design and implementation of a flexible beamline for fs ...€¦ · design and implementation of a...

Nuclear Instruments and Methods in Physics Research A 691 (2012) 113–122

Contents lists available at SciVerse ScienceDirect

Nuclear Instruments and Methods inPhysics Research A

0168-90

http://d

n Corr

Lausann

fax: þ4E-m

journal homepage: www.elsevier.com/locate/nima

Design and implementation of a flexible beamline for fs electrondiffraction experiments

Giulia Fulvia Mancini a,n, Barbara Mansart a, Saverio Pagano a, Bas van der Geer b, Marieke de Loos b,Fabrizio Carbone a

a Laboratory for Ultrafast Microscopy and Electron Scattering, ICMP École Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerlandb Pulsar Physics Eindhoven, The Netherlands

a r t i c l e i n f o

Article history:

Received 29 May 2012

Received in revised form

26 June 2012

Accepted 26 June 2012Available online 11 July 2012

Keywords:

Ultrafast electron diffraction

Femtosecond

Radiofrequency

02/$ - see front matter & 2012 Elsevier B.V. A

x.doi.org/10.1016/j.nima.2012.06.057

esponding author. Present address: École

e, Station 6, CH-1015 Lausanne, Switzer

1 216935875.

ail address: [email protected] (G.F. Manc

a b s t r a c t

Ultrafast Electron Diffraction (UED) has been widely used to investigate the structural dynamics of

molecules and materials. Femtosecond (fs) electron bunches are used to obtain diffraction images of a

specimen upon photo-excitation by a temporally delayed light pulse. The high cross-section of

electrons makes it a very flexible tool for the study of light elements, monolayers and surfaces; at

the same time, electrons can travel down to few nanometers (nm) and structural information from the

bulk can also be retrieved. In this article, we discuss the design and implementation of a flexible

beamline for fs electron diffraction experiments in transmission or reflection geometry. By the use of a

radiofrequency (RF) compression cavity synchronized to our laser system, in combination with a set of

electron optics, we demonstrate that we can control the beam properties in terms of charge per pulse,

transverse spot-size on the sample and temporal duration of the bunches. The characterization of the

beam is performed via a light-electrons cross-correlation experiment and we demonstrate an overall

temporal resolution around 300 fs for bunches containing up to 105 electrons at a repetition rate of

20 kHz.

& 2012 Elsevier B.V. All rights reserved.

1. Introduction

The dual particle-wave nature of electrons gives rise tophenomena like Bragg scattering from materials [1]. Electrondiffraction is commonly used to investigate the surface qualityof thin films [2] or bulk materials more in general [3]. Comparedto similar techniques like X-ray diffraction (XRD) [4], electronsare widely used for in situ characterization of surfaces and thinfilms owing to their higher cross-section for interaction withmatter [2,5] and simpler implementation in table-top set-ups.

Currently, a lot of attention has been devoted to the study ofultrafast phenomena in molecules [6,7] and solids [8–16], anddiffraction techniques have been developed into time-resolvedtools for the investigation of the structural properties of matteracross a phase transition or as a result of an excitation induced bylight pulses [5,17–20]. Both fs X-ray diffraction and absorption[17,18] and fs electron diffraction [19,20] have proven effective indetermining the ionic motions accompanying phase transforma-tions in materials [19,21]. The main limitations for these tools

ll rights reserved.

Polytechnique Fédérale de

land. Tel.: þ41 216933608;

ini).

have been the flux for X-rays based sources [22,23], and space-charge effects limiting time resolution for electron based ones[24]. The advent of the next generation X-ray sources, such as freeelectron lasers [25,26], promises to push the performances ofX-ray based techniques all the way to being able to observedelicate materials structures within one single pulse as short asfew fs [27–29]. At the same time, RF technology has been used toimprove electron diffraction performances, allowing to obtain fewtens to few hundreds of fs electron bunches containing up to 106

electrons at 100 kV acceleration voltage [24], and up to 108 at3.5 MV and 5.4 MV [30,31], also enough for single-shot experi-ments on materials [24,26,30].

In this work, we describe the design and implementation of atable-top electron diffractometer operating at 30 kV accelerationvoltage capable of fs-resolved experiments both in transmissionand reflection geometry. The RF technology, comprising an RFcavity, a temperature control unit and a phase-locked loop circuit,has been developed and demonstrated by Luiten and co-workers[24,32–34] and is currently commercially available from AccTec[35]. The aim of these experiments is to reach a temporalresolution in the 100-fs time-scale in a set-up that can measureboth monolayers [36] and bulk systems. Control of the sampletemperature is obtained via a liquid helium flow-cryostat operat-ing in the range between 1.8 K and 300 K. The whole beamline is

www.elsevier.com/locate/nimawww.elsevier.com/locate/nimadx.doi.org/10.1016/j.nima.2012.06.057dx.doi.org/10.1016/j.nima.2012.06.057dx.doi.org/10.1016/j.nima.2012.06.057mailto:[email protected]/10.1016/j.nima.2012.06.057

G.F. Mancini et al. / Nuclear Instruments and Methods in Physics Research A 691 (2012) 113–122114

designed to reach a vacuum around 10�10 mbar, and standardsurface preparation tools such as a crystal cleaver and a tool formechanical scraping are mounted on a removable feedthrough.

2. Beamline design

In this section, we illustrate some basic principles of electronoptics that were used to design the beamline. We will present abottom-up approach consisting mainly of three steps:

(i)
The beam propagation is first described within a one-to-oneanalogy with geometrical optics, neglecting all space-chargeeffects and the finite dimensions of the electron source. Inthis analogy, electron optics like solenoids and the RF cavitycan be considered as thin lenses with an adjustable focaldistance, with current being the adjusting parameter. Thisallows to coarsely estimate the position and strength of theelectronic lenses needed to obtain the wanted diffractionimage on the detector.
(ii)
The finite size and brilliance (emittance in electron beamphysics) of the source are taken into account, and an estimateis given of both the temporal duration and transverse spot-size that can be expected in a geometrical arrangement suchas the one described in (i). Space-charge effects areneglected.
(iii)
The full simulation of the beam propagation is performedwith the General Particle Tracer (GPT) code [37] that con-siders: the real field profile and emittance of the electronsource, the thickness of the electronic lenses (solenoids andRF cavity), the exact field profile in the RF cavity, and space-charge effects during propagation. These simulations areused to refine the design parameters and are later verifiedby an experiment.
The full simulation of the beamline will be discussed in Section 4,after its experimental implementation has been described inSection 3. The characterization of the beamline performancesand its comparison to the simulations will be shown in Section 5.Finally, conclusions will be discussed in Section 6.

