CYCLOTRONS AND SYNCHROCYCLOTRONS FOR ONCOLOGY THERAPY H. Blosser. D. Johnson. B. Milton. and J. Riedel National Superconducting Cyclotron Laboratory. Michigan State University. East Lansing. MI 48824. and W.E. Powers. R.L. Maughan, C.G. Orton, D.P. Ragan, G.F. Blosser, R.J. Burleigh, and E.B. Jemison, Harper-Grace Hospitals, Detroit, MI 48201 I. Introduction
The use of cyclotrons and synchrocyclotrons to produce ionizing beams for oncology therapy is undergoing a major evolution as a consequence of recent progress in adapting superconducting techniques to each of these accelerator systems. These new devices, the so-called "superconducting" cyclotron or "superconducting" synchrocyclotron, are in fact simply an isochronous cyclotron or a synchrocyclotron with a superconducting main coil. The apparently simple step of making the main coil superconducting has a large impact on the overall accelerator design. The direct effect of making the main coil superconducting is to rather fully free the design from the cost constraints related to main coil current. Cost optimization of the design with these constraints removed leads to much higher magnetic fields, typically in the range around 5 tesla versus the 1.4 to 2.0 tesla typical in room temperature cyclotrons and synchrocyclotrons. The higher magnetic field makes the accelerator smaller and lighter relative to a room temperature cyclotron or synchrocyclotron of the same energy. Typical linear dimensions of a superconducting design are about one-third as large as the corresponding dimensions for a room temperature system and typical weight of a superconducting cyclotron is about one-twentieth of the corresponding room temperature weight. The large decrease in size and weight more than off-sets the added costs which go with buying superconductor, constructing a low temperature vessel, installing super insulation, etc. Overall the superconducting cyclotron is then usually one-third to one-half the cost of a room temperature cyclotron of the same energy. and synchrocyclotrons would behave similarly. At this time (March 1985) only one superconducting cyclotron is in operation in the world, this being the "K500" at the National Superconducting Cyclotron Laboratory in East Lansing and there are no superconducting synchrocyclotrons. The advantages of the superconducting technology are, however, broadly accepted in the physics community--five of eight major cyclotrons now in construction in the world are superconducting and the three which are not predate the introduction of the superconducting technology. (To the author's knowledge, no synchrocyclotrons are under construction at this time.) The reduction in size and cost which makes superconducting accelerators attractive for physics applications is of course also highly important in medical applications. A first such project, a 50 MeV deuteron cyclotron for
-neutron therapy is then already in process. In this application the characteristics of the superconducting cyclotron lead to a greatly simplified design in which the cyclotron is itself mounted in the head of an isocentric rotation system in much the same fashion as a modern electron linear accelerator therapy system. Neutrons are produced in an internal target so that extraction system, beam transport system and isocentric external magnet system are all eliminated. Major features of this project are described in Section II of this paper and in references 1 and 2. Determining the optimum design for a medical proton therapy accelerator is unfortunately a significantly more complicated matter than the neutron application. Three different kinds of accelerators are likely choices namely the cyclotron, the synchrocyclotron, and the synchrotron, and for each both room temperature and superconducting options must be considered. First of all the conventional room temperature isochronous cyclotron meets or exceeds all proton therapy requirements and the technology is firmly developed. Such a cyclotron provides easily variable energy and beam current up to 10 microamps, i.e. a thousand times higher than is conventionally used in therapy. A fairly well optimized version of such a cyclotron has been described in an earlier paper (ref 3). A 250 MeV isochronous cyclotron can also be superconducting but, for protons, focussing and extraction limit the magnetic field which can be used to about 2.5 tesla (reference 4 explains the precise limiting phenomena in some detail). An increase in field to 2.5 tesla is a significant but not a dominating gain relative to the 1.4-1.8 tesla, which would be used in a room temperature cyclotron. The superconducting isochronous cyclotron is then not exceptionally attractive as a proton therapy system and detailed stUdies have not been pursued except to the degree of using scaling relationships to estimate some of the major parameters such as magnet size. cost, etc. The room temperature synchrocyclotron is the accelerator used in presently operating proton therapy programs. It is fairly well matched to the therapy requirements except that energy variation must be accomplished by penetration through degraders, which also reduces beam quality. Room temperature synchrocyclotrons are also massive and bulky. Construction of a new such machine would involve large cost for both the accelerator and the associated building. The synchrocyclotron can also be designed as a superconducting system and this concept is compatible with very high magnetic field values, possibly as high as 7 tesla. As with the room temperature synchrocyclotron the energy is fixed. but the beam current (10-100 na) substantially exceeds the therapy requirement so that energy variation by degrading is feasible. A design study for a superconducting synchrocyclotron is described in Section III of this paper. The proton synchrotron is an accelerator system which easily achieves the desired proton energies. Energy variability is also straight forward. Careful design is required to achieve 10 nanoamps of beam current and the complexity of a synchrotron is a significant possible disadvantage (the need for an injector, the carefully synchronized time variations required by the magnet, the rf frequency, and the systems used to inject and extract, etc.). Synchrotrons of both room temperature and superconducting designs are

