Report on the design of the electron beamline of FLASHForward

C. Behrens for the FLASHForward TeamDeutsches Elektronen-Synchrotron DESY, Notkestr. 85, 22607 Hamburg, Germany

(Dated: April 17, 2015)

Here we report on the electron beamline extension of the free-electron laser (FEL) in Hamburg(FLASH) at DESY towards beam-driven wakefield acceleration (PWFA) experiments. The approvedand well advanced project, officially referred to as FLASHForward, will accommodate a third elec-tron beam extraction line next to the original FLASH facility and in parallel to the FLASH-IIextension. FLASHForward will share the same accelerator tunnel and infrastructure as FLASH-IIbut is planned to operate independent of the FEL user facilities FLASH and FLASH-II. In thisnote, we report on the electron beam extraction strategy, taking into account boundary conditionsand beam dynamics constraints, and discuss the design of the entire electron beam transport up tothe entrance of the plasma cell, where PWFA experiments will finally be carried out.

INTRODUCTION

Here we report on the electron beamline extensionof the free-electron laser (FEL) in Hamburg (FLASH)at DESY towards beam-driven wakefield acceleration(PWFA) experiments. The approved and well advancedproject, officially referred to as FLASHForward, will ac-commodate a third electron beam extraction line next tothe original FLASH facility and in parallel to the FLASH-II extension. FLASHForward will share the same ac-celerator tunnel and infrastructure as FLASH-II but isplanned to operate independent of the FEL user facili-ties FLASH and FLASH-II. In this note, we report on theelectron beam extraction strategy, taking into accountboundary conditions and beam dynamics constraints,and discuss the design of the entire electron beam trans-port up to the entrance of the plasma cell, where PWFAexperiments will finally be carried out.

In order to provide some background for the beam-line design strategy, which is discussed in the next sec-tion, we summarize some of the relevant capabilities ofthe accelerator shared by all three beamlines (FLASH,FLASH-II, and FLASHForward). The accelerator re-lies on superconducting radio-frequency (r.f. ) technol-ogy and is thus able to generate long pulse trains (macropulses) with hundreds of microseconds, accommodatingsingle bunches (micro pulses) with a bunch spacing ofsingle-digit microseconds (MHz micro repetition rate).The macro pulses can be delivered at a repetition rateof maximal 10 Hz. Each of the bunches can contain acharge of about 20 - 1000 pC with a bunch current upto several kiloamperes (see below for more details onthe beam parameters). The bunches are generated ina laser-driven r.f. gun based on photo emission from acathode. The electron beams are accelerated and lon-gitudinally compressed in two stages with a final mainaccelerator downstream of the second bunch compres-sor (BC). Downstream of the final accelerator the elec-tron beams get distributed along the different beamlines,where one of the three, FLASHForward, is dedicated forexploration of PWFA concepts. In order to maintain the

high brightness of the electron beams and provide uniquecapabilities in PWFA at FLASHForward, special care hasto be taken in the design of the electron beam extractionand transport as is discussed in the following.

BEAMLINE DESIGN STRATEGY

The FLASH facility, as has been mentioned above, isoperated with a superconducting r.f. accelerator and canthus provide multiple bunches in a bunch train. Thisprovides the ultimate possibility to split the bunch trainin parts and individually distribute the sub-trains alongdifferent beamlines. FLASH-II uses a fast kicker magnetin conjunction with a septum to extract parts of the upto 800µs long pulse train with up to 800 bunches (at 1µsspacing). A sketch of the electron beam switching area

beam direction

final acceleratorFLASH

FLASH-II

FLASHForward

switch

switch

switch

FLASH-II

FLASHForward

radiation safety wall

FIG. 1: Sketch of the electron beam switching area at theFLASH facility. The electron beam pulse train leaves thefinal accelerator and can be partially distributed either alongFLASH, FLASH-II, or FLASHForward. The fast switchingdipole magnets (kicker + septum magnet for FLASH-II andpulsed bends for FLASHForward) are labeled as ”switch”.The blue frame shows the FLASHForward switching area withmore details (colored boxes indicate individual magnets).

