Note on the LCLS Laser Heater Review Report

Transcription

1 Note on the LCLS Laser Heater Review Report P. Emma, Z. Huang, C. Limborg, J. Schmerge, J. Wu April 15, Introduction This note compiles some initial thoughts and studies motivated by the LCLS laser heater review report, which recommends some clarification and improvement of the study on the microbunching instability and its cure in the LCLS. Please refer to the review report Section 3 for details of the recommendations. 2 Summary on further study 2.1 Origin of the input density modulation The origin of the electron density modulation is caused by the intensity ripples in the photocathode drive laser. Electron shot-noise fluctuation always exists and may play a role in initiating short-wavelength modulations in the absence of any drive laser modulation at these wavelengths (say around 15 µm near the peak gain without any heater [1]), but typically we are concerned with the drive laser modulations that have amplitudes far larger than shotnoise fluctuations. The starting point of our study is from the ultraviolet (uv) drive laser modulation. We use ASTRA/PARMELA codes to simulate the complete space charge dynamics in the rf gun and injector regions, and then use ELEGANT code to simulate the subsequent linac and bunch-compressor dynamics including LSC, CSR and linac wakefields. The start-to-end (S2E) modulation simulations show that ±8% uv laser modulation at wavelengths between 50 to 300 µm is a disaster for the LCLS without the laser heater, 1

2 while a laser heater as proposed smears out the modulation to a large degree and makes these large uv laser modulations almost tolerable [2]. Nevertheless, S2E modulation simulations are computationally intense and are not suitable to guide the design of the laser heater system. Instead, we start with a simplified initial condition assuming only residual density modulation at the end of the photoinjector and use both analytical and ELEGANT approaches to estimate the microbunching gain as well as the induced energy modulation and spread due to the instability. The laser heater parameters are then chosen to minimize the instability effect without degrading the FEL performance. One might argue that the residue density modulation amplitude at the end of the injector may be much smaller than that of the initial uv laser due to space charge oscillations in the gun region. However, ASTRA/PARMELA simulations show that these density modulations are not dissipated but mainly converted to energy modulations in the longitudinal phase space of the electron beam. Without a laser heater to increase the intrinsic energy spread, these small energy modulations can be converted back to much larger density modulations by a bunch compressor. Thus, the initial condition of the starting point may affect the gain estimation but should not change the conclusion that a laser heater is necessary to suppress the microbunching instability in the LCLS. 2.2 Theory and simulation We have ignored the transverse dependence of the LSC impedance because the transverse variation is typically small ( 10%) across the electron beam. The agreement between space charge codes such as ASTRA/PARMELA and the analytical approach (as well as ELEGANT) employing the 1-D LSC impedance is generally very good at higher energies when beam density modulation is basically frozen. For example, shown in Fig. 1 is the energy modulation amplitude at two different wavelengths along a 3-m drift space at energy 120 MeV and current 120 A, starting with a ±5% initial density modulation only. The evolution of the electron beam size along the drift is also taken into account in the 1-D LSC impedance model. Thus, we believe it is adequate to apply this impedance to study LSC effect from the end of the injector (at 135 MeV), which is the case in our study. The discrepancy between simulation and theory seems larger at much lower energy such as 6 MeV (near gun exit), as noted in the report. We try to vary the size of time steps in ASTRA simulations as recommended by the 2

3 Em kev Theory Λ 0 50Μm ASTRA Λ 0 50Μm Theory Λ 0 100Μm ASTRA Λ 0 100Μm s m Figure 1: (Color) Accumulated energy modulation amplitude E m as a function of the drift space distance s for a 120-MeV, 120-A beam with a ±5% density modulation at λ 0 = 50 µm (blue) and λ 0 = 100 µm (red). report. The preliminary study suggests that the simulation results are not very sensitive to changes in step sizes. The transverse and the longitudinal space charge dynamics may be coupled at such low energies, and then the simple 1-D coasting-beam theory for the longitudinal modulation dynamics may be limited here. Therefore we mostly rely on ASTRA/PARMELA simulations for the study of the longitudinal phase space modulation in the photoinjector. 2.3 Injector-accelerator-compressor layout It is possible to relocate the bunch compressors and further manipulate the compression factors in order to reduce the microbunching gain of the entire system. It was pointed out that a larger BC1 compression factor and a lower energy BC2 location might alleviate the gain by amplifying the intrinsic energy spread. For the LCLS, however, the possible variations are quite limited. The strong longitudinal wakefield of the micro-bunch after BC2 (L3-linac) puts a limit on the location of the BC2. A lower energy location forces a longer final linac and thus requires a larger chirp in BC2 in order that the L3-linac wakefield cancels this chirp prior to the FEL undulator. The present BC2 location is at 4.54 GeV. It is possible to move the BC2 chicane to as low as 3 GeV and still cancel the chirp. This increases the intrinsic energy spread by a factor if 1.5, but the BC2 R 56 also decreases by a similar amount and the net Landau damping is not improved. This move 3

4 also increases the rms chirped energy spread in BC2 tightening local magnet tolerances such as alignment and field quality. A larger BC1 compression factor is also possible and will increase the intrinsic energy spread in BC2. A shorter bunch between BC1 and BC2, however, also produced too little chirped energy spread and leaves a net chirp in the FEL undulator, unless the L2-linac phase is moved much closer to the zero-crossing phase, which quickly produces very inefficient acceleration. These manipulations might reduce the gain somewhat, but they cannot be used to solve the problem. (This has been studied again recently with particle tracking.) However, a simple dedicated laser-heater system will eliminate the problem completely by controlling only the one important parameter, namely the intrinsic energy spread. Therefore we believe that the best strategy is to use a laser-heater in the injector and leave the complicated matter of bunch compressor locations as a free decision to be used to minimize several other issues, such as system jitter minimization. 2.4 Experimental study of the LSC instability The GTF laser can be used to produce temporal ripple on the electron beam. Two methods for temporal pulse shaping exist at the GTF. Frequency domain pulse shaping is accomplished by hard aperature filtering in the laser pulse compressor with a highly chirped pulse. The net effect is a IR pulse with steep edges and about 10% peak-to-peak ripple over the flat part of the beam with a total of about four oscillations. The ripple in the UV would be expected to be on the order of 2-4 times greater. The slope of the edges and the ripple are dependent on the chirp of the pulse and the position of the aperature which can be experimentally varied. The UV laser pulse shape can be measured with a streak camera and the electron beam longitudinal phase space can be measured at the GTF using a linac phase scan technique with tomagraphic data analysis to recover the actual phase space distribution. Likewise a time domain pulse shaper based on a Michelson interferometer can be used to generate four independent 2 ps FHWM gaussian pulses in the UV. The delay and amplitude of each pulse is individually adjustable. This technique can produce up to 100% modulation but can produce a maximum of four pulses with existing equipment. The same laser and electron beam diagnostics are available to characterize the photon and electron beams. The experimentally measured results could then be compared with various simulation codes such as ASTRA/PARMELA. Although GTF does not have a bunch compressor 4

5 to study the amplification of input modulations, it can help us understand experimentally the origin of the input density modulation and its evolution in the photoinjector due to space charge effects. References [1] Z. Huang, M. Borland, P. Emma, J. Wu, C. Limborg, G. Stupakov, J. Welch, SLAC-PUB-10334, [2] J. Wu et al., LCLS-TN-XXXX,