22
Particle beam diagnostics and control G. Kube Deutsches Elektronen-Synchrotron DESY Notkestraße 85, 22607 Hamburg, Germany Summary. Beam diagnostics and instrumentation are an essential part of any kind of accelerator. There is a large variety of parameters to be measured for obser- vation of particle beams with the precision required to tune, operate and improve the machine. Depending on the type of accelerator, for the same parameter the working principle of a monitor may strongly differ, and related to it also the requirements for accuracy. This report will mainly focus on electron beam diagnostic monitors presently in use at 4 th generation light sources (single-pass Free Electron Lasers), and present the state-of-the-art diagnostic systems and concepts. 1. – Introduction Nowadays particle accelerators play an important role in a wide number of fields where a primary or secondary beam from an accelerator can be used for industrial or medical applications or for basic and applied research. The interaction of such beam with matter is exploited in order to analyze physical, chemical or biological samples, for a modification of physical, chemical or biological sample properties, or for fundamental research in basic subatomic physics. In order to cover such a wide range of applications different accelerator types are required. Cyclotrons are often used to produce medical isotopes for positron emission to- c Societ` a Italiana di Fisica 1

Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control

G. Kube

Deutsches Elektronen-Synchrotron DESY

Notkestraße 85, 22607 Hamburg, Germany

Summary. — Beam diagnostics and instrumentation are an essential part of anykind of accelerator. There is a large variety of parameters to be measured for obser-vation of particle beams with the precision required to tune, operate and improve themachine. Depending on the type of accelerator, for the same parameter the workingprinciple of a monitor may strongly differ, and related to it also the requirementsfor accuracy. This report will mainly focus on electron beam diagnostic monitorspresently in use at 4th generation light sources (single-pass Free Electron Lasers),and present the state-of-the-art diagnostic systems and concepts.

1. – Introduction

Nowadays particle accelerators play an important role in a wide number of fields

where a primary or secondary beam from an accelerator can be used for industrial or

medical applications or for basic and applied research. The interaction of such beam

with matter is exploited in order to analyze physical, chemical or biological samples, for

a modification of physical, chemical or biological sample properties, or for fundamental

research in basic subatomic physics.

In order to cover such a wide range of applications different accelerator types are

required. Cyclotrons are often used to produce medical isotopes for positron emission to-

c⃝ Societa Italiana di Fisica 1

Page 2: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

2 G. Kube

mography (PET) and single photon emission computed tomography (SPECT). For elec-

tron radiotherapy mainly linear accelerators (linacs) are in operation, while cyclotrons

or synchrotrons are additionally used for proton therapy. Third generation synchrotron

light sources are electron synchrotrons, while the new fourth generation light sources

(free electron lasers) operating at short wavelengths are electron linac based accelera-

tors. Neutrino beams for elementary particle physics are produced with large proton

synchrotrons, and in linear or circular colliders different species of particles are brought

into collision.

As seen from this short compilation there exists a large number of accelerator types

with different properties, and as consequence the demands on beam diagnostics and

instrumentation varies depending on machine type and application. Being aware that

such a wide field will not be summarized in a comprehensive way within a few pages,

this report concentrates on the description of instrumentation and diagnostic concepts

presently in use at 4th generation light sources, i.e. electron linac driven single-pass Free

Electron Lasers (FELs). Examples from the VUV-FEL FLASH at DESY [1, 2] and the

European XFEL (E-XFEL) [3, 4], currently under construction at DESY, will be given.

Monitor concepts applied for particle beam diagnostics rely typically on one of the fol-

lowing physical processes: (i) influence of the particle electromagnetic field, (ii) Coulomb

interaction of charged particles penetrating matter, (iii) nuclear or elementary particle

physics interactions, and (iv) interactions of particles with photon beams. However, there

are fundamental differences in signal generation and underlying physical processes ap-

plied for beam instrumentation between an electron machine and a hadron machine. In

some cases this requires completely different monitor concepts even for the measurement

of the same beam parameter. Therefore the emphasis in the following sections will be on

diagnostics for single–pass FELs.

The reader interested in general aspect of particle beam diagnostics will be referred to

specific textbooks or lecture notes as in refs. [5, 6, 7, 8, 9] and the proceedings from the

DIPAC and BIW conference series. Moreover, a detailed description of the sophisticated

underlying monitor concepts is out of the focus of this report, only short summaries of

the monitor working principles will be given together with references to the appropriate

literature.

This report is organized as follows: in the next section a short introduction to the

instrumentation for beam current measurements will be given. Section 3 presents a brief

overview over the instrumentation for beam position monitoring, while section 4 deals

with transverse phase space diagnostics (i.e. emittance and transverse profile measure-

ments). The last section is dedicated to beam instrumentation for the longitudinal phase

space, i.e. bunch length diagnostics together with energy and energy spread monitors.

In addition, a short introduction to timings systems at FELs will be given.

2. – Beam charge measurements

One of the most important accelerator parameters is the electric beam current resp.

the beam charge. There exist different methods to measure this value, which can roughly

Page 3: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 3

be classified in two categories, intercepting and non–intercepting measurements. In the

following a short description of the most common monitor concepts which are in use at

linacs (AC measurements) will be given: the Faraday cup, the wall current monitor, the

Alternating Current Transformer or Toroid, and as recent development the cavity–based

dark current monitor. For more details the interested reader is referred to the tutorials

about beam current measurements [10, 11, 12] and the recently published overview article

[13].

2.1. Intercepting measurements. – Intercepting measurements are usually destructive

to the beam. The generated monitor signal results from the absorption of a significant

amount of the particle beam energy. Faraday cups are widely used, especially for the

commissioning phase of a linear accelerator and as reference for cross-calibrations. In

order to measure the beam charge, a cup made of conducting material is inserted in the

beam path which is isolated from the beam pipe ground potential. When the beam hits

the cup the charges are collected and integrated, delivering a signal which is proportional

to the primary beam intensity. In case of lepton beams radiative losses may form an

electromagnetic shower which has to be completely absorbed inside the cup material. To

keep the monitor dimensions at reasonable values, Faraday cups are typically deployed

at low particle beam energies, i.e. short behind the gun in a linac. Moreover, due to the

heat–load problems they are usually used only for low current measurements down to the

pA region. In ref. [14] the cup design for an electron beam up to 300 MeV is described.

2.2. Non–intercepting measurements. – Non–intercepting measurements use the elec-

tric or magnetic field coupling of the beam to the measuring instrument to determine the

beam charge by typically integrating the beam current or Wall Image Current (WIC)

coupled inductively or capacitively to the measurement device. The electric field of an

ultra relativistic particle moving inside the vacuum chamber is effectively canceled out-

side the conducting chamber by the WIC induced at the inner chamber diameter, while

the magnetic part of the particle’s field gets strongly attenuated in the non–magnetic

chamber material. As consequence a high–resolution measuring device can only be in-

stalled either in the vacuum chamber (as it is the case e.g. for beam position monitors),

or outside the chamber if an alternative path for the WIC is provided. The latter method

is widely used for beam charge measurements and realized such that a ceramic ring is

soldered at both ends to the beam pipe to form a non–conducting gap through which

the particle’s field leaks out of the vacuum chamber.

A Wall Current Monitor (WCM) is a device with rather high bandwidth up to 5 GHz

and a lower cut–off frequency below hundreds of kHz which is sometimes also used for

longitudinal bunch profile measurements, especially in hadron accelerators with bunch

lengths in the nanosecond region. Here the non–conducting gap is bridged by a resistive

network across the gap. The WCM acts as a current divider, providing separate paths for

the high–frequency WIC component (through the load resistor) and the low–frequency

one. The WCM lower cut–off frequency is proportional to the impedance ratio of the

high–frequency and the low–frequency paths [13]. However, a WCM is prone to noise

Page 4: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

4 G. Kube

because leakage currents may flow directly through the resistors, and therefore a very

good shielding is required. In addition, higher order modes (HOMs) leaking out of the

gap have to be absorbed by ferrites. In case of high intensity beams care has to be taken

that the heat generation due to HOM absorption in the ferrites is effectively dissipated.

