RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 6,671-678 (1992)
Kinetic Energy Measurements of Molecular Ions Ejected into an Electric Field by Matrix-assisted Laser Desorption J. Zhou, W. Ens,* K. G. Standing and A. Verentchikov Department of Physics, University of Manitoba, Winnipeg, R3T 2N2, Canada SPONSOR REFEREE: Professor B. U. R. Sundqvist, Department of Radiation Sciences, Uppsala University, Uppsala, Sweden
Measurements of kinetic energy distributions of molecular ions ejected into an extraction field by matrix-assisted laser desorption are reported. The measurements were made in a time-of-flight mass spectrometer with an electrostatic mirror by measuring the reflected signal as a function of the difference between the accelerating voltage and the voltage applied to the mirror. The molecular ions were found to have less kinetic energy than the extraction field alone would normally provide, i.e., we observed an energy deficit. Under conditions typical for a matrix-assisted laser desorption experiment, the deficit is about 24 eV for molecular ions of insulin. The size of the deficit increases with the intensity of the molecular ion signal, and the molecular weight of the protein; it is also larger for negative molecular ions than for positive molecular ions.
Matrix-assisted laser desorption is an important new method for producing molecular ions from peptides and proteins for analysis by mass spe~trometry.'-~ In this method, the sample is irradiated by a short laser pulse in the presence of a large molar excess of a suitable matrix (e.g., sinapinic acid). The mass spectrum is normally acquired with a time-of-flight spectrometer and a transient recorder; the ion current is measured (at the end of a field-free drift region) as a function of time after the incident laser pulse. The ejection dynamics of the ion production process have an important influence on the mass spectrum in time-of-flight measurements. Measurements of initial time and velocity spreads contribute to the understanding of the ejection mechanism, and provide information relevant to the design and performance of the mass spectrometer. Several experimental measurements of initial velocity distributions in matrix-assisted laser desorption have been reported Experiments at Rockefeller University' and in our laboratory6 have examined the axial velocity distribution of molecular ions of insulin by measuring the peak width as a function of flight path length. In these measurements, the spectrometer was run under normal operating conditions, i.e., the ions were accelerated between the target at high voltage and a grounded grid. The Manitoba measurements indicate that for a laser irradiance about twice threshold, the width of the energy distribution is -50eV for molecular ions of insulin ejected from a sinapinic acid matrix and accelerated to -15 keV. Only the width of the energy distribution can be directly extracted from these experiments; the average energy is not known. In order to avoid a possible effect of an electric field on the energy distribution, recent measurements at Rockefeller University determined the axial velocity distribution for three proteins (MW 1030 u, 5733 u and 15 590 u) when the molecular ions are ejected into a field-free region, and subsequently accelerated.* The * Author to whom correspondence should be addressed. 0951-4198/92/110671-08 $09.00 01992 by John Wiley & Sons, Ltd.
