DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0046] Referring first to FIG. 1 , an apparatus in accordance with the present invention is indicated generally by reference 10 . In known manner, the apparatus 10 includes an ion source 12 , which may be an electrospray, an ion spray, a corona discharge device or any other known ion source. Ions from the ion source 12 are directed through an aperture 14 in an aperture plate 16 . On the other side of the plate 16 , there is a curtain gas chamber 18 , which is supplied with curtain gas from a source (not shown). The curtain gas can be argon, nitrogen or other inert gas, such as described in U.S. Pat. No. 4,861,988, Cornell Research Foundation Inc., which also discloses a suitable ion spray device, and the contents of this patent are hereby incorporated by reference.

[0047] The ions then pass through an orifice 19 in an orifice plate 20 into a differentially pumped vacuum chamber 21 . The ions then pass through aperture 22 in a skimmer plate 24 into a second differentially pumped chamber 26 . Typically, the pressure in the differentially pumped chamber 21 is of the order of 2 torr and the second differentially pumped chamber 26 , often considered to be the first chamber of mass spectrometer, is evacuated to a pressure of about 7 mTorr.

[0048] In the chamber 26 , there is a standard RF-only multipole ion guide Q0. Its function is to cool and focus the ions, and it is assisted by the relatively high gas pressure present in this chamber 26 . This chamber 26 also serves to provide an interface between the atmospheric pressure ion source and the lower pressure vacuum chambers, thereby serving to remove more of the gas from the ion stream, before further processing.

[0049] An interquad aperture IQ1 separates the chamber 26 from the second main vacuum chamber 30 . In the main chamber 30 , there are RF-only rods labeled ST (short for “stubbies”, to indicate rods of short axial extent), which serve as a Brubaker lens. A quadrupole rod set Q1 is located in the vacuum chamber 30 , and this is evacuated to approximately 1 to 3×10 ⁻⁵ torr. A second quadrupole rod set Q2 is located in a collision cell 32 , supplied with collision gas at 34 .The collision cell is designed to provide an axial field toward the exit end as taught by Thomson and Jolliffe in U.S. Pat. No. 6,111,250. The cell 32 is within the chamber 30 and includes interquad apertures IQ2, IQ3 at either end, and typically is maintained at a pressure in the range 5×10 ⁻⁴ to 8×10 ⁻³ torr, more preferably a pressure of 5×10 ⁻³ torr. Following Q2 is located a third quadrupole rod set Q3, indicated at 35, and an exit lens 40 . The pressure in the Q3 region is nominally the same as that for Q1 namely 1 to 3×10 ⁻⁵ torr. A detector 76 is provided for detecting ions exiting through the exit lens 40 .

[0050] Power supplies 36 , for RF and resolving DC, and 38 , for RF, resolving DC and auxiliary AC are provided, connected to the quadrupoles Q1, Q2, and Q3. Q1 is a standard resolving RF/DC quadrupole. The RF and DC voltages are chosen to transmit only the precursor ions of interest into Q2. Q2 is supplied with collision gas from source 34 to dissociate precursor ions or fragment them to produce fragment or product ions. Q3 is operated as a linear ion trap mass spectrometer as described in U.S. Pat. No. 6,177,668, i.e. ions are scanned out of Q3 in a mass-dependent manner, using the axial ejection technique taught in that earlier U.S. patent.

[0051] In the preferred embodiment, ions from ion source 12 are directed into the vacuum chamber 30 where the precursor ion m/z is selected by Q1. Following precursor ion mass selection, the ions are accelerated into Q2 by a suitable voltage drop into Q2, inducing fragmentation. These 1st generation fragment ions are trapped within Q2 by a suitable repulsive voltage applied to IQ3. Once trapped the RF voltage applied to the Q2 rods is adjusted such that all ions above a chosen mass are made unstable, that is there a,q values fall outside the normal Mathieu stability diagram. Removal of ions above the mass of a particular ion of interest is facilitated by the addition of a small amount of resolving DC voltage, here 1.8 volts, applied to the Q2 rods. Next the RF is adjusted so that ions below a particular mass are made to be unstable. These two steps can be accomplished very quickly, on the order of 1-3 ms each. The result is a mass isolated ion population, which can be further collisionally activated.

