Sei sulla pagina 1di 76

Magnetic Lens Design for an Ultrafast Electron Diffraction Beamline

By

Thana Ghunaim

A thesis submitted to McGill University in partial fulfilment of the requirements of the


degree of Master of Science

Department of Physics

McGill University

April, 2009

Thana Ghunaim 2009


Table of Contents
Table of Contents ................................................................................................................................................... i

List of Figures ...................................................................................................................................................... iii

Abstract ................................................................................................................................................................. v

ABREG............................................................................................................................................................ vi

Acknowledgments ............................................................................................................................................... vii

Chapter 1. Introduction................................................................................................................................... 1

1.1. TRANSITION STATES ............................................................................................................................ 1

1.2. PRINCIPLES OF ULTRAFAST ELECTRON DIFFRACTION: ......................................................................... 3

1.3. SPACE-CHARGE EFFECTS ..................................................................................................................... 6

1.4. OVERCOMING SPACE-CHARGE: RADIO FREQUENCY ELECTRON PULSE COMPRESSION ......................... 7

Chapter 2. Fundamentals of Charged-Particle Beam Focusing ................................................................ 11

2.1. THEORY OF BEAM FOCUSING ............................................................................................................. 11

2.1.1. The Lorentz Force and the Paraxial Ray Equation ...................................................................... 11

2.1.2. Electron Beam Focusing by a Short Solenoid............................................................................... 12

2.2. BUSCH'S EQUATION FOR A WEAK LENS .............................................................................................. 16

2.3. THE GLASER MODEL .......................................................................................................................... 17

2.4. ON AXIS FIELD FOR A SOLENOID........................................................................................................ 18

2.5. SATURATION ...................................................................................................................................... 23

2.6. EMITTANCE ........................................................................................................................................ 25

2.6.1. Thermal Emittance........................................................................................................................ 26

Chapter 3. Space-Charge Effects on an Ultra-Fast Electron Diffraction Beam-line ............................... 28

3.1. RF-COMPRESSED ELECTRON BEAMLINE ............................................................................................ 28

3.1.1. Space-Charge Forces in a Spheroid/beam dynamics ................................................................... 29

3.2. SIMULATION OF SPACE-CHARGE FORCES ........................................................................................... 32

3.3. RESULTS AND DISCUSSION ................................................................................................................. 34

i
Chapter 4. Magnetic Lens Design................................................................................................................. 37

4.1. DESIGN REQUIREMENTS ..................................................................................................................... 37

4.2. BUSCHS THEOREM ............................................................................................................................ 38

4.3. MAGNETIC LENS SIMULATION ............................................................................................................ 39

4.4. HEAT EQUATION AND COOLING CONSIDERATIONS:............................................................................. 46

4.4.1. Power dissipation: ........................................................................................................................ 46

4.4.2. Heat Calculations: ........................................................................................................................ 47

4.5. CONCLUSIONS .................................................................................................................................... 51

References ........................................................................................................................................................... 53

APPENDIX ......................................................................................................................................................... 58

ii
List of Figures

Figure 1.1 Schematic of a general reaction potential energy profile ................................................................ 2

Figure 1.2 Schematic of pump probe experiment .............................................................................................. 4

Figure 1.3 Schematic of the UED experimental setup ...................................................................................... 5

Figure 1.4 Propagation of electron pulse through time. .................................................................................... 7

Figure 1.5 The inversion of the longitudinal phase-space distribution due to the oscillating field in the
cavity ...................................................................................................................................................................... 9

Figure 2.1 The cross-section of a typical magnetic lens ................................................................................... 12

Figure 2.2 Effect of magnetic field on the motion of an electron .................................................................... 13

Figure 2.3 Simulation results of the axial magnetic fields produced by a coil with a shield that has 0.5 cm
gap and the same coils with no shield ............................................................................................................... 19

Figure 2.4 The field produced by a lens of radius D, and gap s ...................................................................... 20

Figure 2.5 Focal length as a function of the factor a ........................................................................................ 22

Figure 2.6 Focal length as a function of NI....................................................................................................... 23

Figure 2.7 BH-curves of 9 ferromagnetic materials......................................................................................... 24

Figure 3.1 The proposed setup of beamline ...................................................................................................... 29

Figure 3.2 Superfish design of accelerator DC photogun................................................................................ 29

Figure 3.3 An electron bunch with an ellipsoidal distribution........................................................................ 30

Figure 3.4 Position-momentum distribution of electron bunch at various time steps .................................. 33

Figure 3.5 The RMS beam radius(x) and pulse length (z) as a function of position in the RF compressed
beamline............................................................................................................................................................... 35

Figure 3.6 Magnetic fields required for collimating and focusing the electron bunch ................................. 35

Figure 4.1 Schematic of the proposed UED setup ............................................................................................ 37

Figure 4.2 Design parameters of the lenses ...................................................................................................... 40

Figure 4.3 Fields of the designed lens without a shield and with a shield of 0.5cm gap ............................... 40

Figure 4.4 B(H) curve of the material used in the shield ................................................................................. 41

Figure 4.5 Saturation effect studies of a 1.5 cm thick shield ........................................................................... 42

Figure 4.6 Assembly of the lens ......................................................................................................................... 42

Figure 4.7 The cooling channel .......................................................................................................................... 43


iii
Figure 4.8 The bunch evolution in an optimized setup.................................................................................... 44

Figure 4.9 On-axis magnetic field needed to collimate the bunch .................................................................. 45

Figure 4.10 Measured axial magnetic fields of the machined lens.................................................................. 46

Figure 4.11 The discretization process .............................................................................................................. 48

Figure 4.12 Cross-section of a solenoid ............................................................................................................. 49

Figure 4.13 Thermal conductivity of the lens ................................................................................................... 50

Figure 4.14 Variation of temperature with respect to the thermal conductivity .......................................... 51

iv
Abstract

Ultrafast electron diffraction (UED) is an emerging technique for studies of transition state
and material structural dynamics at the atomic-level. UED is a time-resolved diffraction
technique that uses femtosecond laser pulses to initiate a chemical or material transformation
and an ultrashort electron pulse to follow the structural evolution that results through changes
in the electron diffraction pattern of the sample. Electron source design for UED experiments
is a challenging problem due to the high-charge density inside the ultrashort electron pulses
that are required for these experiments. Coulomb repulsion (space-charge) acts to rapidly
broaden these pulses both transversely and longitudinally. Controlling this space-charge
broadening through appropriate electron optical design is a key aspect of UED
instrumentation. In this thesis a novel electron beamline for UED that uses radio-frequency
electron pulse compression is described. Design criteria for the electron optical system
appropriate for this beamline is determined. The fundamentals of charged particle imaging
with magnetic fields are discussed, and lenses appropriate for use in the space-charge
dominated electron be are described.

v
Abreg

La mthode de diffraction ultra-rapide dlectron (DUE) est une technique mergente visant
les tudes des tats de transition et des dynamiques structurales de la matire au niveau
atomique. La DUE est une technique de diffraction au temps-rsolu qui utilise les pulsations
femto-second du laser pour dclencher une transformation chimique ou matrielle et une
pulsation ultra-courte dlectron pour suivre lvolution structurale qui rsulte des
changements dans le mode de diffraction dlectron de lchantillon. La conception des
sources dlectrons des expriences de la DUE est un dfi vue les densits leves des
charges de pulsations ultra-courtes dlectrons requises par ces expriences. Les rpulsions
de Coulomb (espace-charge) agissent en largissant ces pulsations transversalement et
longitudinalement. Controller cet largissement espace-charge travers une conception
optique adquate de llectron est un facteur crucial de linstrumentation de la DUE. Cette
thse prsente lutilisation de faisceau dlectron compression dlectron radiofrquence
dans la DUE. Les critres de la conception du system optique dlectron adquat pour le
faisceau sont aussi dfinis. Les fondements dimagerie particules charges et champs
magntiques sont abords, et les lentilles appropries utilises pour les lectrons domins par
lespace-charge sont dcrites.

vi
Acknowledgments

I would first like to thank the most important person during the course of my masters degree
at the physics department at McGill University: Prof. Bradley J. Siwick for his unwavering
support and excellent guidance throughout this study. His scientific knowledge, insight and
invaluable suggestions have been extremely encouraging to me and a great source of
inspiration, which are greatly appreciated.

I thank my parents for all their sacrifices for me throughout my life, and I thank my brothers
and sisters for their encouragement and support.

Special Thanks are attributed to all my colleagues, Robert Chatelain, Vance Morrison,
Christopher Godbout and Shona McGawn, who provided great assistance and constructive
suggestions during the course of this research.

I would like to thank all professors and students in the physics department who made my stay
at McGill full of friendship, learning and fun.

vii
Chapter 1. Introduction

Scientific principles and laws do not lie on the surface of nature. They are hidden,
and must be wrested from nature by an active and elaborate technique of inquiry.

~John Dewey, Reconstruction in Philosophy, 1920


.

1.1. Transition States

An important topic in condensed matter physics and chemistry is the study of the transition
states that separate two thermodynamically stables phases in a material phase transformation,
or stable reactant and product states in a chemical reaction. Such transformative processes
involve the rearrangement of atoms from one stable configuration to another. For the past few
decades the goal of understanding the transition state processes at a fundamental level was a
major focus of many scientists. Direct observation of these short lived states was deemed as
the holy grails of chemistry [1] and the study of these transition states was considered to be
the understanding of the Big Bang of biology and chemistry [2].