velocity distribution was found to be independent of the molecular weight of the protein. Similar results for somewhat smaller ions have been reported by Pan and Cotter.' In the Rockefeller University measurements, the measured average velocity in each case was near 700m/s, and the width of the distribution about 500 m/s. For insulin, this corresponds to an average energy of -17eV and a width of about 12eV, considerably narrower than the width observed in the previous measurements described above. At Manitoba we have recently measured similar velocity distributions for proteins ejected into a field-free region, extended the measurements to proteins of molecular weight up to 66 000 u, and demonstrated that the velocity distribution of protein molecular ions under these conditions is independent of laser irradiance .lo For the measurements described below, we have returned to normal operating conditions for the spectrometer, i.e., we report new measurements of the kinetic energy distributions of several molecular ions ejected directly into an extraction field by matrixassisted laser desorption. The measurements were made in a reflecting time-of-flight spectrometer, using a procedure similar to that used previously in measurements of the kinetic energy distributions of small ions produced by laser desorption.", The method gives information on both the width of the energy distribution and the average energy, clearly indicating the effect of the extraction field on the distribution and giving important clues as to the ejection process. Moreover, since measurements are made for normal operating conditions in the spectrometer, knowledge of the velocity distribution for this situation has considerable practical value. EXPERIMENTAL
Measurements were made in the University of Manitoba reflecting time-of-flight mass spectrometer illustrated in Fig. 1 and described in detail e1~ewhere.l~ For these experiments, laser pulses of about 3 n s Received 21 September 1992 Accepted 22 September 1992
ENERGY DISTRIBUTIONS IN MALD
spectrometer, with a single-stage mirror, is independent of the ion energy to first order. Thus the peak does not shift for small changes in the acceleration voltage and this simplifies the integration procedure, especially when daughter ions are present (see below). Although the method described above is straightforward in principle, there are several practical considerations:
,-, :Electrostatic '
.. .. . I
................... I I
..,............ -3 - Detector 2 Converter
Figure 1. Schematic diagram of the reflecting time-of-flight mass spectrometer and data system, showing possible flight paths of a parent ion (O),daughter ion ( 0 ) and neutral ( 0 ) . The angle between the secondary ion path and the spectrometer axis has been exaggerated for the sake of clarity; the actual angle is about 1.4".
duration from a nitrogen laser (Laser Science ND337, Cambridge, MA, USA) are focused with a 50 cm lens onto the sample target at about 77" to the normal. Ions produced by the laser pulse are accelerated by an electric field between the target at high voltage (V, = 15 kV) and a grounded grid, placed 1 cm from the target. Ions are reflected by a uniform electric field in a single-stage electrostatic mirror. The total equivalent flight path is 2 m. The ions are detected at an electron converter coated with CsI; secondary electrons are directed to a chevron microchannel plate detector by a magnetic field. Spectra for this experiment are recorded in a transient recorder (LeCroy 8828D, Chestnut Ridge, NY, USA) and transferred to an Atari Mega ST computer (Sunnyvale, CA, USA) after each shot. Targets were prepared by drying about 1OpL of a 1 0 - 5 solution ~ of protein with a lo4 molar excess of matrix in 0.1% trifluoroacetic acid (TFA) and acetonitrile (1:3, v/v) onto 1cm2 etched silver foil. Sinapinic acid and a-cyano-4-hydroxycinnamic acid (4HCCA)14 were used as matrices. An ideal electrostatic mirror has a sharp energy cutoff defined by the potential, V,, applied to the final electrode. In the ideal case, ions of charge q entering the mirror parallel to the axis with energy less than qV, are reflected, but ions with higher energy are not. A direct method of obtaining the integral energy distribution is therefore to measure the intensity of the reflected signal as a function of the mirror voltage; the derivative of the resulting curve represents the energy distribution. Since we are interested in the contribution to the final energy distribution from the desorption process, it is the difference between the measured ion energy and qV, (i.e., the energy that would normally be obtained from the extraction field alone) that is relevant. We will therefore represent the distributions as a function of V , - V,. In practice, data were acquired by keeping V,,, constant and varying V,. This procedure was used because the flight time in a time-of-flight
Normalization. It is not feasible simply to normalize the measured intensity to the number of laser shots because of the poor reproducibility in matrix-assisted laser desorption. The signal intensity varies from shot to shot on the same target spot, decreases on average with the number of laser shots on the same spot, and varies significantly for different target spots. An alternative method is therefore necessary. We have used two met hods. For most peptides and proteins, a daughter ion at [M - 17]+ is observed in the positive reflected spectrum (Fig. 2) as a result of metastable decay in the field-free region before the mirror. Since the daughter ion has less energy than the parent, it is possible to reduce the mirror voltage so that the parent ion passes through the mirror but the daughter ion is still reflected. Thus the intensity of the daugher ion peak may be used as a measure of the total signal intensity as the mirror voltage is scanned to determine the energy distribution of the parent ion. This is a very direct method of normalization but has some disadvantages. For smaller peptides (-1000 u) the daughter ion peak is quite small compared to the parent peak (Fig. 2(a)). Because of the limited dynamic range of the experiment, it is difficult to obtain both parent and daughter signals for 160
Flight Time (ns)
Figure 2. Molecular ion region of the matrix-assisted laser desorption mass spectrum for (a) substance P (positive) (b) bovine insulin (positive) (c) cytochrome c (positive) and (d) bovine insulin (negative). The spectra were obtained in the spectrometer of Fig. I and show the parent ion peak and the peak(s) resulting from metastable decay in the field-free region before the mirror.