[0052] The subsequent collisional activation step can be accomplished as in a conventional three-dimensional ion trap, that is by application of an appropriate resonant AC waveform. This however requires sophisticated electronics and has the additional requirement that the trapping RF voltage be such that the lowest mass fragment ion and the precursor ion are simultaneously stable within Q2.

[0053] An alternative technique is to simply accelerate the mass isolated ions in to the subsequent mass analyzer. Since Q2 is operated at elevated neutral gas pressure, say 5×10 ⁻³ torr, there is a neutral gas pressure gradient between IQ3 and the subsequent mass analyzer. If the mass isolated ions within Q2 are accelerated through this pressure gradient into the Q3 linear ion trap there will be a sufficient number of collisions to induce further fragmentation. The result is a MS ³ mass spectrum.

[0054] By way of example consider the following set of experimental results obtained using the apparatus in FIG. 1 . A sample of 100 pg/mL of reserpine (MW=608) is introduced into the ion source 12 where it is ionized and directed into the vacuum chamber 30 . The RF and DC voltages of Q1 are adjusted to transmit a 0.7 amu wide beam of the protonated reserpine ions at m/z 609 into Q2. The DC voltage offset of Q2 relative to Q1 is chosen to be 35 volts, which is sufficient to produce extensive fragmentation of the reserpine precursor ion. Q2 is operated as a simple accumulation ion trap by adjusting IQ3 to an appropriately repulsive DC voltage so that none of the entering precursor ions or fragment ion generated therein can exit. Q2 is filled for 50 ms, after which the DC voltage applied to IQ2 is raised to the same value as the trapping IQ3 value. There is now a trapped population of primary fragment and residual precursor ions resident within Q2. If all the ions within Q2 are now allowed into the Q3 linear ion trap mass spectrometer and mass analyzed, the MS ² mass spectrum displayed in FIG. 2 a is obtained. To obtain MS ³ data of the m/z 397 ion), this fragment ion must be isolated and collisionally activated prior to mass analysis by the Q3 linear ion trap mass spectrometer.

[0055] Ion isolation of the m/z 397 fragment ion was accomplished in a step-wise fashion by first adjusting the RF voltage applied to the Q2 rods such that ions above m/z 397 become unstable within Q2 and are lost. The result of this step is displayed in FIG. 2 b . Here, one can see that the ion population within Q2 has been modified such that there is little or no contribution to the MS ² mass spectrum from ions m/z>397.

[0056] Low mass ions may be eliminated from the Q2 ion population by adjusting the RF voltage such that the trapped ions with m/z below 397 become unstable in the Q2 and are also lost. The result of this step prior to mass analysis is displayed in FIG. 2 c , which shows that low mass ions can be effectively eliminated from Q2.

[0057] A combination of these two steps thus provides good mass isolation of the m/z 397 fragment ion within Q2 as is displayed in FIG. 2 d , i.e. these two steps are performed sequentially in Q2. The time penalty for the mass isolation steps is approximately 2×2 ms or a total of 4 ms. As Q2 is a high pressure collision cell, true mass filtering is not possible, and in particular it is not possible to get a sharp cutoff between selected or retained ions, and rejected ions, as is possible in a low pressure mass analysis section, such as Q1. For this reason, it is not possible to apply a narrow window selecting just the desired m/z 397. Any attempt to do this would result in significant loss of the 397 ion. Rather, it has been found that by sequential rejection of masses above and below the mass of interest, the bulk of the unwanted ions can be rejected. Note that in FIGS. 2 a - 2 e , the vertical scale indicates relative intensity with the most populous ion being indicated as 100%.

[0058] Finally, the m/z 397 ions are accelerated into the Q3 linear ion trap MS by increasing the relative DC voltage offset between Q2 and Q3 from 5 volts (used in FIGS. 2 a - c ) to 25 volts. Collisions at the exit of Q2 and entrance of Q3 lead to fragmentation of the m/z 397 ions and results in the MS ³ spectrum displayed in FIG. 2 d . As expected, a range of masses of secondary fragmentations, with masses below m/z 397, are present in the spectrum. Again, the vertical axis shows relative intensity, and as the residual primary fragment ion 397 is still the most populous, it is shown with an intensity of 100%, with the secondary fragment ions of low masses shown accordingly.