In an Alternating Current Transformers or Toroid the beam couples inductively to

the measurement device. A particle bunch crossing the (ceramic) gap in the vacuum

chamber induces a magnetic flux in a high permeability toroid surrounding the gap, i.e.

it acts as a primary single turn winding in a classical transformer. The flux induces a

secondary current in the transformer secondary windings. This current is a measure for

the bunch current and can be detected as a voltage drop across a resistor. The bandwidth

of a toroid ranges from a few Hz up to a GHz. The low–frequency cut–off is given by the

winding inductance, the high–frequency cut–off by the capacitive coupling between the

windings, stray and eddy currents, the energy loss in the core material, and the loss of

permeability with high frequency [13].

Cavity monitors are also well suited for beam intensity monitoring. Here the ampli-

tude of the monopole mode TM010 is a measure for the bunch current. Recently design

and test of a monitor (originally designed as dark current monitor for the E-XFEL) with

a sensitivity sufficient to resolve few-pC bunches was reported [15].

3. – Beam position monitors

Beam position monitors (BPMs) are the diagnostic devices which are most frequently

used at nearly all types of accelerators like linacs, cyclotrons, synchrotrons operating

with lepton, hadron or heavy–ion beams. They are essential during the phase of beam

commissioning, for accelerator fault finding and trouble shooting, machine optics mea-

surements, and accelerator optimization to achieve and keep the ultimate beam quality.

A series of BPMs is distributed along the accelerator to monitor the beam center of mass

position at the distinct locations and derive position information in both transverse orbit

planes.

Each BPM consists of a pickup coupling the particle electromagnetic field to a mea-

surement device. In contrast to the beam intensity monitors the pickups are installed

inside the vacuum chamber, usually consisting of two or four symmetrically arranged elec-

trodes. The signals induced at the pickups are then transferred via UHV feedthroughs to

the outside of the vacuum chamber, followed by a readout electronics system for signal

conditioning and processing.

In the following standard pickup types are shortly described which are common at

most accelerators. They can roughly be diveded in two groups: broadband BPM pickups

and narrowband resp. resonant pickups. Comprehensive review articles can be found e.g.

in refs. [16, 17, 18, 19, 20] and in ref. [21]. The reader interested in readout electronic

systems is referred to the review article ref. [22] or to ref. [23].

3.1. Broadband BPM pickups. – In case of a broadband BPM pickup the monitor

sensitivity is independent on the frequency. Broadband pickups usually operate in terms

Page 5: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 5

of the image current model [21].

The pickup type mainly in use is the so called button BPM which provides a capacitive

or electrostatic coupling. The small button size (typical button diameters 5 - 20 mm)

and the short vacuum feedthrough allows for a compact installation at a comparable

low price, thus explaining the proliferation of this pickup type. The pickup transfer

impedance has a typical high–pass characteristics with a cut–off frequency ωcut defined

by the pickup capacity and the matching impedance. For frequencies ω ≪ ωcut the

voltage drop measured across a matched resistor is proportional to the time derivative

of the beam current.

In order to increase the signal strength the first idea would be to increase the pickup

area, thus increasing the induced charge. However, if the bunch length is in the order of

the pickup size a signal deformation occurs due to the final propagation time. Therefore

the button size should be smaller than the bunch length which results in a very low

signal strength for short bunches. In this case stripline BPMs are well suited because the

signal propagation is considered in the design as it is the case for transmission–lines in

microwave engineering [20]. In addition, the azimuthal coverage of the stripline can be

larger than that of a button pickup, thus yielding an increased signal strength. A stripline

pickup consists of an electrode with length L of several cm, forming a transmission line

between the electrode and the vacuum chamber wall. A signal is created by the beam at

each end of the line depending on the characteristic impedance Zstrip of the electrode,

which is often chosen to be 50 Ω. Depending on the termination R of the downstream

port the signal there is canceled (R = Zstrip) or appears partially (R = Zstrip). A

complete cancellation at the downstream port happens only if the speed of the beam is

equal to the speed of the signal in the transmission line which is almost true for β ≈ 1.

At the upstream port it is always possible to extract the induced signal and the reflected

inverted one separated in time by ∆t = 2L/c (for β = 1). The transfer impedance of a

stripline BPM is composed of a series of maxima fmax = c/4L · (2n−1) with n = 1, 2, . . .

Therefore, for a given acceleration frequency facc the length L should be chosen to work

close to such a maximum.

3.2. Resonant BPM pickups. – In order to fulfill the high position resolution require-

ments of a linac–based FEL (E–XFEL: in the undulator section a single–shot accuracy of

≤ 1µm, bunch train averaged 0.1µm) an increase in the BPM signal strength is required.

However, the standard BPM scheme relies on subtracting the signals from individual

pickup electrodes to extract a position information, i.e. the signal power of the position

information is typically much smaller than the signal power from an individual elec-

trode. To avoid this reduction of information it is desirable to generate a beam signal

right from the beginning which is proportional to the beam position. These demands

can be fulfilled by excitation of resonator modes within a cavity, i.e. the resonator can

be utilized as passive, beam driven cavity BPM. The cavity BPM is usually based on a

pillbox resonator, and because the short bunch lengths of a linac deliver a wide frequency

spectrum several resonator modes are excited resonating at slightly different frequencies.

Position information is gained by detecting via pin antennas the TM110 dipole mode

Page 6: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

6 G. Kube

which has a node at the cavity center and whose excitation amplitude depends linearly

on the beam displacement. Care has to be taken that no signal leakage of the slightly

frequency–shifted monopole mode TM010 in the dipole mode signal occurs, which can

be suppressed by an appropriate design of an outcoupling waveguide. For the position

determination the amplitude of the dipole mode and its phase have to be processed; the

magnitude represents the value of beam displacement and the sign is reconstructed from

the phase which is defined with respect to the phase of the monopole mode. Therefore

the TM010 mode has to be processed additionally which is usually performed via a ref-

erence cavity. The best resolution achieved so far with a cavity BPM was 8.72 nm [26],

but the projected design resolution of this monitor amounted 2 nm. More information

about cavity BPMs in general can be found e.g. in refs. [24, 25].

4. – Diagnostics for the transverse phase space

Particle beam properties in the transverse phase space are characterized by the trans-

verse beam emittance which is one of the accelerator key parameters: for a synchrotron

light source of 3rd or 4th generation the brilliance defining the accelerator performance

scales inversely proportional to the transverse emittances, and a similar situation occurs

for the luminosity of a particle collider. The transverse emittances are the projected

phase space areas, and according to Liouville’s theorem they are conserved in linear

beam optics. In accelerator physics the transverse phase space variables are chosen to

be the beam position x and the beam angular divergence x′ = ∆px/p.

The transverse emittance is either described in the form of an ellipse equation via the

Courant–Snyder or Twiss parameters as

ε = γx2 + 2αxx′ + βx′2 ,(1)

or alternatively according to ref. [27] there exist a statistical definition

εrms =√< x2 >< x′2 > − < xx′ >2 with < x2 >=

∫ +∞−∞ dxx2ρ(x)∫ +∞−∞ dxρ(x)

, . . .(2)

which is widely used at linacs because of the quite general applicability. The latter

equation is based on the characterization of the beam charge distribution ρ(x) by its

2nd statistical moments, and it was assumend that all first moments vanish (which can

be achieved in any case by a variable substitution because the first moments describe a

static offset). More details about the general beam emittance concept can be found in

each textbook about accelerator physics, or e.g. in refs. [28, 29].

Unfortunately the emittance itself is not directly accessible for beam diagnostics, the

measurable quantities are the projections onto both axes: a beam profile or a beam

divergence. Usual beam emittance measurements rely on the analysis of beam profiles,

therefore after the description of general emittance diagnostic principles a short review

about beam profile measurement schemes at linacs will be given.