described in other papers at this conference and are therefore not discussed further here. Omitting the synchrotrons, Section IV of this paper undertakes to compare major attributes of a number of cyclotron and synchrocyclotron systems of interest in oncology therapy. II. A Superconducting Cyclotron for Neutron Therapy
Figure 1 shows a cutaway view of the superconducting cyclotron which is being constructed at the National Superconducting Cyclotron Laboratory as a neutron therapy system for Detroit's Harper Hospital. The cyclotron uses a "pillbox" yoke so that the steel of the yoke functions as an integral part of the radiation shielding system, protecting the patient from primary neutrons except for the area of the tumor and also protecting personnel from residual radioactivity. Neutrons produced in the internal target are collimated in a conventional collimator system mounted in the yoke and directed at the tumor region. The acceleration system for the cyclotron is a "dee-in-valley" system in which a dee is mounted In each of the thr~e valleys of a three hill, three valley magnet. An ion source is inserted along the axis of the magnet in a manner which gives accurate positioning relative to the acceleration structure. The cryostat for the main coil utilizes a novel, invertible, continuously vented structure and a simple bath cooling design holds cryogens sufficient to provide for a week of coil operation. Figure 2 displays the isocentr;~ mounting system for the Harper Hospital neutron therapy cyclotron. The 25 ton mass of the CyClotron plus a corresponding counterweight are easily supported by a pair of large steel rings which rest on below-the-floor rollers. With box rings constructed of 3/4 inch plates, maXimum stress in the rings is 5,800 lbs/sq. inch and stress deflection of the neutron aiming point as the cyclotron is rotated is small. (The aiming error introduced by the deflection is 0.7 mm.) The location of the counterweight--at zero degrees relative to the direction of the deuteron beam as it strikes the target--also means that the counterweight plays an important role in shielding the most penetrating component of the neutron spectrum. The thickness of shielding walls can then be sizably reduced. Figure 3 shows the overall system as seen by the physician and patient. The patient table mounts outside the ring system on a fixed concrete floor with a canterlevered extension to support the patient. The table system includes all conventional table position adjustments. The floor includes a special custom designed moveable section which moves aside as the cyclotron shifts to the angUlar region immediately below the table. When the cyclotron is at any of the I,pward angUlar locations the special floor provides a convenient and comfortable footing for patient and physician access. The system includes arrangements for quickly and conveniently changing collimators and for verifying patient position.

The complete cyclotron and support system should undergo Laboratory tests in the summer of 1986. Patient treatment using the facility should begin at Harper Hospital early in 1987. III. Superconducting Synchrocyclotron
-Historically, the synchrocyclotron has been the dominant proton therapy accelerator. Discussion at this conference has focused on a 250 MeV proton beam with intensity of 10 nanoamperes as meeting the requirements for proton radiotherapy. Capability for lowering the beam energy to values as low as 70 MeV is also important. The synchrocyclotron in fact usually achieves much higher extracted currents, up to levels of a few microamperes in recently modified synchrocyclotrons, which gives a comfortable margin to cover intensity losses associated with the process of degrading the energy to lower values in situations where lower energy is needed. The room temperature synchrocyclotron has the disadvantage of being qUite massive. The Rochester synchrocyclotron, for example, produced 240 MeV protons and used a 1000 ton magnet (ref 5). The Harvard synchrocyclotron reaches 165 MeV, with a 640 ton magnet. Noting that the cost of machined steel is typically $1-$1.25/lb, the cost of steel for a conventional synchrocyclotron is then of itself an almost prohibitive expense in today's economy. From the point of view of building construction, it is also clearly desirable to reduce the weight of the cyclotron magnet as much as possible. Achieving a weight reduction which would permit isocentric mounting of the cyclotron in much the same manner as the previously described neutron system would offer many significant theraputic advantages, as well as reducing cost. Application of superconducting techniques to the synchrocyclotron leads to structures which are much more compact than the conventional synchrocyclotron and much lighter. Assuming that focussing is derived from the average field gradient in the customary synchrocyclotron way there is in fact no clear limit on the maximum field strength which might be used, and the higher the field the lighter the magnet. In particular, superconducting magnets of this general type and size have been successfully constructed in the range of fields up to and beyond 10 tesla. There is however a general consensus to the effect that the overall cost optimum for such magnets is at somewhat lower fields and the studies described here have therefore used 5 tesla and 7 tesla as illustrative cases. For 250 MeV, the magnet would weigh 80 tons at 5 tesla and 60 tons at 7 tesla both of which are light enough to be compatible with isocentric mounting. Figure 4 and 5 give a plan view and a vertical section view of such a synchrocyclotron and generally illustrate these features. The design assumes a one dee accelerating structure as is normal in synchrocyclotrons, but the high frequency (84 Mhz at a central field of 5.5 tesla and 120 mhz at 7.7 tesla) leads to resonators which end within the magnet if built with the normal "quarter wave" design and for these two cases one then needs "three-quarter" and "five-quarter lambda" systems, respectively, to bring the tuning elements outside the magnet yoke. Designs of this type are however straightforward, the synchrocyclotrons at Berkeley (ref 6) and Cern (ref 7) being examples of three-quarter lambda systems which have functioned smoothly for many years. Beam extraction from the superconducting synchrocyclotron is assumed to be accomplished by a "peeler" induced regenerative system in the fashion which is basically standard for synchrocyclotrons. Since this extraction is accomplished by means of magnetic perturbations one qualitatively expects the behavior of the extraction process to scale With the magnetiC field, i.e. that behavior at high fields will be similar to behavior at low fields.

