Annu. Conf. Mass Spectrom. Allied Topics 1988, 608-609. (34) Ledford, E. B., Jr.; Rempel, D. L.; Gross, M. L. Anal. Chem. 1984, 56, 2744-2748. (35) Wobschall, D.; Graham, J. R., Jr.; Malone, D. P. Phys. Rev. 1963, 131, 1565-1571. (36) Beauchamp, J. L. J. Chem. Phys. 1966, 46, 1231-1243. (37) Byrne, J.; Farago, P. S.Proc. Phys. SOC.London 1985, 86, 801-815. (38) Lee, S. H.; Wanczek, K.-P.; Hartmann, H. Adv. Mass Spectrom. 1980, 88, 1645. (39) Shortley, G. H.; Weller, R. J. Appl. Phys. 1938, 9 , 334-348. (40) Dahl, D. A.; Delmore. J. E. SIMION PC/PS2 Version 4.0; Idaho National Engineering Laboratory (EGGCS-7233, Rev. 2, April, 1988). (41) Grosshans, P. B.; Wang, M.; Marshall, A. G., unpublished results. (42) Marshall, A. G.; Comisarow, M. B.; Parisod, G. J. Chem. Ph,vs. 1979. 71.4434-4444.

RECEIVED for review November 4,1988. Accepted March 10, 1989. This work was supported by grants (to A.G.M.) from the U.S.A. Public Health Service (N.I.H. GM-31683), the National Science Foundation (CHE-87-21498), Nicolet Instrument corporation, and The Ohio State University.

CORRESPONDENCE Fast Atom Bombardment Mass Spectrometry of Some Nonpolar Compounds Sir: Fast atom bombardment mass spectrometry (FABMS) (I) has been widely used for the analysis of polar organic molecules (2),biomolecules of high molecular weight ( 3 ) ,and inorganic and organometallic compounds ( 4 , 5 ) . The use of FAB for the characterization of nonpolar compounds has been limited because of their low solubility in many of the readily available matrices and the low success rates of some of the early studies in glycerol matrices. Some success has been achieved recently in obtaining FAB spectra of nonpolar compounds by using a technique known as electrochemically assisted fast atom bombardment (6). FAB spectra of nonpolar compounds have also been reported without the use of the matrix (7). We have been able to obtain fairly good positive and negative ion FAB spectra for relatively nonpolar compounds such as nitroaromatics (8,9) by using 3-nitrobenzyl alcohol (NBA) as a matrix (IO). This matrix is also very useful for organometallic and coordination compounds as well (11). In this work, we have deomonstrated that NBA may be the matrix of choice for nonpolar compounds. We were able to obtain excellent FAB mass spectra for compounds such as vitamin K1, 1,3,5-trimethoxybenzene, 1-bromohexadecane, and other nonpolar compounds of the type studied by EFAB (6). Nitroaromatic compounds were also shown to give reasonably good FAB spectra in glycerol, when they were sufficiently well dissolved by using a cosolvent such as dimethyl sulfoxide (DMSO). 0003-2700/89/0361-1293$01.50/0

EXPERIMENTAL SECTION FAB mass spectra were obtained on a VG ZAB-E located at the McMaster Regional Mass Spectrometry Facility, using unmodified ZAB single and dual target probe tips, and on our own Kratos MS-30, modified with a Kratos FAB retrofit (12) plus our own specially designed probe tips. Instead of the standard 3.2 mm diameter stainless steel tip, these have a diameter of 4.3 mm and are divided into halves with a 3 mm deep cut, 0.1 mm wide, such that compounds could be placed on either half without being mixed (13,14). The negative ion FAB mass spectra were obtained on the ZAB-E. All of the chemicals were purchased from Aldrich and were used without further purification. Samples for the nitroaromatic compounds were prepared by dissolving the compound in DMSO (-15 mg/mL) and mixing them with glycerol (1:1 ratio by volume) or by directly dissolving them in NBA or nitrophenyl octyl ether (NPOE).

RESULTS AND DISCUSSION Figure 1 shows the positive ion FAB spectra of 1,3,5-trimethoxybenzene (TMB) and 1-bromohexadecane (BHD) obtained without using matrices. Ions corresponding to M+' ( m / z = 168, Figure 1A) dominate the positive ion spectrum of TMB and no ions corresponding to (M + H)+ were observed. The strongest high mass peaks for BHD correspond to (M - H)+ ( m / z = 303,305, Figure 1B) and (M - Br)+. The intensity of the 303 peak (28% with respect to the base peak 0 1989 American Chemical Society