Page 7: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 7

It is important to note that all measurement techniques presented in this section

refer to measurements of the projected emittance. However, an FEL relies on a resonant

energy exchange between the electron beam and the emitted photon field, and the FEL

working principle requires that the electron is slipping back in phase with respect to

the photon field by one radiation wavelength each undulator period. Therefore the FEL

is integrating over the slippage length, and it is not only the projected but also the

slice emittance which is of interest. A method to determine the slice emittance will be

presented in the next section in context with longitudinal phase space diagnostics.

There exist two general schemes for (projected) emittance diagnostics which can

roughly be classified in beam matrix based schemes exploiting the transfer properties

of the beam matrix resp. the Twiss parameters, or a mapping of the phase space where a

small phase space element (beamlet) is separated e.g. by a slit, and the beamlet propa-

gation along a drift space converts the angle information in a position information. Both

schemes will be shortly described in the following subsections. For more details about

transverse emittance diagnostics the reader is referred to the review article ref. [30].

4.1. Beam matrix based schemes. – For simplicity the case of uncoupled motion is

considered, i.e. the motion in horizontal, vertical (and longitudinal) plane is uncoupled

and the following considerations are restricted to a 2 dimensional sub–space, here the

horizontal phase space.

The beam matrix can be expressed either in terms of the Twiss parameters or, based

on the statistical approach, in terms of the moments of the beam distribution as

Σbeam = ε

(β −α−α γ

)=

(< x2 > < xx′ >

< xx′ > < x′2 >

)(3)

with σ =√Σ11 =

√εβ =

√< x2 > the rms beam size, i.e. the parameter which can be

measured for beam diagnostic purposes. The beam emittance is connected to the beam

matrix via

ε =√

|detΣbeam| =√Σ11Σ22 − Σ2

12(4)

in accordance to eqs.(1,2). While the single–particle transformation of the phase space

coordinates (x,x’) from an initial to a final location in the accelerator lattice is represented

by the so-called transport or R–matrix(x

x′

)f

=

(R11 R12

R21 R22

)(x

x′

)i

,(5)

the transformation of an initial beam matrix to the final observation point is performed

according to

Σfbeam = RΣi

beamRT .(6)

Page 8: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

8 G. Kube

For beam diagnostic measurements only the beam matrix element Σ11 is directly acces-

sible and corresponds to the square of the beam size. The remaining matrix elements

required for an emittance determination according to eq.(4), Σ12 = Σ21 and Σ22, have to

be determined otherwise. Exploiting the transformation properties of the beam matrix

eq.(6) the element Σ11 (being the observable) is transformed as follows:

Σf11 = R2

11Σi11 + 2R11R12Σ

i12 +R2

12Σi12 = σ2

f .(7)

From Eq.(7) an emittance measurement scheme can be deduced. A reference location in

the accelerator (indicated by index i) is assumed. By measuring a beam profile σf under

a slightly different situation (index f), a connection between the original and the final

state can be established according to eq.(7). In an accelerator this change in state can

be achieved in a pre–defined way, i.e. the matrix elements of the R–matrix are known

a priori. Then eq.(7) represents an equation with 3 unknown parameters: the beam

matrix elements at the reference point, while all other parameters are known either by

measurement (Σf11) or a priori (R-matrix elements). To determine the 3 unknown beam

matrix elements at the reference location a minimum of 3 beam size measurements under

different situations (index f) is required. With knowledge of the beam matrix elements

at the reference point the emittance at the reference point can be deduced according to

eq.(4), but supposing linear beam optics the emittance itself is a global parameter and

does not depend on the location in the accelerator.

Usually there are two possibilities to establish a slightly different situation for a beam

profile measurement. The first method relies on a profile measurement with a single

profile monitor in the accelerator, but a change in the beam optics (i.e. the elements of

the R matrix) between monitor and reference point. This is the basis of the quadrupole

scan method where the focussing strength of a quadrupole in front of the monitor is

varied, and the beam size is measured as function of the focussing strength. In the

second method the beam optics is kept fix but the beam size is measured at different

locations in the accelerator with different profile monitors. This is the basis of the muli–

profile monitor method. It is common for both emittance schemes that the number of

profile measurements is chosen to be larger than the minimum number of 3, and the

data are subjected to a least–square analysis. However, it is obvious that the number

of available data points from a quadrupole scan is munch larger than from the multi–

profile measurement because the space for profile monitors in an accelerator is limited.

In addition, both methods are invasive and cannot be used parasitically.

4.2. Phase space mapping based schemes. – Beam matrix based emittance diagnostics

relies on the assumption of undisturbed phase space evolution. However, especially at

low particle beam energies this assumption can be violated because space charge effects

may affect the phase space evolution in a non–linear way. To reduce the space charge

influence it is common to subdivide the beam into small samples (beamlets) which are

emittance dominated and not any more space charge dominated, and to investigate the

phase space evolution of each beamlet separately.

Page 9: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 9

The subdivison of the beam into beamlets is performed by a small slit which cut out

a small vertical slice in the beam phase space: all beam particles are absorbed in the

slit material (typically tungsten or tantalum) except the ones crossing the slit opening.

This beamlet is propagating along a drift space, so converting the angle information

into a position information which is finally detected with a profile monitor. By scanning

the slit position through the beam the transverse phase space is scanned and can be

reconstructed to deliver the beam emittance. In ref. [31] the emittance reconstruction

procedure is described in detail.

Instead of a time consuming scanning of the slit in the space coordinate, the absorber

can be designed such that it contains a mult–slit arrangement. The advantage of a

multi–slit device is the capability to perform single–shot emittance measurements in one

transverse plane, i.e. it allows to study shot-to-shot emittance fluctuations. An example

for such kind of design together with operational results is described e.g. in refs. [32, 33].

To extend this method to single–shot capability in both planes the design is extended

to a pepper–pot which consists of a matrix of holes in an absorber plate. Typically only

a tiny fraction of the beam crosses the holes (about 1%), the rest of the beam particles

is absorbed in the plate. Therefore heat load considerations are an important task in the

design phase, and a high sensitivity detection system is required. In ref. [34] some useful

hints concerning pepper–pot emittance analysis are given.

While multi–slit and pepper–pot emittance monitors have the advantage of a single–

shot capability, their range of application is restricted to the region of low particle en-

ergies, at least for lepton beams. In order to work properly only particles crossing the

slit resp. the holes should contribute to the profile measurement, the rest of the beam

particles should be completely absorbed. In order to absorb high energetic electrons

a very thick absorber is required, but in this case particle scattered inside the slits or

holes deteriorate the profile measurement. Therefore, this kind of emittance monitor is

typically used up to electron beam energies of about 100 MeV. However, in ref. [35]

emittance diagnostics with a pepper–pot for 508 MeV electrons is reported.

All emittance diagnostic techniques presented so far rely on measurements of the

beam profile. Therefore in the following subsection various techniques to determine

beam profiles will be shortly reviewed.

4.3. Beam profile measurements. – It is common to all beam profile monitors that a

secondary signal is created with an intensity that is proportional to the charge density

of the particle beam. This secondary signal can be generated in an interaction of the

beam with matter (e.g. wire scanners, scintillation screens), with photons (e.g. laser

wire scanner), or by separating the particle electromagnetic field from the beam particles

(e.g. transition radiation or synchrotron radiation monitors). The secondary signal

can be either a flux of charged particles (secondary electrons) or of electromagnetic

radiation (visible light or γ–radiation), and in the detection scheme spatial resolution

can be achieved via scanning a target or detector, or by spatial resolving detectors.