2

at the FLASH facility is depicted in Fig. 1. The electronbunches that get transported to FLASH-II can be splitagain and further distributed to FLASHForward. In con-trast to the FLASH-II switching, FLASHForward cannotmake use of a kicker + septum combination as there isnot enough distance between both components (in beamoptics terminology: there is not sufficient phase advancein between) in order to gain enough offset in the septummagnet. Hence, FLASHForward must adopt an extrac-tion system made of pulsed dipole magnets, referred toas pulsed bends in following. To significantly reduce thecosts of the extraction system, we decided for a relativelysimple but cheap half-sine pulse generator for the switch-ing fields instead of an expensive and not-well elaboratedflat-top pulse generator. This half-sine pulse restrictsthe FLASHForward beamline to 1 - 3 bunches from thepulse train available in FLASH-2 before the extraction.This, in turn, also reduces the costs and efforts of themachine protection and radiation safety as the averagebeam power of a few bunches is rather low.

The design of the beamline underlies several technicalboundary conditions where some of them are in turn re-lated to beam dynamics constraints. First of all, the vac-uum chamber inside the pulsed bends must allow trans-mission of the pulsed magnetic fields with 111.5µs risetime, which can only be achieved by ceramic chambers ofvery thin thickness. At the same time, the thin ceramicchamber must be rigid enough over the entire length ofthe magnets (∼ 300mm) to resist the atmospheric pres-sure. The design of the magnets with only a few wind-ing turns takes into account the impedance match withrespect the half-sine pulse (at 2.24 kHz) generator. Fig-ure 2 presents a technical drawing (cross section) for onethe pulsed bends (top) and a sketch of the pulsed bendsystem including the ceramic chambers (bottom).

In order to allow independent tunnel access consid-ering safety reasons, a radiation protection wall down-stream of the extraction bends (see Fig 1) encapsulatesthe FLASH-II/FLASHForward tunnel from FLASH. The∼ 2m thick wall prevents installation of focusing magnetsand the space between extraction and wall is thus criti-cal in order to allow proper beam transport. This is ofparticular importance when taking account the requiredoptics conditions, which is discussed below.

To accommodate focusing elements upstream the ra-diation safety wall (two quadrupoles and one sextupolemagnet), the bending angle of the extraction magnetsmust be sufficiently large to gain enough transverse off-set until the beam line crosses the wall. On the otherhand, we are dealing with extremely short bunches (tensto hundreds of femtoseconds) with high peak currents(a few kiloamperes), and coherent synchrotron radiation(CSR) effects can thus significantly degrade the beambrightness and must be considered in the design. In gen-eral, large bending angles lead to stronger CSR effects, sotaking into account the space issue for focusing elements,

pulsed bend

FLASH-II

FLASHForward

FIG. 2: Pulsed bend system for the electron beam extractionat FLASHForward: The upper plot shows a technical drawing(cross section along beam path) for one of the pulsed bends.The bends have a magnetic length of 294 mm and a nomi-nal bending angle of 4 deg. The lower plot depicts the arrayof two subsequent pulsed bends with their ceramic vacuumchamber. The total angle between FLASH-II and FLASH-Forward directly downstream of the extraction is 8 deg.

we decided for a total bending angle of 8 deg (see Fig 2),which turns out to be a good compromise. The relativemagnetic field jitter of the pulsed bend is expected tobe about 1× 10−4, which, in combination with designedbeam optics, fulfills the pointing stability at the plasmatarget. In the following, we describe the required beamoptics conditions and report on our beamline design withits lattice and different optics solutions.

