Capillary Discharges as Plasma Sources

for Wakefield Acceleration Experiments

A research proposal for the Ph.D. qualifying examination

by

Efthymios Kallos
Department of Electrical Engineering - Electrophysics

University of Southern California

Committee Members:

Martin Gundersen (chairman) EE-EP, USC

Thomas Katsouleas EE-EP, USC
Patric Muggli EE-EP, USC
Aluizio Prata EE-EP, USC
Stephen Haas Physics, USC
 2007 A research proposal for the qualifying examination

2 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Abstract
In this proposal several ablative and gas-filled capillary plasma sources are reviewed, relevant
diagnostics that can be utilized to diagnose the peak plasma density and the density evolution
over time are investigated, experimental results are presented, and also methods and
techniques that can be implemented in order to generate and diagnose high density (~1019cm-3)
plasmas are proposed for this realm of densities which is at the moment not reliably available
using capillaries. The work is aimed towards utilizing such plasma sources in high gradient
plasma wakefield accelerators.

Summary and Motivation of the Proposal

Ever since the advent of particle accelerators in the middle of the 20 th century, atom
smashers have been driving our scientific knowledge and research efforts in order to
investigate and understand the fundamental forces and elementary particles that constitute
the cosmos. Currently, conventional particle acceleration technologies rely on microwave radio-
frequency electromagnetic fields that propagate inside metallic waveguides and accelerate
particles such as electrons, protons, positrons and ions to GeV and TeV energies. However
these methods have limitations as very high gradients cannot be utilized since the wall
materials breakdown at electric fields higher than approximately 50MV/m. As a result, any
further increases in energy require lengthy and expensive accelerators, such as the ILC
(International Linear Collider) which is proposed to be built at a cost of $6.5B and it will extend
over 30km long in order to accelerate and collide electrons and positrons at 500GeV energies.

In order to overcome those limitations, in the last quarter of the 20th century alternative
technologies gradually emerged as candidates for miniaturizing particle accelerators. They can
provide orders of magnitude higher acceleration gradients at a fraction of the distance and the
cost of conventional machines. One of the most promising technologies is the Plasma Wakefield
Accelerator (PWFA), where an intense laser beam or a strong electron bunch creates electron
waves in high density plasmas, which in turn support accelerating electric fields on the order of
10-100GeV/m – plasmas are already broken down into ions and electrons. Trailing particles can
then sample these waves or wakes and propel into high energies over extremely short
distances. For instance, in the recent E167 experiments conducted at the 3km long Stanford
Linear Acceleration Center (SLAC), some of the electrons of the initial 42GeV electron bunch
doubled their energy to more than 85GeV after travelling over a distance of just 85cm while
sampling the excited plasma wakes which were on the order of 50GeV/m inside a lithium vapor
plasma with density 2.7x1017cm-3.

Efthymios Kallos 3
 2007 A research proposal for the qualifying examination

In order for the plasma accelerators (either laser or electron bunch driven) to operate
optimally, reliable plasma sources are required. These sources must be able to produce high
density (1016cm-3 – 1019cm-3), low temperature (up to 10eV) and longitudinally uniform plasmas
over relatively long distances (1-100mm) in a controllable, repeatable and consistent manner. A
popular category of such plasma sources are capillary discharges, which can be divided into two
general sub-categories: gas-filled discharges and ablative discharges. In gas-filled capillaries a
cylindrical empty volume is surrounded by an insulating material and it is filled with several
hundred Torr pressure of an easy to ionize gas like hydrogen and then a high voltage discharge
(several kV) breaks down the gas and creates a plasma that typically lasts for several
microseconds. In ablative capillaries the cylindrical volume is empty and surrounded by a
material such as polypropylene ([CH2]n). In this setup, when the high voltage discharge occurs
then the current flows along the walls of the capillary and evaporates the material which turns
into a plasma and fills the capillary.

In this proposal we investigate and review several capillary plasma sources, identify and
present relevant diagnostics that can be utilized to diagnose the plasma density evolution over
time and also suggest methods and propose techniques that can be implemented in order to
generate very high density (~1019cm-3) plasmas, a realm of densities which is at the moment not
readily available using capillaries.

Specifically, the proposal consists of 4 sections. In section 1, the applications of plasma

sources in plasma-based particle accelerators is reviewed in more detail and the different types
of plasma sources are examined (capillaries, gas jets, metal vapors). In section 2, various
diagnostics for the plasma density and the plasma temperature are examined: Stark Broadening
of the Hydrogen Balmer lines, CO2 laser transmission and interferometry, Thompson scattering
and Schlieren photography. In section 3, a large set of experimental data from both ablative
and gas-filled capillaries are presented, which aim at identifying the dependences of the density
and temperature on applied discharge voltage, neutral gas pressure, tube length, tube diameter
and electrode diameter. The time evolution of those quantities is also presented. Those
datasets were obtained at the Pulsed Power Lab at the University of Southern California and at
the Accelerator Test Facility at the Brookhaven National Lab. In section 4, a number of novel
designs are proposed that can improve the current results and produce higher plasma densities.
We investigate a Z-pinch discharge and also the construction of a high pressure chamber that
can be filled with several atmospheres of gas, a neutral density necessary to achieve high
plasma densities. Then we propose a new diagnostic for the current setups based on using a
low power CO2 or Nd:Yag laser to perform Schlieren photography. Finally, various issues of
installing and operating a high pressure, high density capillary discharge in the sensitive vacuum
environment of an accelerator beamline are examined.

4 Efthymios Kallos
 Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Table of contents
List of figures ................................................................................................................................................. 7
List of equations .......................................................................................................................................... 10
Section 0 ..................................................................................................................................................... 11
Motivation for using plasmas as accelerators ............................................................................................ 11
Section 1 ..................................................................................................................................................... 15
Types of Plasma Sources and their Applications to Plasma Accelerators................................................... 15
[1.1] Gas Jets ..................................................................................................................................... 16
[1.2] Metal Vapors ............................................................................................................................. 18
[1.3] Capillary Discharges .................................................................................................................. 20
Section 2 ..................................................................................................................................................... 23
Plasma density and plasma temperature diagnostics ................................................................................ 23
[2.1] Collection of the Plasma Spectrum ........................................................................................... 24
[2.1.1] Stark Broadening of the Hydrogen Balmer Lines ................................................................ 25
[2.1.2] Line Intensity Ratio of Balmer Lines .................................................................................... 30
[2.1.3] Line Intensity to Continuum Ratio ...................................................................................... 32
[2.2] Thompson Scattering ................................................................................................................ 33
[2.3] Laser Interferometry ................................................................................................................. 36
[2.4] Schlieren and Shadowgraph Photography................................................................................ 39
[2.5] Simulations of capillary discharges ........................................................................................... 41
Section 3 ..................................................................................................................................................... 43
Experimental data from capillary discharges.............................................................................................. 43
[3.1] Ablative Capillary Discharges at ATF ......................................................................................... 44
[3.1.1] Capillary Design and Current Characteristics ...................................................................... 44
[3.1.2] Time-Integrated Spectral Analysis of the Discharges ......................................................... 47
[3.1.3] Time-Dependent Spectral Analysis of the Discharges ........................................................ 49
[3.1.4] A Side on the Optics Collection System .............................................................................. 56
[3.2] Gas Filled Capillary Discharges At USC ...................................................................................... 60
[3.2.1] Gas Filled Capillary Design .................................................................................................. 61

Efthymios Kallos 5
 2007 A research proposal for the qualifying examination

[3.2.2] The Effect of Neutral Gas Pressure ..................................................................................... 64

[3.2.3] The Effect of the Capillary Length ....................................................................................... 65
[3.2.4] The Effect of the Charging Voltage ..................................................................................... 67
[3.2.5] The Effect of the Capillary Tube Diameter .......................................................................... 69
[3.2.6] The Effect of the Hollow Electrode Inner Diameter............................................................ 70
[3.2.7] Time Dependent Light Collection........................................................................................ 72
[3.2.8] A Side on the Reliability of the Spectral Analysis ............................................................... 73
[3.3] Gas Filled Capillary Discharges at ATF....................................................................................... 76
Section 4 ..................................................................................................................................................... 83
Suggestions, future designs and issues....................................................................................................... 83
[4.1] Enhancements for the Capillary Discharges at USC ................................................................. 84
[4.2] The Schlieren Diagnostic for Capillary Discharges ................................................................... 87
[4.3] Suggestions for Beamline Installation ...................................................................................... 90
References…................................................................................................................................................ 96

6 Efthymios Kallos
 Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

List of figures
Figure 1 - Aerial view of the underground LHC accelerator at CERN, Switzerland. The diameter is 9km
long. Source: [Link] 11
Figure 2 - Inside an accelerator RF cavity ................................................................................................... 12
Figure 3 - Aerial view of the 3km long SLAC. Source: [Link] ...................... 12
Figure 4 - Physical principles of a Plasma Wakefield Accelerator. Electron oscillations are excited when a
strong beam travels through the plasma. Source: Sung Lee. ..................................................................... 13
Figure 5 - Experimental LWFA setup using a helium gas jet. Mangles et al., 2004 ................................... 16
Figure 6 - Schematic of heat-pipe oven at SLAC E167 experiments. Source: Erdem Oz ............................ 18
Figure 7 - Schematic of the gas distributions inside the heat-pipe oven. Source: Erdem Oz ..................... 19
Figure 8 - Double discharge ablative capillary design layout. Source: Kaganovich et al., 1997. ................ 20
Figure 9 - Measured radial plasma density profiles in a gas-filled discharge capillary. Spence et al., 1999.
.................................................................................................................................................................... 20
Figure 10 - Gas-filled capillary design with dual gas injection point. Spence and Hooker, 2000. .............. 21
Figure 11 - Johannes Stark won the Nobel Prize in Physics for his work on the Stark effect. .................... 25
Figure 12 - Stark Effect: The splitting of the hydrogen energy levels in the presence of an external electric
field. ............................................................................................................................................................ 25
Figure 13 - Spectral emission lines in the hydrogen atom. ......................................................................... 25
Figure 14 - The plasma density as calculated from the line shape of the hydrogen Balmer Hα line for
various temperatures.................................................................................................................................. 26
Figure 15 - Typical graphs of Lorentzian distributions. Source: Wikipedia. ................................................ 27
Figure 16 - The plasma density as calculated from the line shape of the hydrogen Balmer Hβ line for
various temperatures.................................................................................................................................. 28
Figure 17 - Comparison of experimental data of Stark broadening calculations with Griem's theory. ..... 29
Figure 18 - Plasma temperature as a function of the line intensity ratio of the Balmer lines. Source: Patric
Muggli. ........................................................................................................................................................ 31
Figure 19 - Plasma temperature as a function of the line to continuum ratio of the Balmer lines............ 32
Figure 20 - Principle of Bremsstrahlung radiation. Source: Wikipedia ....................................................... 32
Figure 21 - Principle of Thompson scattering. Source: Wikipedia .............................................................. 33
Figure 22 - Plasma density and temperature with respect to the Thompson scattering spectral response
function. Source: Gawron et al., 1988. ....................................................................................................... 35
Figure 23 - Experimental setup for detecting Thompson scattering radiation. Gawron et al., 1988. ........ 35
Figure 24 - A typical setup for an interferometric measurement of the plasma density. Detector #2
serves as an amplitude monitor. ................................................................................................................ 37
Figure 25 - Interferometric images from a capillary plasma discharge. Source: Edison et al., 1993.......... 38
Figure 26 - Typical setup for Schlieren imaging. Source: Settles, 2001. ..................................................... 39
Figure 27 - Density and temperature profiles for an ablative capillary simulation. Source: Loeb and
Kaplan, 1989. .............................................................................................................................................. 41

Efthymios Kallos 7
 2007 A research proposal for the qualifying examination

Figure 28 - Simulated radial density profile for a gas-filled capillary. Source: Bobrova et al., 2001. ......... 41
Figure 29 - Simulated time-evolved profiles of the plasma density, temperature and longitudinal
expansion of a gas-filled capillary. Source: Bobrova et al., 2001. .............................................................. 42
Figure 30 - Photograph of the side electrode of an ablative capillary at ATF. Photo:[Link]. .................. 44
Figure 31 - Schematic layout of the double discharge ablative capillary at ATF. Source: Kaganovich. ...... 44
Figure 32 - Plasma current drawn for different charging voltages in the ablative capillary. ..................... 45
Figure 33 - The plasma light recorded during the discharge for various charging voltages as a function of
time. ............................................................................................................................................................ 46
Figure 34 - Typical plasma spectrum recorded for an ablative discharge. The right peak is the Hα line at
656nm, and towards the left is the tail of the Hβ line at 456nm. .............................................................. 47
Figure 35 - Collection of plasma spectra for various charging voltages. .................................................... 47
Figure 36 - Plasma spectra collected for various charging voltages for a different delay between the
discharges. .................................................................................................................................................. 48
Figure 37 - Detail on the Hα line for various times during a 20kV discharge. The peak of the light is set as
the t=0 point. .............................................................................................................................................. 49
Figure 38 - Total light collected at the spectrometer for various charging voltages as a function of time.
.................................................................................................................................................................... 50
Figure 39 - The intensity peak of the Hα line for various charging voltages as a function of the discharge
time. ............................................................................................................................................................ 51
Figure 40 - The FWHM of the Ηα line for various voltages as a function of the time during the discharge.
.................................................................................................................................................................... 52
Figure 41 - The full width at half maximum of the Hα line for various charging voltages as a function of
the time during the discharge. .................................................................................................................... 53
Figure 42 - Time dependent analysis of the 20kV discharge. ..................................................................... 54
Figure 43 - Comparison of the plasma density evolution during the discharge for various parameters. The
highest density is achieved for a 500ns delay time between the discharges as opposed to 700ns. ......... 55
Figure 44 - Experimental setup layout for the collection of the plasma light for the ablative discharges.
Source: Daniil Stolyarov. ............................................................................................................................. 56
Figure 45 - The FWHM of the mercury lines as a function of the entrance slit width of the spectrograph.
.................................................................................................................................................................... 57
Figure 46 - The transmission function of the optical fibers used at the light collection system. ............... 58
Figure 47 - The spectral response function of the optics collection system. It includes the optical fiber,
the spectrograph, the diffraction grating and the intensified gated camera. ............................................ 59
Figure 48 - At high pressures the electrode was pushed outwards from the hydrogen gas inside the
capillary. When the discharge was fired, the gas still broke down creating a large current on the right
side of the capillary. The current has to find its way towards the other electrode by squeezing inside the
tiny 0.6mm diameter of the capillary, thus increasing the energy density deposited at that region. ....... 60
Figure 49 – Photograph of the hydrogen capillary discharge at USC. Source: Jessica Hao Chen ............... 61
Figure 50 - Photographs of the light collection system at USC. .................................................................. 62
Figure 51 - The plasma light emitted with and without the presence of the 0.6mm capillary. ................. 62