ENERGY DISTRIBUTIONS IN MALD
all intensities. For larger molecular ions the difference in energy between the parent and the daughter (112 x 17v') is smaller because the veiocity v is smaller. As a result the energy distributions of the parent and daughter overlap for large proteins. In this case the highest energy daughter ions will not be reflected even when the lowest energy parent ions are reflected. This is already observed to some extent for insulin (5733 u). For larger peptides, the parent and daughter peaks are not clearly resolved in time (Fig. 2(c)) so it is difficult to integrate the signals separately. Finally, if a suitable daughter ion is not observed, as is the case for negative insulin (Fig. 2(d)), the method is not applicable. A second method of normalization involves alternating the electrical potential of the target between successive laser shots. To generate an energy distribution, the parent ion signal is measured as a function of one value of the acceleration (the higher value); for the alternate value of the acceleration, all parents and daughters are reflected, providing a normalization signal. This method is more general than the previous method but requires longer runs to average the shot-toshot variations. Most of the energy distributions reported here were normalized by alternating the acceleration voltage.
v m- Va(volts)
.->r c v)
Incident angle. A plane electrostatic mirror affects the velocity of an ion only along the direction of the mirror axis; the perpendicular velocity component is unchanged. Therefore if an ion with energy qV, enters the mirror at an angle 8 to the axis, a potential V , cos' 8 is sufficient to reflect the ion. For ions to strike the detector in our spectrometer, they must enter the mirror at an angle between about 0.8" and 2.0" to the mirror axis. A 15 keV ion entering at the average angle of 1.4" is reflected by a voltage of only 14.991 kV. This effect must be considered when interpreting the measured energy distribution. Energy resolution. In practice the energy cut-off of an electrostatic mirror is not perfectly sharp. This is especially true in a mirror like ours where a grid is used to define the final equipotential. Because of penetration of the electric field through the grid spaces, it is possible for an ion with energy less than qVm to pass through the grid, and not be reflected. The energy resolution of planar retarding grids has been treated in some detail by Enloe and Shell.I5They have shown that the energy resolution is better if the electric field is smaller, and if multiple grids at the same potential are used. We therefore added a second grid about l c m behind the existing grid at the back of the mirror and applied the same potential to it. Such experimental conditions may change the position of the energy cut-off by -10 eV, and they will influence the width of the measured energy distribution, but these effects are difficult to estimate accurately. Fortunately, an experimental method is available to measure such effects directly. The initial energy of molecular ions desorbed by particle bombardment is expected to be small (-3eV) with a comparable width.16 Since this is smaller than the expected instrumental effects, it is possible to characterize the mirror for measurement of energy distributions by studying ions produced by particle bombardment. For this purpose we studied the molecular ions of substance P desorbed from a nitrocellulose substrate by 30 keV I-
Net Energy (eV) Figure3. (a) The ratio of the reflected parent ion intensity to the [M- 17]+ daughter ion intensity as a function of the difference in potential between the mirror and the target, for molecular ions of substance P produced by 30 keV I- bombardment. This represents the measured integral energy distribution. (b) The derived net differential energy distribution. The net energy on the horizontal scale represents the energy after the normal contribution from the extraction field (qV,) is subtracted. The dashed line in (a) and (b) represents the energy cut-off for the mirror based on the estimated average incident angle for ions entering the mirror. The width of the peak is dominated by instrumental effects (see text).