ELECTRON BEAMLINE AND OPTICS

The FLASHForward beamline uses the tunnel ofFLASH-II and is therefore in parallel to FLASH-II ata distance of 4 m. After the extraction when the beam-lines are parallel again, the horizontal dispersion, whichis generated by the extraction bends, should be closed,i.e., zero again. This achromatic beam translation sys-tem also generates longitudinal dispersion (also referredto as R56), which can result in bunch compression in thepresence of energy chirp. The compression in first or-der can be expressed as C = (1 − hR56)−1, where h isthe linear energy chirp. The operation of FLASHFor-ward is designed to be as independent as possible of thebeam parameters used in FLASH and FLASH-2. Thisand a generally increased flexibility for dedicated beamtimes lead to the demand for a tunable R56. Besideshight peak current of several kiloamperes (> 2.5 kA), the

3

1

beam waist compatible with FLASH-2

overall small betas

final focus

FIG. 3: Transverse position (top) of the magnets and beta functions (bottom) along the beamline in FLASHForward startingwith the extraction from FLASH to FLASH-II (labeled as FLASH 1 and FLASH 2 in the figure). A horizontal beam waist inthe pulsed bends mitigates emittance growth due to CSR effects, and the overall small beta functions keep chromatic effectsbelow a tolerable level. The final focus section with its large beta functions upstream the plasma cell is clearly recognizable.

reverse bend

+-4mm

FIG. 4: Longitudinal and horizontal transverse dispersion (la-beled as R56 and R16, respectively) along the FLASHForwardbeamline starting upstream of the pulsed bends and ending atthe plasma cell. The position of the reverse bend is indicated,and the approximate range of variable R56 is given.

PWFA experiments also require small transverse beamsizes (< 7µm) in both planes, and a good pointing sta-bility of about 10µm spatially and 0.5 mrad in angle. Thepeak current can be controlled by the chirp and R56, and

the transverse beam size can be changed by optics, wherethe transverse emittance limits the achievable beam sizefor a given optics condition. In general, the small emit-tance of FLASH - a mandatory condition for lasing in anFEL - should be preserved during the beam transport inFLASHForward. Besides CSR effects, which have beenmentioned above, also chromatic effects can degrade thebeam quality and have to be taken into account in thebeamline design. To minimize the magnet field jitter ofthe pulsed bends as a source for pointing jitter at theplasma cell, a sufficient phase advance has been foundfor the optics solution. In this context, the phase ad-vance basically describes how an angular kick gets trans-lated to an offset along the beam transport. Figure 3presents the longitudinal and horizontal dispersion alongthe FLASHForward design beamline starting upstreamof the pulsed bends and ending at the plasma cell. Itfulfills all discussed requirements and also those not par-ticularly mentioned here, e.g., closed angular dispersion,closed second-order transverse dispersion, or a horizon-tal beam waist in the bending magnets. Not shown inFig. 3 are three sextupole magnets to control the chro-matic and second-order effects as well as several steeringmagnets along the beamline for orbit correction.

In order to control the compression via the longi-tudinal dispersion R56 (see the compression factor Cabove), we inserted an additional bend (so-called ”re-verse bend”) within the dispersive region of the extrac-tion beamline. The longitudinal dispersion is defined viaR56 =

∫ s

0D(s′)/ρ(s′)ds′, hence we can tune dispersion

by the dispersion D ≡ R16 at the position of the reversebend and by its bending radius. The latter is fixed how-ever, so effective tuning will be done by changing theoptics upstream of the reverse bend. The bend is re-ferred to as ”reverse” as its bending angle has a differentsign compared to the dispersion, and its R56 contribu-

4

max(Bmag) = 1.03

max(Bmag) = 1.59Horizontal plane Vertical plane

FIG. 5: Transverse phase space in special coordinates ({u,v},see main text for the definition) and the mismatch parameter(Bmag) as a metric for chromatic effects. The simulationsassumes pessimistic ± 1.5% relative energy deviation.

tion is thus negative (see also Fig. 4). That way we canachieve a large range of about ± 4mm total longitudinaldispersion along the FLASHForward beamline, includingclosed longitudinal dispersion (isochronous).

