This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 192.107.52.30

This content was downloaded on 26/08/2015 at 09:34

Please note that terms and conditions apply.

First acceleration of a proton beam in a side coupled drift tube linac

View the table of contents for this issue, or go to the journal homepage for more

2015 EPL 111 14002

(http://iopscience.iop.org/0295-5075/111/1/14002)

Home Search Collections Journals About Contact us My IOPscience

July 2015

EPL, 111 (2015) 14002 www.epljournal.org

doi: 10.1209/0295-5075/111/14002

First acceleration of a proton beam in a side coupleddrift tube linac

C. Ronsivalle1(a)

, L. Picardi1, A. Ampollini

1, G. Bazzano

1, F. Marracino

1, P. Nenzi

1, C. Snels

1,

V. Surrenti1, M. Vadrucci

1 and F. Ambrosini2

1 ENEA - Via Enrico Fermi, 00044 Frascati, Rome, Italy2 University La Sapienza-Roma I - Piazza Aldo Moro 5, 00185 Roma, Italy

received 3 March 2015; accepted in final form 3 July 2015published online 29 July 2015

PACS 41.75.Lx – Other advanced accelerator conceptsPACS 41.75.Ak – Positive-ion beamsPACS 87.56.-v – Radiation therapy equipment

Abstract – We report the first experiment aimed at the demonstration of low-energy protons ac-celeration by a high-efficiency S-band RF linear accelerator. The proton beam has been acceleratedfrom 7 to 11.6 MeV by a 1 meter long SCDTL (Side Coupled Drift Tube Linac) module poweredwith 1.3 MW. The experiment has been done in the framework of the Italian TOP-IMPLART(Oncological Therapy with Protons-Intensity Modulated Proton Therapy Linear Accelerator forRadio-Therapy) project devoted to the realization of a proton therapy centre based on a pro-ton linear accelerator for intensity modulated cancer treatments to be installed at IRE-IFO, thelargest oncological hospital in Rome. It is the first proton therapy facility employing a full linearaccelerator scheme based on high-frequency technology.

Copyright c© EPLA, 2015

The application of S-band (3GHz) linac technology inaccelerators designed for proton therapy has been pro-posed since the first years of the 1990s [1,2]. The useof high frequency increases simultaneously Radio Fre-quency (RF) efficiency (thanks to the relation with shuntimpedance Z ∼ f1/2) and maximum allowed accelerat-ing gradient, providing the design of compact therapyunits. The beam currents required for cancer treatmentsare orders of magnitude lower than the ones required bynuclear physics applications (a few nA vs. a few mA).Such low current magnitudes allow the design of acceler-ating RF structures more compact than UHF or L-band(200–800MHz) commonly employed in high-power beamgeneration and with smaller bore holes.

Previous designs of RF proton accelerators for medi-cal applications used a sequence of S-band coupled cav-ity linacs (CCL) only for the high-energy section (from60–70MeV up to the final energy of 230–250MeV). Theyare conceptually similar to the ones employed in conven-tional electron clinical machines for radio-therapy withphotons and electrons. Their cell length that is equal toβλ/2 (β = relative particle velocity, λ = RF wavelength)becomes prohibitively short (a few mm) at energies below

(a)E-mail: concetta.ronsivalle@enea.it

60MeV (particularly below 30MeV) with a drastic dropin efficiency.

Previous designs used two distinct approaches for thelow-medium energy section (injector) of the accelera-tor: highly invasive injectors such as long UHF linacs orcommercial cyclotrons [3] typically used for radioisotopeproduction. Although proposed and sometimes stronglysupported by several papers, these systems have not beenrealized nor fully tested yet. Our group introduced andpatented in Europe [4] a 3GHz standing wave accelerat-ing structure named SCDTL (Side Coupled Drift TubeLinac) characterized by a high shunt impedance in thelow-energy range (0.1 < β < 0.35) [5–8]. This struc-ture, composed by small Drift Tube Linac (DTL) tankscoupled by side coupling cavities, with inter-cavity focus-ing Permanent Magnet Quadrupoles (PMQs), can be in-cluded in the same development stream as the CCDTL(Cavity Coupled Drift Tube Linac) firstly proposed forhigh-current linacs and patented by Los Alamos [9], buthas been sized for low-intensity beams therefore allowinga more relaxed FODO lattice with more space betweenPMQs, larger number of cells/tank, smaller bore holesand therefore suitable for high RF frequency operation.SCDTL indeed allows using 3 GHz frequency for the wholeaccelerator, down to an energy range of 4–7MeV. The