8 Efthymios Kallos
 Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Figure 52 - A collection of images of the plasma discharges in the gas-filled capillaries. .......................... 63
Figure 53 - Typical hydrogen spectrum recorded with the OceanOptics USB spectrometer. .................... 63
Figure 54 - Spectra recorded for increasing neutral hydrogen pressures. ................................................. 64
Figure 55 - The plasma density for different capillary lengths. .................................................................. 65
Figure 56 - The plasma temperature for various capillary lengths. ............................................................ 66
Figure 57 - The plasma density as a function of the charging voltage. ...................................................... 67
Figure 58 - The plasma temperature as a function of the charging voltage............................................... 68
Figure 59 - The plasma density as a function of the capillary tube diameter............................................. 69
Figure 60 - The plasma density as a function of the electrode hole diameter. .......................................... 70
Figure 61 - The plasma temperature as a function of the electrode hole diameter. ................................. 71
Figure 62 - Time dependent light collection of the gas-filled discharge. A 26mm capillary was tested at
1atm with a 5ns gating time. ...................................................................................................................... 72
Figure 63 - Comparison between plasma densities extracted from the linewidths of the Ηα and Ηβ lines.
.................................................................................................................................................................... 74
Figure 64 - Comparison between the plasma temperatures extracted from the three different methods:
line to background ratio for Hα, line to background ratio for Hβ and the line ratio Hβ/Ηα. ........................ 75
Figure 65 - Plasma current recorded for various voltages and capacitances as a function of gas pressure.
.................................................................................................................................................................... 76
Figure 66 - Plasma current drawn for various voltages and capacitances as a function of charging voltage.
.................................................................................................................................................................... 77
Figure 67 - Plasma density for various voltages and capacitances as a function of gas pressure. ............. 78
Figure 68 - Plasma density for various pressures. The hybrid mode can be seen at low pressures. ......... 79
Figure 69 - The ionization fraction for various pressures. .......................................................................... 80
Figure 70 - Time dependent plasma density evolution for the gas-filled capillary discharge at ATF. ........ 81
Figure 71 - Proposed shielding device for the capillary discharges at ATF. Credit to Jessica Hao Chen. ... 85
Figure 72 - Refraction of a ray from a cylindrical plasma .......................................................................... 88
Figure 73 - Layout of the Accelerator Test Facility beamlines. ................................................................... 90
Figure 74 - Photo of beamline 1 at ATF. Source: ATF website. ................................................................... 91
Figure 75 - Photo of experimental devices on beamline 1 at ATF. Source: ATF website. .......................... 92
Figure 76 - Plasma density for various voltages as a function of neutral gas pressure when the gas is
supplied dynamically................................................................................................................................... 93
Figure 77 - Plasma density as a function of pressure for static supply of neutral gas. ............................... 94

Efthymios Kallos 9
 2007 A research proposal for the qualifying examination

List of equations
Equation 1 - Dependence of the spectral linewidth FWHM on the plasma density and temperature ...... 26
Equation 2 - Probability distribution function of a Lorentzian. .................................................................. 27
Equation 3 - Line intensity ratio of adjacent Balmer lines .......................................................................... 30
Equation 4 ...................................... 30
Equation 5 - The plasma parameter ........................................................................................................... 33
Equation 6 - Power radiated from Thompson scattering in a plasma ........................................................ 34
Equation 7 - The spectral response function according to the Salpeter approximation ............................ 34
Equation 8 - The plasma index of refraction............................................................................................... 36
Equation 9 - Phase difference through the plasma .................................................................................... 37
Equation 10 - Plasma density extracted from the phase shift .................................................................... 37
Equation 11 – Deviation angle through a refractive medium .................................................................... 39
Equation 12 - Intensity ratio at the interferogram ..................................................................................... 39
Equation 13 - Intensity ratio difference in a shadowgraph interferogram................................................. 40
Equation 14 - The eikonal equation ............................................................................................................ 87
Equation 15 - Refraction angle due to perpendicular index gradients ....................................................... 87
Equation 16 - Approximated plasma index of refraction............................................................................ 88
Equation 17 - The critical density for EM waves ......................................................................................... 88
Equation 18 - Refraction angle for plasma density gradients ..................................................................... 88
Equation 19 – Condition for the validity of the geometric approach ......................................................... 88
Equation 20 - Refraction angle for a cylindrical plasma ............................................................................. 89
Equation 21 - Plasma density calculation for a cylindrical geometry ......................................................... 89

10 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Section 0
Motivation for using plasmas as
accelerators
Modern particle accelerators
are the largest and most complicated
machines that humans have ever built.
Their purpose is to increase the velocity
and energy of elementary particles
which can thereafter be used in a
number of applications like medicine,
biology and high energy physics.
Especially in the field of modern
physics, particle accelerators have
shaped our knowledge of the
fundamental forces of nature. Those
accelerators work primarily as colliders,
Figure 1 - Aerial view of the underground LHC accelerator smashing particles together after
at CERN, Switzerland. The diameter is 9km long. Source: accelerating them. The Standard
[Link]
Model, the theory that currently is the
most accurate for describing the forces
of nature and the interactions between the different types of particles is mainly a result of
experimental data that were provided from such accelerators.

However in order for new particles to be discovered, higher energies have to be

achieved. Presently the forefront of particle accelerators is the Large Hadron Collider (LHC)
which is built in Geneva, Switzerland and it is designed such that until now elusive particles like
the Higgs Boson (which is expected to explain how the other elementary particles obtain their
mass) will be discovered. In order for the Higgs to be detected energies of more than 100GeV
must be reached. As a result, the LHC is a circular accelerator with a circumference of 27km and
its building cost is around $500M. The next big collider, currently still in the design phase, is the
International Linear Collider (ILC) which will collide together 500GeV or 1TeV electrons and
positrons. It is scheduled to be 30km long and it has a projected cost of $6.5B [1].

Efthymios Kallos 11
 2007 A research proposal for the qualifying examination

The size, cost and complexity of those accelerators increases greatly along with the raise
of the energy frontier. This is because they rely on a century-old technology for accelerating the
particles, which is to use microwave electromagnetic fields
that propagate along metallic waveguides. The strength of
these fields is currently limited by the breakdown of the
wall material: fields stronger than about 55MV/m strip
electrons off the wall which will interfere with the
accelerated particles and also destroy the waveguides.
Therefore the only solution with this conventional
technology (which has worked remarkably well until now)
is to extend the accelerators over greater lengths. For
example, the Stanford Linear Accelerator Center (SLAC) in
Figure 2 - Inside an accelerator RF California [2] is the longest linear accelerator and 3km of
cavity accelerating cavities are required in order to bring the
energy of electrons to a maximum of 50GeV. In circular
accelerators this limitation can be partially overcome since the particles are rotating multiple
times inside a loop, but another
limiting factor shows up: Synchrotron
radiation is emitted as the particles
turn around the loop (the velocity
changes its direction and hence it is
considered acceleration) [3] . This
radiation is stronger with increasing
energy and as a result the energy
gain of the particles is balanced by
the energy loss towards synchrotron
radiation, yielding zero net energy
gain after a few turns.

Since these conventional Figure 3 - Aerial view of the 3km long SLAC. Source:
technologies seem to be approaching [Link]

their limits and not much room for improvement is available, scientists have been trying to
investigate new ways of efficiently accelerating particles into very high energies. A new idea
that was introduced in the past 30 years is instead of using microwaves, to manipulate the
extremely strong electric fields that can arise in a plasma in order to accelerate particles.
Plasmas are gases that are already broken down into electrons and ions; therefore they can in
principle sustain arbitrarily high electric fields, typically in the order of 10-100GV/m.

12 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Figure 4 - Physical principles of a Plasma Wakefield Accelerator. Electron oscillations are excited when a strong
beam travels through the plasma. Source: Sung Lee.

These wakefields can be generated inside the plasma when a laser pulse or a relativistic
electron bunch is fed into the plasma. The bunch initially displaces the plasma electrons which
are then attracted back from the ions and overshoot their original positions, thereby initiating
plasma electron oscillations. These oscillations are typically in the micron range and propagate
at the speed of the driving beam. Trailing particles can then sample the electric fields of those
oscillations over some distance and accelerate into high energies [4-8]. So far the most
successful experimental verification of this method (Plasma Wakefield Acceleration – PWFA)
was provided by the E167 collaboration at SLAC where some of the electrons of the original
42GeV bunch accelerated into more than 85GeV of energy after propagating in a plasma 85cm
long, thus sampling an average wakefield of 52GV/m [9].

The plasma accelerators are not aimed at replacing conventional accelerators: they will
extend them. Plasma accelerators still require relativistic high energy bunches in order to excite
large amplitude waves. The ultimate goal is to place such an accelerator as an afterburner at
the end of a beamline and boost the incoming energy higher in much shorter distances [10].

Efthymios Kallos 13
 2007 A research proposal for the qualifying examination

14 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Section 1
Types of Plasma Sources and their
Applications to Plasma Accelerators
In this section we will categorize three main types of plasma sources and examine their
different applications, along with their advantages and disadvantages. We will focus on gas jets,
metal vapors and of course on capillary discharges in more detail. Although gas jets can provide
higher densities in general and they have been utilized successfully in a lot of laser wakefield
experiments, they cannot be extended over longer distances, something that is the great
property of capillaries.

Efthymios Kallos 15
 2007 A research proposal for the qualifying examination

[1.1] GAS JETS

Gas jets are used to supply high pressure gas for beam-plasma interaction. There is
typically a tank that holds a gas like helium or hydrogen in very high pressure. This gas is
released into vacuum through a valve and a small nozzle possibly at supersonic speeds.
However the gas is not pre-ionized; depending on the application, the laser pulse or the
electron bunch will have to be strong enough to ionize the gas and create the plasma (typically
at least 1015W/cm2 intensities). Once the plasma is created at the front of the beam,
immediately the rest of the beam will expel the electrons and initiate the plasma oscillations.

An example of gas jets in practice is the laser-driven plasma wakefield experiments

(LWFA) which were conducted in the past 5 years by Malka et al., Faure et al., Geddes et al.,
and Mangles et al. [11-15]. They managed to produce relatively monoenergetically accelerated
electron bunches of good quality in a controllable manner. For instance, Faure et al. used a
helium gas jet that had a nozzle diameter of 2mm, generating a longitudinally flat plasma
density on the order of 1018-1019cm-3 over 1.5mm length with 400μm ramps on each side. Their
laser was focused with an intensity of 1018W/cm2 very close to the nozzle, just 800μm away,
since the gas expands rapidly after it exits the nozzle. The other experiments used similar
setups.

Figure 5 - Experimental LWFA setup using a helium gas jet.

Mangles et al., 2004

Gas jets are very popular plasma sources among the LWFA community because the
lasers are intense enough to ionize the neutral gas themselves. On the other hand, the
interaction length is limited by the density plateau which in turn is limited by the small nozzle
diameters. It is also limited by the Rayleigh length of the lasers since they can only be focused
tightly over a small distance. If those interaction lengths need to be increased to cm-scales,
there has to be a technique that will channel the laser power over long distances, and that is

16 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

something that cannot be achieved with gas jets. Nevertheless, gas jets are an excellent choice
for short-length, high density plasmas.

Efthymios Kallos 17
 2007 A research proposal for the qualifying examination

[1.2] METAL VAPORS

Metal vapors work with a small piece of metal being heated up by a set of coils usually
wrapped around it. As the temperature increases, the metal vaporizes and turns into gas that is
contained inside a tube. Again, as with the gas jets, the gas is initially neutral and another
external source like a strong laser or an electron bunch is needed in order to ionize it and create
the plasma.

Figure 6 - Schematic of heat-pipe oven at SLAC E167 experiments. Source: Erdem Oz

A lithium vapor was used in the SLAC energy doubling experiment [9] which was created
inside a heat-pipe oven [16]. Lithium was used because it has the lowest ionization potential of
all metals; hence it is the easiest to ionize by the powerful SLAC electron bunch which is 50fs
long, carries 1.8x1010 electrons and it is focused down to a 10μm spot size. The ionization of a
gas depends in general on the strength of the electric field and the time duration over which
this field is applied; for the length of the SLAC bunch a field of 5GV/m is enough to ionize the
lithium gas, which is an easy task for the beam. On top of that, the bunch focuses inside the
newly created plasma, increases in intensity and also ionizes the Li+ ion which has a threshold
of 300GV/m [17].

An extra complication arises because the hot lithium gas, if left unconfined, it would
diffuse and reach the oven windows. This can be overcome by inserting neutral cold helium gas
on both sides of the lithium region, between the lithium and the oven windows. Finally, the
neutral lithium gas density can be controlled by the temperature of the oven and the pressure
of the surrounding helium gas. When the bunch comes into the gas, it simply dissociates and
ionizes every lithium molecule creating the plasma.

18 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

The great advantage of the heat-pipe oven is that the plasma can be extended over very
long distances (up to 1.2m in the SLAC experiment), which is necessary for a high energy
afterburner. This is simply done by turning on or off the current that flows on a selected
number of coils, thereby heating and vaporizing more or less lithium.

Figure 7 - Schematic of the gas distributions inside the heat-pipe oven. Source: Erdem Oz

Efthymios Kallos 19
 2007 A research proposal for the qualifying examination

[1.3] CAPILLARY DISCHARGES

The plasma sources that are also the focus
of this proposal are capillary discharges. As
opposed to gas jets and metal vapors, capillary
discharges do not require an external strong driver
to ionize a neutral gas. Rather, they produce
plasma directly by applying a high voltage (typically
in the several kV range) along a capillary tube,
which has usually lengths anywhere from few mm
to several cm and transverse length up to 1mm.
The plasma electrons originate either from the Figure 8 - Double discharge ablative capillary
ionization of a gas that has pre-filled the capillary design layout. Source: Kaganovich et al., 1997.
tube (gas-filled capillaries) or from the breakdown
of the capillary wall material (ablative capillaries). Currents in the kA range may be required to
break down the materials, so capillary discharges need to be supported by high voltage circuits
that deposit enough energy to the capillary. This energy is initially stored high voltage
capacitors with capacitances in the several of nF range.

Capillaries are a great plasma source for extending the interaction lengths, especially in
LWFA experiments because the typically cylindrical symmetry can create a near-parabolic in the
radial dimension profile which favors laser channeling [18, 19]. As a result, lasers can propagate
over distances that are no longer limited to their Rayleigh length and therefore increase the
interaction length between the laser and the plasma, allowing greater energy gain. Capillaries
are also great plasma sources for weak driving electron
beams or lasers that do not have enough intensity to
ionize on their own. In addition, the extra parameters
introduced such as the charging voltage or the initial
gas pressure allow for better control over the plasma
density. Since the discharges may last for several μs
before they die out and the relativistic beams travel
close to the speed of light, a beam will transverse the
capillary length in a matter of ps. The proper timing
between a beam and the discharge also allows for a
Figure 9 - Measured radial plasma density choice in the plasma density. Typically densities in the
profiles in a gas-filled discharge capillary. order of 1014cm-3 – 1019cm-3 can be obtained.
Spence et al., 1999.

20 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

On the other hand, the density produced at a capillary may be sensitive to parameters
that are not easy to reproduce widely, such as the exact material composition, the shape of the
discharging electrodes and the internal impedance of the power supply along with the current
pulse it can produce. Since each
experiment usually builds its own
capillary and power supply, there are
difficulties in the reproducibility of the
results among the researchers and
there seems to be a lack of common
knowledge in the literature. In
addition, due to the complexity and the
violent nature of the discharges there
exists no solid theory that predicts the
produced plasma densities with
accuracy. Only recently a few
simulation codes have been published Figure 10 - Gas-filled capillary design with dual gas injection
[20-23] in an attempt to solidify point. Spence and Hooker, 2000.
experimental results.

A major part of this proposal is to present experimental data that were taken at the
Pulsed Power Lab at the University of Southern California and at the Accelerator Test Facility at
the Brookhaven National Lab; those datasets aim at providing a clear investigation on how
various parameters of the capillary geometry (material, length, diameter) and discharge
characteristics (charging voltage, gas pressure, capacitance, electrodes) affect the plasma
density and temperature that is ultimately produced. Extra effort was made so that no other
parameter but only one changes between successive experimental tests, so that the effect of
each parameter can be isolated.

The last section of this proposal will also be devoted to several improvements and
techniques that can be utilized to push the densities into capillaries beyond the 10 19cm-3. High
densities are favorable in general because the accelerating wakefields increase with higher
density. Also, plasma accelerators that rely on resonant excitation of the plasma through
multiple bunches usually require densities at that range. Generating and diagnosing those
plasma densities requires novel ideas and implementations.