The transition state is the key configuration of a reaction system that exists at the top of the
energy barrier that the reactants must overcome to become the product. It is not strictly a
component of the reaction system, and it cannot be examined directly in the way that an
1
intermediate can, because it lasts no longer than the duration of a molecular collision. Figure
1.1 shows schematically a general one dimensional potential energy profile for such
processes.

Figure 1.1 Schematic of a general reaction potential energy profile. The two minima are the reactant and
product states. The maximum is the transition state or the activated complex that separates the two
thermodynamically stable states.

The transition state may have properties of its own, not reflected in those of the starting
materials or of the products and of the reaction, and so it is of vital importance in determining
the course of reaction [3]. It is regarded as important because the rate of a reaction is
proportional to the concentration of molecules reaching the transition state in a given period
of time. The concept of the transition state was first approached by M. Polanyi and M.G.
Evans [4] and by H. Eyring [5] both in 1935. Since then much work has been done, both
experimentally and theoretically [6]-[11]. Experimental approaches that are capable of
directly measuring the unstable transition state structures (or even short lived transient
intermediate structures) that connect thermodynamically stable states are at the forefront of
current efforts to understand chemical reaction and material phase transformation dynamics.

2
This thesis focuses on several aspects of one promising approach to studying the structural
dynamics associated with such processes; ultrafast electron diffraction

1.2. Principles of Ultrafast Electron Diffraction:

Most elementary motions of molecules during chemical changes, e.g. bond breaking and
bond making, happen on the ultrashort time scale that is picoseconds (ps) to femtoseconds
(fs). Pump-Probe experiments are one of the experimental ways by which information on
processes that occur on ultrashort (< 1 ns) timescales can be determined. The basis of such
experiments is to excite the sample with a laser pulse (pump) initiating a photo-induced
dynamical process. While the sample is changing, it is probed typically by second laser pulse
through which a transient absorption spectrum is obtained. The sample is probed at various
time delays so as to determine a complete picture of the entire process. Ultrafast diffraction
substitutes the probe laser pulse with either an ultrashort electron or x-ray pulse in order to
obtain a transient diffraction pattern in place of a transient absorption spectrum. A general
schematic of this process is shown in Figure 1.1. The time resolution of the experiment can
be expressed as [13]

exp eriment = ( pump ation + geometry )


12
2
+ probe
2
+ synchroniz
2 2
(1.1)

where pump and probe are the physical pulse durations of the pump and probe pulses at the
sample, respectively, synchronization is an experimental time jitter due to the uncertainty in
measuring the temporal overlap (t = 0) between pump and probe pulses at the sample, and
geometry is the contribution to temporal broadening due to the crossing angle ,Figure 1.2,
between the pump and probe pulses at the sample. The wavelength of the probe along with
the quality of the diffraction pattern determines the spatial resolution of this experiment,
since the density of relative atomic positions in the sample and the intensity distribution of
the diffraction pattern are related by a Fourier transform pair.

Diffraction experiments have proven themselves to be a convenient way to capture molecular


structure at the atomic level. At the turn of the twentieth century, the determination of
equilibrium 3-dimensional structures with atomic resolution, using diffraction, in systems
ranging from simple molecules to complex assemblies [14] was possible. Both X-ray and

3
electron diffraction is used widely. The general electron diffraction experiments done by
Mark and Weil [15] were the start of the use of electron diffraction.

Applied in a pump-probe fashion, time-resolved diffraction experiments can provide details


of atomic-level structural dynamics in molecules and materials that are not available with
other approaches. Electron diffraction has several attractive features in time-resolved
applications, including the large atomic scattering cross-section for electrons [16] and the
relative ease of producing bright electron beams through photoemission.

These features mean that ultrafast (time-resolved) electron diffraction experiments are table
top experiments and can be implemented using commercial ultrafast laser sources.

Figure 1.2 Schematic of pump probe experiment. A laser pulse excites the sample, and a delayed electron
pulse is used to probe the changes in the samples molecular structure

The UED technique operates in a stroboscopic manner, using properly timed sequences of
ultrafast laser pulses to start the reaction (pump) and ultrashort electron pulses to probe the
changes in the molecular sample. A phosphorus scintillator coupled to a CCD camera is used
to record the diffraction patterns. The probe electron pulses must be delayed with respect to
the arrival of the laser pump. By doing so, changes in the diffraction patterns as a function
of the time-delay between these two pulses can be determined. A schematic of this approach
is shown in Figure 1.2.
4
Currently the Siwick group is working on developing the first UED experiment with a time
resolution below 100 fs. The proposed setup is shown in Figure 1.3. This system is based on a
commercial (SpectraPhysics XP) Chirped Pulse Amplification (CPA) laser system that
produces laser pulses with an energy of 3 mJ, a pulse duration of 35 fs and center wavelength
of 800 nm. The pulse repetition rate of laser is 1 kHz.

Figure 1.3 Schematic of the UED experimental setup

The output of the laser is split into two beamlines. One beamline is the sample pump and
the other is the photocathode pump. Pulses following the pump-line first enter an Optical
Parametric Amplifier (OPA) for wavelength conversion. The wavelength must be tuned to
the appropriate absorption band of the sample in order to photoinitiate the process of interest.
The technical details of an OPA are discussed by Cerrullo and De Silvestri[17]

Pulses passing through the photocathode pump line are used to emit electrons from a metal
photocathode through photoemission, forming the electron pulse shown schematically in Fig.
1.2. An electron is liberated for each photons absorbed by the cathode, provided that the
photon energy is larger than the work function of the photocathode. Thus, the laser pulses
5
(initially 800 nm) are frequency tripled to 266 nm (4.66 eV photon energy) using standard
nonlinear optical conversion strategies before reaching the gold photocathode. Typical
photoelectron energy speads are ~0.6 eV for metal photocathodes. An accelerating field must
be present to aid in the electron extraction from the metal surface of the photocathode. This
can be done by having the cathode held at negative potential with respect to a grounded
anode. This acceleration field is used to control the electrons kinetic energy (electrons of 100
KeV energy have a deBroglie wavelength of 3.7 pm). These electrons then propagate in
vacuum to the sample where they scatter to form a diffraction pattern at the detector. The
ejected electrons start with a temporal duration and transverse area similar to that of the laser
pulse but as the electrons propagate towards to the sample it will suffer from broadening in
all directions due to Coulomb forces. Magnetic lenses can be used to focus the beam
transversely. Design of a set of magnetic lenses to control the transverse electron beam
dimensions of the Siwick group ultrafast electron beamline is the main subject of this thesis.

1.3. Space-Charge Effects

Because electrons are charged particles, confining a large number of electrons in a small area
(i.e. a pulse) to achieve high brightness electron pulses suitable for UED experiments has its
drawbacks. As a result of Coulomb forces, the electrons repel each other causing beam
broadening effects, commonly known as coulomb explosion. The temporal broadening
(broadening of the pulse along the propagation direction) typically causes the pulse duration
to rapidly deteriorate from femtosecond to picoseconds while the radial spreading will
degrade the electron beam's optical quality.

To prevent such broadening in some applications, electrons are accelerated to relativistic


speeds during very short times [18]. At relativistic speeds the bunch explosion is slowed
down in all directions (transverse and longitudinal) and by this the Coulomb repulsion is
effectively damped. The problem with this approach for UED applications is that electrons
with relativistic energies have short de Broglie wave length, and reduced elastic scattering
cross-sections. Another approach to solving this space charge problem is to use electron
beams that consist of single electron charge bunches [19] . Since there is only one electron or
so in each bunch, this would eliminate the Coulomb forces which cause the broadening.
Using these techniques means that the sample should be pumped and probed much more
times than other approaches to get a useful diffraction pattern. This approach requires an
6
enormous number of repeated (and reproducible) laser exposures, limiting the technique to
truly reversible processes.

Simulations of these space charge effects for an electron pulse with 10000 electrons were
performed by Siwick et al [13]. The results, seen in Figure 1.4, showed that the duration of
the electron pulse becomes more than 500 fs after almost 1 ns propagation time, which is
equivalent to 10 cm propagation distance for 30 keV electrons. These results suggest another
approach. In order to avoid the space-charge induced broadening, the propagation distance
should be minimized. To minimize the propagation distance, compact electron sources are
designed and used. These sources provide electron bunches that consist of several thousand
electrons ~0.001 pC charge and time resolution of ~400 fs at 100 KeV kinetic energy [2].

Figure 1.4 Propagation of electron pulse through time. Results of the N-body simulation for a constant beam
radius (N=510 000, r=5200 m, 0 =50 and 500 fs). (a) Initial pulse duration and propagation time
dependence on the electron pulse duration in the drift region. The FWHM pulse duration is defined as the
pulse length containing 76% of the electrons. (b) The rate of temporal broadening vs propagation time in the
drift region [13].

1.4. Overcoming Space-Charge: Radio Frequency electron pulse


compression

Through the development of compact electron sources, temporal resolution has improved by
several orders of magnitude over the last decade and has reached sub picosecond. UED is
now a well demonstrated technique for studying simple structural changes, such as the
femtosecond laser-induced melting of Aluminum [20], the study of protein structures [21]
7
and transitions of molecular structures [22], etc. To capture the vibrational motion as well as
the bond making and breaking, sub picosecond resolution isn't enough since these processes
occur at a time scale of ~100 fs [2]. In addition, the diffraction data quality is strongly
limited by the relatively small number of electrons per pulse that can be used in state-of-the-
art electron sources.