2.1. Ray tracing design

As a first approximation, we neglect the finite size of theelectron source and all space-charge effects. In these conditions,the evolution of the bunch in space (transverse coordinate) andtime (longitudinal coordinate), can be separated and consideredindependent from one another. The aim of the design is to obtaina beam that is as focused as possible in both space and time onthe sample. For this purpose a set of electronic lenses are used. Inparticular, two solenoids are employed to first collimate the beamtransversally and then focus it on the sample. In first approxima-tion, solenoids can be seen as thin lenses with a current-controlled focal distance. These electronic lenses are made of anumber Nw of copper wire windings, each carrying a certaincurrent Iw. The magnetic field produced along the axis of thesolenoid is given by the linear superposition of the fields of eachwinding, which simply corresponds to the field given by the Biot–Savart law for a current spire:

Bzð0,zÞ ¼m0I2

R2

ðR2þz2Þ3=2ð1Þ

where z is the propagation coordinate, R the average radius of thesolenoid and I¼NwIw, with Nw the number of windings and Iwthe current through a single wire. The focusing distance of asolenoid is related to the applied field Bzð0,zÞ by the following

equation:

f sol ¼8meU

e2R þ1�1 B

2z ð0,zÞ dz

ð2Þ

where me is the mass of the electron, U the electron energy (in ourcase 30 keV) and e is the electron charge in Coulomb.

To treat the longitudinal beam properties, a RF cavity isemployed. The RF cavity is made out of two copper half-cellsbrazed at 750 1C for vacuum compatibility and it is oscillating onthe TM010 mode with a resonant frequency of 3 GHz. The stabilityof the resonant frequency value sent to the cavity is controlled bya temperature control system with a precision of 4 mK. Bysynchronizing the phase offset of the RF field to the photoemittedlaser pulses [34], the first electrons of the bunch entering thecavity (fastest ones) feel a decelerating electric field while thelatest to enter the cavity (slowest) feel an accelerating field. Thenet result is a chirp imparted to the bunches which compensatestheir tendency to broaden during propagation. When the phasebetween the RF field and the electron pulse is the one describedabove, the corresponding magnetic field distribution in the cavityinduces a slight spatial defocus of the beam. When operating on ap shifted phase instead, spatial focusing and temporal defocusingcan be obtained. Further details on the principles behind thiscomponent are given in Ref. [38]. For our purposes, we considerthe cavity as a tunable temporal and spatial lens with a temporalfocal distance which depends on the strength of the RF electricfield in the cavity following the formula:

f L,cav ¼�mev3

eE0dco cos f0ð3Þ

where me is the mass of the electron, v the electron velocity –taken as a third of the velocity of light c for non-relativisticelectrons with 30 keV energy – and e is the electron charge inCoulomb. Eq. (3) shows how the focusing properties depend onthe peculiar parameters of the cavity, like the amplitude of thefield E0 applied, its phase f0, its frequency o ðrad=secÞ and thewidth dc (5 mm) of the electrons-RF field interaction space.

Neglecting space-charge effects, from a simple geometricaloptics consideration one should expect a tighter focus for ashorter focal distance. This suggests that for obtaining the short-est possible bunches in time on the sample, the distance betweenthe cavity and the sample must be minimized. Moreover, limitingthe distance between the cavity and the sample reduces space-charge induced beam degradation and beam expansion, making iteasier to use samples where a much smaller spot-size is desired[39]. In our set-up, mechanical constraints fix this distance at11 cm, which will then be our longitudinal focal distance for theRF cavity. For non-relativistic velocities, the transverse and long-itudinal focal length of the cavity are related in a very simple anddirect way:

f T,cav ¼�2f L,cav ð4Þ

The negative sign for f T,cav in Eq. (4) evidences the transversedefocusing effect of the RF cavity when operating on the phasecapable of compensating the temporal chirp of the pulses.

The overall layout of the beamline is depicted in Fig. 1, wherethe spatial and temporal problems are separated into two insets(top and bottom respectively). To have a spatially concentratedbeam on the sample, the photoemitted electrons are collimatedby a first solenoid (convex lens) and later focused by a secondsolenoid (convex lens). The RF cavity is treated as a defocusingelement (concave lens) with a negative focal distance of �22 cm(given by the above considerations and Eq. (4)). For the bestcollimation of the photoemitted electron beam, the first solenoidshould have the highest possible numerical aperture, i.e. theshortest focal distance. For this purpose, it is placed as close as

Fig. 1. Schematic of the beamline based on the analogy between charged particle optics and geometrical optics. The distances of every component with respect to thephotocathode are shown. (A) Spatial profile: the two solenoids act on the transverse profile collimating and focusing the electron bunch, behaving like convex lenses. The

RF cavity defocuses the electrons along this direction, due to the magnetic field distribution at the temporal compression phase. (B) Temporal profile: the RF cavity acts on

the longitudinal compression of the ellipsoidal bunch, behaving like a convex lens along this direction.

G.F. Mancini et al. / Nuclear Instruments and Methods in Physics Research A 691 (2012) 113–122 115

possible to the electron source; this distance is also mechanicallyconstrained to be 8.4 cm. Using Eqs. (1) and (2), one can obtainthe current needed to achieve a focal length of 8.4 cm in the firstsolenoid. In particular, we find that a total current of 1225 A isneeded with an average radius of the windings of 6 cm. In reality,our solenoid is composed of 50 windings each carrying 24.5 A,with an average radius around 6 cm and a total thickness of about4 cm. The collimated electron beam is then directed to thecombination of lenses formed by the second solenoid and theRF cavity. The distance between the cavity and the sample isfixed, as explained above, to 11 cm, while the cavity behaves as adefocusing lens with a focal distance of �22 cm, as depicted inFig. 1A. Since the aim is to obtain a transverse spot-size as smallas possible on the sample, the second solenoid should have theshortest possible focal distance to both compensate the effect ofthe transverse defocusing of the cavity and compress the beamspatially. In this case also mechanical constraints impose that thesecond solenoid is placed before the RF cavity at approximately8.6 cm from its center. Considering that an ideally collimatedbeam hits the set of lenses composed by the second solenoid andthe RF cavity, geometrical optics dictates that:

1=Zfinal ¼ 1=f T,cav�1=Z2 ð5Þ

where Zfinal is the distance from the second solenoid to theformed image, f T,cav ¼�22 cm is the focal length of the RF cavity,and Z2 ¼ dsc�f sol2 with dsc being the distance between the secondsolenoid and the RF cavity and f sol2 being the focal length of thesecond solenoid. By imposing the mechanical conditions thatdsc ¼ 8:6 cm, resulting in a total distance Zfinal of 19.6 cm, onecan find the focal length of the second solenoid needed tocompensate for the defocusing effect of the RF cavity; for satisfy-ing Eq. (5), Z2 ¼�10:5 cm and f sol2 ¼ 19 cm. To obtain such a focaldistance the second solenoid was designed to carry a total currentof 780 A, distributed over 400 windings with an average radius of3.8 cm (current per winding¼1.95 A). The overall evolution of thetransverse beamsize is pictorially shown in Fig. 1A.

