Characterisation and Application of A Laser-Based Hard X-Ray Source

SE9900161
Characterisation and Application

of a Laser-Based Hard X-ray Source
Matthias Gratz
Lund Reports on Atomic Physics

LRAP - 236
Doctoral Thesis
Department of Physics
Lund Institute of Technology
November 1998
ISBN 91-628-3136-4
LUND
TEKNISKA HOGSKOLANI LUND /WM?'&r\ ^ S 1 1 ™ ™ 0 F TECHNOLOGY
INSTITUTIONEN FOR FYSK ifeijS^, )| DEPARTMENT OF PHYSICS
Characterisation and Application

of a Laser-Based Hard X-ray Source
Matthias Gratz
AKADEMISK AVHANDLING
som for avlaggande av teknologie doktorsexamen

vid tekniska fakulteten vid Lunds Universitet
kommer att offentligen fbrsvaras vid
InstitutionenfdrFysik, Solvegatan 14, Hbrsal B,
fredagen den 6 november 1998, klockan 10.15
Organization Document name
Doctoral Dissertation
Lund University Date of issue
Department of Physics November 1998
Lund Institute of Technology
CODEN
P.O. Box 118, S - 221 00 Sweden
LUTFD2/(TFAF-1038)/l-79/(1988)
Author
Matthias Gratz
Title and subtitle
Characterisation and application of a laser-based hard x-ray source
Abstract
Hard X-rays are generated by focusing 110 fs laser pulses with intensities of about 1017 W-cm-2 onto
solid metal targets. Characteristic properties of this X-ray source are the small source size, the short pulse
duration and the high peak flux. The aim of the present work was to characterise this X-ray source and to
demonstrate possible applications. A comparison with other X-ray sources and conventional imaging
techniques is made.
Characterising measurements were performed, including source size, emission spectrum, temporal
behaviour, source stability and the influence of various laser parameters. The emission spectrum was
measured using both energy-dispersive solid-state detectors and wavelength-dispersive crystal
spectroscopy. The conversion efficiency from laser light to X-ray radiation was measured for different
target materials. The laser ablation from different targets was studied.
The feasibility of special imaging techniques, e.g. differential imaging and time-gated imaging, was
investigated both theoretically and experimentally. Differential imaging allows for selective imaging of
contrast agents, while time-gated imaging can reduce the influence of scattered radiation in X-ray imaging.
Time-gated imaging was demonstrated in different imaging geometries, both for planar imaging and
computed tomography imaging. Reasonable agreement between theoretically calculated values and
experimental results was obtained.
Key words
X-ray sources, laser-produced plasma, X-ray spectroscopy, radiography, medical imaging, differential
imaging, time-gated imaging, scatter reduction, Monte Carlo simulation, computed tomography
Classification system and/or index terms (if any)
PACS: 02.70.Lq, 07.85.Fv, 07.85.Nc, 29.30.Kv, 29.40.Mc, 29.40.Wk, 29.40.Ym, 32.80.Cy, 42.62.Be,
52.70.Nr, 78.70.Ck, 87.59.Bh, 87.59.Fm, 87.59.Hp
Supplementary bibliographical information Language
English
ISSN and key title ISBN
91-628-3136-4
Recipient's notes Number of pages Price
151
Security classification
Distribution by (name and address)

Laila Lewin, Department of Physics, P.O. Box 118, S - 221 00 Lund, Sweden
I, the undersigned, being the copyright owner of the abstract of the above mentioned dissertation, hereby grant to
all reference sources permission to publish and disseminate the abstract of the above-mentioned dissertation.
Signature Date
TABLE OF CONTENTS
Abstract 1
List of Papers 3
1. Introduction 5
2. Physics of Laser-Produced Plasmas , 9
2.1 General definitions and notations 9
2.2 Laser light absorption and related processes 10
2.3 X-ray emission from laser-produced plasmas 13
3. Experimental Setup 15
3.1 Laser systems 15
3.2 X-ray generation 18
3.3 X-ray detection 21
3.3.1 Image plates 21
3.3.2 Scintillator-CCD camera combination 21
3.3.3 Scintillator-photomultiplier combination 23
3.3.4 Absolute calibration of the scintillator-based detectors 24
3.3.5 Time-resolved X-ray detection 25
4. Source Characterisation 27
4.1 X-ray source size 27
4.2 X-ray flux 28
4.3 X-ray flux fluctuations 35
4.4 Ablation phenomena 37
5. Spectral Characterisation 39
5.1 Energy-dispersive single-photon counting with conventional detectors 39
5.1.1 Principles 39
5.1.2 Experiments 40
5.2 Energy-dispersive single-photon counting with CCD cameras 41
5.2.1 Principles 42
5.3 Crystal spectroscopy 46
5.3.1 Principles 46
5.3.3 Conversion efficiency and hot electron temperature 51
6. Comparison with other X-ray Sources 53
7. Differential Imaging 57
7.1 Principles 57
7.2 Experiments 58
8. Time-Gated Imaging 61
8.1 The scattering problem in radiology 61
8.2 Monte Carlo simulations of time-gated imaging 63
8.2.1 Methods 63
8.2.2 Simulation geometries 65
8.2.3 Results 66
8.3 Experiments 69
8.3.1 Time-gated planar imaging 69
8.3.2 Time-gated computed tomography 69
8.4 Comparison with conventional scatter-reduction techniques 73
9. Outlook 75
10. Summaries of the Papers 77
Acknowledgements 79
References 81
ABSTRACT
Hard X-rays are generated by focusing 110 fs laser pulses with intensities of about
1017 W-cm"2 onto solid metal targets. Characteristic properties of this X-ray source are the
small source size, the short pulse duration and the high peak flux. The aim of the present
work was to characterise this X-ray source and to demonstrate possible applications. A
comparison with other X-ray sources and conventional imaging techniques is made.
Characterising measurements were performed, including source size, emission spectrum,
temporal behaviour, source stability and the influence of various laser parameters. The
emission spectrum was measured using both energy-dispersive solid-state detectors and
wavelength-dispersive crystal spectroscopy. The conversion efficiency from laser light to
X-ray radiation was measured for different target materials. The laser ablation from different
targets was studied.
The feasibility of special imaging techniques, e.g. differential imaging and time-gated
imaging, was investigated both theoretically and experimentally. Differential imaging allows
for selective imaging of contrast agents, while time-gated imaging can reduce the influence
of scattered radiation in X-ray imaging. Time-gated imaging was demonstrated in different
imaging geometries, both for planar imaging and computed tomography imaging.
Reasonable agreement between theoretically calculated values and experimental results was
obtained.
LIST OF PAPERS
Paper I
M. Gratz, C. Tillman, I. Mercer and S. Svanberg, X-ray generation for medical applications
from a laser-produced plasma, Applied Surface Science 96-98, 443-447 (1996).
Paper II
G. Holzer, E. Forster, M. Gratz, C. Tillman and S. Svanberg, X-ray crystal spectroscopy of
sub-picosecond laser-produced plasmas beyond 50 keV, Journal of X-ray Science and
Technology 7, 50-70 (1997).
Paper III
C. Tillman, S. Johansson, B. Erlandsson, ML Gratz, B. Hemdal, A. Almen, S. Mattson and S.
Svanberg, High-resolution spectroscopy of laser-produced plasmas in the photon energy
range above 10 keV, Nuclear Instruments and Methods A 394, 387-396 (1997).
Paper IV
K. Herrlin, C. Tillman, M. Gratz, C. Olsson, H. Pettersson, G. Svahn, C.-G. Wahlstrom and
S. Svanberg, Contrast-enhanced radiography by differential absorption, using a laser-
produced X-ray source, Investigative Radiology 32, 306-310 (1997).
Paper V
M. Gratz, A. Pifferi, C.-G. Wahlstrom and S. Svanberg, Time-gated imaging in radiology:
Theoretical and experimental studies, IEEE Journal of Selected Topics in Quantum
Electronics 2, 1041-1048 (1996).
Paper VI
M. Gratz, L. Kiernan and K. Herrlin, Time-gated imaging in planar and tomographic x-ray
imaging, accepted for publication in Medical Physics.
1. INTRODUCTION
General aspects of X-ray generation and applications

X-ray technology has found widespread applications since W.C. Roentgen's discovery of
X-ray radiation in 1896. Since then, the main application of X-ray radiation has been
imaging, both in medical and technical contexts [1]. However, the method of X-ray
generation has, in principle, remained the same over the past 100 years. Inside an evacuated
housing, electrons are accelerated towards a metallic target by an electric field, and upon
their deceleration within the solid material X-ray radiation is emitted. The X-ray emission
consists of continuous Bremsstrahlung and characteristic line emission, determined by the
target material.
Besides conventional X-ray tube technology, new X-ray generation methods have been
developed, mainly during recent decades, e.g. laser-produced plasmas (both X-ray lasers and
incoherent broadband X-ray sources) (see, e.g. [2-4]) and synchrotron radiation from
electron storage rings (see, e.g. [5])[6]. In the extreme ultra-violet (XUV) and soft X-ray
regions, high-order harmonic generation from intense laser pulses has been extensively
investigated [7J. More recently, the concept of the free-electron laser (FEL) as a source of
XUV and soft X-ray radiation has been proposed, and planned systems are predicted to
reach the lOkeV energy region [8]. Relativistic Thompson scattering of visible photons
from energetic electrons in a storage ring has been shown to produce extremely short pulses
of hard X-ray radiation [9]. Channelling of energetic electrons in crystals provides another
technique for the generation of short X-ray pulses [10,11]. All these X-ray sources have
widely different properties regarding photon energy, intensity and X-ray pulse duration.
Different novel applications, making use of the characteristic properties of these new
X-ray sources, have been discussed and to some extent experimentally demonstrated. A
variety of basic research applications have been developed, e.g. in the field of spectroscopy
and structure probing [6]. The short pulse duration of some of these sources has been applied
to fast probing of structural changes. In the medical field, monochromatized radiation (e.g.
from synchrotrons) can be used to intrinsically improve the contrast of certain details in
imaging [12,13] or for dual-energy angiography (or dichromography) [14]. X-ray radiation
from a laser-produced plasma has been used in a similar way for contrast-agent-specific
differential imaging [15], and the small size of laser-produced X-ray sources has been
exploited for magnification imaging [16]. In contrast, the potential of the short pulse
duration of, e.g., synchrotron radiation and laser-produced X-ray sources (both down to the
order of picoseconds) has not yet been well exploited in medical imaging. However, the
potential of time-gated detection for scatter-reduced imaging has recently been investi-
gated [17, Paper V, Paper VI].
The introduction of novel and sometimes rather complex X-ray generation techniques in
medical imaging is motivated by the limitations of standard techniques. The requirements
for imaging with high spatial resolution, element-specific contrast and decreased dose
exposure, e.g. through scatter-reduced imaging, are not completely met by conventional
technology, thus requiring X-ray source development. The analysis of microstructures in
Introduction
material inspection or microcalcifications in mammography requires spatial resolution of the
order of 10 p , and correspondingly small source sizes. Dose reduction is an important
aspect in medical imaging, in order to decrease the risk of X-ray induced tumours. This is
especially important in combination with mass screening procedures, such as in
mammography. In fact, a recent study [17] suggests that at least 150 million dollars could be
saved annually in Europe alone, not to mention human suffering, if the collective patient
dose could be reduced to one half. Conventional scatter-reduction techniques impose an
increased dose on the patient. This could, in principle, be avoided by time-gated imaging in
combination with an ultra-short-pulse X-ray source. The selective imaging of a contrast
agent, e.g. in coronary angiography, is accompanied by certain health risks due to the high
concentrations of contrast agent required. These concentrations can be lowered when using
differential imaging with nearly monochromatic radiation, as has been shown in studies
performed with synchrotron radiation [14]. However, the use of synchrotrons is hampered
by their complexity and cost.
Laser-produced plasmas as X-ray sources

Research in laser-produced plasmas, which has been pursued for a number of decades, has
been intimately coupled to the evolution of short-pulse laser techniques. Many early
investigations were carried out in long pulse regimes (10"9 s - 10"6 s) at relatively low laser
intensities. Soft X-ray emission from such plasmas was often used only as a diagnostic tool
in plasma physics. With the development of laser systems delivering pulses in the
picosecond range, laser-produced plasmas gained greater interest as short-pulse X-ray
sources in basic research. With the development of high-intensity, ultra-short-pulse laser
systems (sub-picosecond) in the late 1980s, based on the chirped-pulse amplification
technique, it was soon found that plasmas produced with these laser systems emitted
extremely short and intense bursts of hard X-ray radiation [3,18].
Such a laser-produced plasma X-ray source can, in principle, fulfil the requirements
mentioned above. It has a small source size, is partly monochromatic and emits bursts of
X-ray radiation with extremely short pulse duration. Investigations into the physics of such
X-ray sources is a very active field. However, most laser-plasma physics groups are focusing
on basic plasma physics and fundamental laser-matter interactions at high intensities [4,18-
24]. Only a few have considered medical applications (see, e.g. [25])[26,27].
A laser-produced plasma X-ray source based on a sub-picosecond, high-intensity laser
was built and partly characterised in Lund by C. Tillman [25] a few years ago. Possible
applications were pointed out, such as short-pulse, standard and magnification imaging [28],
element-selective differential imaging [15] and alternative projection geometries [16]. The
biological effects of laser-produced plasma X-ray radiation were addressed in a comparative
study of cell survival rates [29]. The further characterisation of this source and investigations
of its potential in medical imaging, time-gated imaging in particular, are the main topics of
this thesis.
Introduction
Contents of this thesis
To begin with, a theoretical overview of X-ray generation from laser-produced plasmas is
given in Chapter 2, followed by the experimental details of X-ray generation and
measurement techniques, which are presented in Chapter 3.
The first main topic of this thesis is the characterisation of the X-ray source. Important
parameters such as X-ray source size, X-ray spectrum, conversion efficiency and X-ray flux
fluctuations, as well as their dependence on various laser parameters, were measured and are
presented in Chapters 4 and 5. For these measurements, special techniques had to be adopted
to match the particular characteristics of the source, e.g. the ultra-short and very bright X-ray
emission. A comparison with some other X-ray sources is given in Chapter 6.
The second main topic is the potential application of the X-ray source in medical
imaging. Two possible imaging techniques were investigated. Differential imaging, a
technique for element-selective imaging, is presented in Chapter 7. Time-gated imaging, a
technique for scattering reduction, was investigated as an example of an application making
use of the short-pulse nature of the source. Both theoretical and experimental aspects of this
technique are presented in Chapter 8.
Finally, the outlook regarding possible future developments of this X-ray source and
necessary technical improvements is given in Chapter 9. The papers, on which this thesis is
based, are summarised in Chapter 10.
NEXT
left BLANtC
2. PHYSICS OF LASER-PRODUCED PLASMAS
The generation of hard X-rays from a laser-produced plasma is a very complex process. The
main steps in this process are the absorption of laser light in the target with subsequent
heating of the plasma created, the production of energetic electrons in the plasma, electron
migration and deceleration in the target material, and the subsequent emission of X-ray
radiation.
This chapter introduces some basic concepts in the physics of plasmas, with special
respect to laser-produced plasmas. Different laser-light absorption processes and plasma
heating mechanisms are described and related to the X-ray generation process. Characteristic
properties of the X-ray emission from laser-produced plasmas, which are of importance for
various applications, can be deduced from these theoretical considerations.
2.1 General definitions and notations

A plasma is an electrically conducting medium consisting of roughly equal numbers of free
positive and negative ions and charges. Plasmas can be generated by nearly complete
ionisation of gases, liquids or solids. Some examples of plasmas are discharge lamps,
welding arcs, the sun's corona, electrical breakdown and laser-produced plasmas. A plasma
consists of moving free charged particles, therefore it is affected by magnetic and electrical
fields, including those generated within the plasma itself. Plasmas exhibit strong collective
effects, such as plasma waves, due to the long interaction range of these forces.
One of these collective effects is the oscillation of many electrons in the presence of
nearly stationary ions. These electron oscillations, being of essential interest for laser-
produced plasma physics, have an associated frequency, the so-called plasma frequency, cop,
given by:
or cop[s-ll = 56Ajne[cm-3l (1)
where ne is the electron density, me is the electron mass and e the electron charge.
Electromagnetic radiation with a frequency co will propagate through the plasma only as
long as co < cop If the frequency of the electromagnetic wave matches the plasma frequency,
resonant coupling to plasma waves occurs. At higher frequencies no transmission of
electromagnetic waves occurs. The electron density at which cop = co is called the critical
density. A plasma with cop > co is called overdense, and in the opposite case underdense.
Other plasma waves can play an important role in laser-produced plasmas, for example ion-
acoustic waves or sound waves [30].
A plasma can be described by the spatial and temporal distributions of electron density
ne, ion densities ri~m for all ionisation stages m, and corresponding electron and ion energy
distributions. The ions can be considered as stationary in most short-pulse laser-produced
plasmas, i.e. when the laser pulse duration is short compared with the time scale of ion
10 Physics of Laser-Produced Plasmas
a) Before interaction b) Start of interaction c) Somewhat later
Laser beam
n-
Laser beam
Target Target Target

surface surface surface
Figure 2.1: Illustration of the formation of an electron density gradient upon irradiation of a solid
target with a short, intense laser pulse. A laser pulse, incident from the right onto a solid target
surface (a), creates a plasma. Due to plasma expansion, the electron density profile is time dependent
(b) and (c). The laser beam is reflected at the critical density.
movements. In this case, most of the plasma dynamics is connected to the interaction of the
laser light with the plasma electrons. The concept of an electron plasma temperature is often
used to characterise laser-produced plasmas. The definition of temperature assumes a
Maxwellian electron energy distribution, which is not necessarily the case. However, near-
Maxwellian distributions can be attributed to sub-populations of the electrons, thereby
dividing them into, for example, thermal plasma electrons and hot electrons (see also next
section). The hot electron population determines the hard X-ray emission from the laser-
produced plasma.
During the interaction of a sufficiently intense laser pulse (i.e. I > 10H W/cm2) with a
solid target, a plasma is created. The electron density, ne, being highest at the target surface,
decreases with increasing distance from the target (cf. Figure 2.1). The rate of change of the
electron density in the vicinity of the critical density, ncr, is of importance for the interaction
between the laser light and the plasma. This rate of change can be characterised by the
gradient scale length Ln, formally defined as [31]:
(2)
ne dx
The gradient starts to form during the rise time of the laser pulse. Therefore, the
remaining part of the pulse will not interact with an undisturbed solid target surface, but with
an expanding plasma front with a time-dependent gradient scale length. A variety of
absorption mechanisms is responsible for laser light absorption during different stages of
plasma evolution, as presented in the next section.
2.2 Laser light absorption and related processes

