1 4921298 PDF

Negative hydrogen ion production mechanisms
M. Bacal and M. Wada

Citation: Applied Physics Reviews 2, 021305 (2015); doi: 10.1063/1.4921298
View online: http://dx.doi.org/10.1063/1.4921298
View Table of Contents: http://scitation.aip.org/content/aip/journal/apr2/2/2?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Simulation of cesium injection and distribution in rf-driven ion sources for negative hydrogen ion generationa)
Rev. Sci. Instrum. 81, 02A706 (2010); 10.1063/1.3258607
Plasma diagnostic tools for optimizing negative hydrogen ion sources
Rev. Sci. Instrum. 77, 03A516 (2006); 10.1063/1.2165769
Increasing the space-charge limit and other effects of cesium seeding in hydrogen negative ion sources
Rev. Sci. Instrum. 71, 872 (2000); 10.1063/1.1150317
Negative ion enhancement in caesium-seeded hydrogen dischargesa volume or surface effect?
Appl. Phys. Lett. 75, 2737 (1999); 10.1063/1.125133
Hydrogen negative ion production in an electron cyclotron resonance driven plasma
Rev. Sci. Instrum. 69, 950 (1998); 10.1063/1.1148618
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
APPLIED PHYSICS REVIEWS 2, 021305 (2015)
APPLIED PHYSICS REVIEWS
Negative hydrogen ion production mechanisms

M. Bacal1 and M. Wada2
1
UPMC, LPP, Ecole Polytechnique, UMR CNRS 7648, Palaiseau, France

School of Science and Engineering, Doshisha University, Kyoto 610-0321, Japan
(Received 1 March 2015; accepted 28 April 2015; published online 2 June 2015)
Negative hydrogen/deuterium ions can be formed by processes occurring in the plasma volume and
on surfaces facing the plasma. The principal mechanisms leading to the formation of these negative
ions are dissociative electron attachment to ro-vibrationally excited hydrogen/deuterium molecules
when the reaction takes place in the plasma volume, and the direct electron transfer from the low
work function metal surface to the hydrogen/deuterium atoms when formation occurs on the surface. The existing theoretical models and reported experimental results on these two mechanisms
are summarized. Performance of the negative hydrogen/deuterium ion sources that emerged from
studies of these mechanisms is reviewed. Contemporary negative ion sources do not have negative
ion production electrodes of original surface type sources but are operated with caesium with their
structures nearly identical to volume production type sources. Reasons for enhanced negative ion
C 2015 AIP Publishing LLC.
current due to caesium addition to these sources are discussed. V
[http://dx.doi.org/10.1063/1.4921298]
TABLE OF CONTENTS
I. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A. Need of D negative ion beams for
magnetic confinement fusion . . . . . . . . . . . . .
B. Need of H negative ion beams for high
energy accelerators . . . . . . . . . . . . . . . . . . . . . .
II. THE VOLUME PRODUCTION MECHANISM. .
A. Early direct extraction negative ion sources
(before 1980) . . . . . . . . . . . . . . . . . . . . . . . . . . .
B. Observation of a new H ion formation
mechanism in the plasma volume. . . . . . . . . .
C. Theoretical and experimental work on H
formation by DA to excited H2 molecules . .
D. Rovibrationally excited H2 molecule
populations in low temperature plasmas . . . .
E. Experimental validation of the volume
production mechanism . . . . . . . . . . . . . . . . . . .
F. Effect of wall material . . . . . . . . . . . . . . . . . . .
G. Isotope effect in negative ion production in
H2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
III. VOLUME PRODUCTION H ION SOURCES .
A. The Penning volume negative ion source . . .
B. The magnetically filtered multicusp volume
negative ion source . . . . . . . . . . . . . . . . . . . . . .
1. Role of the magnetic filter . . . . . . . . . . . . .
2. Role of the plasma electrode . . . . . . . . . . .
C. Volume negative ion sources for fusion
applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
D. Volume negative ion sources for accelerator
applications . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1931-9401/2015/2(2)/021305/31/$30.00
IV. SURFACE PRODUCTION MECHANISM . . . . .

A. Early observations of surface produced H
ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
B. Theoretical work on H surface production .
C. Experiments on fundamental H surface
production process. . . . . . . . . . . . . . . . . . . . . . .
V. SURFACE PRODUCTION H ION SOURCES .
A. Surface sources for fusion applications . . . . .
B. Surface sources for accelerator applications .
VI. VOLUME SOURCE OPERATION WITH
CAESIUM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A. Experimental observations . . . . . . . . . . . . . . . .
B. Cs seeded source operation for fusion
experiments and accelerators. . . . . . . . . . . . . .
C. Optimization of Cs seeded operation of H
sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
VII. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
8
I. INTRODUCTION
1
1
2
3
3
3
4
9
10
10
11
11
12
15
18
20
20
20
21
22
23
24
24
25
25
26
27
A. Need of D2 negative ion beams for magnetic

confinement fusion
This paper discusses the negative hydrogen/deuterium

(H /D) ion production mechanisms in the context of negative hydrogen ion sources for magnetic confinement fusion
research and high energy proton accelerator applications.
Toward the route of sustainable power from magnetic
confinement fusion reactions, the International Tokamak
Experimental Reactor (ITER) is currently under construction
at Cadarache in France. ITER will operate to produce net
output of fusion power that exceeds the heating power by a
2, 021305-1
C 2015 AIP Publishing LLC

V
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-2
factor of Q 10 and should produce a self-sustaining fusion

burning plasma for several hundred seconds. However, ITER
is still an experimental device and will not produce any
electricity.
After ITER, the DEMO machine will produce electricity
and demonstrate the requisite technologies to allow commercial production of electrical power. The pathway toward
fusion power generation starts from todays tokamaks such
as JET, passing through ITER and possibly DEMO to reach
a commercial fusion reactor. Stellarators may be a possible
long term alternative to the tokamak (see the European
Fusion Roadmap). The stage of DEMO will link the nuclear
fusion research with electricity generation and DEMO will
be the last machine before a commercial fusion reactor.
The European Fusion Roadmap calls for construction of
DEMO to commence in 2030 at the point when ITER has
successfully demonstrated the Q 10 performance. The interest for H/D ion beams appears due to the need for energetic beams of neutral hydrogen/deuterium atoms for heating
the plasma in tokamaks up to the burning temperature and
sustaining the pulse length by current drive. Neutral beam
heating power as high as 33 MW is planned for ITER, while
135 MW will be necessary for the steady state operation of
DEMO. McAdams1 discusses the physics and technological
challenges for the neutral beam systems to meet the requirements for DEMO and beyond.
In ITER, a D ion current of 40 A will be accelerated up
to 1 MV before negative ions are neutralized to pass through
the confinement magnetic field. The pulse length will be
3600 s. The development of negative ion sources and extractors for long-pulse and high-power ion beams are pursued in
fusion research institutes worldwide, e.g., IPP Garching2
(Germany), Consorzio RFX (Padova, Italy),3 JAEA4,5 and
NIFS6 (Japan).
A beam of energetic hydrogen atoms, also called neutral
beam, can be generated from positive hydrogen ion beams
by letting them pass through a gas neutralizer. However, the
efficiency of neutralization of positive hydrogen ions
declines rapidly as the ion kinetic energy exceeds 60 keV/
amu. The neutralization efficiency becomes prohibitively
low above this energy7 (see Figure 1). The neutralization
efficiency of negative hydrogen ions in a neutralizer cell of
FIG. 1. Neutralization efficiency of positive and negative ions versus their

energy. Reproduced with permission from R. S. Hemsworth and T. Inoue,
IEEE Trans. Plasma Sci. 33(6), 1799 (2005). Copyright 2005 IEEE.
Appl. Phys. Rev. 2, 021305 (2015)
optimum thickness remains acceptable at higher kinetic

energy and is nearly independent on beam energy above
100 keV/nucleon.
Since high energy atomic beams are required in fusion
research, the need for producing their precursors, the negative ion beams, in sufficient intensity becomes urgent.
However, the current densities of negative ion beams are
much lower, by orders of magnitude, than those of positive
ion beams. This imposes the intensive development of negative hydrogen ion sources, which are based on two types of
processes:
in the plasma volume, H/D ions are formed by dissociative attachment (DA) of electrons to ro-vibrationally
excited hydrogen/deuterium molecules
on surfaces, H/D ions are formed by the interaction
between hydrogen/deuterium plasma and low work function surface facing the plasma.
B. Need of H2 negative ion beams for high energy

accelerators
The use of negative ion sources began in the 1950s, e.g.,

in Tandem Van-de-Graaf accelerators.8 In the 1960s, negative ion sources were used in cyclotrons9 and the use of
charge-exchange injection of H ions into high-energy proton accelerators was started.10,11 (See a review by Dimov.12)
The implementation of charge-exchange injection in numerous circular high energy accelerators led to the development
of hydrogen negative ion source research and development
worldwide. The first generation of negative hydrogen ion
sources was charge exchange sources, using a single aperture
proton source. To increase the intensity, multi-aperture sources were used in the early 1970s.13,14 Further increase of the
intensity was obtained by using double charge exchange in
alkali metal vapor.15,16
However, in the 1970s, a new stage in the development
of negative ion sources with parameters sufficient for
modern accelerators was achieved by the invention of the
plasma-surface sources. In these sources, the addition of caesium vapor to the hydrogen discharge generates low work
function surfaces capable to emit H ions under bombardment by plasma atoms or ions. At the Institute of Nuclear
Physics in Novosibirsk (USSR),17 small discharge devices in
planatron (magnetron) and Penning geometry were studied,
while at Lawrence Berkeley National Laboratory (LBNL)18
a surface-plasma source (SPS) was based on a negatively
biased caesiated electrode called converter, collecting the
positive ions extracted from the plasma confined in a magnetic multipole. The H ion production processes in these
sources are designated as the surface production mechanism.
In 1977 at Ecole Polytechnique (Palaiseau, France), the
presence of unexpectedly high fraction of H ions in low
pressure hydrogen discharges was identified.19 This result
was confirmed in 1979 using the photodetachment technique.20 This was the demonstration of the existence of an
efficient volume production mechanism. The H/D ion
source research in the following decades was dedicated to
the understanding of physics underlying these two
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-3
mechanisms. In the meantime, the embodiment of these