To scale UED to more challenging structural problems it is important to determine alternative


approaches to handling the space charge problem that would allow for many more electrons
per pulse to be used while simultaneously reducing the electron pulse duration. It is predicted
by theory that 100 pC bunches shorter than 100 fs, and energy of 5MeV will generate Tera-
Hertz radiation of more than 5.7MV/cm. For single-shot, time-resolved electron diffraction
bunches are required of typically 0.1 pC charge, 200 m radius, 100 fs duration, and 100 keV
energy. Electron pulses containing more than 106 electrons with durations below 100 fs are
desirable, all in a high-brightness beam (long transverse coherence length) suitable for
diffraction experiments.

A possible way to create such an ideal beam is by initially producing bunches that have linear
space charge forces. It is known that a uniformly filled ellipsoidal electron bunch has such
properties. Such beams were considered to be unattainable in practice, until Luiten et. al. [23]
outlined an easily implemented approach to attain them through photoemission. Since the
internal space-charge fields are linear in these bunches, space charge broadening occurs in a
controlled and predictable way. The pulses naturally develop a phase space distribution with
a perfect axial velocity-position correlation. Thus, the coulomb expansion in both transverse
and longitudinal directions can be reversed by applying external electric fields. [24].
Longitudinally, the bunch can be compressed by RF fields, while transversely it can be
focused and defocused using a magnetic solenoid lenses.

8
Figure 1.5 The inversion of the longitudinal phase-space distribution due to the oscillating field in the cavity.
(a) The bunch suffers from expansion as it leaves the photocathode after entering the RF cavity, the bunch
momentum is inverted and the bunch is compressed as seen in b), c) The time focus is at the beam waist, d)
The bunch expands in the longitudinal direction

Space-charge forces generate a bunch in which the leading electrons have higher momentum
than those tailing the bunch. To recompress such a bunch, the momentum distribution of the
electrons needs to be inverted. The idea is to apply a time-dependent electric field to the
bunch such that electrons at the front are slowed down while those at the back are sped up.
This will result in the bunch undergoing ballistic recompression. The time-focus is the
position where the bunch reaches a minimum length. Figure 1.5 shows a schematic of the
momentum inversion that can be obtained by injecting a velocity-position correlated electron
bunch into a radiofrequency cavity at the appropriate zero-crossing of the RF cycle. After
interaction with the cavity the bunch will begin to temporally recompress.

The main subject of this thesis is the magnetic lens design for transverse beam focusing in a
radio-frequency compressed ultrafast electron diffraction beamline. The basics of this RF
compression approach were described above. An outline of the remainder of the thesis is as
follows. Chapter 2 will discuss the fundamentals of transverse electron beam focusing using
magnetic fields. Chapter 3 will investigate the electron optical system (i.e. magnetic lens
focal lengths) required to achieve appropriate beam properties under the specific conditions
of the RF compressed beamline being constructed by the Siwick group. The beam is strongly
9
space charge dominated in this scheme and simple ray tracing approaches to the design of the
electron optics are inappropriate. Finally, the actual design of magnetic lenses suitable for
this application will be discussed in chapter 4.

10
Chapter 2. Fundamentals of Charged-Particle Beam
Focusing

In the year 1896 John A. Fleming noticed that a short current carrying coil had a
concentrating action on electrons passing axially, but it was not until 1926 that the general
theory of charged particle image formation was formulated by Hans Busch. Promoted by de
Broglies hypothesis, made two years earlier, he laid the basis of electron optical theory and
derived the conditions in which the electric and magnetic fields - separately or together -
could exert a focusing effect on the trajectories similar to that of a glass lens on light rays and
thus act as electron lenses. Electron lenses have since been used as means of electron beam
focusing.

2.1. Theory of Beam Focusing

2.1.1. The Lorentz Force and the Paraxial Ray Equation


A charged particle passing through an electromagnetic field experiences a Lorentz force. This
is defined by,
r r r r
F = eE + e(v B) (2.1.1)

r r r
The vectors E and B are the electrical and magnetic field vectors, respectively, and v is the
velocity vector of the particle. Such forces can be applied to aid in guiding a charged particle
beam along a well defined path and to help focus the beam, confining it to a narrow vicinity
of the optical axis. Both the electrical and the magnetic components of the Lorentz force can
have an equally important effect on guiding and focusing the beam. Relativistic particles (v
c) need stronger forces to focus them. For such particles, a 1 Tesla magnetic field has the
equivalent effect of a 300MV/m electric field [25]. Considering these numbers, such
magnetic fields are easier to produce compared to electric fields. Thus magnetic lenses are
the most common mean for high energy charged particle beam focusing. Electrostatic
focusing of electrons require potential difference of similar magnitude to the beam energy,
which is problematic at high energies

11
Figure 2.1 The cross-section of a typical magnetic lens. An iron shielding is used to produce stronger and
more focused magnetic fields

The easiest and most convenient way to produce strong magnetic fields is by passing current
through a solenoid, where the magnetic field produced on axis is not only proportional to the
current and the number of turns in the solenoid but also depends on its geometry. In a
solenoid, a large field is produced parallel to the axis of the solenoid. The solenoidal fields
can be confined to a narrower axial region by adding a shield around the coil made of a
material with high magnetic permeability (referred to in what follows as soft-iron), Figure
2.1 The shield helps concentrate the field to a narrower region of the optical axis and can
increase the maximum axial field significantly. as will be discuses in section 2.4

2.1.2. Electron Beam Focusing by a Short Solenoid


Following the derivations C. E. Hall, Introduction to Electron Microscopy, [26], we
demonstrate that a short solenoid has a focusing effect on a paraxial electron beam.
r
Consider an electron passing through an axially symmetric magnetic field. B = Br r + Bz z (i.e.
B =0) in cylindrical coordinates

It is convenient express the Lorentz force (Eq. 2.1.1) in cylindrical coordinates as follows:

12
ir i iz
r
F = er& er& ez& (2.1.3)
Br 0 Bz

Where ir , i and iz are unit vectors, the dots indicate differentiation with respect to time and
the | | denotes a determinant.

We notice that the azimuthal component of the electron velocity combines with the axial
magnetic field exerting a radial force, Fr,

Fr = er&Bz (2.1.4)

on the electrons bringing them closer to axis of symmetry.

Figure 2.2 Effect of magnetic field on the motion of an electron. As a result of the radial and centrifugal
forces acting on the electron, the electron takes a spiralling path along the optical axis

The electron will also experience an angular force, F, perpendicular to is motion and to the
radial component of the magnetic field

F = ez&Br + er&B z (2.1.5)

13
causing the electron to take a circular path. Combining both these forces will cause the
electron to move in a spiral motion about the optic axis (Figure 2.2).

In order to determine the electron's path in the magnetic field we note that from Newton's
second law the radial force, Fr, can also be expressed as the net result of the difference
between the inward force m&r& and the centrifugal force mr 2& ,

Fr = m&r& mr 2& (2.1.6)

Equations (2.1.4) and (2.1.6) can be combined to obtain:

m&r& = er&B z + m& 2 r (2.1.7)

In order to simplify the above, it is useful to find an expression relating & to Bz . To do so,

consider the time change of the angular momentum mrv = mr 2& 2 is equal to the moment of
the force with respect to the axis of rotation:

d
(mr 2&) = rF , (2.1.8)
dt

this along with equation (2.1.5) can be expressed as

d
mr 2& = err&B z erBr z& (2.1.9)
dt
r r
Using Laplace equation to solve for both radial and axial fields, since B = 0 , one can find
that

r
Br = Bz , (2.1.10)
2 z

Substitute for Br in equation (2.1.9), we get

d d e
(mr 2&) = r 2 Bz (2.1.11)
dt dt 2

this equation can be easily integrated to give

14
e 2
mr 2& = r Bz + C (2.1.12)
2

The integration constant C, is not zero for all rays, and a further integration is not possible in
general without a relation between r and z. For simplicity we consider meridional rays
passing through a field free region before they enter the lens fields. In this case, the initial
conditions are & = 0 and Bz = 0 and the value of the constant C is found to be zero (the rays
for which C is nonzero are skew rays in field free space, and C is related to the angular
momentum about the axis in such a region). We now can express the angular velocity & as

e
& = Bz (2.1.13)
2m

This indicates the angular velocity depends only on the axial component of the magnetic
field.

If we substitute equation (2.1.13) in (2.1.7), we find that the radial acceleration is,

e2 2
&r& = B r
2 z
(2.1.14)
4m

This shows that electrons are directed towards the optical axis, and the forces that drive the
electrons towards the axis are directly proportional to the distance r from the axis. This is
the principle of a focusing field; i.e. solenoidal fields act as a charged particle lens.

It is easier to handle Eq. (2.1.4.) if it is put into a form that depends on z instead of t. We can
use the expression of the axial velocity from the kinetic energy equation,

1 2
mv z = eV (2.1.15)
2

Rearranging equation (2.1.15), we find that

dz 2eV
vz = = , (2.1.16)
dt m

where V is the accelerating potential. So equation (2.1.14) becomes

15
e 2
r = Bz r (2.1.17)
8mV

dr
Where r =
dz

It is important to note that the force is proportional to the square of the magnetic field. This
means that if we reverse the direction of the magnetic field by reversing the current, the
direction of the force does not change. The force on the electron is directed toward the optical
axis regardless of the field direction; i.e. only converging lenses are possible using magnetic
fields.

