Design of a Superconducting Cyclotron for Exotic Nuclei ... tions is not to build a suitable...

Mario Maggiore

Design of a SuperconductingCyclotron for Exotic Nucleiproduction and Therapy

Dottorato di Ricerca in FisicaXVIII ciclo

Advisors:prof. Emilio Mignecodott. Luciano Calabretta

Coordinator:prof. Francesco Riggi

Universita degli Studi di Catania

Dipartimento di Fisica e Astronomia

This study has been made in collaboration with

Laboratori Nazionali del SudIstituto Nazionale di Fisica Nucleare64, Via S.Sofia I-95123Catania, Italy

ACKNOWLEDGMENTS

I would like to thank my advisors Prof. E.Migneco and Dr. L.Calabrettawithout whom this thesis would not be there.

My special thanks go to Dr. L.Calabretta for his continuous ad-vice and comments on my research work. His guidance and deeplyknowledge of the accelerators’ physics have made the completion ofthis work possible.

I also want to thank the LNS accelerator staff, in special way Dr.D.Rifuggiato for giving me the opportunity to train with the Super-conducting Cyclotron operations at LNS.

I want to thank Drs D.Battaglia, A.Calanna, D.Campo, L.Piazzaand INFN Genova people, without their work this thesis would beseveral pages shorter.

I thank Drs A.Caruso, M. Di Giacomo and J. Sura for their helpand advice about the Radio Frequency studies.

I would like to thank the staff of the Physics Department of theUniversity of Catania and of the INFN for their support during mygraduate career.

I am grateful to my parents for their patience and for the way theyhave always accepted my decisions.

Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Accelerators in Cancer Therapy . . . . . . . . . . . . . . . . . . . . 52.1 Requirements of ion beams for cancer treatment . . . . . . 92.2 Type of Accelerators for Therapy . . . . . . . . . . . . . . . . . . . 13

2.2.1 Cyclotrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.2.2 Linear accelerators . . . . . . . . . . . . . . . . . . . . . . . . . . . 142.2.3 Synchrotrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.3 Facilities treating patients with Hadrons . . . . . . . . . . . . . 142.3.1 Status of the Hadrontherapy in Italy . . . . . . . . . . . 17

3 SCENT project description . . . . . . . . . . . . . . . . . . . . . . . . . 213.1 The physics with Exotic Nuclei . . . . . . . . . . . . . . . . . . . . . 21

3.1.1 Status of the EXCYT project in Catania . . . . . . . . 253.2 The Preliminary Cyclotron’s design as driver for Exotic

Nuclei Facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273.2.1 Main parameters of the machine . . . . . . . . . . . . . . . 27

3.3 The Final Cyclotron’s design for Medical Applications . 313.3.1 Main parameters of the machine . . . . . . . . . . . . . . . 33

4 Magnet Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374.1 Main parameters of the Cyclotron . . . . . . . . . . . . . . . . . . . 374.2 Magnetic Design (Analytical approximation) . . . . . . . . . . 384.3 Stability conditions and spiral definition (matrix method) 404.4 Preliminary design with 3D code . . . . . . . . . . . . . . . . . . . . 424.5 Magnetic Structure description . . . . . . . . . . . . . . . . . . . . . 474.6 Superconducting Coils . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

4.6.1 Conductor Definition . . . . . . . . . . . . . . . . . . . . . . . . . 55

6 Contents

4.6.2 Conductor Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . 574.7 The Magnetic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 604.8 Magnetic Field Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 634.9 3D simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694.10Fine Tuning of the Magnetic Field . . . . . . . . . . . . . . . . . . 744.11Magnetic Forces calculation . . . . . . . . . . . . . . . . . . . . . . . . 75

5 Radio Frequency System . . . . . . . . . . . . . . . . . . . . . . . . . . . 795.1 The RF resonant cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 795.2 The modelling method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 815.3 The cavity geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 875.4 The simulation results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 885.5 Coupler Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925.6 Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

6 Injection and extraction design . . . . . . . . . . . . . . . . . . . . . 956.1 Injection line study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 956.2 Inflector Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

6.2.1 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 986.3 Central Region Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1026.4 Preliminary study of the extraction process . . . . . . . . . . . 106

6.4.1 Extraction by ED . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1076.4.2 Extraction by Stripping . . . . . . . . . . . . . . . . . . . . . . . 107

6.5 Beam losses and vacuum requirements . . . . . . . . . . . . . . . 110

7 Conclusions and Further Developments . . . . . . . . . . . . . 113

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

1

Introduction

The cyclotron is not only a very compact and hence economical ac-celerator but also a versatile tool. This makes it the preferable choicefor both, research, where flexibility is important, and applied work,where it is important to optimally adapt the accelerator to its partic-ular use. Past decades have shown that cyclotrons can well be adaptedto a large variety of applications in very different fields. Cyclotronsworldwide are employed in scientific experiments in atomic, nuclear,particle and solid-state physics, but also in medical and industrial ap-plications. As the most compact accelerator the cyclotron is the idealchoice whenever size or economical factors become essential.

Since the 1931, when the first cyclotron was developed by E.O.Lawrence and M.S. Livingstone [1], the development of this kind ofmachine has seen many limitations that seemed to hinder further de-velopment and a successful use of this type of accelerator.New strate-gies were found and new ideas were employed in order to remove thelimiting problems. In 1937 H.E.Bethe and M.E.Rose [2] predicted that12 MeV protons or 34 MeV alpha particles would be the highest ob-tainable energy from a cyclotron, because the flat pole in the classicalcyclotrons could not provide both axial focusing and the isochronousacceleration of relativistic particles. In response to user demand (thepredicted existence of pions and muons) the concept of frequency mod-ulation was introduced. In 1945 the first synchrocyclotron was builtand allowed to accelerate beams up to 1 GeV, but intensities werelimited due to the pulsed operation. Only with the introduction ofThomas focusing by azimuthally varying magnetic fields both limita-tions could be removed [3] [4] . Now the sector focused cyclotron was a

2 1 Introduction

unique and versatile instrument that could be tailored to a large vari-ety of uses. Separated sector cyclotrons were built in order to producevery high fluxes of secondary particles (pions, muons and neutrons)and also used in large facilities for heavy ion physics. Superconductingcyclotrons allowed to reduce the size and weight of cyclotron magnetsby more than an order of magnitude which was important for largeheavy ion beam facilities and an essential advantage in some typicalmedical applications. Finally the exploding number of cyclotrons forcommercial application demonstrates the actual trend of cyclotron de-velopment.

In the present job, made in collaboration with the LaboratoriNazionali del Sud (LNS) of Catania [5], the feasibility study of anew Superconducting Cyclotron is discussed.This machine is the subject of the project SCENT, approved by theIstituto Nazionale di Fisica Nucleare (INFN) in 2002. SCENT is theacronym of Superconducting Cyclotron for Exotic Nuclei productionand Therapy. The aim of the preliminary design of SCENT was torealize an high intensity machine able to deliver protons (extractedby means of the stripping of H+

2 ) and light ions, as Carbon, used asprimary beams to produce exotic nuclei in the radioactive ion beams(RIBs) facilities. The purpose was to upgrade the EXCYT [6] facilityat LNS.In a second step, the design of the machine was modified to meet themedical application requirements. The challenge of medical applica-tions is not to build a suitable cyclotron, but to find the technicalsolutions, that best account for economical factors like low price, highreliability and simplicity of operation and maintenance.

Finally, a superconducting compact cyclotron able to accelerateions with mass to charge ratio of 0.5, up to 250 AMeV was designedand developed during this PhD work.

Here a brief overview of the sections is presented.

No section reviews the basic concepts underlying isochronous cy-clotron physics. This because a wide presentation about the Cy-clotrons’ Physics had been already done in my previous thesis de-

1 Introduction 3

gree [7].

A wide discussion concerning the status of the Accelerators in Can-cer Therapy is reported in chapter 1. A brief description of the fun-damental concepts concerning the hadrontherapy and a description ofthe present facilities worldwide have been done.

The section 2 describes the SCENT project, and the two differentconfigurations of the machine are presented. An introduction aboutthe Physics with Exotic Nuclei is also discussed.

From the section 3 up to the section 5, the design methods usedto define the Magnetic Field of the machine, the RF system and theinjection and extraction characteristics are outlined.

Conclusions and further developments are discussed in section 7.

2

Accelerators in Cancer Therapy

Radiation therapy has become one of the most important modalitiesin the treatment of cancer. It is estimated that a person has one out ofthree chance to be confronted with the disease during his or her life-time and that less than half of them will be cured. While the surgeryis still the most successful treatment, radiation therapy either alone orin combination with other modalities contributes to about 40% of theoverall cure rates. It is interesting to note that chemotherapy aloneresults in a rather small part of cancer cures; it is used mostly as anadjuvant therapy. All other modalities contribute only a few percentto the cure rates.Ideally, the objective of any cancer treatment method is to remove ordestroy the tumor while preserving at the same time the healthy tissueas much as possible. It was with this idea in mind that almost a hun-dred years ago low-energy X-rays began to be used for this purpose,although their penetration was poor and therapeutic effect debatable.In early 1920s, radium units came into use, producing deeper pene-trating gamma rays; this was followed by electron accelerators pro-viding higher energy X-rays. Nuclear reactors made radioactive cobaltsources available and they became a standard gamma-ray source forradiotherapy, used until the present (e.g., gamma knife). Most modernand very widely used machines for X-ray therapy are compact linearaccelerators and it is estimated that there are up to 4000 of themaround the world. Over the years, this technique has been constantlyimproved, the machines have been adapted to the hospital environ-ment and the delivery of the radiation to the tumor has become moreand more accurate, trying at the same time to spare the healthy tis-sues. However, there are still many cases where it is not possible to

6 2 Accelerators in Cancer Therapy

avoid irradiation of critical organs in the vicinity of the tumor; themaximum dose allowed for critical organs would in such cases limitthe dose given to the tumor, leading to a possible failure of the localcontrol.About fifty years ago, R. Wilson remarked that the Bragg peak ofmonoenergetic protons (and other, heavier ions) would allow the ra-diation dose to be preferentially delivered at the end of their path, inthe tumor itself where most damage has to be done. By modulatingthe proton (or ion) energy it would, in principle, be possible to irradi-ate the whole volume of the tumor with a uniform and sufficient dose,while keeping the dose delivered to other organs at a lower value. Thischaracteristic, together with a high lateral beam accuracy, is the basisof conformal treatment of tumors, which is an important step towardthe ideal method. Since this first proposal, there were a number ofproton machines either adapted or specifically built for tumor treat-ment.

Fig. 2.1. Depth dependence of the deposited dose for different radiations. Because ofthe Bragg peak the dose distribution is ’inverted’ with respect to the almost exponential,and much less favourable, behavior produced by a beam of high energy protons.

The most recent and quite promising introduction into the rangeof types of radiation for cancer treatment have been energetic ions in

2 Accelerators in Cancer Therapy 7

the mass range from carbon to neon. They will be referred to as lightions, although in the medical literature they are usually called heavyions. In addition to the advantage of showing a Bragg peak which hasa similar characteristic of protons, and an even better lateral beam ac-curacy than protons, ions have other characteristics which could makethem more suitable for treatment of some types of tumors than anyother radiation. The linear energy transfer (LET) or the rate of en-ergy deposition along the path of a particle is higher for light ions (fastneutrons have a similar property) than it is for conventional radiation,including protons; the relative biological effectiveness (RBE) tends tobe higher if LET values are higher (see Fig.2.3). Furthermore, sometumors cells are anoxic and as such are more resistant to conventionalradiation due to the oxygen effect, characterized by the oxygen en-hancement ratio (OER)(see Fig.2.2). There are also indications thatthe effects of light ion radiation in the tumor do not depend as muchon the cell cycle as they do for conventional radiation.However, possible advantages of light ions compared to conventional

Fig. 2.2. Range of measured oxygen enhancement ratio (OER) values for different typeof radiation.

radiation result in a more complex system for beam production and,especially, for beam delivery to the patient. With high LET particlesand with a large part of their energy delivered at the end of the track,in the Bragg peak, it becomes extremely important to properly adjustnot only the shape of the beam but its energy and the time spent


Fig. 2.3. Range of experimental data for the relative biological effectinees (RBE) factoras a function of linear energy transfer (LET) values.

irradiating a certain part of the tumor as well; otherwise, healthy tis-sues may be exposed unnecessarily while the tumor may not get therequired dose. The proper utilization of light ions requires a state-of-art application of medical diagnostics (CT, MRI, PET) to determinethe exact shape and location of the tumor, a full computer controlof the accelerator and beam delivery system and a fast and accuratemeasurement of the beam dose delivered at any instant to the patient.Until very recently, this complexity of the system has been one of thereasons why light ions have found a very limited use in medicine, par-ticularly in the treatment of cancer, so that at present there is just twodedicated facility in the operation (Chiba, Japan and GSI, Germany).Another reason for the lack of interest was the fact that in the past thefew accelerators capable of producing light ion beams with parametersadequate for medical applications were designed for a totally differentpurpose (nuclear and particle physics), with energies and intensitiesnot matched to the needs of patient treatment, complicated to useand expensive to operate. Compared to light ion accelerators, electronlinacs for photon production have enjoyed a long history of develop-ment and present designs are well adapted to a hospital environment.

2.1 Requirements of ion beams for cancer treatment 9

2.1 Requirements of ion beams for cancertreatment

At the Chiba facility, most of the research of the effects of light ionson cells and almost all clinical trials have been done with ions up toneon, although the their facility has been designed for ions up to argon.There is a general agreement that carbon ions offer a very good com-promise between advantages in the treatment (a very favorable ratioof the dose delivered to the tumor and the entrance dose, good ra-diobiological properties) and disadvantages that should be minimized(fragmentation, distal dose). For a certain desired penetration depth(or position of the Bragg peak), the energy of ions delivered to thepatient will depend on the ion species (Fig. 2.4); the ion energy willthen determine the size of the machine and its cost. While for protons

Fig. 2.4. Range energy curves for several ion species of interest in cancer therapy.

an energy of 250 MeV is sufficient for irradiation of tumors seatedup to a depth of 30 cm (water equivalent), light ions require a higherenergy for the same penetration. Carbon ions with an energy of 290AMeV will penetrate only 15 cm deep and for 30 cm an energy above


400 AMeV would be required. For even heavier ions, such as neon,energies about 650 AMeV are needed. Once the range of ion specieshas been selected, the top energy of the heaviest ion will determinethe size of the machine and its cost. There are some trade-offs avail-able in considering these parameters: a machine designed for a certainion species and the full penetration depth (highest energy envisaged)is capable of delivering even heavier ions at the similar or somewhatlower energy per nucleon (the maximum energy will depend on thecharge to mass ratio of ions); although for heavier ions the penetra-tion would not be as deep, they could still be used for treatment ofthose tumors located closer to the surface of the body. For compari-son, at an energy of 400 AMeV carbon ions would have a penetrationdepth of 28 cm, oxygen ions 20 cm and neon ions only 17 cm. Ionslike neon and heavier have a relatively higher plateau (dose deliveredbetween the entrance and the tumor) and are preferred for shallowtumor locations to limit the damage to healthy tissues; a lower energyper nucleon may, therefore, still be satisfactory. Should a deeper rangebe required, the machine should be designed for a higher energy. Therewill be additional requirements on the precision of the desired beamenergy (position of the Bragg peak) and on the allowed energy spreadof the beam (broadening of the Bragg peak); it is possible to reducethese two effects by using energy defining collimators in the transportline, but this necessarily results in the loss of beam intensity. However,modern accelerators have already achieved the desired accuracy in en-ergy setting and control (0.1% or better); the beam energy spread isalso within the limits required for treatment. The knowledge of theproperties of tissues in the path of the beam is more critical becausethis will affect the range (or position of the Bragg peak) and will haveto be included into the planning of the treatment.The treatment of larger volume tumors with protons or light ions re-quires scanning of the volume with ions of varying energy, to achieve aspread-out Bragg peak. There are two methods to achieve the modula-tion of beam energy, one utilizes a fixed output energy from the acceler-ator (or, possibly, a few energies at large steps) when the changes of thebeam energy are accomplished by energy degraders in the treatmentroom (passive systems), while the other is based on the energy mod-ulation of the accelerator itself (active scanning). The first method,does not pose any additional requirement on the accelerator and thebeam transfer line except for a fixed and steady output energy, but

2.1 Requirements of ion beams for cancer treatment 11

it does need very carefully designed energy degraders as well as ele-ments for collimation; it may, however, result in a deterioration of thebeam quality when passing through the degrader (scattering, fragmen-tation) and in additional background radiation in the treatment room.For some types of accelerators this is the only method applicable. Theother method, by modulating the output energy of the accelerator it-self, moves the burden of complexity from the final part of the transferline to the machine itself. This method allows the tumor volume to bescanned a small element by a small element (voxel), always with anappropriate energy and intensity. In spite of the additional complex-ity of the active scanning, further developments of the accelerator andbeam transport control systems will soon make possible the improve-ment of this method for tests and patient treatment.Finally, in order to accurately deliver the required dose throughoutthe tumor, the accelerator has to provide the desired ion species withvery little contamination. In the case of light ions, accelerators maynot be able to distinguish between different species having the samecharge-to-mass ratio and the selection will have to be done in the in-jector stage.