Efthymios Kallos 21
 2007 A research proposal for the qualifying examination

22 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Section 2
Plasma density and plasma temperature
diagnostics
In this section we will briefly review a number of diagnostics that can be used to
measure the plasma density and the plasma temperature in plasma sources. Although the
diagnostics are not in general device-specific, we will focus more on diagnostics that work well
on capillary discharges. The major diagnostic that was used experimentally is the Stark
Broadening of the spectral lines of the light produced by the plasma during discharges. Other
interesting choices are Thompson Scattering, laser interferometry and Schlieren and
shadowgraph photography.

Efthymios Kallos 23
 2007 A research proposal for the qualifying examination

[2.1] COLLECTION OF THE PLASMA SPECTRUM

Plasmas emit light; the light emitted at different wavelengths is a signature of the
plasma and carries information about the processes that occur inside it. By collecting that light
one can work backwards and estimate the plasma parameters such as the density and the
temperature. Unfortunately, both the plasma density and its temperature, i.e. the thermal
random motion of the plasma electrons, will contribute to the light emitted, so one needs to
take extra care and isolate the density effects from the temperature effects.

A hydrogen plasma (such as the one used in this proposal’s capillary discharges) will
emit light and a part of it will be at the optical wavelengths (also called Balmer lines). The light
typically originates from electronic transitions between the atomic shells inside the hydrogen
atom. This light can be collected with a series of lenses and then fed into an optical
spectrometer for further analysis.

3 different spectral techniques were used. First, the Stark broadening of the Balmer Hα
(656nm) and Ηβ (486nm) lines: The linewidth of the lines yields some information on the
density and the temperature. Second, the line intensity ratio between those two lines can
provide information on the temperature. Finally, the temperature can be estimated from the
background light that is emitted from hydrogen.

24 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[2.1.1] STARK BROADENING OF THE HYDROGEN BALMER LINES

Stark Broadening is the electric field analog of the Zeeman effect. In the
Zeeman effect, a strong magnetic field can alter the energy levels of the
electronic shells inside the atoms [24]. The Stark effect describes how a strong
electric field can perturb the energy levels instead [25]. It was first discovered by
Johannes Stark, who was also awarded the 1919 Nobel Prize in physics for this
Figure 11 - work.
Johannes Stark
won the Nobel The spectral signature of
Prize in the Stark effect is called
Physics for his
Stark Broadening.
work on the Transitions of electrons
Stark effect.
between atomic levels
that would normally
produce a spectral line almost at a
single wavelength, now occur between
perturbed levels and as a result the
spectral line becomes wider. The
widening of the lines is caused by the
electric fields generated by the free plasma Figure 12 - Stark Effect: The splitting of the hydrogen
electrons, therefore in general a larger plasma energy levels in the presence of an external electric
density will cause a larger widening and a field.

bigger Stark broadening. High plasma

temperatures (more than 3eV) in general also have an effect on the Stark broadening.

Figure 13 - Spectral emission lines in the

hydrogen atom.

Efthymios Kallos 25
 2007 A research proposal for the qualifying examination

Hans Griem was one of the first scientists to fully work out the details of Stark
broadening in Hydrogen lines and he has published a significant amount of work on diagnosing
the plasma density from spectroscopic measurements [26-31]. The most investigated Stark-
broadened spectral line in the literature is the Hydrogen Balmer Ηα line. It is very popular
because it lies in the visible spectrum and it is very reliable even at high densities. An example
on the dependence of the Ηα linewidth on the plasma density is given in the following graph:

20
10
1eV
3eV
5eV
19
10
Plasma Density

18
10

17
10
0 5 10 15 20 25 30 35
FWHM of H line
Figure 14 - The plasma density as calculated from the line shape of the hydrogen Balmer Hα line for various
temperatures.

The mathematical formula that is used to generate those graphs can be found in [28, 32] and
reads

Equation 1 - Dependence of the spectral linewidth FWHM on the plasma density and temperature

26 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

The plasma density n is given in cm-3 and the Full Width at Half Maximum (FWHM) of the Ηα
line Δλ1/2 is given in angstroms. The parameter α1/2 depends both on the density of the plasma
and weakly on its temperature; it is also different for each spectral line. It can be found
tabulated in Griem’s book [28]. In order to calculate the density one would have to estimate or
measure the temperature using a different diagnostic and then solve the above nonlinear
equation as a function of the plasma density n. Calculations of the density and the temperature
using this technique are limited to densities up to a few times 1019cm-3 and a few eV
temperatures. Griem’s theory is valid for densities up to 1019cm-3 and temperatures of 5eV;
values above those can be derived through extrapolations, but their validity is questionable. In
fact, the Stark broadening theory at higher densities and temperatures is still an open field of
research where people have been trying to make ends meet between experiments, simulations
and theoretical calculations [27, 29-31, 33-48].

The above formula and graph are based on a number of assumptions on the state and
behavior of the plasma, such as local thermal equilibrium (LTE) for the plasma and validity of
the quasistatic approximation, which assumes that the broadening is dominated by the effect
of the electrons and that the ions remain static during that time and produce no broadening
themselves. The shape of the line under those assumptions is predicted by the theory [26, 28]
and it is a lorentzian, which follows the probability distribution

Equation 2 - Probability distribution function of a Lorentzian.

Figure 15 - Typical graphs of Lorentzian distributions. Source:

Wikipedia.

Efthymios Kallos 27
 2007 A research proposal for the qualifying examination

The shape of the line is actually an excellent way to determine whether the plasma satisfies
LTE: if it fits a Lorentzian shape then the odds are that the theory of Stark broadening will give
accurate results; if not, then the error margin increases (although bulk park estimation is still
possible).

A similar graph for the dependence of the linewidth on the density and temperature can be
generated for the Hβ Balmer line, by using the same equation but different parameter α1/2 :

20 FWHM of H Line vs Plasma Density

10
1ev
2ev
19
Plasma Density [cm-3]

10 3ev

18
10

17
10

16
10 0 1 2
10 10 10
FWHM of H Line [nm]
Figure 16 - The plasma density as calculated from the line shape of the hydrogen Balmer Hβ line for various
temperatures.

In the literature the broadening of the Hα line is preferred over the broadening of the Ηβ line
because the it is less sensitive to ion contributions (hence the quasistatic approximation is more
valid) and also because it remains valid even at high densities. The graph below summarizes
recent experiments that have been performed to diagnose Stark Broadening. In all these cases
the density was also measured with an independent diagnostic and the results were compared
with Stark broadening of the Hα line [34, 38, 40, 41, 45]. The yellow line represents the density
from Stark broadening as predicted by Griem’s theory [28, 32]; the error bars account for
temperature variations.

28 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Experiments on Stark Broadening of Hα Line

Boddeker, 1993 Escarguel, 2000
Flih, Oks, Vitel, 2002 Buscher, 2002
Flih, Vitel 2001 Griem Theory, T=2eV
FWHM of Ha Line [nm]

30
25
20
15
10
5
0
1.00E+17 1.00E+18 1.00E+19
Plasma Density [cm-3]
Figure 17 - Comparison of experimental data of Stark broadening calculations with Griem's theory.

It is evident that densities close or above 1019cm-3 have not been investigated thoroughly, and
the accuracy of the diagnostic is low at high densities. In section 3 of this proposal we will
present a set of data exactly at that range of high densities that were obtained with capillary
discharges.

Efthymios Kallos 29
 2007 A research proposal for the qualifying examination

[2.1.2] LINE INTENSITY RATIO OF BALMER LINES

Another useful diagnostic that can be applied when the plasma spectrum is captured is
the ratio of adjacent spectral lines. Specifically, one of the most common tools is the ratio of
the two major Balmer lines, Ηα and Ηβ. If the lines have Lorentzian shapes then the quasistatic
theory can probably be applied and then the population on each atomic shell can be assumed
to follow simply Boltzmann distributions, which is dependent on the temperature of the
plasma. Then the ratio of the populations will be equal to the ratio of the spectral intensities of
the lines, and the ratio can be given as the ratio of the two Boltzmann distributions as [49]:

Equation 3 - Line intensity ratio of adjacent Balmer lines

The coefficients in this formula are tabulated in [50] and are related to the strength of each
transition. For completion reasons the sums are given here for the 2 aforementioned Balmer
lines:

Equation 4

Finally, the energy of each level in eV is and , while

and . Then one can solve for the plasma electron temperature in eV
(kB is Boltzmann’s constant). This formula is valid as long as the wavelength-integrated line
intensity ratio is between 0.01 and 0.68, which yields temperatures between 0.1eV and 10eV.
Extra care has to be taken if the spectral lines are “polluted” with background radiation; then
the pure line intensity has to be extracted first by removing the background contributions. The
following graph shows the temperature dependence on this ratio for the two main Balmer
lines.

30 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

10
Plasma Temperature (eV) T(LineRatio)

0.1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Line Ratio H /H
Figure 18 - Plasma temperature as a function of the line intensity ratio of the Balmer
lines. Source: Patric Muggli.

Efthymios Kallos 31
 2007 A research proposal for the qualifying examination

[2.1.3] LINE INTENSITY TO CONTINUUM RATIO

The plasma temperature can also be obtained
when there is enough continuum background at the
plasma spectrum. This continuum is usually
Bremsstrahlung Radiation, which is radiation emitted
from charged particles such as electrons when they are
deflected in the electric field of another charged particle
such as an atomic nucleus [51].

In general higher temperatures will cause larger

and stronger background radiation which usually blends Figure 20 - Principle of Bremsstrahlung
with the spectral lines and makes the determination of radiation. Source: Wikipedia
the linewidths more difficult. However, in some cases
both the background and the spectral line may be clearly visible and can be separated into their
two contributions. Then the ratio of the total line intensity divided by the background intensity
in a 10nm band around the line’s central wavelength can be calculated. Griem has developed a
theory for predicting the density from this ratio [26]; the theory was extended to lower
temperatures (down to 0.5eV) later by Jones et al., 1987.

Figure 19 - Plasma temperature as a function of the line to continuum ratio of the Balmer lines.

32 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[2.2] THOMPSON SCATTERING

A popular method for diagnosing both the
plasma density and temperature is Thompson
Scattering. The physics of this method is based on the
interaction between an electromagnetic wave and a
charged particle. The setup requires an external laser
source that is focused onto the plasma. The
electromagnetic field of the laser will cause the plasma
electrons to start oscillating, hence they will be
accelerated. As the electrons accelerate they emit more
radiation around the wavelength of the incident field
and at various angles (this is the scattered wave). This Figure 21 - Principle of Thompson
radiation is much weaker than the incident radiation but scattering. Source: Wikipedia
if collected and imaged properly it provides information
on the plasma conditions. Ions also scatter some light but since they are heavier they do not
move easily and the scattering efficiency (or cross section) is smaller that of the electrons. Still,
the ion scattering is useful when the electron scattered light blends with other light emitted
from the plasma.

The quantum mechanical extension of Thompson Scattering is required for very

energetic and high-momentum photons where id called Compton Scattering. Here
we will only focus on Thompson Scattering diagnostics. The plasma parameter α is defined for
Thompson scattering and distinguishes between coherent ( and incoherent
scattering:

Equation 5 - The plasma parameter

Here k is the laser wavenumber and λD the plasma Debye length. The plasma typically will
scatter coherently (hence more strongly) in the forward direction and more incoherently in the
perpendicular direction. For high plasma densities ( and low temperatures
though coherent radiation can be detected even at the perpendicular angles. So,
we will focus our attention on this regime for coherent scattering only. The perpendicular
scattering angle is also favored because in the forward direction the scattered light cannot be

Efthymios Kallos 33
 2007 A research proposal for the qualifying examination

distinguished from the incident light. The general expression for the power scattered at a
distance R from a unmagnetized, collisionless, low temperature plasma of length L around a
solid angle dΩ and frequency dωs around ωs is given by the general expression [52]:

Equation 6 - Power radiated from Thompson scattering in a plasma

The scattering power is proportional to the incident power Pi, the scattering length L, the
plasma density ne and the scattering cross section that is defined through the classical electron
radius . The spectral density function S determines the spectral response
of the scattered spectrum. For incoherent scattering (small α’s) the spectral response is simply
proportional to the velocity distribution function of the electrons. For high alphas the scattered
light emits radiation at wavelengths larger than the Debye length of the plasma; then the
radiation is sensitive to the shielding effects of the electrons around the ions (Debye shielding).
Then the spectral response has contributions from two factors, the so called “electron feature”
and the “ion feature”.

The spectral response function can be simplified when the electron temperature and
the ion temperatures are comparable, . This is the Salpeter approximation [53] and the
ion feature is the contributing part of the distribution since the electrons have very slow
motions (the electron feature becomes important at higher temperatures).This approximation
is valid in high densities and low temperatures because there is a high electron-ion collision rate
(hence the similar temperatures). In that case the shape of the scattered spectrum recorded
provides information on the plasma temperature and its calibrated peak yields the parameter
alpha (once the temperature is known), and therefore the density (from the above equation).
Under this Salpeter approximation the ion feature of the spectral response function is (for
hydrogen plasma) [52, 54]

Equation 7 - The spectral response function according to the Salpeter approximation

The quantity recorded with a spectrometer is . The following tables provide a guide to
calculating the plasma density and temperature for a hydrogen plasma under those
assumptions for a 900 scattering angle.

34 Efthymios Kallos
 Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Figure 22 - Plasma density and temperature with respect to the Thompson scattering spectral response function.
Source: Gawron et al., 1988.

A number of issues arise when an experimental setup has to be decided for Thompson
Scattering. First of all, there is choice of the light source. In order for the laser light to be
transmitted its frequency must by higher than the plasma frequency that is created. For high
densities in the 1019cm-3 range therefore a laser with wavelength smaller than CO2 has to be
used or else the plasma will reflect the radiation. Second, the source has to be powerful enough
to overcome the tiny cross section, which is on the order of 10-13 [52] which implies that a
pulsed power laser has to be used. The Ruby (λ=0.694μm) laser usually satisfies all those
conditions; it has been used successfully in several experiments.

The choice of the scattering angle has to do mostly with parasitic radiation; the input
laser light has to be filtered out or avoided completely by choosing a wide scattering angle
(which of course will reduce
the amount of coherent
radiation recorded). Also,
depending on the plasma
density the background plasma
light may blend with the
scattered light and blind the
detectors.

Figure 23 - Experimental setup for detecting Thompson scattering

radiation. Gawron et al., 1988.

Efthymios Kallos 35
 2007 A research proposal for the qualifying examination

[2.3] LASER INTERFEROMETRY

A widely implemented method for diagnosing the plasma density is interferometry.

Essentially a Mach-Zender interferometer has to be build with its one arm passing through the
plasma. Since the plasma has an index of refraction that depends on the plasma density
according to the expression:

ne 1 ne
p 1 1
nc 2 nc

Equation 8 - The plasma index of refraction

Then when the two signals from each interferometer arm are recombined then there will be
shift in the fringes of the combined signal. Here nc is the critical density which corresponds to
the wavelength of the laser used, which typically is much less than the plasma density, hence
the above linearization of the equation. For example, a HeNe CW laser at 0.633μm can be used
for the interferometry to identify densities at the 1019cm-3 range which correspond to 10.6μm
wavelengths.

This method has the great advantage that it provides full time dependent evolution of
the density during the plasma discharge; its drawback is that the laser light may be deflected or
absorbed by the plasma and pollute the interference signal. The density has to be extracted
from the number of fringe shifts with extra care. A typical setup of the interferometer is shown
below.

36 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Figure 24 - A typical setup for an interferometric measurement of the plasma density. Detector #2 serves as an
amplitude monitor.

In this setup the variable attenuator adjusts for the attenuation of the light inside the plasma
(the two beam need to have as equal as possible intensities when they are recombined). The
second detector records the single-arm time dependent amplitude variations of the beam due
to the plasma; these variations need to be removed from the interferometer signal (detector
#1) before calculating fringe shifts.