mechanisms in the design of new sources was actively
developed.
This paper summarizes the present understanding on
both volume and surface H/D ion production mechanisms.
In Sec. II, concerning the volume production mechanism, we
will present: (a) early experiments with direct extraction
negative ion sources (before 1978) which indicated that H
production in plasmas was not due to mechanisms known up
to the point, (b) H ion formation mechanisms in the plasma
volume consisting of the dissociative electron attachment to
ro-vibrationally excited hydrogen molecules that appeared to
be the principal mechanism leading to negative ion production in plasmas in pure hydrogen/deuterium, and (c) experimental verification of the volume production mechanism by
comparing predictions from the physical models to the measured characteristics in a magnetic multipole device, a typical
plasma source operated with pure hydrogen.
In Sec. III, we will present structures and functions of
components of volume production H ion sources for fusion
and accelerator applications. The H/D surface production
mechanisms have been continuously compared and cross
checked with data obtained from experiments on fundamental processes and those predicted from theory. The efforts
made to explain the measured performance of surface sources out of fundamental data will be discussed in Sec. IV.
Surface production through reflection of positive hydrogen/
deuterium ions and atoms at a low work function metal
surface was theoretically predicted and experimentally
proved as an efficient process to produce H/D ions.
Measured results of energy/momentum spectra of H ions
extracted from a surface production type ion source revealed
that the reflection component had a large angular spread
causing an increased emittance of the extracted beam.
Procedures for optimizing surface condition were investigated and adopted to design ion sources of better beam quality. Descriptions of some surface production H ion sources
dedicated to fusion plasma heating and accelerator applications are given in Sec. V.
In Sec. VI, the caesium seeded operation of recent ion
sources is described. It will be shown how both volume and
surface production mechanism interact in the actual operation of caesiated negative ion sources. Conclusions are
presented in Sec. VII.
Appl. Phys. Rev. 2, 021305 (2015)
source. However, the negative ion currents were in the tens

of microamps range. In the same year (1965), Ehlers at the
Lawrence Berkeley National Laboratory (US) reported a
continuous H current of 5.3 mA (with the current density
up to 40 mA/cm2) from a hot-cathode Penning-type source.18
In 1973, a magnetron-type source, the planotron, was developed by Belchenko et al.24,25 in the Budker Institute for
Nuclear Physics in Novosibirsk. In this source, the discharge
chamber has the shape of a race track and the plasma is kept
circulating in the narrow space by the ExB drift motion. A
short pulsed beam (1 ls, 3 Hz) of H ions with a current of
20 mA was obtained through an emission slit of 1 10 mm2.
B. Observation of a new H2 ion formation mechanism
in the plasma volume
In 1977, Nicolopoulou et al.19 calculated the equilibrium density of hydrogen negative ions in a low pressure
hydrogen plasma by taking the known production and
destruction processes into account. At that time, the known
production processes due to low energy electrons were dissociative electron attachment to hydrogen molecules in their
ground state and dissociative recombination of molecular
positive ions H2. Measurement of the negative ion density
in a plasma produced in a magnetic multipole using a probe
diagnostic technique indicated that the measured H ion
density was higher by a factor of 100 than the equilibrium
density predicted from the calculation based upon the data of
known formation and destruction processes. This result was
confirmed in 1979 by Bacal and Hamilton20 using a photodetachment diagnostic technique for the measurement of the
H ion density. They also extended the measurements to a
deuterium plasma. Figure 2 shows the dependence of the
negative ion density n upon the electron density ne in H2
and D2 plasmas confined only by their ambipolar potential.
The neutral gas density was 2 1014 cm3. Figure 2 indicates that the negative-ion density is not isotope dependent
but increases approximately in proportion to ne3 in the studied range of electron density. In H2, the electron temperature
varied in the range from 0.10 to 0.40 eV, while at maximum
density the ratio of n/ne was 0.35.
II. THE VOLUME PRODUCTION MECHANISM

A. Early direct extraction negative ion sources
(before 1980)
A review by Prelec and Sluyters21 describes the early

direct extraction negative ion sources in pure hydrogen
through categorizing sources into four types: duoplasmatrons, hollow discharge duoplasmatrons, Penning sources,
and magnetron sources. Moak et al.22 from Oak Ridge first
reported in 1959 the possibility of extracting an H beam
directly from a duoplasmatron. In 1965, Lawrence et al.23
found that the intensities of H ion currents could be
increased while those of co-extracted electron currents
reduced by off axis extraction from their duoplasmatron
FIG. 2. Dependence of the negative ion density n- upon the electron density
ne in H2 and D2. Reproduced with permission from M. Bacal and G. W.
Hamilton, Phys. Rev. Lett. 42, 1538 (1979). Copyright 1979 American
Physical Society.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-4
Appl. Phys. Rev. 2, 021305 (2015)
Two mechanisms, unknown at that time, involving intermediate states to form H were proposed19,20 in order to
explain the observed nonlinear dependence of the H density
upon the plasma density and the high H density: (1) DA of
electrons to vibrationally excited hydrogen molecules,2628
and (2) DA to electronically excited long-lived states of
hydrogen molecules H2(C3Pu), with assumed higher DA
cross sections.29
C. Theoretical and experimental work on H2 formation
by DA to excited H2 molecules
The theoretical work on DA started in the 1960 s. The

resonance theory of DA, involving the concept of an intermediate resonance negative ion state, was formulated in the
fundamental papers of Bardsley30 and OMalley.31 The
theory of dissociative electron attachment to hydrogen molecules relies on the model of resonance or temporary negative
ion.32 These terms designate a compound state formed by
the interaction of an incident electron with a target molecule
in which the incident electron is temporarily captured in the
neighborhood of the molecule. The decay of this compound
state can lead to DA with formation of a negative ion H
and a neutral atom H
H2 v00 e ! H2 ! H H;
(1)
or to the vibrational excitation of the molecule (denoted by

e-V excitation)
H2 v00 e ! H2 ! H2 v00 Dv00 e:
FIG. 3. Dissociative attachment cross section for various vibrational states

of H2, each in the rotational state J 0. Reproduced with permission from J.
M. Wadehra and J. N. Bardsley, Phys. Rev. Lett. 41, 1795 (1978). Copyright
1978 American Physical Society.
(2)
In 1979, theoretical26 and experimental27 work showed

that the cross sections for DA to vibrationally excited H2
molecules should increase by four orders of magnitude from
v00 0 to v00 4. The cross sections calculated by Wadehra
and Bardsley26 are shown in Figure 3 for the vibrational excitation, and in Figure 4 for the rotational excitation. Figure
5 shows the direct comparison of the internal-state dependence of threshold energy for dissociative attachment cross
sections in H2 and D2 determined in the experiment of Allan
and Wong27 with the results of the theoretical calculation of
Wadehra and Bardsley.26 This figure shows that the increase
is even stronger for deuterium. Note on Figure 5 that the
cross section enhancement by vibrational excitation is much
larger than that by rotational excitation at the same internal
energy.
In 1984, Wadehra33 found that the ground electronic
state of H2 can support at least 294 rovibrational levels and
calculated the rate coefficients of dissociative electron
attachment to these levels. He found that the maximum possible rate coefficient is about 108 cm3/s, as can be seen in
Figure 6, which shows these rate coefficients for all possible
J levels for v00 0, 2, and 4 at the electron temperature of
1 eV. An important conclusion of Wadehra33 is that the crucial factor that determines the attachment rates and average
energy of the extracted negative ions is the total internal
energy of the gas molecule.
According to Wadehra,34 the vibrational and rotational
enhancement of the DA cross sections arises from an increase
FIG. 4. Dissociative attachment cross section for various rotational states of

the ground vibrational state of H2. Reproduced with permission from J. M.
Wadehra and J. N. Bardsley, Phys. Rev. Lett. 41, 1795 (1978). Copyright
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-5
FIG. 5. Internal state dependence of threshold dissociative attachment cross

sections in H2 and D2. Reproduced with permission from M. Allan and S. F.
Wong, Phys. Rev. Lett. 41, 1791 (1978). Copyright 1978 American Physical
Society.
Appl. Phys. Rev. 2, 021305 (2015)
These rate coefficients can be as high as 108 cm3 s1 when

the molecules are in the highest vibrational states.
The DA cross sections for these excitation levels of the
rovibrational hydrogen and deuterium molecules were calculated independently by Gauyacq35 near the energetic threshold. The results account well for the increase of the threshold
attachment with the internal energy of the molecule,
observed by Allan and Wong.27 The vibrational and rotational enhancement of the DA process is similar to that found
by Bardsley and Wadehra.26
The theory of DA was further developed more recently
by Fabrikant et al.36 and Domcke.37 Domckes review37
described the application of the resonance model to the
theory of inelastic electron-molecule collisions. An

improved nonlocal resonance model proposed by Cizek
et al.38 was used by Horacek et al.39 for the calculation of
cross sections of electron DA of molecular hydrogen by the
impact of low-energy electrons. This model also shows that
the magnitude of the DA cross section in the low-energy
range increases rapidly with increasing vibrational quantum
number of the target molecule; the DA cross section
increases by nearly five orders of magnitude from v00 0 to
v00 89. This feature is illustrated in Figure 7 where DA
in the survival factor, which is the probability that the resonant state dissociates without autoionization. This factor is
increased because electron capture can occur at larger distance due to the larger amplitude of the vibrational motion in
excited vibrational states and to the centrifugal stretching in
excited rotational states. The dependence of the DA rate coefficients on the electron temperature is reported by Wadehra34
for all the vibrational and rotational states of H2 and D2.
FIG. 6. Rate coefficients of dissociative electron attachment to all possible J

levels for v 0, 2, and 4 at electron temperature of 1 eV. The vertical line at
3.994 eV defines the internal energy above which the attachment process
becomes exoergic. Reproduced with permission from J. M. Wadehra, Phys.
Rev. A 29(1), 106 (1984). Copyright 1984 American Physical Society.
FIG. 7. Dissociative attachment cross sections calculated by Horacek et al.

for a set of initial target states v00 . The rotational quantum number is J 0.
Reproduced with permission from Horacek et al., Phys. Rev. A 70, 052712
(2004). Copyright 2004 American Physical Society.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-6
cross sections calculated by this theory are plotted for a series of initial vibrational states, v00 011.
However, the theory of Horacek et al.39 leads to a new
view on the effect of rotational excitation on the DA cross
section. Figure 8 shows the effect of rotational excitation of
the target for v 0 as obtained from this theory. It shows
that with increasing J, the rotational heating increases the
cross section more efficiently than the vibrational heating,
opposite to what is well known at low J. The highest cross
section for an endothermic DA process is obtained for
v00 2, J 23. This reaction opens at 113 meV and the cross
2 at 249 meV.
section reaches the value of 20.1 A
D. Rovibrationally excited H2 molecule populations in
low temperature plasmas
Following the above pioneering works, it was questioned what processes are generating sufficient density of
excited hydrogen molecules at higher ro-vibrational levels.
A review on the physics aspects of negative ion sources40
indicated that both electron-molecule collisions in the
plasma volume and recombinative desorption of atomic
hydrogen from wall surfaces are contributing to the
FIG. 8. Dissociative attachment cross sections calculated by Horacek et al.

for v 0 and J 1, 2, , 30. Reproduced with permission from Horacek
et al., Phys. Rev. A 70, 052712 (2004). Copyright 2004 American Physical
Society.
Appl. Phys. Rev. 2, 021305 (2015)
generation of ro-vibrationally excited molecules. An effective source of highly vibrationally excited molecules, generated by electron-molecule collisions, was identified in 1980
by Kunkel:41 the radiative decays from higher electronic
states excited by collisions of ground state molecules with
energetic primary electrons
H2 1 Rg ; v00 0 e ! H2 B1 Ru ; C1 Pu e;
(3)
H2 B1 Ru ; C1 Pu ! H2 X1 Rg ; v00 h:
(4)
This process is denoted by E-V singlet excitation.