The temporal evolution of the bunches in the beamline isdisplayed in Fig. 1B. The RF cavity is the only element acting on

the compression of the ellipsoidal bunches in time and simplyprovides a temporally focused beam at the sample position; thefocal distance of the RF cavity is controlled via the strength of theRF field injected in it. In particular, for obtaining a focal distancearound 11 cm, the field strength in the cavity should be around0.54 MV/m, according to Eq. (3).

2.2. Propagation of a beam with finite size and emittance

In this section, we take into account the finite size of theelectron source and its brilliance, called emittance in electronoptics. An estimate of the final spot-size in both time and spacecan be provided based on these considerations and neglecting thespace-charge interaction. For an optical beam, brilliance or bright-ness can be defined as the energy stored in the beam per unit timeand solid angle. This is a quantity that is conserved during lightpropagation in a passive medium, and gives information on thefocusability of an optical beam [40]. Analogously, one can define atransverse or a longitudinal emittance for an electron beam,which depends respectively on its spatial and temporal dimen-sions as well as on its momentum spread in space (angulardivergence) and time (temporal divergence). The emittance ofan electron bunch is also a conserved quantity (considering idealelectron optics), and like brilliance for optics it gives a measure ofthe focusability in both space and time of the beam itself. Thebeam energy spread is a less critical parameter for diffractionexperiments, an account of its influence on the propagation canbe found in Ref. [41]. The rigorous definition and the statisticalderivation of the emittance for an electron beam composed of ahigh number of particles are beyond the scope of this article andcan be found in Refs. [38,40]. For our purposes, we use theconcept and the simplified expressions that apply to our case.Like in the previous section, the transverse and longitudinalevolution of the beam during propagation are considered sepa-rately for simplicity.

The transverse emittance of a monochromatic beam of inde-pendent particles at a waist (neglecting space-charge effects) can


be defined simply as:

ex ¼ dx � sy ð6Þ

Where dx is the transverse dimension of the beam and sy itsangular divergence. The transverse emittance has the units of mand measures how tightly a beam can be focused. When a beamwith lateral dimensions di impinges on a solenoid, the latterbends the trajectories of the electrons and focuses the beam to aspot. The angular divergence imparted by the solenoid to thebeam is simply sy ¼ di=f sol, with f sol being the focal length of thesolenoid. Combining the equations for ex and sy, one can estimatethe spot-size dx produced by a solenoid of a given focal length f sol.The emittance properties of our photoelectron gun have beenmeasured, and a value for ex around 10�6 m was estimated. InFig. 2A, the transverse dimension of the spot as a function of thebeam emittance and diameter at the entrance of a solenoid with afocal length around 19 cm, like the one we use to focus our beamon the sample, is plotted. It is visible that for an emittance around10�6 m and a beamsize of approximately 1 mm one shouldexpect a focused spot between 100 and 150 mm.

Fig. 2. (A) Electron pulses final transverse dimension dx (mm) as a function of boththe transverse emittance ex and the beam lateral dimension di at the entrance of asolenoid of 19 cm focal length. Regions with different colors have the same value

of dx for a set of different input values of ex and di. (B) Final electron pulsesduration tf (ps) as a function of the longitudinal emittance ez and initial pulseduration ti at the entrance of a RF cavity with a focal length of 11 cm. Regions withdifferent colors have the same value of tf for a set of different input values of ezand ti . (For interpretation of the references to color in this figure legend, thereader is referred to the web version of this article.)

As far as the temporal evolution of the beam is concerned, thelongitudinal emittance can be defined as:

ez ¼sz � spzme � c

ð7Þ

Where sz is the beam dimension in the z direction (propagationdirection), which corresponds to the Root-Mean-Square (RMS)bunch length, and spz is the temporal momentum spread of thebunch. The longitudinal emittance measures the temporal com-pressibility of an electron pulse. Typical values for this quantityrange between 10�9 m and 10�8 m. The final pulse duration isrelated to the longitudinal emittance and the RF cavity longitudinalfocal distance f L,cav via the formula:

tf ¼f L,cav � ezti � v2

ð8Þ

where ti is the duration of the pulses at the entrance of the RFcavity and v is the velocity of the electrons, in our case v� c=3. Inanalogy to the transverse evolution of the beam, and to whatgeometrical optics considerations suggest, the temporal spot-sizedepends on the focal distance of the focusing lens, in this case theRF cavity. As expected, the shorter the focal distance, the shorterthe achievable pulse duration. This is an important point whichconstraints the design of the final set-up because it suggests thatthe RF cavity should be as close as possible to the sample forhaving the shortest possible pulses and therefore the best timeresolution. In Fig. 2B, the pulse duration obtained after a RF cavitywith a focal distance of 11 cm for a range of longitudinal emit-tances between 10�8 m and 10�7 m and input pulses durationsbetween 1 ps and 10 ps is shown. With a longitudinal emittancearound 5�10�8 m and an initial pulse at the entrance of the cavitylonger than 5 ps, we expect to obtain electron pulses compressedbelow 100 fs.

To summarize, keeping into account the geometries describedin the previous section, and the realistic performances of ourelectron source in terms of beamsize and emittance, we estimatethat we should be able to obtain o100 fs pulses of electronsfocused into approximately 100 mm. All these considerationsneglect the effect of space-charge, which will be treated in detailsin the next sections.