A certain interferometry pattern will appear when the plasma is not fired. When the
plasma is then fired the fringe pattern will shift due to the extra phase shift that the plasma will
introduce. This phase shift is approximately

 klaser L p

Equation 9 - Phase difference through the plasma

Here L is the length of the plasma, assuming that the plasma density remains constant
longitudinally. Combining the two equations we get

4 0 m
ne 2
e laser L

Equation 10 - Plasma density extracted from the phase shift

Efthymios Kallos 37
 2007 A research proposal for the qualifying examination

For example, using a HeNe laser over 3mm of plasma a full phase shift of 360 0 will occur for
every 1.2 1018 cm 3
change in the plasma density, which is ideal for measuring plasma densities
at that range.

Another great advantage of this method is that when a 2D spatial detector is used to
record the interferometer signal, then information about the transverse profile of the plasma
can be also gained [35]. Due to this reason interferometry is very popular for characterizing
capillary discharges since the radial profile of the plasma is of interest for the laser acceleration
community due to its laser channeling and guiding abilities.

Figure 25 - Interferometric images from a capillary plasma

discharge. Source: Edison et al., 1993.

38 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[2.4] SCHLIEREN AND SHADOWGRAPH PHOTOGRAPHY

Schlieren photography is extremely simple to implement; it is based on the fact that a
laser beam will be deflected when it travels through a plasma due to the density variations
along the beam path. These variations can be due to electrons, ions or neutrals. Depending on
the wavelength of the probe radiation, different types densities can be determined. Visible
radiation will resonate also with bound electrons, neutrals and the ions and will yield
information about those densities (unless the gas is fully ionized); infrared beams like CO2 lasers
will deflect from the plasma electrons only and so they can diagnose their density profile.

As opposed to previous methods, Schlieren techniques require a density gradient in

order to work. Specifically, in the shadowgraph technique, light rays will bend by an angle θx
d
when travelling through a plasma with refractive index variation according to [55]
dx

d d
x ( x) dl dl
dx dx

Equation 11 – Deviation angle through a refractive medium

The effect of the plasma will be to transfer electromagnetic energy from the point x that it
d
would have fallen without plasma, to the point x x L x x L dl . L is the distance
dx
travelled by the beam. The initial 1D intensity detected was Idx which now becomes I ,
according to Idx I dx . Then

I d2
1 L dl
I dx 2

Equation 12 - Intensity ratio at the interferogram

Figure 26 - Typical setup for Schlieren imaging. Source: Settles, 2001.

Efthymios Kallos 39
 2007 A research proposal for the qualifying examination

For small fractional intensity variations this yields

I d2
L dl
I dx 2

Equation 13 - Intensity ratio difference in a shadowgraph interferogram.

As a result a signal proportional to the second derivative of the plasma refractive index is
produced, the index being proportional to the plasma density for small angle deviations. This
method can produce relative plasma density deviations (averaged over the length of the
plasma), but it is not straightforward to obtain an absolute plasma density measurement.

In the Schlieren technique the refracted light is not captured directly but it is first
focused through a lens onto an image plane, thus forming a complete imaging system. A sharp
knife edge is then inserted at the focal point of the focusing lens which blocks part of the
forming image. When the plasma is active the rays will refract and miss the focal point, and
some of them will be blocked by the sharp edge. The amount of rays that is blocked is now
directly proportional to the refraction angle θ. A new interference pattern will then form at the
point where previously (without the edge) all the rays reached successfully. For that point the
d
phase difference causing the gradient has in essence transformed into an amplitude
dx
difference which translates into an interference pattern, without the presence of a second
beam. The plasma works as a beam splitter and the focusing lens as the interference element,
thus constituting the device essentially an interferometer.

Both of these techniques require a sensitive enough instrument to measure the

intensity variations of the incident beam. High density plasmas will refract more thus creating
stronger variations in the detected pattern. On the other hand, if large deviations are expected
then it must be taken into consideration the fact that only phase differences up to 1800 will be
detectable. Also, diffraction may wash out any intensity variations before they reach the
imaging screen. In general, Shadowgraphy can handle larger phase shifts but it is less sensitive
that Schlieren [56]. In section 4.2 we will discuss in more detail the application of the Schlieren
diagnostic to the high-density capillary plasmas that we are interested at.

40 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[2.5] SIMULATIONS OF CAPILLARY DISCHARGES

Due to the inherent uncertainty
and instability of the capillary discharges,
a number of simulation codes have risen
recently in order to shed some light in the
processes that happen during the
discharge. Here we will briefly review
some significant models that can be
found in the literature, describing both
ablative and gas-filled discharges. The
physics behind the simulations and
physical processes are quite different,
despite the fact that the measured
quantities (plasma density, temperature)
are similar. In ablative discharges there is
a mass flow that originates from the
capillary walls when the strong current Figure 27 - Density and temperature profiles for an ablative
flows through it. In gas-filled discharges on capillary simulation. Source: Loeb and Kaplan, 1989.
the other hand the ionization takes place
inside the capillary volume through heating and then the electrons move outwards to the walls
were they cool down. As a result, in gas-filled capillaries the temperature drops at the walls and
peaks at the center, which in turn causes the density to have a minimum at the center and
increase near the walls.

The first major attempt to simulate

ablative discharges was made by Loeb and Kaplan
[57] in their 1989 paper. They assumed an energy
balance equation where the inward mass flow
from the ablated wall equals in steady state the
diffusion of the plasma mass through the
capillary ends. They also include Ohmic heating
due to radiation and turbulence. Their model is
limited but it shows that both the density and the
temperature are mostly flat along the radial
Figure 28 - Simulated radial density profile for a
gas-filled capillary. Source: Bobrova et al., 2001. dimension of the capillary with the exception of

Efthymios Kallos 41
 2007 A research proposal for the qualifying examination

the areas very close to the walls. Other more recent models [22, 23, 58-60] deal with simulating
slow discharges (in the ms scale) that have uses in electrothermal launchers and soft x-ray
lasers. However they do not directly apply to PWFA plasmas which require fast (ns scale) and
high current pulses in order to ionize a high density gas.

For gas-filled capillaries a reliable model has been developed by Bobrova et al. [21] and
later by Broks et al. [20, 61]. They use a number of conservation equations to describe their
plasma discharge, which include continuity equation, momentum conservation equation and
species conservation equations, along with the appropriate boundary conditions at the walls.
The time-dependent model verifies the fact that electron densities on the order of 10 18cm-3 can
be produced which do have an approximately parabolic radial profile. The uncertainty in the
gas-filled models is less than the ablative ones because the processes are less violent. The main
uncertainty originates from the source terms that provide the dissociation and ionization
parameters for the energy input to the plasma over which some unverifiable assumptions have
to be made.

Figure 29 - Simulated time-evolved profiles of the plasma density,

temperature and longitudinal expansion of a gas-filled capillary.
Source: Bobrova et al., 2001.

42 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Section 3
Experimental data from capillary
discharges
In this section we will review a number of experimental results that were obtained over
the period of the last 2.5 years (September 2005 to April 2007). They deal with capillary
discharges and the measurements regarding plasma density, plasma temperature, plasma
current, plasma light emitted and various other parameters. The measurements were made at
the Accelerator Test Facility of the Brookhaven National Lab for ablative and gas-filled
capillaries and at the Pulsed Power Lab at the University of Southern California for gas-filled
glass capillaries. The data taken at Brookhaven are more scarce and do not lead specifically to
insight about the physics of the discharges; the data taken at USC are more systematic as we
tried to isolate one parameter only at a time and measure its effect on the plasma parameters.

Efthymios Kallos 43
 2007 A research proposal for the qualifying examination

[3.1] ABLATIVE CAPILLARY DISCHARGES AT ATF

This section is divided into three parts. First, we will review the capillary design and its
current characteristics. Second, we will present a set of time-integrated data collected for Stark
broadening measurements. Finally, we will investigate the time-dependent measurements that
shed some light on the evolution of the discharge and plasma density versus time.

[3.1.1] CAPILLARY DESIGN AND CURRENT CHARACTERISTICS

The ablative capillary constructed was a
double-discharge polypropylene [CH2] capillary
based on the design by Kaganovich et al. [62]. A side
view is shown on Figure 30 and a schematic layout
on Figure 31.

The capillary has 0.5mm diameter and

consists of two regions. The first or trigger region is
short in length, only 3mm, and a first discharge pulse
is deposited onto it first. The charging voltage for
that pulse is 36kV. This trigger discharge creates a Figure 30 - Photograph of the side electrode of
an ablative capillary at ATF. Photo:[Link].
low density plasma that expands and fills the region
of the main capillary, and it serves as an enhancement
for the plasma density. About 1μs (which is adjustable)
after the trigger pulse the main discharge pulse is
deposited across the main electrodes that are on the
ends of the main 14mm section of the capillary. The
charging voltage for this main pulse is between 5kV and
20kV. The discharge current flows along the walls of the
main tube and ablates them, breaking down the
polypropylene into its constituents carbon and
hydrogen, which are in turn dissociated and ionized.
The pre-plasma that was formed from the trigger
discharge reduces the amount of ablation required to
Figure 31 - Schematic layout of the double increase the plasma density.
discharge ablative capillary at ATF. Source:
Kaganovich.
44 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Depending on the charging voltage and the amount of current that is drawn (and hence
the amount of ablation that occurs) the lifetime of the capillary may vary from a few dozen to a
few hundreds of shots. The purpose of the trigger part of the capillary also is to reduce the
ablation of the main part, so that the main part will not need replacement sooner. During the
time period that the data here were taken (which covers a few months) the capillary body was
replaced when it was realized that for example reduced current was drawn.

Plasma Discharge Current for various voltages

800
14kV
16kV
600 18kV
20kV
Discharge Current [A]

400

200

-200

-400
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Discharge time [ns]
Figure 32 - Plasma current drawn for different charging voltages in the ablative capillary.

Figure 32 shows a number of current traces versus time for a number of different charging
voltages. The trigger discharge occurs at some point early in time (on the left side of the graph)
and 0.9μs later the main discharge occurs. During the main discharge the current peaks, and
the maximum current recorded was 0.8kA for a charging voltage of 20kV. In general, the
current increases with the charging voltage that is applied. The current dies completely roughly
2μs after the initiation of the trigger discharge.

Efthymios Kallos 45
 2007 A research proposal for the qualifying examination

Plasma light on photodiode for various discharge voltages

40
10kV
12kV
20
14kV
Plasma Light on Photodiode

16kV
0 18kV
20kV
-20 Discharge Current

-40

-60

-80

-100
0 500 1000 1500 2000 2500
Discharge Time [ns]

Figure 33 - The plasma light recorded during the discharge for various charging voltages as a
function of time.

Figure 33 shows the wavelength-integrated light that was collected after it was focused out of
the capillary and onto a photodiode detector for various charging voltages, along with a typical
discharge current trace. As expected, the amount of light emitted from the plasma increases
with higher voltage and in fact this increase is not linear with the voltage but it seems to be
larger at higher voltages. Also, it is evident that the peak of the light (and probably the peak of
the plasma density) is 200ns after the peak of the current.

46 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[3.1.2] TIME-INTEGRATED SPECTRAL ANALYSIS OF THE DISCHARGES

In this section we will present data that were taken using an optical spectrometer
to analyze the spectrum of the plasma light emitted during the discharge. The goal was to
indentify the linewidths of the hydrogen Balmer lines so that they can be used to estimate the
density using the Stark broadening technique, as described in section 2.1.1.

Figure 34 - Typical plasma spectrum recorded for an ablative discharge. The right peak is the Hα line at 656nm, and
towards the left is the tail of the Hβ line at 456nm.

A typical raw spectrum collected with the optical Spectrograph (SPEX 270M) is shown on Figure
34. The spectra were calibrated later using a mercury lamp. The strong peak on the right side is
the Hα line at 656nm, and the bright spot on the left side is the Hβ line at 486nm. There are also
some other lines visible that are signatures of the carbon atoms in the polypropylene, and also

Figure 35 - Collection of plasma spectra for various charging voltages.

Efthymios Kallos 47
 2007 A research proposal for the qualifying examination

a lot of continuous background light.

Figure 35 shows a number of spectra collected using a small Oceanoptics USB2000

spectrometer for various charging voltages and 1.1μs delay between the two discharges (trigger
and main). It is evident that higher voltages create more background light and wider lines. The
Ηα, Ηβ, Ηγ and Ηδ Balmer lines are visible in these discharges, the rest of the lines being
identified as carbon emissions. There also seems to be background radiation that decreases
towards the higher wavelengths.

Figure 36 - Plasma spectra collected for various charging voltages for a different delay between the discharges.

Figure 36 shows again a number of spectra collected for different voltages but now the delay
between the two discharges is 0.7μs. Because most of those data were taken at low voltages
between 5-12kV, there is not much increase in the light and there is a big jump when 20kV
were applied. Comparing the two graphs for the two different delays, we see that the
maximum linewidth is reached when the delay is 0.7μs and the charging voltage is 20kV. This
result hints that the plasma density is higher when the two discharges are closer together in
time.

48 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[3.1.3] TIME-DEPENDENT SPECTRAL ANALYSIS OF THE DISCHARGES

Now we focus on time-dependent measurements of the ablative discharges. The
measurements were made possible with the use of a pockel cell. It is a piece of nonlinear
material that when a voltage is applied across it, it rotates the polarization of the light travelling
through it. By crossing two polarizers such that there is no light coming through normally and
then by placing the pockel cell between them, we can apply a voltage for a short (few 100ns)
period which will rotate the polarization and let the light only from that period of time go
through to the spectrograph. By carefully timing the discharge and the pockel cell voltage pulse
we can select light from a specific time period of the discharge to be imaged. Of course the light
attenuates due to the presence of the polarizers and also due to the short time gate, so the
gate has to be adjusted such that there is enough light to yield useful information for the Stark
broadening diagnostic. Typical gates used were 100ns-200ns. In future experiments instead of
the pockel cell an intensified gated camera was installed at the exit of the spectrometer.

Ha Line for various delays during the plasma discharge, 180ns gate, 50 m slit
1
-300ns
0.9 0ns
300ns
0.8 600ns
900ns
0.7 1200ns
Normalized Intensity

0.6

0.5

0.4

0.3

0.2

0.1

0
620 630 640 650 660 670 680 690
Wavelength[nm]

Figure 37 - Detail on the Hα line for various times during a 20kV discharge. The peak of the light is set as the t=0
point.

Typical spectra of the Hα line after background subtraction that were collected are shown on
Figure 37. Time t=0 was set at the peak of the light collected, hence the negative values. There
is clear dependence on the linewidth of the line with respect to the time during the discharge,

Efthymios Kallos 49
 2007 A research proposal for the qualifying examination

showing that peak densities are probably available only close to the beginning of the discharge.
On this data set the charging voltage is 20kV, the gate is 180ns and the entrance slit of the
spectrometer is 50μm.

Plasma light on spectrograph for various charging

voltages
3.E+06
14kV 16kV
Total Intensity Counts on

2.E+06 18kV 20kV

Spectrograph

2.E+06

1.E+06

5.E+05

0.E+00
-200 0 200 400 600 800
Time after peak of main discharge [ns]

Figure 38 - Total light collected at the spectrometer for various charging voltages as a function of time.

Figure 38 shows the total light collected at the spectrograph as a function of time for various
charging voltages. It is clear that there is more light with increasing voltage, and also that the
peak of the light occurs later and later in time for higher voltage, maximizing at 200ns after the
discharge peak for 20kV voltage. This is consistent with the data in the section 3.1.1 where the
light was recorded with a photodiode detector. There is 3-4 times more light collected when
the voltage is increased from 14kV to 20kV.

50 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Peak of Ha line for various charging voltages

7000 14kV 16kV

Total Intensity Counts on

6000 18kV 20kV

Spectrograph

5000

4000

3000

2000

1000

0
-200 0 200 400 600 800
Time after peak of main discharge [ns]

Figure 39 - The intensity peak of the Hα line for various charging voltages as a function of the discharge time.