Electrons colliding to H2 molecules in their ground electronic state, X1Rg(v), will generate excitations to all members of both the singlet and the triplet electronic levels. A
major portion of the singlet excitations will return to the
ground state by direct or successive radiative decays.
Though this process, a final vibrational population distribution, X 1Rg(v00 ), that differs from the original v distribution
is generated. The triplet decays will not result in molecular
final products but will generate atomic fragments via the
dissociation process instead. The electron excitation cross
sections leading from v00 0 of the ground state through
the B and C singlet states to v00 levels of the ground state
(Eqs. (3) and (4)) were calculated by Hiskes.42,43 These and
other earlier calculations44,45 were extended to include excitations from v00 > 0 initial levels.43 The principal excitations
in this excitation-radiative sequence (the E-V process) proceed through the B1Ru and C1Pu singlet states. Their cross
sections are shown on Figure 9.43 The maxima of all cross
sections decrease monotonically with v00 . The v00 3, 4
results are not shown. The populations of H2 X1Rg (v00 ) in
the upper ten vibrational levels will provide the major portion of the total dissociative attachment leading to negative
FIG. 9. Cross section r(v00 ), for the excitation of the vibrational levels of
H2X1Rg by electron collisional excitation through the B and C states.
Reproduced with permission from J. Appl. Phys. 70, 3409 (1991). Copyright
1991 AIP Publishing LLC.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-7
ions in a hydrogen discharge. The effective cross section for

excitation of this group of vibrational levels is approximately
3 1017 cm2 at 100 eV (Ref. 42) and is large enough to
assume that the excitation process considered here should
contribute significantly to the total negative ion formation.
Concerning the cross sections for E-V excitation for
higher initial levels, Table V in Ref. 43 shows that the largest
r(v, v00 ) excitations occur along the diagonal, v00 v, with
cross sections diminishing from these maximum values as
one moves away from the diagonal. The population distributions of Cacciatore et al.45 for v 6 and 10 show quite similar trends, with the notable maxima at v00 v. The analysis of
the collisional attenuation of the E-V process, summarized in
Table VI of Ref. 43, showed that for electron densities above
1015 electrons/cm3 and the electron temperature kT 35 eV,
the E-V process is effectively quenched, due to secondary
excitation and ionization.
The vacuum ultraviolet (VUV) emission from a low
pressure hydrogen plasma, associated with the process
described by Eq. (4), has been used by Graham46 to study
the production mechanism for vibrationally excited ground
state molecules. The most interesting observation is that
there is a substantial contribution to the high vibrational
levels of the ground state from both B1Ru and C1Pu states.
Several experimental techniques were dedicated to in situ
observation of ro-vibrationally excited molecules in low
pressure hydrogen plasma. Pealat et al.47 used Coherent
Antistokes Raman Scattering to observe ro-vibrationally
excited molecule populations in a multicusp discharge. In
this experiment, molecules with v00 3 only were detected.
Hall et al.48 observed vibrational excitation up to v00 9 in a
gas cell without a discharge, when a tungsten or tantalum filament was heated. An electron beam of fixed energy crosses
the molecular beam of hydrogen flowing out from the cell to
measure the H ion yield due to dissociative electron attachment. Then, the relative vibrational populations were determined from the observed peak intensities by use of the
dissociative attachment cross sections of Gauyacq.35 Figure
10 presents the vibrational populations from Ref. 48 as a
function of vibrational energy for hot tungsten and tantalum
filaments. The proposed mechanism is recombinative
desorption of atomic hydrogen on the cell surfaces following
atomization of H2 on the filament.
This detection technique was used by Schermann
et al.49 to observe the effect of the discharge on the vibrational populations of H2 detected up to v00 6. It was
observed that the discharge reduced excitation to v00 4 levels. When the tungsten filaments were replaced by tantalum
filaments, higher populations for v00 4 were observed. The
tantalum covered surfaces were irreversibly modified by the
discharge and were losing much of their ability to populate
high vibrational levels. The initial populations are recovered
only after a long bake out of the walls. With higher discharge
currents, the vibrational distributions are approaching a
Boltzmann distribution.
Vacuum ultraviolet laser absorption spectroscopy has
been employed by Young et al.50 to measure the populations
and temperatures of ground state H atoms and vibrationally
excited H2 molecules in a volume H source. The
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 10. Vibrational populations as a function of vibrational energy for

tungsten and tantalum filaments. Reproduced with permission from Hall
et al., Phys. Rev. Lett. 60(4), 337 (1988). Copyright 1988 American
Physical Society.
experimental apparatus has been described in detail by

Stutzin et al.51,52 The vibrational population distribution was
found to exhibit a Boltzmann distribution for v00 1 through
v00 8, with Tvib 4150 K (see Figure 11). This result is in
contrast to the prediction from the source chemistry for the
same source conditions by Hiskes and Karo53 exhibiting a
plateau for vibrational levels v00 5, which disagrees with
this experiment, overestimating the higher v00 levels (v00 3).
For determining the densities of rovibrationally excited
molecules in a low-temperature plasma produced in a
FIG. 11. Vibrational population distribution for a volume source. The experimental points (the closed circles) were obtained at 25 A, 120 V, and 8
mTorr H2. The solid line corresponds to a vibrational temperature of
4150 K. The triangles represent the calculated values of Hiskes and Karo,53
normalized to the experimental data at v00 1. Reproduced with permission
from AIP Conf. Proc. 210, 450 (1991). Copyright 1991 AIP Publishing
LLC.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-8
Appl. Phys. Rev. 2, 021305 (2015)
magnetic multipole plasma source with dc filament excitation, Mosbach54 applied laser-induced fluorescence spectroscopy in the VUV (for molecules with high v00 ) and optical
emission spectroscopy (for molecules with v00 < 4). Mosbach
reports that the population distribution is non-Boltzmann.
For high vibrational states, a plateau is formed, and for small
discharge current a hump appears (see Figure 12).
E. Experimental validation of the volume production
mechanism
The existence of an afterglow peak in the extracted negative ion current from a pulsed hydrogen discharge in a multicusp ion source was used as an experimental validation of
the volume production mechanism.55 In the experiment of
Hopkins and Mellon,55 the vibrationally excited molecules
are produced by energetic electrons and they live some time
after the end of the discharge pulse. Low energy electrons
attach to these molecules in the afterglow, while electron
detachment by fast electrons is reduced. As a result, an
increase of the negative ion density and extracted current
occurs as shown in Figure 13.
An important diagnostic effort by Mosbach allowed one
to compare the H ion density measured by laser induced
photodetachment to that calculated from the measured spectrum of ro-vibrationally excited levels of hydrogen molecules (shown in Figure 12). The calculated density of the
negative ions is in good agreement with the measured H
density, as shown in Figure 14.54 This agreement proves the
validity of the volume production mechanism, on which the
model used in the calculation is based.
F. Effect of wall material
Vibrational level distributions of hydrogen molecules

change depending upon discharge filament materials as the
result of the filament material deposition on the source wall.
The effect of wall material on the H ion production has
been studied by two approaches. In the first, Leung et al.56
and Fukumasa and Saeki57 installed thin metal liners on the
chamber wall and compared the extracted negative ion current measured with liners made of different materials. They
FIG. 12. Total density of vibrational states as a function of discharge current. Reproduced with permission from T. Mosbach, Plasma Sources Sci.
Technol. 14, 610 (2005). Copyright 2005 Institute of Physics.
FIG. 13. The extracted negative ion current from a pulsed hydrogen discharge. The gas pressure is 2.4 mTorr, the discharge pulse length is 2.7 ms,
the repetition rate is 87 Hz, and the discharge current is 15 A. Reproduced
with permission from M. B. Hopkins and K. N. Mellon, Phys. Rev. Lett.
67(4), 449 (1991). Copyright 1991 American Physical Society.
found that Al and Cu liners produced higher extracted H

current compared to other metals. Stainless steel produced
the lowest H current. In the second approach, Inoue et al.58
modified the way to change the wall material by evaporating
filament material and found that the extracted H current
increased up to 50% larger, while the extracted electron current reduced to 20% lower when tungsten filaments were
replaced with tantalum filaments. The work of Hall et al.48
and Schermann et al.49 as well as the work with H ion sources at TRIUMF (TRI University Meson Facility)59 and that
at Culham confirmed the observations by Inoue et al.58 that
the performance of the sources was better with tantalum covered walls than that with tungsten covered walls. At DESY,
Peters60 showed that a tantalum cylinder inserted in the collar of the extraction opening increased the extracted current
by 15% with respect to that found with a steel collar. Spence
and Lykke61 sputtered a thin layer of tantalum on the inner
surface of a microwave-driven ion source to enhance H
production and avoid the difficulties with a bulk liner used in
their earlier experiments. Bacal et al.62 studied plasma chambers in which the walls were covered by tungsten or tantalum
films, resulting from evaporation and/or sputtering of
FIG. 14. Comparison of H ion densities measured by laser induced photodetachment (full squares) and calculated from vibrational population distributions of ground state hydrogen molecules (open squares). Reproduced
with permission from T. Mosbach, Plasma Sources Sci. Technol. 14, 610
(2005). Copyright 2005 Institute of Physics.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-9
corresponding filaments, and also covered by caesium, as a