Calculations checking this point have however not been made. Such calculations should clearly be an early element in any further design study. Other elements of the superconducting synchrocyclotron system are reasonably evident in the figures. The ion source enters axially through the magnet, the main superconducting coil is in an annular cryostat. room temperature penetrations through this cryostat provide for the dee stem and the extraction path, etc. The superconducting coil is supported by a network of thermally insulating tension links as is normal for such coils, the coil is electrically driven thru a standard cryogenic lead system, a normal super insulated radiation shield is provided, etc. Since the stored magnetic energy of such a system is fairly high--seven megajoules, for example, for the 5 tesla system--the coil would be designed to be cryogenically stable to avoid the possibility of damage to the coil in an inadvertent quench. Overall, a synchrocyclotron such as described would be categorized as a new application of existing technology rather than as requiring development of new technology. Information on other details of the design is available. IV. System Comparisons
Given the studies of superconducting synchrocyclotrons described in the previous section and utilizing an earlier study of a room tempe~ature variable energy isochronous cyclotron (ref 3), it is possible to assemble a summary list of proton cyclotrons and synchrocyclotrons which might be of interest for the medical application. Table I lists some of the important parameters which result. In this table Case #1 is based on the 1972 engineering study of a room temperature isochronous cyclotron. Cases 7 and 8 are based on the less complete recent studies of the superconducting synchrocyclotron, described in section III above. Other entries in the Table are interpolated, or estimated on the basis of experience, using applicable scaling rules for cyclotrons. Costs given in Table I are intended to represent the accelerator system only, where the accelerator system is ~aken to include all necessary contrOlS, power supplies, etc. The accelerator also includes a beam extraction system out to a first beam stop at the exit port of the magnet but does not include beam transport elements beyond that point. Costs do not include bUildings, shielding, patient facilities, normal utilities such as cooling water, primary electric service disconnects, etc. Prices do include, for the superconducting systems, a refrigerator-liquifier of capacity adequate to cool down the coil in a 10 day period and to maintain the cold mass at liquid helium temperature on an indefinite basis.

The absolute value of costs in Table I are undoubtedly laboratory dependent and any serious consideration of an actual project should ')bviously involve a careful engineering re-estimate based on the cost structure of the site at which the work would be done. The relative comparisons between different types of accelerators should have much broader general validity and from these comparisons one sees that the superconducting synchrocyclotron would have a very substantial cost advantage relative to the isochronous cyclotron. A similar conclusion of course also holds relative to the room temperature synchrocyclotron (case 6).
-It should be noted that the "isochronous cyclotrons" in the Table produce very much higher beams than are required, i.e. external beams of up to 10 microamps. A variable energy isochronous cyclotron, such as in Cases 1, 2, and 3, also provides beams whose energy can be arbitrarily selected at any value within the specified range. The high current of the isochronous cyclotron is, of course, largely of no help in the therapy application and variable energy is useful but perhaps not to a sufficient degree to justify the sizeable additional cost. With respect to the superconducting synchrocyclotron one notes from the Table that the 7 tesla design (case 8) is slightly less expensive than the 5 tesla design (case 7) and slightly lighter, but the differences are small enough that one might well prefer the more conservative 5 tesla choice, this being the field used in the present generation of superconducting research cyclotrons. Case 9 of Table I is the neutron therapy cyclotron described in Section II, while Cases la, 11, and 12 are possible cyclotrons for so-called "stripped nucleus" therapy, a therapy modality which, though expensive, is expected to combine the benefits of both proton and neutron modalities. Case la, in particular, is the cyclotron now under construction at NSCL for physics applications, except with the variable energy feature suppressed. This cyclotron is expected to come into operation early in 1987 and as a national user facility could be available for biological and medical studies if appropriately persuasive proposals were submitted to the Program Advisory Committee. In conclusion, we note from Table I that a number of the accelerator options are apparently now in a cost range comparable to modern photon therapy units. If this conclusion is confirmed, a major change in the direction of oncology therapy would seem an expected consequence. This expectation follows from the observation that if neutrons, protons, and photons were equal in cost, the photon would never be selected as the radiation of choice, since the proton matches the photon in biological characteristics but is much better in physical characteristics, while the neutron matches the photon in physical characteristics but is significantly better in biOlogical characteristics. There is then no situation in which the photon is superior overall. (In this statement, "physical characteristics" refers to the fraction of dose delivered to the tumor area relative to the fraction delivered to normal tissue, while "biological characteristics" refers to the ability to lethally damage tumor cells relative to the number of normal cells which are lethally damaged.) We then may well be at the beginning of a period of quite significant change in radiation oncology therapy. REFERENCES:

H. Blosser, W. Powers, R. Maughan, C. Orton, D. Reagan, R. Burleigh, E. Jemison, Proceedings of Tenth International Conference on Cyclotrons (1984)431. H. Blosser, D. Johnson, E. Kashy, B. Milton, and J. Riedel, Proceedings of the Tenth International Conference on Cyclotrons (1984)436.
3M Gordon, H. Blosser, D. Johnson, AIP'Conference Proceedings 9(1972)78.
H. Blosser, Proceedings of the Ninth International Conference on Cyclotrons (1981)147.
5F Howard, Oak Ridge National Laboratory Report 2644(1958). 6R Thornton, Proceedings of the CERN Symposium on High Energy Accelerators (1956)413. 7F. Krinen, Proceedings of the CERN Symposium on High Energy Accelerators (1956)425.

- C Fermilab

Fermi National Accelerator Laboratory P.O. Box 500. Batavia, Illinois. 60510
Neutron Therapy Department

July 1, 1985

Re: Medical Workshop on Accelerators for Charged-Particle Beam Therapy held at Fermilab, January, 1985.
We have just learned that Table I from "Cyclotrons and Synchrocyclotrons for Oncology Therapy" by H. Blosser, et al., has been unintentionally omitted. A copy of this table is enclosed. Please add it to your proceedings after page 114.
If you did not pick up a copy of the Fermilab Proton Beam Therapy Facility Proposal at the workshop, they are available upon request.
Phone: (312) 840-3865 From Chicago: 261-1910/ext. 3865 FTS: 370-3865 From Suburbs: 231-6040/ext. 3865

zw -a:

MAIN SUPPORT RING

C'(CLOTRON (

10' POSITION)

I-' I-'

,";t,:;. ,.-1'
SUPPORT RING ROLLER LOCATIONS (ROLLERS OMITTED)
COUNTERWEIGHT ( t 80 POSITION). COUNTERWEIGHT (

10 POSITION)

Figure 2

Figure 3

--.-PEELER MAGNETIC CHANNELS

-ION SOURCE

MAGNET ~YOKE / HELIUM VESSEL
MAGNETIC ~e:::.~-#----. CHANNELS
COIL CRYOSTAT DEE EXTENSION MOUNTING SCREWS
.,.::..t..9lm3B~~~~--,./4~ __ DEE STEM

INSULATORS

LUL~~--77~~----

DEE STEM

YOKE HOLE

~~::::::=

VACUUM ------SEAL DEE STEM INSULATOR
EXTEANAL BEAM VACUUM PUMP

.."'\

RF COUPLING LINE ___ OSCILLATOR

Figure 4

' - - - R F COUPLING LINE
, , \ VACUUM SEAL COIL SUPPORT ,
MAGNETIC CHANNEL POLE TIP RF LINER DEE COIL LEADS AND CRYOGEN SUPPLY & RETURN

"'

VACUUM SEAL TUNER DRIVE DEE STEM INSULATORS

tJV---

VACUUM SEAL TUNER HOUSING DEE STEM INSULATOR
DUMMY DEE COIL CRYOSTAT HELIUM VESSEL

~---+; ~---._._--I~[

;-c=t= \tr

ACCELERATION PLANE

VIBRATING BLADES VACUUM SEAL TUNER DRIVE
COIL SUPPORT ( VACUUM SEAL \ \.--ION SOURCE POLE TIP , ,
YOKE HOLE DEE STEM VACUUM SEAL

Figure 5

Fig. 1.--Cutaway view of 50 MeV, internal target, deuteron cyclotron for neutron therapy. Fig. 2.--Isocentric mounting system for neutron therapy cyclotron. The system provides full 360 degree rotation of the cyclotron. Fig. 3.--The neutron therapy cyclotron system as seen by the physician and patient. The floor below the patient automatically moves aside when the cyclotron moves to locations in the lower quadrant. Fig. 4.--Plan view of a 250 MeV superconducting proton therapy synchrocyclotron (view as seen from Section A-A Fig. 5). For a magnetic field of 5 tesla at the extraction radius, the overall outer diameter of the yoke is 100", the extraction radius is 19" and the central magnetic field is approximately 5.5 tesla (corresponding to a maximum rf frequency of 84 Mhz). Fig. 5.--Vertical section view through 250 MeV superconducting synchrocyclotron (view as seen from Section B-B Fig. 4). For a magnetic field of 5 tesla at the extraction radius, the overall yoke height is approximately 90"

TABLE I:

CYCLOTRONS AND SYNCHROCYCLOTRONS fOR ONCOLOGY THF.RAPY

Energy Casel Par.