Another important parameter of the beam is the intensity. Beam inten-sity (or flux), required from the accelerator, is determined by severalfactors, among them the desired duration of the treatment, prescribeddose, method of beam energy modulation and size and location of thetarget. To minimize the effects of the motion of the patient duringthe irradiation it is desirable that treatment times be no longer thanat most few minutes (in some cases it may be even necessary to syn-chronize the radiation pulses with breathing or heart beats). Thereis some flexibility in the choice of the length of the treatment time;one may e.g. ask for a dose of 5 Gy/min to be delivered to a targetvolume of 2 liters, at the full beam energy. Corresponding values oflight ion beam intensities for such a rate of irradiation are of the or-der of 109 particles per second, less for heavier ion species because oftheir higher relative biological efficiency. The intensity of the beam atthe exit of the accelerator will have to be higher because of losses inbeam handling and transport from the machine to the patient. Passivesystems in principle will have higher losses, possibly up to 80%, butone cannot expect a transport efficiency much better than 50% evenwith active scanning systems. The latter systems, however, have the


advantage that the fraction of the beam, which is not delivered to thepatient, is dumped outside the treatment room without contributingto the background radiation. The accelerator, together with the ionsource and the injector stage, should be designed for the required out-put beam intensities over the whole range of ion species. For a certainmachine design, the beam output will be lower for heavier ions, but sowill be the required target beam intensity. Beam intensities given inTable 2.1 should be considered as upper limits; even lower values couldbe acceptable if this would lead to a simpler and less costly design orto a better beam quality because many tumors are smaller than twoliters or a longer irradiation time could be allowed.An accurate dose delivery over the volume of the tumor requires a

Table 2.1. Values of the beam intensities required for the treatment of patients.

well defined time structure of the beam from the accelerator. If a pas-sive beam delivery system is used, the target is irradiated slice by slicewith a broad beam and the time structure is less critical as long as itis possible to monitor and control the time a slice is exposed to thebeam. An active beam scanning system poses stricter requirements onthe time structure unless there is an on-line beam detection system toaccurately measure the dose delivered to any volume element and todeliver a signal to move the beam to the next element once the ap-propriate dose has been reached. Without such an on-line system, theextracted beam from the accelerator should be as steady as possible,with intensity fluctuations within a few percent only, which leads tovery tight tolerances on accelerator and beam line elements and theirpower supplies.In addition to the stability of the beam, there are other considera-tions. Regardless of the type of the machine, whether it is a cyclotron,linac or synchrotron, the beam will have an inherent intensity struc-ture related to the rf accelerating voltage; depending on the type ofthe machine the beam may be available with a macroscopic duty fac-tor less than 100%. A cyclotron operates with a duty factor of 100%

2.2 Type of Accelerators for Therapy 13

and there is an rf structure corresponding to the frequency of the ac-celerating voltage; the beam intensity control is done best at the lowenergy end, ion source or ion injector. A linac usually has a very lowduty factor, with pulses of a few milliseconds duration, but a highintensity and, again, showing an rf structure corresponding to the ac-celerating voltage. The low duty factor makes linacs less suitable forion delivery systems. Finally, a synchrotron with a slow extractionsystem can have a duty factor up to 50% and its rf structure can inprinciple be removed by debunching the beam although there are nospecial advantages in doing this. A synchrotron, as a pulsed device, iswell matched to both, passive and active beam delivery systems. Thelength of the extracted beam pulse could be made to correspond tothe time needed to fully or partially irradiate a slice of the target, andthe energy can then easily be changed for the next slice. However, theextracted beam intensity in a synchrotron is quite susceptible to anyfluctuations or noise in magnet power supplies and efforts have to bemade to address this problem.

2.2 Type of Accelerators for Therapy

2.2.1 Cyclotrons

Cyclotrons are very compact machines with a constant magnetic fieldand a fixed frequency of the accelerating voltage. The injection of thebeam from the ion source, its acceleration in the machine and ejectionare a continuous process; the extracted beam has a fixed energy andits intensity can also be continuous which can have advantages whenscanning the tumor. Other extracted beam parameters, such as emit-tance, energy spread or time structure, can also be matched to theneeds of the beam delivery system. While cyclotrons delivering pro-ton beams with energies up to 230 MeV have already been developedby industry to operate in a hospital environment, their application asaccelerators for light ion therapy is now under study [8]. The energyper nucleon needed for the same penetration depth is higher, also thecharge to mass ratio of ions compared to protons is lower; becauseof these factors, a standard-design light-ion cyclotron for all cancertreatment would have a prohibitively large magnet.


2.2.2 Linear accelerators

Conventional linear accelerators are usually very low duty-factor ma-chines, delivering high ion beam currents in short pulses (of about amillisecond duration), often for injection into the next stage accelera-tor such as a synchrotron. They can accept and accelerate ions havinga certain ratio of the charge to mass and deliver a beam with an energyfixed or, at best, variable in large steps. Although the extraction effi-ciency is close to 100%, there are presently no linear accelerators usedfor either proton or light ion therapy. Linear accelerators are machinesrequiring a large space, they are expensive to build and to maintainand beam characteristics are not most favourable for radiotherapy. Im-proved performance (broader spectrum of ion species, a higher dutyfactor, some flexibility in output energy, reduced size) can be achievedby using superconducting cavities but this is again a sophisticatedtechnology which is not very suitable for a hospital environment.

2.2.3 Synchrotrons

A synchrotron is a pulsed accelerator, with particles moving on aclosed, approximately circular trajectory where the magnetic field andthe frequency of the accelerating voltage vary in time as the energyof particles increases. The pulse rate of a synchrotron is of the orderof 1 per second or less except for very large machines, and the dutyfactor can be as high as 50%. The energy of the extracted beam de-pends on the final value of the magnetic field and can be changed ona pulse-to-pulse basis, which makes this type of a machine very wellmatched to the depth scanning by beam energy modulation. Althoughthe extracted beam intensity is lower than from either a cyclotron or alinear accelerator, by a proper design it can be made sufficiently highfor any ion species and for treatment of tumors at any depth.

2.3 Facilities treating patients with Hadrons

Most existing particle therapies use protons delivered in fixed horizon-tal beam lines combined with passive scattering systems. In the compe-tition for a better dose conformity, the IMRT with photons reached thesame quality as horizontal passive modulated proton beams. Conse-quently, the protontherapy centers of recent conception feature isocen-tric gantries to improve the conformity of the treatment avoiding the

2.3 Facilities treating patients with Hadrons 15

Table 2.2. Number of patients irradiated with protons by January 2005.

healthy tissues. In this way, radiotherapists have the option of rotatingthe direction of the therapeutic proton beam around the patient justas they do for x-ray treatments.The magnetic rigidity of 200 MeV protons is such that a standardmagnetic channel capable of doing so has a typical total radius of 45m. For this reason, fixed (mainly horizontal) proton beams have beenused worldwide till 1992, when the first hospital-based center becameoperational at the Loma Linda Medical Center (California). As shownin table 2.2, the new proton facilities usually have more gantries, whichare large and heavy mechanical structures and which rotate around ahorizontal axis supporting bending magnets and quadrupoles.Most of the 40 000 patients have been treated with proton beams, pro-


duced by accelerators built for nuclear and sub-nuclear physics. Sincethe turn of the century things have changed. Now, five commercialcompanies offer turnkey centers of proton therapy featuring isocentricgantries (two based on cyclotrons and three on synchrotrons): ACCEL[9], IBA [10], Hitachi [11], Mitsubishi [12] and Optivus [13].These are hospital-based centers in the sense that they have more thanone irradiation room devoted to hadrontherapy of deep-seated tumorsand are run in strict connection with one or more hospitals. Typically,in such a center between 15000 and 25000 irradiation sessions (lasting20÷30 min each) are held every year. Since an average proton treat-ment needs 20÷25 sessions, a typical center will provide protontherapyto 1000 patients and more every year. At the end of 2004, there werethree dedicated hospital-based centers for deep protontherapy in theUnited States and four in Japan. In USA, the second hospital cen-ter (NPTC) is located close to the Massachusetts General Hospital(Boston) and the third one is the Midwest Proton Radiotherapy inBloomington (Indiana). The centers under construction are in Hous-ton (Texas)where the Hitachi medical synchrotron has been selectedby the MD Anderson Cancer Centerand in Gainesville (Florida) wherethe Universty of Florida Shands Medical Center has chosen the IBACyclone 230. In Europe, the Paul Scherrer Institute launched a newproject (PROSCAN) at the end of the year 2000. The proton beam,from the superconducting cyclotron produced by ACCEL, serves boththe existing eccentric gantry and a new isocentric gantry and, to ac-tively distribute the dose, an improved version of the PSI spot scanningtechnique will be implemented [14]. The Rinecker Proton TherapyCenter in Munich (Germany) has selected the same accelerator. Inthis center, to be completed in 2005, the ACCEL cyclotron serves fourgantries and a horizontal beam [13]. Three centers based on the IBAcyclotron were being built in the East, two of them in China and oneat the National Cancer Center in Seoul. The Chinese locations are theWanjie Tumor Hospital in Zibo, opened at the end of 2004, and theSinoJapanese Friendship Hospital in Beijing.

In 1994, the first patient was treated at NIRS in Japan with carbonions where the construction of HIMAC (Heavy Ion Medical Acceler-ator in Chiba) was promoted by Y Hirao. More than 2000 patientshave been treated and very promising results have been obtained. AtHIMAC, the choice was made not to construct rotating gantries but


to have a horizontal and a vertical beam in a single treatment room.The other two rooms feature horizontal beams. Passive spreading sys-tems were used for many years, but in 2003 a simplified active systemwas implemented called layer stacking. Presently, the introduction ofactive scanning is in preparation. In the Center HIBMC in Hyogo,the first patient was treated with protons in May 2001. This center,constructed by Mitsubishi Electric, is based on a single linac injectorand a 29 m diameter synchrotron for protons and carbon ions. It fea-tures three treatment rooms for protons (two of them with gantries)and two rooms for ions featuring a horizontal, a vertical beam andalso an inclined beam. By the end of 2004, about 200 patients hadbeen treated with protons and 50 patients with carbon ions. Startingin 1988, G.Kraft proposed a two-step project: installation of an irra-diation unit for experimental patient treatment at the new heavy ionaccelerator SIS of GSI and, as a second phase, the construction of adedicated heavy ion therapy unit at the Heidelberg clinic. In the sum-mer of 1993, the construction of the therapy cave at GSI started and,in December 1997, the first two patients were treated with the novelintensity modulated raster scanning technique [15]. With the start ofthe patient treatment the project leadership turned to the HeidelbergClinic.

2.3.1 Status of the Hadrontherapy in Italy

In Italy, the first and actually unique protontherapy facility, namedCATANA (Centro di AdroTerapia e Applicazioni Nucleari Avanzate)was built in Catania [16], at the Istituto Nazionale di Fisica Nucleare-Laboratori Nazionali del Sud (INFN-LNS). Here a 62 MeV protonbeam, produced by a Superconducting Cyclotron (SC), is used forthe treatment of shallow tumors like those of the ocular region. TheCATANA project was developed to treat ocular pathologies like uvealmelanoma, which is the most frequent eye tumor in adults. More-over, it is possible to treat other less frequent lesions like choroidalhaemangioma, conjunctiva melanoma, eyelid tumors and embryonalsarcoma. After three years of activity 78 patients have been treatedinside the CATANA (Centro di AdroTerapia ed Applicazioni NucleariAvanzate) facility. Inside CATANA new absolute and relative dosi-metric techniques have been developed in order to achieve the bestresults in terms of treatment precision and dose release accuracy. The


follow-up results for 42 patients demonstrated the efficacy of high en-ergy protons in the radiotherapeutic field and encouraged us in ouractivity in the battle against cancer.

Fig. 2.5. Ocular Melanoma Treatment station operating at the LNS in Catania. 1:treatment chair for patient immobilization; 2: final collimator; 3: positioning laser; 4:light field simulator; 5: monitors chambers; 6:intermediate collimator; 7: box for thelocation of modular wheel and range shifter.

Concerning the therapy with light ions, in Italy the CNAO (CentroNazionale di Adroterapia Oncologica) center has been founded andalso the Trentino country has decided to found an agency to promotethe construction of a proton therapy center. The CNAO will be ahadrontherapy center based on a 400 AMeV synchrotron for Carbonions, and equipped with one fixed gantry (preliminary phase) and (sec-ond phase) two room rotating gantries. The facility is foreseen to beready by the end of 2007.

As alternative choice the LNS group (Catania) suggests a medicalcenter for hadrotherapy based on a superconducting cyclotron ableto deliver proton beams and light ions beam up to the energy of 250AMeV in order to cover the most part of tumors [8].The details of this machine, subject of the SCENT project is discussedin chapter 3 of this job.


The figure 2.6 under shows the layout of this possible medical cen-ter. It is based on the superconducting cyclotron above mentioned.

Fig. 2.6. Layout of an hadrotherapy center based on SCENT project of LNS (Catania).

The treatment building will consists of three floor. The undergroundfloor dedicated to the technical plant for the cyclotron and the othertechnical devices. The ground floor is dedicated to the preparation ofthe patients, to the treatments and to the control of the accelerator.The upper floor is dedicated to host the room for the medical andtechnical staff and for additional technical plants.The beam acceler-ated and extracted from the cyclotron has a fixed energy. To allow atreatment of the tumors with the so called multi-painting techniqueand to avoid the beam energy degradation in front of the patients,we decided to use an energy degrader and a selector magnet systemsimilar to the devices already under construction for the RPTC centerin Munich and at IUCF [17]. The variation of the beam is achievedby a continuously variable double wedge beryllium energy degrader,which permit an energy variation of beam from 250 MeV down to 70MeV in less than 10 sec. The purpose of the Energy Selection BeamLine is to select and transport particles with desired energy from theenergy degrader target to the entrance of either a gantry or a fixedhorizontal treatment beam line. The design of the energy selectionbeam line will be similar to the IUCF devices but the magnetic rigid-ity would be doubled to allow the selection of the carbon ions too. For


this reason we select magnets with bending angle of 45 deg instead ofthe original 63 deg of the IUCF project. The reduction in the resolvingpower due to the reduction of the bending angle is compensated bythe larger bending radius of the magnets.Momentum resolution lowerthan 2× 10−3, corresponding at energy resolution of 250 KeV for theminimum energy of 65 MeV for proton, is expected. The room 1 isdedicated to the treatment of eye melanoma and more generally toun-deep tumors. The cyclotron beam, thanks to the stripper extrac-tion, is feed to this room independently from the Energy Selector line.The beam delivered to this room is energy degraded down to 150/70MeV by fixed energy absorbers. These absorbers will be placed beforethe last bending magnet. The final range shifter and range modulatorwill be placed inside the treatment room, but a local concrete shieldwill protect the patient.

The gantries for the center are the same of the so called PSI GantryII [18]. This novel concept Gantry rotates over an angle range of 180deg. The treatment from every prescribed beam direction is feasibleby rotating the patient position of 180 def. The scanning magnets areinstalled upstream of the last bending magnet, allowing for scanningwith quasi parallel proton beams. The maximum irradiation field-sizebeing of about 10 cm by 20 cm. Others advantage of this new kind ofgantry are:

• Maximum gantries radius is reduced;• There is enough space in the treatment area for installing a CT-

scanner for position verification. To allow the preparation of a cer-tain number of medical team and also for economic reasons one ofthe two gantries will be installed in a second phase.

So despite the center is mainly a protontherapy center it could takeadvantage of the carbon ion availability to dedicate a 10-20% of thebeam time to treat patient with carbon ion or other lighter ions likeLitium. Moreover the room 1 was designed large enough to allow theinstallation of an extra beam line for radioisotopes production to beperformed during the night or weekend.