14002-p1

C. Ronsivalle et al.

Fig. 1: (Colour on-line) Sketch of the interior of a SCDTLstructure with 4 gaps/tank.

dedicated S-band linear accelerator is considered a majorimprovement for proton therapy, with respect to exist-ing commercial facilities based on circular machines (cy-clotrons and synchrotrons). The main advantages of thistechnology are: modularity of the structure, standard andcost effective RF technology, low weight, beam losses onlyat low energies, pulsed operation at high repetition rateand rapid energy variation, well suited to support scan-ning applications. The low transverse beam emittance(typically 0.2π · mm · mrad rms normalized) allows tinygaps in the dipole magnets of the transport lines, thusconsiderably reducing their weight and cost. Moreover,the construction procedures (machining and brazing) arewell known and technology transfer from research centresto industry can be performed effectively. In this context,the good performance of the SCDTL medium energy partof the linac is a key issue.

This structure (fig. 1) consists of short DTL tanks witha small number of cells (4–7) resonantly coupled by sidecavities.

The operating mode of the total structure is π/2 withcoupling cavities field vanishing because of the excitationwith opposite fields by the neighboring tanks. The tanksoperate in phase opposition with a relative distance of(2n+1)βλ/2 (n = integer). The cells inside a tank operatein zero mode with a length βλ. The bore hole radius isvery small (2–2.5mm) and also the drift tubes are verysmall (12mm in diameter), and therefore the peak powerconsumption can be drastically reduced with respect toan equivalent CCL structure. Figure 2 compares CCLand SCDTL structure dimensions for different energies:at low energy CCL efficiency is dramatically affected bythe losses on the walls separating the very tiny βλ/2 longcells, whereas DTL tanks, containing several cells, can bebuilt longer providing high RF shunt impedance, reducedby the stem supporting the drift tubes by 15–20%. Theonly drawback is the reduction of the average gradient dueto the larger cell length (βλ instead of βλ/2).

A comparison between the shunt impedance of SCDTLand CCL structures vs. particle relative velocity is re-ported in [5]. Unlike the low-frequency DTL in which thefocusing is provided by quadrupoles placed inside the drifttubes, in this type of structure, where the small dimension

Fig. 2: Comparison of CCL and SCDTL at different energies.

of drift tubes forbids this arrangement, short PMQs (3 cmlong, max gradient ∼ 200T/m) are accommodated in theinter-tank space. No degradation of PMQ magnetic field isexpected because of the very low accelerated current andconsequently negligible stray radiation field all around themachine. The fixed gradient is not a problem for the beamdynamics if the tolerances in terms of strength (±4%) andalignment (±50μm) are satisfied. However, the use of ex-ternal PMQs allows an easy exchange both for reparationand for optimization of beam dynamics.

The side coupling cavities have to be long enough topermit the insertion of PMQs on the axis. Their lengthrange between 70 and 100mm, being very close to onewavelength, brings the TM011 mode into the pass-bandof the system [10]. A central axial post coupler, whosedimensions are optimized by 3D electromagnetic analysis,is used to move it out of the operational pass-band.

Beam dynamics design for the SCDTL is accomplishedin two steps. First, the DESIGN code [11] performssingle-particle dynamics that receives as input the valuesof the synchronous phase, the maximum PMQs gradientand the dependence of transit time factors on particleβ in the DTL tanks (as retrieved by the SUPERFISHcode [12]), and determines the main parameters of thestand-alone structure in terms of tanks length, num-ber of cells/tank, quadrupoles gradient. Then, by theLINAC code [11], multi-particle dynamics calculations inthe DESIGN-generated structure are done in order tocheck the quality of the design, to optimize the match-ing with the injector and to perform error and tolerancesstudies. Both codes DESIGN and LINAC were specifi-cally developed by Crandall on the basis of the models im-plemented in the well-known PARMILA code [13] largelyused for DTLs design. The absence of space charge effects,due to the very low currents required by proton therapy,simplifies the dynamics, but other problems arise becauseof the use of high frequency at very low energy, specifi-cally the coupling of longitudinal and transverse motionthat can increase the beam emittance. The dependenceon phase of the RF defocusing force given by [14]

Fr =E0I1ωr sinφ

βcγ(1)

14002-p2

First acceleration of a proton beam in a side coupled drift tube linac

Fig. 3: (Colour on-line) Layout of the 150MeV TOP-IMPLART proton linear accelerator at ENEA-Frascati.