3. Experimental implementation

Thanks to the simple ingredients described above, the positionof the elements composing the beamline can be coarsely decidedand the experiment can be physically implemented. The refine-ment of the parameters governing the beam propagation will beperformed simulating the whole experiment under realistic con-ditions via the GPT code in Section 4. The overall layout of ourbeamline is presented in Fig. 3. In panel A the design of theexperiment is shown, while its implementation is depicted inpanel B. Our UED apparatus is powered by a KMLabs WyvernTi:sapphire amplified laser delivering 50 fs pulses at a wavelengthof 800 nm and with an energy of 700 mJ per pulse at a repetitionrate of 20 kHz. The shot-to-shot noise on the energy per pulse inthis system is below 0.2%. The amplified fs laser beam is dividedinto two paths with a beam-splitter. One beam is used tophotoexcite the sample at 800 nm. The other beam is fre-quency-tripled by third harmonic generation in nonlinear crys-tals, and the 266 nm beam thus obtained is used to generate theprobing electron pulses inside a DC gun. The DC gun is made of ahigh voltage cathode (30 kV) on top of which a silver-coatedsapphire window is glued to guarantee electrical connection.The cathode is isolated from the anode, which is grounded tothe UHV shield of the DC gun. Electrons are photoemitted by

Fig. 3. Schematics of the UED setup compared to its performances calculated withGPT. (A) Design and (B) implementation of the UED setup consisting of a DC

photogun, two solenoidal magnetic lenses, two steering coils, and a RF cavity. The

beamline is designed to allow surface properties control by cleaving or scraping

the sample, and the use of a liquid helium flow-cryostat allows to explore a

temperature range from 1.8 K up to 300 K. The diffraction pattern is captured by a

single-electron-counting CCD camera. (C) Longitudinal evolution of the electron

beam (blue line) together with its diameter profile (pink line) is displayed as a

function of the distance from the cathode. (For interpretation of the references to

color in this figure legend, the reader is referred to the web version of this article.)


back-illuminating the photocathode with the 266 nm opticalpulses, and are accelerated to 30 keV and directed through aPt–Ir pinhole of 150 mm diameter. The electron probe pulses aredirected to the sample with the system of magnetic lensesdescribed above. This includes the two solenoids used respec-tively to collimate and focus the beam outcoming the DC gun andtwo steering plates for the orthogonal deflection of the beam; theRF cavity for the temporal compression of the electron probe

pulses is placed right before the experimental chamber and itsaverage distance from the sample is of 11 cm (depending on themanipulator position). All distances and positions of the electro-nic lenses are shown in Fig. 3A and B. The sample chamber hasbeen designed to have openings for the incoming probe pulses,the pump optical access, the single-electron-counting Charge-Coupled Device (CCD) detector exit and a cryogenic system thatallows to explore a range of temperatures from 1.8 K up to 300 K.Additional openings were designed for preparation tools, internalvision and pressure gauges. A high-precision goniometer allowsthe motion of the sample along three axis (x,y,z) and one angle(the rotation around the cryostat axis). The angular rotation of thesample around its surface normal is used in Reflection High-Energy Electron Diffraction (RHEED) experiments and is obtainedvia a piezo-electric support, which is in thermal contact with thecryostat. An optical delay line allows to change the time of arrivalon the sample between pump and probe pulses. The electronpulses scatter from the sample and a diffraction image is detectedby a CCD capable of a one-to-one conversion efficiency betweenelectrons and output counts. The automation of the experimentalsetup was performed by controlling the instruments viacomputer.

4. Propagation of the electron beam in the presence of space-charge effects

The complete simulation of the beamline performances wascarried out via the commercial software GPT [37]. This softwareuses the realistic field maps calculated ab-initio starting fromMaxwell equations for the electron gun, the RF cavity and thesolenoids, and takes into account the effect of mutual Coulombrepulsion between electrons during propagation, i.e. space-chargeeffects. The simulations were done with the Particle-In-Cellmethod of GPT version 2.8 [42]. Stochastic effects for photoemis-sion were assumed to be in the few percent range and thereforeneglected. In the simulations, a gun voltage of 30 kV over 5 mmwas used, resulting in an extraction field of 6 MV/m. The accel-eration section ends abruptly with a mesh and this is modeled byusing a GPT element with a uniform Ez field. Any additionalemittance growth due to the mesh was not taken into account.The acceleration section is followed directly by a pipe and apinhole with a diameter of 150 mm. The simulated initial bunchcharge was 100 fC with a Gaussian transverse distribution withstandard deviation s¼ 250 mm and a cut-off at one s point. Withthe pinhole considered, this results in approximately 10% ofelectrons being delivered to the sample. The initial bunch lengthis 50 fs FWHM with a Gaussian temporal profile. The initialthermal emittance was set at approximately 0:45 mm per mmradius. The evolution of the beam in the whole beamline underthe above-mentioned experimental conditions is shown for boththe transverse and longitudinal dimension in Fig. 3C. In thissimulation we use the highest charge we can obtain from ourDC gun (6�105 electrons per pulse). In this plot, the evolution ofthe beam diameter, pink line, shows the initial expansion afteremission from the gun aperture, position z¼0; the first collimat-ing solenoid is found at z¼8.4 cm from the cathode and compen-sates for the initial beam divergence. An almost 0.5 mm beamreaches the second focusing solenoid at z ¼25.4 cm and entersthe RF cavity at z¼34 cm receiving a slight transverse defocusing.The transverse beam dimension between z¼44 cm and z¼47 cm,where the sample will be positioned, is around 160 mm, in goodagreement with previous estimates. In the same figure, the long-itudinal evolution of the beam is displayed (blue line). Afteremission from the cathode, the electron bunch spreads in time,and it reaches the RF cavity with a duration longer than 10 ps, at


this high charge. The longitudinal focus is at the same propaga-tion distance as the transverse focus for afield in the cavityaround 0.54 MV/m, and the final bunch length is approximately200 fs.

In summary, realistic simulations of our beamline suggest thepossibility to perform UED experiments with approximately 105

electrons per pulse confined in a spot of 160 mm size and 200 fsduration. These numbers are in good agreement with the crudeestimates we made based on simple geometrical optics argu-ments and show that space-charge effects can dominate propaga-tion only at extreme values of charge, as will be discussed infurther details later on.

5. Experimental characterization of the beamline

To characterize the performances of our beamline, a cross-correlation experiment between electron and photon pulses wascarried out [43,44]. A shadow image of a copper grid was createdby focusing the electrons before the same grid and intense laserpulses were used to photoemit electrons via a multi-photonprocess from the copper. The charge created by light pulsesproduces a lensing effect on the electrons which is visible as adistortion of the grid image. The experiment is based on thesimple schematic layout depicted in Fig. 4. The pump and theprobe beams are directed to the sample chamber, and an opticaldelay line allows to change the timing between them. When theoverlap between the pump and the probe beam at t¼0 takesplace, charge is created on the copper grid which deflects theelectrons of the probe. This causes the appearance of a hole in theimage of the grid on the CCD, as shown in Fig. 4B.

