Concept Study of a Compact, High Gradient X-band Travelling Wave RF Photogun with Novel Laser Coupling Scheme
CCONCEPT STUDY OF A COMPACT, HIGH GRADIENT X-BANDTRAVELLING WAVE RF PHOTOGUN WITH NOVEL LASER COUPLINGSCHEME
T. G. Lucas ∗ , P. H. A. Mutsaers, O. J. LuitenEindhoven University of Technology, The Netherlands. Abstract
This paper presents the design of a travelling wave RFphotogun operating at the European X-band frequency of11.994 GHz. The design is based on a CLIC prototypestructure milled from copper halves and uses the unique gapintroduced in the geometry to couple in the laser withoutobstructing the outgoing electron beam. With a modest peakinput power of 19 MW and a fill time of 48 ns, the gun hasthe ability to operate at very high RF pulse repetition rates incomparison to standing wave RF photoguns. For its nominalinput power, the cathode has a peak gradient of 91 MV/mand the entire gun has a gradient of 65 MV/m. The gun isillustrated to provide an 80 pC bunch at a 14.15 MeV with a0.14 % RMS energy spread and an emittance of 0.6 mm mradat a repetition rate of at least 450 Hz.
INTRODUCTION
Future compact light sources are looking towards highgradient linear accelerators (linacs) to reduce their overallfootprint as well as improve beam quality. Such linear accel-erators have been the focus of the CLIC programme, lookingat a future linear collider, which has illustrated that X-bandlinacs can operate at gradients in excess of 100 MV/m [1–3].For this reason, several compact X-ray source projects havefocused on using X-band technology in their design [4–6].Injecting into such facilities is typically addressed through astanding-wave (SW) RF photogun or using a DC gun with abunching section. Each of these are robust methods but alsohave their drawbacks. For example, SW RF photoguns havelong fill times and consequently are limited in their repeti-tion rate. On the other hand, DC guns can pulse continously,or even produce DC beam, but their low cathode fields, incomparison to RF photoguns, limit the bunch charge pos-sible for a given emittance. A more recent concept is thedevelopment of the travelling wave RF photogun which hasbecome more feasible through the development of high gra-dient linacs [7–9]. In [9], such a gun could be developedfrom an existing accelerating structure design through twomodifications:1. Adjust the length of the first cells to allow low energycapture, and2. Modify the input coupler to house a cathode.This paper details the design of the first x-band travellingwave RF photogun developed from a prototype high gradient ∗ Email: [email protected]
Figure 1: Cross-section of the cell geometry used for theCLIC-G Open structure with its "race-track" design and gapto house the joint [10].Figure 2: One half of the completed 24 cell structure of theCLIC-G Open structure [10].accelerating structure designed for the CLIC programme,the CLIC-G Open [10]. The gun will use the unique geom-etry of this prototype to couple in the laser through a gapincorporated into the cells’ design. The paper will beginwith a brief review of the RF design of the original accelerat-ing structure followed by a description of the two significantmodifications made to the CLIC-G Open to transform itinto a travelling wave RF photogun. Following will be adescription of the unique laser coupling scheme illustratingthe benefits and consequences of the method. The RF fieldsfrom the gun design will be incorporated into a particle trac-ing software where a main solenoid and bucking coil will beintroduced to combat space-charge effects in the first sectionof the gun. Finally, the paper will conclude by discussingthe beam quality produced by the gun.