result of Cs seeding. They showed that tantalum and caesium
films deposited on the walls lead to an enhancement of
extracted H current (by a factor of 2) and to a large reduction of the extracted electron current compared to a deposited tungsten film (see Figure 15). It was also shown that a
fresh tantalum film leads to the enhancement of the negative
ion density by 60%90%. Addition of argon leads to a slow
decrease of hydrogen negative ion density, while the electron
density stays constant. The final minimum negative ion density is approximately 40% lower than the initial one in pure
hydrogen plasma.62
The effect of surface reactions in volume hydrogen negative ion sources has been extensively discussed.56,6366
Leung et al.56 attributed the effect of the wall material to different secondary electron emission coefficients. Inoue
et al.58 considered that two surface reactions were important
in this context: the collisional wall de-excitation of vibrationally excited molecules and the wall recombination of atoms
(followed by the recombinative desorption of hydrogen
molecules). In low-power operation, the de-excitation of
vibrationally excited molecules through wall collisions may
be the dominant destruction mechanism for these precursors
of the H ions.
G. Isotope effect in negative ion production in H2
Figures 16 and 17 show the dissociative attachment

cross sections for D2 near threshold calculated by Bardsley
FIG. 15. Variation of extracted negative ion and electron currents on plasma
electrode bias. The effect of films of tantalum, tungsten, and caesium deposited on the walls is shown. Reproduced with permission from Rev. Sci.
Instrum. 75, 1699 (2004). Copyright 2004 AIP Publishing LLC.
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 16. Dissociative attachment cross sections for various rotationless

vibrational states of deuterium molecules. Reproduced with permission from
J. N. Bardsley and J. M. Wadehra, Phys. Rev. A 20, 1398 (1979). Copyright
and Wadehra26,67 for various vibrational and rotational states

of the D2 molecule. In both cases, the enhanced cross sections associated with excited states arise from an increase in
the survival factor, which is the probability that D2 dissociates before losing the third electron. These figures are equivalent to those shown in Figures 3 and 4 for dissociative
attachment in H2. Figure 5 shows the comparison of the mentioned theoretical calculations with the experimental results
of Allan and Wong27 for the threshold DA cross section.
A pronounced isotope effect was observed in the DA
cross section for formation of H and D from H2, HD, and
FIG. 17. Dissociative Attachment cross sections for various rotational levels
of the ground vibrational state of the deuterium molecule. Reproduced with
permission from J. N. Bardsley and J. M. Wadehra, Phys. Rev. A 20, 1398
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-10
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 18. The energy dependence of the

total cross section for dissociative
attachment in H2, HD, and D2 near
3.7 eV. Part (a) shows the experimental
results of Schulz and Asundi (1967)
and part (b) shows the unfolded cross
section as reported by Chen and
Peacher (1968). Reproduced with permission from G. J. Schulz, Rev. Mod.
Phys. 45, 423 (1973). Copyright 1973
American Physical Society.
D2. See Ref. 32 for a review and Figure 18 for the experimental results of Schulz and Asundi68 in the threshold range
from 3.75 to 5 eV. The experimental results measured in
the range from 7 to 18 eV by Rapp et al.69 are shown in
Figure 19. It should be noted that the experimental curves of
Schulz and Asundi shown on the left part of Figure 18
have peak cross sections differing by orders of magnitude
(1.6 10 21 cm2 for H2, 1 1022 cm2 for HD, and 8
1024 cm2 for D2). Whereas the H/H2 cross section was
obtained with an electron energy distribution of 0.1 eV, the
curves for HD and D2 had to be taken with an energy distribution of 0.45 eV in order to gain sensitivity. This accounts
for the difference in threshold behavior.
The isotope effect was first interpreted by Demkov70,71
as the consequence of the smaller survival probability factor
for heavier atoms. More recent theoretical studies of the
isotope effect have been reported by Mundel, Berman, and

Domcke72 and Horacek et al.39
III. VOLUME PRODUCTION H2 ION SOURCES
A review presenting the H ion sources based on volume production by dissociative attachment (denoted Volume
Production Sources, VPS) was published in 2005.73 The
present article covers only a brief introduction to the development of VPS. We refer the readers to the mentioned
review for a detailed presentation of this development, as
well as the brief review dedicated to fusion experiment application.74 Shortly, after the early attempts (1959) to extract
directly negative ions from a duoplasmatron, three main
types of VPS became available:73 the Penning discharge ion
source9 (1965), the tandem multicusp plasma container75,76
(1980), and several versions of RF sources.60,61 Magnetic
multicusp VPS for fusion applications have been developed
in the US,75 France,76 UK,77 and Japan.78,79
A. The Penning volume negative ion source
FIG. 19. Measured total cross sections for negative-ion formation in H2,
HD, and D2 as functions of electron energy in a total-ionization tube.
Reproduced with permission from Rapp et al., Phys. Rev. Lett. 14, 533
In 19631965 at Berkeley, the hot-cathode reflex-type

or Penning plasma source of D and H ions with continuous current of 2 and 5.3 mA (current density of 40 mA/cm2),
respectively, was developed by Ehlers.9,80 The scheme of a
reflex-type ion source is shown in Figure 20. The Ehlers
source can easily be built into the center of cyclotrons and is
used in many cyclotrons for accelerating negative ions.
In Dimovs review,12 the Ehlers H ion source was
reported to be used until 1987 in the TRIUMF cyclotron
(meson factory), but the source was installed outside. The
beam of H ions was injected along the cyclotron axis with
an energy of 300 keV. The Ehlers type source was also used
in a Van de Graaf accelerator which accelerated H ions up
to 1.5 MeV, with the current of 1 mA and the pulse duration
of 1.5 ms.81 Later, a current of 11 mA has been obtained by
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-11
Appl. Phys. Rev. 2, 021305 (2015)
in the caesium-mode operation (350 mA/cm2 was obtained

with an arc current of 55 A).
B. The magnetically filtered multicusp volume
negative ion source
An important step in the development of the negative

ion sources was the understanding that the volume production was due to the following two-step process:
FIG. 20. Scheme of reflex-type ion source. (1) Heated cathode, (2) cathode
aperture, (3) anode chamber, (4) extraction electrode, (5) holes for gas supply, (6) discharge column, and (7) anticathode. Reproduced with permission
from Rev. Sci. Instrum. 67, 1622 (1996). Copyright 1996 AIP Publishing
LLC.
Osher with a Calutron source (see Ref. 82) with an extraction

slit of 3.2 mm 57 mm.
Twenty years later, some basic understanding of the
operation of the Penning type H ion source was presented
by Jimbo et al.83,84 and by Goretsky et al.85 based on the volume production mechanism. The plasma of a Penning-type
discharge forms two regions: the central hot plasma region,
containing fast electrons oscillating between the hot cathode
and the anticathode (reflector), and the cold plasma region
between the hot plasma column and the inner cylindrical
wall of the source (Figure 20). The ionization and the vibrational excitation of molecules occur in the central region.
Outside this central hot plasma column, the electron temperature decreases in the radial direction. H ions are formed by
dissociative attachment to vibrationally excited molecules in
the relatively cold surrounding region, close to the emission
slit. Goretsky et al.85 reported the extraction ion current of
40 mA, with the current density of 80 mA/cm2. It should be
noted that in both the initial source of Ehlers,9 as in the
source of Jimbo et al.,84 the hot cathode and the anticathode
were made of tantalum, which, as shown later by Inoue
et al.58 improves the volume production when deposited as a
thin film on the wall. This may explain in part the good performance of these sources.
An intense volume H ion source based on a Penningtype discharge was developed at Oak Ridge National
Laboratory: the Volume Ionization with Transverse
Extraction (VITEX).86 The extracted current was 25 mA,
with a current density of 125 mA/cm2. The Penning discharge H ion source has been operated with caesium in
Novosibirsk leading to the discovery of the plasma-surface
production of these ions by Belchenko, Dimov, and
Dudnikov87 and the development of the SPS. The SPS operation will be discussed in Sec. V. The operation of these sources without caesium, in pure hydrogen, i.e., in the volume
production regime, in a version of SPS called planatron, was
described by Belchenko and Kupriyanov.88 In an attempt to
replace caesium with a low work function and low vapor
pressure material, Leung et al.89 operated a Penning source
with LaB6 cathodes in a caesium-free discharge. They found
that the extracted H current density was comparable to that
(1) Excitation of molecules in collisions with electrons to a

ro-vibrationally excited state, by the e-V process (Eq.
(2)) or the E-V process (Eqs. (3) (4)).
(2) Dissociative electron attachment to form the negative
ion or the DA process, Eq. (1). Since the negative ion
destruction rate in collision with electrons increases rapidly with the increasing electron temperature, it follows
that low electron temperature plasmas are needed to produce high negative ion densities. A direct consequence
of this is the concept of tandem or magnetically filtered
negative ion source. The magnetic filter provides a limited region with the field lines of force traversing perpendicularly the axis of the beam extraction, which divides
the ion source volume into two regions (this is the reason
for this source to be denoted as tandem source):
(a)
(b)
The driver or source region containing either

the filaments in arc discharge sources or the RF
antenna in RF sources.
The target or extraction region, limited by an
electrode containing the extraction openings, which
is denoted as a plasma electrode (PE) or a plasma
grid.
1. Role of the magnetic filter
The magnetic filter field should be strong enough to prevent the energetic primary electrons in the driver region
from being transported into the extraction region. However,
positive and negative ions of larger Lamor radii and cold
electrons reflected at the source wall can move across the
magnetic filter. Due to the confinement by the magnetic filter
field, a relatively large concentration of fast electrons builds
up in the driver region, while a low temperature plasma fills
the extraction region. Vibrationally excited molecules can be
formed in the driver region, essentially by the E-V process
(Eqs. (3) (4)). They can be converted to negative ions by
DA in collision with slow electrons in both regions, but the
chance of the negative ions to survive is larger in the extraction region.
The first experiments with the negative ion tandem source
were performed in the early eighties at the Lawrence Berkeley
National Laboratory90 and at Ecole Polytechnique.91 The magnetic filter was of internal rod-type, i.e., was constructed by
inserting square permanent magnet rods into copper tubes
through which a square hole for the magnet was broached.
This magnetic filter was inserted into a cylindrical multicusp
ion source geometry, formed by ten columns of samariumcobalt magnets, forming continuous line cusps parallel to the
source axis. These magnets formed closed magnetic line cusp
field with four extra rows of magnets attached at the end plate.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-12
A steady-state plasma was produced by primary electrons