Wire scanners deliver a very direct and reliable measurement with an achievable

resolution down to 1 µm. The operation principle is such that a thin wire (material C,

Page 10: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

10 G. Kube

W, Be, . . . and wire size down to a few microns) is scanned across the beam, and the signal

from the beam–wire interaction is detected as function of the wire position. For intense

and high brilliant beams a fast scan speed of 5–20 m/s is required in order to minimize

the emittance blow–up and to reduce the heat load on the wire. The secondary signal

from the beam–wire interaction consists either from scattered beam particles and/or

bremsstrahlung which is measured outside the vacuum chamber by a fast scintillation

counter. Alternatively the secondary electron emission (SEM) signal from the beam–wire

interaction can be used. The latter method is often applied for low energy beams where

the scattered particles cannot penetrate the vacuum chamber and the bremsstrahlung

intensity is to low. However, in this case one has to be careful, because if the wire

temperature exceeds the thermionic threshold thermal electrons may superimpose the

SEM signal. Wire scanners have the drawback that they deliver projected beam profiles

and not the full 2D information. In some setups a third wire mounted under 45 with

respect to the other 2 wires is used to have an estimate about the coupling between

both transverse planes. An example for such a setup is described in ref. [36]. Instead of

scanning a wire across a beam a grid of wires can be installed. Such a device is called a

harp because of the appearance. Due to the minimum achievable distance between the

individual wires in the order of a few 100 µm, a harp is suited to measure beam profiles

in the order of a few mm and therefore is mainly used at hadron linacs.

The operational principle of a laser wire scanner is very similar to the one of a

conventional wire scanner with the exception that a laser beam is scanned across the

particle beam, and the secondary signal consists of forward-scattered Compton–γ’s, see

e.g. ref. [37]. In principle a resolution down to 1 µm can be achieved, limited by the

minimum laser spot size at the interaction point which can be realized. The advantage of

a laser wire scanner is that it is an almost non–invasive measurement technique which can

in principle operate parasitically, and that the heat load problem in conventional wires

can be avoided, but again the monitor delivers only projected beam profiles but not the

full 2D information. So far only a few laser wire scanners are installed at accelerators,

but they operate more like an experimental system than a beam monitor for daily use.

Taking advantage of the rapid development and the huge market for commercial

available optical sensors, in the past years optical measuring techniques took on greater

significance. Nowadays area scan CCD or CMOS sensors are widely used in beam diag-

nostics because they provide the full 2D information about the transverse particle beam

distribution, allowing in principle to investigate shot-to-shot profile fluctuations at mod-

erate repetition rates. To apply optical measuring techniques the information about the

particle beam charge distribution has to be converted in an optical intensity distribu-

tion which can be recorded by a standard area scan detector. In the selection of this

conversion process, care has to be taken that (i) any resolution broadening introduced

by the basic underlying physical process has to be small (i.e. the Point Spread Func-

tion PSF of the physical process corresponding to the single particle resolution function

should not dominate the total spatial resolution), and that (ii) the conversion process

should be linear to avoid any deformation of the intensity distribution. There exist two

principle possibilities for this conversion process, either to exploit the interaction of the

Page 11: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 11

beam particles with matter (used e.g. for scintillation screens or residual gas luminescent

monitors), or the particle electromagnetic field has to be separated from the beam to be

detected in the far field as radiation.

Optical beam profile monitors probing the particle electromagnetic field are widely

used, therefore the process of radiation generation will be shortly explained in the follow-

ing. Considering an ultra–relativistic electron with an electric field which is relativistic

contracted (i.e. mainly transversal). The degree of contraction is described by the field

opening angle 1/γ with γ = E/m0c2 the Lorentz factor. In the limiting case γ → ∞

the field would be completely transversal and correspond to a plane wave (classical de-

scription of a photon). This situation occurs either by considering a particle with zero

rest mass (i.e. a photon), or in the limiting case if the beam energy is increased into

the relativistic regime. Due to the similarity between a real photon and the field of

an ultra–relativistic particle, the action of this particle is described by so called virtual

photons. However, to measure radiation in the far field the virtual photon field bound to

the beam particle has to be separated from the particle. In case of a circular accelerator

this is achieved by a force acting on the charged particle which is caused by the magnetic

field of accelerator (bending) magnets, and the resulting radiation is called synchrotron

radiation. However, synchrotron radiation based profile monitors are out of focus of

this paper, they are reviewed e.g. in ref. [38]. In case of a linear accelerator there is

(per definition) no particle bending, but the separation can be achieved by acting on

the virtual photons itself via structures that diffract the particle electromagnetic field

away from the particle. For better understanding the analogy between real and virtual

photons can be exploited. Real photons can be refracted resp. reflected at a surface, the

same holds for virtual photons. In this case the radiation is named Forward/Backward

Transition Radiation. In classical optics the effect of edge diffraction is known, in the

case of virtual photons the radiation effect is called Diffraction Radiation. Real photons

can be diffracted at a grating, the same hold for virtual photons and the effect is called

Smith–Purcell Radiation. Finally, high–energetic real photons (X–rays) are diffracted

at a 3D structure of a crystal, and if a charged particle beam traverses such crystal

Parametric–X Radiation is emitted. It is common to all these radiation phenomena de-

pending on the diffraction of virtual photons that the radiation angular distribution has

an on–axis minimum which reflects the property of the incoming virtual photon field and

therefore is a fundamental one.

Optical Transition Radiation (OTR) monitors are widely used for profile measure-

ments at linacs. The radiation is emitted when a charged particle beam crosses the

the boundary between two media with different optical properties, here a thin reflecting

screen (e.g. a silicon wafer covered with a thin layer of aluminum or silver) and vacuum.

For beam diagnostic purposes the visible part of the radiation is used and an observation

geometry in backward direction is mainly chosen which corresponds to the reflection of

virtual photons at the screen which acts as mirror. The screen has an inclination angle

of 45 with respect to the beam axis, and observation is performed under 90. In a

typical monitor setup the beam is imaged via OTR using standard lens optics, and the

recorded intensity profile is a measure of the particle beam spot. The principle achiev-

Page 12: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

12 G. Kube

able resolution of an OTR monitor is given by δr = 1.12 Mλ/θm with λ the wavelength

of observation, θm the lens acceptance angle, and M the magnification of the optical

system (see e.g. ref. [39]). The best resolution achieved so far amounted 1 µm. With

this monitor a vertical beam size of about 5 µm was measured at the accelerator test

facility ATF at KEK (Japan) [40]. Advantages of OTR are the instantaneous emission

process enabling fast single shot measurements, and the good linearity (neglecting coher-

ent effects). Disadvantages are that the process of radiation generation is invasive, i.e.

a screen has to be inserted in the beam path, and that the radiation intensity is much

lower in comparison to scintillation screens. Therefore for low–energetic electron beams

luminescent screen monitors are used instead of OTR screens.

For high intensity or high brilliant electron beams and OTR monitors the interaction

of the beam with the screen material may lead to a screen degradation or even a damage.

In these cases the applicability of OTR monitors is usually restricted to single or few

bunch operation, and permanent beam observation is not possible. In order to overcome

this drawback Optical Diffraction Radiation (ODR) monitors might be a suitable alterna-

tive. Diffraction radiation is emitted when a charged particle passes through an aperture

on a boundary of two media with different optical properties, usually through a thin slit

is a screen. The radiation is generated in an interaction of the particle electromagnetic

field with the screen, taking advantage that the field has a certain transverse extension

characterized by the impact parameter h = λγ/2π with λ the wavelength of observation

and γ the Lorentz factor. If the beam is imaged via ODR as it was the case with OTR,

the result would be an illuminated slit image but no beam profile. In order to extract

information about the beam profile the visibility of the interference fringes of the angular

distribution is exploited. So far ODR based diagnostics is still in an experimental stage,

results of an ODR experiment performed at FLASH are described e.g. in ref. [41]. In

the meantime the experiment was extended to perform ODR interferometry generated

at two slits to increase the sensitivity on the beam parameters [42].

Recently the experience from modern linac based light sources showed that profile

diagnostics based on the detection of radiation originating from the particle electromag-

netic field might fail because of coherence effects in the emission process. Cause of this

coherent emission is the Microbunching Instability, i.e. some unstable micro structures in

the electron bunch that compromise the use of this kind of monitors as reliable diagnos-

tic tool. Coherent OTR (COTR) was observed e.g. at the Linac Coherent Light Source

LCLS in Stanford (USA) and at the free–electron laser FLASH at DESY in Hamburg

(Germany). In ref. [43] the present knowledge about COTR observations is summarized.