[0059] This procedure can be carried out separately on the major fragment ions in the reference reserpine MS ² spectrum of FIG. 2 a . The result is displayed in FIG. 3 where the highest mass peak in each spectrum corresponds to the isolated MS ² primary fragment ion used to obtain the MS ³ spectrum. Thus, FIG. 3 a again shows the complete MS ² spectrum for m/z 609; FIGS. 3 b - 3 e show the MS ³ spectra for the primary fragment ions 448, 397 (equivalent to FIG. 2 e ), 195 and 174, respectively.

[0060] For this technique to be widely applicable the collisional activation step must be sufficiently energetic to provide a wide range of MS ³ fragment ions. The ability to fragment the m/z 609 reserpine ion is a good measure of the energetics of fragmentation since approximately 30 eV _lab of energy is required to observe the m/z 174 and 195 ions.

[0061] FIG. 4 shows the MS ³ mass spectrum obtained after isolation of the residual m/z 609 ions in Q2, i.e. here the residual precursor ions 609 were retained and all the primary fragment ions were rejected. These residual precursor ions 609 were then subjected to collisional activation using a 30-volt potential drop between Q2 and Q3. One can see that all of the major fragments in the MS ² spectrum ( FIG. 2 a ) are present in FIG. 4 , although the relative intensities differ, as the relative intensities, in known manner, will vary depending upon variations in the collision energy of the fragmentation process. This demonstrates that the method for obtaining MS ³ provides sufficiently energetic collisions to generate fragmentation for many potentially important compounds.

[0062] It is understood that the ion isolation step can be accomplished via notched broadband isolation techniques. This entails subjecting the trapped ions to a plurality of excitation signals uniformly spaced in the frequency domain with a notch of no excitation signals corresponding to the resonant frequencies of the ions to be isolated within the ion trap as described by Douglas et al. in WO 00/33350.

[0063] The present inventors have also discovered and identified that one of the important experimental parameters in the transfer of ions from the Q2 linear ion trap to the Q3 linear ion trap is the RF voltage value applied to the Q3 linear ion trap during the Q2-to-Q3 ion acceleration process. Ions received in Q3 can only be successfully trapped within Q3 if their associated q-value is less than ˜0.9. FIG. 5 shows that when the reserpine molecular ion at m/z 609 is accelerated from Q2 into Q3 while the RF voltage is set such that only ions with m/z>350 have a q-value<0.9, only product ions with mass-to-charge values greater than 350 are observed in the final mass spectrum. The m/z value associated with the q=0.9 RF voltage during the Q3 fill step is referred to as the “Q3 fill mass”; and while this suggests a single mass, as FIG. 5 shows it really defines a lower limit to a range of masses.

[0064] The inventors have found that another important parameter is the time for which the Q3 RF voltage is held at the fill mass, referred to as the “Q3 fill time”. This Q3 fill time is in general longer than the actual time required to empty the Q2 ion trap. Ions can be removed from Q2 very rapidly by using an axial DC field as taught by Thomson and Jolliffe in U.S. Pat. No. 6,111,250. At the pressures and voltages used in the current instrument all the ions within Q2 should be transferred to the Q3 ion trap in less than 2 ms, which can be identified as a “transfer time”. Any time in excess of this 2 ms or other transfer time but less than the Q3 fill time is referred to as the “delay time”.

[0065] The Q3 fill time for the experiment that resulted in the spectrum displayed in FIG. 5 was 50 milliseconds (i.e. 2 ms transfer time and 48 ms delay time). If this value is reduced to 5 milliseconds (i.e. 2 ms transfer time and 3 ms delay time) then the mass spectrum in FIG. 6 results. The most obvious difference between the mass spectra in FIGS. 5 and 6 is the appearance of low mass product ions below the Q3 fill mass in FIG. 6 .