emitted by four tungsten filaments located in the central, magnetic field-free region, and the chamber wall served as the anode. The rod filter divides the ion source volume to two
regions, the target and extraction regions, which are often
called target chamber and extraction chamber, respectively.
Sources of other configuration did not contain the rod
filter. An example of this source configuration was
CAMEMBERT II of 25.4-cm diameter and 23.6-cm depth.
In the following decades, CAMEMBERT II and a larger
source (44-cm in diameter 45-cm depth) named
CAMEMBERT III were used as plasma generators to study
the low pressure hydrogen plasma by laser photodetachment,
Langmuir probes, and ion extraction. The primary electrons
were provided by electron emission from tungsten filaments9294 (Figure 21(a)) or by inserting small microwave
applicators that excite plasma with the permanent magnetic
field corresponding to Electron Cyclotron Resonance (ECR:
Figure 21(b)).95,96 These applicator modules will be
described in Sec. III B 2. The multicusp ion source geometry
also evolved from the tandem configuration91 toward the
hybrid configuration, in which the primary electrons are
confined in the multicusp magnetic field.93
Appl. Phys. Rev. 2, 021305 (2015)
2. Role of the plasma electrode
Leung and Bacal91 insulated the electrode having the

extraction hole, called plasma electrode, from walls of other
sides of the ion source and applied separate bias, Vb, with
respect to other walls serving as the anode for discharge of
CAMEMBERT II. They showed that positive bias voltage of
the proper range to the plasma electrode under the presence
of the magnetic filter enhanced the extracted negative ion
current and reduced the co-extracted electron current in accordance with the decreased electron density in the target
chamber. The study of the negative ion density by photodetachment91 showed that the plasma electrode bias could
maximize the relative negative ion density (i.e., the ratio
N/Ne) in the target chamber, with a maximum value
N/Ne 0.13 measured with a discharge of 50 V 1 A (see
Figure 22). This is in contrast with the relative negative ion
density in the driver chamber which exhibited only a modest
variation around the value of N/Ne 1.15 102. The
observed reduction of the plasma potential gradient across
the filter field when the plasma electrode is positively biased
allows the flow of the negative ions from the driver into the
target chamber, which causes the increase of the negative
ion density in this chamber. An important consequence of
FIG. 21. Design of the source CAMEMBERT III equipped with: (a) 16 filaments or (b) seven ECR modules. The design of an ECR module is also shown in
(b). Fig. 21(a) was reproduced with permission from Rev. Sci. Instrum. 64, 3205 (1993). Copyright 1993 AIP Publishing LLC. Fig. 21(b) was reproduced with
permission from Phys. Plasmas 20, 101601 (2013). Copyright 2013 AIP Publishing LLC.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-13
FIG. 22. H ion relative density (n-/ne) in the target (extraction) chamber of
a tandem source versus the plasma electrode bias voltage. Reproduced with
permission from Rev. Sci. Instrum. 55(3), 338 (1984). Copyright 1984 AIP
Publishing LLC.
Appl. Phys. Rev. 2, 021305 (2015)
magnetic field line of force, while the extraction region is

formed in the central part of the plasma container.
Replacing one of the end walls by the PE equipped with
an extraction electrode, the effects due to a weak magnetic
field at the PE surface upon the extracted negative ion and
electron currents were studied.98 This magnetic field is the
stray magnetic field produced by the magnets enclosed in the
extractor and can be idealized as parallel to the PE. The field
is denoted as the PE magnetic field hereafter in this section.
The PE magnetic field magnetically insulates the PE from
the plasma. In Figure 24, the dependence of the extracted
negative ion and electron currents on the PE bias Vb is shown
for three values of the intensity of the PE magnetic field. The
negative ion current I exhibits the most pronounced maximum and attains the highest value, while the electron current
Ie exhibits the most considerable drop when the PE magnetic
field is the strongest. The fraction of the electrons in negatively charged particles present in the plasma is reduced by
the PE magnetic field as larger field intensity restricts direct
the increase of the plasma potential in the target chamber

with positive Vb is the situation that the negative ions can
reach the plasma electrode and the extractor. This feature
was observed in other volume sources, e.g., in the microwave driven source described by Svarnas et al.97 Figure 23
(from Ref. 97) shows a typical dependence upon Vb of the
plasma electrode current and those of the extracted negativeion and electron currents.
An early example of tandem source concept for a hybrid
configuration was described in 1980.76 This source contained
a virtual magnetic filter located next to the plasma electrode
which became widely used in the following years. The multicusp structure of this ion source was formed by ten external
magnet columns aligned parallel to the source axis. Ten filaments were located inside the source, in the multicusp magnetic field of the external magnets. The multicusp magnetic
field acted as a virtual magnetic filter for the electrons emitted by the filaments. The driver plasma region is located near
the wall confining electrons from the filament along the
FIG. 23. Typical evolution of the extracted electron and negative-ion currents as a function of the plasma electrode bias. The plasma electrode current is also shown. Reproduced with permission from Svarnas et al., IEEE
Trans. Plasma Sci. 35, 11561162 (2007). Copyright 2007 IEEE.
FIG. 24. Dependence of the extracted negative ion (a) and electron (b) currents upon the PE bias, with different stray magnetic fields in front of the
PE: open circlehigh B, full circle medium B, and crosslow B. 50 V,
10 A, 2.5 mTorr discharge. Extraction voltage: 1 kV. Reproduced with permission from Rev. Sci. Instrum. 59, 2152 (1988). Copyright 1988 AIP
Publishing LLC.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-14
electron transport from the plasma to reach the PE.

Therefore, the extracted electron current is reduced when the
PE magnetic field is enlarged.
When biased positively with respect to the walls, the PE
collects a large electron current, as can be seen in Fig. 23.
Since the region close to the PE cannot be easily reached by
electrons from the main plasma, the electron population in
this region is strongly depleted when the applied PE bias
potential allows the local electrons to flow up to the PE
traversing the magnetic field. It was shown in Ref. 98 that
the negative ion density increased in the region where the
electron density decreased. This suggests that the negative
ions penetrate from the main plasma into the magnetized
region close to the PE, where the plasma neutrality requires
negatively charged species capable of passing across the
magnetic field to compensate depleted local electron density.
The negative ions are unaffected by the weak magnetic field.
This explains the enhancement of the extracted negative ion
current as the coextracted electron current goes down.
Figure 25 shows the comparison of the dependence
upon Vb of the positive ion and that of negative ion extracted
currents. It can be seen in Fig. 25 that the extracted positive
ion current decreases when the positive bias potential of the
PE is increased, while the negative ion current rises to a
maximum value, which is close to the value of the positive
ion current, and then both currents decrease, conserving very
close values. This occurs because the plasma neutrality in
this region requires equal positive and negative ion densities
with ne being very low.
Figure 26 shows the measured variation of the electron
density ne and negative ion density n at 8 mm from the PE
against Vb in a 50 V, 50 A discharge at 3 mTorr (Ref. 99) in
the larger source Camembert III. A very high ratio
n/ne 12 occurs in these conditions at Vb 2.5 V, as shown
FIG. 25. Dependence upon the PE bias of the extracted currents (positive
ion I, negative ions I, and electron Ie) and of the current to the PE, Ib. The
range of the plasma potential variation is indicated on the top of the figure.
These results are measured for a discharge of 50 V, 5 A, 2 mTorr.
Reproduced with permission from Rev. Sci. Instrum. 59, 2152 (1988).
Copyright 1988 AIP Publishing LLC.
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 26. Variation of the negative ion density n- and electron density ne versus plasma electrode bias Vb at 8 mm from the PE. The measurement is
made at 8 mm from the PE where a weak transverse magnetic field is present. The discharge is operated at 50 V, 50 A, 3 mTorr. Reproduced with permission from Rev. Sci. Instrum. 67, 2221 (1996). Copyright 1996 AIP
Publishing LLC.
in Fig. 27. A plasma with such high n/ne ratio contains a

very low electron fraction and the negative ion density is
only slightly lower than the positive ion density, namely,
n (12/13) n. Therefore, such a plasma is designated as
ion-ion plasma. Hatayama et al.100 reported simulation
results for the negative ion and vibrationally excited molecule production in the source Camembert III, using twodimensional Monte Carlo simulation codes. Sakurabayashi
et al.101 studied the H ion trajectories of both volume and
surface produced H ions, showing that the wall loss is significant at low pressure (1 mTorr). The study by Kuppel
et al.102 focused on the combined effects of the weak magnetic field and the PE bias voltage upon the extraction process of H ions and electrons based on numerical
simulations. Their results indicate that applying the magnetic
FIG. 27. Variation with plasma electrode bias Vb of the ratio n/ne measured
in Camembert III at 8 mm from the PE. Same discharge operation conditions
as in Fig. 26. Reproduced with permission from Rev. Sci. Instrum. 67, 2221
(1996). Copyright 1996 AIP Publishing LLC.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-15
Appl. Phys. Rev. 2, 021305 (2015)
field (without the PE bias) enhances H ion extraction, while

it drastically reduces the extracted electron current. Second,
the extracted H ion current has a maximum when the PE
bias is equal to the plasma potential, while the extracted
electron current is significantly reduced by applying the PE
bias. These are in agreement with the experimental results.
The various consequences of biasing the plasma
electrode in multicusp H ion sources were summarized in
Ref. 103. The PE bias controls the current collected by the
PE, the extracted negative ion and electron currents, and the
potential drop across the magnetic filter. Moreover, the positive bias of the PE depletes the electron population and
concentrates the H ions near the surface forming an ion-ion
plasma. The plasma parameters in front of the PE also determine the local H production rate coefficient from vibrationally excited hydrogen molecules which are shown to be
produced from recycling of hydrogen atoms at the PE surface by Hall et al.,48 Schermann et al.,49 and Young et al.50
C. Volume negative ion sources for fusion
applications
Early in the development of negative ion sources