As consequence for the new 4th generation light sources new reliable tools for transverse

beam profile measurements are discussed. One option would be to use transition radi-

ation (TR) profile diagnostics at higher photon energies where coherence effects should

be absent. In ref. [44] a test experiment is described dedicated to the investigation of

the TR intensity in the EUV region.

A further option currently under discussion is to use Luminescent Screens. In a

luminescent screen the energy from the beam particles deposited in the screen material

is converted into atomic excitations which are followed by radiative relaxations. The

Page 13: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 13

light intensity of these optical transitions is a measure for the particle beam profile. In

contrast to OTR the scintillation light is emitted isotropically, i.e. there is no restriction

on the observation geometry and screen and camera can be placed under arbitrary angles

with respect to each other. Inorganic scintillators are of special interest for electron

beam diagnostics because of their good radiation resistance, high stopping power for

high light yield, and short decay times of the excited atomic levels. The application of

scintillation screens for particle beam diagnostics was recently reviewed in ref. [45], and

ref. [46] describes the results of a study about the achieveable spatial resolution for profile

measurements with different scintillator materials and micro–focussed ultra–relativistic

electron beams. According to this study LYSO:Ce and BGO scintillators could be a

suitable screen material for electron beam diagnostics. Moreover it was demonstrated

that the observation geometry strongly affects the achievable resolution and therefore

has to be considered carefully. Furthermore care has to be taken that additionally OTR

is produced at the boundary between screen and vacuum which might be reflected from

the screen surface to the camera. Usually the intensity of this OTR contribution is to

neglect, but in the case of COTR emission it may even compromise the use of scintillating

screen monitors as shown in ref. [47]. As it was demonstrated there exist to possibilities

to suppress the COTR contribution: (i) a temporal suppression by using a fast gated

camera, and (ii) a spatial one by tilting away the screen surface from the camera. While

the first method was already tested successfully the possibility of spatial suppression is

presently under investigation.

5. – Diagnostics for the longitudinal phase space

As it was the case for the transverse phase spaces, the longitudinal beam properties

are characterized by the longitudinal emittance. In case of the longitudinal phase space

the phase space variables are the phase deviation ϕ (measured with respect to the syn-

chronous particle), and the momentum deviation δ = ∆p/p. In case of ultra–relativistic

particles as it is usually the case for electrons, the momentum deviation can be identified

with the energy deviation, i.e. δ = ∆E/E. Considering a particle bunch rather than

a single particle, the rms values are taken to characterize the particle ensemble. The

characteristic values are then the phase spread σϕ =√< ϕ2 > and the energy resp.

momentum spread σδ =√< δ2 >. In beam diagnostics the phase spread is a rather

cumbersome parameter because it is not directly accessible. Instead of the bunch length

σz is usually taken which is related to the phase spread according to σz = c/ωRF · σϕ.The formalism for the treatment of the longitudinal phase space is rather similar to the

one for the transverse phase spaces, i.e. the beam matrix is constructed according to

eq.(3) simply by replacing x by ϕ and x’ by δ, and the emittance is deduced from the

beam matrix by eq.(4). Because of the similarity between longitudinal and transverse

phase space, in principle similar techniques for emittance diagnostics could be applied:

instead of a quadrupole scan with transverse profile monitor a longitudinal focussing de-

vice with bunch length monitor could be used. Such longitudinal focussing device exist

and is called buncher, however a buncher works effectively only at low particle beam

Page 14: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

14 G. Kube

energies (short behind the gun) and is used mainly at hadron accelerators. Instead of

measuring directly the longitudinal emittance, usually their projections onto the phase

space axes are investigated, i.e. the bunch length and the energy spread. Together with

diagnostic schemes for both of these parameters, in the following particle beam energy

measurements and timing and synchronization issues are briefly presented.

5.1. Bunch length measurements. – Depending on the type of accelerator the length of

a bunch is spread from the nanosecond region (hadron storage ring) over the picosecond

region (lepton storage ring) to the range of several femtoseconds (linac driven FEL with

bunch compression). To cover such a wide range a number of different techniques are

applied which can briefly be classified in (i) electromagnetic monitors, (ii) optical meth-

ods, (iii) methods based on the bunch frequency spectrum, (iv) laser based monitors,

and (v) monitors based on RF cavity manipulation. Due to the bandwidth limitation

of electromagnetic monitors typically in the range of several GHz, these techniques are

mainly applied at hadron accelerators and will not be considered in the following.

Optical Methods are widely applied because standard instrumentation is available,

typically also for reasonable costs. There exist some sophisticated schemes like non–

linear mixing [48] or the analysis of the shot noise spectrum [49] which will not be

discussed in the following. The device which is mainly in use at various accelerators is

the Streak Camera (SC). In a SC visible light generated in an instantaneous process from

the beam (synchrotron radiation, OTR, Cherenkov radiation, . . .) hits a photo cathode

where the light pulses are converted into a number of electrons proportional to the

incident intensity distribution. The photo electrons are then accelerated along a streak

tube and transversely swept by deflecting plates so that the incident time distribution is

converted in a spatial distribution on a Micro Channel Plate (MCP). The photo-electrons

amplified by the MCP are impinging on a phosphor screen where they are reconverted in

visible light which is detected with a scientific grade CCD detector. A SC is a powerful

tool commercially available to study bunch lengths and longitudinal beam dynamics.

However, with a resolution limit of about 200 fs [50] it is not sufficient to resolve beam

profiles from fully compressed beams of a FEL. According to ref. [51] there are 3 factors

limiting the time resolution: (i) the initial velocity distribution of the photo electrons

from the photo cathode (see also ref. [52]), the temporal spread occuring in the deflection

field, and (iii) the time spread because of space charge effects inside the streak camera.

Therefore a SC should be operated with a narrow–bandwidth interference filter in front

of the camera and at a low incoming intensity level which can be adjusted by the entrance

slit in front of the SC system. Furthermore, in order to achieve a resolution of about 200

ps a reflective light collecting optical system should be used because dispersion effects in

the lenses spoil the time resolution.

To overcome the resolution limit imposed by a streak camera different diagnostics

schemes can be applied, e.g. based on the detection of the bunch frequency spectrum:

as shorter the bunch length, as broader the frequency spectrum of the electron bunch.

Therefore the bandwidth of the radiation produced from a Gaussian bunch may extend

to the THz region for ultra short bunch lengths. One option to measure the bunch length

Page 15: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 15

via the frequency spectrum is to use an RF pickup as described in ref. [53]. However,

methods widely applied in particle beam diagnostics are based on Coherent Radiation

Diagnostics (CRD) [54]. Radiation is emitted coherently if the wavelength is in the order

of the bunch length, i.e. information about bunch length and shape is encoded in the

emission spectrum which is exploited in CRD. In case of coherent emission the spectral

intensity is strongly amplified which can be expressed in the following form:

dU

dλ=

(dU

)1

(N +N(N − 1)|F (λ)|2

)with F (λ) =

∫ ∞

−∞dz S(z) e−2πiz/λ .(8)

Here (dU/dλ)1 is the single particle emission spectrum, N the number of particles in the

bunch, and F (λ) the bunch form factor which is related to the normalized bunch profile

S(z) via a Fourier transform. According to eq.(8), from a measurement of the spectral

intensity and with knowledge of the single electron spectrum together with the bunch

charge, the form factor can be determined. Inverting the Fourier transform results in

the reconstructed bunch profile S(z). The situation is more complex because it is the

magnitude |F (λ)| of the form factor which is determined rather than the complex form

factor itself. Reconstruction is possible only if both amplitude and phase are available.