[0066] It is necessary to consider the details of the scanning procedure to understand the reason for the appearance of the low mass-to-charge product ions in the FIG. 6 mass spectrum. The particular scan function employed here is shown in FIG. 7 , which shows the timing steps from the Q3 fill step onward. During the Q3 fill step the value of IQ3 is set to allow ions to flow from Q2 into Q3, as indicated at 20. Simultaneously, an RF voltage 22 is supplied to the rod set Q3. The value of the Q2 to Q3 DC voltage rod offset (not shown in FIG. 7 ) is simultaneously adjusted to the value of the desired laboratory reference frame collision energy. The exit lens 40 is provided with a high voltage, indicated at 24, during the Q3 fill step, so as to provide an appropriate trapping voltage. The drive RF voltage 20, and thus Q3 fill mass, is set to some optimum value during the Q3 fill step, and at the end of the fill step, is then rapidly changed (in less than 100 microseconds as indicated at 26) to an RF voltage 28 to be used at the beginning of the mass scan.

[0067] As indicated at 30, at the end of the fill time, the voltage on the interquad aperture IQ3 is increased to a potential indicated at 32. Simultaneously, the voltage on the exit lens 40 is maintained, so that Q3 then acts as an ion trap.

[0068] At the end of the Q3 fill time, the voltage on the exit lens 40 is dropped as indicated at 34 to a voltage 36, and both the RF voltage and the AC excitation voltage for Q3 are ramped up as shown at 38 and 40, respectively. This then provides a mass spectrum of the ions trapped in the Q3 linear ion trap. At the end of the scanning phase the voltage at IQ3 drops at 42 to a lower voltage 44. Simultaneously, the RF and AC voltages are dropped as shown at 46 and 48 respectively, to final voltages 50 and 52.

[0069] The inventors have found that a very important factor influencing whether or not ions with mass-to-charge ratios below that of the Q3 fill mass are observed is the duration of the Q3 fill step, i.e. the Q3 fill time up to the voltage changes indicated at 26 and 30 in FIG. 7 . This is shown by the differences between the product ion mass spectra for the protonated reserpine molecular ion at m/z 609 in FIGS. 5 and 6 . The only differences between the spectra are the Q3 fill time which is 50 ms (i.e. 2 ms Q2-to-Q3 transfer time and 48 ms delay time) for FIG. 5 and 5 ms (i.e. 2 ms Q2-to-Q3 transfer time and 3 ms delay time) in FIG. 6 , all other parameters are the same: Q2-to-Q3 acceleration energy=35 volts and Q3 fill mass=350.

[0070] It is believed that the reason for the observation of ions with q-values seemingly greater than the first stability region limit of ˜0.908 is the unique Q2-to-Q3 fragmentation environment. The pulse of ions was introduced into the Q3 linear ion trap at a translational energy of 35 eV _lab . Since the neutral gas pressure within Q3 is relatively low, approximately 3×10 ⁻⁵ torr, the corresponding collision frequency is also low. Thus, in a short time frame there will be few momentum dissipating collisions within Q3, at least compared to the conventional high pressure collision cell (B. A. Thomson et al. Anal. Chem. 1995, 34, 1696-1704 ). A considerable amount of translational kinetic energy will remain in any unfragmented precursor ions after a short Q3 fill time of 5 ms. The end of the Q3 fill period is marked by a rapid reduction in the Q3 RF voltage at 26, i.e. a reduction in the lowest m/z ion that is now stable within the Q3 linear ion trap. If any precursor ion within the Q3 ion trap has retained sufficient internal energy, it may collide with a neutral gas atom or molecule to produce a product ion with a q-value that falls within the first stability region defined by the RF voltage during the cooling portion (shown at 28 in the FIG. 7 timing diagram), this product ion can be trapped and detected during the subsequent mass scan. The presence of low mass product ions in the 5 ms Q3 fill time spectrum in FIG. 6 is clear evidence that sufficient energy was retained by the precursor ion population trapped within the Q3 ion trap, so that when the RF voltage was reduced in the “cooling time” step, these precursor ions could provide efficient fragmentation and the fragment ions would then be stable in Q3. In contrast, the 50 ms Q3 fill time spectrum in FIG. 5 , shows that the amount of energy dissipated between the time ions are injected into Q3 and the time when the Q3 RF voltage is reduced to the lower level of the cool step is too long for a sufficient number of precursor ions to retain a high enough kinetic energy for the production of fragment ions. Also, if any fragment/product ions are generated during the fill time, the higher mass cutoff will cause them to be rejected. Consequently, with a long delay time, the precursor ions have experienced enough collisions within the Q3 linear ion trap to preclude the formation of any significant quantity of low mass-to-charge product ions of reserpine. Thus, this method allows one to vary the average amount of internal energy deposited into a precursor ion and more significantly retained until the start of the cooling step when the lighter ions will be stable within Q3. This variation is effected simply by changing the delay time between the 2 ms Q2-to-Q3 transfer time and the time at which the Q3 RF amplitude is reduced, terminating the Q3 fill time and starting the cooling time.