External Beam Current (nanoamps) 10,000 10,000 10,000 10,000 2,000 1,000 500
Cyclotron Type Isochronous
Accelerating System dees in gap

'_'_ _ 11_'_'

Hagne.:.t conventional

. (tesla)

Iron Wt. tons 325

~<tO

Pole diam. 125" 136" 126" 125" 76" 130"

Proc. Nadl. Acad. Sci. USA Vol. 89, pp. 5701-5704, July 1992 Chemistry
Laser desorption studies of high mass biomolecules in Fouriertransform ion cyclotron resonance mass spectrometry
(insulin/nitroaniline/matrix-assisted laser desorption/kinetic energy)
TOURADJ SOLOUKI AND DAVID H. RUSSELL
Department of Chemistry, Texas A&M University, College Station, TX 77843
Communicated by Fred W. McLafferty, February 21, 1992 (received for review December 12, 1991)
Matrix-assisted laser desorption ionization is ABSTRACT used to obtain Fourier-transform ion cyclotron resonance mass spectra of model peptides (e.g., gramicidin S, angiotensin I, renin substrate, melittin, and bovine insulin). Matrix-assisted laser desorption ionization yields ions having appreciable kinetic energies. Two methods for trapping the high kinetic energy ions are described: (') the ion signal for [M + H]+ ions is shown to increase with increasing trapping voltages, and (ii) collisional relaxation is used for the detection of [M + H]+ ions of bovine insulin.
insulin (m/z = 5734) requires collisional relaxation of ions prior to their introduction into the ion cell.

MATERIALS AND METHODS

FTICR Mass Spectrometer. The FTICR mass spectrometer used in this study consists of a home-built ICR cell and a vacuum system. The mass spectrometer is equipped with a 7.0-T superconducting Oxford magnet and Extrel FTMS2001 data acquisition system. A Laser Science cartridge-type pulsed N2 laser (model VSL-33ND) is used for ionization. The laser delivers =200 p.J of pulse energy at 337.1 nm per 3-ns pulse. The maximum repetition rate of the laser is 20 Hz with intensity stability of 4% at 10 Hz. A 200.0-cm focal length lens mounted outside the vacuum chamber focuses the laser beam through a UV-grade sapphire window to a second focusing lens of 20.0-cm focal length that is mounted onto the ICR cell (Fig. 1). The second lens focuses the collimated laser beam to a spot size approximately 20 Am by 20 ,Am onto a direct insertion sample probe surface. The laser beam is incident on the target surface in a perpendicular fashion. The desorbed ions are trapped in a 16-cm-long, 4-cm-diameter cylindrical cell made of oxygen-free copper. For typical experiments, the sample probe tip is placed about 5 mm outside the ion cell, and ions are transferred into the cell through an aperture of 0.45 cm in diameter. MALD ionization mass spectra were obtained at trapping voltages of 1-9 V (6.25 V/m-56.25 V/m). The bias voltage applied to the sample probe was kept constant at 0.0 V. Fig. 2 illustrates the sequence of events for a single laser shot. First, positive and negative ions inside the cell are removed in quench events. The removal of ions is accomplished by first applying +9.75 V to both trap plates for 50 ms (negative ion quench), which is then followed by a 50-ms positive ion quench event where trapping plates are set at -9.75 V. After a delay of 50 Aus (D1) a -5 V logic pulse from the FTICR cell controller initiates a single firing of the laser. At the time of the laser pulse, the trapping plates are at ground potential. A second variable delay (D2) permits ions to enter into the cell before the trapping voltage (VTR) is applied to the trap plates. Positive or negative ions can be selectively trapped by applying positive (for positive ions) or negative (for negative ions) voltages to the trap plates. Ion detection is accomplished by chirp excitation with a radiofrequency field. Ion detection is performed at medium resolution (16,000-32,000 data points) in the direct broadband mode.
Abbreviations: VTR, trapping voltage; MALD, matrix-assisted laser desorption; ESI, electrospray ionization; FTICR, Fourier-transform ion cyclotron resonance; ICR, ion cyclotron resonance. *Laude, D. A., Jr., Riegner, D. & Beu, S. C., Toward Quantitative Ion Dissociation Measurements in the FTICR Cell, Federation of Analytical Chemistry and Spectroscopy Societies and Pacific Conference on Chemistry and Spectroscopy (FACSS XVIII/PCCS XIII), September 1991, Anaheim, CA.