2.2. Busch's Equation for a Weak lens

An approximation for the focal length of a magnetic lens is easily found if the focal length is
long compared with the axial extent of the field. Such a lens is also called a weak (or thin)
lens. Under such conditions it can be assumed that the radial distance r is constant within the
lens to a fair approximation. Then by integrating equation (2.1.17)


e
ri ro = rB z dz
2
(2.2.1)
8mV

If the rays in the object space (r0) are parallel to the axis, then ro = 0. Since we assumed that
r = r0 is constant through the field we have,


ri e 1
= B dz =
2
z (2.2.2)
ro 8mV
f

and the focal length of a thin lens can be expressed as


1 e
= B dz
2
z . (2.2.3)
f 8mV

Equation 2.2.3 is valid for low energy electrons, if the voltage is raised so as the electrons
have relativistic speed, then the value of V should be corrected:

16
e
(
Vr = V 1 + mo c 2V = V 1 + 0.987 10 6 V ) (2.2.4)
2

The variation of the magnetic field intensity along the axis Bz(z) is a function of the magnetic
lens geometry and design. The focal length of the lens depends very much on the axial
distribution of the field. In particular, note that the focal length depends on the integral of Bz
squared. This means that for a given on-axis integrated flux, the more strongly peaked the on
axis field distribution the shorter the focal length.

2.3. The Glaser Model

Many attempts have been made to compute the magnetic lens properties analytically, by
assuming a hypothetical field distribution along the axis of the lens for which the integral in
Eq. 2.2.3 can be solved exactly. Some of the most commonly used hypothetical fields are the
ones proposed by Ramberg and Glaser.

Ramberg suggested that B(z ) = Bo sec h 2 bz , where b = 2.63 / D and D is the bore
diameter[27], [28]. This field was assumed for a lens with an indefinitely small pole piece
gap.

On the other hand Glaser assumed

Bo
B( z ) = (2.3.1)
1 + ( z a) 2

where is a parameter related to the FWHM of the on axis field distribution. The field
distribution assumed that the pole pieces of the lens are close to saturation [27], [28]. Glaser's
bell shaped field is very convenient, not merely because its a good approximation to the field
in many common types of lenses and can be fitted to many practical field distributions (see
Figure 2.3), but also because the trajectory equations can be solved explicitly in terms of
circular functions giving a complete qualitative analysis of the lens.

Then focal length according to Glaser's field is

1 1
= sin (2.3.2)
f a K 2 +1

17
where K 2 is the excitation parameter

eBo2
K2 = (2.3.3)
8mVr

The smallest focal length is achieved when


sin =1 (2.3.4)
K 2 +1

or when K 2 = 3 , hence according to Glasers hypothetical field the minimum focal length is
equal to a.

2.4. On Axis Field for a Solenoid

For Thick solenoids without a shield the axial magnetic field can be found by integrating the
magnetic fields of a single loop over length and thickness of the solenoid [29],

+ [ 2 + ( + c) 2 ] + [ 2 + (c ) 2 ]
1/ 2 1/ 2
o
B(0, z ) = NI ln ( + c) + ( c )
( 1) 1 + [1 + ( + c) 2 ] 1 + [1 + (c ) 2 ]
1/ 2 1/ 2

(2.4.1)

rout L z
Where N is the number of turns, I is the current carried by the wire = , = , c=
rin 2rin rin
and rout ,r in and L are the outer radius, inner radius and the length of the solenoid respectively.

The above expression no longer holds when a shield is introduced to the solenoid.
The benefit of adding a shield is to dramatically increase the maximum field produced, Bo
(see Figure 2.3). Since the focal length, f , is inversely proportional to the integral of the
magnetic field squared (Eq. 2.2.3) one can produce stronger lenses with much lower drive
currents. The magnetic field produced depends on the geometry of the shield. The simplest
illustration of this can be understood by analyzing the C-magnet geometry shown in Figure
2.4. This picture shows a transverse cross-section of a solenoidal magnet, with a dipole field
excited by the coils that carry a certain total current with an inner radius D, and a gap of
length s. The current is usually contained in N windings so that one often writes the total
current as NI, the number of windings times the current I in each, resulting in a total indicated

18
by the unit Ampere-turns. The planar surfaces at the bottom and the top of the gap set the
dipole field inside the gap.

Figure 2.3 Simulation results of the axial magnetic fields produced by a coil with a shield that has 0.5 cm gap
and the same coils with no shield. Both coils where driven by a 600 A/cm2 current density. Notice that the
shielded magnetic coil produces an axial field that is much stronger and more focused than the same coil
with no shield

19
Figure 2.4 The field produced by a lens of radius D, and gap s

The relationship between the total excitation current and the field in the shield gap can be
simply estimated by use of the integral form of ampere's law,
r
H d l = I enc (2.4.2)

for the geometry shown in Figure 2.4,this can be written approximately as the sum of two
components,

Bp 1 r r
o
s+ dl = I enc = NI
B
shield
(2.4.3)

r
where B p is the dipole field strength in the gap. Because the magnitude of B inside the

shield is not too different from B p , while the permeability is orders of magnitude larger in

the iron, the second term of the left hand side is typically ignored, and we have

NI
B p = o (2.4.4)
s

To determine the focal length of such a shielded lens we must now determine a relationship
between the maximum magnetic field on the optical axis, Bo, and the dipole field in the gap,

20
Bp. In order to do so it is convenient to introduce a function, f(z), which rises to a maximum
height of unity,

B( z )
f ( z) = , (2.4.5)
Bo

The function f(z) determines the axial extend of the magnetic field and will be seen to depend
on the lens geometry. A straightforward application of Amperes law around a closed loop
involving the entire optical axis, but the same enclosed excitation current as (2.3.6) then
provides the connection between Bo and Bp as follow:


sB p = Bo f ( z )dz

(2.4.6)

From Glaser's hypothetical field, we can say that

1
f ( z) = (2.4.7)
1+ z2 a2

So, combining (2.39) and (2.37) we obtain the result

NI NI
Bo = o = o (2.4.8)

a
(1 + z a 2 ) 1 dz
2

21
Figure 2.5 Focal length as a function of the factor a. The graph shows the focal length of a lens with 250
turns and the driving current passing trough the wire is 8A acting on electrons with 100kV accelerating
potential.

The axial flux density distribution was defined in terms of a maximum flux density Bo, on the
axis and the half width 2a of the field distribution, that is, the distance between the points
where the field decreases to B/2. The dependence of the focal length on the parameter a is
shown in Fig. 2.5. Durandeau and Fert(1956) found that at flux densities in the iron not
exceeding 1.26 T the half width of the axial field distribution could be related to pole piece
geometry S and D by the relation [30], [31]

2a = 0.97 s 2 + 0.45D 2 (2.4.9)

However, if the flux density in the shield exceeds saturation limit, the above approximation
for the half width at full maximum is not longer valid and should be adjusted accordingly.
Appropriate expressions can be found in several papers [32], [27].

The focal length according to Glasers assumption depends on NI and the geometry of the
lens. A stronger lens can be designed by providing more current-turns, see Figure 2.6.

22
Figure 2.6 Focal length as a function of NI. The focal length depends on the number of turns and current
passing through the wire. The stronger the current the shorter the focal length, The graph shows the focal
length of a lens with 254 turns, an inner radius of 3.6 cm and a length of 1.6 cm, the acceleration potential is
100 KV

2.5. Saturation

Saturation is a phenomenon limited to shielded magnetic solenoids. Air core magnetic


solenoids do not saturate, but on the other hand they do not produce as nearly as much
magnetic flux as shielded ones for the same number and wire turns and current. Saturation
limits the performance of the ferromagnetic material used in shield, which is commonly
known as soft-iron. Saturation can be understood qualitatively since the soft-iron aids in the
flow of the B-field by aligning its microscopic dipole moments to the field, and relaying the
field without causing loss of the integrated H- field. Once all of the dipole moments are
optimally aligned, the B-field in the iron does not increase with additional H in the same
proportion as for low values. This means that the effective incremental permeability,

dB
= (2.5.1)
dH

23
falls from values as high as 104o when H << Hsat , to a value that descends ultimately to near
o for H >> Hsat , the fields are no longer well confined in the shield above saturation
[33].Technically, above saturation the B-field continues to increase, but at the paramagnetic
rate, which is 3 orders of magnitude smaller than the ferromagnetic rate seen below
saturation[34].

Figure 2.7 BH-curves of 9 ferromagnetic materials, 1.Sheet steel, 2.Silicon steel, 3.Cast steel, 4.Tungsten
steel, 5.Magnet steel, 6.Cast iron, 7.Nickel, 8.Cobalt, 9.Magnetite, for all materials permeability (slope of BH-
cruve) increases with H to a maximum, Eventually, however, all the magnetic domains align with the applied
field, and the curve flattens out as the ferromagnet becomes magnetically saturated. [35]

This implies that the shield used will not behave the same at large excitations. In particular, at
certain level of the magnetic field all atoms are aligned and further increase in magnetic flux
causes no change, as illustrated by typical B(H)curve.

This is not easily exceeded in ferromagnetic material, which is known as the saturation field.

24
The effect of saturation in the pole piece causes a field broadening effect and is equivalent to
an apparent increase in lens dimensions apart from a loss of magnetic motive
force(M.M.F.)in the iron which depends on the design of the magnetic circuit as a whole[36],
[37]. These saturation effects depend in general on shield geometry in a complex way.
Particularly the best approach to determining saturation levels for particular lens geometry is
to numerically solve the field equations using available simulation packages, such as
SuperFish etc. This is discussed further in Chapter 4.

2.6. Emittance

It is important to have a measure of the quality of the beam in terms of their ability to be
transported over long distance and to be focused into small space with minimum divergence.
A beam of particles can be characterized in detail by its density in the six dimensional phase
space (x, Px, y, Py, z, Pz), where P is the canonical momentum. The extent of the beam in
phase space, termed its emittance, is constant in time under ideal conditions [38].