A large variety of different absorption mechanisms is responsible for the heating of the
laser-produced plasma. The relative contribution of these mechanisms depends on the laser
intensity and the gradient scale length of the plasma. During the initial phase of the laser-
target interaction, atoms become rapidly ionised in the increasing light field. Laser light
Physics of Laser-Produced Plasmas 11
penetrates a short distance (called the skin depth) into the target material, of the order of a
few nanometres, and interacts with the electrons.
The predominant process during the initial phase of laser light absorption is inverse
Bremsstrahlung absorption [32]. Electrons extract energy from the light field and heat the
plasma by subsequent collisions with the ions. The inverse Bremsstrahlung mechanism is
most efficient at low plasma temperatures, high electron densities and high atomic
numbers [32]. The cross-section for electron-ion collisions decreases with the plasma
temperature, so that inverse Bremsstrahlung becomes less efficient after initial heating.
Resonance absorption at the critical density becomes the dominant absorption
mechanism in a preheated plasma [32-34]. Laser light penetrates the plasma until it reaches
the critical density. Electron oscillations are driven resonantly by a laser field with an
electric field component parallel to the density gradient [34]. This makes resonance
absorption sensitive to the laser angle of incidence and the laser polarisation. The electrons,
accelerated in the plasma wave, dissipate their energy through collisions and wave-breaking
phenomena [33]. This gives rise to an approximately Maxwellian hot electron distribution,
with a temperature Thot (in keV) given by [32]
[ 2 "I 1/3
T I X
hot[keV] ~ 1 0 - [Tcoid[keV] •
where TcoW (in keV) is the initial electron temperature, I is the intensity in units of
1015W-cm"2 and X the laser wavelength in um. Resonance absorption is most effective up to
laser irradiances (=I-A,2) of about 1015 W-cm"2-um2 [33]. A sufficiently large gradient scale
length is necessary for resonance absorption to work.
If the gradient scale length is much shorter than the laser wavelength, resonant build-up
of a plasma wave cannot take place. In this case, a process of "not-so-resonant" resonance
absorption (Brunei-effect, vacuum heating) was proposed by Brunei [35]. Within half a laser
cycle, electrons can be "pulled out" of the plasma by the strong laser field and accelerated
back into the plasma with high energy. This heating process is effective at high laser
irradiances, up to 1018 W-cm"2-um2, and steep density gradients (Ln < 0.1-A.).
The anomalous skin effect is a heating mechanism which becomes effective at high
electron densities and low laser intensities [33]. Electrons moving from the inner part of the
plasma towards the critical density will enter the skin layer, i.e. the layer in the overdense
plasma where laser light penetrates as an evanescent wave. Electrons having the right
velocity and the right arrival time with respect to the oscillating laser field can be reflected
back into the plasma with increased energy. This mechanism can be important during the
falling edge of the laser pulse.
Very intense light pulses produce a high light pressure, PL, on the target. Its magnitude is
given by [32]:
PL[PO] * 3.3 • 1013 • /
10,
For typical experimental parameters used in this work, a light pressure of about 1013 Pa
(= 0.1 Gbar) can be obtained. The action of the light pressure is largest close to the critical
density. Provided the laser intensity is high enough, the light pressure can be larger than the
12 Physics of Laser-Produced Plasmas
thermal pressure of the plasma, and the critical surface can be pushed inwards [31,36]. This
leads to gradient steepening. For very high laser irradiances, i.e. I-A.2 > 1020 W/cm2-um2, an
ion shock front can be created which additionally heats the plasma [32].
Besides these effects, which all take place at the critical density, a number of different
processes, often summarised as parametric instabilities, can contribute to laser light
absorption. The incident laser light can parametrically excite coupled plasma waves, as long
as energy and momentum are conserved . For example, laser light at a frequency a>o can
generate two electron waves at co0/2 at the quarter critical density (cf. Eq. (1)), which is
called a two-plasmon decay [34]. Other examples are stimulated Raman scattering
(03o = COscat,ered light + ©electron wave) Or Stimulated Brillouin Scattering (©o = ^scattered light +
<»ion-acoustic wave)- The waves that are generated by parametric processes are usually damped
quite efficiently in the plasma and contribute, therefore, to the plasma heating. However,
they also lead to undesired effects. If, for example, the two-plasmon decay becomes very
efficient, then the plasma becomes reflective at the quarter-critical density, thereby
preventing the light from reaching the critical-density region [34].
Apart from these absorption and heating mechanisms, a number of other processes also
influence plasma generation. One of them is the generation of magnetic fields, which was
observed already in the 1970s in plasmas generated with relatively long laser pulses [37,38].
The magnetic field generation is mainly attributed to the stream of fast electrons that is
generated by the absorption processes mentioned above [34]. The magnetic field shape is
torroidal around the laser focus, and its magnitude can reach extremely high values of
several hundred Tesla [39]. During the laser pulse, these fields can have noticeable effects
on the plasma dynamics. In ferromagnetic materials, these magnetic fields can last
significantly longer than the laser pulse and thereby affect the plasma dynamics even after
the pulse [40].
The transport of electrons within the plasma and towards the surrounding cold target
material is an important issue for hard X-ray generation. Both experiments and simulations
indicate that fewer electrons than expected propagate into the target [41]. This phenomenon
is called electron inhibition (or flux limitation). It is attributed to the large electric and
magnetic fields that are created by the electrons moving into the target. These fields
constitute restoring forces that limit the electron flux [41], This effect could potentially also
influence the duration of hard X-ray emission.
Most existing short-pulse laser systems produce laser pulses with some kind of small
prepulse before the main pulse, with prepulse-to-main-pulse intensity ratios in the range of
10"4- 10"7. When working at high intensities, the intensity of these prepulses exceeds the
plasma formation threshold on the target surface, and a preplasma is formed. Consequently,
the main pulse interacts with an expanding preplasma, which is characterised by a gradient
scale length depending on the temporal separation of prepulse and main pulse and their
contrast ratio. As a result, the absorption of the main pulse can be increased. However, the
hot electron temperature achievable can be reduced if the scale length becomes too large. In
the soft X-ray region, a prepulse has a strong influence on the emission spectrum [42].
Physics of Laser-Produced Plasmas 13
2.3 X-ray emission from laser-produced plasmas

X-ray emission from laser-produced plasmas has two main origins: (1) radiation from the
hot plasma and (2) secondary radiation from electrons penetrating into the surrounding
target material. The radiation from the plasma consists mainly of thermal radiation and line
emission from the highly excited target atoms. This X-ray emission is situated mainly in the
soft X-ray region, i.e. below a few keV, and has been widely investigated as a tool for
plasma diagnostics (see, e.g., [43]).
The radiation produced by the hot electrons consists of a Bremsstrahlung continuum and
the characteristic line radiation from atoms within the cold target material. The spectral
distribution is mainly determined by the electron temperature in the plasma. Due to the high
electron energies that can be reached in the plasma, X-ray photons with energies up to the
MeV range can be produced [16,18,44]. Assuming a Maxwellian hot-electron energy
distribution, the X-ray spectrum can be approximated by an exponential function at high
X-ray photon energies [45,46]. The hot-electron temperature can thus be accessed
experimentally by measuring the X-ray emission spectrum.
The time scale of X-ray emission from the plasma is determined by plasma dynamics.
The fast expansion of the plasma leads to efficient cooling, thereby rapidly terminating the
X-ray emission from the plasma. Using sub-picosecond laser pulses, the X-ray pulse
duration is of the order of several tens of picoseconds in the 300 eV region [24], and as short
as a few picoseconds in the range of a few keV [47]. In the case of X-ray radiation produced
by hot electrons, the X-ray pulse duration is determined by either the lifetime of the hot
electrons that are penetrating into the target material or the duration of the laser pulse,
depending on which is larger. Simulations suggest hard X-ray pulse durations of about
1 ps [48], for laser pulse durations of 0.7 ps. However, flux-limiting effects (as described in
the previous section) could possibly extend the hard X-ray pulse duration.
The short X-ray pulse duration is the key property for time-gated imaging, which is
described in Chapter 8. Unfortunately, hard X-ray pulses with durations in the picosecond
range are very difficult to measure (see Section 3.3.5). Experimentally determined X-ray
pulse durations of a few picoseconds in the keV region have been reported in the
literature (see e.g. [49]). However, no reliable measurements at significantly higher photon
energies have been published.
S8EXT PAGl(S)
left BLANK
15
3. EXPERIMENTAL SETUP
3.1 Laser systems

The generation of short, intense pulses of hard X-rays from a laser-produced plasma, as
discussed in this thesis, is based on the use of ultra-short, very intense laser pulses. Such
pulses are most frequently generated in compact solid-state laser systems based on the
chirped-pulse amplification (CPA) technique. This technique was developed around
1986 [50]. The chirped-pulse amplification technique made it possible to generate high peak
powers with small-scale laser systems (thus the name T3 lasers: Table-Top Terawatt).
Chirped-pulse amplification advanced rapidly with the introduction of Tksapphire amplifiers
around 1990, when pulses of 100 fs duration on the terawatt level were produced. Chirped-
pulse amplification is now a well established technique [51]. The relatively small size of
T3 systems allowed their installation at many university laboratories and made investigations
in strong-field physics independent of the large laser facilities. The development of compact
T3 laser systems was a requirement for the typetyj of X-ray source discussed in this thesis,
especially with regard to potential applications.
Chirped-pulse amplification
The peak power available from a pulsed laser system is limited, among other things, by non-
linear effects and the damage threshold of the optical materials in the laser system. In ultra-
short pulse laser systems these limitations are reached even at moderate pulse energies,
preventing further amplification. This problem can, in principle, be solved by enlarging the
laser-beam diameter and thus reducing the peak intensity. However, such systems easily
become complex and expensive and have a limited pulse repetition rate.
The concept of chirped-pulse amplification circumvents this problem by temporal
dispersion of the pulse energy. The basic principle of chirped-pulse amplification is
illustrated in Figure 3.1. A short seeding pulse is temporally stretched in a well defined way.
Stretching factors range from 1000 to 100 000, thereby reducing the peak power by the same
factor. These stretched pulses can be amplified up to the damage threshold of the optical
materials in the laser system. Finally, the pulse is expanded in size recompressed in time by
inversion of the stretching operation, concentrating the energy into a pulse (ideally) as short
as the original pulse. In most existing systems, stretching and compression are based on a
wavelength-to-time-delay conversion by grating arrangements, making use of the large
bandwidth of the short pulses [51]. More recently, chirped-mirror techniques have been used
in the sub-20 fs regime [52].
The configuration of the Lund High-Power Laser Facility's terawatt laser [53] is shown
in Figure 3.2. The laser system is based on chirped-pulse amplification in Ti:sapphire.
Seeding laser pulses from a mode-locked oscillator with a duration of about 100 fs, a pulse
energy of a few nJ and a diameter of 1 mm are temporally stretched (2500*) in a grating
arrangement. The stretched pulses are coupled into a regenerative amplifier by polarisation
switching and amplified to a level of about 9 mJ. The amplified pulse is switched out of the
16 Experimental Setup
Stretching
Time Amplification
•.Rccompression
Figure 3.1: Schematic illustration of chirped-pulse amplification. Note that stretching and
recompression are not to scale. Common stretching factors are in the range of 1000 to 100 000.
regenerative amplifier, again by polarisation switching, expanded to a diameter of 8 mm,
and is then fed into a multi-pass amplifier, where it is amplified to several hundreds of mJ.
The beam is then further expanded to a diameter of about 50 mm before the pulses are
temporally recompressed. The upgraded system, which was used in the most recent part of
this work, splits off approximately 100 mJ of the stretched pulse after the multi-pass
amplifier. The beam profile is smoothened by a spatial filter and amplified up to the 1 J level
in an additional multi-pass amplifier. Due to the high peak power, the pulses must be
compressed and guided to the experiment in vacuum to avoid pulse distortion due to non-
linear effects. The laser system typically delivers pulses at about 800 nm at a repetition rate
of 10 Hz.
During the course of the work presented in this thesis, the laser system was modified and
improved several times. During the early part of this work (Papers I and II), the duration of
the compressed pulses was 150 fs, with pulse energies of at most 150 mJ on target. After
modification of the pulse stretcher, the pulse length was reduced to 110 fs (Papers III, IV, V
and VI). The most recent experiments were performed with the upgraded laser system, with
a pulse length of 110 fs and energies of up to 500 mJ delivered to the experiment.
Temporal structure
The temporal structure of the pulses from the laser system consists of both the intense main
pulse, generated by the mechanisms mentioned above, and some weak prepulses, originating
from technical imperfections in the laser setup.
The temporal shape of the main pulse depends on both the stretcher-compressor
configuration and spectral changes introduced by the pulse amplifiers. If the spectral shape
of the pulse is changed during amplification, the original temporal pulse shape cannot be
restored by the compressor. This leads, in most cases, to a temporal broadening of the
generated pulse [51].
Experimental Setup 17
To the experiment:
220 mJ, 110 fs
-800 nm
To the experiment:
500 mJ, 110 fs
-800 nm
Nd:YAG 3-L Multipass

Laser amplifier
(532 nm)
Compressor
Upgraded laser system
Figure 3.2: Schematic overview of the Lund terawatt laser system. Both the original system and the
upgraded laser system are shown.
1000 3
/\
mull!!lILImJ UUil i illilJ

-4 -2 0
Time fps|
Figure 3.3: Example of the laser main-pulse shape, as measured with a third-order autocorrelator.
Several prepulses are generated in the laser system. Some of them are due to the
imperfect pulse-switching from the regenerative amplifier. The Pockels cells, responsible for
switching, are capable of only a finite suppression of the pulses preceding the desired pulse
and therefore leak prepulses at a minimum level of about 10"6 of the main pulse. The
temporal separation between the prepulse and the main pulse is 11.6 ns (the roundtrip time
in the cavity of the regenerative amplifier). Other prepulses are present on the nanosecond
time scale before the main pulse. Studies on the influence of such prepulses on the X-ray
generation are presented in Section 4.2.
The temporal shape of the main pulse was measured with a third-order autocorrelator,
and an example is shown in Figure 3.3. Other measurements of the pulse length were made
with a second-order autocorrelator. No routine measurements of prepulses on the picosecond
time scale were performed. Prepulses on the nanosecond time scale were measured with fast
photodiodes during certain experiments. The laser system had pulse-to-pulse energy
fluctuations of about 7% (see also Section 4.3).
3.2 X-ray generation

Basic setup
The basic setup used for X-ray generation is shown in Figure 3.4. The laser pulses are
directed into a target chamber with gold-coated steering mirrors. The target chamber must be
evacuated in order to avoid electrical breakdown or other non-linear phenomena in air when
the laser pulses are focused. In the target chamber, the laser pulses are focused onto a
rotating metallic target by means of a diamond-turned, parabolic mirror. The parabolic
mirror (Janos Technology Inc., USA) has an effective focal length of 50 mm and a surface
accuracy of A/5 at 800 nm. The pulses are incident at an angle of 30° with respect to the
target surface normal. Intensities of about 1017 W-cm"2 are reached in the focal region. A hot
plasma is generated and emits extremely bright, ultra-short X-ray pulses (as discussed in
Chapter 2). The target chamber, made of aluminium, is shielded by lead bricks (5 cm thick),
in order to protect the operators from harmful X-ray radiation and to reduce the measured
X-ray background.
Without protective measures, sputtered target material from the laser-produced plasma
will be deposited on the focusing optics. Therefore, the pressure in the target chamber is
maintained at up to 20 hPa (1 hPa = 1 mbar) in order to slow down these sputtered particles
and to reduce particle deposition on optical elements. Additionally, thin glass slides
(microscopy slides, 100 um thick) are placed in front of the target in order to intercept
sputtered particles. Inserting such a glass slide had no observable effect on the overall X-ray
yield. Ablation phenomena could be observed by studying deposition patterns on the glass
slides, as described in Section 4.4.
Target materials and setup

Tantalum was used as the standard target material, both in the form of solid plates (thickness
> 3 mm) and foils (thickness 250 um) glued onto steel discs. Other targets were also used,
including solid plates of copper, steel, tin, antimony and cerium, as well as foils of tungsten
and gadolinium (thickness 250 urn). Both fresh and "recycled" targets were flattened,
machined and roughly polished before use, in order to reduce surface irregularities and laser-
produced craters, respectively. Solid targets were found to offer better mechanical stability
compared with glued foils. Target diameters were between 51 mm and 72 mm. A fresh
target surface was exposed to every laser pulse by rotation and translation of the target.
However, it was observed that "reusing" the same tracks or running the target at very low
Rotating
Glass target Vacuum
slides
Parabolic chamber
mirror Lead
bricks
Target
positioning
X-ray
output window Position
sensor beam
Figure 3.4: Basic arrangement for the generation of hard X-rays, showing the components in the
vacuum chamber.
rotation speed (in the case of solid targets) noticeably enhanced the X-ray yield (see
Section 4.2 for further details). However, this operation mode was not used under normal
conditions, in favour of more controlled X-ray generation.
Beam transport considerations

High-intensity laser pulses which propagate through a medium with a non-linear refractive
index, n2, experience an intensity-dependent phase shift. Large values of this phase shift
correspond to a strong wave-front distortion of the laser pulse and can result in an increased
focal spot size. The introduced phase shift should therefore be kept as low as possible. Given
the laser intensity, /, and the laser wavelength, X, the phase shift, A<p, (or B-integral) after
propagation through a medium of length L is [54]:
2n-n2-L-I
(5)
I
For example, for the experiments described in Paper II, the phase shift in the centre of the
20
Unstabilised:
Standard deviation: ± 1 Ijum
Stabilised:
A Standard deviation: ± 1.
0.0 0.5 1.0 1.5 2.0

Number of revolutions
Figure 3.5: Position of the target surface with and without stabilisation for a solid target.
beam was approximately 0.35-2TC before reaching the vacuum chamber entrance window
( I = 6 m, 1 = 800 nm, n2 = 4.7-10"19 cm2-W"' (in air), / = 1011 W-cm"2). An additional phase
shift of about 0.34-2rc was added by this window (10 mm fused silica, n2 = 2.7 cm^W 1 ). For
the experiments described in Papers V and VI, the laser beam was therefore guided through
a vacuum tube and a thinner entrance window (5 mm fused silica). This reduced the total
phase shift by a factor of about 2. The compressor of the upgraded laser system and the
corresponding beam transport system were evacuated, with a remaining phase shift, only due
to the target-chamber window, of about 0.38-271.
Target stabilisation
Assuming an ideal f/1 parabolic mirror and a Gaussian laser beam profile, the laser pulses
would be focused down to spot sizes of about 1 um in diameter, within a focal depth of
about 2 um along the optical axis. Due to technical limitations of the laser setup, the laser
beam was approximately 3x diffraction limited. Therefore, the focal diameter and focal
depth could be expected to increase to about 3 um and 18 um, respectively. Consequently,
the target must be held at the focus position with a correspondingly high precision, in order
to work with well defined X-ray generation parameters. In practice, mechanical irregularities
in the target structure and wobbling due to rotation occur. The wobbling amplitude was
reduced down to about ± 20 um by mechanical adjustment. However, this did not satisfy the
criteria above (much better than ± 9 urn). An active target-position stabilisation was
therefore developed, using a piezo translator stage, to compensate for mechanical
fluctuations.
The target position was measured using a mechanical sensor (Mu-Checker, Mitutoya,
Japan), placed on the target surface (see Figure 3.4). This sensor is easy to adjust and
insensitive to disturbances from the plasma. Relative position changes down to 0.3 um can
be readily measured. The mechanical influence of the sensing head on the target seemed to
Experimental Setup •_ 21
be negligible, apart from small abrasions on very soft target materials (e.g. tin and
antimony).
Different target-position regulation algorithms were tested, until a satisfactory solution
was found. Mechanical backlash, non-linearities and "creeping" of the piezo stage prevented
the use of conventional PID algorithms. Instead, an approach was chosen, in which only the
change in the correction signal is proportional to the difference between the desired position
and the real position, independent of the actual value of the correction signal. In this way,
the problems of imprecision in target positioning were solved. The regulation system was
implemented as a LabView program. Typical position data for a stabilised and a non-
stabilised target are shown in Figure 3.5. New solid targets can be stabilised down to about
± 0.4 um, and recycled foil targets to about ± 3 um.
3.3 X-ray detection

3.3.1 Image plates
The X-ray imaging of large objects was performed using image plates (Fuji BAS 200,
20 cm><40 cm). These image plates have several advantages over conventional film. The
complete dynamical range is 4-5 orders of magnitude, with a sensitivity nearly two orders of
magnitude higher than for conventional X-ray film [55,56]. Readout of the image plates
results directly in digitised images, advantageous for further image processing. The pixel
size is 200 um x 200 urn, with 10 bits intensity resolution. Pixel intensities are stored in
logarithmic form. Exposure times using the image plate system were about 1 to 10 minutes,
depending on the geometry and application. Accepting a lower signal-to-noise ratio, even
single-pulse exposures could be obtained.
A comparison of the image plate's spectral sensitivity to that of other imaging detectors
requires knowledge of its chemical composition. However, data on the exact composition
are not available from the manufacturer. Instead, an approximate composition of F: 10%,
Br:40%, 1:10%, Ba:40% (atomic fractions) as found in [56], was used to when obtaining
interaction coefficients from the National Nuclear Data Center (NNDC, Brookhaven, USA)
X-ray database [57]. The calculated absorption of the image plates is shown in Figure 3.7.
3.3.2 Scintillator-CCD camera combination