for fusion applications, a conventional rod filter was
used.78,79 Soon after the experiments in LBNL and Ecole
Polytechnique with the rod filter, high current negative ion
beams were produced using a similar magnetic filter geometry by Okumura et al.,79 who demonstrated in 1986 that a
high current H ion beam of more than 1 A can be extracted
from a volume ion source. Beams of H ions formed by volume processes in a magnetically filtered multicusp plasma
generator were extracted for 0.2 s through a multi-aperture
extractor with a total extraction area of 133 cm2. The discharge was created by electron emission from eight tungsten
filaments. The magnetic filter (an array of five water cooled
copper tubes 8 mm in diameter, in which small SmCo magnets are inserted) was located at 3 cm above the plasma grid.
Unlike in LBNL, the source developed in 1985 at the
Culham Laboratory (Oxfordshire, UK)104 employed a rectangular chamber (20 24 cm2 in area and 20 cm deep)
surrounded by several rows of SmCo magnets on each sidewall which form continuous line cusps perpendicular to the
beam axis. There are also several rows of SmCo magnets on
the back end plate. The energetic electrons trapped in the
multicusp magnetic field drift between the continuous line
cusps to make ionization, excitation, and dissociation collisions. In this source, a virtual dipole magnetic filter was created using magnets attached on the outside of the plasma
source. The bias potential of the plasma electrode has a
major effect on reducing the extraction of electrons. The
existence of an electron trap in the second electrode of the
extractor separates the H and electron beams and these currents are measured independently. It was shown that these
beams have different parametric dependences on the bias
potential of the plasma electrode (shown in Figure 28),
which allows to reduce the extracted electron current by
applying a suitable positive PE bias. In 1986, a large negative ion source (55 30 20 cm3) with a tent filter field in
operation at Culham was described by McAdams et al.105
FIG. 28. Variation of extracted H and electron currents with plasma electrode bias potential in the Culham source. The current I3 is the current which
breaks through the electron suppression electrode. Reproduced with permission from Rev. Sci. Instrum. 56, 1697 (1985). Copyright 1985 AIP
Publishing LLC.
The explanation of the terms dipole and tent filter is

given in Ref. 106. The maximum extracted H current
through an aperture in the plasma electrode 24 mm in diameter was 145 mA at a beam energy of 83 keV, while the ratio
of the electron current measured at the second grid to the
extracted H current was approximately four. Two designs
of electron suppressors were tested in a volume negative ion
source operating dc with an extraction aperture diameter of
16 mm.106
The Culham source inspired the design of sources built
in other laboratories and used in various applications. Such a
source was built and studied at FOM Institute for Atomic
and Molecular Physics in Amsterdam (The Netherlands).107
The H ion density was measured by laser photodetachment
versus the hydrogen pressure, at various values of the discharge current (Fig. 29). The fractional H density n/ne
varied between 5% and 40%. One observes an initial
increase of the fraction with increasing pressure followed by
a saturation or a decrease. For larger values of the discharge
current, the maximum in the n curve shifts to higher pressures. This source was acquired from FOM by the Dublin
City University and operated as the DENISE (Deuterium
Negative Ion Source Experiment) ion source.108
The pressure dependence of the H density and
extracted H ion current is an important characteristic of
volume sources. As shown by Eenshuistra et al.,107 the optimum pressure for the extracted H ion current depends on
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-16
FIG. 29. H ion density in a source with a magnetic filter as a function of

source pressure for various discharge currents. Discharge voltage of 100 V.
Reproduced with permission from J. Appl. Phys. 67, 85 (1990). Copyright
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 31. H ion density dependence upon the gas pressure in the magnetic
multipole source operated with a filament discharge measured by Mosbach.
The pressure dependence of the population of vibrationally excited molecules in the vibrationally excited state v00 10 is also shown. Open symbols
are calculated values, and solid symbols are measured values. Reproduced
with permission from Rev. Sci. Instrum. 77, 03A532 (2006). Copyright 2006
AIP Publishing LLC.
the discharge current. The optimum pressure also depends

on the ion source: it is 4 to 5 mTorr in the microwave driven
magnetic multipole Camembert III109 (Fig. 30), while it is 10
mTorr in the magnetic multipole operated with a filament
discharge by Mosbach (Fig. 31).110
The study by Paunska et al.111 on the radial profile of
the negative ion density in cylindrical RF plasma driven
chambers with different diameters leads to the conclusion
that a small radius driver provides high negative-hydrogen
ion density with a peak at the chamber center. This traced
out a concept for the design of an effective volumeproduction based source of negative hydrogen ions as a matrix of small radius discharges, 23 cm in diameter, with an
extraction hole in front of each small radius discharge. This
concept was developed in other works from the University of
Sofia (Bulgaria).112117 This concept was earlier predicted by
Hiskes et al.118
Several research efforts were made to study the effect
on the H ion density by mixing argon and other inert gases
to hydrogen.119122 Work on the performance improvement
of large RF driven hydrogen negative ion sources for fusion
integrated the idea of mixing argon or other inert gases to
hydrogen and was initiated at IPP and described in 1999 by

Speth et al.123 In this source, the RF antenna was located
near the back plate. It was found that the negative ion yield
increased with pressure but saturated with rf power. With the
addition of 5% to 15% of argon, the H yield increased up to
8.5 mA/cm2 (Figure 32). In the source described by Vollmer
et al.124,125 in 2000, the plasma is generated in a source with
a small RF driver (alumina cylinder of 24.5 cm diameter)
supplying an expansion volume of 30 58 19 cm3 (see
Figure 33). The rf coil is wound around the alumina cylinder.
Two different filter field configurations were tested. Shown
in Fig. 33 is the configuration in which the filter magnets
were installed in a flange connecting the expansion chamber
and the plasma grid. The RF coil was driven by a 1 MHz
push-pull oscillator that delivered up to 170 kW. The source
FIG. 30. Electron and H ion density dependence on the gas pressure in the
microwave driven source Camembert III. Reproduced with permission from
Rev. Sci. Instrum. 77, 03A352 (2006). Copyright 2006 AIP Publishing LLC.
FIG. 32. H yield vs RF power in pure hydrogen and in hydrogen with addition of argon. Reproduced with permission from Speth et al., Fusion Eng.
Des. 46, 383 (1999). Copyright 1999 Elsevier.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-17
FIG. 33. IPP RF source with extraction system. Reproduced with permission
from Vollmer et al., Fusion Eng. Des. 5657, 465 (2001). Copyright 2001
Elsevier.
operated reliably up to 150 kW RF power and at source pressure as low as 0.2 Pa. For negative ion extraction, the plasma
source was combined with a three grid extraction system,
developed by CEA Cadarache (an uncooled molybdenum
plasma grid and two cooled copper grids). The 44 extraction
holes (14 mm diameter) were arranged in an area of 13 cm
14 cm. The total net extraction area of the grid system is
67 cm2. For pure volume production with about 20% admixture of argon, up to 9 mA/cm2 H current density at 0.65 Pa
was measured.
The source described by Vollmer is still used at the
BAvarian Test MAchine for Negative ions (BATMAN) testbed and operates reliably with the input power as high as
70 kW from the 100 kW oscillator.126 In 2002, Kraus
et al.127,128 described the development at IPP of a large rf
source for negative ion production. To protect the ceramic insulator of the driver from thermal stress, an internal copper
Faraday screen has been added. This prolonged the pulse duration of the source operation up to 15 s. Because there is no
filament material covering the inner surface of the RF source,
the use of different plasma grid materials is possible.
Therefore, Kraus et al.127,128 tested eight different plasma
grid materials as potential options for caesium-free operation,
as well as the effect of the addition of several noble gases, of
which the largest improvement of the plasma density was
found with xenon. With a fast scanning Langmuir probe, the
profile of plasma density was measured from the back plate to
0.8 cm from the plasma grid in one pulse. Recently, Kraus126
pointed out that the reported Argon effect was not caused
by a higher production of negative ions, but by a higher neutralization, which increased the H ion current densities
measured using a calorimeter, while the admixture of rare
gases did not affect significantly the extracted ion current.
An overview of the RF source development program at
IPP Garching aiming at demonstration for ITER-relevant
source parameters was given in 2006 by Speth et al.129 This
implies a current density of 200 A m2 accelerated D ions
at a source filling pressure of 0.3 Pa and an electron-to-ion
Appl. Phys. Rev. 2, 021305 (2015)
ratio of 1 from an extraction area similar to positive-ion

based sources at JET for pulse length of up to 1 h.
Experiments in caesium-free operation showed that the H
ion current density was by a factor of 510 lower than with
caesium. The electron-to-ion ratio was above 10 in the low
pressure range. The optimization of negative ion formation
in source hydrogen plasma requires knowledge of the plasma
parameters. Fantz et al.130 introduced the optical emission
spectroscopy methods for the IPP source as a diagnostic tool
to obtain line of sight averaged plasma parameters.
In 2011, Franzen et al.131 analyzed the effect due to the
magnetic filter field upon the performance of the RF driven
IPP prototype source for negative hydrogen ions. The magnetic filter field in front of the extractor plays an important
role for the transport of the negative hydrogen ions to the
extractor and the suppression of the co-extracted electrons. It
is worth noting that the optimum location of the magnetic
filter is not at the center of the source, as initially proposed
by Leung et al.,90 but next to the plasma electrode, as proposed by Bacal et al.76 (see Sec. III B). For the large ITER
source, the filter field will be generated by a current of up to
4 kA flowing through the first grid of the extractor.132,133
However, the optimization of the magnetic filter field was
done using a frame that surrounds the source and attaches
permanent magnets.
Schiesko et al.133 investigated the homogeneity of the
plasma parameters in front of the dummy grid of the source on
the RADI (RADial Injector) test-bed and reported the result in
2012. The source of RADI is a scaling of the IPP prototype
source with four drivers in a 2 2 arrangement. In addition to
its increased size, the main difference between RADI and the
other test-beds devoted to the optimization of negative ion
source for ITER neutral beam heating is the way the magnetic
field is generated near the grid. At RADI, a current flowing
through the grid generates the filter magnetic field like in the
ITER source, while permanent magnets are used on the other
test beds equipped with small prototype sources. In order to
avoid the overlapping of magnetic field and caesium influences on the homogeneity, no caesium was evaporated in RADI.
The result of the study shows that the grid of the large source
is irradiated by the plasma more homogeneously than smaller
test-bed sources for various plasma parameters. This study
also showed that an electron temperature of 1 eV could be easily achieved in the extraction region, which is important to
minimize the destruction of the negative ions by electron
collisions.
Other problems associated with high power operation of
negative ion sources were discussed in the publications of
the IPP scientists, such as the effect of back-streaming positive ions,133 the magnetic field dependence of the plasma
properties,134 optical emission spectroscopy at the large RF
driven test facility ELISE (Extraction from a Large Ion
Source Experiment),135 and results of the ITER-relevant negative ion beam test facility ELISE.136,137 ELISE is equipped
with a half-size ITER RF source and is intended to demonstrate large-scale extraction at ITER parameters in terms of
negative ion current densities, co-extracted electrons, and
beam uniformity. The plasma is generated in four cylindrical
drivers (instead of eight as for the ITER source) with a
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-18
diameter of 300 mm each mounted on the back plate of the