Usually, the six dimensional description is split into two dimensional sub-states (x, Px),
(y,Py), and (z, Pz), where z is often taken as the optic axis of the beam. The emittance is
defined in terms of the area occupied in the two dimensional spaces. If the area of the beam
in the (x, Px) space is x then the normalized emittance is defined as [39].

nx x mo c (2.6.1)

Commonly Px is replaced by x' = dx/dz = vx/vz. The emittance can then be expressed
geometrically in terms of the area Ax occupied by the beam in the trace space (x,x').

Ax
x (2.6.2)

From the definitions it is seen that nx = z x where z = v z c and is the Lorentz factor

= 1 1 2 . Similar definitions apply to the y and z spaces. These definitions assume a


hard-edged distribution. However, they are easily generalized to more realistic distribution. In
hard-edged bunches the area is generally taken to be that of the smallest ellipse enclosing the
particles. In phase space (x, Px), the semi-axes of the ellipse can be estimated as x and px

25
respectively, where (a a ) 2 is the variance of quantity a. The RMS-emittance is can
then be expressed, and is

x = x px (2.6.3)

A more generalized definition of emittance which is applicable to general beam distribution


without an easily defined shape, this is based on the concept of the equivalent perfect
beam, and may be shown to be[38]

x = x 2 x 2 xx 2 (2.6.4)

A low emittance particle beam is a beam where the particles are confined to a small distance
and have nearly the same momentum. Smaller emittance is usually preferred. Beam
brightness in terms of emittance is given by:

Q
B= (2.6.5)
mc n , x n , y n , z

2.6.1. Thermal Emittance


Thermal emittance is the measure of the electron distribution in the transverse position-
momentum phase space at its generation point. The thermal emittance of an electron beam
generated by a photoemission imposes a lower limit for the normalized emittance on the
electron beams used in UED experiments.

The normalized thermal emittance depends on spot size, momentum distribution and angular
distribution of the emitted electrons, where the angular distribution and energy are functions
of the cathode material and photon energy. At the source , is zero so the normalized
emittance can therefore be written as [40], [41]

,
, (2.6.1)

It is important to note that the thermal emittance is a quantity influenced not only by the
photocathode material properties, but also the parameters of cathode laser and accelerating
field amplitude.

26
For electrons emitted with initial kinetic energy EKin, the emittance can be expressed as


, (2.6.2)

Where the initial kinetic energy of the emitted electrons is basically the difference in the
energy of the incident photon h and the effective work function e

(2.6.3)

Substituting equation (2.6.3) into equation(2.6.2), thermal emittance becomes


, (2.6.4)

The presence of a strong electric field at the cathode will significantly lower the work
function of the photo-emissive material, and the effective work function will be

(2.6.5)

Here is the lowering of the work function by the electric field Eacc and is given by [40].

(2.6.6)

Where o is the permittivity of free space. The lowering of the work function is also known as
the Schottky effect.

The thermal emittance of electrons emitted with an electric field of 100keV from a copper
photocathode is 0.58 mm-mrad,

27
Chapter 3. Space-Charge Effects on an Ultra-Fast Electron
Diffraction Beam-line

In the previous chapter the essential fundamentals of charged particle imaging using magnetic
fields has been presented. In this chapter the particular electron beamline that is being
implemented as a part of the Siwick groups ultrafast electron apparatus is described. With
the aid of state of the art particle tracking simulations, the progression of an electron bunch in
this apparatus is simulated from its point of birth at the cathode, through the process of space-
charge blow-out, and finally to its recompression. The effects of all externally applied fields
from different elements along the beamline and internally generated electron bunch space
charge fields are taken into consideration. These particle tracking simulations define the
design requirements for the magnetic lenses in the beamline.

3.1. RF-Compressed Electron Beamline

The proposed setup of the RF-Compressed beamline is shown in Figure 3.1.The setup
consists of a DC photogun, two solenoidal magnetic lenses S1 and S2, and an RF cavity. Two
solenoids are used to control the transverse dimension. The first solenoid S1 collimates the
bunch after photoemission and the second solenoid S2 is used to transversely focus the bunch
to the required spotsize. The field strength and the position of the first solenoid S1 is
determined during the simulations.

The electron pulse is accelerated by the DC photoelectron accelerator. The initial bunch
dimensions are determined by the focus and duration of the laser pulse that is used to emit the
electrons. Initial space charge (coulomb repulsion) acts to broaden pulse both transversely
and longitudinally. The other elements in the beamline are to control those effects; the
Solenoids are used for transverse focusing of the beam, while the RF cavity is used for
longitudinal (temporal) focusing.

The goal of this chapter is to determine the focal length requirements for S1 and S2 in this
beamline under conditions where the internal space charge forces of the bunch determine the
electron beam properties i.e. simple ray-tracing is completely inadequate. These are the

28
conditions in which the beamline is expected to operate during UED experiments. Discussion
of the temporal focusing properties of the RF cavity is deferred to section 3.3.

Figure 3.1 The proposed setup of beamline consists of the following elements: (a) DC accelerator, (b)
Collimating magnetic lens S1, (c) Focusing magnetic lens S2, (d) RF cavity

Figure 3.2 SuperFish design of accelerator DC photogun. Purple lines are equipotential[42]

3.1.1. Space-Charge Forces in a Spheroid/beam dynamics


In this section we consider the effect of space charge (coulomb repulsion) on a photoelectron
pulse within a simple analytical model. Conditions equivalent to this model can actually be
achieved in practise.

29
The density distribution of particles over the two transverse coordinates, x and y, are called
the beam profile. Beam profiles can be derived from the two-dimensional subspaces of the
six dimensional phase space, occupied by the ensemble of the particles.

Figure 3.3 An electron bunch with an ellipsoidal distribution can maintain linear space charge distribution

To study the space charge effect with respect to measured beam profiles, one has to consider

the action of the internal field components Ex, Ey, Ez (or Er, Ez, since r = x 2 + y 2 ) caused by
the space charge as well as the external fields acting on the electrons.

Uniform three-dimensional (3D) ellipsoidal distributions (Figure 3.3) have particularly


simple internal space-charge fields that are governed by Poisson equation:
r r r
2 (r ) = k (r ) (3.1.1)

A rotationally symmetrical ellipsoid with homogeneous charge distribution is characterized


by a constant charge density (x, y, z) inside a volume that is bound by the following
equation:

x2 y2 z2
+ + =1 (3.1.2)
R 2 R 2 L2

30
The charge distribution is comparable to a homogenously charged sphere and is expressed as
[43]:

3 N e
o = (3.1.3)
4R 2 L

Where N is the number of electrons in the bunch, is their charge, e = 1.602x10-19 is the
elementary charge, R is the maximum radius and L the half of the total bunch length, see
Figure 3.3

For an ellipsoid the with uniform charge distribution, the potential is:

o
M (r , z ) =
2 o
(MR 2 M r r 2 M z z 2 ) (3.1.4)

Thus the electric field inside the ellipsoid can be characterized as:

r
E = (E x , E y , E z ) = o ( M x x, M y y, M z Z ), (3.1.5)
o

Where Mx, My, and Mz are the following form factors, which add up to unity and depends on
the ratios of the primary axes of the ellipsoid and are given by [42], [38]

1+
Mz = ( arctan ) (3.1.6)
3

1
Mx = My = (1 M z ) (3.1.7)
2

where = R 2 L2 1 is the eccentricity of a spheroid. The potential is defined such that it


r r
equals zero if o = 0 , in other words if R,L . Using E = V , the space-charge field can
be found to be:

o
Er (r ) = Mrr (3.1.8)
o

o
Ez ( z) = M z
o z

31
The most important feature of the above expressions is that the space-charge fields inside a
uniformly filled ellipsoid are linear function of position. These linear space-charge fields give
rise to particle velocities which are also linear functions of position.

It needs to be emphasized that due to the space charge effect, the simple paraxial ray equation
alone is insufficient to determine the beam properties. In the simple case of a uniformly filled
ellipsoid, space-charge fields act as a negative lens, whose properties depend on the charge
density (or position in the beamline) in a non trivial way. In order to know how exactly the
space-charge affect the beam profile, particle tracking simulations including these effects
need to be performed.

3.2. Simulation of Space-Charge Forces

In this section we seek to define the transverse focusing requirements for the RF compressed
electron beamline being developed in the Siwick group. In particular, we seek to determine
the on axis magnetic fields required in order to longitudinally focus the space charge
dominated beams present in this system.

The dynamics of a large number of charged particles in electromagnetic fields can be


simulated and studied by General Particle Tracer (GPT), a powerful tool used for designing
accelerators and beam lines. GPT has a built in fifth order Runge-Kutta method which solves
the 3-D relativistic equation of motion for large number of charged particles in externally
applied electromagnetic fields. The position of the beam line components and the external
fields can be easily manipulated [44]

The simulations where performed to determine the position and field strength of the
collimating solenoid S1. The position of the RF-cavity was determined according to the
changes in the bunch. The simulations were also done for various values of the RF field
amplitudes so as to optimize the results and to get the required results at the temporal
focusing. Both the position and the field strength of the second magnetic solenoid S2 were
varied to obtain electron bunches with 0.1pC charge with duration less than 100fs and a
bunch radius less than 200 m at the temporal focus.

32
Figure 3.4 Position-momentum distribution of electron bunch at various time steps. As the beam propagates,
its velocity momentum is reversed resulting in a linear chirp. z is the velocity normalized to the speed of
light, z is the coordinate of propagation, and bars represent averages.