A scintillator, coupled to an image-intensified CCD (charge coupled device) camera, was
used as an alternative imaging system. A terbium-doped fibre-optical scintillating plate,
coated with aluminium at the X-ray entrance side, was employed (LKH-6, Collimated Holes
Inc., USA). The use of a scintillating medium in the form of fibre-optical plates results in
good spatial resolution, even for large thicknesses of the scintillator (in contrast, the spatial
resolution of an amorphous scintillator system is approximately limited by its thickness).
In the work presented in Paper II, the fibre plate was placed directly against an image
intensifier. The output face of the image intensifier was lens-coupled to a CCD system
(Star I, Photometries, USA), with a pixel size of 53 urn. The collection efficiency of light
emitted from the image intensifier was rather low, being less than 0.1%. The dynamical
range of the system was about 4000:1.
Figure 3.6: Examples of images obtained with a scintillator-CCD detector, (a) Screws exposed to WO
X-ray pulses (lens-coupled system), and (b) a pair of pliers exposed to one X-ray pulse (fibre-taper-
coupled system).
100
10 100 1000
Energy [keV]
Figure 3.7: Photo-absorption as a function of energy for the image plate (thickness 150 fim) and the
LKH-6 scintillator (thickness 5 mm). Attenuation data for the scintillating fibre plate were provided by
the manufacturer [58].
In later studies, this detection system was significantly improved. The fibre-optical plate
was placed directly against the image intensifier of an intensified CCD camera (Flamestar
Ilf, LaVision, Germany), with an effective pixel size of 36 urn. This camera system had a
fibre optical taper connecting the output of the image intensifier to the CCD camera, thereby
providing a much higher light collection efficiency. The dynamical range of this system was
about 10 000:1. Images obtained with the two systems are shown in Figure 3.6.
The absorption data for the scintillator are shown in Figure 3.7, together with those from
the image-plate detector for comparison. The maximum absorption of the scintillating fibre
plate is only ~ 85% as the plate also contains non-scintillating fibre cladding and internal
light-absorbing material between the fibres. The scintillator-based system is especially
appropriate for X-ray detection at high X-ray photon energies.
i,
Figure 3.8: Image of a metal pin obtained M ith X fays impinging directly onto a bare CCD detector.
Light-tight Photomultiplier
housing
Filter
Scintillator
Figure 3.9: Geometrical setup of the scintiUator-photomultiplier combination used for online
monitoring of the X-ray flux. A filter was placed in front of one of the detectors in order to
simultaneously measure total and hard X-ray flux.
The scintillator-based system was absolutely calibrated according to a procedure
described in Section 3.3.4. This calibrated system was then used to measure absolute
conversion efficiencies from laser light into X-rays, as summarised in Section 5.3,3.
CCD detectors can also be used for direct imaging, as shown in Figure 3.8. However, this
technique was not used due to the low overall sensitivity to hard X-rays.
3.3.3 Scintillator-photomultiplier combination

A scintillator-photomultiplier combination was used for online monitoring of the X-ray
emission. The scintillator is similar to the one used in the scintillator-CCD camera
combination described in Section 3.3.2. It consists of a 10 mm x io mm large, 4 mm thick
terbium-doped fibre optical scintillator (LG-9 TAR, Schott Fibre Optics Inc., USA), which
is placed directly on a miniaturised photomultiplier tube (R5600U, Hamamatsu, Japan), as
illustrated in Figure 3.9. Two of these detectors are placed close together, with one of them
filtered by various amounts of copper foil. Simultaneous observation of both the total X-ray
flux and the hard X-ray flux can thus be achieved. This system was used in investigations of
0.0
50 100 150
Energy [keV]
Figure 3.10: Energy dependence of the CCD camera output signal on the incident photon energy.
the X-ray source as presented in Sections 4.2 and 4.3. The calibration of these detectors is
described in the next section.
3.3.4 Absolute calibration of the scintillator-based detectors

Quantitative measurements of the X-ray flux require absolute calibration of the detector
system. Therefore, the scintillator-CCD camera detection system was calibrated with the
help of reference radionuclide samples. It was then used to measure conversion efficiencies
from laser light into characteristic line emission (see Section 5.3.3).
The radionuclide samples were either placed directly onto the scintillator or mounted at a
distance of 90 mm. The resulting signal was integrated during 0.2 s - 1 s for the samples on
the scintillator, and during 30 s for the samples positioned away from the detector. Samples
of 24IAm, 57Co, B3Ba, 137Cs, 54Mn, 60Co and 22Na were used, with activities in the range
200 kBq to 480 kBq. The most relevant samples for the work presented here are the 241Am
and the 57Co samples, producing X-rays at predominantly 59.5 keV and 122.1 keV,
respectively. The photon energies and intensities of the different decay paths were obtained
from the Table of Isotopes Online Database [59].
The exact composition of the scintillating material could.not be obtained from the
manufacturer. However, some data were provided on the total absorption within the
scintillator (see Figure 3.7). Therefore, a simplified model of absorption and scattering was
constructed which fitted the given absorption data. By using these simplified interaction
coefficients together with the radionuclide sample emission data, theoretical values of the
signal could be calculated and compared with the corresponding experimental values. Using
this procedure, a photon-energy-dependent conversion factor was determined, which related
the number of counts in the CCD camera image to the number of absorbed X-ray photons. In
the range 10 keV to 80 keV, the CCD signal is approximately proportional to the energy of
the incident photon. At higher energies, the decrease in absorption leads to a slower increase
in signal, as shown in Figure 3.10. The scintillator-photomultiplier combination was
Electron
focussing Time sweep
optics electrodes
X-rays <l CCD

readout
Extraction Vacuum housing Output

mesh phosphor
screen
Figure 3.11: Schematic view of the different parts of a streak camera. Note that the whole system
(apartfrom the readout system) is under vacuum.
calibrated in the same way. It was thus possible to measure the X-ray flux with this detector
on a single-pulse basis.
The error in this calibration procedure is rather large. The radionuclide samples
themselves are calibrated with an accuracy of ±5%. Residual errors of about ±15% were
obtained in the adjustment procedures described above. Additional systematic errors result
from several assumptions. During calibration, all the energy absorbed in the scintillator is
taken into account, i.e. also Compton-scattered radiation which is absorbed at different
positions in the detector after the first scattering event. This leads to problems when
measuring the X-ray intensity of very localised signals, e.g. X-ray spectra in the detection
plane of a crystal spectrometer (see Section 5.3.2). In this case, the Compton-scattered
photons will produce a diffuse background which is difficult to evaluate. This will lead to an
underestimation of the X-ray intensity of several percent. At photon energies below 10 keV,
the scintillator becomes less efficient than assumed in the simple model. This could be due
to an intrinsic decrease in scintillation-light production at low X-ray photon energies [60].
3.3.5 Time-resolved X-ray detection

An X-ray-sensitive streak camera was utilised for time-resolved detection of the laser-
produced X-rays. A schematical overview of a streak camera is shown in Figure 3.11.
X-rays impinge on the front of a thin cathode where they generate electrons. These electrons
then exit from the rear of the cathode where they are accelerated in a strong static electric
field. A geometrical image of the cathode (normally with a narrow line shape) is formed on
the output phosphor screen by means of static field electron optics. Time resolution is
achieved by using additional deflection electrodes, which sweep the electrons across the
output screen. The optical image generated on the phosphor screen contains both one-
dimensional spatial data (along the cathode) and temporal data (along the sweep direction).
The temporal resolution is determined mainly by two factors, the sweep speed and the
geometrical width of the cathode image (spatial width transforms into temporal width via the
sweep speed).
The X-ray streak camera used in this work (Kentech Instruments Ltd., UK) allowed for
the insertion of custom-made cathodes. Different cathodes were made and tested for their
time resolution and hard X-ray detection capabilities. Caesium iodine was chosen as the
cathode material because of its high secondary electron yield [61]. The cathode was made by
sputtering in vacuum, either with or without argon buffering gas, for low-density or near-
solid-density Csl cathodes, respectively. Low-density cathodes were not further used after
some initial trials, due to sputtering reproducibility problems and mechanical instability of
the cathodes.
The cathode sensitivity can be partly optimised for a given X-ray photon energy by
choosing an appropriate cathode thickness [62]. The escape length of secondary electrons in
solid-density Csl is of the order of 25 nm [62], so that no significant increase in electron
output should be expected when increasing the cathode thickness above a few escape
lengths, i.e. ~ 0.3 urn. However, empirical observations on the signal strength (which is
directly coupled to the electron output) were made with the home-made Csl cathodes,
showing a signal increase up to a cathode thickness of a few micrometres. This might be
explained by a lower density of the home-made cathodes, resulting in an increase in the
secondary electron escape depth.
Thin brass foils (50 um) were used as cathode substrate. Before reaching the Csl cathode,
the X-ray had to traverse the brass foil. This resulted in an effective soft X-ray filtering for
photon energies below 20 keV. Another advantage of the optically opaque brass foil is, that
no visible straylight enters the streak camera. The minimum achievable cathode width was
about 0.9 mm, which corresponds to a temporal contribution of about 12 ps when working at
the fastest streak speed (10.8 ps/mm). Smaller cathode widths, corresponding to about 3 ps,
were obtained using a different technique, where the Csl was deposited on free-standing thin
wires. However, the sensitivity of these cathodes was much lower than for the brass-foil
based cathodes.
It was difficult to measure accurately the time resolution of the streak camera, because no
ultra-short reference pulses, as are usually present in the optical or VUV domain, were
available in the hard X-ray energy region. The X-ray pulses from the laser-produced X-ray
source had to be used as their own reference, which was rather unsatisfactory. When using
the smallest available cathode mentioned above (3 ps contribution), the efficiency was much
too low for single-pulse recording. Integrating over many X-ray pulses resulted in pulse
widths of 20 ps. The internal trigger jitter of the streak camera is assumed to give the
dominant contribution. In addition, it is unclear to what extent processes within the cathode,
e.g. secondary electron scattering, contribute to the temporal broadening of the measured
signal. Due to this limitation in time resolution, it was not possible to determine the X-ray
pulse duration below the upper limit of 20 ps. This is significantly longer than the expected
pulse duration, which is of the order of only a picosecond [48].
27
4. SOURCE CHARACTERISATION
The X-ray source described in this thesis had to be characterised and stabilised as well as
possible in order for it to become a useful tool in, for example, basic physical research or
medical imaging. Important characteristic parameters are the X-ray source size, the emission
spectrum, the X-ray pulse duration and the X-ray flux. These properties depend on various
experimental parameters, e.g. target material, laser pulse characteristics and ambient gas
pressure in the vacuum chamber. Additionally, they can vary with the energy of the emitted
photons.
The X-ray source properties can be partly chosen according to the user's needs by an
appropriate choice of these experimental parameters. This is especially important when
certain properties are to be optimised for special applications, e.g. the source size for
magnification imaging or the X-ray spectrum for contrast imaging. However, there will be
limitations when trying to optimise the different properties simultaneously.
Various characterisation experiments on different properties of the X-ray source were
carried out, as described below.
4.1 X-ray source size

Different methods exist for the determination of X-ray source sizes, e.g. imaging of star
patterns or fixed line patterns, knife-edge and wire imaging methods [63], and pinhole
imaging [64]. The size of the laser-produced X-ray source from a tantalum target had
previously been measured to be smaller 60 um, using a lead star-test pattern with a
maximum spatial frequency of 10 linepairs/mm [16]. Attempts to measure the size with
knife-edge methods (as is common in soft X-ray or optical source size determination) turned
out to be unsatisfactory, basically due to technical problems in aligning knife-edges of
sufficient thickness.
During the course of the present work, the source size was further investigated using an
X-ray pinhole setup. A hole of diameter 5 \im in a platinum substrate (175 um thick) was
used. A typical image for a tantalum target, obtained with the scintillator-CCD system
described in Section 3.3.2, is shown in Figure 4.1. The pinhole was placed at a distance of
20 mm from the source, and the magnification factor was about 34. The X-ray beam was
filtered with 10 mm of opaque plastic (~10% transmission at 30 keV). From these images,
the width of the X-ray source was determined. The measured widths varied with, for
example, focusing conditions [65]. Minimum values of about 33 um were obtained. The
intensity profile of the source might give additional information on plasma processes as far
as they manifest themselves in intensity distribution changes.
The pinhole measurements were supported by additional measurements with a lead star-
test pattern with a maximum spatial frequency of 25 linepairs/mm. From these
measurements, a source size of 45 ]um was deduced [65]. The source size did not change
upon varying the laser pulse energy within a factor of 5, neither upon a change in prepulse-
to-main-pulse ratio from about 10"6 to 5-10". The average detected photon energy could be
28 Source Characterisation
j%\
ensit y |arb. un its]

j \ FWHM: 33 [im
> \
•
c «
-150 -100 -50 0 50 100 150

Position [urn]
Figure 4.1: X-ray source image, with an intensity profile showing a width of about 33 ftm, at an
average photon energy of about 40 keV.
changed by placing different filters in front of the detector. No change in source size was
observed for average photon energies in the range 30 keV to 70 keV . However, an increase
in X-ray source size, to about 80 um, was observed occasionally without satisfactory
explanation.
4.2 X-ray flux

Dependence on laser-pulse parameters
The X-ray flux from a tantalum target was measured as a function of different laser
parameters, e.g. pulse energy and prepulse-to-main-pulse ratio. These measurements were
performed on a single-pulse basis, measuring all parameters for every laser pulse. The X-ray
flux was measured with the scintillator-photomultiplier combination (see Section 3.3.3). One
of the detectors was filtered with a copper foil (1.3 mm thick), having 20% transmission at
60 keV. The other detector was not filtered, apart from the effect of a plastic housing (1 mm
thick).
A typical dependence of the hard X-ray flux, Ix-hard, on the laser pulse energy, IL, is
shown in Figure 4.2. This dependence can be described by a power law of the form:
' X-hard (6)

However, at higher laser-pulse energies, the hard X-ray flux seems to increase slower with
increasing laser-pulse energy, an effect which has also been reported in the literature by
other groups [66]. Different power laws have been reported in the literature, for experiments
performed with aluminium targets, with values of the exponent of about 2.4 [66].
Differences in laser parameters, target setup and detector calibration can partly explain the
difference obtained in the present work. The hard X-ray flux increases faster than the total
X-ray flux, Ix-wtai, (shown in Figure 4.3), with
Source Characterisation 29
1 '
ux [arb. unit
5 5,
*
2 •
I 10 " 3 '
• ! •'•*••'
X • .' '' '••:

. ' ; *•'. •
in-.
10 100 1000
Laser energy on target [mJJ
Figure 4.2: Hard X-ray flux from a tantalum target as afunction of laser-pulse energy.
i" 10'1
1O" 2 1
1O" 3 1
1 10--!
10"
10" 10"** 10° 10" 10"' '
Total x-ray flux [arb. units]
Figure 4.3: Hard X-ray flux versus the total X-ray flux, upon varying the laser-pulse energy
(measurements with a tantalum target).
T A11 (7)
1 J
X-hard ~ X-total
The prepulse-to-main-pulse ratio was also found to affect the X-ray yield. An increase in
X-ray yield with increasing prepulse-to-main-pulse ratio was observed, for a prepulse-to-
main-pulse ratio in the observed range 5-10"6 to 8-10'5. This is shown in Figure 4.4 on a
single-shot basis.
Dependence on ambient gas and pressure

The influence of ambient pressure was measured and is shown in Figure 4.5. The X-ray flux
did not change significantly at pressures below 5 hPa. At higher pressures, the X-ray flux
decreased with increasing pressure, probably due to ionisation-induced defocusing of the
-
»°o8» *!
3 o o o 00 oO °
< o
go o J*
-
o o o
o
e o
* ' • . .J ^ »«P>« 0 0 o
•
i2H
o
•
^ •: -3:- 0
0 •+««--«-«
5-10 10-10
Prepulse-to-main-pulse ratio
0.0
5-10 10-10""
Prepulse-to-main-pulse ratio
Figure 4.4: Total (upper) and hard (lower) X-ray flux from a tantalum target versus prepulse-to-main-
pulse ratio. The lines shown are only for guiding the eye.
focused laser beam and electric breakdown in air just before reaching the target. The X-ray
yield was also measured for different ambient gases (helium, nitrogen and air), and no
significant change in X-ray flux was found.
Dependence on target position

The dependence of the X-ray flux on the target position was measured. Such measurements
are shown in Figure 4.6 for a tantalum target. Both at relatively soft X-ray energies and at
very hard X-ray energies the curve has a full width at half maximum of about 80 um. The
target surface must be held at the focus position with a better precision. The regulation
mechanism (see Section 3.2), with its residual position error of ± 3 um at maximum, met this
requirement.
20 40 60 80 100
Presure [hPa]
Figure 4.5: Total X-ray flux from a tantalum target versus ambient pressure (air) in the vacuum
chamber. A guiding-the-eye line is also shown, based on a linear fit to averaged data points.
Cu: 0.5 mm ; Pb:5mm

, ,- Hi %.
units]
1 1 , • 111
.£
JL
$ • •
i.
2?
w •• I-1 •'
r- : 1.
ly mtensi
• & *
><
-150 -100 -50 0 50

Si,100 150 20 -150 -100 -50 0 50 100
Target Position [fjm] Target Position [pm]
Figure 4.6: X-ray flux as a function of target position for two different filters corresponding to X-rays
above -35 keV (a) and above ~ 100 keV (b). Data were taken during ~ 17 target rotations.
Dependence on target structure

Under normal conditions, a fresh target surface was exposed to every laser pulse. However,
it was found empirically, that reusing the same track could significantly enhance the X-ray
flux. In addition, this effect showed a clear dependence on the ambient gas pressure when
using foil targets. This led to more systematic investigations, in which a number of laser
pulses were focused onto the same target spot, while recording the X-ray flux for every
pulse. The results of these measurements are shown in Figure 4.7, for tantalum foil targets,
and in Figure 4.8, for solid tantalum targets, for different ambient pressures.
— 10 --r
3 5 7 9 11 13 15 3 5 7 9 11 13 15
Number of laser pulses Number of laser pulses
1 3 5 7 9 11 13 15 1 3 5 7 9 11 13 15
— 10 -
S 25hPa 40hPa
-
a
re 6 • 6- -
2
( l f& —
3 5 7 9 11 13 15 1 3 5 7 9 11 13 15
Figure 4.7: X-ray flux as a function of laser pulse number when focusing the pulses onto the same
target spot on a tantalum foil target. The X-ray flux scale is the same for all pressures.
Measurements for the foil targets show an immediate decrease in X-ray flux at pressures
below 15 hPa, whereas a signal maximum can be observed after a few laser pulses at higher
pressures. Hardly any X-ray emission is observed after about ten laser pulses. In contrast,
solid targets show very different behaviour. An increase in X-ray flux is observed up to
about 12 laser pulses, whereafter the signal decreases within about 50 pulses. No pressure
dependence of this behaviour was observed, in contrast to the study with the foil targets.
3 -
1 11 21 31 41 51 11 21 31 41 51
1 11 21 31 41 51 1 11 21 31 41 51
Figure 4.8: X-ray flux as a function of number of laser pulses when focusing the pulses onto the same
target spot on a solid tantalum target. The flux scale is the same for all pressures, but not comparable
to the scale in Figure 4.7.
The rapid decrease in X-ray flux with the foil targets can be attributed to a burn-through
of the 250 jam thick tantalum foil, thus reducing the volume in which electrons can
efficiently generate hard X-rays. The formation of deep craters was observed with
secondary-electron microscopy, as shown in Figure 4.9. Each laser pulse on a foil target
produced a crater of about 40 |im - 50 \im in depth, as measured with an optical microscope.
The pressure dependence is difficult to interpret. With the solid targets, no burn-through
effects would be expected, as the target is 3 mm thick. X-ray emission can therefore be
obtained even after a large number of pulses.
A possible explanation of the X-ray flux increase for the first laser pulses in the solid-
target case could be enhanced laser light absorption in the already existing crater. The laser
light would be captured in the crater by multiple reflections. Additionally, resonance
absorption would be increased compared with a flat target surface. As described in
Section 2.2, resonance absorption requires an E-field component of the laser pulse
perpendicular to the density gradient. This condition is fulfilled at many places in a crater
with irregular walls.
Figure 4.9: Craters formed by one laser pulse (left) and by 10 laser pules (right) on a tantalum foil
target (250 pm thick). Note the different scales in the images. For one laser pulse, the hole is about
70 /urn in diameter, wheras it is about 150 jxm in diameter after 10 laser pulses.
• : .*•/ ' "• *"v .*• i
Time [arb. units]
Figure 4.10: X-ray flux from a solid tantalum target, for two different target rotation speeds (based on
single-pulse measurements). The spacing between the craters was approximately 10 jxm and 400 pm
at low and high rotation speed, respectively.
With solid targets, the effects described above can have practical implications. By
drastically reducing the target rotation speed, each laser pulse was focused onto almost the
same spot as the preceding laser pulse. An increase in X-ray flux, similar to that described
above, was observed, as shown in Figure 4.10. An average increase in X-ray flux of a factor
of 3 was measured. No X-ray source size increase was observed when changing from normal
target rotation speed to slow rotation speed.
0.6
•g 0 . 4 -
'. -• • • ', •>'•'•'" ' " • ; »V
2 0.2 - -. . .* - . -
"E • ' • • • ' . • .