The transverse evolution of the beam is not critical and can beeasily monitored by observing the beamsize on the CCD uponvarying the current in the collimating and focusing solenoidsrespectively. The obtained spot-sizes on the CCD agree with thepredictions described in the previous sections. For this reason, wefocus our attention on the longitudinal evolution of the beam,which is governed by the RF cavity and determines the final time-resolution of the set-up. As discussed above, the RF cavity can betreated as a temporal thin lens whose focal distance and behavior(temporally focusing or defocusing) are governed by the injectedRF field strength and phase. These dependences are verified at

Fig. 4. Schematic layout of the experiment. (A) The electron pulses are collimatedand focused before the grid to create a shadow image of the copper grid on the

CCD. (B) Coulomb scattering with an electron cloud that is photoemitted from the

TEM grid, taking place at t¼0. The electron image is distorted in correspondenceof the laser beam position and a hole appears in the image of the grid recorded by

the CCD.

two different acceleration voltages of the electron bunches, 25and 30 kV.

5.1. Propagation of 25 kV electrons

For slower electrons, a lower field is expected to be needed inthe cavity to compress the electron pulses. In Fig. 5 a set ofexperiments at 25 keV energy is reported. The RF cavity to sampledistance was 15 cm. Our experiments were carried out at 6�104

electrons per pulse. The estimate was performed integrating thecounts from our single-electron-counting CCD. At such a lowcharge level this estimate is not a critical parameter as will bediscussed below. In panel A the temporal evolution of the widthof the hole produced by the laser pulses on the grid image (seeFig. 4B) is displayed at the optimal compression field (0.30 MV/mfor 25 kV electrons). An estimate of the field needed to focustemporally 25 keV electron pulses at 15 cm from the RF cavity canbe obtained from Eq. (3) for different values of the RF phase. Inour set-up, the phase of the RF field can be controlled via a voltage(0–10 V) with a total dynamical range of approximately 7001,therefore two values of the phase leading to the bunches optimaltemporal compression are expected to be found. For technicalreasons, the voltage control of the RF phase is not linear; a phasecan be associated to a voltage in Fig. 5B where the cavity’stemporal and spatial focusing properties are displayed as afunction of the phase, as discussed below. In Fig. 5B, thetransverse beam dimension observed on the CCD without currentin the focusing solenoid (i.e. for a spatially collimated beamentering in the RF cavity, blue symbols) and the rise time of theelectron-laser cross-correlation (red symbols) are displayed. Asshown by the minus sign in Eq. (4), the RF cavity behaves as atemporally focusing lens for a given phase, at which its spatialeffect is to defocus the beam. On the contrary, when the phase issuch that the RF cavity focuses spatially, it will defocus the beamtemporally. For this reason, in Fig. 5B the minimum rise-time ofthe cross-correlation effect, corresponding to the maximumcompression of the pulses, happens in coincidence with the phasethat gives the largest transverse spot-size. This experimentdemonstrates in a simple way the principle by which the RFcompression cavity works. In Fig. 5C, the effect of the fieldstrength in the cavity is also shown. For the optimal field of0.30 MV/m the rise-time of the cross-correlation is the shortestwhile it becomes longer for both a higher and a lower fieldbecause the temporal focus of the RF cavity does not correspondanymore to the position of the sample. At lower accelerationvoltages, 25 kV in this case, compression is not optimal because alower field is needed resulting in a longer focal length, see Eq. (3),and space-charge effect and emittance-related effects are ofcourse enhanced. In these conditions the rise-time of the widthof the laser-induced hole in the image is approximately 700 fs,which gives us a rough estimate of the overall temporal resolu-tion of the experiment at this energy. The experiments at 25 kVwere used simply to test the trends of the bunch length as afunction of the different parameters like the phase and the fieldstrength. The choice of slower electrons was done to enhance thesensitivity of the propagation to these parameters and show thetrends more clearly. The precise characterization of the beamlineperformance was obtained via experiments at 30 kV accelerationvoltage.

5.2. Propagation of 30 kV electrons

In these experiments, the copper grid was positioned around47.5 cm from the cathode. The optimal focal length of the RFcavity for compression was therefore 13.5 cm, which could beobtained using a field around 0.54 MV/m. The charge in the

Fig. 5. Experimental results for electron pulses with 25 keV energy. (A) Fit of the width of the laser-induced hole for different phases at the optimized RF field valueE0 ¼ 0:30 MV=m. (B) The transverse beam dimension observed on the CCD with no current in the focusing solenoid (blue symbols) for different phases of the RF field in thecavity is shown together with the phase dependence of the rise-time of the electron–laser cross-correlation (red symbols). Due to the characteristics of the RF cavity, which

defocuses transversally the electron pulses when their temporal chirp is compensated, the minimum rise-time of the cross-correlation effect corresponds to the phase that

gives the largest beamsize on the CCD. (C) Effect of the RF field strength in the cavity: the evolution of the laser-induced hole in the image is displayed for three different

field values E0¼0.20, 0.30 and 0.36 MV/m, at a constant phase value. (For interpretation of the references to color in this figure legend, the reader is referred to the webversion of this article.)

Fig. 6. Experimental results for electron pulses with 30 keV energy. (A) Normalized cross-correlation of every delay image with respect to the first image at negative timefor the phases corresponding to the values expressed in voltage of 2.5, 4 and 6 V, at the optimized RF field strength value of 0.54 MV/m. (B) Results from a complementary

data analysis method, based on the fit of the area of the dip created by the pump on the copper grid and imaged on the CCD. (C) The effect of the RF field strength in the

cavity is displayed at a constant phase value.