is stronger at low energies. In addition, an unsustain-able emittance growth, that would eventually produce acomplete loss of the beam could be induced by paramet-ric resonances [15] if the longitudinal phase advance in aperiod L

σl =2πqE0T sin(−ϕs)L

mc2(βγ)3λ(2)

is a multiple of the transverse phase advance that is typi-cally around 70◦. This means that a particular attentionis required in the choice of the number of cells/tank, tankinter-distance and average gradient. In the initial part ofthe SCDTL we limit the average gradient and the num-ber of cells per tank, and gradually increase both of themwith the energy. This choice maximizes the transverse ac-ceptance minimizing the FODO period compatibly withthe space required for the allocation of external PMQs(>68.5mm, considering also other mechanical constraints)and the maximum gradient of PMQs (200T/m). The syn-chronous phase has been chosen as ϕs = −19.5◦ corre-sponding to a phase acceptance of 3|ϕs| = 58.5◦. This isa good compromise between the need to get a good phaseacceptance and to limit the RF defocusing in the tanks.The resulting 100% transverse unnormalized acceptance ofthe SCDTL structure for a bore hole radius of r = 2mmis given by

A =r2

βTwiss max= 8.7π · mm · mrad, (3)

where βTwiss max is the maximum value of the Twiss pa-rameter in the middle of the first PMQ where αTwiss = 0.

The above criteria constitute the basis of the SCDTLstructure design whose first operation is described in thisletter.

The experiment has been done in the framework ofthe TOP-IMPLART (Oncological Therapy with Protons-Intensity Modulated Proton Therapy Linear Acceleratorfor Radio-Therapy) project aiming at realizing a compacthigh-frequency 230MeV linac for proton therapy [16]. TheIMPLART segment up to 150MeV, whose basic layout isshown in fig. 3, is under construction and installation atENEA-Frascati, chosen as test site for its validation beforethe transfer to IRE-IFO-Rome hospital, that will be theclinical user.

5 6 7 8 9 10 11 120

1000

2000

3000

4000

5000

6000

energy(MeV)

Fig. 4: (Colour on-line) Computed energy spectrum at theSCDTL-1 exit.

Table 1: TOP-IMPLART SCDTL parameters.

Module Module Module Module1 2 3 4

Number of Tanks 9 7 5 5Gaps/tank 4 5 6 6Bore hole radius 2 2 2.5 3Distance between 5.5 4.5 3.5 3.5tanks/βλEnergy out, MeV 11.6 18 27 35Power, MW 1.3 1.6 2.2 2Length, m 1.109 1.082 1.353 1.105

The proton beam is delivered by a commercial injec-tor (AccSys-HITACHI Model PL-7) operating at 425MHzconsisting in a 30 keV duoplasmatron source, a 3MeVRFQ and a 7MeV DTL [17] followed by a sequence of2997.92MHz linear modules boosting the beam up to thefinal energy. In the current configuration the injector isable to deliver pulses with currents up to 160μA. Theinjector is followed by a Low-Energy Beam Transport(LEBT) consisting of four electromagnetic quadrupolesmatching the beam to the high-frequency booster. The7–35MeV section of the high-frequency linac employs fourSCDTL modules whose main parameters are reported intable 1.

The first module SCDTL-1 consists in 9 DTL tankscoupled by 8 side cavities with an array of 9 removablePMQs with a gradient around 197T/m [18] and an ef-fective length of 30mm arranged in a DOFO-like schemealigned with a precision greater than ±50μm: the firstPMQ is placed at the entrance of the structure and theother eight are in the inter-tank space. The proton beamenergy spectrum at the SCDTL-1 exit computed by beamdynamics simulations is shown in fig. 4. Up to 42% of theexit beam is effectively accelerated to the design averageenergy of 11.63MeV, while the remaining portion is a tailat low energy composed by off-phase protons that will belost in the next module.

14002-p3

C. Ronsivalle et al.

Fig. 5: (Colour on-line) SCDTL-1 module.

The computed transmission, defined as the ratio be-tween the accelerated and the injected particles in theSCDTL, is 33%, 16%, 13%, 12% at the end of eachSCDTL module. Beam dynamics simulations show thatthe beam can be transported without no further lossesup to 150MeV. The low value of transmission is causedby the longitudinal mismatch between the injector andthe high-frequency booster. It is produced by the largefrequency difference between the two structures and is in-creased by the lengthening of the injector bunch caused bythe velocity spread in the 2.5m long LEBT. The result-ing bunch length at the SCDTL-1 entrance covers threeperiods of 2997.92MHz frequency. Since the two frequen-cies (425MHz and 2997.92MHz) are not in harmonic re-lation and therefore are not locked in phase, the quitelong injector bunch allows to limit the jitter of the chargeaccelerated by the SCDTL structure. The drawback is astrong reduction of the capture efficiency. However, thisissue is not critical because the required output current isvery low for the injector current capabilities: average cur-rents of 2–5 nA required by proton therapy are achievedby 2.5–6.5μA in a 4μs long pulse at a repetition frequencyof 200Hz.