• • • ' .
0.0
100 125 150 175 200
Laser energy on target [mJ]
Figure 4.11: X-ray flux fluctuations versus laser energy.
4.3 X-ray flux fluctuations

The laser-produced plasma X-ray source exhibits fluctuations in X-ray flux, both on a pulse-
to-pulse level and on longer time scales. Various measurements were performed in order to
identify the origin of these fluctuations and to reduce them. Both the hard and the total X-ray
flux were measured using the scintillator-photomultiplier combination (see Section 3.3.3). In
most measurements, the hard X-ray flux was measured by filtering with 1.3 mm of copper.
Fluctuations in the hard X-ray flux were much less pronounced than in the total X-ray flux.
In the first place, mechanical wobbling of the target surface was expected to produce
short-term variations in the X-ray yield. Having implemented the target stabilisation system
(see Section 3.2) it could be shown that the X-ray flux was nearly constant within a target
position range of about 20 urn (see Figure 4.6). Target wobbling could thus be excluded as
the source of fluctuations, when the stabilisation mechanism was in operation.
Laser-pulse energy fluctuations were also suspected of making a significant contribution
to the observed fluctuations, due to the strong dependence of X-ray flux on the laser pulse
energy (see Figure 4.2). The scatter plot in Figure 4.11, based on single-pulse measurements,
shows the distribution of hard X-ray flux with the fluctuating laser pulse energy. No
correlation can be seen, apart from the overall increase according to Eq. (6).
Laser prepulse fluctuations are another possible source of X-ray pulse fluctuations.
However, the influence of the prepulses at 11.6 ns and at 3.6 ns before the main pulse were
measured, and no correlation could be found between their fluctuations and the X-ray flux
fluctuations.
The fluctuations could also be due to due fluctuations in the laser pulse intensity. In the
previous measurements, only the pulse energy was measured. Fluctuations in pulse duration
could lead to intensity fluctuations, even with constant pulse energy. The pulse-width
fluctuations were measured using two photodiodes, one measuring the energy of the laser
pulse, Eo, the other measuring the energy in a frequency-doubled pulse, E2(o. The ratio
U.D •
'.
i •
•gO.4-
JS. • •
| -
. ' ' '
Hard x-ra
to
o
f
'"'' '\'r. ,•'"••'

1
• •'• -. :
o
1
o
0.70 0.85 1.00 1.15 1.30

Ratio of pulse length to average pulse length
Figure 4.12: X-ray flux fluctuations versus laser pulse width fluctuations.
(Ec0)2/E2(o is proportional to the pulse width assuming Gaussian-shaped laser pulses with
constant energy [67]. Measurements of this type are shown in Figure 4.12, where a subset of
data points with laser pulse energies in a 5% region on either side of the average pulse
energy value is plotted. No correlation between the fluctuations in X-ray flux and the
fluctuations in laser pulse width could be observed.
Microscopic irregularities in the target surface, being different for every laser pulse,
might lead to fluctuations. This possibility was at least partly excluded by measuring
fluctuations from both polished and unpolished targets. No difference in the fluctuation
could be seen in these measurements.
Fluctuations were also measured with different gases (nitrogen, air and helium), at
different ambient pressures (0.1 hPa, 1 hPa and 20 hPa). The magnitude of the fluctuations
at different ambient pressures of air, measured as the standard deviation of the values
divided by their average value, is given in Table 4.1. At low pressures, the magnitude of the
fluctuations for the total X-ray flux had a tendency to decrease, whereas no significant
change was observed for the hard X-ray flux. Similar behaviour was observed with the other
ambient gases.
The increase in fluctuation with increasing pressure could be due to a ablated particles
remaining in the beam path. Particles of different sizes are ejected from the plasma for every
laser pulse. The movement of larger ablated particles (> 0.5 um) is hardly influenced by the
Table 4.1: Magnitude of fluctuations for different ambient pressures of air. The values given are for
about 500 laser pulses. Values are given for both total X-ray fluctuations and hard X-ray fluctuations.
Pressure Total X-ray flux Hard X-ray flux

0.1 hPa 30.4% 12.0%
1.0 hPa 32.9% 12.1%
20hPa 45.2% 12.5%
Figure 4.13: Ablated particles from a tantalum target, deposited on two different glass slides. Note the
different scales in the images.
change in pressure, whereas smaller particles (< 0.3 urn) are efficiently slowed down in an
ambient gas [68]. If the diffusion of such small particles is too slow (at high pressure), they
might remain in the beam path until the next laser pulse arrives and thus influence beam
propagation and focusing.
4.4 Ablation phenomena

Ablation phenomena have been studied by means of deposition patterns on thin (100 |um)
glass slides [Paper I]. These slides, intended to protect the sensitive parabolic mirror surface
from the debris ablated from the target, were placed approximately 15 to 20 mm away from
the target, parallel to the target surface. In a previous study, the type and size of the
deposited particles were determined with the help of scanning electron microscopy [25].
Measurable particle sizes ranged from nearly spherical sub-urn particles to relatively large
irregular fragments with sizes up to a few tens of um. Particle appearance on two different
glass slides is shown in Figure 4.13.
The appearance of relatively clear circular structures in the deposition patterns on the
glass slides motivated more systematic investigations of this phenomenon, using different
target materials [Paper I]. Two glass slides, on which the ring patterns are clearly visible, are
shown in Figure 4.14. In certain cases even the central spot shows a ring structure. In more
recently observed deposition patterns, the inner onset of the outer ring is very sharp
compared with its outer region. However, it must be pointed out, that the ring structures
were often less distinct than those shown in Figure 4.14.
Further studies have been performed using a scanning electron microscope, in an effort to
determine differences in the type of particles deposited within and outside the observed ring
structures. However, no difference in the type or the size of the particles could be found. The
observed structures must therefore be due either to very small particles, not resolvable with
the scanning electron microscope, or to a particle-size-independent pattern-generation
process.
.....
Figure 4.14: Images of two glass slides, showing distinct deposition patterns in rings around the
target surface normal. The white spots visible on the right side of the slides correspond to the position
at which the laser pulses passed through the glass slides and are attributed to laser cleaning effects.
The height of both images corresponds to 24 mm.
We have found no satisfactory explanation of this phenomenon. To our knowledge, no

similar observations have been made at the intensities which are normally used in ablation
studies (approximately 109W-cm'2 - 10!4W-cm~2. At these intensities, the angular
distribution of particle deposition can be described by [69]
F(3)~cosn(&), (8)
with n ~ 1 ..8 and S being the angle to the target surface normal.
It may be that the strong torroidal magnetic field created around the plasma
(see Section 2.2) is responsible for this deposition pattern, in analogy with a miniature mass
spectrometer.
39
5. SPECTRAL CHARACTERISATION
The X-ray emission spectrum is an important property of an X-ray source. For many
applications, the spectrum must be known for quantitative analysis. For example, the
estimation of the achievable contrast in imaging relies on knowledge of the spectral
distribution. In more research-oriented applications it is of interest to know absolute photon
numbers as a function of energy. Knowledge of the spectrum can provide information about
the plasma and the X-ray generation process, for example the electron temperature in the
plasma. Absolute measurements of the X-ray yield can be compared with theoretical
calculations in order to test theories.
Different spectral measurement techniques, both energy and wavelength dispersive, were
investigated in this work and are presented in the following sections. A different technique,
based on Kedge filtering [25,70], has previously been explored in Lund, but was not
employed in the work presented in this thesis. The choice of technique depends on a trade-
off between the achievable spectral range, energy resolution and acquisition time.
5.1 Energy-dispersive single-photon counting with conventional detectors

5.1.1 Principles
Energy-dispersive single-photon counting, e.g. with germanium or sodium iodide (Nal)
detectors, is a common technique for measuring X-ray and y spectra [71]. In germanium
detectors, single X-ray photons are absorbed in the semiconductor detector volume. A
number of electron-hole pairs are generated proportional to the absorbed photon energy. The
electron-hole pairs generate a voltage pulse with a height proportional to the energy of the
absorbed photon. In a Nal detector, single X-rays photons are absorbed in the detector and
generate a number of visible photons, proportional to the absorbed photon energy. The light
is detected in a photomultiplier. Again, the output signal is a voltage pulse with a height
proportional to the absorbed photon energy. The output pulses of such detectors are
registered in a pulse-height multi-channel analyser, whose function is to "sort" different
pulse heights, corresponding to different photon energies, into a number of energy bins.
After the detection of many photons a spectrum histogram is built up.
With this technique, a large spectral range extending up to the MeV region can be
measured simultaneously. However, the photon flux must be kept below a value given by
the registration time of the detector. If twp X-ray photons were absorbed within this time,
their sum-energy would be erroneously registered as a single photon event at higher
energies, an effect called pile-up. Therefore, the probability of absorbing more than one
photon at a time must be kept as low as possible. Applying single-photon counting to an
X-ray source driven by a 10 Hz laser system therefore necessarily limits the count rates to
significantly less than 10 photons per second. This leads to very long data-acquisition times,
of the order of hours.
Obviously, energy-dispersive single-photon counting is severely hampered by the intense
bursts of X-rays from the laser-produced plasma X-ray source. The photon flux from the
40 Spectral Characterisation
10'
>io 8
3
Q.
io 6
50 100 150 200
Energy [keV]
Figure 5.1: X-ray photon flux from a tin target, as measured with a germanium detector in Comptfew
geometry. The laser energy was 150 mJ on target.
source must be substantially reduced, so that the photon detection probability per laser pulse
is kept much smaller than one. In the investigations presented in Paper III, we used two
similar germanium detectors to simultaneously record the spectrum in two different
geometries. One detector was used in a Compton scattering geometry, whereas the other was
placed at a long distance from the X-ray source (which corresponds to a small solid angle),
in order to achieve the required reduction in X-ray flux.
The absolute efficiencies of both germanium detectors were calibrated using radionuclide
samples. Additionally, Compton scattering of the photons within the detector must be taken
into account. Considering monochromatic radiation, most of the incident photons will be
completely absorbed within the detector volume, producing an output signal proportional to
the photon energy. However, some of the scattered photons will escape the detector volume,
and consequently only a fraction of the incident energy is registered. This leads to an
erroneous spectral contribution at lower energies. For the direct detection geometry, the
contribution of this effect was measured with radionuclide samples and accounted for in a
suitable way [Paper III]. The detector used in the Compton geometry was corrected for this
effect by the manufacturer.
5.1.2 Experiments
Spectra of tin, antimony, tantalum and bismuth targets were recorded, both in the direct and
in the Compton geometry. The spectra for tin and tantalum, measured in the Compton
geometry, are shown in Figure 5.1 and Figure 5.2, for example. The different components of
the characteristic K emission, i.e. Ka and Kp lines, were resolved.
Absolute conversion efficiencies from laser light into characteristic line emission
(EX-K/EL) and into total X-ray emission above 50 keV (EX/EL) were measured by integration
of the spectrum. For the tantalum target, we found E X -K/EL to be about 6-10"6, while Ex/EL
was about 7-10"5. For the tin target, these numbers were about 8-10"6 and about 5-10"5,
Spectral Characterisation 41
10v
>io 8
-107
I io 6
50 100 150 200
Energy [keV]
Figure 5.2: X-ray photon flux from a tantalum target, as measured with a germanium detector in
Compton geometry. The laser energy was 150 mJ on target.
respectively [Paper III]. Hot electron temperatures could be determined by fitting the
exponential decrease of the spectrum at high photon energies [45,46]. Hot electron
temperatures of about 90 keV for tantalum and bismuth targets and of about 1 l.O.keV for tin
and antimony targets were determined. No characteristic radiation could be observed from
the bismuth target. This could be an indication that the electron temperature in the plasma
was not high enough to produce a significant excitation of bismuth K radiation (around
77 keV).
Different conversion efficiencies have been reported in the literature by other groups,
with values of 3-10'4 (>20keV, Ta target, EL=40 mJ, IL=1017Wcm"2) [18], 2-10"4 (MoKa,
EL=400 mJ, IL=5-1017 Wcm'2) [26] and 8-10"5 (TaKa, E L =43 J, IL=3-1017 Wcm"2) [66]. These
values are about a factor of 5 higher than the values presented in Paper III. Differences in
laser parameters, e.g. laser and prepulse energy, and detector calibration can partly explain
this deviation.
5.2 Energy-dispersive single-photon counting with CCD cameras

The long measurement time due to pile-up is the main disadvantage of conventional energy-
dispersive detectors when working with intense X-ray pulses at 10 Hz repetition rate from
our laser-produced plasma X-ray source. Many conventional detectors can be used
simultaneously to satisfy the need for fast, but still accurate, spectral characterisation. Each
of the detectors is allowed to detect at most one photon per X-ray pulse, but the number of
detectors compensates for this low counting rate. In practice, however, it is rather
inconvenient and expensive to use more than a few conventional detectors.
An alternative approach to overcome this problem is to treat every single pixel in a CCD
camera as an energy-dispersive detector. Instead of using the CCD for imaging, i.e. two-
dimensional intensity mapping, the energy absorbed in each pixel can be evaluated and
sorted into a histogram, which corresponds to the X-ray spectrum. The result is an an
X-ray X-ray
X-ray Photon photon
photon
Pixel electrode
structures
Charges are confined and' | Depletion

completely detected layer
Charges can diffuse
and recombine Bulk
(field free)
Figure 5.3: Schematic structure of a front-illuminated CCD detector. Incident X-ray photons generate
electron-hole pairs both in the depletion layer and in the bulk material, leading to differences in
charge collection.
X-ray
photon
Charges are confined ana =. Depletion

completely detected layer
Pixel electrode
structures
Figure 5.4: Schematic structure of a back-illuminated CCD detector. All X-ray absorption events take
place in the depletion layer. The depletion layer is often thicker than for front-illuminated CCDs.
overwhelming number of detectors; 3-105 to 2-107 detection elements for devices with
512x512 pixels and 4096x4096 pixels, respectively. This technique has been developed
mainly for astronomy and space-borne spectroscopic instruments operating below
10 keV [72,73], but it has also been used for synchrotron beam characterisation [74], neutron
detection [75] and plasma spectroscopy below 10 keV [76].
Within this work, the use of a CCD camera for spectroscopy in the hard X-ray region was
investigated. The principles of operation and some experimental examples are presented
below.
5.2.7 Principles
When an X-ray photon is absorbed within the CCD material (most often silicon), a number
of electron-hole pairs, proportional to the absorbed energy, are created (see Figure 5.3).
Recombination and spatial diffusion will be negligible, if these charges are created within
the depletion layer (active zone) of the CCD, and all charges will be registered within one or
a few pixels. If, however, the charges are created in the bulk substrate of the CCD (the
inactive zone), they may diffuse, leading to partial recombination. Only a certain fraction of
the charge generated will diffuse into the active zone and be registered, often smeared out
over several pixels. The amount of recombination and diffusion depends on the depth at
which charge creation occured. This effect will lead to an erroneous spectral redistribution
towards lower energies.
In back-illuminated CCDs, the effect of diffusing charges can be almost entirely
eliminated. In such detectors, the bulk material is etched away, and the radiation illuminates
the detector "from the back". In this way, the depletion layer is directly exposed to the
incoming radiation, as illustrated in Figure 5.4. Back-illuminated CCD detectors are often
used in soft X-ray imaging [77] or dental imaging [78].
A well-known problem in energy-dispersive X-ray spectroscopy is Compton escape,
which can be severe in the case of CCD spectroscopy. At energies above 52 keV, the
scattering cross section of silicon becomes larger than the photoelectric absorption cross
section, and scattering is by far the dominant interaction at even higher energies. In a small
detector volume like a CCD pixel, a Compton scattered photon will deposit only a small part
of its energy. From the observer's point of view such an event will not be distinguishable
from the photoelectric absorption of a photon with this lower energy. This phenomenon
results in an erroneous decrease in spectral intensity at higher photon energies. This problem
is not so pronounced in conventional detectors, which have much larger detection volumes.
In those detectors the scattered photon will most probably be completely absorbed
somewhere else in the detection volume.
The problem of low overall efficiency in a CCD detector, due to the small thickness of
the absorbing silicon, can be solved by using efficiently absorbing materials, e.g. CdTe, for
CCD fabrication [79].
5.2.2 Experiments
A typical image upon exposure of a front-illuminated CCD detector (TH 7883, Thomson,
France, 576><378 pixels, pixel size 23x23 urn2) to one X-ray pulse from the laser-produced
plasma is shown in Figure 5.5. Each group of bright pixels corresponds to an X-ray
interaction within the CCD.
In order to recover spectral information from images such as the one shown in Figure 5.5,
an algorithm was written for image processing. In this algorithm, all pixels belonging to an
absorption event are grouped together. The integrated intensity in each event was calculated
and stored in a histogram for further data analysis. The pixel intensity distribution obtained
in this way had to be corrected for the energy-dependent absorption efficiency of silicon
(assuming a depletion zone depth of 2 jam).
Reconstructed spectra for different target materials are shown in Figure 5.6, for
measurements with a front-illuminated CCD detector (TH 7883, Thomson, France). For
each spectrum, about 60 frames were taken, each being exposed by two X-ray pulses, with
an average event rate of 500/pulse. This corresponds to an average event rate of
approximately 5-10"3 events/pixel, so no pile-up problems were encountered. Compared with
a conventional energy-dispersive detector (at 0.1 events/pulse), the data acquisition rate is
increased by a factor of 5 000.
Figure 5.5: Image obtained upon irradiating a CCD detector with one X-ray pulse from the laser-
produced plasma source (only half a frame is shown). Every bright spot corresponds to an X-ray
absorption event. The magnified detail shows that each absorption event can be smeared out over
more than one pixel.
50 100 150 50 100 150

Energy [keV] Energy [keV]
- Sn Target | - Cu Target |
i.. snK &K B -
a .Ml
-
'c
xi
3
/I
f -
w i
u
• * • *
3 V 1. I-
- u
- -I
50 100 150 50 100 150

