Application of a PWFA to an X-ray FEL Yasmine Israeli, Jorge Vieira, Sven Reiche, Marco Pedrozzi and Patric Muggli Max Planck Institute for Physics, Instituto Superior T´ecnico, Paul Scherrer Institute Abstract There is a growing demand for X-ray Free-electron lasers (FELs) in various science fields, in particular for those with 6 short pulses, larger photon fluxes and shorter wavelengths. The level of X-ray power and the pulse energy depend on 1 0the amount of electron bunch energy. Increasing the latter will increase the power of the radiating X-rays. 2Usingnumericalsimulationsweexplorethepossibilityofusingaplasmawakefieldaccelerator(PWFA)schemetoincrease n theelectronbeamenergyofanexistingFELfacilitywithoutsignificantlyincreasingtheacceleratorlength. Inthispaper awe use parameters of the SwissFEL beam. The simulations are carried out in 2D cylindrical symmetry using the code JOSIRIS. Initial results show an energy gain of ∼2 GeV after propagation of 0.5 m in the plasma with a relative energy 4spread of ∼1%. ]Keywords: Plasma wake Free Electron Laser FEL PWFA h p - c1. Introduction K, equal to 1.2 [1]. The radiation wavelength is given by c a s. Free Electron Lasers (FELs) provide very intense and λ = λU (cid:18)1+ K2(cid:19), (1) ctightly focused X-ray beams. These X-rays can be used ph 2γ2 2 i sto map the atomic structure of materials, including ybio-molecules and nanometer scale structures. The laser where γ is the electrons energy in units of the rest energy h m c2. Therefore, for this operation the energy of the power and the radiation wavelength are determined by e p electron beam is 5.8 GeV. the energy and brightness of the electron bunches. In [ this paper, we present an FEL scheme which includes 1 The accelerator facility enables different operation a plasma wakefield accelerator (PWFA) after the linear v modes. The beam charge can typically range between 10 accelerator (linac). 0 pC and 200 pC and the bunch size (σ ,σ ), from 10 µm z r 8 to 20 µm. Beams with 200 pC are characterized by nor- 4 PWFAs can provide accelerating gradients up to 100 malized projected emittance (cid:15) = 0.43 mm.mrad and a 0GV/m, orders of magnitudes higher than gradients that N 0 350 keV energy spread. The saturation power can be esti- can be produced by conventional radio frequency linacs. . mated via P ∼ρP , with P =γmc2·I /e. 1By adding a PWFA after the linac, we may be able to sat beam beam e 0double the electron energy in a much shorter distance ρ is the Pierce parameter, defined as [2]: 6 than that of the linac and potentially generate a pulse 1with higher energy and a shorter wavelength. (cid:34) 1 I (cid:18)K[JJ]λ (cid:19)2 1 (cid:35)1/3 : ρ= e U , (2) v 16I σ 2π γ3 i A r X One of the promising future FELs, SwissFEL, is being rconstructed by Paul Scherrer Institute (PSI). The beam where I is the electron beam peak current, I is Alfven aparameters in this study have been chosen from within e A current (≈ 17 kA) and the Bessel function factor [JJ] is the range of possible operating modes of SwissFEL. equal to J (cid:0)K2/(4+2K2)(cid:1)−J (cid:0)K2/(4+2K2)(cid:1). 0 1 By increasing the electron beam energy to 13.6 GeV 2. Swiss FEL Beam Parameters with λ =30 mm and K=2.751, we can increase the sat- U uration power by a factor of 2.5. In addition, the lasing The SwissFEL baseline design seeks to provide a wave- length range from 0.1 nm to 7 nm. The undulator design, chosen for the minimum wavelength (0.1 nm), has a pe- 1UsingK=0.934·B[T]·λU[cm],B isthepeakmagneticfieldof riod length, λU, of 15 mm and an undulator parameter, theundulatorontheaxis. Preprint submitted to Nuclear Physics B January 5, 2016 15 10 10 m] m] 5 V/ 5 V/ G G ds [ 0 ds [ 0 kefiel -5 kefiel W@ r≈0 Wa Wz@ r≈0 Wa -5 Wz@ r=σDrive -10 Wr@ r=σrDrive Drrive r Drive Witness -15 -10 4.6 4.7 4.8 4.9 5 4.6 4.7 4.8 4.9 5 z[m] ×10-3 z[m] ×10-3 Figure 1. Lineouts of the longitudinal wakefield W on Figure 2. Lineouts of the longitudinal wakefield W on z z axis (–) and the transverse wakefield W at σDrive (-.) axis (–) and the transverse wakefield W at σDrive (-.) r r r r after 5 mm propagation in the plasma. Result from 2D after 5 mm propagation in the plasma. The drive and cylindrical OSIRIS simulation with n /n =2.245. The witness bunches are marked by dotted (.) and solid lines b pe simulation parameters are summarized in Table 1. respectively. Result from 2D cylindrical OSIRIS simulation with n /n =2.245. The simulation b pe parameters are summarized in Table 1. σ (cid:15) λ requirements: γ <ρ and N (cid:54) ph, where σ is the lo- γ γ 4π γ n 3.53·1016 cm−3 cal intrinsic energy spread, are better fulfilled with higher pe peaks separation 100 µm beam energies. Using a PWFA might make this energy initial energy 5.8 GeV jump possible within a compact acceleration distance. σDrive 20 µm z σDrive 10 µm r 3. The PWFA Scheme QDrive 400 pC σWitness 10 µm z A PWFA fires a driving bunch into a plasma and uses σWitness 10 µm r the resulting oscillation of plasma electrons to accelerate QWitness 170 pC a witness bunch. An effective acceleration depends on the bunch parameters (charge Q, longitudinal size σz Table 1. Parameters of the preliminary