A small focus at the plasma cell entrance is essentialfor performing effective PWFA experiments, hence pre-serving the emittance and beam matching (re-matchingupstream the plasma cell is possible though) along theFLASHForward beamline is required. Besides collectivebeam effect such as CSR or space charge forces, also chro-matic effects can degrade the beam quality. Off-energyelectrons simply see different focusing fields, which canchange the transverse phase space ({x,x’} and {y,y’} forthe horizontal and vertical plane, respectively) signifi-cantly. A metric for chromatic effects along a beamtransport line can be the mismatch parameter definedas Bmag = 1/2(β̃γ−2α̃α+βγ̃) with the Courant-Snyder(CS) parameters α, β, and γ describing the phase spaceellipse. The CS parameters with and without tilde rep-resent the actual and the design case, respectively. De-viations arise from chromatic effects and describe a mis-match (mismatched beams have Bmag > 1). Beside avalue describing the mismatch, there is also a graphi-cal representation. By transforming the transverse phasespace coordinates (in the following only shown for thehorizontal plane; likewise for the vertical case) via thedesign CS parameters downstream of the extraction with(

uv

)=

(1/√β 0

α/√β√β

)·(xx′

),

a perfectly matched beam that is not subject to chro-matic effects would appear as a perfect circle with unityradius. Any beam mismatch would result in a phasespace ellipse with semi-axes 6= 1. Figure 5 shows a calcu-lation for both the mismatch parameter Bmag and itsgraphical representation in the {u,v}-phase space. Arather pessimistic energy deviation of ± 1.5% has beenassumed in the calculations but still the chromatic ef-fects are in a reasonable limit. The vertical plane shows

FIG. 6: Matching and final focus section upstream of theplasma cell. The last, dispersion-closing bends of the extrac-tion are indicated as triangles, and the section ends at theentrance of the plasma cell. The optical transition radiation(OTR) label indicates the imaging station for beam size mea-surements needed for matching. The steering magnets andother diagnostics are not shown for the sake of simplicity.

a larger impact compared to the horizontal plane, butthe latter is the critical plane anyway, as CSR effects actin this bending plane of the extraction bends.

The last section downstream of the dispersive extrac-tion beam and thus directly upstream of the plasma cellis the ”matching and final focus” section. A sketch ofthis section with its magnets is depicted in Fig. 6. Af-ter the last, dispersion-closing bends of the extraction,we planned with a matching section consisting of fourquadrupoles plus one quadrupole for doing a scan in or-der to measure the emittance and the actual optics. Thebeam size measurements for this purpose will be carriedout using an imaging station (labeled as OTR in Fig. 6),which accommodates different screens. Once the beamis matched, it gets focused down, by using the final fo-cusing quadrupoles, to beta functions at the plasma cellentrance of 25 and 35 mm in the horizontal and verticalplane, respectively. The achieved steering resolution atthe plasma cell entrance amounts to better than 1µm inposition and to better than 10µrad in angle.

SUMMARY AND CONCLUSIONS

In summary, we presented the design consideration ofthe FLASHForward electron beamline for PWFA experi-ments. The optics and beam dynamics calculations meetall requirements. Chromatic effects are not significant,and CSR effects are the limiting factor. Nevertheless,the beamline allows mitigation of the degrading CSR ef-fects by careful control of the compression (via energychirp and tunable longitudinal dispersion). The techni-cal drawings are almost completed, and the magnets andbeams diagnostics are either available or ordered.

The first components of the FLASHForward beam-line will be installed during a coming shutdown in May2015, and further shutdowns are scheduled for 2016. Thepost-plasma beamline design, which includes electronbeam capturing, special beam diagnostics, and subse-quent beam transport to undulator magnets, is also ingood progress and will be finalized soon.