3

SCENT project description

A compact superconducting cyclotron is the subject of a project calledSCENT (Superconducting Cyclotron for Exotic Nuclei and Therapy)[8]. The aim of the preliminary design of SCENT was to realize anhigh intensity machine able to deliver protons (extracted by means ofthe stripping of the accelerated H+

2 beam) and light ions, as Carbon,used as primary beams to produce exotic nuclei in the radioactive ionbeams (RIBs) facilities. The maximum energy was fixed at 250 MeVfor protons and 210 AMeV for Carbon ions. Two pairs of supercon-ducting coils and the use of trim coils to refine the magnetic field wereexpected for this machine. Because of the design complexity, this cy-clotron is not optimized for easy operation. But the purpose was toupgrade the EXCYT [6] facility at LNS, where expert people alreadyoperates.In a second step, the same machine was modified to meet the medi-cal application requirements. In this case the cyclotron was redefinedto simplify the technologic choices and it was searched for achieve amore easy reliability and operativity. Thus, the energy was fixed at250 AMeV for ions with charge to mass ratio constant (0.5), in orderto use only a pair of coils, to fixe the RF frequency at 93 MHz in 4th

mode and to avoid the use of complex system of trim coils.

3.1 The physics with Exotic Nuclei

The study of exotic nuclei opens new opportunities, excitement, andchallenges. The opportunities arise because it should be possible toselect specific nuclei from a greatly enhanced gene pool in order toisolate and/or amplify specific interactions, nucleonic correlations, ex-

22 3 SCENT project description

citation modes, and symmetries. The challenges arise because exoticnuclei will present new and radically different manifestations of nucle-onic matter that arise near the bounds of nuclear existence, where thespecial features of weakly bound, quantal systems come into promi-nence, and because these nuclei are key to understanding the cosmos.We already see glimpses of the exciting physics, for example, in theappearance of Borromean halo nuclei, and in the breakdown of thelong cherished magic numbers as benchmarks for structural evolution,but a much broader range of new phenomena is expected to emergebeyond the present limits of experimental accessibility. The nuclear

Fig. 3.1. Table of stable and unstable nuclei.

landscape (Fig.3.1) defines the territory of Radioactive Ion Beams(RIBs) research. Most of what we know about nuclei today comesfrom studies with stable nuclei: these are the black squares on thefigure. By adding either protons or neutrons to one of these stablenuclei, one moves away from the line of stability, first producing un-stable nuclei and finally reaching the drip lines where nuclear bindingforces are no longer strong enough to hold nuclei together. The yellowsquares indicate unstable nuclei that have been produced and studiedin the laboratory. But many thousands of radioactive nuclei have yetto be explored: this nuclear terra incognita is indicated in green. RIBswill expand the investigations into the nature of nucleonic matter byproviding experimental access to these nuclei. It will define and map

3.1 The physics with Exotic Nuclei 23

the limits of nuclear existence and allow us to explore the structure ofthe exotic systems that inhabit these boundaries.

A remarkable level of interest has developed with accelerated radioac-tive ion beams (RIBs) in a short time. This interest has led to theconstruction, development, and proposed construction of first- andsecond-generation RIB facilities in Asia, Europe, and North Amer-ica. This interest stems directly from the scientific opportunities RIBsprovide for nuclear, solid-state, and astrophysics and applied research,and the fact that the acceleration of RIBs has become technically fea-sible over the last two decades. RIBs will greatly increase the numberof nuclei available for study, particularly nuclei with extreme valuesof neutron and proton number. Altogether, between the proton andneutron drip lines and the fission limits for heavy nuclei, there areapproximately 3000 particle-stable nuclei that can be studied. Someinformation exists on about 40% of these nuclei. Most of our nuclearstructure information is from nuclei near beta stability that were firstproduced with accelerated light-ion stable beams incident on stabletargets and then with heavy-ion beams, particularly heavy-ion fusionreactions producing proton-rich nuclei. Much of the nuclear informa-tion on neutron-rich nuclei was obtained from reactor produced fissionfragments. The development of the RIB facilities presently underwaywill allow new regions of proton- and neutron-rich nuclei to be studied,providing exciting new research opportunities. Of course not all un-studied nuclei will prove to be interesting; however, much of the newinformation on new interesting phenomena will come from nuclei stud-ied with RIBs. Another extremely interesting research area for RIBsis astrophysics, and in particular nucleosynthesis. Most of the heavynuclei in the universe have been produced in explosive stellar eventswith short-time scales through nuclear reactions on unstable nuclei.It is only with RIBs that these reaction processes can be studied indetail in the laboratory. The reaction network for explosive hydrogenburning, the rapid proton or RP process, is believed to have producedthe proton rich stable nuclei between oxygen and the iron region.

In general, RIBs can be obtained by two very different and com-plimentary methods: The Projectile Fragmentation, PF, method andthe Isotope Separator On Line, ISOL, method. The PF method pro-duces RIBs in peripheral collisions between heavy-ion projectiles and


light-target nuclei [19]. The incident beam is typically in the 50 to 200AMeV range and the radioactive fragments recoil in a forward-anglecone with velocities similar to that of the incoming beam.The PF method is fast, so short half-life RIBs can be made, anddoes not suffer from the severe target beam-ion-source chemistry con-straints of the ISOL method. However, RIBs are produced at high en-ergies and the RIB intensities, because of thinner targets and smallerprimary beam intensities, can be several orders of magnitude lowerthan those from the ISOL method. The PF method requires only oneaccelerator system.The ISOL method requires two accelerator systems, a driver accel-erator to produce radioactive atoms at rest and a post acceleratorto accelerate these radioactive atoms to energies of interest. The twoaccelerators are coupled by a target-ion source and mass separator.This method is equivalent to selecting a low-energy RIB from an online isotope separator and accelerating this beam to energies of in-terest. The crucial component for the ISOL method is the target-ionsource. Radioactive atoms are produced by bombarding thick targetswith high-intensity beams of protons or other light ions. The result-ing radioactive atoms will diffuse and desorb from the target material,which typically is a refractory ceramic in powder form which is heatedto over 2000 Celsius deg. The radioactive atoms will effuse through aheated transport tube into the ion source where they are ionized byelectrons from a heated cathode and accelerated. Following ionizationand acceleration, the desired mass can be separated with a magneticmass separator. A very good separator, with a mass resolution greaterthan 1/10,000, can separate isobars from a single mass thus providingisotopically pure beams on target. A significant disadvantage of theISOL method, is that the diffusion and desorbtion of the radioactiveatoms from the target material and the effusion and surface desorbtionin the transport tube and ion source are sensitively dependent on thehigh-temperature chemistry of the target material, the beam element,and construction materials. The ISOL method is also much slowerthan the PF method, limiting RIBs to a half-life of at least 100ms orlonger. However, for the longer lived species, the ISOL method is ca-pable of producing more intense beams at energies needed for nuclearspectroscopy and astrophysics studies.

3.1 The physics with Exotic Nuclei 25

3.1.1 Status of the EXCYT project in Catania

The Laboratorio Nazionale del Sud (LNS) is equipped with a 15MV Tandem accelerator and with a K=800 MeV superconductingcyclotron (SC). These accelerators will be reconfigured in order toprovide an ISOL type RIB facility. In order to increase the energy andthe intensity from the SC cyclotron, a new 14.5 GHz, 1.4 T, supercon-ducting ECR ion source was constructed for axial injection. The aim isto extract from the SC cyclotron beam currents of the order of 1 µAp.This is a challenge for this type of cyclotron, and the electrostatic de-flector will have to stand very high power of the order of 3 kW comingfrom the extracted beam. The negative ions will be produced with anew 150 kV injector, similar to the HRIBF concept and injected intothe 15 MV tandem accelerator. RIB up to mass 40 will be acceleratedabove the Coulomb barrier. The EXCYT [6] (Exotic Nuclei produc-tion with Cyclotron and Tandem) project has been funded in 1995and the commissioning of the facility will start in the middle of 1998and operation in 1999. The layout of the project is shown in Fig.3.2.

Fig. 3.2. Scheme of the EXCYT project at LNS.

On June 10, 2005 a 8Li beam was successfully produced and ex-tracted at the EXCYT facility of the LNS. The primary beam of 13C4+


(45 AMeV) was provided by the K-800 Superconducting Cyclotron. Abeam power of 18 watts was delivered and focused on the Target IonSource Assembly (TIS) to produce the radioactive atoms. The TISconsists on a 8.5 mm thick graphite (C) target on a tantalum con-tainer and a Positive Ion Source (PIS) both heated up to 2100 Celsiusdegrees. The PIS, being a surface ionisation source, is highly selectiveand efficient for alkaline atoms. The 8Li beam was extracted at 25keV form the TIS and transported through the first (pre-separator)and the second platform (first isobaric separator). The 8Li beam wasmeasured by means of a LEBI diagnostic device which detects the8Li beta decay emission. Taking into account the LEBI detection effi-ciency and subtracting the background, the 8Li production yield wasof 1.5×104 atoms/s (see Fig.3.3).

Fig. 3.3. First counts of the radioactive ions 8Li at LNS. (courtesy of LNS)

3.2 The Preliminary Cyclotron’s design as driver for Exotic Nuclei Facilities 27

3.2 The Preliminary Cyclotron’s design as driverfor Exotic Nuclei Facilities

Accelerators of medium power (10-50 kW) are suitable drivers of facili-ties to produce radioactive ion beams. Therefore a widespread interesthas arisen around them. At LNS the EXCYT project is in progress.This project is based on a Superconducting Cyclotron to be used asa primary accelerator to produce radioactive beams to be acceleratedby a Tandem accelerator. One main limitation of the project is theprimary beam intensity deliverable by the present cyclotron. The EX-CYT facility could be significantly upgraded if a primary beam witha 20÷40 times higher power would be available.For this reason we investigated on the design of a new superconductingcyclotron able to deliver a beam power of about 50 kW. Extractionis the critical feature for a cyclotron of this kind, since it is generallyperformed by electrostatic deflectors. To be an efficient process, thisextraction method requires well separated turns, which is achievedwith a big radius of the machine and a high accelerating voltage [3].Both these features make the cyclotron expensive. Following the de-sign of commercial cyclotrons often used to produce radioisotopes, wepropose to overcome the limitations set by electrostatic extraction byusing extraction by stripping.In the following the preliminary study of the superconducting cy-clotron is presented.This cyclotron is able to accelerate H+

2 up to 250 AMeV. The samemachine was thought to accelerate light ions like 12C5+ at lower en-ergy, to be extracted by stripping and in general light ion beams withcharge state qaccel=Z-1, Z being the charge of the nucleus (Z=6-10).

3.2.1 Main parameters of the machine

The study of the cyclotron has started from the main parameters ofthe K1200 cyclotron operating in the Michigan State University, USA[20], and we have modified them to design a new machine, whosecharacteristics is shown in Table 3.1.

The main differences are the number of sectors and the radius size.As compared to the K1200 cyclotron, the larger extraction radius al-lows, at the same time, to increase the flutter, so as to achieve a Kfoc=500 and to maintain the magnetic field at acceptable values. The main


Table 3.1. Main Parameters of the Cyclotron for RIBs facilities.

characteristics, i.e. spiral angle, sector width and other important pa-rameters of the machine have been preliminarily calculated by ana-lytical approach applying a simple first-order formalism and assumingthe uniform iron saturation. Several changes of the preliminary modelhave been made using the 3D electromagnetic code OPERA [21].

A parametric model of the cyclotron was built, and an iterativeprocess has been realized in order to optimize automatically severalparameters of the machine like sectors width, hill and valley gaps, spi-ral angle to fulfill the beam requirements, like the isochronous fieldand vertical focusing. The sector width varies from 25 deg at the cen-ter of the machine up to 40 deg at extraction radius. The maximumgap height of the valley has been fixed to 45 cm and an iron shim of10 cm high has been introduced up to 90 cm of radius to better shapethe isochronous field. Four holes were located at the center of the val-leys (70 cm from the center and 29 cm of diameter) to lodge the RFcavity stems. A large gap (74 mm) between the poles was providedfor possible beam growth in the vertical plane and allows to introducethe trim coils system. An adequate shimming of the hill gaps at the

3.2 The Preliminary Cyclotron’s design as driver for Exotic Nuclei Facilities 29

Fig. 3.4. Top view of the cyclotron.

Fig. 3.5. Isochronous field for two different ions accelerated.

extraction region should ensure a good vertical focusing. Differencein gap height at the minimum and maximum radius is about 6 mm.Decreasing the hill gaps at the last centimeters of the pole, the spi-ral constant can be reduced, which is advantageous also from the RFpoint of view [22]. Two pairs (α and β) of superconducting coils (see


Table 3.1), symmetrically placed above and below the median plane,are supposed to generate the isochronous fields for light ion beams.The position, geometry and currents of the coils have been optimizedwith a process similar to the above mentioned to cover the whole rangeof energy. A big vertical space (200 mm) between the coils is requestedto allow access to the cyclotron median plane for the beam stripper,the extraction elements and the beam diagnostic probe.The design of the machine model has been done to accelerate H+

2

molecules up to 250 AMeV and 12C5+ ions up to 210 AMeV, corre-sponding respectively to a maximum average field at an extractionradius of 3.8 tesla and 4.2 tesla. Figure 3.5 shows the isochronous fieldfor both ion species. Fine tuning of the fields should be obtained byusing trim coils system whose fields are estimated to be less than a fewhundreds of gauss. The beam dynamic properties have been studiedby the code GENSPE1 [23], which calculates the equilibrium orbits.In Fig.3.6, the vertical and radial focusing frequencies are plotted forboth the ion types vs. energy. The preliminary trajectories of extrac-tion by stripping have been traced for H+

2 and 12C5+ ions (see Fig.3.7).

Fig. 3.6. Frequency tunes for Carbon ions and Hydrogen molecules.

3.3 The Final Cyclotron’s design for Medical Applications 31

Fig. 3.7. Extraction path of Carbon ions and Protons by means of stripping process.

3.3 The Final Cyclotron’s design for MedicalApplications

The preliminary layout of the machine above described, was modifiedto meet the medical application requirements.The radiation oncologist community is more and more interested inusing light ions like carbon for tumor treatments. Usually the maxi-mum energy of carbon beam to treat the deepest tumors is 400 AMeV.But according to the distribution of the number of patients vs targetdepth treated at Chiba shown in Fig. 3.8, it is quite evident that anenergy of 400 Mev is necessary only for treatment of prostate anduterus tumors which are just 30% of the total cases. The treatmentof 50 % or of 70 % of cases is feasible with an energy of 250 and 300AMeV respectively. Particulary the eligible cases in this energy rangeare mainly head and neck, Non Small Cell Lung Tumor, soft tissuesarcoma and spinal cord tumors. These lower energies are achievableat cyclotron accelerators which are smaller and cheaper than a syn-chrotron. Moreover a cyclotron is easier o be operated and needs asmaller staff than a synchrotron. Its continuous beam and better cur-rent control are two crucial features allowing to deliver the right doseon the target.


In the past, studies for medical cyclotrons able to deliver up to 400

Fig. 3.8. Statistical data concerning the number of targets versus the depth from CHIBA,Japan.. The arrows indicate the depth limit achievable with the proton beam of 250 MeVand the Carbon ion at the same energy. (courtesy of HIMAC center, Japan)

AMeV were proposed, but the proposed designs were not appealingdue to the technological difficulties [24]. To achieve a conservative andreliable design, we decided to use only technical solutions already usedand/or tested in existing machines. According to these constraints atthe beginning of the study, the maximum energy of the acceleratorwas fixed to be 250 AMeV.We want to emphasize that this cyclotron is able to accelerate bothprotons and light ions to be used for radiotherapy. Although low beamintensities are generally required for therapy, it is also possible to ex-tract the H+

2 beam by stripping with a 100 % extraction efficiency.So a beam power of 10 kW or more can be delivered to produce newmedical radioisotopes.This is an additional advantage of this kind of cyclotron which couldwork 24 h per day, from 8 a.m. till 8 p.m. for patients treatments, andover night for radioisotope production. A real interest towards these


new radioisotopes is proved by the amount of radioisotopes producedby TRIUMF and other research centers around the world [25].