In Figure 39 we plot the intensity peak of the Hα line over time for the various voltages. Again, a
similar behavior is observed: the line peaks later with higher voltage and the peak is larger for
larger voltages.

Efthymios Kallos 51
 2007 A research proposal for the qualifying examination

FWHM of Ha line for various charging voltages

w/ background

14kV 16kV
25
18kV 20kV
FWHM of Ha Line [nm]

20

15

10

0
-200 0 200 400 600 800
Time after peak of main discharge [ns]

Figure 40 - The FWHM of the Ηα line for various voltages as a function of the time during the discharge.

Now we focus on the full width at half maximum (FWHM) of the H α line which translates
directly to plasma density. Figure 40 shows the FWHM dependence when the background light
is also included. Again, higher voltages lead to larger linewidths. For 20kV the FWHM peaks
roughly where the total light peaks, 200ns after the peak of the main discharge current. There is
also an interesting dip in the linewidth that appears 420ns after the current peak, which is
evident for at least the two higher voltages. It is not clear though whether this is also a real dip
in the plasma density or there is some other light collection issue that may cause this dip to
appear. In order to figure out the plasma density however the background has to be removed
and then the FWHM has to be re-evaluated. The linewidths after this processing are shown in
the next figure.

52 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

FWHM of Ha line for various discharge voltages

No background

14kV 16kV
25
FWHM of Ha Line [nm]

18kV 20kV
20

15

10

0
-200 0 200 400 600 800
Time after peak of main discharge [ns]

Figure 41 - The full width at half maximum of the Hα line for various charging voltages as a function of the time
during the discharge.

In Figure 41 the background has been removed before evaluating the linewidths. We conclude
that the apparent increase in plasma density with charging voltage is not so sharp anymore;
although the density decreases with time for all datasets, the difference between different
voltages is much smaller. This means really that increasing the charging voltage of the ablative
discharge causes a small increase in the density but also a big increase in the plasma
temperature. Remember from section 2.1.3 that in general more background translates to
higher temperature. Finally, a collective picture of all the discharge parameters versus time is
presented in the following figure.

Efthymios Kallos 53
 2007 A research proposal for the qualifying examination

Plasma light during 20kV discharge

Total Light on Spectrograph Peak of Ha Line
FWHM of Ha Line [no back] Peak of 1st Carbon Line
Normalized Light Intensity, Discharge

Peak of 2nd Carbon Line Discharge Current

Light on Photodiode
Current and Ha Linewidth [au]

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-100 100 300 500 700 900
Time after peak of main discharge [ns]

Figure 42 - Time dependent analysis of the 20kV discharge.

In Figure 42 we see how the spectral information relates to the current and light produced in
the discharge. The charging voltage was 20kV. Indeed we see that the linewidth peaks at the
same time as the plasma light. This occurs 200ns after the peak of the discharge current. Also,
there seems to be a small bump in the plasma light shortly after a peak (positive or negative) in
the current oscillation. Sharp variations in the plasma density occur only later in time though.

54 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Figure 43 - Comparison of the plasma density evolution during the discharge for various
parameters. The highest density is achieved for a 500ns delay time between the discharges as
opposed to 700ns.

Finally, in Figure 43 we observe the dependence of the linewidths when experimenting with
different entrance slits and gates. A longer gate allows more light to be imaged, which allows us
to decrease the entrance slit and therefore increase the resolution of the spectrometer. The
results for those different parameters are more or less consistent with each other, which means
that they do not affect our plasma density measurement too much. The most interesting fact
though is that decreasing the delay between the two discharges caused a big increase in the
linewidth compared to the 1.1μs delay that we have been using until then. The FWHM peaked
at 22nm which translates to a density between 4-6x1018cm-3, depending on the temperature
which was unknown for this measurement.

Efthymios Kallos 55
 2007 A research proposal for the qualifying examination

[3.1.4] A SIDE ON THE OPTICS COLLECTION SYSTEM

Figure 44 shows the experimental setup layout that was used during the data recording
for the ablative discharges. The plasma light is emitted and collected through a focusing lens
onto a fiber, which then transmits the light to the spectrograph for imaging. The length of the
fiber is around 30m, as it transfers the light between two rooms that are far apart. The question
then to be answered is how the optics collection system affects the signals that we analyze.
How does the fiber affect the plasma spectrum? How does the spectrograph grating affect the
light amplitude?

Figure 44 - Experimental setup layout for the collection of the plasma light for the ablative discharges. Source:
Daniil Stolyarov.

The following figure shows the dependence of the light recorded at the camera when the
calibrated mercury lamp is fed as an input of the spectrograph. Specifically, we plot the FWHM
of the three visible mercury lines as a function of the entrance slit widening, while having the
gain in the camera constant.

56 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Dependence of Mercury Line Width vs the spectrograph entrance slit width

16

14

12
Mercury Line FWHM[nm]

10

Center Line, 546nm

2 Left Edge Line, 435nm
Right Edge Line, 479nm
0
0 50 100 150 200 250 300 350 400 450 500
Spectrograph Slit Width [ m]

Figure 45 - The FWHM of the mercury lines as a function of the entrance slit width of the spectrograph.

This graph reveals the resolution of the spectrometer for the specific grating used when taking
those data: it is at best 2nm, and it becomes worse when a larger slit is set. This is not a
problem at high densities above 1018cm-3 since the linewidths are measured in the tenths of
nanometers at that range.

Efthymios Kallos 57
 2007 A research proposal for the qualifying examination

Transmittance of Optical Fiber and plasma spectrum

1.4
Optical Fiber Transmittance
Plasma Spectrum
1.2

0.8
Intensity

0.6

0.4

0.2

0
350 400 450 500 550 600 650 700
Wavelength[nm]

Figure 46 - The transmission function of the optical fibers used at the light collection system.

Next, we used a halogen lamp that has a very wide and somewhat flat spectrum in order to see
any effects of the optical fiber on the light imaged. The spectrum was measured both with and
without the fiber, and the above spectral response of the fiber was obtained. The plasma
spectrum is also plotted in the same axes for reference.

58 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Transmittance of optics collection system

2

1.8

1.6

1.4
Transmittance[au]

1.2

0.8

0.6

0.4

0.2

0
450 500 550 600 650 700
Wavelength[nm]

Figure 47 - The spectral response function of the optics collection system. It includes the optical fiber, the
spectrograph, the diffraction grating and the intensified gated camera.

Finally, the spectrum of the halogen lamp was measured right at the capillary and then also
using the optics collection system with all elements included: fiber, spectrograph, grating, and
the camera. The spectral response shown in Figure 47 was obtained. The “real” shape of the
hydrogen lines is therefore slightly altered due to the optics, and the above response can be
used to account for it by inverting it and then multiplying the recorded spectrum with that
inverse function. The difference at the FWHM of the lines when this is performed is not great,
usually around 1-2nm at most.

Efthymios Kallos 59
 2007 A research proposal for the qualifying examination

[3.2] GAS FILLED CAPILLARY DISCHARGES AT USC

In this section we will review a set of data that deals with capillary discharges that we
performed at the Pulsed Power Lab of the University of Southern California. We will investigate
the dependence of the plasma density and temperature on the pressure, the capillary length,
the capillary diameter, the charging voltage, the hollow electrode diameter and finally time.
The capillary was made out of pure glass and in was filled with neutral hydrogen up to 2atm in
pressure.

Figure 48 - At high pressures the electrode was pushed outwards from the hydrogen gas inside the capillary. When
the discharge was fired, the gas still broke down creating a large current on the right side of the capillary. The
current has to find its way towards the other electrode by squeezing inside the tiny 0.6mm diameter of the
capillary, thus increasing the energy density deposited at that region.

60 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[3.2.1] GAS FILLED CAPILLARY DESIGN

After a number of attempts and design reviews, since we decided to built a capillary
from scratch we chose to build the whole capillary shape out of borosilicate glass. A picture of a
typical capillary is shown on Figure 49. We built several capillaries of various lengths and
diameters for our tests. The lengths ranged from 4mm up to 40mm and the diameters from
0.6mm up to 3mm. This glass design offers several advantages. First, glass is non-conducting
with very high breakdown voltage and does not ablate at all for our parameters, resulting in a
clean pure hydrogen spectrum (Balmer lines + background). Second, it is extremely fast and
cheap to manufacture at almost any diameter and length. Third, it is transparent to the light
emitted so it is straightforward to collect and focus the light onto an optical fiber and then to an
optical spectrometer. The OceanOptics USB4000 spectrometer was used for the data collection.

Figure 49 – Photograph of the hydrogen capillary discharge at USC. Source: Jessica

Hao Chen

The electrodes are made out of copper and their ends are made out of Molybdenum. We were
able to see absorption line of Molybdenum in the discharge spectrum. They are hollow, with
holes ranging from 0mm to 3mm because we needed to test whether there is an enhancement
due to the hollow discharge as opposed to just having flat electrodes. The power supply
provided a 75ns FWHM current pulse that peaked up to 1kA, depending on the discharge
parameters. The charging voltage could be adjusted between 24kV and 90kV.

In Figure 50 the experimental setup with the optics collection system is presented.

Efthymios Kallos 61
 2007 A research proposal for the qualifying examination

Figure 50 - Photographs of the light collection system at USC.

Figure 51 - The plasma light emitted with and without the presence of the 0.6mm capillary.

Figure 51 depicts the effect of the capillary into the spectrum and the increase in plasma
density: the black spectrum was taken with a simple glass tube with a diameter larger than the
outer diameter of the electrodes (1cm). The gas pressure was 1atm and the charging voltage
54kV. The red spectrum was recorded with the same discharge parameters but with a 1cm long
capillary with 0.6mm diameter between the electrodes. The presence of the capillary
“squeezes” the current through the tiny diameter, thereby increasing the current density and
accordingly the plasma density also.

62 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Figure 52 presents a number of time-integrated photographs of various discharges, in

order to point out symmetries and intensity gradients along the discharges. Note that much of
the light recorded comes from reflections off the glass volume.

Figure 52 - A collection of images of the plasma discharges in the gas-filled capillaries.

Figure 53 - Typical hydrogen spectrum recorded with the OceanOptics USB spectrometer.

Efthymios Kallos 63
 2007 A research proposal for the qualifying examination

[3.2.2] THE EFFECT OF NEUTRAL GAS PRESSURE

In this section we present a comparison of the spectra recorded for different neutral
hydrogen pressures, ranging from 0-2atm. The data are presented in Figure 54, where the
spectra are normalized to the Balmer Hα line.

Figure 54 - Spectra recorded for increasing neutral hydrogen pressures.

It is evident from this figure that higher plasma densities and stronger light emission can be
obtained for higher pressures. The linewidths of both lines increase, suggesting an increase in
density. This is also accompanied by a change in the ratio of the two line intensities since we
observe that the Hβ/Ηα ratio increases with pressure, suggesting an increase in the plasma
temperature with pressure, according to the line-ratio technique that was described in section
2.1.2. Quantitative measurements of the density and temperature are presented in the
following sections.

For low pressures and densities the line shape of the Hα line remains mostly Lorentzian,
which strengthens the density calculations. For larger pressures the shape is not purely
Lorentzian; however the right side of the Hα line is still Lorentzian enough. In those cases the
double of the half width at half maximum was used to estimate the density.

64 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[3.2.3] THE EFFECT OF THE CAPILLARY LENGTH

Figure 55 presents a set of data that depict the plasma density dependence as
calculated through the Stark broadening of the linewidth of the Hα Line, for different capillary
lengths. Each data point is an average of 3-5 consecutive shots with the error bar being on the
order of 10%, which implied quite consistent discharges. The inaccuracy due to the density
extraction from linewidth and temperature uncertainty is higher than this shot to shot
variation.

Plasma Density for Various Capillary Lengths

60kV Discharge, 0.6mm Diameter
1.1E+19
1E+19
9E+18
Plasma DEnsity [cm-3]

8E+18
7E+18
20mm
6E+18
16mm
5E+18 12mm new pulser
4E+18 8mm new pulser
3E+18 6mm
2E+18 4mm new pulser
1E+18
0
0.0 0.5 1.0 1.5 2.0
Neutral Gas Pressure [atm]

Figure 55 - The plasma density for different capillary lengths.

We observe that in general the density increases with pressure, and at the same time it also
increases with the capillary length, up to a maximum at 12mm length and 9.5x10 18cm-3 for
2atm of pressure. For larger lengths the density drops again, and capillaries that had more than
30mm of length did not always breakdown for out voltage settings due to the Paschen curve
limitations.

Efthymios Kallos 65
 2007 A research proposal for the qualifying examination

A possible explanation for the increase of density with length can be the following. Since
the discharge is based on an avalanche effect, where each electron suffers a number of
collisions along his path from one electrode to the other, a lengthier capillary would imply that
each individual electron is involved in more collisions. In a short capillary the electrons diffuse
out quickly before colliding with enough other particles to increase the plasma density enough.
Then, after some optimal length is crossed, it is tougher to ionize the same number of particles
per unit volume because the acceleration and speed of the electrons dies out before they reach
from one electron to the other. Unfortunately this is just an intuitive explanation which is not
backed up by either a definite experimental proof or a simulation result yet.

Figure 56 shows the temperature dependence for the same capillary lengths as the
previous graph. The temperature was extracted from line to background intensity ratio of the
Ηα line. It is interesting to observe that the temperature also increases with length, and the
maximum temperature is reached at 16mm, while the maximum density was reached at 12mm.
In fact, the absolute linewidth was larger for the 16mm capillary, but according to these results
the large linewidth was related to a big temperature increase as opposed to a plasma density
increase. Since these plasma densities are close to 1019cm-3, the temperature affects the
linewidth more strongly as compared to low densities where the temperature effect is not as
important.

Plasma Temperature for Various Capillary Lengths

60kV Discharge, 0.6mm Diameter
7
20mm
Plasma Temperature [eV]

6
16mm
5
12mm new
4 pulser
8mm new
3
pulser
2 6mm

1 4mm new
pulser
0
0.0 0.5 1.0 1.5 2.0
Neutral Gas Pressure [atm]

Figure 56 - The plasma temperature for various capillary lengths.

66 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[3.2.4] THE EFFECT OF THE CHARGING VOLTAGE

Figure 57 shows a number of data for different charging voltages, 54kV, 66kV and 78kV,
with the same capillary, everything else remaining constant. We used the 20mm long capillary
with 0.6mm diameter.

Plasma Density vs Gas Pressure

1.E+19
Plasma Density [cm-3]

54kV

66kV

78kV

1.E+18
0.0 0.5 1.0 1.5 2.0 2.5
Pressure [atm]

Figure 57 - The plasma density as a function of the charging voltage.

The plasma density reached at each pressure is slightly higher when the charging voltage is
increased from 54kV to 66kV, but there is no significant increase when the voltage is further
increased to 78kV. The extra energy that is deposited seems once again to be transferred to
temperature in the plasma rather than density, as Figure 58 below shows.

Efthymios Kallos 67
 2007 A research proposal for the qualifying examination

Plasma Temperature vs Gas Pressure

5
54kV
Plasma Temperature [eV]

4
66kV

3 78kV

0
0.0 0.5 1.0 1.5 2.0 2.5
Pressure [atm]

Figure 58 - The plasma temperature as a function of the charging voltage.

It is clear that further increase in voltage causes more background and hence more
temperature increase. In addition, higher charging voltages create larger instabilities in the
current pulse delivered to the plasma, since we saw large variations from shot to shot even at
low pressures. As a result, we decided to fix the charging voltage to 60kV for the majority of the
discharges thereafter.

68 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[3.2.5] THE EFFECT OF THE CAPILLARY TUBE DIAMETER

Figure 59 shows a set of data that were taken with three different 20mm long capillaries
at 60kV charging voltage their only difference being the different capillary tube diameter, which
varied between 0.6mm, 1.2mm and 3mm.