The SCDTL-1 module (fig. 5) has been mounted,aligned and tested with protons. Figure 6 shows on theleft one of the PMQs mounted in the inter-tank spaceand on the right the beam spot on a fluorescent screenat the exit. The elliptical beam shape is determined bythe last PMQ traveled that focuses the protons in verticaldirection.

The transmitted proton charge was measured using aFaraday cup placed 22.8mm from the exit of the last tankof SCDTL-1. The output design energy of 11.63MeV isachieved for an input power of 1.3MW, corresponding tothe maximum beam transmission according to simulations(fig. 7).

The energy has been measured from the proton rangein aluminum. The measurement is based on the detectionof the transmitted charge through aluminum targets of

Fig. 6: (Colour on-line) (Left) PMQ mounted in the inter-tankspace; (right) beam spot at the SCDTL-1 exit: the pink areais a 10mm diameter alumina disk.

0

20

40

60

80

100

120

0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6

Acce

lera

ted

char

ge n

orm

alize

d to

max

imum

val

ue (%

)Power (MW)

Simula�on

Measurements

Fig. 7: (Colour on-line) Charge at the SCDTL-1 exit after analuminum foil 700 μm thick vs. RF power.

different thicknesses. Protons gradually lose their energydue to interactions with atomic electrons and nuclei andstop if the absorber is thick enough. The beam averageenergy is related to the projected range defined as theabsorber thickness corresponding to a transmission of50% (R50). We used a stack of different aluminum foilsacquired from LEBOW Company [19] with nominalthickness of 500μm, 100μm, 30μm, 7μm, 4μm and acertified purity of 99.95% for the first three and 99% forthe last two. The effective thickness was measured with abalance from the weight/area ratio with a precision within1%. In the experimental setup the beam exiting fromSCDTL-1 passes through a 50 μm thick Kapton window,6mm of air, the variable thickness of aluminum target and22.5mm of air before being collected by the Faraday cup.

Beam dynamics calculations show that the PMQ ar-ray is able to transport a fraction of the injected pro-tons even in the absence of acceleration, so two measure-ment sets were taken, with and without RF power ap-plied to the structure. The results are reported in fig. 8and are compared with SRIM [20] code simulations ofthe experimental setup. In the absence of acceleratingfield, the beam is completely absorbed after 315μm of alu-minum. With acceleration, the measurements start froman aluminum thickness of 660μm in order to completelyabsorb the low-energy tail and select the exiting beamportion consisting of effectively accelerated protons. Asmany as 810μm are needed to stop the beam completely.The experimental data for a 36μA pulsed beam without

14002-p4

First acceleration of a proton beam in a side coupled drift tube linac

0

5

10

15

20

25

30

35

40

45

100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850

Tran

smi�

ed c

urr

ent

(µµA

)

Aluminium thickness (µA)

measurements RF off

measurements RF on

SRIM 11.63 MeV

SRIM 7 MeV

Fig. 8: (Colour on-line) Energy measurement: charge trans-mission curves with and without acceleration.

acceleration and a 11μA beam when 1.3MW RF poweris put into the structure are in agreement with simulatedcurves corresponding respectively to an average energy of7MeV (the injector energy) and 11.63MeV (the designenergy).

In this letter we have reported the experimental demon-stration of the acceleration capability of the 3GHz SCDTLstructure of low β protons.

An experimental test of proton acceleration by 3GHzstructures was already pursued in the past [21] but onCCL structures and with an energy jump of 11MeV, muchlower than the injection energy (62MeV), with a reallynegligible current (a few nA, pulsed), and with no trans-verse focusing lattice installed. Here the demonstration iscomplete and effective from the point of view of the furtheracceleration stages.

Further activities will concern the operation of the fol-lowing SCDTL modules up to 35MeV and on the study ofa simpler injection subsystem consisting of a lower-energysub-harmonic injector followed by a short matching sec-tion in order to enhance the transmission from the injec-tor also up to more than 90% with a better longitudinalmatching [6] to the high-frequency booster.