3.3.1 Main parameters of the machine

Based on the above described machine for RIB facilities, the super-conducting cyclotron is able to deliver ions beams with Z/A=0.5, inparticular the H+

2 molecule and 12C6+ ion (it should be possible to ac-celerate all fully stripped light ions up to Neon), up to the maximumenergy of 250 AMeV. The main layout of the cyclotron was modifiedand the main parameters resumed in Table 3.2.This machine is a compact cyclotron whose magnetic field is gen-

Fig. 3.9. Layout of the SCENT machine.

erated by means of a pair of superconducting coils instead of twopair described in the previously version. These coils are symmetricallyplaced above and below the median plane, and their maximum currentdensity should be less than 50 amp/mm2. The coils should also providea rather sharp fringing field fall-off to facilitate the beam extraction.This requirement again calls for coils placed as close as possible to thepole and to the mid-plane and for a high aspect ratio height/width ofthe coils cross-section.


Table 3.2. Main parameters of the SCENT machine.

The selected values, compatible with all the constraints of the coilconstruction, are a cross section of 150×210 mm2, with a distance tothe median plane (MP) of 75 mm. Four sectors are expected in order toavoid harmful resonances and to guarantee the necessary flutter. Theprimary goal of the sector design is however to assure an adequateaxial focusing for the accelerated beam. The bending limit of the ma-chine was preliminary set as 1080 MeV, corresponding to a maximumaverage field 3.75 tesla and to an extraction radius of approximately1330 mm. As focusing limit of the machine it was previously calcu-lated the value of 500 MeV. The operating RF harmonic mode should


Fig. 3.10. Artistic view of a superconducting cyclotron similar to SCENT machine(courtesy of MSU -USA- and ACCEL Gmbh -Germany-).

be the 4th. This choice needs a very precise isochronous magnetic field.The fine tuning of the magnetic field will be done by means the passivemagnetic elements as the trim rods (TR).The general structure of the machine for all aspects relevant to themagnetic field design, like the iron geometry, the superconducting coilsis outlined in the chapter 4.

Four RF resonant cavities operating in 4th harmonic mode are ex-pected to generate the electric field needed to accelerate the particles.The RF frequency was set at 93 MHZ, and the maximum averagevalue of voltage should be 120 kV at the extraction radii.


The RF system is discussed in chapter 5.

The extraction system is composed by a stripper foil appropriatelypositioned inside the cyclotron in order to extract the proton beamby separation of the H+

2 molecule and two electrostatic deflectors forlight ions extraction. Such system should be followed by a set of passivemagnetic channels, in order to optimize the extraction efficiency. Theinjection system foresees an axial injection through an spiral inflector.The layout of the injection system is discussed in chapter 6.

4

Magnet Design

A superconducting cyclotron able to deliver ions beams with charge tomass ratio Z/A=0.5, in particular the H2

+ molecule and the 12C6+ ion(it should be possible to accelerate all fully stripped light ions up toNeon), up to the maximum energy of 250 AMeV is the subject of thisstudy. Such machine is a compact cyclotron whose magnetic field isgenerated by means of a pair of superconducting coils, symmetricallyplaced above and below the median plane, whose maximum currentdensity should be less then 50 amp/mm2. Four sectors are expectedin order to guarantee the necessary focusing and to avoid harmfulresonance of νr=N/2 (N is the number of sector), and such sectorsshould be spiraled for adequate vertical focusing of the beam.

4.1 Main parameters of the Cyclotron

The design guidelines of such a machine are based on the existingK1200 superconducting cyclotron operating at Michigan State Uni-versity [20]. The key differences are that there are four sector insteadof three, and the size of the radius is larger due to the higher extrac-tion energy of the accelerated ions (250 AMeV instead of 200 AMeV),which cause an increase in the dimensions of the whole machine. Theextraction radius has been fixed at 1.3÷1.33 m in order to keep accept-able the overall weight of the machine. Consequently the maximumaverage value of the isochronous magnetic field 〈B(REXT )〉 at the ex-traction region will be about 3.75 tesla. The K bending parameter forthis machine should be:

38 4 Magnet Design

Kbend =e2 · c2 · 〈B(REXT )〉2 ·R2

EXT

2 · Eo

where:e :electron chargec :speed lightEo :rest energywhich gives, from the relation,

T

A≤ (

T

A)2 ·Kbend

the maximum theoretical energy achievable (T ) of 270 AMeV by ac-celerating the ions with Z/A = 0.5.Assuming the isochronous field shape along the radius,

Biso(r) =Bo√

1− ( rωo

c)2

where

fo =ωo

2π=

Z · e ·Bo

A ·mo · 2πwe can determine the magnetic field at the center of the cyclotronas Bo= 2.96 tesla and the revolution frequency of the ions inside themachine as f0 ∼23 MHz.The gap between the magnet poles has been fixed, preliminary, as7.4 cm as set during the design of the machine for RIBs facilities. Asmaller gap should reduce the required number of ampere-turns in thecyclotron main coils and allow the last orbit of accelerated particles tobe brought closer to the pole edge, in order to simplify the extractionoperation with the electrostatic deflectors.The table 4.1 resumes the main parameters of the cyclotron.

4.2 Magnetic Design (Analytical approximation)

A fourfold symmetric compact magnetic structure has been chosen toproduce the required azimuthal varying magnetic field. The magneticfield has to guide the particles on isochronous trajectories and providethe required focusing in order to maintain good internal beam char-acteristics.

4.2 Magnetic Design (Analytical approximation) 39

Table 4.1. Preliminary parameters of the cyclotron

These properties are the results of two effects:

• The spatial field variations due to the shape of four spiral sectors;• The positive radial gradient of the isochronous field is required to

compensate the relativistic mass increasing of the ions with energy.This gradient is obtained by increasing the angular span of thesectors and by an adequate position of the superconducting coils.

Once the main parameters are set, it is necessary to establish thedimensions of the machine. As first approximation, it is possible toscale the dimensions of the K1200 superconducting cyclotron in orderto set the radial width of the flux return yoke. To validate the scalinglaw, the total magnetic flux in the cyclotron is also estimated usingthe average magnetic field and the pole radius, in order to make equalthe magnetic flux through the pole area and the return yoke equal:

Φ(〈BPOLE〉) = Φ(〈BY OKE〉)

π ·R2EXT · 〈BISO(r)〉θ = AreaY OKE ·Bsaturation

A value of yoke radius RY OKE = 2×RPOLE should be a conservativechoice. The chosen average magnetic field is the basis of the first esti-mation of the magnetic field in the hill Bhill(R), in the valley Bvalley(R)and the fraction of the hill α on one symmetry period according tothe equation:

40 4 Magnet Design

〈B(r)〉 = α · 〈Bhill(r)〉+ (1− α) · 〈Bvalley(r)〉

The Fig. 4.1 shows the first approximation of the average value of themagnetic field along the radius. The values of Bhill and Bvalley are de-

Fig. 4.1. Magnetic field approximation along the radius. The average value on the hill(red line) and on the valley (blue line) is shown with the average value (black line).

termined by using an sinusoidal shape of the magnetic field along theazimuthal direction:

B(r, θ) = B(r)(1 + f · sin(4 · θ))

On the graph shown in Fig. 4.2, the analytical approximation of themagnetic field at the extraction radius is plotted, where the main valueis 3.75 tesla and the peak values are respectively Bhill = 4.5 tesla andBvalley = 3 tesla respectively. In this case, if f = 0.2, the flutter F ,which measures the change field strength between the hills and val-leys and which we assumed constant as first approximation, gets thefollowing value:

F =〈B2〉 − 〈B〉2

〈B〉2=

f 2

2= 0.02

4.3 Stability conditions and spiral definition(matrix method)

In order to ensure the axial stability of the beam during acceleration,it is necessary to shape the edges of the sectors as spirals. To define

4.3 Stability conditions and spiral definition (matrix method) 41

Fig. 4.2. Azimuthally varying field aproximation

the spiral angle, the matrix method was applied for approximate de-scription of the dynamic of the beam inside the cyclotron, withoutconsidering the acceleration phase [7].Taking the previously shown magnetic field design, the transfer matrixof the radial Mr(r, θ) and axial Mz(r, θ) dynamic were defined. Then,the stability condition |Tr(M)| ≤ 2 (where M is the transfer matrix)was applied [7], [3], [26].The results shown in table 4.2 were produced after the preliminaryanalysis:

Radius (m) Energy (AMeV) Biso (tesla) ξ(deg)

0.024 0.060 2.933 1.0130.152 2.393 2.940 10.9980.279 8.192 2.959 23.2640.407 17.657 2.989 35.0420.534 31.124 3.031 44.9960.662 49.105 3.088 52.8700.790 72.348 3.162 58.9490.917 101.928 3.255 63.6401.045 139.414 3.374 67.2981.172 187.15 3.525 70.1941.3 248.774 3.72 72.522

Table 4.2. Energy of the ions, Isochronous Field and Spiral angle ξ versus the radius ofthe orbit.

42 4 Magnet Design

Fig. 4.3. Layout of the spiraled sectors.

4.4 Preliminary design with 3D code

The calculations described above are simple, analytical and allow foran initial layout of the cyclotron magnet. The overall structure wasdesigned with CAD codes as shown in Fig. 4.4.The first step is to define the azimuthal span angle of the sectors

for a preliminary matching of the isochronous field. For such reason aparametric model of the cyclotron has been made for changing severalparameters of the structure as the coordinates of the points on theedge of the spiral sectors, and/or the vertical coordinate of eventualvalley shims and their extension radius. The size of the coils has beenfixed preliminary as shown in tab.4.3.

4.4 Preliminary design with 3D code 43

Fig. 4.4. Side view of 1/4 of the cyclotron with measures (cm).

Table 4.3.

After the first simulations, the preliminary results are shown in thefigure 4.5 and 4.6.Any discrepancy between the average cyclotron magnetic field and

the isochronous magnetic field results in a shift of accelerated parti-cles with respect to the chosen phase of the cyclotron RF system asdescribed in the follows: 4φ ≈ 2π · h · 4B

B· n.

An acceptable phase shift of the particle of 20÷40 deg, accelerating in

44 4 Magnet Design

Fig. 4.5. Comparison between the isochronous field and the average one

Fig. 4.6. Relative differences between the magnetic fields

4th harmonic, with an average number of turns n = 400, determinesthe tolerance of the magnetic field close to the value of4B\B = 10−4.The cyclotron magnetic field at the given radius r, having the N-fold

4.4 Preliminary design with 3D code 45

rotational symmetry, can be presented as Fourier series:

B(r, θ) = B0(r) +∞∑

n=1

Bn(r) · cos(n(θ − φn(r))

Bn are the Fourier coefficients, where n=0 corresponds to the aver-age magnetic field, and n=4 to the main harmonic. The flutter of themagnetic field can also be defined using coefficients of Fourier series:

F (r) =1

B20(r)

·∞∑

n=1

B2n(r)

The main harmonic coefficient (plotted in Fig. 4.8) is necessary to

Fig. 4.7. Flutter behavior along the radius

calculate the preliminary focusing limit (KFOCUSING) of the cyclotron.This limit is therefore mainly dependent on the spiral term and in-creases almost linearly with the spiral constant.In the figure 4.9, the solid line shows the upper focusing limit of themachine here described. The decreasing above 120 cm is compensated

46 4 Magnet Design

Fig. 4.8. Main harmonic coefficient along the radius

Fig. 4.9. Focusing limit of the cyclotron along the radius. The blue square shown theupper limit of KFOC=500 at extraction radius

by negative gradient of the magnetic field (this term does not appearin the KFOCUSING equation) , resulting in an increase of focusing fre-quency νz.In order to make sure of the radial and vertical stability, the betatron

4.5 Magnetic Structure description 47

oscillations have been determined by means of a dedicated trackingcode, named ORBITCODE [23], and shown in Fig. 4.10.

Fig. 4.10. Radial and vertical betatron oscillations vs the radius.

4.5 Magnetic Structure description

The whole magnet is mainly a cylinder with a radius of 2450 mm andan height of 2900 mm. The magnet structure can be separated intothree different regions:

1. the sectors region;2. the poles region;3. the return yoke region;

The first region (see Fig. 4.13) includes the spiraled sectors and theplug of the central region designed for the axial injection of the beamin order to compensate the hole effect on the magnetic field at the midplane. This region extends up to z=525 mm from the median plane(valley floor) and radially up to 1330 mm, where the sectors end. The

48 4 Magnet Design

Fig. 4.11. Top view of the cyclotron

Fig. 4.12. Side view of the cyclotron with measures (cm)

height of the sectors (hills) is 500 mm, in order to keep a half gap of25 mm (the height of the gap is set at 50 mm).The section a of the table 4.4 shows the azimuthal hill width as a

function of the radius, whilst the section b of the same table shows the


Fig. 4.13. Spiraled sectors highlighted in the model

Fig. 4.14. Spiral angles description.

parameters used for defining the spiral angles (as shown in Fig. 4.14).On the valley floor, one shim is placed at a different height (z=510

mm) and it extends radially up to 800 mm, for producing the desiredfield shape (see Fig. 4.15). Another shim is placed between R=1300mm and R=1325 mm, and extends vertically up to 100 mm from themedian plane (see Fig. 4.16). This shim together with the inner wallof the vacuum chamber (which extends up to 60 mm from the medianplane) are necessary to gain the magnetic field in the extraction region,shifting the peak average value towards the geometric limit of the pole.

50 4 Magnet Design

(a)

Radius width Radius width(cm) (deg) (cm) (deg)

5 34 70 3910 34 75 39.515 35.1 80 4020 35.6 85 40.325 35.8 90 40.530 36 95 40.735 36.2 100 40.940 36.5 105 4145 36.8 110 41.250 37.1 115 41.355 37.6 120 41.860 38.1 125 4365 38.6 130 44

(b)

Radius (cm) PHI (deg) ξ (deg)

10 5.01 6.520 12.33 15.630 20.89 25.440 30.37 34.650 40.59 42.660 51.45 49.470 62.87 5580 74.79 59.590 87.16 63.2100 99.96 66.2110 113.15 68.7120 126.7 70.8130 140.59 72.6

Table 4.4. Main Parameters of the sectors

Fig. 4.15. The shim positioned on the valley floor.

This feature provide an increase in the bending limit and helps to bringthe coils radially closer to the magnetic pole.The poles are characterized by the presence of the holes provided forcooling systems, cryogenic insertions (the biggest with a diameter ofφ=32 cm at r=90 cm) and the trimming and coupling capacitors (thesmallest with φ=20 cm at r=100 cm) as shown in Fig. 4.17.The yoke region has to guide the return magnetic flux. The design

was performed looking at two parameters:


Fig. 4.16. The second shim positioned at the outer radii.

Fig. 4.17. The holes expected in the valleys.

1. the fringing field in the area around the cyclotron and mainly inthe control room;

2. the shape of the fringing field around the cyclotron.

By means of EM codes, the magnet yoke has been shaped in order tominimize the leakage of the magnetic flux at 10 Gauss at a distanceof 10 m from the cyclotron and 500 Gauss at 1 m from the pole caps(see Fig. 4.18).Symmetric holes have been made across the median plane of the yokein order to simulate the apertures for the beam exit and for radialsupport of the superconducting coils. As for the coils, a set of 8 holes(diameter =100 mm) has been designed vertically across the yoke for

52 4 Magnet Design

Fig. 4.18. Magnetic fringing field distribution outside the cyclotron. A lower limit of500 gauss is shown.

Fig. 4.19. Holes drilled on the yoke magnet.

the axial coil supports, the currents leads and liquid helium feeder (seeFig. 4.19.

4.6 Superconducting Coils 53

4.6 Superconducting Coils

Two superconducting coils symmetrically placed above and below themedian plane are expected to generate the magnetic field. A crosssection of the main coils is shown in Fig. 4.20 and the geometricalparameters are also defined in tab.4.5.

Fig. 4.20. Coils Layout.

When performing the design of the cold mass of the superconduct-ing cyclotron, we had to consider that the cooling system could notinvolve the use of cryogenic liquids (namely LHe), due to the negativeimpact they would have on the medical environment (because of thecomplexity of the operations). As a result, we have worked under thehypothesis of cooling the winding with a cryo-cooler, which can beoperated very simply. Nevertheless, the impact on the magnet designis quite strong:

• Due to the relatively low available cooling power at 4.5 K (a fewWatts),we had to design a superconducting winding based on thecriterion of indirect cooling.This means that the cooling system will

54 4 Magnet Design

Table 4.5. Coils parameters.

be able to remove the limited heat loads (due to cryostat heat lossand to current leads), but it will not be able to protect the windingfrom thermal disturbances.The release of local energy for mechan-ical or electrical reasons will be recovered by the winding itself. Inorder to survive these localized disturbances, the winding will havea good enthalpy margin. Consequently, we had to keep the overallcurrent density as low as the total temperature increase is withinthe limit. As a superconducting material, we chose NbTi because, inspite of the relatively low critical temperature (9.2K), it is the mostwell-known material involved in superconducting magnets (Nb3Snis too brittle, High Tc has not yet been developed at an industriallevel).