Ionization methods such as californium plasma desorption (1), matrix-assisted laser desorption (MALD) (2), and electrospray ionization (ESI) (3) have greatly expanded the role of mass spectrometry of high mass biomolecules. Biomolecules having molecular masses >100,000 Da can be ionized to yield a single-charged (e.g., [M + H]+) or multiple-charged (e.g., [M + nH]n+) ion. Although most of the MALD and ESI work has utilized time-of-flight, quadrupole mass spectrometers and/or quadrupole ion traps, MALD (4) and ESI are adaptable to Fourier-transform ion cyclotron resonance (FTICR). The advantage of FTICR (and the quadrupole ion trap) is that tandem mass spectrometry experiments [e.g., collision-induced dissociation and photodissociation (5)] and ion molecule reactions can be implemented for structural characterization. The best high mass performance [-60,000 mass resolution at m/z 6000 (6)] of FTICR has been achieved with laser desorption ionization of nonpolar polymers. McLafferty and coworkers (7) have made considerable progress with the development of ESI for FTICR, and a highsensitivity ESI/FTICR system was recently described.* Several groups (9, 10) have suggested that the performance of FTICR with biological molecules (samples that have relatively low ionization probabilities) and molecular masses greater than -2500 Da may be adversely affected by inefficient trapping of the ions or inherent limitations of the ion detection. For example, the mass resolution for samples with molecular masses greater than -2500 Da is far less than the theoretically predicted values (11). On the basis of our experience with desorption ionization and high mass FTICR, we suggested that the major problem with high-resolution FTICR at high mass is due to the high kinetic energies of desorbed ions (9). We have recently evaluated two experimental methods for improving the trapping efficiency of high mass ions: (i) the effects of increasing the electric trapping field and (ii) the effects of collisional relaxation of the laser-desorbed species prior to trapping the ions in the ion cyclotron resonance (ICR) cell. In this article we present the results from both experimental methods. The results show that increasing the trapping potentials improves the ion trapping efficiency, but detection of [M + H]+ ions of bovine
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. 1734 solely to indicate this fact.
Chemistry: Solouki and Russell
Proc. Natl. Acad. Sci. USA 89 (1992) resolution are due to the volatility of the matrix; matrices such as nicotinic acid and sinapic acid are simply too volatile for the (10-8 torr; 1 torr = 133 Pa) vacuum requirement for FTICR. It is also important to consider the effects ofnonthermal kinetic energies on the detection of high m/z ions (13). Recent work by Beavis and Chait (14), Pan and Cotter (15), and Standing and coworkers (16) suggests that the ions formed by MALD have relatively large kinetic energies. For example, Beavis and Chait (14) report an average axial velocity of 750 m/s for polypeptide molecular ions (independent of m/z) produced by MALD. A value of 2 eV of axial translational energy for molecular ions of [Arg8]vasopressin (m/z = 1085) with an energy spread of 8 eV (full width at half-maximum) has been reported.t Measurements of Pan and Cotter (15) also suggest that higher mass ions are desorbed with increased kinetic energies. The initial velocity distribution measured by Standing and coworkers (16) indicates a radial velocity of about 2.4 eV and a much higher axial velocity (e.g., 50 eV for bovine insulin). Fig. 4 contains the MALD/FTICR mass spectrum of gramicidin S, angiotensin I, renin substrate, and melittin. In each case the spectra were obtained at different trapping voltages, and the optimum trapping voltage increases as the mass of the analyte is increased (see Fig. 5). Note that gramicidin S is efficiently trapped at voltages >0.5 V, whereas melittin requires a minimum trapping voltage of =7 V. Ion yields (Fig. 5) represent relative values of ions of interest trapped in the cell and are normalized to the largest peak in the spectrum. In an effort to check the reproducibility of ion yields, at least three experiments (each experiment being an average of 50 laser shots) were performed for each data point; variations between measurements were <10%6. It is interesting to note that the optimum VTR of each sample correlates reasonably well with the kinetic energy measurements reported by Beavis and Chait (14). However, the optimum VTRs change as the laser power density changes. At lower laser power densities, the number of desorbed species decreases. A decrease in the number of desorbed species results in (i) decreased ion yield and (ii) decreased number of gas phase collisions. Both conditions (i and ii) would require higher VTRs. First, higher VTRs are required to trap a measurable number of ions at decreased ion yields. Second, a decreased number of postdesorption collisions alters the translational energy distributions of desorbed partides (17-21), and higher VTRs are required. The effects of using high VTRs to trap high kinetic energy ions are manifested in two ways. First, the absolute abundance of high m/z [M + H]+ ions changes with VTRs. It is worth noting that the absolute signal intensity increases as the VTR is increased beyond the minimum voltages required for trapping a particular ion (i.e., larger numbers of ions are trapped in the ICR cell as the VTRs are increased). For instance, trapping of melittin [M + H]+ ions requires +7 V, and increasing the VTR results in an increase in the [M + H]+ ion signal (Fig. 5). On the other hand, at much higher VTR (e.g., greater than +20 V), ion signals are lost. The higher voltage ion trapping method works well only for ions up to approximately m/z 3000. The upper mass limit for this method of ion trapping is determined by the axial ion ejection and increased magnetron radius of the trapped ions (22-24). Second, the experimentally measured ICR frequencies of ions vary with the VTRs (22, 25, 26). It has been shown that "screened" (26) and "field-corrected" (8, 27) ion traps can be used to minimize the frequency shifts induced by trapping potentials. To account