Although a strict solution of this phase–reconstruction problem is not possible, a so–

called minimal phase can be constructed with the Kramers–Kronig relation which gives

a handle to solve this problem satisfactory. A detailed treatment of this problem can

be found e.g. in ref. [55]. In principle any kind of coherent radiation can be used as

a radiation source. Measurements were performed with coherent synchrotron radiation,

transition radiation, and diffraction radiation. The resolution of CRD is limited to about

100 fs, mainly caused because of uncertainties in the spectral reconstruction. Drawback

of CRD is that the radiation sources are polychromatic, i.e. a spectrometer is required

for the spectral decomposition which is usually a scanning device and does not allow

single–shot measurements. In this context the development of multi–stage spectrometers

(see e.g. ref. [56]) might be an interesting alternative, or the application of Smith–Purcell

radiation [57] because the radiation source is dispersive by itself.

To achieve better resolution laser–based methods can be applied. The laser can be

used either by directly scanning the particle beam profile (laser wire scanner, see ref.

[58]), or by probing the action of the particle electromagnetic field (similar as an elec-

tromagnetic pickup does, but with much higher bandwidth providing better resolution).

The latter scheme is exploited in Electro–Optical (EO) techniques. EO detection schemes

can be applied for ultra–relativistic electrons where the particle Coulomb field is purely

transversal, i.e. the field strength of the non–propagating particle field is a measure of the

longitudinal bunch profile. If the bunch passes close to an electro–optical crystal (ZnTe

or GaP), its Coulomb field induces a change in the crystal refractive index (so called

Pockels effect). The information about the longitudinal profile is therefore encoded in an

refractive index change which can be converted into an intensity variation by means of a

laser together with polarizers. In the simplest scheme of EO sampling a polarized laser

beam is scanned along the bunch, and the change in intensity is recorded as function of

Page 16: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

16 G. Kube

the time delay. There exist more sophisticated schemes with even the capability of single

shot resolution like spectrally [59], temporally [60], or spatially [61] resolved detection.

Principle resolution limits of EO bunch lengths measurements are crystal phonon reso-

nances, dispersion in the crystal leading to a gradual distortion and lengthening of the

THz pulses, and a laser pulse broadening which can be neglected for crystal thicknesses

smaller than 200 µm. The shortest bunch length resolved so far was about 60 fs with

temporally resolved EO detection, being close to the principle limit of a GaP crystal.

Detailed information about EO techniques and measurements performed at FLASH can

be found in ref. [62].

The most state-of-the-art instrument for bunch length measurements and even more is

a Transverse Deflecting Structure (TDS). A TDS is an iris loaded RF waveguide structure

designed to provide hybrid deflecting modes (HEM11), a linear combination of TM11 and

TE11 dipole modes resulting in a transverse force that acts on the synchronously moving

relativistic particle bunch. They exist as traveling wave structures [63, 64] or as standing

wave structures [65], and their working principle resembles that of an intra beam streak

camera: a single bunch inside the bunch train, traversing the structure at an appropriate

RF phase experiences a vertical kick which depends linearly on time and vanishes in

the bunch center. Due to the vertical deflection, the vertical position of the electrons

inside this bunch are linearly correlated to their longitudinal coordinates. Usually a fast

horizontal kicker deflects the bunch onto an off–axis screen. The spot at the screen in

vertical direction is a measure of the longitudinal bunch profile. The resolution limit of

a TDS is determined by

σres =E/e

V02πfRF cosψ·

√ε√

βtds sin∆ϕ(9)

with E the beam energy, V0 the deflecting voltage (should be high for good resolution),

fRF the TDS RF frequency (should be high), ψ the phase between bunch centroid and

RF wave (should be operated at zero–crossing), ε the transverse beam emittance, βtdsthe β function at the location of the TDS (should be as large as possible for most effective

kick), and ∆ϕ the phase advance between TDS and location of the screen (should amount

90 or 270). At FLASH a traveling wave–type vertical deflecting S–band (2.865 GHz)

RF structure is operated [66]. For the E–XFEL several TDS setups are planned with a

resolution of about 10 fs/sin∆ϕ [67].

Besides the excellent time resolution, the advantage of operating a TDS is to have

the possibility to measure additional beam parameters in a time resolved way which are

not accessible with any other diagnostic instrument. For example, the streaked image

onto the screen in horizontal direction contains information about the horizontal beam

size, but now for each longitudinal position inside the bunch. If a quadrupole scan is

performed with one of the upstream quadrupoles in front of the structure, the horizontal

beam size in each slice is determined as function of the quadrupole settings, and this

technique gives access to the horizontal slice emittance [68]. The quadrupole scanning

gives rise to a rotation of the beam in the transverse phase space. Applying tomographic

Page 17: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 17

reconstruction schemes (see e.g. ref. [69]) it is even possible to reconstruct the complete

phase space information in one transverse plane [70]. In addition, with a screen located in

a horizontally dispersive section behind the deflecting structure it is possible to measure

directly the longitudinal phase space distribution, i.e. to have access also to the energy

spread [71].

5.2. Energy spread . – The energy spread is a phase space variable of the longitudinal

phase space. However, the dispersion in an accelerator results in a coupling between the

transverse (i.e. usually the horizontal one) and the longitudinal phase space. Therefore

measurements of the energy spread can be performed either with longitudinal or with

transverse phase space diagnostics schemes.

Longitudinal schemes are widely used at circular accelerators where the rms energy

spread linearly depends on the bunch length. By changing the synchrotron frequency

via the accelerator RF and measuring the bunch length with a streak camera the energy

spread can be deduced, see e.g. ref. [72].

Measuring the impact of the energy spread in the transverse phase space is simply

achieved by a transverse profile measurement. Taking the closed orbit contribution of

an off–momentum particle into account in the displacement of a particle from the ideal

trajectory, the rms beam size is affected by an additional contribution:

σx =

√εβ +

(ησδE

)2

(10)

with η the dispersion. According to eq.(10) a direct way to measure the energy spread in

a linac is a proper selection of the transverse profile monitor location in the accelerator

where the beam spot size is fully determined by the dispersion contribution. With

knowledge of η the energy spread can easily be deduced. An example of such kind of

measurement is described in ref. [73].

A completely different and non–invasive detection scheme of the energy spread is de-

scribed in ref. [74]. In this Beam Energy Spread Monitor (BESM) the multipole moments

induced by the passage of a charged particle beam in the multi–stripline electrodes (here:

8 units) of an electromagnetic pickup are exploited. The induced quadrupole moments

contain information about the beam size which depends on the energy spread according

to eq.(10).

Finally the time resolved energy spread can be measured with a TDS if the screen is

located in a horizontally dispersive section behind the deflecting structure as described

in the previous subsection.

5.3. Beam energy . – The precise knowledge of the beam energy is of vital interest

for the experiments performed at an accelerator. For high energy physics experiments

the accuracy in the particle beam energy determines the accuracy in the location of

particle resonances and their widths, while for synchrotron light sources the energy is

important to characterize the parameters of the insertion devices. Moreover, for the

operation of an accelerator it is important to know the absolute beam energy to precisely

Page 18: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

18 G. Kube

understand the particle beam optics. Ref. [75] gives an overview over different methods to

determine the beam energy. According to this reference the different schemes for energy

determination can roughly be classified in (i) measurements via dipole spectrometers, (ii)

measurements via particle or nuclear physics processes, (iii) photon based methods, (iv)

energy measurements from the central frequency, (v) energy losses due to synchrotron

radiation, and (vi) resonant depolarization. While the highest accuracy is achieved with

the last method (∆E/E ≈ 10−5 − 10−6) it is only applicable for circular accelerators.

The measurement schemes in use at linacs are either dipole spectrometers or photon

based methods, and in the following these techniques will be briefly discussed.

The principle of a Dipole Spectrometer relies on the fact that a dipole transforms a

momentum resp. momentum spread into a position resp. position spread. The deflec-

tion angle is proportional to the integral over the magnetic field seen by the particle,

and inversely proportional to the particle momentum: θ = p−10

∫dsB. The detection

scheme relies on spatial resolving detectors, i.e. BPMs or screens. For high resolution

measurements one has to take into account that the beam position in dispersive direc-

tion behind the dipole not only depends on the position x0 in front of the dipole, but

also on the angular divergence x’0. Therefore at least 2 position sensitive monitors in

front of the dipole are required for a simultaneous measurement of x0 and x’0. To define

both parameters very precisely in case of large emittance beams an entrance slit in front

of the spectrometer is recommended. An example for a spectrometer setup is the LEP

spectrometer [76]: with 1 µm position accuracy and a field stability of ∆B/B ≈ 10−5,

an energy resolution of about ∆E/E ≈ 10−4 could be achieved.