RF DESIGN
The RF design started with the CLIC-G Open structurewhich has been well-documented in [10] therefore only anbrief description will be discussed below. The design started a r X i v : . [ phy s i c s . acc - ph ] S e p igure 3: One symmetric half of the RF design (vacuum) of the travelling wave RF photogun with the notable featureslabelled.with the CLIC-G prototype, the most widely tested CLICprototype, which operates at a 120 degrees phase advanceand the European X-band frequency of 11.994 GHz. Withthe aim of joining the cells through the milling of two halves,a gap was introduced into the design to house the brazingjoint (Figure 1).The gap’s width was crucial to cutting off the fundamen-tal mode frequency preventing the fields from propagatinginto the gap and reaching the joint. This allows a variety ofjoining techniques to be used. The cells were designed witha "race-track" geometry where the cross-sectional profileincludes straight regions of half-length Fx and Fy (Figure 1).These cells were ultimately arranged into a 24 cell structure,completed with a waveguide coupler at each end. With thisas the initial geometry, the design of the travelling wave RFphotogun could begin (Figure 2).The first task was to reduce the length of the first three reg-ular cells, noting that the first cell in the structure is theinput coupler cell where the cathode will later be housed.An analysis performed by X. Stragier illustrated that thelengths 6.832mm, 7.332 mm and 7.832 mm for the first,second and third cells, respectively, were an appropriatereduction in the cells’ lengths for capture of the low betaelectrons [11]. The lengths of these cells were adjusted inthe CLIC-G Open and then tuned back to the correct oper-ational frequency. Following, the input coupler was to bemodified to house a cathode. To do so, the input coupler wascompletely removed and replaced with a compact couplerdesign used in CLIC G* Structures [12]. Additionally, a1 mm gap was added on the plane normal to the RF input,as per the original CLIC-G Open coupler design, and thebeampipe was replaced with a flat plane where the cathodewould be placed. This RF design does not look at how thecathode would be installed although this has been addressedin [9]. The input coupler’s dimensions were then optimiseduntil the reflection was below -45 dB. Various input couplercell lengths were tested using General Particle Tracer (GPT).A length of 5 mm produced the lowest emittance for an input power of 19 MW. The input coupler was not optimised toreduce surface fields. The electric field along the z-axis ofthe final design is illustrated in Figure 4. The electric fieldat the cathode (length = 0 m) is observed to be 91 MV/m.Along with this, the first three cells are observed to have areduced axial field strength. The RF frequency spectrum ofthe accelerator is plotted in Figure 5, with the operationalfrequency marked with a red line, and Table 1 lists the im-portant RF parameters. The fill time is similar to that in theCLIC-G Open as expected. Given that the fill time is only 48ns (Table 1), the design opens up the possibility to operateat very high repetition rates well above those possible ina current state-of-the-art SW RF photoguns. The CLIC-GOpen structure was tested to an average power of 445 W -where it operated at a 44.5MW with a 200 ns RF pulse at 50Hz - without any sign that a limitation in the average powerwas being reached [10]. For the nominal fill time, this wouldallow a repetition rate of 490 Hz at an input power of 19 MWwhich would need to be reduced to 450 Hz to be a harmonicof the 50 Hz mains. It is expected that the repetition rate willbe able to exceed this. PSI’s TW gun has been proposed tooperate with an average power of 2.64 kW [9] which wouldequal a repetition rate of 2.9 kHz. Further thermal analy-sis will be performed to understand the average power, andtherefore repetition rate, limitations of the gun. POWER SYSTEM
Whether this repetition rates could be achieved is depen-dent on the RF power system. The current design is opti-mised for an input RF power of 19 MW, giving a gradient of65 MV/m. The design could be powered by a Canon E371136 MW klystron feeding a pulse compressor operating at again of 3.2 [13]. Such klystrons are designed to operate at400 Hz for a 5 µ s pulse and proposed to operate at 1 kHzwith a 2 µ s RF pulse. A simulation of the pulse compressorused in CERN’s Xbox 3, displayed in Figure 7, illustratesthat this gain is achievable for a 2 µ s input RF pulse. Further-more, CPI and Canon have research programmes to developigure 4: The axial longitudinal field distribution for the electric field from the accelerating structure and the magnetic fieldfrom the main solenoid and bucking coil.Figure 5: S11 parameter of the RF photogun. Parameter Value UnitLength 216 mmRegular Cells 24Phase Advance 120 degsFrequency 11.994 GHzAttenuation -2.23 dBPower 19 MWGradient 65 MV/mPeak Cathode field 91 MV/mFill time 48 nsRepetition Rate 450 Hz
Table 1: Summary of the travelling wave gun properties.high repetition rate klystrons operating at 10 MW. Usingthese it may become possible to generate the power throughthe weaving of two klystron pulses as is also at CERN’s thirdX-band Test Stand (Figure 6) [3]. Figure 6: Xbox 3 test stand at CERN which uses 400 HzCanon E37113 6 MW klystron.Figure 7: A simulation of the Xbox 3 pulse compressoroutputting a 19 MW pulse from a 6 MW, 2 µ s input pulse. Bunch Charge [pC] 40 80Laser spot max. radius [um] 282 400norm. intrinsic emittance [mm mrad/mm] 0.55 0.55Laser size at 250mm [um] 291 403Laser pulse length RMS [fs] 100 100Bunch length RMS [fs] 510 550norm. emittance [mm mrad] 0.32 0.6Beam Energy [MeV] 14.14 14.14Beam Energy Spread [%] 0.09 0.14