Moreover, the low beam losses that occur after theSCDTL second module (1–2%) and the low average en-ergy of the lost particles mainly around 7MeV suggest apossible evolution to a locally shielded accelerator withheavy shielding limited only to the treatment room. Sincethe impact of shielding is heavy on installation costs, weconsider this arrangement an improvement for a low-cost,compact, single-room proton therapy facility.

∗ ∗ ∗

This work has been granted by Regione Lazio under theagreement “TOP-IMPLART Project”.

REFERENCES

[1] Hamm R. W., Crandall K. R. and Potter J. M., Pro-ceedings of the 1991 IEEE Particle Accelerator Confer-ence, San Francisco, California, 1991, IEEE Catalog N.91CH3038-7, edited by Lizama L. and Chew J. (IEEE,New York) 1991, pp. 2583–2585.

[2] Nightingale M. P. S., Holmes A. J. T. andGrifiths N., Proceedings of the 1992 Linear AccelerationConference on LINAC, Ottawa, Ontario, Canada, 1992,edited by Hoffman C. R. (AECL Research, AcceleratorPhysics Branch, Chalk River Laboratories) 1992, ISSN0171-4546, pp. 398–401.

[3] Amaldi U., Szeless B., Vretenar M., Wilson E.,

Crandall K., Stovall J. and Weis M. S., Proceedingsof the XIX International Linear Accelerator Conference,Chicago, IL, USA, August 23–28 1998 (Argonne NationalLaboratory, Argonne) 1998, pp. 633–635.

[4] Patent No. RM95-A000564, 1995: inventors: Picardi L.,

Ronsivalle C. and Vignati A.; owner: ENEA.[5] Picardi L., Ronsivalle C. and Vignati A., Proceedings

of the 1996 European Particle Accelerator Conference,Barcelona, Spain, 1996, edited by Myers S., ISBN-13:978-0750303873, pp. 352–354.

[6] The TERA Collaboration, The RITA Networkand the design of Compact proton Accelerators, TheGreen Book, edited by Amaldi U., Grandolfo M.

and Picardi L. (INFN-LNF-Reasearch Division) 1996,Chapt. 9, ISBN 88-86409-08-7.

[7] Picardi L., Ronsivalle C. and Spataro B., Nucl.Instrum. Methods A, 425 (1999) 8.

[8] Picardi L., Ronsivalle C. and Spataro B., Nucl.Instrum. Methods B, 170 (2000) 219.

[9] Billen J. H., Krawczyk F. L., Wood R. L., Young

L. M., Proceedings of the 17th International. LinearAccelerator Conference, Tsukuba, Japan, 1994, edited byTakata K., Yamazaki Y. and Nakahara K. (KEK,Tsukuba, Japan) 1994, p. 341.

[10] Picardi L., Ronsivalle C. and Spataro B., Eur. Phys.J. AP, 20 (2002) 61.

[11] Crandall K. R. and Weiss M., TERA 94/34 ACC 20internal note (September 1994).

[12] Los Alamos Accelerator Code Group, ReferenceManual for the POISSON/SUPERFISH Group of Codes,LA-UR-87-126.

[13] Takeda H. and Billen J. H., LANL publication,LA-UR-98-4478, 2005.

[14] Wangler T. P., RF Linear Accelerators (Wiley) 2008,ISBN: 978-3-527-40680-7.

[15] Kapchinskiy I. M., LA-UR-93-4192, Los Alamos, 1993.[16] Ronsivalle C. et al., Eur. Phys. J. Plus, 126 (2011) 68.[17] Picardi L., Ronsivalle C. and Hamm R., Proceed-

ings of the VII European Particle Accelerator Confer-ence, Vienna, Austria, 2000 (European Physical Society,Geneva) 2000, p. 1675.

[18] Ronsivalle C., Picardi L., Vadrucci M. andAmbrosini F., Proceedings of the V International Par-ticle Accelerator Conference (IPAC14), Dresden, Ger-many, 15–20 June 2014, edited by Petit-Jean-Genaz

C., Arduini G., Michel P. and Schaa V. R. W., 2014,ISBN: 9783954501328, pp. 1250–1252.

[19] Lebow company website: http://www.lebowcompany.

com/.[20] Ziegler J. F., Biersack J. P. and Littmark U., The

Stopping and Range of Ions in Solids (Pergamon Press,New York) 1995, ISBN 008021603X.

[21] De Martinis C. et al., Nucl. Instrum. Methods Phys.Res. A, 681 (2012) 10.

14002-p5