bunches was around 6�104; this was estimated via our single-electron-counting detector for which we know that 1 electronproduces around 100 counts. Our beam produces a spot in theCCD of approximately 10�10 pixels (100 pixels) whose inte-grated counts per single pulse was 6�106. In Fig. 6, the results ofthe electrons-light cross-correlation experiment at 30 kV areshown. The effect of the laser pulses on the image of the coppergrid is analyzed in two different ways. The cross-correlation of theimages at the different time delays t and t0 is evaluated according

to the formula:

gðt0; tÞ ¼ Sx,yCx,yðtÞCx,yðt0Þffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Sx,yCx,yðtÞ2Sx,yCx,yðt0Þ2q ð9Þ

where the contrast Cx,yðtÞ ¼ ½Ix,yðtÞ�IðtÞ�=IðtÞ, with Ix,yðtÞ the inten-sity of the pixel at the position (x,y) at time t, and IðtÞ is the meanof Ix,yðtÞ. By comparing images at different time delays t withrespect to a fixed-image frame at t0 before t¼0, through the


evaluation of the correlation coefficient gðt0; tÞ, the cross-correla-tion method allows to retrieve information on the imagedynamics as a function of time. The results of this analysis areshown in Fig. 6A for three different phases (for 30 kV electronsthe phase corresponding to 4 V on the controller is the onecorresponding to maximum temporal compression). Alterna-tively, the dip caused in the image counts by the laser pulse isfitted to a Gaussian profile whose area (intensity � width) isdisplayed for different values of the phase and field strength inFig. 6B and C. These transients evidence the optimal phase andfield as simulated via GPT, and are fitted to a step functionconvoluted to a Gaussian, the latter simulating the effect oftime-resolution:

F ¼ acffiffiffiffip2

rerfc

b�xffiffiffi2p

cð10Þ

with a and b respectively the height and the position of the stepfunction, and c the smearing of the Gaussian.

These fits yield a temporal resolution of 580 fs for the cross-correlation and 360 fs for the overall area. The temporal evolution ofthe beam in these experiments was simulated in GPT, using the verysame experimental parameters in terms of distances and currents inthe beamline components. In Fig. 7, the bunch length as a function ofthe field strength, phase, and charge in the pulses are shown. Inpanels A to C, the bunches duration dependence on the RF fieldstrength is displayed within 10 cm around the sample position (5before and 5 after) for three different values of the charge in eachpulse. As the number of electrons in every shot is increased, the focaldepth of the RF cavity decreases and the final pulse durationincreases. For less than 105 electrons per pulse and at the optimalfield strength of 0.54 MV/m, which gives the optimal compression atthe sample position (z¼47.5 cm), the pulses duration is found to beshorter than 400 fs over around 6 cm of propagation. The field andphase dependence for less than 105 electrons in every bunch, in

Fig. 7. GPT calculations of the bunch length as a function of the field strength and the pdifferent values of the field E0, at a constant phase value. (D,E) The phase dependence

Fig. 7(panels A, B and D, E), is also found to be relatively weak. Theseresults suggest that for stroboscopic experiments where integrationof the signal over several pulses is carried out, using a smaller chargeper bunch results in a dramatically less critical set-up in which fewhundreds of fs pulses can be obtained even in non ideal conditions. Athigher values of the charge, the focal depth of the cavity becomesvery short and the compression critical. In these conditions, errors orfluctuations in the field strength or/and phase quickly result in acompromised temporal resolution. To better evidence this effect, thepulses duration as a function of the charge per pulse at a particular RFfield strength (0.54 MV/m) is displayed in Fig. 8. These data show thatto have the optimal compression for the focal distance required byour set-up (10–15 cm), the number of electrons in every pulse shouldnot exceed 105. Above this charge, while it is still possible tocompress pulses down to 2–300 fs, the control of the relevantparameters and its required precision becomes impractical. In gen-eral, at high charge values, a stronger RF field would be required tocompensate space-charge effects, which would mean a shorter focaldistance and therefore placing the RF cavity closer to the sample,which is also impractical for construction reasons.

6. Discussion and conclusions

In a UED set-up, the pulses duration on the sample is not theonly limiting factor for the overall performance. Several othereffects can spoil the temporal resolution of the beamline, whichcan be defined as:

t:resðUEDÞ ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiDt2jitterþDt

2durationþDt

2GVM

qð11Þ

In this formula, the statistic average between the arrival timejitter of the pulses, the pulse length and the Group VelocityMismatch (GVM) is taken.

hase for three different values of charge in the pulses. (A–C) Bunch length for three

at three different charges per pulse is displayed at a constant field strength value.

Fig. 8. Electron RMS pulse duration as a function of the number of electrons in asingle pulse at the RF field value of 0.54 MV/m.


In our set-up, the RF cavity is controlled via a phase-locked-loop circuit which receives a driving signal from the laseroscillator and controls the central frequency of a local VoltageControlled Oscillator (VCO) to match the repetition rate of thelaser [34]. When the circuit is locked and the local electronicoscillator is synchronized to the laser, the error signal, taken asthe residual difference between the repetition rate of the laserand the controlled VCO output, can be monitored. The synchro-nizer has been tested as a stand-alone unit and shows anelectronic jitter of 10 fs in the frequency range up to 30 kHz; inRef. [24], in a complete experiment the synchronization of the RFcavity to the laser has been shown to have a pulse-to-pulse jitteraround 80 fs. For this reason, we use Dtjitter ¼ 80 fs in Eq. (11).

The bunches duration in the focus of the cavity are estimated viaGPT simulation, and for the highest charge we considered, 6�105

electrons per pulse, we obtain 288 fs. Therefore, Dtduration ¼ 288 fsin Eq. (11).

The GVM is given by the difference in the time of arrivalbetween the photon pulses and the electron probe in the differentparts of the sample. In our geometry, the photoexcited area had adiameter of 125 mm; with an excitation laser fluence of 48.9 mJ/cm2 the resulting GVM is 242 fs [50] (DtGVM ¼ 242 fs).

According to these estimates, which are done considering thehighest charge we can extract from our cathode, and a non-optimized geometry to limit GVM effects, we obtain an overalltime resolution around 380 fs, in good agreement with the lowestvalue obtained experimentally of 360 fs.