A photon originating from or interacting with the particle beam carries information

about the beam energy which is encoded in either the photon spectrum or the angular

distribution. Therefore Photon based Methods rely on the measurement of photon prop-

erties to draw conclusions about the beam energy. In a circular accelerator for example

synchrotron radiation (SR) is emitted with a spectrum that strongly depends on the

beam energy. From a measurement of the high energetic part of the SR photon spectrum

it was possible to determine the beam energy to a level of about 10−3 [77]. In a linac

the angular distribution of OTR can be exploited. Due to the fact that the angular

distribution possesses characteristic maxima at angles 1/γ with γ the Lorentz factor,

from such a measurement the beam energy can be derived. Instead of using photons

emitted from the beam, laser photons can be backscattered at the particle beam (Comp-

ton backscattering). From the determination of the maximum energy of the scattered

Compton photons the beam energy could be determined to a level of ∆E/E ≈ 10−4 [78].

5.4. Beam synchronous timing . – A beam synchronous timing system (BST) has to

fulfill the following tasks: it has to generate and remotely distribute a phase reference,

(ii) it has to trigger fast and slow sub–systems, and it needs an interface to the control

system. Usually it is based on the following building blocks: (i) the reference or master

oscillator which defines the phase reference for all sub–systems, (ii) the master time–

base or event system generating e.g. trigger and bunch clock signals and delivering

experiment triggers, (iii) the distribution system (either coaxial cables or fiber optics)

Page 19: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 19

to the various sub–systems, and (iv) the control system interface. Typically there exist

two different timing levels in an accelerator environment, a fast timing which is at the

level of individual bunches and a slower one being at the level of a revolution clock in a

circular accelerator resp. on the level of bunch train repetition rate in a linac. A general

review about accelerator timing systems, based on the experience of 3rd generation light

sources, can be found in ref. [79].

The new 4th generation light sources however are capable to generate light pulses even

with a duration below 10 fs, and even shorter pulse durations are envisioned. In order

to take full advantage of these short light pulses, synchronization schemes on a sub–10 fs

scale or even below together with precise measurements and control of the electron bunch

arrival time are required. The required precision which is ideally a small fraction of the

photon pulse duration can only be achieved with laser–based synchronization schemes.

In the following a brief overview over the optical synchronization schemes together with

the sub–systems in use at modern high–gain FELs will be given.

The general concept of an optical synchronization system consists of a laser as timing

reference (master oscillator), a distribution unit to split the timing reference signals, and

actively length stabilized fiber–links that transport the reference signals to the different

remote locations. There are two general concepts to encode the timing information and

perform link stabilization, either to use a cw laser and maintain a constant number of

optical wavelength over the fiber length [80], or to encode the timing information in the

highly accurate repetition rate of a master laser oscillator (MLO) and maintain a constant

repetition rate of forward and reflected pulses over the fiber length. The latter method

which is in use at FLASH and has be designated for the E-XFEL will be presented in

the following. It has the advantage that due to the short laser pulses, non–linear optical

methods can be applied to measure timing changes of the optical pulse train.

The MLO is an erbium fiber laser operating in the soliton regime and producing

a train of sub-100 fs long laser pulses at a repetition rate of 216.7 MHz (sixth sub–

harmonic of the accelerator RF frequency of 1.3 GHz) at a central wavelength of 1500

nm. In this wavelength range, many fiber–optic components are available developed for

the telecommunication industry, and broadband transmission in dispersion compensated

optical fibers is possible. The fiber laser provides excellent stability on short time scales,

but in order to compensate for long time drifts (length change of the laser resonator

due to thermal expansion or contraction, microphonics) the laser frequency is locked

to a long–term stable RF oscillator [81]. However, the existing self–built laser system

has been replaced recently by a commercial available SESAM (semiconductor saturable

absorber mirror)–based erbium laser [82].

For the distribution of the MLO time reference signals to the different sub–systems

over length scales of 400 m (FLASH) up to 3.5 km (E-XFEL) via optical fiber links, active

stabilization is required in order to compensate for optical length changes of the fiber, e.g.

due to changes in the environment temperature. The measurement of the optical length is

performed such that the MLO pulses are transmitted through a dispersion compensated

optical fiber, partly reflected at a Faraday rotating mirror, and then sent back to the

MLO through the same fiber. There they are combined with the laser pulses coming

Page 20: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

20 G. Kube

directly from the laser in a balanced optical cross–correlator [83]. The temporal overlap

between both pulse trains is measured with sub-fs resolution and length changes are

compensated with a feedback loop acting on a piezo fiber stretcher (fine compensation)

and an optical delay stage (coarse compensation) [81].

In order to measure the electron bunch arrival time with respect to the time reference

signal, a Bunch Arrival Time Monitor (BAM) has been developed [84]. It consists of

a commercial available Mach–Zehnder–type electro–optical modulator (EOM) which is

driven by the transient voltage signal from a broadband electromagnetic pickup. The

arrival time of the reference laser pulse is adjusted such that it coincides with the zero–

crossing of the pickup for nominal bunch arrival time. Variations in the electron bunch

arrival time result in a variation of the modulation voltage experienced by the reference

laser pulse, and by detecting the amplitude of the pulse the arrival time can be derived.

The resolution of a BAM installed in FLASH was determined to be 3-5 fs for optimized

settings. Based on precise arrival–time information provided by the BAMs even a bunch

arrival time feedback has been implemented at FLASH which improved the arrival–time

stability from around 180 fs to 25 fs [85].

∗ ∗ ∗

Many thanks to my DESY colleagues C. Behrens, H. Delsim-Hashemi, D. Lipka, D.

Nolle, H. Schlarb, B. Steffen, M. Yan, and K. Wittenburg for their stimulating discussions

and their help in the preparation of this lecture, and also to H. Braun (PSI), M. Ferianis

(Synchrotrone Trieste), P. Forck (GSI), M. Minty (BNL), A.-S. Muller (KIT), U. Raich

(CERN), and V. Schlott (PSI) for their lectures about different task in the field of particle

beam diagnostics which I could use as illustrative material.

REFERENCES

[1] V. Ayvazyan et al., Eur. Phys. J. D 37 (2006) 297.[2] H. Schlarb et al., Proc. FEL’07, Novosibirsk, Russia, 2007, p.211.[3] M. Altarelli et al. (eds.), The European X–ray free–electron laser technical design report,

DESY 2006-097 (2006), http://xfel.desy.de.[4] R. Brinkmann, Proc. FEL’06, Berlin, Germany, 2006, p.24.[5] M.G. Minty and F. Zimmermann, Measurement and control of charged particle beams

(Springer Verlag, Berlin, Heidelberg, 2003).[6] P. Strehl, Beam instrumentation and diagnostics (Springer Verlag, Berlin, Heidelberg,

2006).[7] Proc. CERN Accelerator School on Beam Diagnostics, Dourdan, France, 2008, CERN-

2009-005.[8] V. Smaluk, Particle beam diagnostics for accelerators: Instruments and methods (VDM

Verlag Dr. Muller, 2009).[9] P. Forck, Lecture notes on beam instrumentation and diagnostics, Joint Universities

Accelerator School 2011, Archamps, January–March 2011.[10] R.C. Webber, Proc. of Beam Instrumentation Workshop BIW’94, Vancouver, Canada, AIP

Conf. Proc. 333, p.3, and FERMILAB-Conf-00-119, June 2000.[11] G. Gelato, in CERN-PE-ED 001-92 (revised Nov. 1994), p.185.