Energy [keV] Energy [keV]
Figure 5.6: Reconstructed spectra obtained with a front-illuminated CCD detector for different target
materials (assuming a depletion zone depth of 2 /urn).
Energy calibration was accomplished by the use of different targets (characteristic line
positions) and absorption filters (Kedge positions). Absorption within the electrode
structure (see Figure 5.3) and the optical shutter was not taken into account, leading to the
intensity decrease below 20 keV.
50 100 150
Energy [keV]
Figure 5.7: Reconstructed spectra obtained with a back-illuminated CCD detector for a tantalum
target (assuming a depletion zone depth of 10 nm).
The intensity decreases faster than in spectra recorded with standard energy-dispersive
detectors [Paper III] or crystal spectrometers (see Section 5.3). This is most likely due to the
above mentioned problems of Compton scattering and incomplete charge collection from the
depletion zone.
The spectrum from the tantalum target was also measured with a back-illuminated CCD
detector (SX-TE/512-TBK, SITe, USA) and is shown in Figure 5.7. The use of the back-
illuminated CCD detector clearly results in a better-resolved spectrum, compared with the
spectra obtained with the front-illuminated CCD.
The problems encountered with the CCD spectroscopy technique could partly be
overcome by increasing the absorption of X-ray photons. Specialised back-illuminated CCD
detectors have been recently developed by another group, in order to improve the absorption
efficiency and to reduce the influence of scattered radiation [80]. The depletion zone was
made significantly thicker (about 250 |um), and the pixel size was increased to
150x150 um2. These detectors are intended to be used as high-efficiency detectors in space-
born imaging spectrometers in the energy range up to 20 keV.
Another approach is the use of an effectively absorbing scintillator with good spatial
resolution, for example a fibre-optic plate. This approach is similar to the concept used in
Nal detectors (see Section 5.1.1). The spatial resolution allows for the simultaneous
detection of many X-ray photons, as with the CCD detectors. By measuring the amount of
scintillation light in each absorption event, a histogram of photon energies can be built up.
However, the Compton-escape problem has not been solved and the measurements are
deteriorated in the same way as in the CCD spectroscopy measurements.
Experiments were carried out using the scintillator-CCD camera described in
Section 3.3.2. It was found that the advantage of the fibre-optic plates, i.e. the high spatial
resolution obtained by guiding scintillation light through fibres, prevented its use in
spectroscopic measurements. This is most probably due to internal absorption of the
scintillation light, e.g. in the fibre walls. The amount of light reaching the CCD detector
depends on where in the fibre the X-ray photon absorption occurred. This leads to a massive
broadening of any spectral feature.
5.3 Crystal spectroscopy

Crystal X-ray spectroscopy is a well-established technique. During recent years, different
types of crystal spectrometers have been used for plasma X-ray spectroscopy, mostly in the
energy range up to a few keV. Bent crystals have been increasingly used, both for X-ray
focussing and monochromatic X-ray imaging. Although this technique had been used in the
hard X-ray region up to the MeV range [71], its importance in this energy range was
drastically reduced with the development of energy-dispersive solid-state detectors in the
1950s.
Crystal spectroscopy is not hampered by pile-up problems due to high photon fluxes
from a laser-produced plasma, in contrast to energy-dispersive single-photon counting. Data
acquisition times can therefore be drastically reduced when working with laser-produced
plasma X-ray sources. This was the main reason for investigating crystal spectroscopy of
hard X-rays from a laser-produced plasma in this work.
Additionally, crystal spectroscopy offers a substantial increase in resolving power.
However, there is a trade-off between the observable spectral range and the spectral
resolution. Usually, the simultaneously observable spectral range in crystal spectroscopy is
smaller than for energy-dispersive single-photon counting techniques.
The work on crystal spectroscopy was performed in collaboration with the X-ray Optics
Group, Institute of Optics and Quantum Electronics, University of Jena, Germany.
5.3.1 Principles
X-ray radiation with a wavelength A will be reflected from lattice planes with a spacing d in
reflection order n if the Bragg angle 6B, which is the angle between the incoming radiation
and the reflecting lattice planes, fulfils the Bragg equation,
n • A = d • sin 6B (9)
Monochromatic radiation will be reflected within a small but finite angular region around
6B- This angular dependence of the reflection of monochromatic radiation is described by the
reflection curve (also called the rocking curve). For flat crystals of the type used in this
work, the width of this reflection curve is of the order of a few arcseconds. Integrating the
area under the reflection curve gives the integrated reflectivity. This is a measure of how
efficiently a certain wavelength is reflected for a given crystal reflection (i.e. a set of lattice
planes with a given orientation). The integrated reflectivity can be significantly increased by
bending the crystal. This is due to the increasing penetration depth into the crystal and the
corresponding broadening of the reflection curve. The energy dependence of the integrated
reflectivity for different crystals, reflection orders, bending radii and spectrometer
geometries is described in Paper II.
Two different spectrometer geometries were used, the Johann geometry [81], illustrated
in Figure 5.8, and the Cauchois geometry [82], illustrated in Figure 5.9. A cylindrical
Detector plane
X ra
X-ray point " y focus
source (polychromatic)
Figure 5.8: Crystal spectrometer with a bent crystal in a Johann geometry setup (reflection case).
Bentcrystal
X-ray focus
(polychromatic)
Detector
plane
X-ray point
source
Symmetric
reflection Asymmetric
reflection
Figure 5.9: Crystal spectrometer with a bent crystal in a Cauchois geometry setup (transmission
case). For symmetric reflections, the reflecting lattice planes are normal to the crystal surface. For
asymmetric reflections, the reflecting lattice planes are inclined by an asymmetry angle, s, with
respect to the crystal surface normal.
bending of the crystal was achieved with the crystal bender shown in Figure 5.10, in
combination with the triangular shape of the crystal. The integrated reflectivity in the
Cauchois geometry can be further increased by using an asymmetric reflection, i.e. a
reflection on lattice planes that are inclined to the crystal surface by an asymmetry angle, s,
(see Figure 5.9).
The integrated reflectivity of the crystal as a function of photon energy must be known
for measurements of the absolute photon flux from the X-ray source. As an example, the
integrated reflectivity of a silicon crystal (reflection 111) with an asymmetry angle s=15°,
bent with a radius of curvature of R = 5 m (parameters used in the experiments described
below), was calculated by G. Holzer [83] and is shown in Figure 2.1 . Certain assumptions
in these calculations lead to a decrease in accuracy with increasing photon energy, with error
margins of about 10% at 80 keV.
5.3.2 Experiments
In the first experiments, spectra of tantalum Ka,P and tantalum La,p radiation were
recorded in Johann geometry with a bent germanium crystal, and in Cauchois geometry with
a flat silicon crystal, as described in Paper II. These measurements showed well-resolved
characteristic radiation from the solid target material. However, the detector sensitivity was
Figure 5.10: Crystal holder and bending mechanism used in this work. A silicon crystal is mounted in
the apparatus. By displacing the tip of the triangularly shaped crystal towards the reader, the crystal
can be cylindrically bent.
100-
_ 75 -
•o
2
50
¥ •
25 -
50 100 150
Energy fkeV]
Figure 5.11: Integrated reflectivity as a function of energy for the crystal used in the experimental
work (silicon, 1 mm thick, reflection 111, bent to R = 5 m, asymmetry angle e= 15°).
fairly low in these experiments, and the shape of the Bremsstrahlung contribution was thus
difficult to observe. A tantalum K spectrum is shown as an example in Figure 5.12, obtained
with a bent germanium crystal in Johann geometry and recorded with a scintillator-CCD
camera detector. The various components of the K emission, i.e. the Kot dublett and the
Kp\3 line, are well resolved.
100 -
-22
1 75
"
50 55 60 65 70 75
Energy [keV]
Figure 5.12: Tantalum K spectrum obtainedfrom a bent germanium crystal (reflection 115, R=6.2 m)
in Johann geometry, recorded with a scintillator-CCD camera detector (adaptedfrom Paper II).
Experiments were recently carried out with the upgraded laser system and an improved
and calibrated detection system (see Sections 3.3.2 and 3.3.4). X-ray emission spectra were
measured using a bent, asymmetrically cut silicon crystal (s=15°, reflection 111) in
Cauchois geometry and a bent germanium crystal (Paper II). Knowing the sensitivity of the
detector and the calculated integrated reflectivity of the silicon crystal (see Figure 5.11), the
absolute photon flux from the source could be determined [84].
Calibrated tin and tantalum spectra, obtained with the asymmetrically cut silicon crystal,
are shown in Figure 5.13 and Figure 5.14. The observable spectral range was increased
compared with the earlier measurements [Paper II] by the use of a wider collimator. The
upgraded laser system was used for these measurements. The bent germanium crystal was
used in Johann geometry for measurements of highly resolved spectra. As an example, a tin
Kot spectrum is shown in Figure 5.15. However, only a very small spectral range can be
simultaneously observed in this case. As in the measurements with the energy-dispersive
detectors, no characteristic lines were observed from the bismuth target.
The spectra obtained with the tin target showed a very interesting feature at 26.3 keV
(see Figure 5.13), which coincided with the Lyman a emission of highly ionised, hydrogen-
like tin atoms [83]. This was a strong indication of a rather hot plasma, in which a noticeably
large population of highly excited tin exists.
20 25 30 35
Energy [keV]
Figure 5.13: X-ray photon flux from a tin target, as measured with the asymmetrically cut silicon
crystal (reflection 111, R = 5 m). The laser energy was 250 mJ on target. Notice the indicated spectral
feature at 26.3 keV, which coincides with Lyman a emission from hydrogen-like tin.
20 40 60 80 100
Energy [keV]
Figure 5.14: X-ray photon flux from a tantalum target, as measured with the asymmetrically cut
silicon crystal (reflection 111, R = 5 m). The laser energy was 300 mJ on target.
510"
Kc
f 4-104
• \ T m
\
I/ "
Ka2
1
1
2-10'
•a
1 110" -(
a
24 24.5 25 25.5 26
Energy [keV]
Figure 5.15: Highly resolved Ka spectrum from a tin target. The spectrum was measured with a bent
germanium crystal (reflection 115, R = 8 m). The number of detected photons is shown. The laser
energy was 350 mJ on target.
5.3.3 Conversion efficiency and hot electron temperature

The conversion efficiency of laser light into characteristic X-ray radiation was determined
from the calibrated spectra. The number of photons emitted in the characteristic lines was
obtained by fitting Lorentzian profiles to the spectral line shapes. As an example, the
number of photons emitted as characteristic radiation and the corresponding conversion
efficiency for the spectra shown in Figure 5.13 and Figure 5.14 are summarised in Table 5.1.
The conversion efficiency was determined at different laser pulse energies and different
levels of prepulse intensity. The conversion efficiency showed a tendency to increase with
laser energy. However, this increase was within the experimental errors. An increase in
prepulse-to-main-pulse ratio from about 10'5 to 10"4 did not influence the conversion
efficiency outside the experimental uncertainty, either.
An exponential decay was fitted to various spectra from tantalum targets, in the energy
range of 70 keV to 100 keV, in order to determine the hot electron temperature. The
resulting hot electron temperature was approximately 50 keV, and showed a tendency to
increase with increasing laser energy. Again, this increase was within the experimental
errors.
Within the collaboration with the X-ray Optics Group in Jena, numerical simulations
were carried out by P. Gibbon in order to estimate theoretical conversion efficiencies,
characteristic photon numbers and hot electron temperatures for the parameters used in our
experiment [85]. For a tantalum target and experimental parameters similar to those
corresponding to Figure 5.14 and Table 5.1, these simulations predicted approximately 1-109
Ka photons, which should be compared with the measured value of 6.2-108. The
corresponding values for the tin target were 4-109 Ka photons (simulation) and 1.3-108
Table 5.1: The absolute number of photons emitted as characteristic radiation and the corresponding
conversion efficiency of laser pulse energy into characteristic radiation energy, for the spectra shown
in Figure 5.13 (tin) and Figure 5.14 (tantalum).
Spectral Line Photon Flux [l/(pulse-27csr)] EX/EL EL [mJ]

SnKa 1.310 s 2.2-10-6 250
SnKP 0.810 s 1.5-10-6 250
TaKa 6.2-10s 1.910"5 300
TaK(3 2.0-108 0.7 10'5 300
(experimental), exhibiting a rather large discrepancy. The hot electron energy distribution
for the tantalum target, calculated for 380 mJ laser pulse energy focussed to 1018 W-cm"2,
and a gradient scale length of L = 2-X, could be described by a two-temperature Maxwellian
distribution, with one temperature of about 50 keV and the other around 180 keV. The
50 keV contribution was dominant up to about 100 keV. The experimentally determined
values of the hot electron temperature are in agreement with the values obtained by
simulation. It must be noticed, that the simulation results significantly depend on the
prepulse parameters.
The conversion efficiencies obtained by crystal spectroscopy agree, within the
experimental uncertainty, with the values obtained by energy-dispersive spectroscopy
[Paper II]. However, the values of the hot electron temperature disagree by a factor of 2-3.
This discrepancy can be partly explained by the two-temperature Maxwellian hot electron
distribution. The slopes of the spectra from the energy-dispersive measurements were
evaluated in the energy region from about 80 keV to 200 keV, which is the transition region
between the two temperatures. Thus, the two temperature components may be mixed. In
contrast, with crystal spectroscopy the slopes of the spectra were evaluated in the photon
energy region below 100 keV, where the 50 keV temperature component dominates.
53
6. COMPARISON WITH OTHER X-RAY SOURCES
A large variety of different X-ray sources exists and more are currently being developed.
Conventional X-ray tubes are predominantly used in all kinds of medical radiology, e.g.
planar imaging and computed tomography (CT) imaging. Due to continuous development
these devices have reached a degree of reliability and ease of use which is hardly matched by
any other X-ray source [86]. Various types of X-ray tubes exist, suitable for the different
requirements on resolution and average flux. The standard imaging X-ray tube, which has an
X-ray source size of 300 (am -1200 urn, is used for most kinds of imaging tasks and delivers
relatively high average doses. Mammography X-ray tubes and microfocus tubes, with their
smaller source sizes, meet the demand for detection of small structures, e.g.
microcalcifications or microlesions. Some typical values for X-ray source size and relative
average flux for these different tubes are shown in Table 6.1, including a comparison with
the laser-produced plasma X-ray source discussed in this thesis. Quite obviously, the
average X-ray intensity is fairly low for the current laser-produced plasma X-ray source.
However, recently developed high-intensity short-pulse laser systems with 1 kHz repetition
rate might significantly improve this situation. The X-ray source size is comparable to that
of current microfocus tubes. A decrease in X-ray source size down to about 10 [im should be
possible for the laser-based source by improvements of focusing elements and better control
of laser parameters.
There are characteristic differences in the spectra of a laser-based X-ray source and
conventional X-ray tubes. At high photon energies, the former has an exponential intensity
decrease as a function of photon energy, determined by the plasma electron temperature
(see Section 2.3). Instead, the X-ray tube spectrum has a high-energy cut-off, determined by
the electron acceleration voltage applied to the X-ray tube. The high-energy tail does not
contribute to the imaging information although it increases the patient absorbed dose.
However, for non-medical applications, e.g. in nuclear physics, this high-energy tail can be
of great interest [44].
X-ray tubes can be modified to deliver very short X-ray pulses. Such flash X-ray sources
deliver pulses with durations down to about 20 ns, which is substantially shorter than
conventional pulse durations of several milliseconds to seconds. The spectral range is about
5 keV -100 keV, at repetition rates below 60 Hz. Exposures of about 1 R («9.5 mGy in soft
Table 6.1: Comparison of different X-ray tube systems and the laser-produced plasma X-ray source.
Source type Size Relative average flux

Laser-based X-ray source (10 Hz) ~ 10-30 um 1
Microfocus X-ray tube ~ 10 um -10
Mammography X-ray tube ~ 100 um -100
Standard X-ray tube 3 0 0 - 1200 jim -1000
54 Comparison with other X-ray Sources
tissue) per pulse are achieved, i.e. much higher than for the laser-produced plasma X-ray
source. Flash X-ray sources are used in technical contexts for the imaging of fast flows,
movements or impact studies [87-89]. However, their pulse duration is significantly longer
than the duration of X-ray pulses from a laser-produced plasma X-ray source based on sub-
ps laser systems. Laser-based short-pulse X-ray sources can, for example, be used to image
fast plasma-expansion processes, where a time resolution in the picosecond range is
required.
Synchrotron radiation from dedicated electron storage rings is used in a variety of
applications [5]. Synchrotron radiation is emitted when relativistically moving charges are
radially accelerated [6]. This happens in electron storage rings both in the bending magnets
and in dedicated devices consisting of periodically arranged magnets (wigglers and
undulators). Synchrotron radiation is emitted in a wide spectral range, from a few electron
volts up to the order of 100 keV in modern devices. Due to the temporal structure of the
electron flux in the storage ring, synchrotron radiation is typically emitted as short pulses of
the order of 50 ps duration, at high repetition rates up to several 100 MHz. A very
characteristic property of synchrotron radiation is the small divergence angle; of the order of
a few milliradians. Together with the small beam size, less than 1 mm , this leads to good
efficiencies in monochromators and focusing devices [90-92].
High-average monochromatic photon fluxes can be achieved in synchrotrons, required in,
for example, coronary angiography. In recent experiments, about 3-1011 photons-s"'-mm"2 at
33.2 keV, in a 140 eV bandwidth, were used, corresponding to about 33 Gys' 1 [93]. For
comparison, the average number of tantalum Ka photons emitted from the laser-produced
plasma X-ray source (at 10 Hz) was about 6-109 photons-s"1-(27israd)"'. However, the number
of photons in a single X-ray pulse from the laser-produced plasma can be significantly
higher than in a synchrotron radiation pulse. The small divergence of synchrotron radiation
is in sharp contrast to the wide angular emission into 2TC from a laser-produced plasma X-ray
source. In general, synchrotrons are used when good energy tunability and resolution as well
as high average photon fluxes are required, e.g. in soft X-ray spectroscopy, structure analysis
and surface studies [5]. Synchrotrons have also been used in medical imaging of contrast
agents [5,93,94].
There are several other X-ray sources which are the subject of current investigations.
Most of them are less relevant in the medical context, but have interesting properties for
basic research. Scattering of intense light pulses from relativistic electrons produces very
short X-ray pulses, of the order of 300 fs, with photon energies of about 25 keV. However,
the photon flux of about 105 photons/pulse is rather low in the reported experiments [9]. This
X-ray source could potentially create shorter pulses of hard X-rays than a laser-produced
plasma, at the cost of a significantly lower photon yield and an increased complexity.
Channelling of relativistic electrons passing through a crystal in appropriate directions
generates hard X-rays. Such electrons experience a periodically modulated environment,
similar to the situation in synchrotron radiation generation with wigglers or undulators.
X-ray radiation up to 20 keV has been experimentally observed [11]. Additionally, the
crystal can be periodically bent, e.g. by standing acoustic waves, resulting in a similar
Comparison with other X-ray Sources 55
phenomenon. In such a case, the production of coherent X-ray radiation up to 100 MeV has
been predicted theoretically [10].
In the soft X-ray region high harmonic generation [7] and laser-produced plasmas from
droplet-targets [95] are being investigated as convenient and effective X-ray sources. High-
harmonic radiation from ultra-short intense laser pulses has been generated up to photon
energies in the so-called water window (2.4 nm - 4.2 nm, i.e. 0.3 - 0.5 keV) [52], but only
with very low photon fluxes in this energy region. Radiation at about 100 eV can be
generated at photon fluxes of about 107 photons-pulse"1. High harmonic radiation is coherent,
with pulse widths shorter than the generating laser pulse, thus producing the shortest X-ray
pulse durations among the sources mentioned above. Recently, the generation of pulses in
the sub-femtosecond region by high harmonic generation has been proposed [96-98]. An
X-ray source based on a laser-produced plasma from a droplet target has been demonstrated
in the soft X-ray region just below 1 keV, with photon fluxes of 2-1012 photons-sr"1-pulse"1
and pulse durations below 100 ps [95]. This droplet-target source combines the advantage of
a high conversion efficiency (about 5%) with an extremely low debris emission, which is
otherwise a problem in solid-target plasma X-ray sources (see Section 4.4).
57
7. DIFFERENTIAL IMAGING
7.1 Principles
A common task in medical imaging is the selective imaging of a contrast agent, e.g. iodine
solution in coronary angiography. Conventional standard techniques involve digital
subtraction methods, where images of the region of interest are taken before and after the
administration of the contrast agent. Common problems with this technique are motion
blurring, due to patient movement, and the high concentrations of the contrast agent that are
needed to achieve the required contrast.
A technique which can circumvent this problem is differential imaging, also called
dichromography [99], which relies on the strong absorption change at the K edge of a
contrast agent. This technique can reduce the time delay between the exposures to be
subtracted, thus reducing motion blurring, and increase the sensitivity, thus reducing the
required concentration of the contrast agent. In differential imaging the source spectrum,
ideally monochromatic, is modified between two exposures, so that it samples the absorption
just below the K edge in one image and just above the K edge in the other image
(see Figure 7.1). The contrast agent is administered prior to imaging. This is in contrast to
conventional subtraction imaging, where one image is taken before and the other image after
administering the contrast agent. In this case, the absorption increase is purely due to the
administration of the contrast agent, without changing the source spectrum.
The contrast achievable in differential imaging is sensitive to the source spectrum. Quite
obviously, nearly monochromatic radiation just above and below the K edge is the optimum
10.0 q -
25 30 35 40 45
Energy [keV]
Figure 7.1: Schematical illustration of the change in absorption in differential imaging due to a
change in the source spectrum. Monochromatic radiation just below (A) and just above (B) the
absorption edge of the contrast agent is used for two separate exposures. The change in absorption
for the surrounding soft tissue is small compared with the absorption increase across the K edge of
the contrast agent. Thus soft tissue image information cancels in the subsequent image subtraction
58 Differential Imaging
3
Cg a) b)
1 Emission
C
\ ofGd Attenuation
o Attenuation
ISSI
E inGd inGd
•ation,
B
JV
£ / \
< /
0 40 20 60 80 1Q0 0 20 40 60 80 100
Energy fkeV] Energy [keV]
Figure 7.2: Comparison of the absorption of a gadolinium contrast agent to the modelled emission
spectra of gadolinium (a) and tantalum (b).
choice. Therefore, differential imaging has been mainly investigated using synchrotron
radiation [14,93], but conventional tubes have also been used together with crystal
monochromators [100].
The feasibility of differential imaging with X-ray radiation from a laser-produced plasma
was investigated and is presented in Paper IV. The change in the source spectrum, which is
required for differential imaging, can be easily achieved by changing the target material,
either by physically changing the target itself or by using composite targets with alternating
regions of two target materials. The contrast achieved in differential imaging is then mainly
a function of the source spectrum for the different targets.
7.2 Experiments
Most of the experimental investigations were carried out by imaging solutions of different
elements with known concentrations, which allowed for a comparison to theoretically
calculated values [Paper IV]. In these investigations, tantalum and gadolinium were used as
target materials, with contrast solutions having their K edges in the energy region of the
K emission of these target materials. The measurements are illustrated by the absorption of a
gadolinium solution together with modelled emission spectra of tantalum and gadolinium in
Figure 7.2.
The theoretical estimates of the achievable contrast were made using an X-ray-tube-like
source spectrum, and the results agreed fairly well with the experimentally obtained
values [Paper IV]. The Bremsstrahlung part of the source spectrum decreases the achievable
contrast compared with a situation in which only characteristic radiation from the target
would be used.
In another study, iodine was used as a contrast agent to obtain differential images in an
animal study. A rat bladder filled with iodine solution was imaged with X-rays from a
cerium and from an antimony target, with Kaj emission lines at 34.72 keV and 26.36 keV,
respectively, thereby bridging the iodine Kedge at 33.17 keV. The image obtained with
X-rays from the cerium target is shown in Figure 7.3, together with the differential image
Differential Imaging 59
pr.
A ' "
Figure 7.3: Image of a iodine-solution-filled rat bladder obtained with X-rays from a cerium target
(left) and the differential image (right) after subtraction of the image taken with X-rays from an
antimony target. Bright pixels correspond to high iodine concentrations. The differential image
reveals more information on the spatial distribution of the iodine solution than the image taken with
X-rays from a cerium target.
obtained by subtracting the image obtained with the antimony target from that obtained with
the cerium target.
Sef I BLUMK
61
8. TIME-GATED IMAGING
8.1 The scattering problem in radiology