The design of these experiments has been carried out with theidea to build a flexible beamline in which the propagationparameters, spot-size, pulses duration and charge could be variedto match the experimental requirements. The sample environ-ment is controlled in a separate vacuum chamber and gives thestandard conditions used in more common photoemission orsurface science experiments, in terms of vacuum and temperaturecontrol. In this article, we describe the basic concepts needed forthe design of such a system, but we stress that these ideas aregeneral and apply to different geometries and experimentalarrangements. The aim of this work is to provide the reader withsimple recipes to design a pulsed electron diffraction beamlineand estimate its performances. In our particular case, few hun-dreds of fs time-resolution using up to 105 electrons per pulsewas demonstrated. Compared to existing similar set-ups[43,45–50], our beamline takes advantage of the repetition rateof the laser, 20 kHz, and the high charge per bunch which we can

use to achieve a high signal-to-noise ratio on the diffractionimages. In general, 106 electrons need to be counted to acquirea diffraction pattern [24,26]; however, in a time-resolved experi-ments changes smaller than 1% in the Bragg peaks position orintensity must be distinguished, requiring a higher signal-to-noise ratio which can be obtained only by averaging over severalimages. As an example, the diffraction images recorded in Refs.[9,10,16,20] contained as many as 3.6�108 electrons. Thesecounts were obtained using approximately 2000 electrons perpulse at 1 kHz repetition rate over three minutes integration time.Using the RF technology to store more electrons in every shot, upto 105, and a higher repetition rate laser, 20 kHz in our case, it ispossible to reduce significantly the acquisition time and achieve avery high signal-to-noise ratio in the diffraction patterns. Wewould also like to point out that the RF cavity only requires aslightly stronger spatial lens before itself to compensate for itsdefocusing effect, and can be implemented easily on a pre-existing beamline. As far as the limits in the performances ofsuch a system, our estimates suggest that the best overall time-resolution achievable may be below 100 fs. The timing jitter is aninherent parameter of the RF set-up and 80 fs seems to be its limitin terms of performances. The pulse duration instead can betuned by varying the sample position and the charge in the beam;compression down to 60 fs can be easily achieved with 104

electrons per pulse. The ultimate limiting factor for the timeresolution is the GVM which can be particularly severe inreflection geometry [50]. A precise estimate of how effective isthe approach of tilting of the optical wavefront (described in Ref.[50]) is lacking; the shortest time-resolution observed with thisarrangement has been 200 fs in Ref. [19]. In transmission geo-metry instead, reducing the angle between electrons and photonsand controlling the spot-size on the sample one can easily limitthe GVM below 100 fs.

In conclusion, we reported the implementation of a beamlinefor UED capable of both reflection and transmission geometryexperiments, using 30 kV electrons and a state of the art RFcompression cavity. A temporal resolution ranging between 60 fsand few hundreds of fs can be achieved in these experimentsdepending on the beamline parameters and the charge employedin the electron bunches. The design criteria and principles havebeen illustrated and a series of recipes has been provided whichcan be used both to design new ultrafast electron systems and toupgrade pre-existing ones.

Acknowledgments

The authors acknowledge O.J. Luiten, P. Musumeci, M.J. van derWiel, P. Baum and M. Chergui for useful discussions. This projectis founded by the Swiss National Science Foundation (SNSF)through the grant No. PP00P2–128269/1.

References

[1] A. Ichimiya, P.I. Cohen, Reflection High-Energy Electron Diffraction, Cam-bridge University Press, Cambridge, 2004.

[2] A.H. Zewail, Annual Review of Physical Chemistry 57 (2006) 65.[3] D.-S. Yang, N. Gedik, A.H. Zewail, Journal of Physical Chemistry C 111 (2007)

4889.[4] W.H. Bragg, W.L. Bragg, X-rays and Crystal Structure, G. Bell and Sons,

London, 1915.[5] M. Chergui, A.H. Zewail, ChemPhysChem 10 (2009) 28.[6] A. Paarmann, T. Hayashi, S. Mukamel, R.J.D. Miller, Journal of Chemical

Physics 130 (2009) 204110.[7] E.F. Aziz, M.H. Rittmann-Frank, K.M. Lange, S. Bonhommeau, M. Chergui,

Nature Chemistry 2 (2010) 853.[8] B. Mansart, M.J.G. Cottet, T.J. Penfold, S.B. Dugdale, R. Tediosi, M. Chergui,

F. Carbone, Proceedings of National Academy of Science USA 109 (2012)5603.


[9] F. Carbone, D.S. Yang, E. Giannini, A.H. Zewail, Proceedings of NationalAcademy of Science USA 105 (2008) 20161.

[10] F. Carbone, Chemical Physics Letters 496 (2010) 291.[11] C. Giannetti, F. Cilento, S. Dal Conte, G. Coslovich, G. Ferrini, H. Molegraaf,

M. Raichle, R. Liang, H. Eisaki, M. Greven, A. Damascelli, D. van der Marel,F. Parmigiani, Nature Communications 2 (2011) 353.

[12] N. Gedik, M. Langner, J. Orenstein, S. Ono, Y. Abe, Y. Ando, Physical ReviewLetters 95 (2005) 117005.

[13] H. Schäfer, V.V. Kabanov, M. Beyer, K. Biljakovic, J. Demsar, Physical ReviewLetters 105 (2010) 066402.

[14] M. Beck, M. Klammer, S. Lang, P. Leiderer, V.V. Kabanov, G.N. Gol’tsman,J. Demsar, Physical Review Letters 107 (2011) 177007.

[15] D. Fausti, R.I. Tobey, N. Dean, S. Kaiser, A. Dienst, M.C. Hoffmann, S. Pyon,T. Takayama, H. Takagi, A. Cavalleri, Science 331 (2011) 189.

[16] F. Carbone, O.-H. Kwon, A.H. Zewail, Science 325 (2009) 181.[17] A. Cavalleri, S. Wall, C. Simpson, E. Statz, D.W. Ward, K.A. Nelson, M. Rini,

R.W. Schoenlein, Nature 442 (2006) 664.[18] V.-T. Pham, T.J. Penfold, R.M. van der Veen, F. Lima, A. El Nahhas, S.L. Johnson,

P. Beaud, R. Abela, C. Bressler, I. Tavernelli, C.J. Milne, M. Chergui, Journal ofAmerican Chemical Society 133 (2011) 12740.

[19] P. Baum, D.-S. Yang, A.H. Zewail, Science 318 (2007) 788.[20] F. Carbone, P. Baum, P. Rudolf, A.H. Zewail, Physical Review Letters 100

(2008) 035501.[21] A. Cavalleri, Th. Dekorsy, H.H.W. Chong, J.C. Kieffer, R.W. Schoenlein, Physical

Review B 70 (2004) 161102. (R).[22] Ch. Bressler, C. Milne, V.-T. Pham, A. ElNahhas, R.M. van der Veen,

W. Gawelda, S. Johnson, P. Beaud, D. Grolimund, M. Kaiser, C.N. Borca,G. Ingold, R. Abela, M. Chergui, Science 323 (2009) 489.

[23] S.L. Johnson, P. Beaud, E. Vorobeva, C.J. Milne, É.D. Murray, S. Fahy, G. Ingold,Physical Review Letters 102 (2009) 175503.