main chamber. The drivers are built-in a so-called dome
such that the RF coils can be operated in vacuum.
The RF power is coupled inductively into the plasma via
a six-turn copper coil wound around an Al2O3 insulator. An
actively cooled copper Faraday screen protects the insulator
from plasma sputtering. The plasma expands from the four
drivers into the main chamber. A magnetic filter field with a
strength of less than 10 mT is used (i) to cool the electron
temperature from around 10 eV (driver) down to about 1 eV
(close to extraction region) and thus minimize the H
destruction by electron collisions and enhance the negative
ion production by dissociative attachment, (ii) to reduce the
amount of coextracted electrons, and (iii) to enhance the
extraction probability for the surface produced negative ions.
In contrast to small sources, filter field is created by running
a current through the plasma grid in the line of design extension from the IPP prototype source produced with permanent
magnets mounted near the plasma grid.
Although the source ELISE is intended to operate
with caesium seeding, the cited reports132,136 refer to
non-caesiated operation. The investigation on the electron
extraction for ELISE was conducted also in non-caesiated
condition.138 It reports that the amount of co-extracted
electrons could be suppressed sufficiently with a moderate
magnetic filter field and by plasma grid bias. The coextracted electrons have to be filtered out of the beam before
being accelerated to high energy; this is done by deflecting
the electrons at the extraction gap with the magnetic field
created by magnets embedded inside the extraction grid.
Several countermeasures have to be undertaken to limit the
co-extracted electron current and its power and power
density: (1) a sufficiently low extraction voltage of 10 kV at
maximum, (2) biasing the plasma electrode positively with
respect to the source walls by a few tens of volts, and (3) a
magnetic filter field of the order of some mT in front of the
plasma electrode with the integrated magnetic field distance
of about 1 mT m through the source to realize reductions of
the electron temperature and density.
D. Volume negative ion sources for accelerator
applications
In the late 1980s, LBNL developed a small filamentdriven volume source,139 with a 75 mm inner diameter using
Appl. Phys. Rev. 2, 021305 (2015)
all permanent magnets as shown in Fig. 34(a). 14 rows of

water-cooled SmCo5 magnets provided radial confinement,
and another 4 rows in the back flange provided rear confinement. The filter field was generated by a pair of water-cooled
permanent magnets mounted on the outlet flange. Placing a
pair of SmCo5 magnets 40 mm apart produced the highest
H current when the plasma electrode was biased to about
1 V with respect to the chamber wall. From this compact
source, an H current density greater than 250 mA/cm2 has
been extracted.139
In 1991, Leung et al.140 reported the development of an
RF-driven H ion source at LBNL for use in the superconducting super collider. In a later publication, Leung141 presented the advantages of the RF-driven source over the dc
filament discharge source:
(1) there are no short life components in the source,
(2) a clean plasma free from contaminants from the cathode
material can be maintained,
(3) the RF power supplies can be operated at ground potential, and
(4) the RF source can be operated in either cw or pulsed
mode.
With these reasons, the rf-driven multicusp source is
considered attractive as the H ion source for particle accelerator application.
A schematic diagram of the LBL RF ion source140 is
shown in Figure 34(b). The source chamber is a copper cylinder (10 cm diameter 10 cm long) surrounded by 20 columns of samarium-cobalt magnets that form a longitudinal
line cusp configuration. The RF internal antenna is fabricated
from 4.7 mm diameter copper tubing and is coated with a
thin layer of hard porcelain material. An H current of
40 mA can be obtained from a 5.6 mm aperture with the
source operated at a pressure of about 12 mTorr and with
50 kW of RF power. In order to reduce the co-extracted electron current, the filter magnetic field was coupled to an
11.6 mm diameter 9 mm long collar to make the extraction
region. The ratio of electrons to H ions in the extracted
beam varied from 8 to 12 as the RF power changed from
20 to 50 kW. This RF source was the starting point for the
development of RF sources with internal antennae, e.g., at
Oak Ridge National Laboratory.142,143
For the DESY-HERA source, an almost maintenance
free system with 0.1% duty factor and 40 mA H current was
FIG. 34. Schematic of the small (a) filament driven and (b) RF driven volume H source developed at LBNL.
Figure (a) was reproduced with permission from Rev. Sci. Instrum. 62,
100 (1991). Copyright 1991 AIP
Publishing LLC. Figure (b) was reproduced with permission from Rev. Sci.
Instrum. 60, 531 (1989). Copyright
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-19
demanded. A copy of the LBNL H RF source manufactured

by AccSys was acquired in 1994 but the antenna lifetime
appeared to be unacceptable.144 Despite testing other configurations, DESY was unable to produce robust antennas, which
would yield adequate H beam current with an adequate lifetime.60,144 Finally, in the late 1990s, DESY decided to develop an external antenna H source where the antenna is
wound around an L-shaped Al2O3 ceramic, which forms the
wall of the plasma chamber (Figure 35). The resulting source
was very successful, producing a 40 mA H ion beam for
0.15 ms at 6 Hz for over 100 weeks.145 This was the start for
new developments of RF sources with external antenna, e.g.,
at Oak Ridge National Laboratory and CERN.
In 2008, Peters reviewed the development of multicusp
volume H ion sources for high energy accelerators, providing a table with source data and a graph illustrating the history of H volume sources.145 Another review of negative
hydrogen ion sources for accelerators by Moehs et al.146
summarized in 2005 work on VPS (i.e., caesium-free sources) effected to this date.
As part of the superconducting proton linac study at
CERN a new H plasma generator147150 driven by an external 2 MHz RF antenna has been developed from the RF powered, uncaesiated DESY-Linac4 H source.60,144 More
recently, improvements of VPS H ion sources were proposed in many laboratories, of which we will mention only a
few. At TRIUMF, a 15 mA dc H multicusp source has been
developed for injection into a cyclotron.59 This source was
also used with a 900 kV tandem accelerator to obtain 10 mA
protons at 1.8 MV. A major effort was dedicated to the search
for optimal filament materials, shape, and location. The chosen option was to use four half-ring tantalum filaments far
from the extraction region. An elaborate arrangement of virtual filter magnet bars was made. The higher vacuum
improves the beam intensity which has been lost in the stripping process. More than 15 mA cw dc H currents at 5 kW of
arc power with low e/H ratio were reported.
The caesium-free H ion sources developed at Japan
Proton Accelerator Research Complex (J-PARC)
Appl. Phys. Rev. 2, 021305 (2015)
demonstrated an H ion current of 38 mA.151,152 This was

achieved by an optimal shape and a high temperature of the
plasma electrode and use of a lanthanum hexaboride filament.152 The H ion current could not be increased by seeding caesium when the lanthanum hexaboride filament was
used. However, it was increased to more than 70 mA if Cs
was added into an arc discharge sustained by a tungsten filament cathode.
At Seoul National University, the enhancement of H
ion surface production was attempted by introducing a negatively biased additional electrode.153,154 The RF wavefrequency dependence of H ion production and extraction
in a DESY type ion source was also studied at this
University.155,156 The plasma density and temperature determined using a Langmuir probe are correlated to the extracted
H beam current at different frequencies.155
In 2002, a new concept of ECR plasma production that
may be suitable for an accelerator size source has been proposed by Lacoste et al.156 The design of this source, designated as dipolar plasma source, is shown in the inset in
Figure 21(b). It consists of two main parts: (i) a cylindrical
drilled permanent magnet with an azimuthal symmetry
around its magnetization axis and (ii) a microwave
(2.45 GHz) applicator constituted by a coaxial line in the
direction of the magnetization vector, which ends at the rear
pole of the magnet. The microwave power can thus be transmitted through the plasma and be absorbed near the ECR
region. At the microwave frequency of 2.45 GHz, the magnetic field required for the ECR condition is 875 1024 T. A
2D network of seven units of dipolar sources was inserted in
the large plasma chamber Camembert III in which negative
hydrogen ion production was studied by photodetachment
and extraction.95,157 The effect of the PE bias on the plasma
parameters in the center of the chamber Camembert III was
reported in Ref. 97, while those near the PE was described
by Svarnas et al.158
An overview of the fundamental results obtained with:
(i) a single dipolar source and (ii) a network of seven dipolar
plasma sources inserted into the magnetic multipole
FIG. 35. (a) Schematic and (b) detailed

outlet collar of the external antenna
H source developed at DESY. Fig.
35(a) was reproduced with permission
from Rev. Sci. Instrum. 71, 1069
(2000).
Copyright
2000
AIP
Publishing LLC. Fig. 35(b) was reproduced with permission from Rev. Sci.
Instrum. 73, 900 (2002). Copyright
2002 AIP Publishing LLC.264
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-20
Camembert III was presented by Bechu et al.159 Aleiferis

et al.160 described the magnetic field of the network of five
dipolar sources forming the source Prometheus and present
the photo of the produced plasma.
IV. SURFACE PRODUCTION MECHANISM
A. Early observations of surface produced H2 ions
Emission of H ions from solid surface was observed