• Current leads are sources of dissipation. The maximum current wasthen set around 1 kA. This is also the minimum current to limit thenumber of turns and the coil inductance, which plays an importantrole in case of quench.

• Since the winding is indirectly cooled, it is highly recommended toinvolve structural material with good transport thermal properties.As a consequence, aluminium alloy was implemented in the me-chanical structure. In this case we can have a containing structurewith high thermal conductivity (an advantage for the cooling) andthermal contraction not far from the winding (copper plus resin),so we can limit the mechanical stress. Of course, the aluminiumwill be a suitable alloy with excellent mechanical properties (highyield and tensile stress, good elongation).


4.6.1 Conductor Definition

In order to choose the conductor, the protection criterion based onthe hot spot temperature has been followed. Let us schematize themagnet as an inductance Lm protected by placing the resistance RD

in parallel. The current flowing in the circuit in operating conditionsis Iop. Let us consider that a small portion of the conductor becomesnormal, with a resistance r(T ) � RD, and that the protection systemopens the switch S in such a way that the magnet is disconnected fromthe power supply (see Fig. 4.21). In these conditions, starting from

Fig. 4.21. Layout of the protection circuit of the magnet.

a temperature Ti, the normal zone warms up to the temperature Tf ,following the approximated relation:

Z(Ti, Tf ) =1 + α

2

∫Cp(T )

ρm(T )dT =

1

AtotAm

I2opLm

2RD

with

α =Am + Asc

Atot

whereCp(T ): specific heatρm(T ): matrix resistivityAm: area matrixAsc: area superconduting insertAtot = Am + Asc + Aiso where Aiso is the insulation area

56 4 Magnet Design

The Z(Ti, Tf ) function is shown in Fig. 4.22 for several materials.Let us introduce some parameters: the magnetic energy E = 0.5 ·

Fig. 4.22. Z(4.2 K, Tf ) function for several materials: 1) silver 99.99%; 2) copper RRR200; 3) copper RRR 100; 4) copper RRR 50; aluminium 99.99 %

LI2op, the maximum voltage in case of quench V = RDIop, the critical

current density at the maximum field on the conductor Jc(Bmax)and the ratio between this parameter and the current density in thesuperconductor in operating conditions g = Jc(Bmax)/Jop. From therelation Z(Ti, Tf ) we can write:

Am =E

V · Z(Ti, Tf )

2α

g(1 + α)(1 + λ)· Jc(Bmax)

It is worth noting that the magnetic energy is determined by the di-mensions of the magnet and the operating magnetic field, whilst themaximum voltage and the Z function are partially arbitrary and con-nected to the safety. The critical current density at the maximum fieldis fixed, so that only the parameter λ = Am/Asc and the cross sectionalarea of the copper matrix will descend from the operating conditions


of the system.As regard the overall current density, its value is typically less than100 A/mm2. In order to have a satisfactory enthalpy margin we setthis value to 50 A/mm2 so that:

Io

Atot

=Jc(Bmax · α)

g(1 + λ)= 50

Then we can choose to have a wide stability margin, for instance 1/4 ofthe critical current, setting g=4. Knowing that Bmax=3.8 T, Jc ≈3100A/mm2 (see Fig. 4.23) and typically α=0.73, we find λ=10. Finally, we

Fig. 4.23. NbTi critical current density.

can choose a protection voltage V=1000 volts and Z(Ti = 4.5, Tf =300)=1.25e17 A2 s/m4 (supposing we have RRR=50 copper matrixand limiting the hot spot temperature to 300 K). Knowing that E=32MJ, we obtain Am=15.22 mm2. Is is possible to calculate the areaof the superconductor as Asc=15.22 mm2 and the operative currentIop = Asc ·Jc(B = 4.7 tesla, T = 4.5 K)/g=1070 amp. The inductanceis L=56 H.

4.6.2 Conductor Layout

The total section of the conductor is Am+Asc=16.6 mm2. A reasonablelayout is a Rutherford cable composed of 14 filaments, 1.25 mm indiameter, leading to a overall dimensions 2.3×8.5 mm2. The main

58 4 Magnet Design

characteristics of the conductor are summarized in the following table.

Table 4.6. Main characteristic of the conductor.

The designed dimensions of the two solenoids are the following:Ri=1.42 m, Re=1.57 m, h=0.21 m, d=0.85×2=1.7 m, where Ri isthe inner radium, Re the outer radius, h the height and d the axialdistance. Considering that the coils have to be wound with a conductor2.6×8.8 mm2 in section, the real dimensions will be slightly differentfrom the ones designed:

• Number of layers = 150/2.6=57.7 =⇒58• Number of turns per layer = 210/8.8=23.9=⇒ 24

The total length of conductor will be: 58×24×2×1.495=13080 m. Thisconductor can be reasonably supplied in 5 lengths, 2700 m each. Asthe joints cannot be immersed in the winding, it follows that thenumber of layers has to be divisible by 5. The result is a wind-ing made up of 60 layers (156 mm in thickness), with 24 turns perlayer (211.2 mm in height). The total length of the conductor is then60×24×2×1.498=13560 m (supplied in 5 lengths, 2.75 km each). The


Fig. 4.24. Load line and critical current at he maximum field of 3.8 tesla.

magnet load line is shown in Fig. 4.24. At the maximum field 3.8 T,the critical current is 4280 A. The sharing temperature, that is thetemperature at which the current starts flowing in the copper matrix,is 6.9 K. This means that there is a temperature margin of 2.4 K withrespect to disturbances or heating of the cold mass, corresponding to aheat input of 0.35 J/kg (43 mJ/m). The magnetic analysis shows thaton each coil there is a total radial force Fr=24.5 MN (corresponding toa radial pressure of 13 MPa), and a total axial force Fz=11.3 MN (inthe sense that the coils get compressed one against the other), whosedistributions are shown in Fig. 4.25 and 4.26.

60 4 Magnet Design

Fig. 4.25. Radial forces distribution on the coil.

Fig. 4.26. Axial forces distribution on the coil.

4.7 The Magnetic Model

The final structure of the cyclotron was built step by step, startingfrom the simplest model shown in section 4.4. Computational limita-tions, finite boundary conditions and material property are all possiblelimitations on the accuracy of the paper design. Considerable effortwas put into refining computer calculations as the design progressedto obtain reproducible convergence to very high accuracy.The 3-D model, limited by symmetry, tangential and normal bound-aries is restricted to 1/8 of the machine as shown in Fig. 4.27. The

4.7 The Magnetic Model 61

Fig. 4.27. Physical Model simulated.

complete model could be defined with eventual asymmetry, but thiswould be unnecessarily expensive for CPU time analysis.The boundary conditions have been set as:

1. Normal magnetic on the median plane;2. Periodic boundaries on the lateral sides of the box air enclosing

the model;3. Tangential magnetic on upper and external sides of the box air.

Generally, the iron used for the magnet have a low carbon content.In our experience with 3-D simulations [27], for iron specificationsranging from very high to normal commercial quality carbon content,no major effects should be found on any important parameter. Wehave, therefore, chosen an high quality steel whose B-H curve used forthe analysis is shown in Fig. 4.28.The air box limit was set at 3×RY OKE along R and 3×HY OKE alongZ. In practice, at long distances from the device producing field, thedistribution will be modified by the local environment (tangential oropen boundaries), but this will not effect the field close to the orbitplane of the cyclotron.

62 4 Magnet Design

Fig. 4.28. B-H magnetization curve of iron used during the simulations.

The first model was meshed with hexahedral elements in order to speedup the analysis time. The number of cells during the first simulationsdid not exceed 300’000.The last model was simulated with the tetrahedal meshing in order toincrease the geometrical accuracy. In this case the simulations achievedmore than 1’000’000 mesh cells and nodes.

4.8 Magnetic Field Design 63

4.8 Magnetic Field Design

The corresponding optimal iron field was calculated and the sectorgeometry shaped in order to meet this field as close as possible to theisochronous one. The sector shape which produces the best magneticfield was obtained in different steps. The layout of such an iterativeprocess is presented in Fig. 4.29.

Fig. 4.29. Magnet design optimization

From the analytical approach we have designed the preliminary modelof the machine, assuming the width angle of the sectors to be constantalong the radius. The matrix method (discussed in section 4.2) wasapplied to define the spiral angle in order to ensure the axial stabil-ity of the beam. Once the dimensions of the structure were fixed, thesimulations with 3D dedicated codes were carried out, to refine themagnetic field. Several variables of the model of the magnetic circuitwere parameterized in order to meet the best field design. In the firstmodel, the edge spiral line on the sector was parameterized as shownin the Fig. 4.30.By means of this parameterization it was possible to change the sec-tor width in order to vary the magnetic field value locally and at the

64 4 Magnet Design

Fig. 4.30. Parametric model simulated. The hill-edge points, which we have parameter-ized, are pointed out with red circles

same time, if necessary it allowed us to modify the spiral angle of thesector in order to increase the axial focusing effect. The focusing limitis therefore mainly dependent on the spiral term and increases almostlinearly with the spiral constant.A limit is set however by the reduced spiral efficiency since in reality,the magnetic focusing term does not increase as the geometric one forhigh value of the spiral constant. The difference between the geomet-ric and magnetic spiral can be better appreciated by considering therespective spiral focusing terms which are presented in Fig. 4.31.The closeness of the two curves, up to R=120 cm, shows that the

magnetic spiral depends mostly on the hill shape with negligible effectfrom either the valley shims or the holes drilled in the poles. Differ-ences exist in the central region but they are not relevant since therethe spiral focusing is negligible there. The two curves are instead quitedifferent near the extraction region because of the radial shape of thesector and the last valley shims positioned between the R=130 cm and132.5 cm. This discrepancy points out that in the extraction region,


Fig. 4.31. Geometric spiral function (dashed line) compared with the magnetic spiralfunction (solid line), along the radius in cm.

where the axial focusing, the isochronism and the fringing field fall-offplay simultaneously a role in limiting the machine performances, asatisfactory design of the sector geometry can be reached only by trialand error.

Fig. 4.32. Frequency tunes comparison. νr and νz during an unstable configuration(black and magenta lines) and after some modification of the spiraled sectors in order tomeet the dynamic stability (red and blue lines). The radius (m) is shown on the abscissaaxis.

66 4 Magnet Design

For example, at radius R =125 cm, where the axial focusing frequencyreaches the minimum value, the magnetic spiral focusing term is 15%lower than the corresponding geometric term. This discrepancy had tobe compensated by increasing the spiral angle and shaping the sectorwidth. The figure 4.32 shows the results after the preliminary modifi-cations of the spiral had been carried out in order to meet the requiredaxial focusing.In order to decrease the number of ampere-turns in the main coils andto allow for the last orbit of the accelerated particles to be broughtcloser to the pole edge and simplify the extraction operation, the gaphill size was reduced to 5 cm. Such a modification already allows usto increase the flutter and generally, to raise the axial focusing effect.Compared to the previous value of 7 cm, the new flutter curve is shownin Fig. 4.33.

Fig. 4.33. Flutter versus radius (cm) plot. The flutter function before the gap wasreduced (red line) and afterwards (blue line)

An increasing of almost 20 % of the focusing properties of the ma-chine allows us to reduce the spiral angle. In fact, the vertical focusingtune is usually set to be greater than 0.15. The previously chosen spiral


line was rather prudent in that the tune was over 0.4. The hill-edgeline is thus unspiraled, which is beneficial when reducing capacitiveload to lower RF power losses.The hill-spiral lines of two designs and their corresponding focusingtunes are shown in the Fig. 4.34 and Fig. 4.35.

Fig. 4.34. Comparison between two different sectors with the spiral shape modified. Asshown in plot the new spiral (red sector) reduces the angular span of the sector by 13deg. The length of the hill-edge is reduced by 6% compared to the old shape.

Once the spiral hill edge and the angular width of the sector havebeen set, it is possible to refine the magnetic field by means of theshims located on the valley. A second model of the magnet was built,in which more parameters can be set in order to fulfil the magneticfield requirements.A file, containing the parameterized structure, was written in ASCIIformat. Part of this file, concerning the design of the valley shims isshown in detail and explained in the Figures .4.37 and 4.38.

68 4 Magnet Design

Fig. 4.35. vertical focusing tunes comparison before (blue line) and after (cyclamin line)introducing the new spiraled sector.

Fig. 4.36. Final model of the superconducting cyclotron.

4.9 3D simulation results 69

Fig. 4.37. This shown a part of the ASCII file. Each section is introduced by a titledescribing the part of the machine. Then there are a set of parameters which it is possibleto vary, in this case the radial span of the shims and their related height.

Fig. 4.38. This is part of the setup file. The setup conditions for the simulation and themeshing parameters are defined in another ASCII file. Once the file has been written, itis possible to run the code by means of a batch file containing only commands and callsto files where variables are defined.

4.9 3D simulation results

A good agreement between the isochronous field and the average fieldwas achieved without the help of any correction system (trim coils ortrim rods). The discrepancy is under 10 gauss from the central andthe extraction region as shown in Fig. 4.40.

70 4 Magnet Design

Fig. 4.39. Average magnetic field (gauss) simulated along the radius (cm).

A cone field in the central region is necessary to provide an axialmagnetic focusing and an appropriate starting phase for the beam, soas to ensure axial electric focusing.The amplitude of the main harmonics of the field as a function of theradius are plotted in Fig. 4.41. The peak value of the fourth harmonicC4 is in excess of more than 10 Kgauss around R=100 cm and inthe extraction region drops to about 7 Kgauss. The amplitude of theother harmonics are much lower. The validity for beam accelerationof the computed field is tested with codes calculating the equilibriumorbit [23] which provide the phase curve and the focusing frequency

Fig. 4.40. Relative difference between the average and the isochronous magnetic field.


Fig. 4.41. Harmonic amplitudes of the main C4 coefficient (red line) and C8 (blue line).

(plotted in Figures .4.42 and 4.43).The working path in the (νr ,νz) plane is shown in Fig. 4.44 where themajor resonance lines are plotted. Examination of the curve shows thatthe extraction, via the excitation of the νr=1 resonance with a first

Fig. 4.42. Betatron frequencies radial (blue line) and vertical (magenta line) versus theenergy of the ions (AMeV).

72 4 Magnet Design

Fig. 4.43. Phase History (deg) versus the energy of the ions (AMeV).

Fig. 4.44. Working diagram and resonances.

harmonic coil (eventually), takes place in proximity to the νr +2νz = 0resonance which, however, should not be crossed. The second dan-gerous resonance νr = 2νz (Walkinshaw resonance), should not be aproblem, because it is crossed during the final turns.On the median plane, the maximum magnetic field of 4.6 tesla is lo-cated in proximity to the hills, whilst the magnetic field does notexceed the 5.4 tesla on the structure as shown in the Figures 4.45 and4.46.Such a magnetic field was achieved by setting a coil density current

of 40.5 amp/mm2. This low value of current crossing the coils allowsus to maintain quite conservative the maximum strength on the coils


Fig. 4.45. Magnetic field map (gauss) on the median plane.

as previously described (see section 4.6)and shown in tab.4.7.

(a) section

MAGNETIC VALUES

Bhill peak 4.64 teslaBvalley min 2.37 tesla

B peak 5.4 teslaB min 0.24 tesla

Energy stored 28 MJ

(b) section

COILS VALUES

Size 150×210 mm2

Internal radius 142 mmDistance to the M.P. 7.5 cm

Current density 40.5 amp/ mm2

Max Strength Axial 11 MNMax Strength Radial 24 MNMax Magnetic Field 5 tesla

Table 4.7. Magnetic values calculated from the simulation (a) and coils parameters (b)

74 4 Magnet Design

Fig. 4.46. Magnetic field (gauss) distribution on the magnet.

4.10 Fine Tuning of the Magnetic Field

Isochronization of the magnetic field, as previously mentioned, mustbe obtained with the independent excitation of two coils and with theshape of the sectors, since the correction capability of whatever sys-tem is limited to a few hundred gauss. Because the cyclotron is ableto accelerate one set of ions , in our opinion is better to make the finetuning of the magnetic field by means of few mechanical iron moveableshims instead of several trim coils.The final setting of the magnetic field was achieved by adding thecontribution of a set of trim rods (TR). Eleven TR per each hill,placed along the center of the hill and with radial distance of 10 cmamong them, were simulated to obtain the individual magnetic fieldmap whose form factor is shown in Fig. 4.48. Inside each sector thecylinder holes are drilled starting from 6 cm to the median plane upto 4.5 cm (height = 1.5 cm) with a diameter of 5 cm.