N2 LASER

FIG. 1. Diagram of the MALD/FTICR s3ystem.
In the pulsed-valve experiments, the gaseous reagents (iert collisional gases such as He and Ar) are adrr utted into the vacuum system by using a General Valve (motdel 9-337-900) pulsed valve. The timing of the reagent gas pullse sequence is controlled in such a manner that molecules of thLe collision gas collide with desorbed species immediately after the laser is fired. "Waiting room" experiments of laser-desorbed ions are accomplished by placing the sample probe iinside a brass housing. A close-up view of the probe, trapiping cell, and external housing (waiting room) is shown in Fig 3. The direct insertion sample probe forms a tight seal againsAt the housing, and ions and neutrals formed by the incident laLser irradiation escape via a 0.25-cm diameter aperture. After lasser irradiation, desorbed species inside the waiting room underggo collisions to lose most of their initial kinetic energies; tranislationally relaxed ions are transferred into the ICR cell and nnass analyzed. The same experimental sequence (as described,above) is used for waiting room experiments except that the sample probe bias voltage is set at positive potentials to forcse the positive ions into the trapping cell. The volume inside the waiting room can be varied by moving the sample probe. Sample Preparation. All chemicals were c btained from Sigma and were used without any further purificration. Analyte samples (gramicidin S, angiotensin I, renin subs trate, melittin, and bovine insulin) of 0.1 mg were dissolved in 1 ml of ethanol. The matrix was prepared by dissolving 1 mg ofF4-nitroaniline in 1 ml ofethanol. Mixtures (1:1, vol/vol) of analyte and matrix solution (-1 tl of each) were delivered to the sajmple probe tip surface. The solvent was allowed to evaporate, leaving a thin film of matrix/analyte on the probe tip. RESULTS AND DISCUSSION Previous studies of MALD with FTICR on bion iolecules such as oligonucleotides and peptides were succe.ssful for only relatively low molecular mass samples (e.g., 10 00-2000 Da or less) (10, 12). In part, the limited mass rarige and mass
MOSITIVE & SUTIVE 10N

QuaN1H

I~~r LLASER FIRE

ION TRAF

INTO THE CEL

ION TRAPPING

RF EXCITATION

ION DETECTION

tPan, Y. & Cotter, R. J., Measurement of Initial Translational
FIG. 2. Event sequence for MALD/FTICR experiments.
Energies of Peptide Ions in Laser Desorption/Ionization Mass Spectrometry, Proceedings of the 39th ASMS Conference on Mass Spectroscopy and Allied Topics, May 19-24, 1991, Nashville, TN, pp. 364-365.
Solouki and Russell Chemistry: Nati. Acad. Sci. USA 89 (1992) ~Proc. Chemistry: Solouki and Russell

Trapping

FIG. 3. (Upper) Diagram of the trapping cell and external waiting room with the sample probe inserted. (Lower) Cross section of the waiting
for the deviation of ICR frequency with trapping potentials, it is necessary to calibrate the mass axis with each VTR setting. The influence of the VTR on the detectable yield for [M + H]+ ions of large peptides is consistent with our previous studies that indicate that the kinetic energy of high mass ions
formed by desorption ionization methods adversely affects the performance of the ICR (9). We have attempted collisional relaxation experiments by introducing a collision gas (He) through a pulsed valve to cool the translationally excited ions; however, these experiments were largely unsuccessful.

Gramicidin S

[M + -Na]+

Renin substrate

r, ct la r.0

L--hjwL-l_

i &-A.a 1.
I.L"m ILI ~.A& AAAi I1

m/z [M

Akngiotensin I

100 -D

Melittin

Lif A-1A

500 1000

A~~~~A

FIG. 4.

MALD/FTICR

at VTR

spectra (50

scans,

12%.5)

2 V. (C)

substrate (m/z

1 V. (B) Angiotensin I (m/z 1141.5) at VTR 32,000 data points). (A) Gramicidin S (m/z 9 V. 4-Nitroaniline was 6 V. (D) Melittin (m/z 2846.6) at VTR 1759.0) at VTR
the matrix used for each spectrum.
Proc. Natl. Acad. Sci. USA 89 (1992)
probe. Ions entering the ICR cell have lost most of their kinetic energies and can be trapped at lower (less than +9 V for positive ions) voltages. In conclusion, we have shown that high VTRs can be used to trap high mass/high kinetic energy ions produced by MALD ionization. This result is consistent with the measured kinetic energy of ions formed by MALD (14). The limitation in using high VTRs is due to complications ofthe ion motions (e.g., increased magnetron motion). We have also shown that to circumvent the problems associated with higher VTRs, collisional relaxation in a small volume chamber can be used effectively for trapping high mass ions. Therefore, detailed studies of high mass biomolecule ions produced by MALD ionization can be performed.