The electrons where produced by simulated photoemission with an initial bunch duration of
o = 30 fs (FWHM), and an initial radius ro=xy= 50 m. The charge follows a Gaussian
distributions in both space and time. The photocathode is held at -120kV with respect to a
grounded anode. The electromagnetic fields of the accelerator and the RF-cavity used were
simulated using SuperFish and are described by the so-called fieldmaps (see Figure 3.2).
These fieldmaps include all the EM field components and thus approximate the real fields in
the laboratory setup. The photoemitted electron bunches develop a linear chirp in the
longitudinal direction, that is the front electrons are accelerated and the back electrons are
decelerated through space-charge interactions (see Figure 3.4). The function of the solenoidal

33
magnets, S1 and S2, is to control the transverse properties of the bunch before it enter the RF-
cavity.

GPT has a built in capability to model the fields produced by solenoids with a rectangular
cross-section. The solenoids are modeled by a 4th order Taylor series expansion from the
analytical expression for the on-axis field Bz where

z+a o Ir 2
Bz (z , r = 0 ) =
Rout

Rin
z a 2(r 2 + z 2 )
3
2
dr dz (3.2.1)

The effect of space-charge is accounted for by a Point-to-Point method, where the fields
generated are calculated directly from relativistic particle-particle interaction.

In the simulations the first solenoid, S1, has an inner diameter Rin=36mm, an outer diameter
Rout=56mm and the total length L=16mm and is positioned close to the accelerator, at z = 50
mm. Dimensions of the second solenoid S2 are similar to that of S1; Rin=36mm, Rout=5.6mm
and L=16mm. The position of S2 and its magnetic field were varied to achieve the desired
results.

3.3. Results and discussion

The GPT simulations were performed to study the total RF compression and the bunch
evolution in an optimized setup using GPTs built in solenoid (rectcoil). By injecting a bunch
just before the field goes through zero, the position and magnetic field strength of the second
solenoid are chosen such that both the transverse and longitudinal foci coincide, thereby
determining the location for the UED experiment.

An acceleration field of approximately 10 MV/m is achieved by applying a DC voltage of


120 kV between the cathode and the anode. A 30 fs laser pulse is used to excite the
photodiode, thus releasing the electrons. These electrons are then accelerated through the
diode to energy of 120 keV. Because of its linear space-charge fields, the photoemitted bunch
will evolve such that its phase-space distribution becomes linearly chirped with faster
electrons towards the front and slower electrons towards the back of the bunch.

34
Figure 3.5 The RMS beam radius (x) and pulse length (z) as a function of position in the RF compressed
beamline

Figure 3.6 Magnetic fields required for collimating and focusing the electron bunch. The green line refers to
the fields produced by S1, and the blue line show the fields produced by S2

The RMS radius as a function of position is shown in Figure 3.5. Simulation results using the
rectcoil showed that the collimating lens, S1, should provide a B2z = 1.44x10-4 T2m2, and
from equation (2.2.3) the focal length would be f = 4.9 mm, while the focusing lens, S2,
35
should provide B2z = 1.24x10-5 T2m2 a focal length f = 42.4 mm. The lenses S1 and S2
would need a current density of 1063 A/cm2 to collimate the beam and a current density of
480A/cm2 to focus it. The on axis magnetic field of the collimating lens is shown in Figure
3.6. The RF cavity was positioned at z = 430mm, and hence the position of S2 was at z =
490mm.

36
Chapter 4. Magnetic Lens Design

Two magnetic lenses are needed in our proposed experimental setup to control the transverse
properties of the electron bunches, mainly one for collimating the beam after photoemission
and another to ensure that the beam focuses at the sample or detector. In this chapter we will
discuss the parameters needed in order to achieve the required fields while satisfying other
experimental constraints. Figure 4.1shows a schematic of the entire proposed UED setup.

Figure 4.1 Schematic of the proposed UED setup

4.1. Design Requirements

From the GPT simulation results presented in the previous chapter we determined that the
magnetic lenses S1 and S2 must provide a focal length of approximately 4.9 mm and 42.4
mm respectively. Both lenses must also have an inner diameter which is larger than 1.5
inches, since the beam tube chosen for the vacuum pumping considerations has an outer
diameter of 1.5 inches. The physical length of lens S1 should also be as short as possible so
as to minimize the propagation distance of the electron bunch from photocathode to sample in
experimental configurations that do not include the RF cavity (i.e. the compact electron
beamline shown in fig. 4.1). To maintain the beam quality, the magnetic field at the
photocathode must also be very small as will be discussed in the next section.

In summary, design parameters that need to be determined are:

Number of Ampere-turns,
37
Total thickness of the shield,
Size of the gap, and
Total length along the z- axis.

4.2. Buschs Theorem

When charges are emitted from a surface immersed in an axial magnetic field in cylindrically
symmetric problems the conservation of canonical angular momentum has ramifications for
the dynamics of the emitted particles. Non-zero fields results in the beam being born with a
non-zero angular momentum2.1.1). Solenoidal magnets will produce an initial angular
momentum that manifests in an initial emittance that cannot be eliminated.

Busch theorem states that the angular velocity of any electron at any point depends only on
the difference between the total magnetic fluxes linked by the charge rings at two electron
positions [38]. The angular velocity does not depend on the trajectory between these points,
which could be written as

q
& = (o z ) (4.1.1)
2r 2 m

where o is the magnetic flux at the cathode. Busch's theorem actually represents
conservation of angular momentum. We can note that the angular rotation varies linearly with
the cathode magnetic field.

The beam emittance due to the fields at the photocathode alone can be shown to be [45]

ero2
=
x
n Bo (4.1.2)
8mc

Thus the magnetic field at the photocathode will not significantly deteriorate the initial
emittance when

8mc Th
Bo << n (4.1.3)
ero2

38
For a 100 KeV electron beam, the emittance should not exceed 0.58 mm mrad from equation
(2.6.4). Taking into account an initial beam radius of 100 m, the magnetic field at the
photocathode should be Bo<< 79 G. This is an important design requirement for the lens S1.

4.3. Magnetic lens simulation

Simulations of the magnetic lens were performed using Poission SuperFish. SuperFish is
one of the main solver programs in a collection of programs for calculating static magnetic
and electric fields and radio-frequency electromagnetic fields in either 2-D Cartesian
coordinates or axially symmetric cylindrical coordinates. The programs generate a triangular
mesh fitted to the boundaries of different materials in the problem geometry. Poission
SuperFish solves Poissions equation (or Laplaces) for the vector (scalar) potential with
nonlinear isotropic iron (dielectric) and electric current (charge) distributions for 2-D
Cartesian or 3-D cylindrically symmetry.

The Poission group of codes solves Maxwells equations (static) in integral form and in two
dimensions. The Maxwell equations (static), taken together with the boundary conditions for
the particular problem, are equivalent to a generalized form of Poissions equations in 2-D.
Poission uses successive point-over relaxation method to solve the equations [47]

The lens inner radius was kept at a minimum of 1.8 so that it could fit and be positioned
around a 1.5 beamtube. The other parameters that were adjusted in the design process were
the excitation current density, the thickness of the shield, the size of the gap and the total
length of the lens along the z-axis; these parameters are shown in Figure 4.2.

To achieve required fields mentioned in the previous chapter, the excitation coil designed has
an inner radius of 3.4cm, an outer radius 5.6cm and is 1.6cm long. Using copper wires of
1.3mm diameter (16 AWG), the expected number of turns is about 200 turns. A shield of
sufficient thickness and a 0.5cm gap was used on the first lens to help concentrate the
magnetic fields and provide the required focal length using current that can be easily
provided.

39
Figure 4.2 Design parameters of the lenses. The lens's inner diameter was fixed to a minimum of 1.8", while
the dimensions of the gap s, the length L, the shield thickness t, number of turns and current density where
varied so as to achieve the desired focal length while maintaining short propagation distance

Figure 4.3 Fields of the designed lens without a shield (left) and with a shield of 0.5cm gap (right). Note that
the fields produced by the shielded lens are more focused and stronger than the non-shielded solenoid,
although both have equivalent integrals; the shielded lens produces significantly shorter focal lengths.

The material used in the shield is a non-oriented 80% nickel-iron-molybdenum alloy,


provided and machined by MuShield, which offers extremely high initial permeability and
maximum permeability with minimum hysteresis loss [48]. To relieve the strain after
40
machining and to re-obtain isotropic properties the finished parts are annealed in an oxygen-
free dry hydrogen atmosphere. The lens thickness and geometry was varied and optimized to
provide strong magnetic fields and prevent saturation. The saturation effect was studied, by
comparing the field broadening from simulations using different current densities. The
normalized axial field is plotted in Figure 4.3, it is noticed that the shield starts to saturate at a
1000A/cm2 current density.

Figure 4.4 B(H) curve of the material used in the shield. The material starts to saturate at4000 G (0.4T) [49]

41
Figure 4.5 Saturation effect studies of a 1.5 cm thick shield. The magnetic fields begins to broaden at
1000A/cm2

The shield thickness of 1.5 cm according to simulations would start to saturate at a current
density of 1000A/cm2, which is equivalent to 13.27 A, see Figure 4.5. The total length of the
lens with the shield included is 4.6cm.

Figure 4.6 Assembly of the lens. The lens consists of three parts: a spool, which also acts as a cooling
channel for the coil, and two parts which make up the shield. These slides on top of the coil, covering the
whole coil except for 0.5 cm gap. For more detailed drawings refer to the appendix.

42
Figure 4.7 The cooling channel. The spool on which the wire is wrapped around. This acts as a cooling
channel. For more detailed drawings refer to the appendix.