4.11 Magnetic Forces calculation 75

Fig. 4.47. Trim Rods positioning .

Fig. 4.48. Individual form factor of the magnetic field expected from TR. The magneticfield intensity (KG) is shown in ordinata axes

By the preliminary trimming of the field the phase growth has beenreduced within 10 deg as shown in the plot shown in Fig. 4.49.

4.11 Magnetic Forces calculation

In order to evaluate the forces due to the magnetic fields, we use the

following formula which gives the resultant−→R on a magnetized element

Λ:

76 4 Magnet Design

Fig. 4.49. Difference between phase curve before and after the magnetic fine tuning bymeans of trim rods.

−→R = µ0

∫Σ

((−→H · −→n )

−→H − 1

2H2−→n )dΣ

where Σ is the surface enclosing Λ,−→H is the magnetic field intensity,

µ0 is the vacuum permeability, −→n is the unit vector normal to thesurface Σ.For the preliminary study we assumed a uniform axial magnetic field.The formula described above reduces:

F =1

2µ0

∆B2 ·Σ

where B2 is the difference between the square of the magnetic fieldinduction at the inner and outer surface. Supposing Bouter ≈ 5 teslaand Bouter ≈ 0.3 tesla, the attractive force between the poles is:

F =π

2µ0

(52 − 0.32) · 1.32 = 5.262 · 107newton ≈ 5300 tons

The knowledge of the maximum stress and the consequent deformationexperienced by the poles is important for two reason:

• first of all, the maximum stress must be well below the elastic limitof the iron;

• second, the median plane gap variation must be kept within 0.1÷0.4 mm for a reliable operation of the cyclotron.

This in order to maintain within a few gauss the first harmonic com-ponent of the field arising from a possible asymmetry in the pole de-formation. Moreover, this variation must be taken into account for all

4.11 Magnetic Forces calculation 77

Fig. 4.50. Axial force distribution on the sectors (newton/m2).

those parts of the magnet yoke which have a mechanical interactionwith other components of the cyclotron, like the vacuum chamber ofthe cryostat.By simulating the magnetic structure, a more detailed evaluation ofthe stress was carried out. The attractive force between the poles is4300 tons instead of 5300 calculated above. While the forces acting onthe sectors was evaluated as 472 tons, with an maximum strength of76 kg/cm2, located at the outer radii of the hills where the magneticfield is higher, as shown in Fig. 4.50.

5

Radio Frequency System

To accelerate particle beams, four normal conducting Radio Frequency(RF) resonant cavities are foreseen. The magnetic field and the massto charge ratio being constants (Z/A=0.5), the orbital frequency of theaccelerated ions is fixed at ∼23 MHz. In order to ensure the maximumenergy gain per turn and thus reduce the number of turns inside themachine, it has been decided to operate the RF cavities in 4th harmonicmode. Consequently, the resonant frequency of the cavities has beenset as fRF =93 MHz.The preliminary design was accomplished using advanced 3D codes[21] [28] in order to better predict the behavior of the resonators andto optimize their performance.

5.1 The RF resonant cavity

The resonant cavity proposed for the RF system of SCENT, is thestandard double gap delta cavity. It can be considered as a deformedλ/2 coaxial line with a capacitive delta shaped electrode (the DEE)defining the accelerating gaps and an inductive part forming the ver-tical supporting stem.The resonator geometry depends strongly on the shape of the mag-netic circuit of the cyclotron. In a compact superconducting cyclotron,the cavities are installed completely inside the machine and they arehoused inside the so-called valley regions (see Fig. 5.1). Since theseregions are spiralled to increase the axial focusing effects on the accel-erated beams, the copper electrodes of the cavities have to be shapedin the same way (see Fig. 5.2).Generally RF systems in cyclotrons work in the variable range 10-50

80 5 Radio Frequency System

Fig. 5.1. Top view of the SCENT machine. The black regions are the valleys where thecavities will be housed.

Fig. 5.2. Layout of a single DEE between the poles.

MHz, so they need large vertical apertures in the yoke of the magneticcircuit to house the stems and the moveable short-circuits. In the caseof the SCENT machine, the operating fixed frequency of 93 MHz al-lows to house the resonators inside the valleys, where the available

5.2 The modelling method 81

depth is 45 cm from the median plane and moveable short-circuits arenot required. To minimize the risk of discharge, voltages lower than 70kV are required in the central region, whilst voltages around 120 kVor higher are required at the outer radii to achieve an adequate turnseparation of the beam in the last turns, in order to optimize the ex-traction efficiency performed by the electrostatic deflectors. Due to thelarge spiral angle, the electrode equivalent length is around 2.5 meterswhich results in a LINER-DEE equivalent capacitance of more than200 pF. The resonant frequency of a resonant cavity was calculatedas:

fo =1

2π√

L · CIn order to compensate the high capacitance C and to attain thehigh resonant frequency fo, it is necessary to decrease the inductivecontribution L by increasing the diameter of the stem. On the otherhand, since the diameter extension of a single stem is limited by therestricted angular width of the DEE (∼35 deg), the value of the res-onance frequency will have an upper limit. In order to increase thislimit, it is necessary to introduce new stems. The advantages of usingmore stems are as follows:

• It is possible to shape the voltage distribution along the acceleratinggap with voltage increasing along the radius;

• The three supports for the heavy copper DEE (instead of one)provide a better mechanical stability of the cavity;

• The currents are split and flow towards the nearest stem, reducingthe local power dissipation (i.e. the current density for each short-circuit is lower) and increasing the cavity performance in some case.

Finally in order to control the radial voltage distribution and to obtainthe required resonant frequency, a multi-stem configuration appears tobe the only feasible solution [29], [30], [31].

5.2 The modelling method

The classical method used to study the cyclotron cavities has beenwell explained by J. Vincent [32]. The modeling technique consistsof slicing the TEM-mode resonator up into pieces that can be accu-rately described as equivalent linear circuit elements. The response


of the resulting circuit is then equivalent to the response of the ac-tual resonator within some predictable margin of error. The successof this method depends on the ability and experience of the modelerwho has to know in advance the electro-magnetic (EM) field distri-bution. In the past, it was convenient to apply such a method to thestudy of the resonant cavities, since the numerical codes dedicated tosolving these problems were difficult to use, too expensive to buy andrequired great computational resources. Today, 3D codes have becomemore user friendly, are not as expensive and the hardware costs havedecreased considerably.To study complex structures like multi stem cavities, it is necessary touse numerical 3D codes, Finite Elements Method (FEM) based, which

Fig. 5.3. Average voltage distribution along the gap of the same cavity with two differentpositions of the stem. The peak value of the voltage is located where currents are higher.The EM field configurations of both cavities are strongly unbalanced, and small shiftsof the stems cause considerable changes about the resonant frequency and the voltagedistribution.


allow to predict the electro-magnetic fields distribution inside the res-onator. By introducing one or more stems inside the cavity, it becomesvery difficult to know the electric field distribution along the gap, andafterwards the voltage configuration of the DEE. In such a cavity it isdesirable to get the maximum voltage in proximity to the extractionregion, i.e. an electric field distribution arising from the inner regionsof the DEE towards the outer ones. To obtain this, it is necessary toplace the stems at appropriate positions on the DEE. Yet, the diffi-culty lies in achieving the required voltage distribution without harmthe resonance frequency and vice-versa. We noted, indeed, with thefirst results of the simulations, that small shifts of the stems alongthe DEE cause different distributions of the EM fields, also modifyingthe resonance frequency of the resonator. In some case, it was pos-sible to invert the electric field distribution on the DEE completely(see Fig. 5.3). Therefore, introducing more stems inside the resonatormakes the structure itself electro-magnetically unpredictable.To overcome such a difficulty, we have adopted a dedicated modelingmethod which will now be described.

For simplicity’s sake, we shall consider the median symmetry planeand consider quarter-wave resonators (QWR). A multi-stem QW res-onator can in principle be modeled as several paralleled QWR (orsub-system), as shown in Fig. 5.4,all having the same eigen-frequency(fi ≈ fo)and whose EM field distribution stays almost unchanged ifthey are coupled so that they don’t interact with each other (mi ≈ 0).This is feasible when the stems are far enough apart, and each QWR issimulated with the proper boundary conditions, as is shown in Fig. 5.5.

Fig. 5.4. The equivalent circuit representing three sub-systems of the cavity magneticallycoupled.


Fig. 5.5. Example of cavity subdivision. Normal magnetic boundaries are selected onthe vertical separation surfaces.

Each QWR consists of one stem, charged on one section of theelectrode. Such a structure usually has a voltage distribution with aminimum near the stem base, increasing towards the DEE extremi-ties. In order to ensure that each QWR resonates at the same resonantfrequency and to find an arrangement such that the whole structureremains electro-magnetically stable, the diameter and the position ofthe stem on the DEE have been parameterized and optimized sepa-rately (see figures 5.6 and 5.7). The best position was sought in orderto maximize the local quality factor Q which means optimize the res-onant frequency fo value, because

Q = 2πf0U

P

whereU is the EM energy storedP is the Average Power dissipated

and minimize the magnetic interaction with the adjacent sub-systems. Once the position had been determined, the diameter of thestem was increased in order to attain the required resonant frequency.

Because the coupling between sub-systems is very low, combiningthem produces an entire structure resonating at the fixed frequency;one that is optimized so that all of Q is maximized. Moreover, the


Fig. 5.6. Example of sub-system optimization. In this case the inner part of the cavitywas optimized. In order to maximize the resonant frequency (right side plot), the positionof the stem was varied along the DEE. The stem was located where the frequency attainsthe maximum value.

Fig. 5.7. Example of sub-system optimization.In this case the outer part of the cavitywas optimized. Once the position of the stem was fixed, the diameter of the cylinder hadto be increased to intercept the required value of resonant frequency, i.e. 93 MHz .

combination of two or more sub-structures like this, gives a voltagedistribution where the adjacent maxima join in a plateau as no cur-rent has to flow in either direction. By cutting the DEE into sectionsof different length, it is possible to obtain several different profiles,where the lower voltage corresponds to the minimum of the sectionwith higher capacity.To simulate each section with the proper boundary conditions, we

tested two different solutions, both giving similar results.In the first case, a magnetic symmetry condition is set among two ad-jacent stems. This method greatly reduces the number of mesh cells.In the second case, tested on a simplified structure, each stem with its


Fig. 5.8. Results of the simulations of different stem configuration of the same cavity. Inboth of cases the modelling method described has been applied. The resonant frequencyis the same and voltage distribution results stable. The blue regions on the both DEEsindicate the current separation in agreement with cuts made when the optimisation wasdone. In such a way EM configuration results very stable. By moving the stems fromtheir position, the voltage distribution does not vary significantly.

section of DEE, is simulated in the whole cavity volume. The well splitcurrents on the DEE surface of the whole model lead to a good sepa-ration and low magnetic coupling between the different QWRs. Thenthe whole structure is simulated using the geometry found and fewlocal adjustments are made to compensate the effects of the method’sapproximations. At least one simulation of the whole volume is re-quired to check the position of other resonant modes that should notbe too close.In the case of a 3-stem structure, for instance, the two nearest modestypically have voltages of opposite sign on the DEE extremities, orlower voltages on the extremities rather than on the central part.By means of such a method it should be possible to optimize res-onator with three or more stems (see Fig. 5.8).Increasing number ofstems improves the control of the voltage distribution along the gap,

5.3 The cavity geometry 87

but it could harm the performances of the cavities in terms of lossesand power consumption. Indeed, increasing the number of stems, inorder to maintain the frequency value, the diameter dimension willbe smaller, reducing the lateral surface of the cylinder. Because of thecurrents flowing through the surfaces of the stems, the local power den-sity could increase although the current value is reduced. The choiceof two, three or more stems, depends on the geometric structure of thecavity and on the voltage profile desired.

5.3 The cavity geometry

As regards the cavities we expect to use in SCENT, in order to attainthe high resonant frequency, we have been forced to insert at least threestems. Thus, we have chosen to have higher electrode voltages on theinner and outer extremities and to keep a lower voltage in most of theDEE, in order to optimize the shunt impedance (Rsh = ωoLQ). Byapplying the above described method, we have obtained the resonatorshown in Fig. 5.9.The large central stem (low inductance) concentrates the currents

Fig. 5.9. Layout of the cavity.

coming from most of the electrode, while the two lateral ones (highinductance) resonate with the much smaller capacitances of the DEE


extremities. The total height of the half-wave cavity is 80 cm, whilsta vertical aperture of 30 mm along the median plane is left for thebeam. In order to reduce the DEE capacitance and the electric field,the acceleration gap size ranges from 1.5 cm at inner radii to 6 cmat the outer end. The electrode’s angular width gradually increasesto 39 degrees near the first stem, then it is rather constant in theproximity of the main stem and decreases to 23 degrees at full radius(see Fig. 5.2). The three stems are centered at radii 30, 90 and 120cm. Their diameters are 8.6 cm for the external stem and 10 cm forthe internal stem. The main stem extends 32 cm radially and up to18 cm as minor axis.

Fig. 5.10. Layout of the cavity inside the cyclotron.

5.4 The simulation results

The simulations have mainly been performed with the eigenmode mod-ule of Microwave Studio (MWS) [28] and Table 5.1 reports the calcu-lated parameters.The effective voltage distribution, going from 60 kV in the injection

region to a peak value of 120 kV in the extraction region, is shownin Fig. 5.11. The values are obtained by integrating the electric fieldprofiles across the gaps in the median plane of the cavity.The current distribution on the DEE surface is shown in Fig. 5.12,

5.4 The simulation results 89

Table 5.1. RF cavity parameters.

Fig. 5.11. Effective voltage vs radius: the internal gap distribution (squares), the externalone (diamonds) and the average (triangles).

which shows very clearly how the different sections interact with thestems and how the currents are well split next to the cuts previouslymade for optimizing each sub-system. The currents, coming from thelarge radii, gather along the gaps towards the stem and heat theelectrode along the gap. The maximum current density is about 69amp/cm and is located on the edges of the two gaps near the externalstem. In this region the maximum power density is estimated at 16watt/cm2, for an average maximum voltage of 120 kV (see Fig. 5.13).High currents are also visible at the base of the external stem, requir-


Fig. 5.12. Current distribution on the complex DEE-stems and on the LINER. As shown,the critical regions are the outer stem, where the voltage is highest and the short circuitof the same stem.

ing special care for the connection to the electrode and on the outerstem.Loss calculations were also carried out with the SOPRANO [21] codeand compared with the MWS results in table 5.2.

5.4 The simulation results 91

The data did not show remarkable difference between the results of

Table 5.2. Comparison between losses calculated with SOPRANO and MWS for differ-ent cavity elements.

the two codes. Power losses did not exceed the 44 kW per cavity at120 kV of maximum average voltage (160 kV of peak). To take into ac-count the surface roughness and the average working temperature, thecopper conductivity has been set as 5.0×107 S/m (instead of 5.8×107).We have neglected the quality of the contacts and the beam loadingeffect (the current of the beam accelerated is relatively low).

Fig. 5.13. Power density distribution (watt/m2) for 150 kV of peak voltage. The peakvalue is 25 watt/cm2.


To prevent sparking at regions of high surface electric field, we haveadjusted the nose shape and gap length so that the surface electricfields do not exceed the Kilpatrick field limit [33] (see Fig. 5.14).

Fig. 5.14. Electric field distribution on the surface of the cavity. The maximum value isless than 11 MV/m which represents the Kilpatrick limit at 93 MHz.

5.5 Coupler Design

To feed the power from the RF amplifier into the cavities, we havechosen an inductive coupling, placing a coupler at the top of the liner,near one of the stems. We have chosen these areas as they are easilyaccessible from the outside and have a high magnetic field. In fact in-ductive coupling is achieved by perturbing an elevated magnetic fieldarea of the desired mode with a current coil (loop).The interaction of such field with the coil, allows the transfer of powerfrom the source to the cavity.