PWFA scheme. and transverse size σ ) as well as on the plasma density r n . The most important is the drive bunch longitudinal pe size, which should be in the same order of the plasma witness bunch parameters is based on the beam loading wavelength. principles [5] in order to minimize the energy-spread. In thecurrentdesigna170pCwitnessbunchwithtransverse This study focuses on the non-linear regime (or the and longitudinal lengths of 10 µm is placed 5·σDrive from z bubble regime), where the drive bunch density n exceeds the drive bunch. For simplicity, the bunches are assumed b the plasma density. In the non-linear regime the trans- to be mono-energetic with an energy of 5.8 GeV at verse wakefield, W , inside each bubble is independent of injection in the plasma. The parameters for this scenario r the propagation direction z (Figure 1) and linear with are summarized in Table 1. the radius r [3]. Therefore, the bunch experiences a constant focusing force inside the bubble, minimizing the Figure 2 presents the longitudinal and transverse wake- emittance growth through the propagation. In addition, fields after propagation of 5 mm of the drive and witness within each bubble the longitudinal wakefield, W , is bunches in the plasma . We can see that the accelerat- z independent of the radius and thus equally accelerates ing gradient G reaches about 3.5 GeV/m and the field is particles with the same longitudinal position. However, relatively constant in the witness bunch. From this ini- the accelerating field varies linearly with the longitudinal tial accelerating gradient, we can estimate the average position, leading to a correlated energy spread for an energy gain of the witness bunch after L =0.5 m to be p accelerated witness bunch. By means of simulations we G·L =1.75 GeV. P investigate the loading of the wakefield by the witness bunch in order to reduce the final energy spread. Figure 3 presents the energy distributions of the drive and the witness bunches at 0.5 m. The witness bunch dis- For the PWFA design, we use OSIRIS [4] code to per- tribution has a maximum value at 7.94 GeV and a full formnumericalsimulationswith2Dcylindricalsymmetry. width at half maximum (FWHM) of 80.32 MeV with 57% Our preliminary PWFA scheme includes a plasma with ofthebunchparticles. Themeanvalueofthedrivebunch n = 3.53·1016 cm−3 and a 400 pC drive bunch with is 4.13 GeV and the FWHM is 2.1 GeV. Accordingly, the pe σDrive=20 µm and σDrive=10 µm. The selection of the witness bunch gained 2.14 GeV while the drive bunch lost z r 2 ×10-11 ×10-11 4 4 V] V] e e M M C/3 C/3 n [ n [ bi bi y- y- erg2 erg2 n n e e er er e p1 e p1 g g ar ar h h c c 0 0 5 5.5 6 6.5 7 7.5 8 8.5 9 2.5 3 3.5 4 4.5 5 5.5 6 6.5 E [GeV] E [GeV] Figure 3. Energy distributions of the witness (left) and drive (right) bunches after propagation of 0.5 m inside the plasma. The witness bunch distribution has a maximum value at 7.94 GeV and FWHM of 80.32 MeV (includes 57% of particles of the initial beam). The drive bunch mean value is 4.13 GeV with FWHM of 2.2 GeV (52% of the particles). The bin width is 40 MeV. 1.67GeV.Therelativeenergyspreadofthewitnessbunch References is about 1%, yet does not satisfy the SwissFEL require- [1] PaulScherrerInstitute. SwissFELCDR. 2012. ments. [2] Z.HuangandK.J.Kim. Reviewofx-rayfree-electronlaserthe- ory. Physical Review Special Topics - Accelerators and Beams, 10,2007. 4. Conclusions [3] J.B.Rosenzweigetal. Accelerationandfocusingofelectronsin two-dimensionalnonlinearplasmawakefields. Phys.Rev.A.,44, FEL saturation power depends on the energy of the R6189(R),1991. electron bunches. In the case of SwissFEL, reaching an [4] R.A.Fonsecaetal. Osiris: Athree-dimensional,fullyrelativis- tic particle in cell code for modeling plasma based accelerators. energy level of 13.6 GeV will increase the power by a Lecture Notes in Computer Science,2331:342–351,2002. factor of 2.5. A PWFA scheme with SwissFEL bunch [5] M. Tzoufras et al. Beam loading in the nonlinear regime of parameters and n = 3.53 · 1016 cm−3 can reach an plasma-basedacceleration. Phys.Rev.Lett.,101,145002,2008. pe accelerated field with a multi GeV/m scale. In this first studyweshowanenergygainof∼2GeVin0.5mforthe witness bunch. Consequently, we can assume a doubling of the beam energy in very few meters. We loaded the plasma wave and minimized the energy spread of the witness bunch to 80.3 MeV. However SwissFEL operation requires an energy spread of 350 keV. An useful approach would be to investigate particles with an energy range of ±175 keV around the maximum distribution value. This would insure a suitable energy spread, while reducing the applicable charge for lasing. Wenoteherethatloadingofthelongitudinalfieldleads to a modification of the transverse field in the witness bunch as can be seen in Figure 2. Future studies will aim at reducing the witness bunch energy spread as well as at minimizing potential emittance growth due to ”loading” of the transverse wakefields. 5. Acknowledgment TheauthorswouldliketoacknowledgetheOSIRISCon- sortium, consisting of UCLA and IST (Lisbon, Portugal) for the use of OSIRIS, for providing access to the OSIRIS framework. 3