Table 2: Summary of the travelling wave gun’s laser andelectron beam properties.igure 8: One symmetric half of the RF design (vacuum) of the travelling wave RF photogun with the notable featureslabelled.Figure 9: The longitudinal distribution of a 80 pC bunch
LASER COUPLING
A possible criticism of this style of RF photogun is theacceptance angle of the accelerator for a laser. If one wereto modify a typical CLIC-G design, which has no gap, into aTW RF photogun an acceptance angle of 0.6 degrees wouldbe possible if the laser were to be coupled through the irises.Using such a method may make it difficult to couple in thelaser without obstructing outgoing electrons, given the com-pactness of the X-band structure’s geometry. The additionof the gap has opened up this angle quite significantly toan angle up to 3.6 degrees. This angle could be increasedgreatly by placing a window on the side of the gun and cou-pling through the side of the structure, again using the gap.A limitation of the coupling is the width of the gap. Thebeam must be carefully thread through the 1 mm gap withminimal interception of the wall. This ultimately limits thelargest spot size one can achieve. Table 2 illustrates the laserradius on the cathode and the radius of the laser at 250 mm,which is approximately the length from the cathode to theend of the structure. Calculating the Rayleigh Range andthe beam envelope, it is found that the greatest bunch chargeachievable is approximately 120 pC. This could be increasedthrough an increase to the charge density on the cathode butmay lead to an increased emittance. Figure 10: The phase space distribution of a 80 pC bunch
MAIN SOLENOID AND BUCKING COIL
After the RF design was finalised, it was incorporated intoGPT and a solenoid was added to counter the defocusingeffect at the end of the structure, to keep the beam from col-liding with the cell irises and to maintain the low emittancein the structure. Along with the main solenoid, a buckingcoil was added to reduce the field at the cathode to preventan increase in the intrinsic emittance caused by local mag-netic fields. GPT was used to optimise the length and fieldstrength of the main solenoid as well as the bucking coil.The conditions for the optimisation were that the beam’sdivergence was less than 100 µ rad, the field at the cathodewas less than 20 µ T, and the emittance was less than thatwithout the solenoid. The axial magnetic field distributionis illustrated in Figure 4 where the B field at the cathodeis observed to be approximately zero (6 mu T) and the fo-cusing field of the main solenoid covers the first half of thestructure.
BEAM DYNAMICS
With the optimised solenoids from GPT and the axialelectric field distribution from the CST model, the entire gunwas simulated from start-to-end in GPT. The off-axis electricigure 11: The evolution of the emittance over the gun.fields were extrapolated from the imported on-axis fielddistribution to match that of a TM mode structure. These donot include the quadrupole component expected from a dual-fed input coupler which will be investigate in future work.The cathode was simulated with an intrinsic emittance of0.55 mm mrad/mm as has been used in [8] and the laser spotsize on the cathode was set to 400 µ m for an 80 pC bunch. Fordifferent charges the laser spot size was scaled according to r ∝ √ Q to keep a constant charge density. GPT’s optimiserwas used to find the appropriate RF power (between 15 and20 MW), RF phase and laser pulse length (between 50 fsand 1 ps) to minimise the emittance and energy spread. Forvalues between 50 and 500 fs the pulse length was found tohave little effect on the beam quality. The spatial distributionand phase-space distributions of the bunch which resultedfrom these optimisations are illustrated in Figures 9 and 10,respectively. Figure 11 illustrates the emittance over thelength of the gun. Table 2 summarises the results of theoutput electron beam quality for bunch charges of 40 and 80pC. CONCLUSION AND FUTURE STEPS
A design for a travelling wave X-band photogun based ona CLIC prototype made from milled halves was studied. Theunique gap in the geometry was used to couple in the laser tothe cathode without obstructing the output electrons. With ahigh field solenoid and bucking coil, the start-to-end beam-line simulations were performed in GPT illustrating that thegun could produce 80 pC electron bunches at 14.15 MeVwith a low energy spread of 0.14 % energy spread at an emit-tance of 0.6 mm mrad. The short fill time of 48 ns allowsthe gun to operate at very high repetition rates. There arevarious future steps for this gun. To finalise this design ofthis gun, the surface fields on the input coupler will be opti-mised and an investigation into the effects of the quadrupolecomponent at the cathode will be studied. Separate to this,a high power version of this gun will be developed in theaim of producing a lower emittance gun which can competewith the current stare-of-the-art standing wave guns.
ACKNOWLEDGEMENTS
The authors would sincerely like to thank Xavier Stragier,Marco van der Sluis and Harry Doorn from Eindhoven Uni-versity of Technology, and Walter Wuensch and Nuria Cata- lan Lasheras from CERN for their input and discussionsabout the project.