Page 21: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

Particle beam diagnostics and control 21

[12] J.-C. Denard, Proc. CERN Accelerator School, CERN-2009-05, p.141.[13] D. Belohrad, Proc. DIPAC’11, Hamburg, Germany, 2011, WEOC01.[14] T. Suwada et al., Proc. PAC’99, New York, USA, 1999, p.2238.[15] D. Lipka et al., Proc. DIPAC’11, Hamburg, Germany, 2011, WEOC03.[16] G.R. Lambertson, Proc. Anacapri, Isola di Capri, Italy 1988, Lecture Notes in Physics

(Springer Verlag, Berlin, Heidelberg, 1988), p.380.[17] R.E. Shafer, Proc. of Accelerator Instrumentation Workshop 1989, Upton (New York),

AIP Conf. Proc. 212, p.26.[18] E. Schulte, in CERN-PE-ED 001-92 (revised Nov. 1994), p.129.[19] S.R. Smith, Proc. of Beam Instrumentation Workshop BIW’96, Argonne (Illinois), AIP

Conf. Proc. 390, p.50.[20] P. Forck et al., Proc. CERN Accelerator School, CERN-2009-05, p.187.[21] M. Wendt, Proc. DIPAC’11, Hamburg, Germany, 2011, MOOC01.[22] G. Vismara, Proc. of Beam Instrumentation Workshop BIW’00, Cambridge (Maryland),

AIP Conf. Proc. 546, p.36.[23] B. Keil, Proc. DIPAC’09, Basel, Switzerland, 2009, p.275.[24] R. Lorenz, Proc. of Beam Instrumentation Workshop BIW’98, Stanford (California), AIP

Conf. Proc. 451, p.53.[25] D. Lipka, Proc. DIPAC’09, Basel, Switzerland, 2009, p.260.[26] Y. Honda et al., Phys. Rev. ST Accel. Beams 11 (2008) 062801.[27] P.M. Lapostolle, IEEE Trans. Nucl. Sci. 18 (1971) 1101.[28] J. Buon, Proc. CERN Accelerator School, CERN 94-01, p.89.[29] K. Flottmann, Phys. Rev. ST Accel. Beams 6 (2003) 034202.[30] M.P. Stockli, Proc. of Beam Instrumentation Workshop BIW’06, Batavia (Illinois), AIP

Conf. Proc. 868, p.25.[31] M. Zhang, Fermilab Report FERMILAB-TM-1988 (1996).[32] A. Cianchi et al., Proc. EPAC’04, Luzerne, Switzerland, 2004, p.2622.[33] D. Alesini et al., Proc. PAC’07, Albuquerque (New Mexico), USA, 2007, p.80.[34] S. Jolly et al., Proc. DIPAC’09, Basel, Switzerland, 2009, p.421.[35] N. Delerue et al., Proc. PAC’09, Vancouver, Canada, 2009, p.3597.[36] H. Hayano, Proc. LINAC’00, Monterey (California), USA, 2000, p.146.[37] I. Agapov et al., Phys. Rev. ST Accel. Beams 10 (2007) 112801.[38] G. Kube, Proc. DIPAC’07, Venice, Italy, 2007, p.6.[39] G. Kube, TESLA-FEL Report 2008-01, (2008).[40] M.C. Ross et al., Proc. of Beam Instrumentation Workshop BIW’02, Upton (New York),

AIP Conf. Proc. 648, p.237.[41] E. Chiadroni et al., Nucl. Instr. and Meth. B 266 (2008) 3789.[42] E. Chiadroni et al., Proc. DIPAC’09, Basel, Switzerland, 2009, p.151.[43] S. Wesch and B. Schmidt, Proc. DIPAC’11, Hamburg, Germany, 2011, WEOA01.[44] L.G. Sukhikh, Y. Popov, G. Kube et al., Proc. DIPAC’11, Hamburg, Germany, 2011,

WEOA02.[45] B. Walasek–Hohne and G. Kube, Proc. DIPAC’11, Hamburg, Germany, 2011, WEOB01.[46] G. Kube, C. Behrens and W. Lauth, Proc. IPAC’10, Kyoto, Japan, 2010, p.906.[47] M. Yan et al., Proc. DIPAC’11, Hamburg, Germany, 2011, TUPD59.[48] M. Zolotorev et al., Proc. PAC’03, Portland (Oregon), USA, 2003, p.2530.[49] P. Catravas et al., Phys. Rev. Lett. 82 (1999) 5261.[50] M. Uesaka et al., Nucl. Instr. and Meth. A 406 (1998) 371.[51] T. Watanabe et al., Nucl. Instr. and Meth. A 480 (2002) 315.[52] K. Kinoshita et al., Rev. Sci. Instrum. 58 (1987) 932.[53] C. Martinez, CLIC note 2000-020 (1999).

Page 22: Particle beam diagnostics and control · to the primary beam intensity. In case of lepton beams radiative losses may form an electromagnetic shower which has to be completely absorbed

22 G. Kube

[54] O. Grimm, Proc. PAC’07, Albuquerque (New Mexico), USA, 2007, p.2653.[55] O. Grimm and P. Schmuser, TESLA-FEL Report 2006-03 (2006).[56] B. Schmidt et al., Proc. EPAC’08, Genoa, Italy, 2008, p.130.[57] G. Doucas et al., Phys. Rev. ST Accel. Beams 9 (2006) 092801.[58] W.P. Leemans et al., Phys. Rev. Lett. 77 (1996) 4182.[59] I. Wilke et al., Phys. Rev. Lett. 88 (2002) 124801.[60] G. Berden et al., Phys. Rev. Lett. 93 (2004) 114802.[61] A.L. Cavalieri et al., Phys. Rev. Lett. 94 (2005) 114801.[62] B. Steffen et al., Phys. Rev. ST Accel. Beams 12 (2009) 032802.[63] G. A. Loew and O. H. Altenmueller, SLAC Technical Report SLAC-PUB-135 (1965).[64] P. Emma, J. Frisch, and P. Krejcik, LCLS Technical Report LCLS-TN-00-12 (2000).[65] D. Alesini et al., Nucl. Instr. and Meth. A 568 (2006) 488.[66] M. Rohrs et al., Proc. FEL’06, Berlin, Germany, 2006, p.300.[67] M. Rohrs et al., Proc. FEL’08, Gyeongju, Korea, 2008, p.90.[68] M. Rohrs et al., Proc. EPAC’06, Edinburgh, Scotland, 2006, p.77.[69] M. Stratakis et al., Phys. Rev. ST Accel. Beams 9 (2006) 112801.[70] M. Rohrs et al., Proc. PAC’07, Albuquerque (New Mexico), USA, 2007, p.104.[71] M. Rohrs et al., Proc. EPAC’06, Edinburgh, Scotland, 2006, p.80.[72] A. Fisher et al., Proc. of Beam Instrumentation Workshop BIW’98, Stanford, USA, AIP

Conf. Proc. 451, p.471.[73] Ch. Gerth, Proc. DIPAC’07, Venice, Italy, 2007, p.150.[74] T. Suwada et al., Phys. Rev. ST Accel. Beams 6 (2003) 032801.[75] A.-S. Muller, Proc. CERN Accelerator School, CERN-2009-05, p.427.[76] B. Dehning et al., Proc. EPAC’00, Vienna, Austria, 2000, p.436.[77] E. Tegler and G. Ulm, Nucl. Instr. and Meth. A 266 (1988) 185.[78] R. Klein et al., Nucl. Instr. and Meth. A 384 (1997) 293.[79] T. Korhonen, Proc. ICALEPCS’99, Trieste, Italy, 1999, p.167.[80] R. Wilcox et al., Optics Letters 34 (2009) 3050.[81] F. Lohl, TESLA-FEL Report 2009-08, (2009).[82] S. Schulz et al., Proc. FEL’10, Malmo, Sweden, 2010, p.581.[83] F. Lohl et al., Proc. PAC’07, Albuquerque (New Mexico), USA, 2007, p.3804.[84] F. Lohl , Proc. of Beam Instrumentation Workshop BIW’10, Santa Fe (New Mexico), USA,

2010, p.1.[85] F. Lohl et al., Phys. Rev. Lett. 104 (2010) 144801.