bombardment. These measurements were made using single-ion counting and a time-to-digital converter, so the difficulty with dynamic range mentioned above was not relevant, and normalization was accomplished using the [M- 17]+ metastable daughter ion. Fig. 3(a) shows the ratio of the parent-ion yield to the daughter ion yield as a function of the mirror-to-target voltage difference (Vm- V , ) . The derived differential kinetic energy distribution is shown in Fig. 3(b). For ions with energy qV, a sharp peak is expected at the position of the dashed line in Fig. 3. It is displaced from V , - V ,= 0, in Fig. 3(a), because of the non-zero incident angle of the ions entering the mirror as explained above. The measured distribution is within a few eV of this line and has a width of about 12 eV. This distribution represents the instrumental limitation on resolution and accuracy.
ENERGY DISTRIBUTIONS IN MALD
o.o! . . -150
Flight Time (ns) Figure 4. The molecular ion region of the reflected MALD spectrum for bovine insulin with the mirror voltage (a) above, (b) 60 V below and (c) 90V below the target voltage. The spectra show the parent molecular-ion peak and the [M - 17]+ daughter-ion peak.
RESULTS 1. Energy deficit To illustrate the principle of the measurement, the molecular ion region of the mass spectrum for bovine insulin is shown in Fig. 4 for three values of the mirrorto-target voltage difference. The spectra were obtained using sinapinic acid as the matrix and an irradiance about 3 times threshold. In each case the [M - 17]+ daughter ion should be completely reflected. In contrast to what is expected if the energy from the ejection process simply adds to the energy obtained from the extraction field, the data show that about half of the parent ions are reflected when the mirror voltage is 60 V lower than the acceleration voltage (Va) applied to the target (Fig. 4(b)). Even when the mirror voltage is 9 0 V less than the target voltage (Fig. 4(c)) a small fraction of the parent ions are reflected. Thus the parent ions have energy significantly less than qV, , indicating an energy deficit, as observed previously for small atomic ions desorbed by laser irradiation.". Figure 5(a) shows the molecular ion intensity for bovine insulin as a function of V,,,- V,. The measurements are relative to the intensity obtained on alternate
' . . -100
' . . . . ' 0 50
v m- va(volts)
Net Energy (eV) Figure 5. (a) Integral and (b) differential energy distributions for molecular ions of insulin ejected from a sinapinic acid matrix with near threshold irradiance. The intensity in (a) is normalized to alternate laser shots in which V , - V , = +35 V. The dashed line represents the energy cut-off for the mirror.
laser shots when V,,,- V ,= +35 V, as described in the Experimental section. Figure 5(b) shows the corresponding differential energy distribution corrected for the average incident angle of the ions entering the mirror. For these measurements sinapinic acid was used as the matrix and the irradiance was near threshold. The differential energy distribution peaks at approximately 22 eV below the value expected, and has a width of about 24 eV. A similar, but smaller energy deficit is observed when the matrix 4HCCA is used, as illustrated in Fig. 6. Because of better reproducibility and longer persistence of the spectra obtained with 4HCCA, most of the subsequent data were obtained with this matrix. 2. Intensity dependence We initially intended to examine separately the effect of the laser irradiance and target conditions on the energy distribution, but it became clear that these conditions have only an indirect influence on the distribution. The dominant effect seems to be related to the intensity of the signal in a given laser shot. This is demonstrated in the series of spectra shown in Fig. 7. Here each spectrum represents the sum of selected transients in which the signal intensity is within a
ENERGY DISTRIBUTIONS IN MALD
Net Energy (eV) 1 15000 Figure 6. (a) Integral and (b) differential energy distributions for molecular ions of insulin ejected from a 4HCCA matrix with near threshold irradiance. The intensity in (a) is normalized to alternate laser shots in which V,,,- V, = +35 V. The dashed line represents the energy cut-off for the mirror.