5.5 Coupler Design 93

The dimensioning of the loop has been performed with the aid of elec-tromagnetic codes CST MWS and ANSOFT HFSS [34], varying theloop’s area; in fact, varying the section of the loop, the interactionof the magnetic field with the coupler and therefore the transfer ofpower. In short, the efficiency of the coupler varies.By studying Scattering’s parameters by means of Eigenmode-DriverModal simulations, we get critical coupling in terms of minimal reflec-tion at the coupler’s port by studying VSWR, impedance adaptationby analysing Smith Chart and critical coupling by comparing the cav-ity’s quality factors such as Q0 and Qext.Critical coupling is achieved if the cavity is excited from a source whoseinner impedance is equal to the characteristic impedance of the cavity.This condition implies that the energy supplied from the generator,is evenly distributed between source and cavity, with the consequentequality of the values of Q0 and Qext. We evaluated the values of Q0

and Qext applying a numerical method on the S11 parameter obtainedfrom the simulations and obtained the curves shown in Fig. 5.16.We can get critical coupling by positioning a 22 cm2 area loop near

the main stem and, with a similar analysis, by positioning a 6 cm2 loopnear the external stem. The picture in Fig. 5.15 shows the simulatedloop configuration near the central stem.A power amplifier prototype able to deliver up to 50 kW in the fre-

Fig. 5.15. Inductive loop layout.


Fig. 5.16. Intercepting Qext value with the Q0 one, we found best loop area whichoptimizes the coupling of the feeder with the cavity.

quency range 60-110 MHZ has been built over the past few years atLNS to feed the cavity named Chopper-500 [35], and is under test-ing. So the cavities could be driven by four independent RF amplifierssimilar to our inexpensive prototype. According to our experience theuse of four amplifiers instead of one allows us to operate with mediumsize RF transmission lines and feasible and reliable couplers.

5.6 Considerations

Multi-stem cavities are a good solution to the design of cyclotron res-onators with challenging DEE shapes and dimensions. Although thedegrees of freedom of the structure grow with the number of stems, theRF design can be approached with some simple but realistic approxi-mations that, with the help of modern 3D EM codes, lead to a quickdefinition of the stem diameters and positions. Then, multi-stem cav-ities can easily be designed to provide the voltage distribution neededto optimize the power consumption, to reach high resonant frequenciesin spite of the DEE dimensions, and to reduce the current densitiesand the total losses. Last but not least, using more stems will also helpto increase the mechanical stability of the structure and allows us toprovide an easier path for the DEE cooling pipes.

6

Injection and extraction design

The preliminary results of the study concerning the injection and ex-traction processes, are shown in the following section.The proposed cyclotron will be equipped with an external ECR ionsource able to deliver both H+

2 and Carbon ions fully stripped. A so-lution with two separated sources to reduce the sweeping time withinproton and carbon treatment, is under study. The ions are deliveredaxially, through an injection line, into the machine, where an elec-trostatic inflector bends the beam by 90 deg from its axial path intothe cyclotron median plane. After this, the optimized central regionaccelerates for the first turns the beam, in order to maximize the ac-ceptance of the machine. The beam is accelerated by means of theresonant cavities up to the external radii, where a stripper foil is po-sitioned to extract the protons if the H+

2 is accelerated, or two elec-trostatic devices deflect the carbon ions from its last accelerated orbitoutside the machine.

6.1 Injection line study

The function of the injection line is to transfer the beam deliveredby the source to the center of the cyclotron, and to achieve matchingbetween the source emittance and the cyclotron acceptance. A typicallayout of the injection line is shown in Fig. 6.1. It consists of thesource coupled to a solenoid lens, which focus the beam at the objectpoint of a bending magnet. This allows us to perform the charge stateanalysis. After the analysis, the beam is focused at the entrance of thecyclotron’s inflector by a second solenoid lens. Faraday cups have to beinstalled at the object and image points of the analyzing magnet and

96 6 Injection and extraction design

just at the entrance of the cyclotron yoke to check the injected beamcurrent. In the present case, the beams delivered to the cyclotron willonly be H+

2 and 12C6+. The H+2 beam will probably be quite free of

contaminants, whilst the 12C6+ beam could be delivered together withother contaminants like H+

2 , 14N7+, 16O8+ and so on. These contami-nants have the same charge to mass ratio so it is impossible to cleanthe beam by means of the analyzing magnet. For this reason, we donot consider the use of the magnetic analyzer necessary. Fortunately,the isochronous magnetic field of the cyclotron is able to select therequired beam. Therefore, in order to reduce the cost and simplifythe injection line, we propose installing the source directly along thecyclotron axis, as shown in figure 6.1. Only if it is considered usefulto plan the installation of two sources, will it be necessary to install asmall ± 30 deg bending magnet to allow the sweeping from one sourceto the other in a short space of time. The final decision of whetherto install 1 or 2 sources depends very much on the treatment center’sworking plan. It is not crucial to the design of the whole facility. Asa result, this decision has been postponed. In this configuration just

Fig. 6.1. Two different layout (with and without magnet analyzer) of the injection line.

one solenoid lens is necessary to focus the beam at the entrance ofthe cyclotron’s inflector. (It might not be possible to achieve a goodemittance matching, and 50% of the injected beam could be lost dueto the mismatch. This will be dealt with later. Fortunately, the ability

6.2 Inflector Design 97

of the source to deliver beam current in excess of the amount requiredallows us to tolerate a large amount of beam losses.) Just one Faradaycup will be installed before the entrance of the axial hole of the cy-clotron. The position and the characteristics of the solenoid have beenselected to minimize the beam spot at the entrance of the inflector.The layout of the injection line and of the beam envelope along it,does not change significantly if we replace the solenoid lens with anelectrostatic quadrupole doublet. The are some advantages to usingan electrostatic quadrupole (EQ) versus a magnetic solenoid. The EQis lighter and cheaper than the solenoid magnet, it is not influencedby the magnetic field of the cyclotron, and its rotation along the axisof the beam line allows control of the radial and vertical mixing afterthe spiral inflector. Of course, it is necessary to shield the electrodesto prevent possible impact on them from the halo and spurious beams.

6.2 Inflector Design

The properties of the spiral inflectors are described in historical refer-ences [36–38]. Here we repeat them briefly.We will first define a radius of curvature A (which the ion would haveif it were acted upon by an electric field in the absence of any mag-netic field -in fact it is the height of the spiral infector) and a radiusof curvature Rm(which the ion would have if it were acted upon by amagnetic field in the absence of any electric field) as

A =T

qEe

Rm =p

qB

where T is the kinetic energy of the ion, p the total momentum of theion, B the magnetic field value, q the electric charge of the ion, andEe the magnitude of the electric field.In length units p is given by

p =106 · Eo

ηBc

√T

Eo

(2 +T

Eo

)

where η is the specific charge of the ion, Eo the rest energy of the ionper nucleon and c the speed of light. The total momentum of the ionp through the inflector is constant.The spiral inflector has two free parameters. One parameter is A, and


the second one is k′ (a parameter which is related to the direction ofthe electric field, called tilt).It is convenient to de”ne a new parameter K as

K =A

2Rm

+k′

2

It is assumed that the direction of the electric force on an ion locatedat the entrance of the inflector is in the u direction (u is one of theunit vectors of the so-called optical coordinate system. The other twounit vectors of this coordinate system are h and v).Given that we wish to apply an electric field perpendicular to thedirection of motion (so there is no energy gain for the central ray)then the electric field vector must lie in the u-h plane. In the simplestinflector (i.e. no tilt) the appropriate direction for the electric forcewould be given by the u vector and the spacing between the twoelectrode surfaces (the gap width) is held constant and the magnitudeof electric field in the u direction is a constant. In the case of a tiltedinflector, the idea is to use a component of the electric field to modifythe beam centering. In this case the direction of the electric force is theur direction (ur is one of the unit vectors of the so-called rotated opticalcoordinate system. The other two unit vectors of this coordinate systemare hr and v), the gap width d is narrowed in order to maintain aconstant electric field in the u direction. The magnitude of the electricfield in the ur direction, Eur, is raised as the ion proceeds along thetrajectory in a way described by

Eur =2T

ηA

√1 + k′2sin2b

where 0 ≤ b ≤ π/2 is the angle of inflection (the instantaneous anglebetween the velocity vector and the vertical).

6.2.1 Numerical Results

The preliminary calculations have been done by analytical approxi-mations described above by means of MATLAB code [39]. Once thegeometrical structure was determined, the inflector was designed andsimulated by means of 3D code OPERA to calculate the effective elec-tric field distribution. The trajectory of the ions inside the inflector


was studied by using the tracking code of OPERA itself.We have selected, preliminary, the spiral inflector with:

Rm = 1.356 cm A = 2.2 cm, k’ = 0

the electrode spacing at is entrance is 0.6 cm, and the width of itselectrodes is 1.2 cm (aspect ratio of 2).Different configurations of the injected beam were considered in orderto study the behavior of the beam at the inflector’s exit.In our calculations of the paraxial ion trajectory through the spiralinflector we have been trying to consider the case when the followingthree requirements should be satisfied:

• the maximal size of the beam emittances at the inflector entranceshould be 150 π mm mrad;

• the beam envelopes through the spiral inflector have to be containedwithin a small volume to prevent the beam being affected by thebad field quality in the fringing field;

• the size of the beam emittances at the inflector exit should bereasonable for acceptance of the central region of the cyclotron.

Fig. 6.2. This picture shows a layout of the beam trajectory (8 particles) starting at 13cm from the median plane. In this case a defocusing configuration (red line) was takeninto account. Because of the lens effect of the main magnetic field, the beam arrives atthe entrance of the inflector with a convergent distribution on the transversal plane (blueline). The beam delivered through the inflector maintains the focused distribution, andthe beam size is well contained (black line).


Fig. 6.3. In this case, a waist distribution was considered at the starting point. In thiscase the beam at the exit of the inflector appears with a wide angular dispersion.

During the simulation of the trajectory, the magnetic field dueto the presence of the magnet was considered also. As mentioned

Fig. 6.4. In this case a beam with an angular dispersion was set as initial condition.Like the previously case, the beam exits from the inflector with a bad transversal phasespace configuration.


above, the presence of the solenoid lens positioned before the cyclotronchanges the phase of the ions arriving at the aperture. This phase thendetermines which rays (and thereby starting conditions) will enter theinflector.The Figures 6.2 6.3 6.4 show three different envelopes of the beamdelivered through the same inflector. The beam envelope shown inthe Fig. 6.2 seems to be adequate to fulfil the expected requirements.For this case, a more refined simulation has been done and results areshown in Fig. 6.5.

Fig. 6.5. Transversal phase space distribution of much more particles at the start point(blue line) and at the exit of the inflector (red line). In this accurate simulation, it ispossible to see the non-linear effects on the beam envelope during the transport throughthe inflector.

As shown from the plots described above, due to the fringing fieldeffect, few displacements from the central trajectory were observed.We can compensate the fringe field of the inflector by decreasing thevoltage between the inflector, by cutting certain pieces of the inflectoror inserting two collimators before and after the inflector. The finalconfiguration and the optimized inflector is shown in Fig. 6.6 andparameters are described in Table 6.1.


Fig. 6.6. Layout of the electrodes’ inflector and the position of the collimators.

Table 6.1. Final parameters of the inflector.

6.3 Central Region Design

Once the inflector shape was determined, a four harmonic central re-gion was designed. The accelerating structures in the SCENT machineconsist of four DEEs occupying the valley region of the magnet gap.These DEEs extend radially from some centimeter out to extractionradius. The space at the center of the machine (up to 5 cm) is usedfor the central region. The central region itself consists of a set of es-pecially shaped electrodes attached to the DEEs which accelerate theparticles from the inner region out to the main, electrode-free DEEstructures. The central region electrodes consist of posts mounted onthe center plug and electrodes mounted on the tips of the DEEs (andthus called DEE tips). the placements and shape of the various postsand electrodes define a range of starting conditions for the ions which

6.3 Central Region Design 103

can be accelerated out of the region. Further, the electrodes are posi-tioned so as to place the beam as close to its accelerated equilibriumorbit, upon exiting the central region, as possible. These electrodesthen limit the amount of beam which the machine can accelerate anddictate the quality of the beam.All electrode shapes and positions had to accommodate the inflectorhousing and guide the beam through the central region leaving it rea-sonably well centered. This placement, orientation and shape of theposts, inflector and DEE tips need to satisfy the following conditions:

• the inflector orientation must provide good vertical focusing forthe emerging ions. Since the first few gaps will provide the criticalvertical focusing, the inflector exit must be placed to assure thatthe ions will pass through the first few gaps with hte proper latephase required for electric focusing.

• Each DEE must be isolated, thai is, the field produced by a DEEmust be restricted to the area immediately surrounding that struc-ture and not allowed to bleed into the fields of neighboring DEEs.This prohibits cross-talk between the RF amplifiers and preventsthe formation of destructive radial fields along the beam path.

• the size and position of the DEEs electrodes nust provide neededstructural support at the DEE tips as well as preventing the cre-ation of radial fields in the beam path. By placing posts at the tipsof each DEE and larger secondary posts at a larger radial distancealong the DEE, the radial fields are excluded from the ion orbitand structural strength of the DEE is enhanced.

• the beam should be placed upon existing the central region suchthat during acceleration through the machine, it is well centered(a beam is well centered if it exhibits no coherent radial oscilla-tions). In this way any small centering errors produced by misplace-ments in the mechanical construction of the central region should becorrectable through the use of a possible first harmonic magneticbump. To achieve such beam placement the shape and positionof aforementioned posts are adjusted to misshape the acceleratinggaps during the first turn. The distortion of the gap shapes whichresults can be chosen to correctly place the beam at the end of thefirst turn.

Added to all the above conditions, are the constraints that the electricfields be kept as low as possible to limit sparking (Kilpatrick limit [33]),


and that the transmitted current be as large as possible. Since the DEEvoltage will be 70 kV for ions injection, an electrode separation of 8mm at least was chosen, limiting fields to 88 kV/cm.With this considerations in mind a preliminary central region was de-signed (see Fig. 6.7 that produced a well-centered beam (see Fig. 6.10)with good vertical focusing properties.

Fig. 6.7. Central region layout.

A code to calculate the trajectory of the beam was written byMATLAB ??. This code evaluates the beam dynamics (see Fig. 6.8,

integrating the standard equation−→F = q(

−→E +−→v ×

−→B ) using the carte-

sian coordinates and the time as variable of integration (the step ∆tis a small fraction of the RF period. The electric and magnetic fieldmaps were evaluated by OPERA code.A general procedure for the design of the central region, once definedthe accelerating structure (number and extension of the DEEs, har-monic number mode, gap width and height), can be schematized asfollows:

6.3 Central Region Design 105

Fig. 6.8. The 4th harmonic central region of SCENT. Central trajectories with startingtimes 110 deg to 120 deg are plotted.

1. track backward the accelerated equilibrium orbit up to the 1st DEEor 1st gap;

2. track forward from source or inflector up to the 1st gap or DEE;3. match in energy, phase and coordinates by geometrical reposition-

ing of the source, rotation of the inflector and DEEs or gap modi-fication;

4. check for axial focusing, phase acceptance, modify accordingly andrestar from point 2.

Generally, when space is available, the design of the central region isdetermined by the shape of the first two gap. The matching the ac-celerated equilibrium orbit (beam centering) is obtained within a fewmm. This error can be corrected using a 1st harmonic of the magneticfield, produced by tunable coil or trim rod, since the beam stays closeto νr = 1 for many turns in the central region.


Fig. 6.9. Energy gain for the first turns.

Fig. 6.10. The plot shows the X (solid line) and Y (dashed line) coordinates of thebending center. After few turns the beam seems to converge toward the magnetic center.

6.4 Preliminary study of the extraction process

Two kind of extraction are expected for this cyclotron.To have the proton beam, it is necessary to extract the accelerated H+

2

molecules by means of the stripping process. The carbon ions are de-

6.4 Preliminary study of the extraction process 107

livered outside the machine through the electrostatic deflectors (ED).In both cases the final energy will be 250 AMeV.

6.4.1 Extraction by ED

The extraction of a fully stripped light ion like C6+ needs the use ofED. According to our experience ED devices are reliable if electricfield and voltage applied are lower than 120 kV/cm and 60 kV respec-tively. The choice to operate the cyclotron with 4 RF cavities forcedus to install ED inside two hills. The housing of the ED have beendesigned to stay inside the 50 mm gap of the hill. The length of thetwo ED are 36 deg and 32 deg respectively. We remember that thehill width is 44 deg, so these length are quite conservative and theends of ED 1 are at about 8 cm from the boundary of the iron hill. Agap of 5 mm between the electrode and the septum was selected. Thesimulation studies show that an electric field of 110 kV is sufficientto extract the beam. To increase the margin of reliability our simu-lation studies were accomplished not with the last orbit but with a 2mm inner orbit. Following the second ED, placed on the same hill and4 deg long, the beam cross the first Magnetic Channel (MC), whichincreases the steering towards the outer radii. The distance betweenthe last accelerated orbit and the extracted trajectory at the positionof the first MC is 21 mm. A total of 10 MC have been expected tomaintain the beam envelope along all the extraction path at valueslower than ±1.5 cm radially. Vertical beam size is ever smaller than±1 cm. These beam envelopes were evaluated assuming a conservativeemittance value of 2πmm mrad at the extraction radius. The stripperangular straggling was left out due to the high energy of the beam andthe small thickness of the stripper. The number and the positions ofMC needs to be optimized.