[24] T. van Oudheusden, P.L.E.M. Pasmans, S.B. van der Geer, M.J. de Loos, M.J. vander Wiel, O.J. Luiten, Physical Review Letters 105 (2010) 264801.

[25] B. McNeil, Nature Photonics 3 (2009) 375.[26] F. Carbone, P. Musumeci, O.J. Luiten, C. Hebert, Chemical Physics 392 (2012) 1.[27] H.N. Chapman, A. Barty, M.J. Bogan, S. Boutet, M. Frank, S.P. Hau-Riege,

S. Marchesini, B.W. Woods, S. Bajt, W.H. Benner, R.A. London, E. Plönjes,M. Kuhlmann, R. Treusch, S. Düsterer, T. Tschentscher, J.R. Schneider,E. Spiller, T. Möller, C. Bostedt, M. Hoener, D.A. Shapiro, K.O. Hodgson,D. van der Spoel, F. Burmeister, M. Bergh, C. Caleman, G. Huldt,M.M. Seibert, F.R.N.C. Maia, R.W. Lee, A. Szöke, N. Timneanu, J. Hajdu, NaturePhysics 2 (2006) 839.

[28] Y. Ding, A. Brachmann, F.-J. Decker, D. Dowell, P. Emma, J. Frisch, S. Gilevich,G. Hays, Ph. Hering, Z. Huang, R. Iverson, H. Loos, A. Miahnahri, H.-D. Nuhn,D. Ratner, J. Turner, J. Welch, W. White, J. Wu, Physical Review Letters 102(2009) 254801.

[29] P. Emma, R. Akre, J. Arthur, R. Bionta, C. Bostedt, J. Bozek, A. Brachmann,P. Bucksbaum, R. Coffee, F.-J. Decker, Y. Ding, D. Dowell, S. Edstrom, A. Fisher,J. Frisch, S. Gilevich, J. Hastings, G. Hays, Ph. Hering, Z. Huang, R. Iverson,

H. Loos, M. Messerschmidt, A. Miahnahri, S. Moeller, H.-D. Nuhn, G. Pile,D. Ratner, J. Rzepiela, D. Schultz, T. Smith, P. Stefan, H. Tompkins, J. Turner,J. Welch, W. White, J. Wu, G. Yocky, J. Galayda, Nature Photonics 4 (2010) 641.

[30] P. Musumeci, J.T. Moody, C.M. Scoby, M.S. Gutierrez, M. Westfall, R.K. Li,Journal of Applied Physics 108 (2010) 114513.

[31] J.B. Hastings, F.M. Rudakov, D.H. Dowell, J.F. Schmerge, J.D. Cardoza,J.M. Castro, S.M. Gierman, H. Loos, P.M. Weber, Applied Physics Letters 89(2006) 184109.

[32] O.J. Luiten, S.B. van der Geer, M.J. de Loos, F.B. Kiewiet, M.J. van der Wiel,Physical Review Letters 93 (2004) 094802.

[33] T. van Oudheusden, E.F. de Jong, B.J. Siwick, S.B. van der Geer, W.P.E.M. Op’tRoot, O.J. Luiten, Journal of Applied Physics 102 (2007) 093501.

[34] F.B. Kiewiet, A.H. Kemper, O.J. Luiten, G.J.H. Brussaard, M.J. van der Wiel,Nuclear Instruments and Methods in Physics Research A 484 (2002) 619.

[35] /http://www.acctec.nlS.[36] D.-S. Yang, A.H. Zewail, Proceedings of National Academy of Science USA 106

(2009) 4122.[37] S.B. van der Geer, M.J. de Loos, The General Particle Tracer code, /http://

www.pulsar.nl/gptS.[38] T. van Oudheusden, Electron source for sub-relativistic single-shot femtose-

cond diffraction, Thesis TU, Eindhoven, 2010.[39] R.P. Chatelain, V. Morrison, C. Godbout, S.B. van der Geer, M.J. de Loos,

B.J. Siwick, Ultramicroscopy 116 (2012) 86.[40] C. Lejeune, J. Aubert, Emittance and brightness: definitions and Measure-

ments, in: A. Septier (Ed.), Applied Charged Particle Optics, Part A, AcademicPress, New York, 1980, p. 159.

[41] Y.W. Parc, I.S. Ko, Journal of Korean Physical Society 54 (2009) 2247.[42] G. Pöplau, U. van Rienen, S.B. van der Geer, M.J. de Loos, IEEE Transactions on

Magnetics 40 (2004) 714.[43] H. Park, Z. Hao, X. Wang, S. Nie, R. Clinite, J. Cao, Review of Scientific

Instruments 76 (2005) 083905.[44] J.R. Dwyer, C.T. Hebeisen, R. Ernstorfer, M. Harb, V.B. Deyirmenjian,

R.E. Jordan, R.J.D. Miller, Philosophical Transactions of the Royal Society A364 (2006) 741.

[45] A. Gahlmann, S.T. Park, A.H. Zewail, Physical Chemistry Chemical Physics 10(2008) 2894.

[46] C.-Y. Ruan, Y. Murooka, R.K. Raman, R.A. Murdick, R.J. Worhatch, A. Pell,Microscopy and Microanalysis 15 (2009) 323.

[47] Y. Wang, N. Gedik, IEEE Journal of Selected Topics in Quantum Electronics 18(2012) 140.

[48] G. Sciaini, M. Harb, S.G. Kruglik, T. Payer, C.T. Hebeisen, F.-J. Meyer zuHeringdorf, M. Yamaguchi, M. Horn-von Hoegen, R. Ernstorfer, R.J.D. Miller,Nature 458 (2009) 56.

[49] M. Aidelsburger, F.O. Kirchner, F. Krausz, P. Baum, Proceedings of NationalAcademy of Science USA 107 (2010) 19714.

[50] P. Baum, A.H. Zewail, Proceedings of National Academy of Science USA 103(2006) 16105.

http://www.acctec.nlhttp://www.pulsar.nl/gpthttp://www.pulsar.nl/gpt

Design and implementation of a flexible beamline for fs electron diffraction experimentsIntroductionBeamline designRay tracing designPropagation of a beam with finite size and emittance

Experimental implementationPropagation of the electron beam in the presence of space-charge effectsExperimental characterization of the beamlinePropagation of 25kV electronsPropagation of 30kV electrons

Discussion and conclusionsAcknowledgmentsReferences

design and implementation of a flexible beamline for fs ...€¦ · design and implementation of a...

Documents