specific range. Thus the top spectrum represents transients for which the signal intensity was high and the bottom spectrum is a sum of low intensity transients. The laser irradiance was constant for these measurements but the laser spot was moved to produce transients of various intensities; the computer then determined the signal intensity and averaged the transient with the appropriate spectrum. The daughter ions are all reflected for the selected mirror voltage (Vm- V , = - 15 V) so the daughter ion peak height was used to determine the intensity of the transient. The spectra indicate that for more intense signals, a larger fraction of the parent-ion signal is reflected, i.e., the energy of the parent is lower for more intense signals, even when the same irradiance is used throughout. Similar results are obtained when only one target spot is used, but the laser irradiance is changed to produce different ion intensities. The energy distributions for different molecular ion intensities are shown in Fig. 8 for molecular ions of bovine insulin. The data show that the energy deficit is considerably larger for an intense molecular ion signal than it is for a weaker signal; the distribution is also considerably broader. The weakest signal (Fig. 8(c)) was produced by first irradiating the target with several hundred laser shots. For such very weak signals from a
Flight Time (ns) Figure 7. The molecular ion region of the reflected MALD spectrum for bovine insulin for selected transients: (a) intense molecular ion signals, (b) typical transients in MALD, and ( c ) very weak molecular ion signals. The irradiance was about twice threshold for each spectrum, but the laser spot was moved around on the target to produce transients of different intensity.
4HCCA matrix, a small energy surplus is actually observed. These measurements were normalized by alternating the acceleration potential, as described in the Experimental section. The selection of molecular ion intensity was achieved on the alternate laser pulses where all the parent ions were reflected. Thus it is assumed that, on average, adjacent pulses have similar ion intensity. Some subtle differences in the energy distribution were observed between molecular ion signals that were weak as a result of low laser irradiance and those that were similarly weak as a result of an unfavorable or damaged target spot. These differences are currently under investigation but they may be related to differences in the ratio of the matrix signal to the molecularion signal. 3. Mass effect In the previous energy measurements discussed in the Introduction, where proteins are ejected into a fieldfree region prior to acceleration, the average kinetic
E N E R G Y DISTRIBUTIONS IN M A L D
0.01 ' . -150
I . -100
. . 0
vm - va (volts) Figure 8. Integral energy distribution for molecular ions of bovine insulin ejected from 4HCCA matrix for selected transients: (a) intense molecular-ion signals, (b) typical transients in MALD and (c) very weak molecular-ion signals. The same irradiance was used for all the data. Transients of different intensity were produced by moving the laser spot on the target. The weakest signals were produced by first damaging a spot on the target with several hundred laser shots.
0.0- . . . . -150
. . -50 1
energy of the protein molecular ions was found to increase in proportion to the mass.'-'" The situation is very different when the proteins are ejected directly into an electric field as shown in Fig. 9. In this case, the average energy decreases (or the energy deficit increases) with increasing mass. Since the energy distribution is strongly dependent on the intensity of the molecular ion signal, the data of Fig. 9 were obtained by running mixtures of the proteins for which the molecular ions gave comparable signal heights. The same laser shots were then used to extract energy distributions for the molecular ions. of the different components of the mixture. With the available dynamic range it was difficult to obtain all four distributions with the same target, so the proteins were run in pairs: substance P with insulin, insulin with cytochrome c, and cytochrome c with trypsin. In each case the heavier protein showed a larger energy deficit. The same effect was observed whether intense or weak signals were selected.