The parts of the lens are shown in Figure 4.6. The shield consists of two parts that slide over
the coil, which is wrapped around a spool that acts as a cooling channel. The cooling channel
was designed so as to pass water slowly around the wire and allow heat dissipation. The
schematic of the cooling channel is shown in Figure 4.7. Water passing through the spool is
maintained at a fixed temperature. Temperature calculations and considerations are discussed
in the next section.

43
Figure 4.8 The bunch evolution in an optimized setup. RMS radius as a function of position

Results from SuperFish simulations of the lens designed were used in GPT simulations and
compared with the results presented in the previous chapter. It was found that a shielded lens
with a current density of 600 A/cm2 provides the desired focal length for collimating the
beam, while an unshielded lens with a current density of 480 A/cm2 provides the needed
magnetic field to focus the bunch. The behaviour of the bunch as it propagates through the
setup is shown in Figure 4.8 using both the SuperFish lens design and the rectcoil. From the
figure, the time focus and beam waist are both positioned at the same point, z = 900mm, and
the bunch has RMS width of 275 m The on axis magnetic field of the collimating lens is
shown in Figure 4.9

In order to measure diffraction patterns of sufficient quality in a single-shot with temporal


resolution of atomic motion, the bunch should[2]:

1- contain more than 106 electrons which implies that Bunch charge Q > 0.1 pC,
2- RMS bunch duration t < 100 fs

44
3- RMS radius x < 200 m

In addition to above bunch requirements it is also important that time focus and beam
waist are both positioned at same point, as explain in section 3.1.

The simulation results showed that the designed lenses fulfill the requirements for such
electron bunches. The beam waist can be chosen to coincide with the time focus or the
detector position.

Figure 4.9 On-axis magnetic field needed to collimate the bunch. Red line shows the field produced by an
unshielded solenoid and the blue line shows the field produced by a shielded solenoid. The shielded solenoid
needs less current to provide the desired focusing of the bunch.

Measurements for both lenses were performed after they were machined. A Tesla-meter was
mounted on a sliding stage and the magnetic field produced by the lens was measured from a
distance of 9 cm to 1.9 cm with an interval of 1mm. The geometry of the Tesla-meters probe
used in these measurements limited the readings to dimensions outside of the lens, hence the
field at z=0 could not be measured and was estimated to fit the curve.

45
These measurements were done for several driving currents and are shown in Figure 4.10.
The measurements showed that the lens had 130 thus requiring more driving current to
achieve the required focal lens (~19 A for collimating lens S1) .

Figure 4.10 Measured axial magnetic fields of the machined lens. The input currents used were, from bottom
to top: 6A, 7A, 8A, 10A, 21A, 23A, 25A, 27A

4.4. Heat equation and cooling considerations:

4.4.1. Power dissipation:


It is important to known the amount of electrical power required to generate a given magnetic
field as this power will be dissipated in the coil and appear as heat. For a typical magnetic
lens, the power can be calculated by integrating the power in an elemental cross-section of
the coil. If the current density, j, and the resistivity r are constant over the volume [50] then,

W = dW = j 2 dV , (4.4.1)

and
46
W = j 2 r rin3 2 ( 2 1), (4.4.2)

rout L
where = , = , and rout ,r in and L are the outer radius, inner radius and the length of
rin rin
the solenoid respectively, and is the space factor and is defined by

active sec tion of the winding


= (4.4.3)
total sec tion of the winding

The current density is

NI
j= (4.4.4)
L (rout rin )

The highest possible space factor in the case of wires with circular cross-section is /4 [46].

The power can be written in terms of actual applied current instead of current density,

2
rout rin2
W= r ( NI ) 2 (4.4.5)
2 (rout rin ) 2

If the power is too high the temperature of the solenoid can get above the melting point of the
wire, so cooling of the coil must be provided.

4.4.2. Heat Calculations:


To estimate the temperature within the conductor, we need to know the temperature rise due
to thermal gradients in the conductor and across any thermal barrier between the point and
the cooling surface. The temperature profile within a body depends upon the rate of its
internally-generated heat, its capacity to store some of this heat, and its rate of thermal
conduction to its boundaries. Mathematically this is stated by the Heat Equation,

r 1 T 1
2T = Q (4.4.6)
t k

where Q is the power generated per unit volume, k is the thermal conductivity, and is the
thermal diffusivity, and it is related to the thermal conductivity, k , specific heat capacity, and
density by ,
47
k
= (4.4.7)
C p

when the temperature is time independent, then this is know as the steady-state, and the
equation then becomes

r 1
2T = Q (4.4.8)
k

In the case when heat is being generated within an electrical element,

Q = j2 (4.4.9)

Where j is the current density and is the electrical resistivity.

A number of analytical solutions are available, but in most practical situations, these
solutions cannot be applied because of the geometry and the boundary conditions of the
problem which will then involve complex series solutions and would be difficult to
implement. In such case, the problem can be solved numerically.

Figure 4.11 The discretization process. The numerical accuracy of these calculations depends strongly on the
number of designated nodal points, which control the number of elements generated

In contrast to an analytical solution that allows for the temperature determination at any point
in the medium, a numerical solution allows the determination of temperature only at discrete
points. The first step in any numerical analysis must therefore be to select these points. This
is done by dividing the region of interest into a number of smaller regions, a process known
as discretization, as shown in Figure 4.11. These regions are bounded by points. These
48
reference points are termed nodal points and their assembly results in a grid or mesh. It is
important to note that each node represents a certain region surrounding it, and its
temperature is a measure of the temperature distribution in that region [51]. In this study
Finite Element Method (FEM) was used which involves the discretization of both the domain
and the governing equations. The FEM considers that the solution region comprises many
small elements and gives a piece-wise approximation to the governing equations, that is, the
complex partial differential equations are reduced to either linear or nonlinear simultaneous
equations. Thus, the finite element discretization procedure reduces the continuum problem,
which has an infinite number of unknowns, to one with a finite number of unknowns at
specified points referred to as nodes which then can be solved to find the temperature at a
certain given point.

Since the lens is axially symmetric, and there is no significant variation in temperature
expected in the circumferential direction (), it can be treated as a two dimensional
rectangular problem, see Figure 4.12. The cooling channel provides a water supply
maintained at 22oC, thus the boundary conditions at 3 sides is considered to be T=22 oC. The
fourth side was also taken to be at T=22 oC for simplicity. From equation (4.4.9) we find that
Q, the power generated per unit volume is equal to 0.604 W/cm3, provided that the current
density is 600 A/cm2 and the resistivity of copper wire is 1.72 cm.

Figure 4.12 Cross-section of a solenoid. Since the lens is rotationally symmetric, a cross-sectional area can
be used to solve the heat equation in 2-D, the color scale used shows the results using thermal conductivity of
grease kgrease.

49
Figure 4.13 Thermal conductivity of the lens. The lens consists of different material with different thermal
conductivity. Copper wires are coated with polymer and silicon grease is used as a thermal conductive
material

Thermal conductivity of the coil cannot be determined easily, since the coil consists of
different materials with significantly different thermal conductivities, see Figure 4.13. The
copper wire is insulated by a polymer and silicone grease was used as thermal grease. Heat
calculations where done for kGrease< k< kcopper , Using k = kGrease, the temperature rise was not
significant; the temperature rises to a maximum of 32.8oC. The wire insulation holds up to a
temperature of 240oC, for additional precaution k = kGrease/10 was also taken into
consideration and the maximum temperature rise was fond out to be 130.2 oC, see Figure
4.14. The values of thermal conductivities and corresponding maximum temperatures are
given in table below.

50
Figure 4.14 Variation of temperature with respect to the thermal conductivity. The temperature rise in the
lens depends basically on its thermal conductivity. The above graph shows the heat distribution in the XY-
plane in the case of k = kGrease/10

4.5. Conclusions

Ultrafast electron diffraction resolution was limited due to space charge broadening
effects in the electron bunch. In the proposed approach, the broadening due to space
charge can be overcome by manipulating an ellipsoidal bunch using RF and magnetic
fields to focus it temporally and transversely, thus achieving sub-fs temporal
resolution. Magnetic solenoid lenses are used for focusing and defocusing of the
electron bunch transversely. The simple paraxial ray equation (2.1.17) alone is
insufficient to determine the beam properties due to space charge effects. Therefore,
the bunchs behaviour was studied as it propagates along the beamline, taking into
consideration the space charge effects using GPT simulations with the built in lenses.
This then provided us with the required focal length for both collimating and the
focusing lens which were 4.9mm and 42.9mm respectively. The beam tube chosen for
the vacuum pumping considerations has an outer diameter of 3.81 cm which limits the
inner diameter of both lenses. Therefore 4.2 cm inner diameter was provided for both
lenses to allow easy position adjustment. The physical length of the collimating lens
should also be as short as possible so as to minimize the propagation distance of the
electron bunch from photocathode to sample in the experimental configurations that
do not include the RF cavity. The design length of the collimating lens used was 5.1
cm. To achieve required focal length the excitation coil designed was 1.6 cm long,
had an inner diameter of 6.8 cm and an outer diameter of 11.2 cm. A shield of 1.5cm
51
thickness and a 0.5cm gap was used on the collimating lens to help concentrate the
magnetic fields and was driven by current density of 600 A/cm2 to provide the
required focal length, while a current density of 480 A/cm2 was used to drive the
focusing lens.
The above lens was modeled using SuperFish which simulated the magnetic field
lines produced by the lens. These magnetic fields were then used to simulate the
behaviour of the electron bunch along the beamline using GPT. The results of these
simulations also produced similar focusing effect as those produced by previous GPT
simulations and the electron bunch produced had a 0.1 pC charge and sub-fs temporal
resolution.