Plasma Density for various capillary tube

diameters

1.E+19 3mm

1.2mm
Plasma Density [cm-3]

8.E+18
0.6mm

6.E+18

4.E+18

2.E+18

0.E+00
0 0.5 1 1.5 2 2.5
Gas Pressure [atm]

Figure 59 - The plasma density as a function of the capillary tube diameter.

The data shows that narrower capillaries hugely enhance the plasma density. When the
diameter decreases 2.5 times from 3mm down to 1.2mm the density increases 3.25 times, from
0.8x1018cm-3 to 2.6x1018cm-3. Then, when the diameter is decreased further by a factor of 2,
the density jumps to 7x1018cm-3 which is an increase by 2.7 times. This is probably due to the
enhanced current density (in A/cm2). Assuming that the current supplied is the same over those
discharges, ideally the current density should be inversely proportional to the square of the
diameter of the capillary. The enhancement observed is density is not linear to the current
density but it is still very strong, suggesting that in order for high densities to be achieved a
diameter as small as possible will need to be used.

Efthymios Kallos 69
 2007 A research proposal for the qualifying examination

[3.2.6] THE EFFECT OF THE HOLLOW ELECTRODE INNER DIAMETER

The electrodes used for these discharges are hollow. Be tested 3 pairs of electrodes, one
pair with 3mm hole diameter, one pair with 0.6mm hole diameter (equal to the diameter of the
capillary) and one pair with no holes at all. The capillary that was tested was 16mm long (for
which we measured the highest plasma density earlier) and the charging voltage at 60kV.

Plasma Density for various Electrode Hole Diameters

1.E+19
Plasma Density [cm-3]

3.0mm
0.6mm
0.0mm

1.E+18
0 0.5 1 1.5 2 2.5
Gas Pressure [atm]

Figure 60 - The plasma density as a function of the electrode hole diameter.

Figure 60 shows the dependence versus pressure for the 3 hole diameters tested. We observe
that the “matched” diameter does not create much difference in plasma density, while the
3mm and 0mm hole diameters do not differ much. The density at 7.5x1018cm-3 is measured
with the flat electrodes at a pressure of 1.8atm.

70 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Plasma Temperature for various Electrode Hole

Diameters
7
Plasma Temperature [eV]

6
5
4 3.0mm
3
0.6mm
2
1 0.0mm

0
0 0.5 1 1.5 2 2.5
Gas Pressure [atm]

Figure 61 - The plasma temperature as a function of the electrode hole diameter.

Figure 61 shows the plasma temperature dependence as a function of the pressure for the 3
hole diameters. We observe that the “matched” diameter seems to create the lowest
temperature in the plasma, although the results do not differ much between the three choices
of electrode holes.

Efthymios Kallos 71
 2007 A research proposal for the qualifying examination

[3.2.7] TIME DEPENDENT LIGHT COLLECTION

In the late stages of the experimental data collection we were able to use an intensified
gated camera (Lavison PicoStar HR12) to obtain time-dependent snapshots of the discharge.
The images recorded with a 5ns gate time for different delays after the beginning of the
discharge are shown on the left. The integrated light is shown in a plot as a function of the
delay time in Figure 62 below. Similar to the ablative discharges data, there is a clear
dependence on the time and the total light drops to zero roughly 1.7μs after the discharge. At
this point we were not able to setup a time-dependent optical system that also analyzes the
spectrum (so that we can extract the [plasma density as a function of time) but there are
immediate plans to do so in the near future.

Figure 62 - Time dependent light collection of the gas-filled discharge. A 26mm capillary was tested at 1atm with a
5ns gating time.

72 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[3.2.8] A SIDE ON THE RELIABILITY OF THE SPECTRAL ANALYSIS

There is of course some concern when analyzing the spectra whether the Stark
broadening analysis is valid for those high densities. Can the density and temperature data be
trusted? In this section we will attempt to partially shed some light into that matter.

The main issue associated with Stark broadening at high densities is self absorption; it is
advised that in high densities the actual linewidths may not increase further because the
electrons are so packed together that re-absorb the photons that are emitted from nearby
electrons. It is not accurately known whether this is an important problem. For our
experimental data we measured linewidths (after background subtraction) of more than 30nm,
which correlate to densities at least 1019cm-3. Maybe in fact the actual density of the plasma is
higher but the broadening is not sensitive to it; it is difficult to diagnose whether this is true or
not.

Another issue is the comparison between the Hα and Ηβ lines, and the densities that
occur from their linewidths. In an ideal world if the two lines could be images at the same time,
then their linewidths would be such that the corresponding densities match. Figure 63 below
shows some data from an 8mm long 60kV capillary discharge. For this length the plasma
density is relatively low so that the background light is limited and both lines are approximately
Lorentzian in shape. We then extracted the density derived from those measurements.

Efthymios Kallos 73
 2007 A research proposal for the qualifying examination

Plasma Density vs Gas Pressure

1.E+19
Plasma Density [cm-3]

Hα Hβ

1.E+18
0.0 0.5 1.0 1.5 2.0 2.5
Pressure [atm]

Figure 63 - Comparison between plasma densities extracted from the linewidths of the Ηα and Ηβ lines.

The graph indicates that Hβ always yields a lower density than Hα. In fact this makes sense
because Griem’s theory that extracts the densities from the linewidths does not give the
necessary parameters for densities larger than 1018cm-3 for the Hβ line; we are simply forcing
the theory to gives us an output number. On top of that, the difference between the two lines
is up to a factor of two and keeps increasing for larger densities, indicating that the H α line
seems more appropriate for that density range.

Next, we extracted the temperature from the same set of data using all three available
ways: line to continuum ratio for Hα, line to continuum ratio for Hβ and line ratio Hβ/Ηα.

74 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Plasma Temperature for 8mm Length Capillary,

60kV Discharge, 0.6mm Diameter
7
Plasma Temperature [eV]

6
5 Ha
Hb
4
Ha&Hb
3
2
1
0
0.0 0.5 1.0 1.5 2.0
Neutral Gas Pressure [atm]

Figure 64 - Comparison between the plasma temperatures extracted from the three different methods: line to
background ratio for Hα, line to background ratio for Hβ and the line ratio Hβ/Ηα.

It is encouraging to see that as long as the spectrum looks reasonable and there isn’t a huge
amount of background, then the three methods yield similar results for the plasma
temperature. The line ratio gives the lowest number but it can also provide an answer when
there is not enough background at all in the other lines to extract a temperature. In addition,
low temperatures (less than 2eV) do not affect the density calculation significantly.

Efthymios Kallos 75
 2007 A research proposal for the qualifying examination

[3.3] GAS FILLED CAPILLARY DISCHARGES AT ATF

A limited amount of data was taken with a hydrogen-filled capillary at ATF. A 10mm long
capillary with 1mm diameter was constructed out of Macor, and a discharge voltage from 3kV
up to 25kV was applied for a range of pressures from 30Torr up to 500Torr. Also, 3 different
capacitors were installed at the discharge circuit in order to investigate the effect of the
capacitance (and hence the deposited energy) on the plasma density. Once again, the plasma
density was measured using Stark broadening, and it was found to have a peak at 3x1018cm-3
even for high voltages and pressures. This was attributed to the low ionization fraction of the
gas. The following graphs briefly summarize the findings.

Current Drawn vs Gas Pressure

8kV - C=2.7nF 12kV - C=2.7nF
8kV - C=9nF 12kV - C=9nF
8kV - C=40nF 12kV - C=40nF
2.0
Current [kA]

1.5
1.0
0.5
0.0
0 100 200 300 400 500 600
Pressure [Torr]
Figure 65 - Plasma current recorded for various voltages and capacitances as a function of gas pressure.

Figure 65 depicts the peak current recorded for a different set of charging voltages and 3
different capacitors over the range of neutral gas pressures. We observe that higher voltages
and higher capacitances will generate higher currents. The current peaked at 1.24kA for these
voltages, but it increases only marginally as the pressure is raised 5 times from 100 to 500 Torr.
The maximum current recorded was 2kA for 25kV voltage and 500 Torr pressure.

76 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Another interesting plot is the current drawn as a function of the charging voltage for the 3
different capacitors, shown in Figure 66.

Current Drawn vs Voltage

10mm, 1mm diam
100Torr - C=2.7nF 200Torr - C=2.7nF
500Torr - C=2.7nF 100Torr - C=9nF
200Torr - C=9nF 500Torr - C=9nF
100Torr - C=40nF 200Torr - C=40nF
2.0
Current [kA]

1.5
1.0
0.5
0.0
0 10 20 30
Voltage [kV]
Figure 66 - Plasma current drawn for various voltages and capacitances as a function of charging voltage.

It is interesting to observe that the data fall on straight lines with 3 different slopes for each
capacitance – probably hinting at the constant resistances that occur during each discharge for
the 3 settings.

The corresponding plasma density is shown in Figure 67.

Efthymios Kallos 77
 2007 A research proposal for the qualifying examination

Plasma Density vs Gas Pressure

10mm Length, 1mm diam

8kV - C=9nF 8kV - C=2.7nF 8kV - C=40nF

12kV - C=9nF 12kV - C=2.7nF 12kV - C=40nF

3.0E+18
Plasma density [cm-3 ]

2.5E+18
2.0E+18
1.5E+18
1.0E+18
5.0E+17
0.0E+00

0 100 200 300 400 500 600

Pressure [Torr]
Figure 67 - Plasma density for various voltages and capacitances as a function of gas pressure.

We observe that the density increases consistently with pressure and voltage. Higher voltage
will lead to higher density, as will higher capacitance. The problem is though that Macor ablates
relatively easily at high currents – its resistance is lower than the plasma resistance at low
pressures and as a result the current will flow through the capillary walls also. This is the reason
why the for the highest capacitance the density raises above 2.5x1018cm-3 and then drop again
at high pressure: for low pressure there is ablation as well as gas ionization, and the ablation is
gone at high pressures as the current flows through the gas only. At 12kV and above it was not
possible to get electron density - it was a pure ablation regime. This can be seen in the
spectrum since there is no visible Hα line, but there are other lines and a lot of background light.

This hybrid ablation-ionization mode is seen clearly in the following figure, where a
number of points were taken at low pressures.

78 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Density vs (Static) Pressure

C=10nF

1.60E+18
Plasma Density [cm-3]

1.40E+18
1.20E+18
1.00E+18
8.00E+17 20kV

6.00E+17 12kV
4.00E+17
2.00E+17

0 50 100 150 200 250

Pressure [torr]

Figure 68 - Plasma density for various pressures. The hybrid mode can be seen at low pressures.

We observe that below 15 Torr there exists the hybrid mode where there is both ablation and
ionization – this is because the density is decreasing with pressure but there is still H α line
visible which means that the gas ionizes a little. Then there is a minimum in the density, and as
the pressure is increased more the density increases almost linearly with pressure as expected.
The spectrum at this range consists of a clean Ηα line only.

The ionization fraction for the above set of points is shown in the next figure.

Efthymios Kallos 79
 2007 A research proposal for the qualifying examination

Ionization Fraction vs (Static) Pressure

C=10nF
Ionization Fraction [%]

10000

20kV
1000
12kV

100

10

0 50 100 150 200 250

Pressure [Torr]

Figure 69 - The ionization fraction for various pressures.

Clearly the hybrid/ablation mode is confirmed here since there is more electron density than
the neutral gas can provide at low pressures – obviously the extra electrons have to be
originating from the capillary walls.

Finally, Figure 70 shows a comparison between the plasma current, the plasma light and
the plasma density as a function of time. The data were taken at 24kV voltage and 200 Torr.

80 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

18
x 10 1cm Capillary, 24kV, 200Torr
2
H Mean Light
Mean Current
Plasma Density [cm-3]

1.5 Plasma Density

0.5

0
0 500 1000 1500 2000
Time [ns]
Figure 70 - Time dependent plasma density evolution for the gas-filled capillary discharge at ATF.

We observe that the plasma density peaks approximately were the light also peaks and that
happens several ns after the peak of the current.

Efthymios Kallos 81
 2007 A research proposal for the qualifying examination

82 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Section 4
Suggestions, future designs and issues
In this section we will propose a few designs towards building a better plasma source.
We will not concentrate so much on ablative capillary discharges because they have a number
of disadvantages. First, they are violent processes with non uniform currents produced. Since
the ultimate purpose of the capillary is an electron beam to pass through so that the electrons
get accelerated, it is important to have as small plasma current when the beam goes through
the plasma: strong currents create magnetic fields that deflect the beams out of their
propagation direction, thus destroying any alignment between the beam and the detectors and
also weakening the beam-plasma interaction. Gas-filled discharges on the other hand probably
produce more uniform and symmetric currents, so the deflection effect on the beam should be
weaker. Second, ablative discharges require an extremely high voltage to produce a high
plasma density since the voltage is the only knob available that can tune the density. High
voltages though create a lot of background, ablate large amount of material, pollute the
vacuum of the beamline and have a very short lifetime. For example, the polypropylene
ablative capillary that was tested at Brookhaven only lasted a few dozen shots at 20kV of
charging voltage before we saw the current characteristics decaying.

Producing plasma densities in the range of 1019cm-3 and higher are of interest for the
multibunched resonant PWFA experiments. In those experiments the electron beam is
modulated through the IFEL effect [63-65] to form 100’s of mini bunches that resonantly can
excite the plasma waves [66, 67]. Since the IFEL typically works with a CO2 laser (at least at ATF)
[68], then the separation of the bunches and hence the plasma wavelength is at the wavelength
of the laser, which is 10.6μm, and that corresponds to a plasma density close to 10 19cm-3. In
addition, larger densities are always welcome because the peak amplitude of a plasma
wakefield scales with the square root of the plasma density [69]: the higher the plasma density,
the more electrons exist to support the oscillations and a stronger electric field can be build.

The capillary discharges, as it is clear from the data presented in this proposal, are quite
reliable in generating plasma densities in the 1017-5x1018cm-3 range. In this section we will try to
address issues that arise when we try to push the densities even higher. We will suggest designs
that can support for high pressure discharges, diagnostics that can measure those high
densities, and ultimately we will discuss a number of considerations for installing those plasma
sources and diagnostics online in an accelerator beamline.

Efthymios Kallos 83
 2007 A research proposal for the qualifying examination

[4.1] ENHANCEMENTS FOR THE CAPILLARY DISCHARGES AT USC

With the present setup at the pulsed power lab at USC densities of up to (but not
including) 1019cm-3 can be achieved. For the resonant PWFA experiment at the CO2 wavelength
though, peak densities of at least 1019cm-3 need to be reliably achieved. A few innovations need
to be added so that this extra step to break the 1019 barrier is made. It doesn’t necessarily mean
that radical designs have to be implemented; we simply need some new tricks to push the
density by a factor of 30%-50%.

The first obvious thing to try is to simply push the neutral gas pressure by 0.5-1atm
more; according to the graphs presented in section 3.2.3, if the pressure scaling stays the same
then this is more than enough to yield densities above 10 19cm-3. However the problem is that
the capillary setup and mounting was designed as a vacuum device; it was utilized as a back-lit-
thyratron, a high voltage gas switch which operates in near-vacuum pressures. The glass tube of
the capillary is mounted to the copper electrodes with a rubber O-ring. Though O-rings have
significant friction to hold the electrodes together with the glass, when the inner pressure is
higher than 1atm then the gas pushes the electrodes outwards, creating a larger distance
between the pair of electrodes (thus reducing the plasma density during the discharge) and
eventually creating a leak between the tube and the surrounding air environment. This effect
was pronounced when it was attempted to push the gas pressure beyond 2atm; in every such
case the electrodes were pushed out and the hydrogen gas expanded out with a loud noise.