[0071] One advantage to operating the instrument with a high Q3 fill mass is a higher intensity product ion mass spectrum relative to that obtained with a low Q3 fill mass. FIG. 8 shows the product ion mass spectrum of the protonated reserpine ion at m/z 609 obtained with a Q3 fill mass of 180. Comparison of this mass spectrum with that in FIG. 6 (which was obtained under the same conditions except that the Q3 fill mass was 350) shows that the higher Q3 fill mass of 350 results in a sensitivity increase of about 20×. The increased in sensitivity for the Q3 fill mass of 350 mass spectrum is likely due to a larger radial well depth that better confines any scattered ions during the Q3 fill step. Intensity is maximized when the Q3 fill mass is approximately ½ that of the precursor ion mass-to-charge ratio, although the optimization characteristics are broad.

[0072] A further advantage to the use of an elevated Q3 fill mass is that the ions with m/z<Q3 fill mass are produced at a later time (after the cooling time) than those with m/z>Q3 fill mass, as they are products of precursor ions with lower kinetic energy since some collisional relaxation of the precursor ion during the delay time. That is, the energy of the precursor ion has been reduced by some of the relatively infrequent collisions within Q3 during the fill time. Thus consecutive fragmentation processes producing these ions with m/z<Q3 fill mass are less favoured since the precursor ion has less internal energy at the time at which the lower mass product ions are collected. The resulting product ions in turn have less internal energy and thus reduced probability of further fragmentation, leading to suppression of second generation product ion precursor-to-product ion pairs. This can make it easier to identify first generation precursor-to-product ion pairs, which can be especially useful in the identification and differentiation of different dissociation pathways.

[0073] An example is the mapping of the product ions of bosentan studied by Hopfgartner et. al. (J. Mass Spectrom. 1996, 31, 69-76 ). Hopfgartner et. al. found that the major m/z 280 product ions ion in the product ion spectrum of the m/z 552 bosentan molecular ion does not arise directly from the molecular ion, but rather from a two step process involving fragmentation of the m/z 508 ion to the m/z 311 ion and finally to the m/z 280 product ion. The product ion mass spectrum of the m/z 552 molecular ion is displayed in FIG. 9 . This spectrum was obtained by mass selecting the m/z 552 precursor ion with Q1 and accelerating this ion into the conventional Q2 collision cell and trapping the resultant product and residual precursor ions in the Q3 linear ion trap, from which they were mass selectively scanned out. This mass spectrum is virtually identical with that reported by Hopfgartner et al. Note the strong product ion feature at m/z 280.

[0074] A product ion mass spectrum for bosentan was obtained using the method described herein. Once again the precursor ion was mass selected by Q1 and then, in accordance with the present invention, it was introduced into and trapped within Q2, this time at low energy in order to eliminate fragmentation. Next, the ions trapped within Q2 were accelerated into the Q3 linear ion trap at a laboratory collision energy of 30 eV, a Q3 fill mass of 400, and a Q3 fill time of 5 ms (i.e. 2 ms transfer time and 3 ms delay time). Thus, the only product ions that would be stable during the 5 ms fill time in the Q3 ion trap have m/z>400. Immediately after the Q3 fill time (at 26 in FIG. 7 ) the Q3 RF voltage was reduced to that corresponding to m/z 100, which would allow trapping of any product ions with m/z<400. As the delay time is short, precursor ions and first generation fragment ions should have retained sufficient energy, to collide and fragment, forming lighter ions which are now stable. The result is a somewhat different product ion mass spectrum from the one in FIG. 10 , in that the relative intensity of the m/z 280 product ions ion is significantly reduced from that in FIG. 9 .