Scattered radiation has always been of major concern in medical radiology, because it
reduces the contrast and the detectability of structures of interest. Image degradation due to
scattering is more pronounced when thick tissues and large areas are imaged. Experimentally
determined ratios between scattered and unscattered radiation are in the range 4 to 20 for
abdomen or spine examinations and 0,4 to 1.5 for mammographic examinations [101].
The advantages of scatter-reduced imaging are not only improved image quality (with
unchanged absorbed dose to the patient), but also a possible reduction in the absorbed dose
to the patient (while preserving the image quality). The reduction in dose would be a highly
appreciated feature. Numerous experimental and theoretical studies have been carried out in
order to quantify the amount of scattered radiation and its undesired effects (see
e.g. [102,103]). Monte Carlo simulations are an often used tool in this context (see
e.g. [104,105]).
Different approaches may be adopted to reduce the amount of detected scattered
radiation, all based on the different properties of scattered and unscattered (primary)
photons. Firstly, Compton-scattered photons have slightly lower energies than unscattered
ones, due to energy losses in Compton scattering (which is the predominant type of
scattering in the medically relevant photon energy range). Secondly, scattered photons are
less directional than unscattered photons from a small primary source. Finally, their transit
time through the imaged object is longer, as their pathlengths are longer due to scattering. In
principle, all these differences can be used to differentiate between scattered and unscattered
photons and thus to suppress scattered radiation.
Energy discrimination as a scatter-rejection method is based on the use of a
monochromatic X-ray source in combination with an energy-resolving detector. A
significant number of Compton-scattered photons can then be suppressed by setting the
detection threshold just below the initial photon energy [106]. The anti-scatter grid technique
makes use of the decrease in directionality and is commonly used in radiology. The principle
of scatter rejection with such grids is shown in Figure 8.1. Unfortunately, the collimating
grid structure also absorbs a significant proportion of the unscattered radiation (30%-
45% [101]), resulting in an increase in the overall X-ray dose required for an exposure.
Scatter-rejection concepts based on temporal discrimination have been intensively
investigated in the optical domain (see e.g. [107,108]). However, temporal discrimination in
the X-ray region was not realised for a long time, due to the lack of suitable short-pulse
X-ray sources and sufficiently fast detection techniques. With the recent development of
ultra-short-pulse laser-plasma-based X-ray sources, it became feasible to explore the
potential of time-gated imaging in the hard X-ray domain [27]. Time-gated imaging in the
hard X-ray region will be the topic of the rest of this chapter.
62 Time-Gated Imaging
X-ray Source
Tissue
Grid with
absorber strips
Image Detector
Figure 8.1: Principle of scatter rejection by an anti-scatter grid. Most X-rays coming from other
directions than straight from the source are geometrically blocked by absorber strips (e.g. lead strips
embedded in aluminium).
After transmission of a short X-ray pulse through a scattering medium, the temporal
shape consists of a short, unscattered component, followed by an extended tail of scattered
radiation. This is schematically shown in Figure 8.2. Using a gatable detector, with a
temporal acceptance window centred on the unscattered peak, the scattered radiation can be
efficiently suppressed. The integrated intensity in the long-lasting tail of scattered radiation
can be comparable to or even exceed the unscattered component.
An important issue when discussing improvements by scatter reduction is quantification
of the quality of radiological images, e.g. in terms of signal level, noise level and contrast of
the observed details. Therefore, a concept of image quality was introduced in Paper V which
takes both the contrast and the signal-to-noise ratio into account. For practical medical
applications, more subjective visibility criteria are often used [1], which are not applicable in
this context.
The scatter reduction performance of time-gated imaging was investigated both
theoretically and experimentally. The theoretical investigations were performed using Monte
Carlo simulations, which are described in Section 8.2. These simulations provided an
analysis of the influence of the most important factors in time-gated imaging, e.g., tissue
properties, photon energy and gating performance, without being limited by experimental
considerations. The experimental study is presented in Section 8.3, together with a summary
of the results. Rather good agreement was obtained between the image improvements
calculated by simulations and those observed in experiments.
Time-Gated Imaging 63
Input x-ray pulse
I
3 ray pulse
•e
es -
Unscattered
photons
Scattered tail
-+-
Gate window Time [arb. units]
Figure 8.2: Schematic illustration of the temporal structure of short X-ray pulses before and after
transmission through a tissue sample, showing the short-lived unseattered component and the much
longer tail of scattered radiation.
8.2 Monte Carlo simulations of time-gated imaging

8.2.1 Methods
Monte Carlo simulations were used to determine photon propagation in random-scattering
media. In the simulation, the trajectories of photons are sampled by random walks within a
model medium, depending on material constants and initial conditions. The photon is moved
in discrete steps, s, with the photon intensity distribution following a Beer-Lambert law with
a total interaction coefficient (j.totai given by:
f-t-total Mphotoabsorption "** Mincoherent scatter "*" ^coherent scatter (10)
At each interaction point, the photon is either absorbed or scattered. The relative
probabilities for scattering, Pscatt, and absorption, P abs , are given by:
s
_ 1 ~ exP(.~(Mcoherent + ^incoherent ) ' ) exp( /Jphotoabsorption ' s)
rp P
scatt ~ abs ~ (11)
1 - exp(-jutotai • s) 1 - exp(-// t0/a/ • s)
In the case of scattering, a further distinction into coherent scattering and incoherent
(Compton) scattering is made, depending on the relative magnitude of the scattering
coefficients. The polar scattering angle <p is sampled in the interval [0,7i] according to the
angular distributions Pjncoherent(9) and PCOherent (<p) given by the Klein-Nishina differential
cross section [109] and the coherent (F) and incoherent (S) atomic form factors as described
and tabulated in [110].
Interaction
point
Figure 8.3: Schematic illustration of the geometrical description of photon propagation in tissue. The
photon is moved in steps of size s. At each interaction point the photon is either absorbed or scattered.
If scattering occurs, the angles q> and y/ are sampled according to their probability distribution.
1 2 a -{I- COS <p)

•wzcoherent' 1 + cos <p + sin ^?- S{x,X)
a(\ -cos<p) (12)
^coherent («?) ~ 0 + C °S 2 <P) ' s i n <P '

The azimuthal angle, \\>, is evenly sampled from the interval [0, 2TT). The geometry is
illustrated in Figure 8.3.
The energy loss due to Compton scattering at an angle cp is taken into account according
to:
^before ^before
(1-cos^), (13)
E,after m •c
where m is the electron mass, c the speed of light, and Ebefore an Eaf,er are the photon energy
before and after the scattering event, respectively.
All angles and step sizes are sampled from their probability density function using the
distribution function method [104]. This method makes it possible to sample non-uniform
probability functions by using a uniform random number generator which is much easier to
implement. A normalised probability density function f(x), defined in an interval [a,b] has a
distribution function F(x) which is given by:
= \f(y)dy, with F(x)e [0,1] (14)
Sampling a uniformly distributed random number £ € [0,1] and finding x so that F(x)= "t,
gives the desired value of x. If F(x) can be inverted, as is the case for the photon step size
sampling, x is given by:
x = F~\t) (15)
If F(x) can not be inverted, this problem can be solved numerically by tabulating the
values of F(x) in a sufficiently fine grid and performing a search for x, so that F(x)= \. This
technique was used for sampling of the angular distributions.
If a photon is absorbed, all the information about its path prior to the absorption event is
lost. Therefore, in order to make the simulation more efficient, a model of photon packages
with a continuous weight was used. Upon an interaction in tissue, the weight of the photon
package is reduced by a factor proportional to Pabs. The remaining photon package
undergoes scattering. This concept leads to a better statistical accuracy or less computational
effort. It was carefully checked that the photon package concept gave the same simulation
results as the single-photon concept.
All interaction and cross-section data were obtained from the databases of the National
Nuclear Data Centre (NNDC), Brookhaven [57] and the National Institute of Standards and
Technology (NIST) [111]. Interaction coefficients for compounds and mixtures were
obtained by averaging the interaction cross sections of the atomic constituents according to
their relative abundance. Data on the average composition of biologically relevant materials
were taken from the International Commission on Radiation Units and Measurements
(ICRU) [112]. Values of the interaction coefficients at non-tabulated energies were
calculated by logarithmic interpolation.
8.2.2 Simulation geometries

The amount of scattered radiation and its temporal and spatial distribution is dependent on
geometric factors, i.e. the form of the scattering volume, the illuminated field of view and
the source and detector positions [101-103,113,114]. The aim of the studies presented in this
thesis was, in the first place, to demonstrate the concept of time-gated imaging and its
scatter-suppression performance in general, as independently as possible from the above
mentioned factors. In addition, direct comparison with the experimental values had to be
possible. Therefore, two different simulation geometries were chosen.
An X-ray point source emitting only in the forward direction ("a pencil beam"), incident
on a laterally infinite homogeneous medium of defined thickness, was used in the first set of
simulations [Paper V]. Behind the medium, the radial and temporal distribution of photons
was recorded. This is illustrated in Figure 8.4 for an example of adipose (fat) tissue. The data
from this simulation correspond to the point spread function of the system. The temporal
distribution of transmitted X-ray photons for homogeneous illumination of the tissue layer
can be obtained by spatial convolution of the data in Figure 8.4 (see also Paper V and
Paper VI). Such simulations are useful when investigating the influence of different
parameters, e.g. photon energy, tissue thickness and temporal resolution in time-gated
imaging. However, due to the assumption of a laterally infinite medium, the amount of
scattered radiation will be slightly overestimated.
A different simulation geometry was chosen in order to make the results comparable to
experimental results. These simulations were modified to simulate "real" imaging: An X-ray
point source illuminates a defined area on a laterally limited multi-layered medium. Flat
absorbing non-scattering objects can be placed inside this medium, with a given
transmission. No scattering occurs in these objects. In this geometry, photons are registered
on a 2-dimensional detector, resulting in a set of images at different times. The output is
schematically illustrated in Figure 8.5. This simulation geometry allows the direct evaluation
of contrast improvement in a real geometry and was therefore used in Paper VI to compare
experimental and simulated contrast improvement values.
10"
400 &
10 *• M
25 600 ^
Radial Position [cm]
Figure 8.4: Example of output from the Monte Carlo simulation in pencil-beam geometry, for
illumination of a 12 cm adipose (fat) tissue layer, at a photon energy of 60 keV.
Figure 8.5: Example of output from the Monte Carlo simulation in spatially resolved geometry, for
illumination of an 18 cm soft tissue layer, at a photon energy of 80 keV. The images are 1 ps apart, the
image to the right shows the time-integrated image.
8.2.3 Results
The influence of X-ray photon energy and tissue thickness on the amount of scattered
radiation and the temporal shape of the transmitted signal was calculated. From these data,
the scatter suppression by time-gated detection with a given gate width was numerically
determined (see Paper V and Paper VI). As an example, the amount of scattered radiation in
different thicknesses of breast tissue and the achievable scatter suppression for different
thicknesses of the same tissue are shown in Figure 8.6.
50 100 150 50 100 150 50 100 150

Energy [keV] Energy [keV] Energy jkeV]
Figure 8.6: (a) Ratio of the intensity of scattered radiation to the total intensity of transmitted
radiation as a function of X-ray photon energy for different thicknesses of breast tissue, (b) The
fraction of scattered radiation that is suppressed by time-gated imaging for a 5 cm thick layer of
breast tissue, for different gate widths and photon energies, (c) as in (b), but for a 20 cm thick layer of
breast tissue [adaptedfrom Paper VI].
The data shown in Figure 8.6 were calculated without taking the finite X-ray pulse width
and limited temporal resolution into account. The situation becomes much more complicated
as soon as more realistic assumptions, e.g., on the temporal resolution are made. In the
idealised case, the gate window should be open only during the arrival of the unscattered
photons. In this case scatter reduction similar to that shown in Figure 8.6 would be obtained.
The important point here is that all unscattered photons can be collected within a short gate
width, while suppressing most of the scattered radiation. In a more realistic case, the
unscattered and the scattered part of the transmitted radiation overlap in time. Thus there is a
trade-off between a) collecting as much unscattered radiation as possible in a wide window
to achieve a good signal-to-noise ratio, and b) achieving the best contrast available, which
requires a narrow window during the early part of the pulse. In order to quantify these
considerations, a concept of image quality was formalised [Paper V], which takes both
signal-to-noise ratio and contrast into account. The problem of insufficient signal-to-noise
ratio can be easily visualised with the output images of the imaging type of Monte Carlo
simulations, as shown in Figure 8.7.
The imaging-type simulations allow direct evaluation of the contrast achievable for a
given experimental geometry, which is difficult to achieve with the other type of
simulations. The imaging-type simulations were therefore used to compare the theoretically
calculated contrast improvements with the experimental results of the time-gated
tomography study (see Section 8.3.2). It was necessary to weight the simulation results with
the spectral distribution from the X-ray source. The emission spectrum from a tantalum
target [Paper II] was used for this purpose. Good agreement between experimental and
theoretical contrast improvement values was obtained. This type of simulation can therefore
be used to predict the contrast improvement with time-gated imaging for realistic tissue
a) 60 keV 140 keV
ballistic xl.5
150 ps xl.2
total
Figure 8.7: Simulated images (10 * 3 mm2) of a totally absorbing small rod imaged through 18 cm of
water, at different gate widths; (a) for a photon energy of 60 keV, and (b) for a photon energy of
140 keV. Note the different intensities, as indicated by the multiplication factors (adapted from
Paper VI).
50 150 200
Gate
Figure 8.8: Contrast improvement factor for imaging of a 1.9 mm x 30 mm opaque object, immersed
at the centre of a block of water 17 cm x 13 cm x 9 cm, as a function of gate width for different X-ray
photon energies. The temporal response function of the detection system was assumed to have a
Gaussian shape with 50 psfull width at half maximum.
thicknesses, as described in Paper VI. As an example, the contrast improvement for imaging
through 18 cm of adipose tissue is shown in Figure 8.8.
In the practical implementation of time-gated imaging with high temporal resolution
path-length differences from the source to different regions of the detector must be taken
into account (Paper V). This problem can be visualised using imaging-type simulation. A
sequence of images, with short time intervals between them, is shown in Figure 8.9. The
unscattered photons arrive first in the centre of the detector, while there is a temporal delay
in the outer regions. The contrast in the central parts of the image can be degraded by the
1 ps 3 ps 5 ps
Figure 8.9: Simulated imaging of a cross-shaped object with a very short X-ray pulse, shown at short
times after the arrival of the first photons. Central photons arrive at the detector earlier than non-
central photons which must travel a longer distance to the flat detector (cross imbedded in 18 cm of
soft tissue, 80 keVphoton energy).
time the ideal contrast is obtained in the outer regions. Suitable detector shapes, as discussed
in Paper V, could overcome such problems.
8.3 Experiments
8.3.1 Time-gated planar imaging
The first experiments in time-gated imaging were carried out using a small lead object,
which was imaged through water containers of various dimensions [Paper V]. The difference
between time-integrated and time-resolved images is illustrated in Figure 8.10. In time-
integrated imaging, the shadow of the lead object becomes less visible with increasing water
thickness. In the time-resolved images, the shadow is clearly visible at early times, when
mainly unscattered photons reach the detector. Similar experiments were carried out using a
mammography phantom [Paper VI]. Contrast improvement factors of about 6 were obtained
for a 15 cm thick scattering medium, and up to about 1.7 for the mammography phantom.
Small systematic errors were intrinsic in these measurements, as can be seen from the
illustration of the geometrical arrangement in Figure 8.11. The X-ray source, the object and
the streak camera were held at fixed positions, thereby leading to a varying air-gap between
the object and the scattering volume when the amount of water was varied.
Two additional factors must be taken into account when comparing these results with
more realistic situations. Firstly, the object was placed in front of the scattering volume,
which led to maximum scattering degradation in the image. Less scattering degradation
would have been obtained by placing the object in the middle of the scattering volume.
Secondly, in these experiments, the streak camera had an acceptance angle for incoming
X-ray radiation of about 1.1 sr, compared with 2% sr for a flat imaging detector. Therefore,
the amount of scattered radiation is underestimated, because scattered radiation can be
emitted into relatively large solid angles when leaving the scattering volume. These two
factors partly cancel each other.
8.3.2 Time-gated computed tomography