A future design will utilize a more stable setup where the electrodes are tightly hold
together to sustain pressures of several atmospheres. In addition, the new design will have to
address the issue of electromagnetic radiation: the several kV and several kA short (~75ns
FWHM) current pulses create a large electromagnetic wave that radiates into the room when
the discharge occurs. This radiation interferes with the electronic equipment causing
malfunctions – for example, our USB spectrometer crashed multiple times when the current
drawn was large. In order to avoid that case the two conductors that transmit the current
(towards the anode and the cathode of the) discharge need to be constructed in a symmetric
way, typically one enclosing the other, so that the EM radiation cancels out outside the wire.
This is taken care of by the high voltage cable that transfers the energy to the capillary, but
eventually near the capillary the two conductors have to be split to reach each electrode.
Finally, this shielding design needs to be transparent at least in some small region so that the
light of the discharge can be collected for processing.

84 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

Figure 71 - Proposed shielding device for the capillary discharges at ATF. Credit to
Jessica Hao Chen.

Figure 71 shows a proposed shielding mechanism for the capillary, designed by Jessica
Hao Chen at USC. It takes care of the radiation problem since the ground is built in a symmetric
way cylindrically around the capillary; the current pulse will flow through the anode and the
plasma and then it will return through the shielding copper material back to the high voltage
cable. One portion of the outside copper shield is made out of a thin copper wire grid, which
allows the current to flow and at the same time the light to be collected. The whole design will
be movable in the longitudinal direction such that capillary of different lengths can be
accommodated. This is the most promising solution for increasing the plasma density through
increasing the neutral pressure inside the capillary.

The plasma density produced will probably also increase if a Z-pinch is created during
the discharge. In a Z-pinch the current flowing is extremely large (in the several kA range) and
this creates a strong azimuthal magnetic field, which it turn creates a radially inward Lorentz
force that tends to collapse the electrons towards the axis and as a result greatly enhance the
plasma density at this region. In order to create a large current density typically a thin tungsten
wire that runs along the capillary center longitudinally can be used. In this case the energy
supplied is forced to follow the path of least resistance, which in now the wire instead of the
breaking down of the gas. The current density along that wire is huge and can thereafter start
to ionize the gas. The problem with this technique is that the wire vaporizes after each shot, so
the discharge is not reproducible quickly since another wire has to be inserted. However it will
be a good test to try and see if high plasma densities can be achieved if everything else fails.

Improvements in the current pulse delivered can also be made by modifying the pulse
forming network. The plasma exhibits a certain resistance when the current flows through it,

Efthymios Kallos 85
 2007 A research proposal for the qualifying examination

depending on the capillary and the pressure applied, which also varies with time. The power
supply has its own fixed impedance. Unless the two impedances are equal, the network will not
be matched and the energy transfer will be less than optimum due to reflections. This can be
seen in the discharge current traces where the current is typically oscillating or it is damped. A
matched network will have a current trace that will have a single half-cycle (zero to a peak and
then back to zero) of current.

Another improvement for the capillary discharges at USC will be to time-resolve the
spectrum, similar to the work that was done at Brookhaven (section 3.1.3). The present work at
USC time resolves the total light emitted from the discharge but does not analyze it spectrally in
time. In order to achieve that a monochromator and/or an optical spectrometer will have to be
used, together with the intensified gated camera. The monochromator will be placed right after
the light from the capillary is collected and focused, allowing only the H α line to be imaged onto
the camera CCD. The width of the line imaged will indicate the width of the spectral line which
can be then used to derive the density at a specific point in time. With the optical spectrometer
(not the USB one since it images the light into its internal CCD) the discharge light will be
focused through the entrance slit, and the gated camera will be placed at the exit slit where the
spectrum is formed. This is a favorable setup since the background light will also be imaged,
and this allows the determination of the plasma temperature at the same time.

Usually in time gated measurements there is concern whether there is enough light left
after the gating to form an image. In our case however the transparent glass design of the
capillary allows for a lot of light to be collected, so that we can decrease the gate even down to
5ns which is a great resolution for the microsecond-long discharges. It will allow to pinpoint the
plasma density exactly which is of great interest in the resonant plasma density experiments
where the density has to be adjusted with an accuracy of 3% for an optimal effect.

Getting the density to reach above 1019cm-3 is one thing. Installing the capillary
discharge in an accelerator beamline is a totally different issue. The problems that arise during
such a process are discussed in section 4.3.

86 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

[4.2] THE SCHLIEREN DIAGNOSTIC FOR CAPILLARY DISCHARGES

In this section we will address in some more detail the possibility of implementing a
Schlieren diagnostic to investigate the capillary discharges. Specifically, it can be utilized to
identify the radial density gradients that occur during the discharge. It can also provide
information on the longitudinal dependence of the density and also information with respect to
time, if a pulsed laser beam with duration much less that the duration of the discharge (which is
several microseconds) can be utilized.

The first choice has to do with the laser that can be used. Basically, there is an upper
and a lower limit to the proper wavelength. On the lower side, near the infrared and visible part
of the spectrum lie the eigenfrequencies of the bound electrons, the neutrals and the ions, and
this will contribute to the plasma refractive index (unless the plasma is fully ionized) [70]. Thus
a lower limit of the laser wavelength is 0.5μm-1μm. On the upper side, the plasma acts as a
cutoff and reflects the lasers that have wavelengths that correspond to densities higher than
the plasma density. Since the plasmas we investigate here aim at high densities above 10 19cm-3
and 10.6μm (CO2) corresponds to that density, it is clear a wavelength smaller than that will
have to be used. CO2 could still be used to diagnose densities maybe up to 10 18cm-3, but beyond
this range the Nd:Yag laser with wavelength 1.06μm is probably the ideal solution.

The next aspect we need to investigate is the sensitivity of the method. How does the
beam refraction depend on the laser wavelength and the density gradient in the plasma? This is
simpler to deduct if for now we assume a slab of plasma with a constant longitudinal density
profile. We can examine the traces of collimated rays that are incident in the plasma, which will
travel a path according to the eikonal equation [71]:

d dr
ds ds

Equation 14 - The eikonal equation

If the index gradient is perpendicular to the ray path s , then the deflection angle α is equal to

1 d L d 1 d
a dl L
dx dx dx

Equation 15 - Refraction angle due to perpendicular index gradients

Efthymios Kallos 87
 2007 A research proposal for the qualifying examination

Here x is simply the direction of the refractive index gradient. The plasma index of refraction ne
can be written (away from the cutoff frequency, ne  nc )

ne 1 ne 1
1 1 1 2
n , C  1.113 1013 cm
0 e
1

nc 2 nc 2C

Equation 16 - Approximated plasma index of refraction

Here 0 is the vacuum laser wavelength and we replaced the cutoff frequency according to
2 2 2
me 0 4 c me 0 1
nc
e2 e2 2
0

Equation 17 - The critical density for EM waves

Then the deflection angle can be written as

2
d C L dne
0
a L 1 2 ne
dx 0 2C dx

Equation 18 - Refraction angle for plasma density gradients

This is the main result of Schlieren photography, where the deflection angle is proportional to
the plasma density gradient and also proportional to the square of the wavelength of the laser
used. Thus a CO2 laser will cause roughly 100 times larger deflection angles than a Nd:Yag laser.

The ray tracing technique is applicable as long as the refractive index gradients are not too high,
according to [71]

d dne n ne
 2 c
dx 0 dx 0

Equation 19 – Condition for the validity of the geometric approach

For example for the CO2 laser where densities up to 1018cm-3

could be diagnosed, the maximum density gradient tolerated
is 1021cm-4.

We need to focus our attention on cylindrical

geometries, where the plasma density as a function of the
Figure 72 - Refraction of a ray from a radial distance is sought after: ne n0 n r , assuming
cylindrical plasma
cylindrical symmetry. This is a more complicated case than the

88 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

simple slab. We define an impact parameter x, which is now the distance shown on Figure 72.
The beam deflection is now given in this cylindrical case by [72]

1
a( x) 2 xn0 dr
rmin r r n (r ) x 2 n02
2 2

Equation 20 - Refraction angle for a cylindrical plasma

xn0
Here rmin is the minimum distance of the ray path from the cylinder axis, rmin . The
n rmin
above integral can be solved numerically for the relative density n(r) once the deflection angle
is recorded, or it can be solved analytically is the function r n r is monotonically increasing
with r, through the following set of parametric equations [72]:
R
1 ( x)
( s) dx
2
s x s2
(s)
n( s ) n0e
(s)
r ( s) se

Equation 21 - Plasma density calculation for a cylindrical geometry

The above formulas provide a method to figure out the density gradients based on the
deflection of a single ray as a function of its impact parameter. In reality it is not always easy to
setup tiny pulses like that, especially if the diameter of the capillary is as small as 0.6mm.
Practically, the beam will illuminate the whole volume of the capillary and then the intensity
changes in the Schlieren interference pattern will be proportional to either the derivative of the
index of refraction or to the second derivative of it, as we showed in section 2.4. Even if a
quantitative result of the density gradients cannot be obtained, the qualitative results will yield
a lot of information too on the discharge. For example, a time-dependent recording of the
Schlieren interferogram will reveal a wealth of information regarding the evolution of the
discharge and the motion of the plasma electrons inside the capillary tube [73-76].

Efthymios Kallos 89
 2007 A research proposal for the qualifying examination

[4.3] SUGGESTIONS FOR BEAMLINE INSTALLATION

In this section we will discuss a number of issues that arise when a capillary discharge is
planned to be installed on an accelerator beamline. Specifically, we will focus on the
Accelerator Test Facility (ATF) at Brookhaven, because many of the capillary measurements
were taken there and they have the ability to explore different PWFA techniques easily. The
“downside” is that the ATF beam quality is not great when compared to SLAC for example. SLAC
generates a 42GeV energy, 3nC bunch with 10μm of longitudinal and transverse spot size which
corresponds to 30kA of beam current. ATF generates a 60MeV energy, 0.5nC bunch with
σr=100μm and σz=300μm, which corresponds to a few hundred amperes of current at best.

The strength of the powerful SLAC beam allows for a number of active diagnostics to be
installed that do not affect the beam significantly. For example, micron-thick titanium foils for
CTR and OTR diagnostics and beryllium windows to confine the plasma have no effect on the
shape and emmitance (a measure of its diverging tendency) of the beam. On the other hand at
ATF the beam is weak so that anything that will be along the path of the beam will probably
affect it is some minor or major way. So, one has to be careful on what is inserted along the
beam path.

Figure 73 - Layout of the Accelerator Test Facility beamlines.

90 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

The other important issue with beamline installation is the quality of the vacuum. Typically the
ATF beamline operates at 10-8 Torr of vacuum, because the beam path needs to be clean of any
residual air and dust particles which will defect the beam otherwise. A number of vacuum
pumps constantly operate so that the vacuum is always kept in good levels. However, a
capillary discharge will introduce inevitably a certain number of neutral and ionized atoms in
this vacuum. First, there are the neutral particles
of the gas in a gas-filled capillary, which may
introduce a density of 1-2atm or more. Second,
there are the plasma particles that are created in
either gas-filled or ablative discharges. Once the
plasma is created inside a capillary, it will
immediately tend to diffuse out of the capillary
ends and into the beamline. The velocity of the
diffusion depends on the plasma conditions; the
plasma may actually expand violently outwards
due to the heating and ionization. Pressure that is
Figure 74 - Photo of beamline 1 at ATF. Source: ATF 3-5 times higher than the neutral gas pressure
website. may occur during this expansion. A possible
solution to this vacuum pollution problem is to
shield the capillary ends with windows, such as diamond windows. These windows will have to
be strong enough to contain the plasma and withstand the expansion velocity and pressure
applied to them, but at the same time they need to be weak enough so that the beam is not
affected significantly by them.

Another issue somewhat related to the vacuum pollution is the necessary pressure that
is required to achieve high densities. Most of the experimental data presented in this proposal
regarding gas-filled capillaries were taken using a capillary installed in a closed chamber which
could be filled with enough static pressure of a certain gas. This is probably not possible in a
beamline because due to the presence of the pumps there is no such thing as a static field
pressure: the gas immediately will flow outward and a static pressure will never be established.
In order to overcome this problem the first idea that comes to mind is to supply the gas into the
capillary through a valve system. There will be enough reservoir pressure behind the valve, and
the valve will open only for a short period of time (maybe a few milliseconds) to supply the
minimum volume of gas required to reach a density high enough. If that is the case, a number
of issues must be first addressed: For how long is it acceptable for the valve to open? Is the
pressure inside the capillary the same as the pressure behind the valve? Is the pressure inside
the capillary constant as long as the valve is open? It is not straightforward to answer those

Efthymios Kallos 91
 2007 A research proposal for the qualifying examination

questions because it is not simple to measure the pressure of a dynamic flow of gas through a
gas feed system, into the capillary and out.

We can suggest three possible techniques to overcome these difficulties. The

first option is again to shield the capillary with windows. This will allow a static pressure to be
built inside the capillary, and the discharge will then create the expected plasma density. Again,
however, there is the problem of destroying the beam quality when the beam will go through a
pair of such windows. Even if there is the desired beam-plasma interaction, if the beam
emmitance is much deteriorated then the differences will not be visible in the final beam
diagnostic or energy spectrum.

Another alternative is to shield

the capillary not with permanently
installed windows but with very fast
moving shutters. The shutters will be
closed while the gas fills the capillary
and they will open only when the
beam is about to pass through the
plasma. Of course, the opening speed
of the shutters will have to be faster
than the expansion speed of the gas
in order for this method to have any
effect. If the gas leak into the vacuum
is a problem, then the shutters will Figure 75 - Photo of experimental devices on beamline 1 at ATF.
Source: ATF website.
have to also close equally fast after
they are open for just enough time
for the beam to go through (this is usually picoseconds) and before the neutral gas and plasma
expands out. In any case, a shutter faster than the gas expansion speed is required, which is
probably a very expensive piece of equipment.

A third option would be to let the gas fill the capillary dynamically, as long as there is
confidence on the actual pressure inside the capillary and the flow of gas is for a short period of
time so that it is pumped out first without affecting the vacuum. A good method to achieve this
level of confidence is to measure the plasma density when the discharge occurs at a specific
static pressure first. Then, the plasma density can be measured during the dynamic gas feed.
Assuming that every other parameter remains identical and that the capillary produces reliable
discharges, then the same density should be reached for the same amount of pressure inside
the capillary even if the feed is dynamic (the discharge occurs in a matter of microseconds

92 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

only). One can then correlate the required reservoir pressure to actual dynamic pressure inside
the capillary through the plasma density measurement. Preliminary tests done at the ATF
facility indicate that in the range 10-500Torr of reservoir pressure, the actual pressure inside
the capillary is 2-7 time smaller, depending on the absolute pressure. In general higher pressure
will diffuse faster than low pressure, so it becomes increasingly harder to achieve high dynamic
pressures inside the capillary; possibly several atmospheres of reservoir pressure would be
required.

Density vs Reservoir Pressure

10mm Length, 1mm Diameter, C=10nF

5kV
1.60E+18
8kV
1.40E+18
Plasma Density [cm^-3]

12kV
1.20E+18
1.00E+18 20kV

8.00E+17 25kV

6.00E+17
4.00E+17
2.00E+17
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Pressure [atm]

Figure 76 - Plasma density for various voltages as a function of neutral gas pressure when the gas is supplied
dynamically.

Figure 76 shows measurements made on a gas filled capillary at ATF, where plasma density (as
derived from Stark broadening) is plotted against reservoir pressure for different charging
voltages. The gas was supplied through a valve and a 1m long tube into the capillary with a
valve opening time of 160ms. The discharge was applied as soon as the gas valve closed.
Against intuition, the density seems to drop with pressure. This happens because the actual
pressure is much less inside the capillary, thus creating a smaller plasma density. Also, the
fraction of the gas that fills the capillary is not constant but it becomes smaller at increasing
pressures; this is probably the reason why the curves decrease with pressure.