[0075] The product ion mass spectrum of the m/z 552 bosentan molecular ion obtained with the Q3 fill mass set at 400 for a 10 ms fill time (i.e. 2 ms transfer time and 8 ms delay time) is displayed in FIG. 11 , with the conditions otherwise being the same as in FIG. 10 . The additional 5 ms spent at the Q3 fill mass has a profound effect on the mass spectrum. This increased delay time allows the precursor ions time to dissipate some energy; thus residual precursor ions and first generation fragments, after commencement of the cooling time with the broader stability band, are much less likely to have sufficient energy for further fragmentation to occur. Most of the same product ions ion peaks are still distinguishable, but at much reduced intensity below the fill mass; note that intensities in the mass range to <m/z 480 are shown magnified by a factor of 10. Notable also is that the mass spectrum shows virtually complete elimination of the m/z 280 product ions ion peak. This is strong evidence that the m/z 280 product ions ion is a secondary fragmentation product, or has a higher appearance energy (i.e. requiring a precursor ion to have a high energy than other product ions ions<m/z 400. These results are in agreement with those of Hopfgartner et. al.

[0076] The only limitation for the use of a variable Q3 fill mass is that the precursor ion must be stable within the Q3 linear ion trap, so the Q3 fill mass must be less than the mass-to-charge ratio of the precursor ion.

[0077] This method has also been found to be useful for the simplification of peptide product ion spectra as is demonstrated in FIG. 12 . This figure displays two product ion spectra of a doubly charged peptide product ions at m/z 1094 from digestion of beta-casein in the presence of trypsin. FIG. 12 a is the optimized product ion spectrum using conventional Q1-to-Q2 acceleration and generation of fragment ions in the Q2 collision cell with subsequent mass analysis using the Q3 linear ion trap. The resulting spectrum is particularly rich in the low mass-to-charge region due to the presence of sequential fragmentation and internal product ions products. FIG. 12 b is a Q2-to-Q3 acceleration product ion mass spectrum of the doubly charged m/z 1094 ion from the same beta casein sample, i.e. with ions passed through Q2 with substantially no fragmentation. FIG. 12 b was obtained with a Q3 fill mass of 600 and a Q3 fill time of 7 ms. The two spectra are similar, however FIG. 12 b is much less congested in the region below the Q3 fill mass. FIG. 13 shows an expanded view of the lower mass-to-charge region of these product ion spectra. The assignments of the mass peaks in the product ion spectra have been included. FIG. 13 b was obtained using the Q2-to-Q3 acceleration method show only y-ions in this mass-to-charge region. The standard Q1-to-Q2 acceleration data in FIG. 13 a displays the same y-ions and many other fragmentation products including b-ions and internal product ions. The congestion in FIG. 13 a makes identification of sequence specific product ions difficult if not impossible. However FIG. 13 b contains only sequence specific y-ions. The discrimination against b-ion products and those resulting from internal fragmentation pathways has been found to be general phenomenon for Q2-to-Q3 acceleration collisional dissociation of peptides resulting from trypsin digestion using an elevated Q3 fill mass.

[0078] The technique of ion isolation within a nominally RF-only collision cell and subsequent ion acceleration with concomitant fragmentation is also applicable to other Qq(MS) (where Q designates the mass selection step via a conventional RF/DC resolving quadrupole mass spectrometer and q the higher pressure nominally RF-only collision cell , here carried out in Q1 and Q2 respectively) instruments, where the MS stage can be another fast scanning mass spectrometer other than a linear ion trap mass spectrometer. One such device is a QqTOF tandem mass spectrometer. The TOF is particularly well suited to be used for the final mass analyzer since it is best used with a pulsed ion source, which is what emerges from the collision cell. Furthermore, a full mass spectrum can be obtained for each ion pulse, giving better overall efficiency.

[0079] Additionally, it may in some circumstances be possible to eliminate the collision cell, and provide the collision gas by some other mechanism to the flow of ions into Q3. Additionally, the basic requirement for the section of containing Q3 is that it will be a lower pressure section capable of collecting and collimating ions. It could include, for example, a multipole rod set that provides just this function without acting as a mass analyzer. Where it is desired to set a fill mass, the multipole rod set must be capable of defining this cut off mass with a required degree of precision. A mass analyzer can then be provided downstream.

[0080] The final step of mass analyzing the MS ³ fragment ions can also be carried out using other mass analyzers that yield full mass spectra for a single pulse of ions such a 3-dimensional ion trap.