4HCCA matrix is very weak so sinapinic acid was used for these measurements. The distribution is shown in Fig. 10 for two different signal intensities. Again the data are normalized by alternating the accelerating potential as described in the Experimental section. The signal intensity was selected on the alternate shots where the parent ions were completely reflected. For the selected laser shots where the signal was weak, the spectrum was erratic so the points for this case were somewhat scattered. Nevertheless, a clear difference from the distributions for positive ions is observed. The energy deficit is considerably larger, and the effect of the signal intensity, if any, is considerably smaller.
4. Negative molecular ions Preliminary measurements of the energy distribution for negative molecular ions have been made for insulin. The negative molecular ion signal obtained with a
DISCUSSION In simple descriptions of ejection and extraction of ions from a surface, it is usually assumed that the initial energy obtained from the ejection process, E,, com-
v,- v, (volts) Figure 9. Integral energy distributions for (a) substance P, (b) bovine insulin, (c) cytochrome c and (d) bovine trypsin, ejected from a 4HCCA matrix. The laser irradiance was near threshold and the intensity of the transients was typical for analysis by MALD.
ENERGY DISTRIBUTIONS IN MALD
Figure 10. Integral energy distribution for negative molecular ions of bovine insulin ejected from a sinapinic acid matrix for (a) intense and (b) weak molecular-ion signals. Because the polarity is opposite, V , - V,,, is plotted on the horizontal scale.
bines with the energy supplied by the extraction field, qVa, to give a total energy qVa+ E,. The observation that molecular ions produced by matrix-assisted laser desorption in the presence of an extraction field, have less energy than the extraction field alone should contribute, indicates that in this case the above assumption does not apply. A more complicated interaction occurs between the ejected ions and the applied electric field. This was already indicated by the apparent contradiction between previous measurements of energy distributions with and without an extraction field at the target surface, as described in the Introduction. In the field-free situation the energy distribution is relatively narrow and independent of irradiance.8, I" On the other hand, an extraction field at the target yields a distribution that is much broader and depends strongly (although indirectly) on the irradiance. The increasing energy deficit with increasing signal intensity explains the previously reported observation that the broadening that occurs with increasing irradiance occurs entirely on the high time side in a time-of-flight The possibilities of an energy deficit6 or a time delay6." were considered to explain the observation, but our results indicate that the broadening can be understood without invoking a time delay. We have considered various plausible mechanisms that could produce an energy deficit for ions ejected into an extraction field. A rapidly expanding gas jet following a rapid phase change of the solid matrix has been proposed to explain matrix-assisted laser d e s ~ r p t i o n . ~ . 'In , ' ~this model, acceleration of molecular ions could be impeded by collisions with other particles in the gas, or neutral molecules could become ionized by collisions at a lower potential some distance above the surface. Alternatively, surface charging caused by the extraction of light ions (or electrons in the case of negative spectra) could reduce the local field
encountered by the heavier molecular ions and cause an energy deficit. The observation that the energy deficit increases with increasing signal intensity is consistent with any of the above descriptions since the number of collisions and the amount of charging should increase with the intensity of the transient. Since, in general, the collision cross-section increases with molecular weight, the two descriptions involving collisions in the gas phase would predict an increasing deficit with increasing molecular weight as we observed. However, this result is not easily reconciled with the explanation involving surface charging. If the charging is dominated by the removal of matrix ions, and the time scale of the reneutralization is considerably longer than the extraction time of the molecular ions, then the energy deficit would be independent of mass for ions larger than the matrix. If the target surface neutralizes in a time comparable to or less than the extraction time, then larger ions that spend more time near the target would be affected less by the transient surface charge. Thus surface charging is expected to cause larger molecular ions to have the same or a smaller energy deficit than smaller molecular ions, contrary to what is observed. On the other hand, the larger deficit observed for negative molecular ions than for positive molecular ions is more easily understood if charging plays a role; the rapid removal of many electrons is likely to produce more pronounced surface charging for negative ions.