Efthymios Kallos 93
 2007 A research proposal for the qualifying examination

Density vs (Static) Pressure

10mm capillary

20kV
3.00E+18
Plasma Density [cm^-3]

20kV - Apr18 data

2.50E+18
12kV
2.00E+18
12kV - Apr18 Data
1.50E+18

1.00E+18

5.00E+17

0.00E+00

0 100 200 300 400 500 600

Pressure [torr]

Figure 77 - Plasma density as a function of pressure for static supply of neutral gas.

Figure 77 shows the exact same setup and discharge parameters, but now the gas is fed
statically and not in a dynamic way. Now that the pressure measurement is accurate, the
density keeps increasing with pressure and it does not drop at high values. This is an indication
that the dynamic gas feed may be adequate for generating low densities but it is not very
reliable for producing high plasma densities.

We now focus on the plasma density diagnostics: we have to consider which diagnostics
can be applied online while the plasma source is present on the beamline. Although the plasma
density can be mapped in advance offline, ideally one would like to have a live measurement of
the plasma density shot by shot. Not all diagnostics can be setup online though. This is where
Stark broadening is extremely useful since it is one of the few diagnostics that passively read
the plasma light emitted to determine the density; as long as there is some window to capture
the light then it can work. Thompson and Raman scattering, although harder to prepare and
setup, they would also work as long as appropriate detectors can be placed at the proper
angles and the laser beam can be aligned with the electron beam to probe the wake. On the
other hand, Schlieren photography and laser interferometry require equipment that can only

94 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

be setup offline. In general diagnostics that need not be along the beam propagation axis would
seem to work better online.

Finally, a reliable plasma source needs to produce a peak density slightly higher than the
required density for a certain PWFA experiment. This is because even for gas-filled discharges
the magnetic field generated by the plasma current than flows along the capillary can
significantly deflect the electron beam and destroy the alignments. Usually if there is even a
small current (100A-200A) that is non uniform, the beam will deflect and it may even hit the
capillary walls before exiting the capillary plasma. However the current dies much faster than
the density drops: the current may die after less than a microsecond, while the plasma density
won’t drop more than an order of magnitude during that time. For example, if the operational
plasma density needs to be 1019cm-3, the peak plasma density will have to be at least 2-
3x1019cm-3 so that when the current dies the density is still not lower than the required
operational density and the beam will travel smoothly through.

Many of the effects mentioned here cannot necessarily have a predictable behavior
beforehand. It is not possible to know whether something will work or cause a problem for
sure. The ultimate test is always the actual experiment which dictates what can and cannot be
done, what should or should not be tried. In many cases something is not supposed to work but
it finally does because of some before unpredicted parameter or setting. Nature always has a
new trick up her sleeve for us!

Efthymios Kallos 95
 2007 A research proposal for the qualifying examination

References…
1. Large Hadron Collider. [cited; Available from:
[Link]
2. SLAC. [cited; Available from: [Link]
3. Jackson, J., Classical Electrodynamics. Classical Electrodynamics. 1998: Wiley.
4. Dawson, J., Nonlinear Electron Oscillations in a Cold Plasma. Physical Review, 1959. 113(2): p.
383.
5. Tajima, T. and J.M. Dawson, Laser Electron Accelerator. Physical Review Letters, 1979. 43(4): p.
267.
6. Ruth, R.D. and P. Chen, Plasma Accelerators, in SLAC-PUB. 1985.
7. Chen, P., et al., Energy Transfer in the Plasma Wake-Field Accelerator. Physical Review Letters,
1986. 56(12): p. 1252.
8. Dawson, J.M., Plasma Particle Accelerators. Scientific American, 1987. 260(3): p. 54-61.
9. Blumenfeld, I., et al., Energy doubling of 42GeV electrons in a metre-scale plasma wakefield
accelerator. Nature. 445(7129): p. 741-744.
10. Lee, S., et al., Energy doubler for a linear collider. Phys. Rev. ST Accel. Beams, 2002. 5(1).
11. Malka, V., et al., Electron Acceleration by a Wake Field Forced by an Intense Ultrashort Laser
Pulse. Science, 2002. 298(5598): p. 1596-1600.
12. Geddes, C.G.R., et al., High-quality electron beams from a laser wakefield accelerator using
plasma-channel guiding. Nature, 2004. 431(7008): p. 538-541.
13. Faure, J., et al., A laser-plasma accelerator producing monoenergetic electron beams. Nature,
2004. 431(7008): p. 541-544.
14. Katsouleas, T., Plasma physics: On the node of a wave. Nature, 2006. 444(7120): p. 688-689.
15. Faure, J., et al., Controlled injection and acceleration of electrons in plasma wakefields by
colliding laser pulses. Nature. 444(7120): p. 737-739.
16. Vidal, C.R. and J. Cooper, Heat-Pipe Oven: A New, Well-Defined Metal Vapor Device for
Spectroscopic Measurements. Journal of Applied Physics, 1969. 40(8): p. 3370-3374.
17. Oz, E., et al., Ionization-Induced Electron Trapping in Ultrarelativistic Plasma Wakes. Physical
Review Letters, 2007. 98(8).
18. Spence, D.J. and S.M. Hooker, Investigation of a hydrogen plasma waveguide. Physical Review E,
2000. 63(1): p. 015401.
19. Gonsalves, A.J., et al., Transverse Interferometry of a Hydrogen-Filled Capillary Discharge
Waveguide. Physical Review Letters, 2007. 98(2).
20. Broks, B.H.P., et al., Modeling of a square pulsed capillary discharge waveguide for
interferometry measurements. Physics of Plasmas, 2007. 14(2).
21. Bobrova, N.A., et al., Simulations of a hydrogen-filled capillary discharge waveguide. Physical
Review E, 2001. 65(1): p. 016407.
22. Li, B. and D. Kwok, Multi-dimensional transient process for a pulse ablating capillary discharge:
modeling and experiment. Journal of Plasma Physics, 2004. 70(4): p. 379-395.
23. Zoler, D. and R. Alimi, A proof of the need for consistent treatment in modelling of capillary
ablative discharges. Journal of Physics D: Applied Physics, 1995. 28(6): p. 1141-1152.
24. Zeeman Effect. [cited; Available from: [Link]
25. Stark Effect. [cited; Available from: [Link]

96 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

26. Griem, H., Plasma Spectroscopy. 1964: McGraw-Hill.

27. Kepple, P. and H. Griem, Improved Stark Profile Calculations for the Hydrogen Lines H alpha , H
beta , H gamma , and H delta. Physical Review, 1968. 173(1): p. 317.
28. Griem, H., Spectral Line Broadening by Plasmas. 1974: Academic.
29. Griem, H.R., Stark Broadening of the Hydrogen Balmer-a Line in Low and High Density Plasmas.
Contributions to Plasma Physics, 2000. 40(1-2): p. 46-56.
30. Griem, R., H. Jacek, and O. Wieslaw, Comparison of hydrogen Balmer-alpha Stark profiles
measured at high electron densities with theoretical results. Journal of Physics B: Atomic,
Molecular and Optical Physics, 2005. 38(7): p. 975-1000.
31. Stambulchik, E., et al., Stark broadening of high principal quantum number hydrogen Balmer
lines in low-density laboratory plasmas. Physical Review E (Statistical, Nonlinear, and Soft Matter
Physics), 2007. 75(1).
32. Ashkenazy, J., R. Kipper, and M. Caner, Spectroscopic measurements of electron density of
capillary plasma based on Stark broadening of hydrogen lines. Physical Review A, 1991. 43(10):
p. 5568.
33. Stamm, R. and D. Voslamber, On the role of ion dynamics in the stark broadening of hydrogen
lines. Journal of Quantitative Spectroscopy and Radiative Transfer, 1979. 22(6): p. 599-609.
34. Boddeker, S., et al., Shift and width of the H_{ alpha } line of hydrogen in dense plasmas. Physical
Review E, 1993. 47(4): p. 2785.
35. Edison, N., et al., Characterization of a capillary-discharge plasma. Physical Review E, 1993.
47(2): p. 1305.
36. Oks, E., Advance in diagnostics for high-temperature plasmas based on the analytical result for
the ion dynamical broadening of hydrogen spectral lines. Physical Review E, 1999. 60(3): p.
R2480.
37. Escarguel, A., et al., Highly nonlinear, sign-varying shift of hydrogen spectral lines in dense
plasmas. Physical Review E, 2000. 62(2): p. 2667.
38. Escarguel, A., et al., A single laser spark in aqueous medium. Journal of Quantitative
Spectroscopy and Radiative Transfer, 2000. 64(4): p. 353-361.
39. Oks, E., A new spectroscopic effect resulting in a narrowing of hydrogen lines in dense plasmas.
Journal of Quantitative Spectroscopy and Radiative Transfer, 2000. 65(1-3): p. 405-414.
40. Flih, S.A. and Y. Vitel. Experimental profiles of hydrogen balmer alpha line emitted in weakly non-
ideal plasmas. in SPECTRAL LINE SHAPES: Volume 11, 15th ICSLS. 2001: AIP.
41. Buscher, S., et al., The Stark width and shift of the hydrogen H \α line. Journal of Physics B:
Atomic, Molecular and Optical Physics, 2002. 35(13): p. 2889-2897.
42. Oks, E., Reduction of spectral line shifts due to the acceleration of electrons by the ion field in
plasmas. Journal of Physics B: Atomic, Molecular and Optical Physics, 2002. 35(10): p. 2251-
2260.
43. Olchawa, W., Computer simulations of hydrogen spectral line shapes in dense plasmas. Journal
of Quantitative Spectroscopy and Radiative Transfer, 2002. 74(4): p. 417-429.
44. Gigosos, M., M. Gonzalez, and V. Cardenoso, Computer simulated Balmer-alpha, -beta and -
gamma Stark line profiles for non-equilibrium plasmas diagnostics. Spectrochimica Acta Part B:
Atomic Spectroscopy, 2003. 58(8): p. 1489-1504.
45. Flih, S.A., E. Oks, and Y. Vitel, Comparison of the Stark widths and shifts of the H-alpha line
measured in a flash tube plasma with theoretical results. Journal of Physics B: Atomic, Molecular
and Optical Physics, 2003. 36(2): p. 283-296.

Efthymios Kallos 97
 2007 A research proposal for the qualifying examination

46. Torres, J., et al., An easy way to determine simultaneously the electron density and temperature
in high-pressure plasmas by using Stark broadening. Journal of Physics D: Applied Physics, 2003.
36(13): p. L55-L59.
47. Vitel, Y., et al., Spectra of dense pure hydrogen plasma in Balmer area. Journal of Quantitative
Spectroscopy and Radiative Transfer, 2004. 83(3-4): p. 387-405.
48. Alexiou, S., Stark broadening of hydrogen lines in dense plasmas: Analysis of recent experiments.
Physical Review E (Statistical, Nonlinear, and Soft Matter Physics), 2005. 71(6).
49. Bekefi, G., Principles of Laser Plasmas. 1976, John Wiley and Sons.
50. NIST Spectral Lines Data. [cited; Available from:
[Link]
51. Bremsstrahlung. [cited; Available from: [Link]
52. Sheffield, J., Plasma Scattering of Electromagnetic Radiation. Plasma Scattering of
Electromagnetic Radiation. 1975: Academic.
53. Salpeter, E.E., Electron Density Fluctuations in a Plasma. Physical Review, 1960. 120(5): p. 1528.
54. Gawron, A., et al., He ii P_{ alpha } Stark broadening and intensity ratio of C iv and C iii lines
calibrated with Thomson scattering for high-density plasma diagnostics. Physical Review A,
1988. 38(9): p. 4737.
55. Settles, G.S., Schlieren and Shadowgraph Techniques. Schlieren and Shadowgraph Techniques.
2001: Springer.
56. Hutchinson, I.H., Principles of Plasma Diagnostics. Principles of Plasma Diagnostics. 1987:
Cambridge.
57. Loeb, A. and Z. Kaplan, A theoretical model for the physical processes in the confined high
pressure discharges of electrothermal launchers. Magnetics, IEEE Transactions on, 1989. 25(1):
p. 342-346.
58. Powell, J.D. and A.E. Zielinski, Capillary discharge in the electrothermal gun. Magnetics, IEEE
Transactions on, 1993. 29(1): p. 591-596.
59. Stolyarov, D., et al. Plasma Density Measurements in Hydrogen-Filled and Ablative Discharge
Capillaries Based on Stark Broadening of Atomic Hydrogen Spectral Lines. in ADVANCED
ACCELERATOR CONCEPTS: 12th Advanced Accelerator Concepts Workshop. 2006: AIP.
60. Zeng, X., et al., Laser-induced shockwave propagation from ablation in a cavity. Applied Physics
Letters, 2006. 88(6).
61. Broks, B.H.P., K. Garloff, and J. van der Mullen, Nonlocal-thermal-equilibrium model of a pulsed
capillary discharge waveguide. Physical Review E (Statistical, Nonlinear, and Soft Matter
Physics), 2005. 71(1).
62. Kaganovich, D., et al., Investigations of double capillary discharge scheme for production of wave
guide in plasma. Applied Physics Letters, 1997. 71(20): p. 2925-2927.
63. Palmer, R., Interaction of Relativistic Particles and Free Electromagnetic Waves in the Presence
of a Static Helical Magnet. Journal of Applied Physics, 1972. 43(7): p. 3014-3023.
64. Kroll, N., P. Morton, and M. Rosenbluth, Free-electron lasers with variable parameter wigglers.
Quantum Electronics, IEEE Journal of, 1981. 17(8): p. 1436-1468.
65. Liu, Y., et al., Experimental Observation of Femtosecond Electron Beam Microbunching by Inverse
Free-Electron-Laser Acceleration. Physical Review Letters, 1998. 80(20): p. 4418.
66. Kallos, E., et al. Resonant Plasma Wakefield Experiment: Plasma Simulations and Multibunched
Electron Beam Diagnostics. in ADVANCED ACCELERATOR CONCEPTS: 12th Advanced Accelerator
Concepts Workshop. 2006: AIP.

98 Efthymios Kallos
Capillary Discharges as Plasma Sources for Wakefield Acceleration Experiments
2007

67. Kallos, E., et al. A Multibunch Plasma Wakefield Accelerator. in Particle Accelerator Conference,
2005. PAC 2005. Proceedings of the. 2005.
68. Pogorelsky, I., et al., Plasma-based advanced accelerators at the Brookhaven Accelerator Test
Facility. Laser Physics, 2006. 16(2): p. 259-266.
69. Lu, W., et al., Limits of linear plasma wakefield theory for electron or positron beams. Physics of
Plasmas, 2005. 12(6).
70. Keilmann, F., An infrared Schlieren interferometer for measuring electron density profiles. Plasma
Physics, 1972. 14(2): p. 111-122.
71. Born, M. and E. Wolf, Principles of Optics. 1975: Pergamon.
72. Dodel, G. and W. Kunz, Diffraction limitations for plasma electron density measurements with
schlieren methods. Applied Optics, 1975. 14(10): p. 2537.
73. Villagran-Muniz, M., H. Sobral, and E. Camps, Shadowgraphy and interferometry using a CW
laser and a CCD of a laser-induced plasma in atmospheric air. Plasma Science, IEEE Transactions
on, 2001. 29(4): p. 613-616.
74. Camacho, J.F., D.E. Bliss, and S.M. Cameron, Temporally resolved schlieren images of a laser-
generated plasma. Plasma Science, IEEE Transactions on, 2002. 30(1): p. 42-43.
75. Redel, T., et al., Schlieren diagnostics of pulsed electron beam ablation of polymethyl-
methacrylate. Applied Physics A: Materials Science & Processing, 1992. 54(6): p. 520-522.
76. Benattar, R., C.J. Walsh, and H.A. Baldis, Time resolved schlieren study of highly refracting CO2
laser produced plasma. Optics Communications, 1983. 47(5): p. 324-328.

Efthymios Kallos 99