52
References

[1] Polanyi, J. C. and Zewail, A. H. Direct Observation of the Transition State Acc.
Chem. Res., 28 (3), 119. (1995)

[2] Dwyer, J. R.; Hebeisen, C. T.; Ernstorfer, R.; Harb, M.; Deyirmenjian, V. B.; Jordan,
R. E. and Miller, R. J. D. Femtosecond Electron Diffraction: Making the Molecular
Movie Phi. Trans. R. Soc. A, 364 (1840), 741. (2006)

[3] Fueno. T. The Transition State: A Theoretical Approach; Gordon and Breach Science
Publishers (1999)

[4] Polanyi, M. and Evans, M.D Some Applications of the Transition State Method to
the Calculation of Reaction Velocities, Especially in Solution Trans. Faraday Soc.,
31, 875. (1935)

[5] Eyring, H.; Gershinowitz, H. and Sun, C. E.; The Absolute Rate of Homogenous
Atomic Reactions J. Chem. Phys., 3, 107. (1935)

[6] Truhlar, D.G.; Garrett, B. C. and Klippenstei, S.J.; Current Status of Transition-State
Theory J. Phys. Chem.,100 (31), 12771. (1996)

[7] Lienhard. G.E.; Enzymatic Catalysis and Transition-State Theory Science, 180
(4082), 149. (1973)

[8] Miller, W.H.; Semiclassical Limit of Quantum Mechanical Transition State Theory
for Nonseparable Systems J. Chem. Phys. 62, 1899. (1975)

[9] Plaxco, K. W.; Simons, K. T. and Baker, D. Contact Order, Transition State
Placement and the Refolding Rates of Single Domain Proteins J. Mol. Bio., 277 (4),
985. (1998)

[10] Wan, C. ; Fiebig, T.; Kelley, S. O.; Treadway, C. R.; Barton, J. K. and Zewail, A. H.
Femtosecond Dynamics of DNA-Mediated Electron Transfer PNAS, 96 (11),
6014. (1999)

53
[11] Siwick, B. J.; Dwyer, J. R.; Jordan, R. E. and Miller, R. J. D. Femtosecond Electron
Diffraction Studies of Strongly Driven Structural Phase Transition J. Chem. Phys.,
299 (2-3), 285. (2004)

[12] http://nobelprize.org/nobel_prizes/chemistry/laureates/1967/porter-bio.html

[13] Siwick, B. J.; Dwyer, J. R.; Jordan, R. E. and Miller, R. J. D. Ultrafast Electron
Optics: Propagation Dynamics of Femtosecond Electron Packets, J. Appl. Phys., 92
(3), 1643. (2002)

[14] Zewail, A. H. 4D Ultrafast Electron Diffraction, Crystallography, and Microscopy


Ann. Rev. of Phys. Chem., 57, 65. (2006)

[15] Mark, H. and Wierl, R. ber Elektronenbeugung Am Einzelnen Molekl


Naturwissenschaften, 18, 205. (1930)

[16] Hargittai, I. and Harigittai, M.; Stereochemical Applications of Gas-Phase Electron


Diffraction, Part. A: The Electron Diffraction Technique, VCH, New York, (1988)

[17] Cerullo, G. and De Silvestri, Ultrafast Optical Parametric Amplifiers, S., Rev. Sci.
Inst., 74(1), 1 (2003)

[18] Hastings, J. B.; Rudakov,F. M.; Dowell, D. H.; Schmerge, J. F.; Cardoza, J. D.;
Castro, J. M.;Gierman, S. M.;Loos, H. and Weber, P. M. Ultrafast Time-Resolved
Electron Diffraction with Megavolt Electron Beams Appl. Phys. Lett., 89 (18),
184109 (2006)

[19] Lobastov, V. A.; Srinivasa, R. and Zewail A. H. Four-Dimensional Ultrafast


Electron Microscopy Proc. Natl Acad. Sci. USA, 102 (20) 7069. (2005)

[20] Siwick, B. J.; Dwyer, J. R.; Jordan, R. E. and Miller, R. J. D. An Atomic-Level View
of Melting Using Femtosecond Electron Diffraction, Science., 302 (5649) 1382.
(2003)

[21] F. Schotte et al., Watching a Protein as it Functions with 150-ps Time-Resolved X-


Ray Crystallography, Science, 300 (5627), 1944. (2003)

[22] H. Ihee, V.A. Lobastov, U.M. Gomez, B.M. Goodson, R. Srinivasan, C.-Y. Ruan,
A.H. Zewail, Direct Imaging of Transient Molecular Structures with Ultrafast
Diffraction, Science, 291(5503), 458. (2001)

54
[23] Luiten, O. J.; van der Geer, S. B.; de Loos, M. J.; Kiewiet, F. B. and van der Wiel, M.
J. How to Realize Uniform Three-Dimensional Ellipsoidal Electron Bunches Phys.
Rev. lett. 93 (9), 094802. (2004)

[24] van der Geer, S.B.; van der Oudheusden, T.; opt Root, W.P.E.M.; van der Wiel, J.;
Luiten, O. J. Longitudinal Phase Manipulation of Ellipsoidal Electron Bunches in
Realistic Fields; Phys. Rev. ST Accel. Beams, 9, 044203. (2006).

[25] Wiedemann , H.; Particle Accelerator Physics, Wiedemann; Springer (2007)

[26] Hall, C. E. Introduction to Electron Microscopy, McGraw Hill, (1953)

[27] Liebmann, G.; and Grad, E. M.; Imaging Properties of a Series of Magnetic Electron
Lenses; Proc. Phys. Soc. B, 64, 956. (1951)

[28] El-Kareh A. B. and El-Kareh. J. C. J. ; Electron Beams, Lenses and Optics, New York
: Academic Press, (1970)

[29] de Lacheisserie, . T.; Gignoux, D.; Magnetism Fundamentals, Norwell,


Mass.: Kluwer Academic Publishers, (2002)

[30] Juma , S. M. and Yahya, A. A. ; Characteristics of Axial Magnetic Fields of


Asymmetric Electron Lenses J. Phys. E: Sci. Instrum., 17, 398 (1984)

[31] Mulvey, T. and Wallington, M. J.; Electron Lenses, Rep. Prog. Phys., 36, 347.
(1973)

[32] Mulvey, T. and Wallington, M. J.; The Focal Properties and Aberrations of Magnetic
Electron Lenses J. Phys. E: Sci. Instrum., 2, 466. (1969)

[33] Lawson, J.D.; The Physics Of Charged-Particle Beams. Oxford Press, (1977)

[34] http://info.ee.surrey.ac.uk/Workshop/advice/coils/mu/#sat

[35] Steinmetz C. ; Theory and Calculation of Electric Circuits; McGraw-Hill, New York,
USA, (1917)

[36] Bauer, M.; Pole Piece Saturation in Magnetic Electron Lenses; J. Phys. E: Sci.
Instrum., 1, 1081. (1968 )

[37] Liebmann, G; The Effect of Pole Piece Saturation in Magnetic Electron Lenses
Proc. Phys. Soc. B, 66, 448. (1953)

55
[38] Reiser., M.; Theory and Design of Charged Particle Beams; New York, Wiley, (1994)

[39] Molokoviskii, S. I. and Sushkov, A.D.; Intense Electron and Ion Beam, Springer,
(2005)

[40] Clendenin, J. E.; Kotseroglou, T.; Mulhollan, G. A.; Palmer, D. T. and Schmerge, J.
F.; Reduction of Thermal Emittance of RF Guns; Stanford Linear Accelerator Center,
Stanford, CA 94309, USA (1999)

[41] Flottmann, K.; Note on Thermal Emittance of Electrons Emitted by Cesium Telluride;
DESY, Notkestr. 85, 22603 Hamburg, Germany. (1997)

[42] van Oudheusden, T. ; de Jong, E.F. ; van der Geer, S.B. ; Op 't Root, W.P.E.M. ;
Luiten, O.J. and Siwick, B.J. ; Electron Source Concept for Single-Shot Sub-100 fs
Electron Diffraction in the 100 KeV Range, J. Appl. Phys., 102 093501. (2007)

[43] Strehl, P. Beam Instrumentation and Diagnostics, Springer (2006)

[44] http://www.pulsar.nl/gpt/

[45] Palmer, D.T.; Wang, X.J.; Ben-Zvi, I.; Miller, R.H.; Skaritka, J. Experimental
Results of a Single Emittance Compensation Solenoidal Magnet, Proc. of the 1997
PAC, 2843. (1997)

[46] Avallone, E. A.; Baumeister, T.; Sadegh, A. and Marks, L. S.; Marks' Standard
Handbook for Mechanical Engineers, McGraw Hill. (1996)

[47] http://laacg1.lanl.gov/laacg/services/download_sf.phtml#ps0

[48] http://www.mushield.com/material-specs.shtml

[49] http://www.mushield.com/documentation/catalog_and_design_guide.pdf

[50] Montgomery, D. B.; Solenoid Magnet Design: The Magnetic and Mechanical Aspects
of Resistive and Superconducting Systems, Reprint edition, Huntington, N.Y. : R. E.
Krieger Pub. Co. (1980)

[51] Lewis, R. W.; Nithiarasu, P. and Seetharamu, K. N.; Fundamentals of the Finite
Element Method for Heat and Fluid Flow, Wiley. (2004)

56
57
APPENDIX

58
59
APPENDIX

Design drawing of the cooling channel, all dimensions are in mm

Side A:
Spool:
Side B:
Water inlets:
Design drawinng of the Shielding, all dimensions are in mm,

Side 1:
Side 2:
Assembly of parts

Cooling channel
Shield Assembly

Potrebbero piacerti anche