Time-gated computed tomography (CT) was investigated with a view to application in
volumetric tomographic imaging. This is a technique with which complete information on
the 3-dimensional structure of an object can be obtained in a single revolution of the object,
instead of being composed of a set of many 2-dimensional slices. The difference is
illustrated in Figure 8.12.
0 cm 3 cm 12 cm
a)
b)
C/3 ii en
OH
O O O
Figure 8.10: The shadow of a small lead object in both time-integrated (a) and time-resolved imaging
(b), for different amounts of scattering water in the X-ray beam path.
Streak
X-ray camera
source
Varying thickness of
scattering volume
Figure 8.11: Geometrical arrangement of the scattering volume in planar time-gated imaging (not to
scale).
This technique has been demonstrated on small objects [58,115,116]. However, it is

associated with problems due to strong scattering degradation when applied to larger
volumes. Anti-scatter grids, apart from their dose-penalty effects, also decrease the
resolution of a volumetric CT imaging system. Time-gated imaging might be used as an
alternative technique in this case, but requires efficient and ultra-fast 2-dimensional gated
detectors [27].
The feasibility of time-gated CT for standard slice imaging was demonstrated in Paper VI
and [117]. The cross section of a small object consisting of wires of different diameters was
imaged, both temporally gated and time-integrated, with and without the presence of a
scattering volume around the object. Instead of rotating the X-ray source and the detector
around the object as in normal CT scanners, the small object was rotated (see Figure 8.13).
For each projection angle, a time-resolved profile was recorded, as illustrated in Figure 8.14.
Figure 8.12: Computed tomography geometries for conventional 2D slice imaging (left) and the
intended 3D volumetric imaging (right).
X-ray
source Streak
camera
Scattering volume
Porcelain
Figure 8.13: Geometrical arrangement of the scattering volume and the imaged object in time-gated
tomographic imaging (not to scale).
Figure 8.14: A time-resolved profile obtained from the time-gated computed tomography
measurements. Good contrast is obtained at early times after irradiation.
Figure 8.15: Reconstructed images of the object through a 9 cm thick water layer, both time-
integrated (a) and time-gated (b). The gray level is proportional to the attenuation value, with lighter
shades corresponding to high attenuation.
Figure 8.16: Image of the object imbedded in water, as obtained with a medical CT scanner (Lund
University Hospital). The absorption of the metallic rods is too high to be correctly imaged with the
medical scanner, which is optimised for the imaging of small absorption differences in soft tissue or
bones. Note that a significantly higher number of projections and an anti-scatter grid were used to
obtain this image.
The images were reconstructed using a convolution backprojection algorithm [118]. This
algorithm was coded in a program taking our experimental conditions into account, e.g.
normalisation to the spatially varying detector sensitivity. The algorithm was tested and
optimised with the aid of theoretically calculated projection data. Reconstructed images of
the object imbedded in a volume of scattering water are shown in Figure 8.15. A contrast
improvement of about 50% was achieved when a gate width of 45 ps was used. This is in
good agreement with the values obtained from simulations (see Section 8.2.3 and
[Paper VI]).
It should be noted that the overall image quality is lower than in commercially available
CT equipment. This is predominantly due to the smaller number of projections which were
used for the reconstruction. To obtain the images shown in Figure 8.15, a total of 36
projections for each image were taken. This should be compared with about 1000
projections which are normally recorded with modern CT scanners. The same object as in
Figure 8.15 was scanned in a conventional CT scanner at Lund University Hospital, with a
much higher number of projections (about 1200) and the use of an anti-scatter grid (see
Figure 8.16).
8.4 Comparison with conventional scatter-reduction techniques

Anti-scatter grids are the most commonly used scatter-reduction devices in radiological
imaging. Other techniques are scanning-beam geometries and air-gap techniques
[1,101,113,119]. For a given imaging situation, the degree of scatter suppression obtainable
with anti-scatter grids is determined by its collimation geometry and material composition
(see Figure 8.1). An increase in collimation increases the scatter-reduction performance. An
undesired effect when using anti-scatter grids is the partial attenuation of unscattered
radiation by the collimating structure. This leads to an undesired dose-penalty effect, which
increases with improved scattering suppression. The dose penalty can be reduced by using
air-spaced grids or scanning beam geometries [101], but these techniques are technically
more complicated than conventional anti-scatter grids.
Air-gap techniques rely on the partial dispersion of scattered radiation in an air-gap
between the object and the detector [119]. However, due to difficulties in the practical
implementation of this technique it is seldom used.
The intrinsic advantage of time-gated imaging is that all unscattered photons exiting the
patient reach the detector, in contrast to the case with anti-scatter grids. In principle, no dose
penalty is therefore introduced by time-gated imaging. Using a gatable 2-dimensional
detector with the same sensitivity as standard imaging detectors, routinely used today, time-
gated imaging should result in superior performance. A comparison of achievable contrast
improvements and corresponding dose penalty is shown in Table 8.1.
However, the present limitations in the X-ray sensitivity of time-gated image detectors
result in an inferior performance compared with anti-scatter grids. Consequently, the
theoretical improvement factors discussed above are counterbalanced by these sensitivity
losses. Substantial improvements in detection techniques are needed to overcome this
problem.
Table 8.1: Comparison of scatter-suppression performance between time-gated imaging (our

experimental values) and anti-scatter grids [101]. Direct comparison with conventional CT imaging is
not possible due to the different geometries.
Time-gated imaging Anti-scatter g rids

Approximate contrast Dose Contrast Dose
improvement factor penalty improvement factor penalty
Mammography 1.7 none 1.3-1.5 2-3
(~ 20 keV, 4.5 cm)
Abdomen 6 none 4-7 6-10
(~60keV, 15 cm)
Tomography (9 cm) 1.5 none - -
75
9. OUTLOOK
Laser-produced plasma X-ray sources have the potential to be useful in many applications,
within both basic research and more applied fields. In order to make these sources more
attractive and usable, some important improvements are necessary. For example, the average
X-ray flux must be increased, the source should be made more user-friendly, and the
sensitivity of gatable 2-dimensional detectors must be significantly increased.
The future development of these sources is intimately coupled to the development of
high-power laser systems. This development is expected to follow two main lines; an
increase in peak power (by increasing the pulse energy and/or decreasing the pulse duration)
and an increase in pulse repetition rate. Which of these improvements is preferrable for
laser-produced plasma X-ray sources depends on the requirements on such a source. An
increase in laser pulse energy and peak power, at constant repetition rate, leads to both an
increase in average X-ray flux and an X-ray spectrum extending towards higher photon
energies. Increasing instead the repetition rate, at constant pulse parameters, results in a
higher average X-ray flux, but does not change the spectral distribution. A number of high-
power laser systems delivering only a few mJ in pulse energy, but operating at repetition
rates of 1 kHz have been, and are being, installed at various laser facilities. These laser
systems are well suited for high-repetition-rate X-ray sources in the X-ray region up to about
50 kV.
From the end-user point of view it is necessary to simplify the practical handling of the
X-ray sources as much as possible. This includes adjustment procedures, on-line monitoring
of the X-ray output and the length of uninterrupted exposure time that is available. Our
present X-ray source must be further improved in order to fulfil such requirements. Optical
alignment and target positioning could be automated by feeding X-ray source size and X-ray
output into a regulation mechanism that optimises these parameters. The rotating target,
which at present has a "lifetime" of at best 3 hours before it has to be changed, could be
replaced by a continuously refreshed wire target, as has been demonstrated by other
groups [120]. The continuous-wire technique is more or less mandatory when using 1 kHz
laser systems. The problem of sputtered target material must also be solved in a better way
than up to now. Several solutions have been investigated by groups working on applications
of laser sputtering in general, e.g. rotating slit wheels which are synchronised with the laser
pulse. Such improvements can be incorporated with engineering effort. However, it will be
difficult to transform a laser-produced plasma X-ray source into a "turn-key" device such as,
for example, conventional X-ray tubes.
Effective detector systems must be developed in order to use the probably most
interesting feature of this X-ray source, i.e. the short pulse duration. Time-resolving
detectors for hard X-ray pulses on the picosecond time scale, such as streak-cameras and
multi-channel plates, are still rather inefficient compared with other detectors such as image
plates, scintillators and solid-state detectors. This presently hampers time-gated imaging
applications in the medical context. However, detector technology is continuously
improving, and suitable devices might be available in the near future.
8SEXT PAGE(S)
left &UIM&
77
10. SUMMARIES OF THE PAPERS
Paper I
Ablation Studies
An overview of X-ray generation and medical applications in general is given in the first
part of this paper, summarising earlier investigations in our group. The second part describes
investigations in ablation phenomena that were observed during the X-ray generation
experiments.
Paper II
Spectroscopy (I) - Crystal Spectroscopy
Crystal spectroscopy was investigated as a tool for the spectral characterisation of hard
X-rays from the laser-produced plasma X-ray source. In the theoretical part of this paper,
considerations regarding the choice of crystal and diffraction geometry in these energy
ranges are presented. In the experimental part, the technique was applied to record well-re-
solved spectra of the characteristic L and K line emission from tantalum targets.
Paper III
Spectroscopy (II) - Energy-dispersive Spectroscopy
Energy-dispersive spectroscopy of the laser-produced plasma X-ray source using germanium
detectors was investigated in this paper. Conversion efficiencies from laser light into X-ray
radiation were measured and plasma electron temperatures could be determined.
Paper IV
Applications (I) - Differential Imaging
Differential imaging of a contrast agent was demonstrated in this paper. Experimental data
were obtained with both a test phantom and animals. The experimental values were
compared with theoretically estimated values.
Paper V
Applications (II) -Planar Time-Gated Imaging
This paper contains a detailed description of the Monte Carlo simulation techniques. The
influence of various parameters, e.g. photon energy, tissue thickness and time resolution, is
analysed. Issues concerning the quantification of image quality are discussed. Results of
experimental studies on the imaging of small objects through water phantoms are presented.
The problem of adequate detection devices is discussed.
Paper VI
Applications (III) - Tomographic Time-Gated Imaging
This paper contains both a review of the simulation results and an experimental section on
time-gated computed tomography. The simulation results were obtained with simulations
78 Summaries of the Papers
using an improved description of photon propagation in tissue. A different simulation
geometry was used in order to directly compare simulation results with experimentally
obtained values. Time-gated tomography of a small test phantom was demonstrated.
Contributions of the author to the articles