TRA1ING VOLTAGE

FIG. S. The normalized relative ion abundances plotted versus VTRs for gramicidin S (m), angiotensin I (+), renin substrate (*), and melittin (o).
This research was supported by the National Science Foundation (CHE 8821780) and the National Institute of General Medical Sciences (RO1-GM33780-5). The Texas Advanced Technology Program provided funds for purchase of the 7-T superconducting magnet and the Extrel FTMS-2001 computer system. The laser system was purchased from funds provided by U.S. Department of Energy, Division of Chemical Sciences, Office of Basic Sciences (DE-AS05-85-ER13023).
1. Sundqvist, B. & Macfarlane, R. D. (1985) Mass Spectrom Rev. 4, 421. 2. Karas, M., Bachmann, D., Bahr, D. & Hillenkamp, F. (1987) Int. J. Mass Spectrom. Ion Processes 78, 53. 3. Fenn, J. B., Mann, M., Meng, C. K. & Wong, S. F. (1990) Mass Spectrom. Rev. 9, 37. 4. Nuwaysir, L. M. & Wilkins, C. L. (1991) SPIE Appl. Spectrosc. Mater. Sci. 1437, 112-123. 5. Hunt, D. F., Shabanowitz, J. & Yates, J. R., III (1987) J. Chem. Soc. Chem. Commun. 8, 548-550. 6. Wilkins, C. L., Wiel, D. A., Yang, C. L. C. & Ijames, C. F. (1985) Anal. Chem. 57, 520-524. 7. Loo, J. A., Quinn, J. P., Ryu, S. I., Henry, K. D., Senko, M. W. & McLafferty, F. W. (1992) Proc. Natl. Acad. Sci. USA 89, 286-289. 8. Wang, M. & Marshall, A. G. (1990) Anal. Chem. 62, 515-520. 9. Hanson, C. D., Castro, M. E., Russell, D. H., Hunt, D. F. & Shabanowitz, J. (1991) in Fourier Transform Mass Spectrometry: Evolution, Innovation, and Applications, American Chemical Society Symposium Series 359, ed. Buchanan, M. V. (Am. Chem. Soc., Washington), pp. 100-115. 10. Hettich, R. L. & Buchanan, M. V. (1991) J. Am. Soc. Mass Spectrom. 2, 22-28. 11. Hunt, D. F., Shabanowitz, J., Yates, J. R., III, Zhu, N., Russell, D. H. & Castro, M. E. (1987) Proc. Natl. Acad. Sci. USA 84, 620-623. 12. Hettich, R. L. & Buchanan, M. V. (1991) J. Am. Soc. Mass Spectrom. 2, 402-412. 13. Hanson, C. D., Kerley, E. L., Castro, M. E. & Russell, D. H. (1989) Anal. Chem. 61, 2040-2046. 14. Beavis, R. C. & Chait, B. D. (1991) Chem. Phys. Lett. 181, 479484. 15. Maruyma, S., Anderson, L. R. & Smalley, R. E. (1990) Rev. Sci. Instrum. 61, 3686. 16. Ens, W., Mao, Y., Mayer, F. & Standing, K. G. (1991) Rapid Commun. Mass Spectrom. 5, 177-123. 17. Fiegl, P., Schueler, B. & Hillenkamp, F. (1983) Int. J. Mass Spectrom. Ion Phys. 47, 15-18. 18. Levis, R. J. & Romano, L. J. (1991) J. Am. Chem. Soc. 113, 7802-7803. 19. Cowin, J. P., Auerbach, D. J., Becker, C. & Wharton, L. (1978) Surf. Sci. 78, 545-564. 20. Noorbatcha, I., Lucchese, R. R. & Zeiri, Y. (1987) J. Chem. Phys. 86, 5816-5824. 21. Cousins, L. M., Levis, R. J. & Leone, S. R. (1989) J. Chem. Phys. 91, 5731-5742. 22. Huang, S. K., Rempel, M. L. & Gross, M. L. (1986) Int. J. Mass Spectrom. Ion Processes 72, 15-31. 23. Grosshans, P. B. & Marshall, A. G. (1990) Int. J. Mass Spectrom. Ion Processes 100, 347-379. 24. Wang, M. & Marshall, A. G. (1990) Int. J. Mass Spectrom. Ion Processes 100, 323-346. 25. Beauchamp, J. L. & Armstrong, T. (1969) Rev. Sci. Instrum. 40, 123-128. 26. Wang, M. & Marshall, A. G. (1989) Anal. Chem. 61, 1288-1293. 27. Hanson, C. D., Castro, M. E., Kerley, E. L. & Russell, D. H. (1990) Anal. Chem. 62, 520-526.

The failure of these pulsed-valve experiments is in part due to inefficient cooling of desorbed ions. In order to have sufficient numbers of gas phase collisions for translational cooling of desorbed ions, a large dose of reagent gas must be introduced into the ion cell. Because ion detection in F1TICR requires a low vacuum (e.g., <1o-8 torr), removal of large doses of reagent gas requires longer pumping times. As the pumping time (after ion formation) increases, the probability of losing ions increases. A method that has been very successful is to perform MALD in a small chamber (waiting room). Laser ablation produces a high density of neutral and charged species, and, in a confined space (=0.25 CM3), the pressure is sufficiently high for collisional relaxation to occur. This is the same general approach involved in the waiting room experiment of Smalley and coworkers (15). The FTICR mass spectrum of bovine insulin shown in Fig. 6 was obtained by using the waiting room to relax translationally excited insulin ions. In this experiment, laserdesorbed species undergo collisional relaxation before entering the ion cell (Fig. 3). Collisionally thermalized ions are drifted into the cell by applying a bias voltage to the sample
FIG. 6. FTICR mass spectrum (1 scan, 16,000 data points) of bovine insulin (m/z = 5734) obtained by using the waiting room for collisional relaxation.