CONCLUSION Molecular ions produced by matrix-assisted laser desorption in the presence of an electric field exhibit an energy deficit; their energy is lower than that of an unimpeded ion accelerated from rest through the same field. For 15 kV acceleration across a 1 cm gap, and for threshold irradiance, the average energy deficit for insulin molecular ions, desorbed from a sinapinic acid matrix, is about 24 eV. The energy deficit increases with increasing signal intensity, and with the molecular weight of the sample. It is also larger for negative molecular ions than for positive molecular ions. The results are consistent with gas-phase interactions occurring above the surface. Acknowledgements This work was supported by grants from the U S National Institutes of Health (Institute of General Medical Sciences), and from the Natural Sciences and Engineering Research Council of Canada, which also provided an International Fellowship for one of us (A.V.).
REFERENCES 1. K. Tanaka, H. Waki, Y.Ido, S. Akita and Y . Yoshida, Rapid
Commun. Mass Spectrom., 2, 151 (1988). 2. M. Karas, D . Bachmann, U. Bahr and F. Hillenkamp, Inf. J . Mass Spectrom. Ion Processes, 78, 53 (1987); Anal. Chem., 60, 2299 (1988). 3. F. Hillenkamp, M. Karas, R. C. Beavis and B. T. Chait, Anal. Chem., 63, 1193 (1991). 4. B. Spengler and R . J . Cotter, Anal. Chem., 62, 793 (1990). 5. R . C. Beavis and B. T. Chait, in Methods and Mechanisms for Producing Ions from Large Molecules, in the NATO AS1 Series B: Physics Vol. 269, ed. by K. G . Standing and W. Ens, p. 227, Plenum Press, New York (1991). 6. W. Ens, Y. Mao, F. Mayer and K. G. Standing, Rapid Commun. Mass Spectrom., 5 , 117 (1991) 7. T. Huth-Fehre and C. H . Becker, Rapid Commun. Mass Spectrom., 5 , 198 (1991).
ENERGY DISTRIBUTIONS IN MALD
8. R. C. Beavis and B. T. Chait, Chem. Phys. Lett., 181, 479 (1991). 9. Y. Pan and R. J. Cotter, Org. Mass Spectrom., 27, 3 (1992). 10. A. Verentchikov, W. Ens, J. Martens and K. G . Standing, Proceedings of the 40th Conference on Mass Spectrometry and Allied Topics, Washington DC, ASMS, East Lansing (1992). 11. T. Mauney and F. Adams, Int. J . Mass Specrrom. Ion Processes, 59, 103 (1984). 12. A. Vertes, P. Juhasz, P. Jani and A. Czitrovszky, Inr. J. Mass Spectrom. Ion Processes, 83, 45 (1988). 13. X. Tang, R. Beavis, W. Ens, F. Lafortune, B. Schueler and K. G. Standing, Int. J . Mass Spectrom. Ion Processes, 85, 43 (1988). 14. R. C. Beavis, T. Chaudhary and B. T. Chait, Org. Mass Spectrom. Lett., 156 (1992).
15. C. L. Enloe and J. R. Shell, Reo. Sci. Instrum., 63, '1788 (1992). 16. S. Widdiyasekera, P. Hikansson and B. U. R. Sundqvist, Nucl. Instrum. and Meth., B33, 836 (1988). 17. A. Hedin, A. Westman, P. Hikansson, B. U. R. Sundqvist and M. Mann, in Methods and Mechanisms for Producing Ions from Large Molecules, in the NATO AS1 Series B: Physics Vol. 269, ed. by K. G. Standing and W. Ens, p. 211, Plenum Press, New York (1991). 18. R. W. Nelson, M. J. Rainbow, D. E. Lohr and P. Williams, Science, 246, 1585 (1989); P. Williams and R. W. Nelson, in Methods and Mechanisms for Producing Ions from Large Molecules, in the NATO AS1 Series B: Physics Vol. 269, ed. by K. G. Standing and W. Ens, p. 265, Plenum Press, New York (1991).