The author was first author and main responsible for Papers I, II (experimental part), V and
VI, and has also contributed to the manuscripts of Paper III and IV. The author carried out
most of the experiments described in Paper I (related to ablation studies), II, V and VI, and
also contributed to the experiments presented in Papers III and IV. The author carried out
most of the theoretical work in Papers IV, V and VI.
79
ACKNOWLEDGEMENTS
Firstly, I would like to thank my supervisors, Claes-Goran Wahlstrom and Sune Svanberg,
for the invaluable support and inspirations they have given me. They gave me the motivation
that kept projects going, practical help with theoretical and experimental problems, and a
great deal of necessary constructive criticism.
I would like to thank people that participated in this work: Carl Tillman, who started the
whole project and who gave me the benefit of his experience in physics, practical matters
and work organisation; Ian Mercer and Laurence Kiernan, who showed me different ways to
treat physical problems and to find different solutions; Antonio Pifferi, who gave me
invaluable advice on the problem of simulating photon migration in tissues; and many
others. During the last year, Christian Delfin, my new room mate, and Anders Sjogren, who
is becoming an X-ray expert, made my work much more delightful. With their sense of
humour and their approach to physics and experiments, they were an indispensable source of
motivation.
In interdisciplinary work such as that presented in this thesis, much expertise is required
in the various fields. Kristian Herrlin, Gudmund Svahn and Claes Ohlsson from the
Department of Radiology, Gustav Grafstrom from the Department of Radiophysics, and
Bengt Erlandsson from the Division of Nuclear Physics were of great help in answering
questions about radiological problems, X-ray dosimetry and conventional X-ray
spectroscopy. Within the collaboration with the X-ray Optics Group, Jena, Gisbert Holzer
helped me to understand crystal diffraction at a level far beyond the Bragg equation.
Working with advanced lasers and computer simulations inevitably required the help of
Anders Persson, who always had solutions for laser or programming problems. Many thanks
also to the staff in administration, electronics and mechanics. They were indispensable in
making things work in practice. It was an enjoyable experience to work in such a stimulating
environment as the Atomic Physics Division.
Many thanks to my parents, who always encouraged me in the work that led to this
thesis. Pernilla was of invaluable help during the last two years, motivating me in desperate
moments and reminding me about many other things besides physics.
81
REFERENCES
[I] E. Krestel, "Imaging systems for medical diagnostics", Siemens AG, Berlin (1990)
[2] S. Svanberg and C.-G. Wahlstrom (eds.), "X-ray lasers 1996", IOP, Bristol (1996)
[3] M.M. Murnane, H.C. Kapteyn, M.D. Rosen and R.W. Falcone, Ultrafast x-ray pulses
from laser-produced plasmas, Science 251, 51-536 (1991)
[4] D.G. Stearns, O.L. Landen, E.M. Campbell and J.H. Scofield, Generation of
ultrashort x-ray pulses, Phys. Rev. A 37, 1684-1690 (1988)
[5] S. Ebashi, M. Koch, and E. Rubenstein (eds.), "Handbook on synchrotron radiation",
North-Holland, Amsterdam (1991)
[6] H. Winick and S. Doniach (eds.), "Synchrotron radiation research", Plenum, New
York (1980)
[7] C.-G. Wahlstrom, J. Larsson, A. Persson, T. Starczewski, S. Svanberg, P. Salieres, Ph.
Balcou and A. L'Huillier, High-order garmonic generation in rare gases with an
intense short-pulse laser, Phys. Rev. A 48, 4709-4720 (1993)
[8] J. Rossbach, E.L. Saldin, E.A. Schneidmiller and M.V. Yurk, Interdependence of
parameters of an x-ray FEL, TESLA-FEL 95-6, (1995)
[9] W.P. Leemans, R.W. Schoenlein, P. Volfbeyn, A.H. Chin, T.E. Glover, P. Balling, M.
Zolotorev, K.-J. Kim, S. Chattopadhyay and C.V. Shank, Interaction ofrelativistic
electrons with ultrashort laser pulses: Generation of femtosecond x-rays and
microprobing of electron beams, IEEE J. Quant. Elect. 33, 1925-1934 (1997)
[10] A.V. Korol, A.V. Solov'yov and W. Greiner, Coherent radiation of an ultrarelativistic
charged particle channelled in a periodically bent crystal, J. Phys. G 24, L45-L53
(1998)
[II] M.Y. Andreyashkin, V.V. Kaplin, S.R. Uglov, V.N. Zabaev and M.A. Piestrup, X-ray
emission by multiple passes of electrons through periodic and crystalline targets
mounted inside a synchrotron, Appl. Phys. Lett. 72, 13 85-13 87 (1998)
[12] J.M. Boone and J.A. Seibert, A figure of merit comparison between bremsstrahlung
and monoenergetic x-ray sources for angiography, J. X-ray Sci. Techn. 4, 334-345
(1994)
[13] F. Carrol, Use of monochromatic x-rays in medical diagnosis and therapy. What is it
going to take ?, J. X-ray Sci. Techn. 4, 323-333 (1994)
[14] W.-R. Dix, Intravenous coronary angiography with synchrotron radiation, Prog.
Biophys. Molec. Biol. 63, 159-191 (1995)
[15] C. Tillman, I. Mercer and S. Svanberg, Elemental biological imaging by differential
absorption with a laser-produced x-ray source, J. Opt. Soc. Am. B 13, 209-215
(1996)
[16] C. Tillman, A. Persson, C.-G. Wahlstrom, S. Svanberg and K. Herrlin, Imaging using
hard x-rays from a laser-produced plasma, Appl. Phys. B 61, 333-338 (1995)
[17] B.M. Moores, material presented at the Luxembourg Workshop on Reference Doses
(1998)
[18] J.D. Kmetec, C.L. Gordon HI, J.J. Macklin, B.E. Lemoff, G.S. Brown and S.E. Harris,
MeV x-ray generation with a femtosecond laser, Phys. Rev. Lett. 68, 1527-1530
(1992)
82 References
[19] A. Rousse, P. Audebert, J.P. Geindre, F. Fallies, J.C. Gauthier, A. Mysyrowicz, G.
Grillon and A. Antonetti, Efficient K a x-ray source from femtosecond laser-produced
plasmas, Phys. Rev. E 50, 2200-2207 (1994)
[20] J.C. Kieffer, M. Chaker, H. Pepin, C.Y. Cote, Y. Beaudoin, C.Y. Chien, S. Coe, G.
Mourou and O. Peyrusse, Ultrafast x-ray sources, Phys. Fluids. B 5, 2676-2681
(1993)
[21 ] H.M. Milchberg, I. Lyubomirsky and C.G. Durfee, Factors controlling the x-ray pulse
emission from an intense femtosecond laser-heated solid, Phys. Rev. Lett. 67, 2654-
2657(1991)
[22] M. Schnurer, M.P. Kalashnikov, P.V. Nickles, Th. Schlegel, W. Sandner, N.N.
Demchenko, R. Nolte and P. Ambrosi, Hard x-ray emission from intense short pulse
laser plasmas, Phys. Plas 2, 3106-3110 (1995)
[23] U. Teubner, J. Bergmann, B. van Wonterghem and F.P. Schafer, Angle-dependent x-
ray emission and resonance absorption in a laser-produced plasma generated by a
high intensity ultrashort pulse, Phys. Rev. Lett. 70, 794-797 (1993)
[24] J. Workman, A. Maksimchuk, X. Liu, U. Ellenberger, J.S. Coe, C.-Y. Chien and D.
Umstadter, Control of bright picosecond x-ray emission from intense subpicosecond
laser-plasma interactions, Phys. Rev. Lett. 75,2324-2327 (1995)
[25] C. Tillman, "Development and characterisation of a laser-based hard x-ray source",
PhD Thesis, Lund Reports on Atomic Physics, LRAP-206, (1996)
[26] J. Krol, A. Ikhlef, J.C. Kieffer, D.A. Bassano, C.C. Chamberlain, Z. Jiang, H. Pepin
and S.C. Prasad, Laser-based microfocused x-ray source for mammography:
Feasibility study, Med. Phys. 24, 725 (1997)
[27] C.L. Gordon III, G.Y. Yin, B.E. Lemoff, P.B. Bell and C.P.J. Barty, Time-gated
imaging with an ultrashort-pulse laser-produced-plasma x-ray source, Opt. Lett. 20 ,
1056-1058(1995)
[28] K. Herrlin, G. Svahn, C. Olsson, H. Pettersson, C. Tillman, A. Persson, C.-G.
Wahlstrom and S. Svanberg, Generation of x-rays for medical imaging by high-power
lasers: Preliminary results, Radiology 189, 65-68 (1993)
[29] C. Tillman, G. Grafstrom, A.-C. Jonsson, B.-A. Jonsson, I. Mercer, S. Mattson, S.-E.
Strand and S. Svanberg, Survival of mammalian cells exposed to ultrahigh dose-rates
from a laser-produced plasma x-ray source, accepted for publication in Radiology
(1998)
[30] F.F. Chen, "Plasma Physics and Controlled Fusion", Plenum Press, New York (1984)
[31] R. Sauerbrey, Acceleration in femtosecond laser-produced plasmas, Phys. Plas 3,
4712-4716(1996)
[32] S.C. Wilks and W.L. Kruer, Absorption of ultrashort, ultra-intense laser light by
solids and overdense plasmas, IEEE J. Quant. Elect. 33, 1954-1968 (1997)
[33] P. Gibbon and E. Forster, Short-pulse laser - plasma interactions, Plasma Phys.
Control. Fusion 38, 769-793 (1996)
[34] E.T. Kennedy, Plasmas and intense laser light, Contemp. Phys. 25, 31 -58 (1984)
[35] F. Brunei, Not-so-resonant, resonant absorption, Phys. Rev. Lett. 59, 52-55 (1987)
[36] M.P. Kalashnikov, P.V. Nickles, Th. Schlegel, M. Schnurer, F. Billhardt, I. Will, W.
Sandner and N.N. Demchenko, Dynamics of laser-plasma interaction at 1018 W/cm2,
Phys. Rev. Lett. 73, 260-263 (1994)
[37] J.A. Stamper, E.A. McLean and B.H. Ripin, Studies of spontaneous magnetic fields in
laser-produced plasmas by Faraday rotation, Phys. Rev. Lett. 40, 1177-1181 (1978)
References 83
[38] A. Raven, O. Willi and P.T. Rumsby, Megagauss magnetic field profiles in laser-
produced plasmas, Phys. Rev. Lett. 41, 554-557 (1978)
[39] R.J. Mason and M. Tabak, Magnetic field generation in high-intensity-laser-matter
interactions, Phys. Rev. Lett. 80, 524-527 (1998)
[40] A.V. Rode, E.G. Gamaly and B. Luther-Davies, Evidence of magnetic fields imprinted
in ferromagnetic targets by the spontaneous magnetic field of a laser-produced
plasma demonstrated by after-pulse pulsations ofKa emission, Phys. Plas 4, 3676-
3683 (1997)
[41] A.R. Bell, J.R. Davies, S. Guerin and H. Ruhl, Fast-electron transport in high-
intensity short-pulse laser-solid experiments, Plasma Phys. Control. Fusion 39, 653-
659(1997)
[42] J. Steingruber, S. Borgstrom, T. Starczewski and U. Litzen, Prepulse dependence ofx-
ray emission from plasmas created by IR femtosecond laser pulses on solids, J. Phys.
B29,L75-L81 (1996)
[43] Z. Jiang, J.C. Kieffer, J.P. Matte, M. Chaker, O. Peyrusse, D. Gilles, G. Kom, A.
Maksimchuk, S. Coe and G. Mourou, X-ray spectroscopy at hot solid density plasmas
produced by subpicosecond high contrast laser pulses at 1018 -101 W/cm2, Phys. Plas
2, 1702-1711 (1995)
[44] H. Schillinger, C. Ziener, R. Sauerbrey, and H. Langhoff, Application ofMeVx.rays
from femtosecond laser-produced plasmas to the study ofphotonuclear effects, in:
"Applications of high field and short wavelength sources VII", A. L'Huillier, L.
DiMauro, and M.M. Murnane (eds.), OSA, Santa Fe (1998)
[45] H. Pueil, H.J. Neusser and W. Kaiser, Temperature and expansion energy of laser
produced plasmas, Z. Naturf. 25a, 1815-1822 (1997)
[46] G.H. McCall, Calculation of x-ray bremsstrahlung and characteristic line emission
produced by a Maxwellian electron distribution, J. Phys. D. 15, 823-831 (1982)
[47] J.C. Kieffer, M. Chaker, C.Y. Cote, Y. Beaudoin, H. Pepin, C.Y. Chien, S. Coe and G.
Mourou, Time-resolvedkiloelectron-volt spectroscopy of ultrashortplasmas, Appl.
Opt. 32,4247-4252(1993)
[48] M. Schnurer, R. Nolte, Th. Schlegel, M.P. Kalachnikov, P.V. Nickles, P. Ambrosi and
W. Sandner, On the distribution of hot electrons produced in short-pulse laser -
plasma interaction, J. Phys. B 30, 4653-4661 (1997)
[49] J.F. Pelletier, M. Chaker and J.C. Kieffer, Picosecond soft x-ray pulses from a high-
intensity laser-plasma source, Opt. Lett. 21, 1040-1042 (1996)
[50] D. Strickland, P. Maine, M. Bouvier, S. Williamson and G. Mourou, Picosecond pulse
amplification using pulse compression techniques, in: "Ultrafast Phenomena V", G.R.
Fleming and S.E. Sigman (eds.), Springer, New York (1986)
[51] S. Backus, C.G. Durfee, M.M. Murnane and H.C. Kapteyn, High power ultrafast
lasers, Rev. Sci. Instr. 69, 1207-1223 (1998)
[52] C. Spielmann, N.H. Burnett, S. Sartania, R. Koppitsch, M. Schnurer, C. Kan, M.
Lenzner, P. Wobrauschek and F. Krausz, Generation of coherent x-rays in the water
window using 5-femtosecond laser pulses, Science 278, 661-664 (1997)
[53] S. Svanberg, J. Larsson, A. Persson and C.-G. Wahlstrom, Lund high-power laser
facility - Systems and first results, Phys. Scr. 49, 187-197 (1994)
[54] R.W. Boyd, "Nonlinear Optics", Academic Press, San Diego (1992)
[55] J. Miyahara, The imaging plate: A new radiation image sensor, Chem. Today 223, 29-
36(1989)
84 References
[56] A.M. Gurvich, C. Hall, LA. Kamenskikh, I.H. Munro, V.V. Mikhailin and J.S.
Worgan, Phosphors for luminescent image plates, J. X-ray Sci. Techn. 6, 48-62
(1996)
[57] Data extracted from XRAY database, http://www.nndc.bnl.gov/, NNDC Online Data
Service, National Nuclear Data Center, Brookhaven, USA (1996)
[58] C. Bueno and M.D. Barker, High-resolution digital radiography and three-
dimensional computed tomography, in: "X-ray Detector Physics and Applications II",
V.J. Orphan (ed.), SPIE Proceedings 2009, San Diego, CA (1993)
[59] S.Y.F. Chu, L.P. Ekstrom and R.B. Firestone, WWW Table of Radioactive Isotopes,
http://nucleardata.nuclear.lu.se/Database/toi/, (1998)
[60] G. Zanella and R. Zannoni, Absolute light yield of plastic scintillators and cerium
scintillating glasses under low energy x-ray excitation, Nucl. Instr. Meth. A 302, 352-
354(1991)
[61] B.L. Henke, J. Liesegang and S.D. Smith, Soft-x-ray-induced secondary electron
emission from semiconductors and insulators: Models and measurements, Phys. Rev.
B 19, 3004-3021 (1979)
[62] B.L. Henke, J.P. Knauer and K. Premaratne, The characterization of x-ray
photocathodes in the 0.1-10-keVphoton energy region, J. Appl. Phys. 52, 1509-1520
(1981)
[63] V. Kaftandjian, Y.M. Zhu, G. Roziere, G. Peix and D. Babot, A comparison of the
ball, wire, edge, and bar/space pattern techniques for modulation transfer function
measurements of linear x-ray detectors, J. X-ray Sci. Techn. 6, 205-221 (1996)
[64] C. Kimme-Smith, L.W. Bassett and R.H. Gold, Focal spot size measurements with
pinhole and slit for microfocus mammography units, Med. Phys. 15, 298-303 (1988)
[65] A. Nykanen, "Spatial characterisation of a laser-produced x-ray source", Diploma
Work, Lund Reports on Atomic Physics, LRAP-210, Lund (1997)
[66] M. Schnurer, P.V. Nickles, M.P. Kalachnikov, W. Sandner, R. Nolte, P. Ambrosi, J.L.
Miquel, A. Dulieu and A. Jolas,Characteristics of hard x-ray emission from
subpicosecond laser-produced plasmas, J. Appl. Phys. 80, 5604-5609 (1996)
[67] S. August, D. Strickland, D.D. Meyerhofer, S.L. Chin and J.H. Eberly, Tunneling
ionization of noble gases in a high-intensity laser field, Phys. Rev. Lett. 63, 2212-
2215(1989)
[68] H.A. Bender, D. O'Connell and W.T. Silfvast, Velocity characterization of paniculate
debris from laser-produced plasmas usedfor extreme-ultraviolet lithography, Appl.
Opt. 34, 6513-6521 (1995)
[69] F. Davanloo, E.M. Juengerman, D.R. Jander, T.J. Lee, C.B. Collins and E. Matthias,
Mass flow in laser-plasma deposition of carbon under oblique angles of incidence,
Appl. Phys. A 54, 369-372 (1992)
[70] K. Chu and A. Fenster, Determinations of x-ray spectral distribution from
transmission measurements using K-edge filters, Med. Phys. 10, 772-777 (1983)
[71] K. Siegbahn (ed.), "Alpha, beta- and gamma-ray spectroscopy", North Holland,
Amsterdam (1965)
[72] CD. Mackay, Charge-coupled devices in astronomy, Ann. Rev. Astron. Astrophys.
24,255-283(1986)
[73] D.H. Lumb, G.R. Hopkinson and A.A. Weels, Performance ofCCDsfor x-ray
imaging and spectroscopy, Nucl. Instr. Meth. 221, 150-158 (1984)
References 85
[74] T. Tuomi, J. Partanen and K. Simomaa, Direct x-ray sensitive charge-coupled device
(CCD) as a detector in the study of synchrotron section topographs, Nucl. Instr. Meth.
Phys. Res. B 61, 569-574 (1991)
[75] G.J. Yates, G.W. Smith, P. Zagarino and M.C. Thomas, Measuring neutron Jluences
and gamma/x-ray fluxes with CCD cameras, IEEE Trans. Nucl. Sci. 39, 1217-1225
(1992)
[76] J. Dunn, B.K. Young, A.D. Conder, and R.E. Stewart, Detection of 1-100 keVx-rays
from high-intensity 500-fs laser-produced plasmas using charge-coupled devices, in:
"Solid State Sensor Arrays and CCD Cameras", C.N. Anagnostopoulos, MM. Blouke,
and M.P. Lesser (eds.), SPIE Proceedings 2654, San Diego, CA (1996)
[77] T. Wilhein, "Gedunnte CCDs: Characterisierung und Anwendungen im Bereich
weicher Rontgenstrahlung", Skaker, Aachen (1994)
[78] J.D. Cox, D.S. Langford, and D.W. Williams, Electronic intraoral dental imaging
system employing a direct sensing CCD array, in: "X-ray Detector Physics and
Applications II", V.J. Orphan (ed.), SPIE Proceedings 2009, San Diego, CA (1993)
[79] D. Grassi, E. Caroli, G. De Cesare, A. Donati and W. Dusi, Study of a CdTe multipixel
spectrometer for X- and gamma-rays, Nucl. Instr. Meth. A 379, 312-316 (1996)
[80] L. Struder, H. Brauninger, U. Briel, R. Hartman, G. Hartner, D. Hauff, N. Krause, B.
Maier, N. Meidinger, E. Pfeffermann, M. Popp, C. Reppin, R. Richter, D. Stotter, J.
Triimper, U. Weber, P. Holl, J. Kemmer, H. Soltau, A. Viehl and C. von Zanthier, A
36 cm2 large monolythic pn-charge coupled device x-ray detector for the European
XMMsatellite mission, Rev. Sci. Instr. 68, 4271-4274 (1997)
[81] H.H. Johan, Die Erzeugung lichtstarker Rontgenspektren mit Hilfe von
Konkavkristallen, Zeitschr. f. Phys. 69, 185-206 (1931)
[82] Y. Cauchois, Spectrographie des rayons xpar transmission d'unfaisceau non
canalise a trovers un cristal courbe (I), J. Phys. Rad. 3, 320-336 (1932)
[83] G. Holzer, privat communication (1998)
[84] G. Holzer, E. Andersson, P. Gibbon, E. Forster, M. Gratz, L. Kiernan and S.
Svanberg, Hard Ka yield from a laser-produced plasma, manuscript in preparation
(1998)
[85] P. Gibbon, private communications (1998)
[86] T.S. Curry III, J.E. Dowdey and R.C. Murry jr, "Christensen's Physics of Diagnostic
Radiology", Lea&Febiger, Philadelphia (1990)
[87] J. Geiswiller, E. Robert, L. Hure, C. Cachonicinlle, R. Viladrosa and J.M. Pouvesle,
Flash x-ray radiography of argon jets in ambient air, Meas. Sci. Technol. 9, 1537-
1542(1998)
[88] A. Khacef, R. Viladrosa, C. Cachonicinlle, E. Robert and J.M. Pouvesle, High
repetition rate compact source of nanosecond pulses of 5-100 keV x-ray photons, Rev.
Sci. Instr. 68, 2292-2297 (1997)
[89] J.M. Pouvesle, C. Cachonicinlle, R. Viladrosa, E. Robert and A. Khacef, Compact
flash x-ray sources and their applications, Nucl. Instr. Meth. Phys. Res. B 113 , 134-
140(1996)
[90] P. Suortti, Focusing monochromators for high energy synchrotron radiation, Rev.
Sci. Instr. 63, 942-945 (1992)
[91] P.J. Eng, M. Rivers, B .X. Yang, and W. Schildkamp, Microfocusing 4-ke V to 65-ke V
x-rays with bent Kirkpatrick-Baez mirrors, in: "X-ray microbeam technology and
applications", W. Yun (ed.), SPIE Proceedings 2516, San Diego (1995)
86 References
[92] A. Snigirev, V. Kohn, I. Snigireva, A. Souvorov and B. Lengeler, Focusing high-
energy x rays by compound refractive lenses, Appl. Opt. 37, 653-662 (1998)
[93] R. Lewis, Medical applications of synchrotron radiation x-rays, Phys. Med. Biol. 42,
1213-1243(1997)
[94] F.A. Dilmanian, X.Y. Wu, E.C. Parsons, B. Ren, J. Kress, T.M. Button, L.D.
Chapman, J.A. Coderre, P. Giron, D. Greenberg, D.J. Krus, Z. Liang, D. Marcovici,
M.J. Petersen, C.T. Roque, M. Shleifer, D.N. Slatkin, W.C. Thomlinson, K.
Yamamoto and J. Zhou, Single- and dual-energy CTwith monochromatic synchrotron
x-rays, Phys. Med. Biol. 42, 371-387 (1997)
[95] L. Malmquist, L. Rymell and H.M. Hertz, Droplet-target laser-plasma source for
proximity x-ray lithography, Appl. Phys. Lett. 68, 2627-2629 (1996)
[96] G. Farkas and C. Toth, Proposal for attosecond light pulse generation using laser-
induced multiple-harmonic conversion processes in rare gases, Physics Letters A 168,
447-450 (1992)
[97] P. Antoine, A. L'Huillier and M. Lewenstein, Attosecond pulse trains using high-
order harmonics, Phys. Rev. Lett. 77, 1234-1237 (1996)
[98] P.B. Corkum, N.H. Burnett and M.Y. Ivanov, Subfemtosecond pulses, Opt. Lett. 19,
1870-1872(1994)
[99] B. Jacobson, Dichromatic absorption radiography: Dichromography, Acta. Radiol.
39,437-449(1953)
[100] Z. Zhong, D. Chapman, R. Menk, J. Richardson, S. Theophanis and W. Thomlinson,
Monochromatic energy-subtraction radiography using a rotating anode source and a
bentLaue monochromator, Phys. Med. Biol. 42, 1751-1762 (1997)
[101] G.T. Barnes, Contrast and scatter in x-ray imaging, Radiographics 11, 307-323
(1991)
[102] G.T. Barnes and LA. Brezovich, The intensity of scattered radiation in
mammography, Radiology 126, 243-247 (1978)
[103] T J . Petrone, K.D. Steidley, A. Appleby, E. Christman and F. Haughey, X-ray beam
energy, scatter, and radiation risk in chest radiography, Health Phys. 70, 448-497
(1996)
[ 104] J. Persliden, A Monte Carlo program for photon transport using analogue sampling of
scattering angle in coherent and incoherent scattering processes, Comp. Prog.
Biomed. 17, 115-128(1983)
[105] M. Sandborg and G.A. Carlsson, Influence of x-ray energy spectrum, contrasting
detail and detector on the signal-to-noise ratio (SNR) and detective quantum
efficiency (DQE) in projection radiography, Phys. Med. Biol. 37, 1245-1263 (1992)
[106] F. Pedersen, C. Ronnquist, K. Fransson, L. Gustafsson, and S. Kullander, Energy
Discrimination with an x-ray Pixel Detector - A Monte Carlo Simulation, presented at
the 3rd Workshop on Pixel Detectors, Bari, Italy (1996)
[107] S. Andersson-Engels, R. Berg, O. Jarlman and S. Svanberg, Time-gated
transillumination for medical diagnostics, Opt. Lett. 15, 1179-1181 (1990)
[108] L.-H. Wang, P.P. Ho, C. Liu, G. Zhang and R.R. Alfano, Ballistic 2D-imaging
through scattering walls using an ultrafast optical Kerr gate, Science 253, 769-771
(1991)
[109] B.K. Agarwal, "X-ray spectroscopy", Springer, Berlin (1991)
References 87_
[110] J.H. Hubbell, W.J. Veigele, E.A. Briggs, R.T. Brown, D.T. Cromer and R.J.
Howerton, Atomic form factors, incoherent scattering functions, and photon
scattering cross sections, J. Phys. Chem. Ref. Data 4, 471-538 (1975)
[111] M.J. Berger and J.H. Hubbell, XCOM: Photon cross section database,
http://physics.nist.gov/PhysRefData/contents.html, National Institute of Standards and
Technology, USA (1998)
[112] International Commission on Radiation Units and Measurements, "Tissue Substitutes
in Radiation Dosimetry and Measurements", ICRU-44, Bethesda, USA (1989)
[113] B. Nielsen and G. Fagerberg, Image quality in mammography with special reference
to anti-scatter grids and the magnification technique, Act. Rad. Diag. 27, 467-479
(1986)
[114] M. Sandborg, "Effective use of x-rays in Diagnostic radiology", PhD Thesis,
Linkoping University, Linkoping (1993)
[115] W. Graeff and K. Engelke, Microradiography and microtomography, in: "Handbook
on Synchrotron Radiation", S. Ebashi, M. Koch, and E. Rubenstein (eds.), North-
Holland, Amsterdam (1991)
[116] U. Bonse, Rontgen-Mikrotomographie, Phys. Bl. 53, 211-214 (1997)
[117] M. Gra'tz, L. Kiernan, K. Herrlin, C.-G. Wahistrom and S. Svanberg, Time-gated x-ray
tomography, Appl. Phys. Lett., November issue (1998)
[118] A. Macovski, Basic concepts of reconstruction algorithms, in: "Radiology of the skull
and the brain", T.H. Newton and D.G. Potts (eds.), Mosby, St.Louis (1981)
[119] G.U.V. Rao, R.L. Clark and B.W. Gayler, Radiographic magnification: a critical,
theoretical and practical analysis (I), Appl. Radiol. 1, 37-63 (1973)
[120] C.P.J. Barty, private communication (1997)

Characterisation and Application of A Laser-Based Hard X-Ray Source

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Characterisation and Application of A Laser-Based Hard X-Ray Source

Uploaded by

Copyright:

Available Formats

SE9900161

Characterisation and Application

Lund Reports on Atomic Physics

Characterisation and Application

som for avlaggande av teknologie doktorsexamen

Distribution by (name and address)

General aspects of X-ray generation and applications

Laser-produced plasmas as X-ray sources

2.1 General definitions and notations

or cop[s-ll = 56Ajne[cm-3l (1)

Target Target Target

2.2 Laser light absorption and related processes

2.3 X-ray emission from laser-produced plasmas

3.1 Laser systems

Nd:YAG 3-L Multipass

mull!!lILImJ UUil i illilJ

3.2 X-ray generation

Target materials and setup

Beam transport considerations

0.0 0.5 1.0 1.5 2.0

3.3 X-ray detection

3.3.2 Scintillator-CCD camera combination

3.3.3 Scintillator-photomultiplier combination

3.3.4 Absolute calibration of the scintillator-based detectors

X-rays <l CCD

Extraction Vacuum housing Output

3.3.5 Time-resolved X-ray detection

4.1 X-ray source size

ensit y |arb. un its]

-150 -100 -50 0 50 100 150

4.2 X-ray flux

' X-hard (6)

X • .' '' '••:

Dependence on ambient gas and pressure

Dependence on target position

Cu: 0.5 mm ; Pb:5mm

-150 -100 -50 0 50

Dependence on target structure

• : .*•/ ' "• *"v .*• i

Time [arb. units]

'. -• • • ', •>'•'•'" ' " • ; »V

"E • ' • • • ' . • .

4.3 X-ray flux fluctuations

'"'' '\'r. ,•'"••'

0.70 0.85 1.00 1.15 1.30

Pressure Total X-ray flux Hard X-ray flux

4.4 Ablation phenomena

We have found no satisfactory explanation of this phenomenon. To our knowledge, no

5.1 Energy-dispersive single-photon counting with conventional detectors

5.2 Energy-dispersive single-photon counting with CCD cameras

Charges are confined and' | Depletion

Charges are confined ana =. Depletion

50 100 150 50 100 150

50 100 150 50 100 150

5.3 Crystal spectroscopy

5.3.3 Conversion efficiency and hot electron temperature

Spectral Line Photon Flux [l/(pulse-27csr)] EX/EL EL [mJ]

6. COMPARISON WITH OTHER X-RAY SOURCES

Source type Size Relative average flux

8.1 The scattering problem in radiology

Input x-ray pulse

8.2 Monte Carlo simulations of time-gated imaging

f-t-total Mphotoabsorption "** Mincoherent scatter "*" ^coherent scatter (10)

• : .•/ ' "• "v .*• i