6.4.2 Extraction by Stripping

The extraction by stripping has been largely used to extract H− =⇒p+2e, at energies as high as 520 MeV at TRIUMF and in many com-mercial cyclotrons [40]. The main limit of this kind of cyclotron is thesmall binding energy (0.7 eV) of H− ions which forbids the use of highmagnetic fields [41]. Because of the binding energy of the electronof the H+

2 molecule is about 20 times stronger than the H− one, itis possible to accelerate the H+

2 up to 250AMeV in an high magnetic


Fig. 6.11. The extraction trajectories by means of the stripping process and the elec-trostatic deflectors.

field machine to reduce its magnetic dimension and consequently thecosts [42]. There are also other differences between the stripping pro-cess for H− and H+

2 . In the first case both the two electrons have to beremoved to extract one proton and the foil has to be thick enough toguarantee the stripping efficiency of 100%, while for H+

2 if the moleculeis not stripped at first cross through the stripper, it turns inside thecyclotron and hits once again the stripper until it is stripped. Thenfor H+

2 it is possible to use a stripper with thickness smaller than forH− and then a longer mean life is expected. Moreover stripping of H+

2

produces electrons but while for H− the electrons are bent towardsthe center of the machine and hit the stripper foil after spiraling inthe magnetic field, for H+

2 the electrons are bent towards the outerradius, so an electron catcher can be installed to remove the electronsemerging from the stripper and strongly reduce the stripper damage.Another important advantage to accelerate H+

2 is the reduced spacecharge effect due to the lower charge to mass ratio as compared withprotons and the better emittance and higher currents of H+

2 sourcesas compared to the H− sources [43]. The main disadvantage to accel-erate H+

2 is its magnetic rigidity which is twice that for protons with

6.4 Preliminary study of the extraction process 109

the same velocity, nevertheless using superconducting magnets it ispossible to maintain the size of cyclotrons for H+

2 at reasonable values.

In Fig. 6.11 the extraction trajectories of the protons produced bystripping of H+

2 on a carbon foil, and of C6+ ions, extracted by electro-static deflectors, are shown. As it is quite evident the two trajectoriescan be delivered to the same direction. This result was achieved quiteeasily adjusting just the angular position of the stripper. Only thelast part of the proton trajectory needs to be steered towards the in-ner side of cyclotron to compensate the effect of the return yoke fieldwhich for proton is stronger than for Carbon beam. The dashed curve,in Fig. 6.11, is the trajectory without MC correction. Just three MCwere required to focus and to steer in the required direction the pro-ton beam. The radial position of the first MC on the proton trajectoryis at R=158 cm, quite far from the last accelerated orbit. This firstchannel is at radius larger than the outer radius of the main coil butinside the cryostat. The gap between the superconducting coils being150 mm, it allows to design and install the MC quite easily.

To validate the extraction studies performed with dedicated code,during the July 2005 a test (STRESS experiment) for a quasi extrac-tion by stripping was performed in the K800 cyclotron operating atLNS.A Ne9+ beam was accelerated up to 45 AMeV and stripped by a 0.1mm (22 mg/cm2) carbon foil. After one turn inside the cyclotron thebeam strike an alumina screen and a Faraday cup as shown in Fig. 6.12As shown in Fig. 6.13, the sizes measured on the screen show a reduc-tion of the vertical size from 11 mm to 9 mm, while the radial sizeincrease from 9 mm to 15 mm. The radial size increase is partially dueto the energy spread introduced by the stripper. Anyway the resultsagree quite well with the simulation. Also the full current measuredbefore and after the stripping scale according to the ratio of the chargestates. No beam losses have been detected.


Fig. 6.12. Simulation of the extraction trajectory of Ne9+ accelerated with CS at LNS.

Fig. 6.13. Picture taken from camera placed inside the CS during the STRESS experi-ment at LNS. This shows the beam halo before and afterward the stripper foil.

6.5 Beam losses and vacuum requirements

Due to the high binding energy of the H+2 molecule, ∼16.3 eV vs.

0.7 eV of H2, it is possible to use magnetic fields as high as 10 Teven at energies as high as 1 A GeV without any beam losses due

6.5 Beam losses and vacuum requirements 111

to electromagnetic dissociation. Beam losses, along the accelerationpath, are due to the interactions of the H+

2 ionized molecule with theresidual gases. These may produce stripping of the electron of H+

2 .In general for a molecule this probability is higher than for a simpleion. The fraction of lost beam particles was evaluated according to theBetz-Bohr model [44], that takes into account the beam cross sectionfor electron lost versus the beam energy, and it is summarized by thefollowing formula:

Ploss = 1− exp(− P

KT

∫σ(E)dl)

where P is the pressure (Pa) , l is the path length (m), σ(E) is thecross section of electron loss. The formula to evaluate σ(E) is:

σ(E) ≈ 4πa2o(vo/v)2Z2

t + Zt

Z2i

where ao and vo are the radius of orbital and the speed of the electron,Zi and Zt are the number of electrons of the incident H+

2 and of theresidual gases respectively, v is the velocity of the accelerated H+

2 .To check the previous formula and the value of the parameter ao weevaluate the beam losses for the Karlsrhue cyclotron and comparedwith the experimental results. Although our simulation overestimatesthe losses at inner radii and underestimates at outer, the agreementwas quite satisfactory. Perhaps the differences are due to our assump-tion of constant accelerating voltage at all radii (240 kV/turn) and ofuniform pressure of 2·10−6 Torr as reported by authors. On the con-trary if the pressure is better in the central region than in the outer,these could give reasonable explanations of the small differences.So we used the previous formula to evaluate the beam losses for thepresent cyclotron design. We assumed an accelerating mean voltage of800 kV per turn and a pressure value of 10−6 Torr. Most of the beamcurrent is lost at small radii along the initial orbits, where the beamhas low speed, while large amount of the power is lost at outer radiiwhere the beam is more energetic.We evaluate both the intensity and power losses either the expectedbeam lost all along the machine should be of 2.7%, which means apower loss of 100 W for an accelerating beam of 10 kW. These powerlosses are quite small and acceptable for the proposed cyclotron. The


tests of extraction from the K800 superconducting cyclotron of Cata-nia show that also with beam power higher than 100 W dissipatedat the extraction radius, very close to the cryostat, there are no de-tectable effects on the cryogenic parameters of the cryostat.Although a vacuum of 10−6 Torr is nowadays feasible also in a cy-clotron, anyway to achieve this goal a careful treatment of all thesurfaces inside the vacuum chamber is required.

7

Conclusions and Further Developments

Radiation therapy using hadrons - protons and ions - was proposedby Robert Wilson some 60 years ago, as these particles have a betterdose-depth distribution in tissue compared with X-rays. This gives animproved conformity of delivered dose on a tumour, allowing the doseto the tumour to be increased, while reducing the risk to healthy tissueand nearby critical organs. However, the relative size and complexityof accelerators for protons and ions mean that most of the 40,000patients treated with hadrons to date have been irradiated at largeresearch institutions where appropriate particle beams are available.Only in the past decade have suitable accelerators and beam-deliverysystems been developed, and the first few dedicated clinical-therapyfacilities for protons have now been built. Many more are in the plan-ning stage.Ions such as carbon are interesting because different biological mech-anisms are involved in their interactions, compared with protons, buttherapy systems using ions are even bigger and more complex thanthose for protons only. Two such facilities are under construction inEurope, one by GSI for the university clinics in Heidelberg, and oneat Italy’s national hadron-therapy centre, the Centro Nazionale diAdroterapia Oncologica (CNAO) in Pavia, with major involvementfrom INFN and Ansaldo Superconduttori.An ion-therapy system usually requires a synchrotron 16-25 m in di-ameter as its main accelerator, while most therapy systems based onlyon protons use compact cyclotrons - being a continuous source, the cy-clotron is more suitable for beam scanning across a tumour. In thiscontext, ACCEL has developed a novel superconducting proton cy-clotron only 3 m in diameter, with superior beam-delivery character-

114 7 Conclusions and Further Developments

istics. This forms the particle source for proton-therapy installationsat the Paul Scherrer Institute (PSI) in Villigen, and for Europe’s firstclinical proton-therapy system at the Rinecker Proton Therapy Center(RPTC) in Munich, which ACCEL is currently commissioning.

To combine the advantages of a superconducting cyclotron withthe goal of accelerating different species of ions in addition to pro-tons, INFN has developed a concept for a multiparticle-therapy cy-clotron. This is based on LNS Catania’s extensive experience both withcyclotron technology and operation, and with its successful proton-therapy programme for eye tumours, in which 87 patients have beentreated since 2002. Combining this experience with commercial andtechnical considerations of size and weight for transport, handling andoperational environment has led to a machine concept for providingbeams of 250 AMeV protons and light ions. Owing to their strongerinteraction in human tissue, carbon ions will have limited penetrationdepth, but will still cover relevant treatment cases, as has been shownby ion-therapy studies in Japan.

Recently Italy’s National Institute for Nuclear Physics (INFN), AC-CEL Instruments GmbH and Ansaldo Superconduttori are to collab-orate on the feasibility study of the machine developed at the INFNLaboratori Nazionali del Sud (LNS) in Catania. The newly formedcollaboration is designing this multiparticle cyclotron as a solution forthe worldwide clinical ion/proton-therapy market that is more costeffective, and less operator and maintenance intensive. While it willhave somewhat reduced energies for heavier ions such as carbon, it willhave superior beam characteristics compared with synchrotron-basedinstallations.

The preliminary design of this Superconducting Cyclotron able toaccelerate ions with charge to mass ratio of 0.5 up to the energy of250 AMeV, for Medical Applications was discussed in this job.The novel superconducting accelerator is a clear example of the ben-efits brought by advances over the past 30 years in the applicationof superconductivity to accelerators in particle and nuclear physicslaboratories.

December, 2005

PUBLICATIONS RELATED TO THIS JOB

• M.Maggiore et al., A superconducting cyclotron as a primary accel-erator for exotic beam facilities, Nukleonika 2003, 48, pp 165–167.

• L.Calabretta, M.Maggiore et al., A superconducting cyclotron asdriver for radioactive beam facilities, Nuclear Physics A734, 2004,pp 378–381.

• M.Maggiore et al., Conceptual design of the RF accelerating cavi-ties for superconducting cyclotron, to be published in Nuclear In-struments and Methods A.

• L.Calabretta, M.Maggiore et al., A novel superconducting cyclotronfor therapy and radioisotope production, to be published in NuclearInstruments and Methods A.

References

1. E.O.Lawrence and M.S.Livingston. Apparatus for the multiple acceleration of lightions to high speed. Physical Review, 1930.

2. H:E:Bethe and M.E.Rose. Maximum energy obtainable from cyclotron. PhysicalReview, 1937.

3. J.J. Livingood. Principle of Cyclic particle Accelerators. Argonne National Labora-toy, 1962.

4. M.S. Livingstone. High Energy Accelerators. Interscience, 1951.5. P.Finocchiaro et al. Laboratorio nazionale del sud. Nuclear Physics News, 1998.6. G. Ciavola et al. Recent developments of the excyt project. Nuclear Physics A,

(616):69–76, 1997.7. M.Maggiore. Studio di un ciclotrone superconduttore per h2+ da 1 a.gev. Degree

thesis on physics, University of Turin, 2000.8. L.Calabretta et al. Lns catania project for therapy and radioisotope production.

pages 179–191, Tokyo, Japan, October 2004. 18th International Conferece on Cy-clotrons and their Applications.

9. www.accel.de.10. www.iba worldwide.com.11. www.pi.hitachi.co.jp/rd-eng/product/industrial-sys/accelerator-sys/proton-therapy

sys/index.html.12. http://global.mitsubishielectric.com/bu/particlebeam.13. www.optivus.com.14. E.Pedroni et al. The 200mev proton therapy project at the paul scherrer institute:

conceptual design and practical realisation. Med. Phys., 1995.15. G.Kraft et al. Magnetic scanning system for heavy ion therapy. Nucl.Instrum.

Methods Phys. Res., 1993.16. G.A.P. Cirrone. The catana proton therapy facility: three years of clinical and dosi-

metric experience. Technical report, Laboratori Nazionali del sud, 2004.17. D.Friesel et al. Design and construction progress on the iucf midwest proton radi-

ation institute. pages 179–191, Paris, France, 2002. Europen Particle AcceleratorsConference 2002.

18. E.Pedroni et al. A novel gantry for proton therapy at the paul scherrer institute.pages 179–191, East Lensing, USA, 2001.

19. T.J.M.Symons et al. Physical Review Lettres, (42):40, 1979.20. J.A. Nolen et al. Commissionig experience with the nscl k1200 sc. In 12th Interna-

tional Conferece on Cyclotrons and their Applications, pages 5–9, 1989.21. Vector Fields Ltd, Oxford,UK. OPERA v.10.5 MANUAL.

118 References

22. H. Blosser and J. Kim. Optimized magnet for a 250 mev proton radiotherapy cy-clotron. pages 345–347, East Lansing, USA, 2001.

23. Michigan State University, NSCL, East Lansing, USA. Orbitcode v.12 MANUAL.24. P.Mandrillon et al. Feasibility studies of the eulima light ion medical accelerator.

pages 179–191, Berln, Germany, March 1992. 3rd European Particle Accelerator Con-ference.

25. Pate BD. Plans for radiopharmaceutical production at triumf. Prog. Nucle. Met.,4:4–61, 1978.

26. S. Humphreys. Principle of charged particle acclerator. Willey Interscience publica-tion.

27. M.Maggiore et al. A superconducting cyclotron as driver for ribs. Nukleonika,2(48):165–167, October 2003.

28. Computer Simulation Technology Gmbh, Darmastadt,Germany. Microwave Studiov.5 MANUAL.

29. Schillo et al. Compact superconducting cyclotron 250 mev proton cyclotron for thepsi-proscan proton therapy project. In 16th International Conferece on Cyclotronsand their Applications, pages 37–39, 2001.

30. E.Acerbi et al. Design of the rf cavity for a 200 mev psc. In 14th InternationalConferece on Cyclotrons and their Applications, pages 245–247, 1995.

31. C.Bieth et al. A very compact protontherapy facility based on use of hts. In 15th

International Conferece on Cyclotrons and their Applications, pages 669–671, 1998.32. J.Vincent. Modeling and analysis of rf structures using an equivalent circuit method-

ology with application to charged particle accelerator rf resonators. In Michigan StateUniversity, editor, Phd Thesis, 1998.

33. S.O.Schriber et al. Factors limiting the operations of structures under high gradient.In LINAC 98, 1998.

34. Ansoft Corporation, Pittsburg, PA USA. High Frequency Simulation System (HFSS)v.12 MANUAL.

35. A. Caruso et al. Chopper 500 status report. In 17th International Conferece onCyclotrons and their Applications, pages 669–671, 2004.

36. L.Belmont and J.L. Pabot. IEEE Trans. Nucl. Sci., (NS-13):191, 1966.37. L.W. Root. Master’s thesis, Univ. Of British Columbia, 1972.38. L.W. Root. PhD thesis, Univ. Of British Columbia, 1974.39. The MathWorks, Inc. Matlab ver 7.40. B.F. Milton. A 30 mev h- cyclotron for isotope production. In 12th International

Conferece on Cyclotrons and their Applications, page 145, 1989.41. G.M. Stinson et al. Nuclear Instruments and methods, (NIM 74):333–341, 1969.42. L.Calabretta et al. Superconducting cyclotron for acceleration of h2+. In 15th

International Conferece on Cyclotrons and their Applications, 1998.43. W. Joho. High intensity problems in cyclotrons. In 9th International Conferece on

Cyclotrons and their Applications, page 337, 1981.44. Betz. Rev. of Mod. Physics, 1972.

Design of a Superconducting Cyclotron for Exotic Nuclei ... tions is not to build a suitable...

Documents

Transcript of Design of a Superconducting Cyclotron for Exotic Nuclei ... tions is not to build a suitable...