for metal samples under exposure of positive lithium ion
beams as early as 1931 by Woodcock.161 Alvarez proposed
to utilize H ions produced by passage of positive hydrogen
ion beam through a thin foil for tandem acceleration with a
Van de Graaff apparatus.8 Foils of 10 lg/cm2 thickness
behave as solid materials against H ions leaving the surface, and the resulting charge state fractions are different
from those for thick gas target.162 The negative ionization
fraction was found determined by the material covering the
surface of the beam exit side.163 Production of polarized H
ions for tandem acceleration was also shown to be possible
by charge exchange through a thin carbon foil.164
These early observations indicated that fractions of H
ions against protons increased with decreasing incident proton energy. Verbeek et al. have shown that the energy distribution of H ions produced from 15.3 keV incident proton
backscattered at Au contained a larger amount of lower
energy component than the distribution of protons produced
in the backscattering process.165 They also observed the ratio
of H ions to protons produced by back scattering decreased
by changing the surface to Ta and increased by changing it
to ThO2. Difference in work function, u, of the surfaces of
Ta (u 5.6 eV), Au (u 4.8 eV), and ThO2 (u 2.0 eV)
was considered to have some correlation to the H ions to
protons fractions.
B. Theoretical work on H2 surface production
Many theoretical works and experimental attempts to

explain fundamental characteristics of H surface production
have been started after Belchenko et al., reported that high
density beam of H ions was extracted from their surfaceplasma source.87 Contrary to the two-step volume production
process, negative surface ionization of hydrogen is a single
Appl. Phys. Rev. 2, 021305 (2015)
step process; it is the direct electron transfer from solid material to a hydrogen atom leaving the surface. A simplified
one-dimensional potential diagram of the system can be
schematically illustrated in Fig. 36. The electron affinity of
hydrogen is only 0.754 eV, while surface work functions of
refractory metals, which can be immersed in dense plasma,
exceed 4 eV. Electron transfer from the metal to the hydrogen atom is not an efficient process under this condition.
When the work function of the metal surface is reduced by
alkali or alkaline earth metal adsorption and a hydrogen
atom approaches to the surface, the hydrogen affinity level
goes below the work function at a small distance from the
metal surface. The hydrogen atom captures an electron from
the metal to the affinity level and can escape from the surface
as an H ion. Thus, reduction of work function by alkali or
alkaline earth metal adsorption was considered responsible
for the enhancement of H current from a Cs loaded surfaceplasma source. By adopting the potential geometry in
Fig. 36, Kishinevskiy has shown that the probability for a
secondary H ion to leave the surface region without losing
the electron in the affinity level depends upon two important
parameters: the surface work function and H outgoing velocity component normal to the surface.166
A theoretical model to calculate destruction probability
of a negative ion near metal surface was formulated by Janev
by introducing parabolic coordinates.167 Hiskes also used
parabolic coordinates for solving complex eigen value problem to compute how the H affinity level shifted and broadened.168 He also pointed out that not only the work function
but also the potential structure near the surface influence the
survival of H ions. Rasser et al. have also computed the
level shift and broadening of the H electron affinity for
bare tungsten (W) substrate, thick Cs case and W covered by
a partial monolayer of Cs. The fractions of H ions against
the sum of neutral hydrogen and H ions were calculated for
the three cases as functions of the H velocity component
perpendicular to the surface.169 They also had suggested that
a simple formula can be used to estimate the negative hydrogen ion yield in the low exit energy condition, which was
earlier proposed by Blandin et al. for more general electron
transfer processes at the surface.170 Yields of H ions for
conditions identical to Rasser et al. but with a simple surface
geometry were calculated by Cui.171
FIG. 36. Electronic potential diagram

around the surface with a hydrogen
atom leaving the surface. In the figure,
symbols designate the following quantities: u, work function; V0, lowest
level of conduction band. Other parameters in the figure signify: EA, electron
affinity of H far from the surface: DE,
level shift of affinity (image potential),
and C, level broadening of affinity.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-21
In early theoretical models, results were tabulated as

functions of the hydrogen velocity component normal to the
surface. Effects due to oblique trajectory were investigated
by Koppers et al.,172 and the probability of negative ion formation was shown to be the function of angle of H ions
leaving the surface. Formation of H ions by grazing surface
collision was studied for Al(111) plane by Zimny.173 He
showed that his theoretical model can recover the results
obtained experimentally by Wyputta et al.174 Maazouz et al.
treated this problem both theoretically and experimentally by
extending the scattering angle up to 40 with respect to the
surface plane.175 Aluminum was chosen as the material to
study the formation of H ions by scattering from a thin film
to investigate the quantization effect along the direction of
the surface normal.176 More recently, survival of H ions
near the cluster surface of Al was studied to investigate the
quantum-size effect.177 The formulation was also given for
grazing angle scattering onto a spherical Al cluster.178
Carbon is another material which has been employed to
study fundamental formation process of H ions, since
Tsumori et al. have shown that the conversion efficiency was
relatively large for positive hydrogen ion incidence.179 A
theoretical explanation was given after some time for H ion
formation by scattering at graphite surface,180 and at Si surface,181 for which the conversion efficiency was also experimentally investigated nearly at the same time when carbon
was tested.182 These theoretical models usually assume ideal
situations even though the importance of the impurity
adsorption to the actual condition of the H ion source environment was recognized.183
C. Experiments on fundamental H2 surface
production process
Experiments to confirm the work function and energy

dependence of H surface production usually employ ultrahigh vacuum system to avoid complexity caused by the impurity adsorption on the surface. The H/D ion production
by reflection of molecular hydrogen/deuterium ions was
studied experimentally by Schneider et al.184 and van
Wunnik et al.185 Results of these experiments clearly indicated the expected dependence upon work function as shown
in Fig. 37. The negative ion yield increases exponentially
against the reduction of work function for alkali metal coverage well less than a single monolayer, as predicted from theories. The experimental results also show that the same value
of the surface work function does not give the same negative
ion yield, but it is affected by the number of Cs evaporation
that is proportional to Cs coverage on the substrate metal.
The particle and energy reflection coefficient of the surface
change depending upon the target material, and the observed
results were attributed to higher reflection coefficients of Cs
against incoming hydrogen ions. Negative ion yields were
also measured as functions of incident hydrogen ion energy.
Comparisons were made with the semi-theoretical/semi-experimental model of the electron transfer during hydrogen
reflection process at the alkali metal target by Hiskes and
Schneider186 for the conditions corresponding to experiments
reported in Ref. 184. Their model gives a good fit to the
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 37. Correlation between surface work function and backscattered D

ion yield. The thickness of Cs layer on nickel increases as more Cs is evaporated to the nickel target. Reprinted with permission from Schneider et al.,
Phys. Rev. B 23, 941 (1981). Copyright 1981 American Physical Society.
experimental results as shown in Fig. 38. The theoretical

model developed by Rasser et al.169 has been compared with
experimental results in a similar manner for different incident angles of hydrogen ions.187,188
Measurement at lower hydrogen/deuterium molecular
ion energies was difficult because of the smaller negative ion
intensity signal, but the researchers made efforts to obtain
fundamental data by reducing incident hydrogen ion
FIG. 38. Experimentally measured H ion yields plotted as functions of

incident energy of hydrogen ions for different target material. The lower
case letters are the results obtained from the model developed by Hiskes and
Schneider. Reprinted with permission J. R. Hiskes and P. J. Schneider, Phys.
Rev. B 23, 949 (1981). Copyright 1981 American Physical Society.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-22
energies.189 This is because the incident hydrogen ion energy

is small in the actual surface ion source conditions. Seidl
et al. had assembled their system to measure H yield from
caesiated Mo surface under bombardment of hydrogen ions
with the energy down to less than 5 eV per atom.190 Their
results shown in Fig. 39 indicate a nearly constant H ion
yield above threshold, which suggests the possibility of large
current H ion production with low incident hydrogen ion
energies. Seidls research group had also reported that work
function and ion yield became different as the conditions for
Cs and hydrogen supply onto his Mo target were changed.
The H ion yield changes by coadsorption of hydrogen on a
Cs adsorbed surface, as Amersfoort et al. have reported.191
Before their ion reflection experiment, Seidl and his coworkers also tried to measure the H formation efficiencies
for processes involving even lower energies: backscattering
of thermal hydrogen atoms192 and desorption induced by Cs
ion sputtering.193 Production of H ions by injecting thermal
H atoms to low work function surface was later confirmed
by Lee and Seidl194 and by Okuyama and Mori.195
Yu experimentally investigated surface production of H
by ion bombardment induced hydrogen desorption, or sputtering of adsorbed hydrogen on the surface.196 He showed
that the H ion yield exhibited dependence upon work function resembling electron tunneling effect. Electron stimulated
desorption197 and photon induced desorption can also produce H ions at the hydrogen adsorbed C and Si surfaces.198
Kawano has found that alkali and alkaline earth hydrides produce H and D ions by electron stimulated desorption.199
The above results from theories and fundamental process experiments are compared with the H ion current from
the surface immersed in actual ion source condition by
Ehlers and Leung.18 They built a multicusp ion source with
the plasma source equipped with a Mo electrode, called converter, facing the extraction aperture. A small magnetic momentum analyzer measured the velocity distribution function
of H ions produced at the surface of the converter which
was biased at several hundred of volts negative with respect
to the hydrogen plasma surrounding the converter. The
measured velocity distribution of H ions transported out of
FIG. 39. Measured H ion yields for low energy hydrogen incident ions.
Reprinted with permission from J. Appl. Phys. 79, 2896 (1996). Copyright
Appl. Phys. Rev. 2, 021305 (2015)
the ion source operated with pure hydrogen clearly indicated

three components: the first one produced from back scattering of H, the second due to back scattering of H2, and the
third due to H3. Addition of Cs into the discharge enhanced
the H ion yield by a factor of 100 and narrowed the energy
spread of H ion beam. They have further investigated the
velocity distribution to show that the surface-generated H
ions contain two groups: one group has energy approximately equal to sheath potential, and the other appears to be
produced by reflection process.18 The former component was
assumed produced by desorption process at the converter
surface.
The presence of hydrogen plasma in front of the converter makes the measurement of work function difficult as it
induces dc electrical current flow to hinder small signal currents arising from minute contact potential difference.
Photoelectric method was employed to investigate the correlation between the work function and the energy resolved
H ion yield from a negatively biased Mo surface immersed
in caesiated hydrogen plasma. The H ion yield exhibited an
exponential dependence upon the work function as shown in
Fig. 40.200 Meanwhile, the energy distribution function
revealed the increase in low energy component in accordance with the work function decrease. The reason for observing a pronounced enhancement in H ion yield of the low
energy component is attributable to the narrow angular
spread of the desorption component compared to the large
spread of the reflection component, as discussed by Hiskes
et al.201 Belchenko et al. and Dimov et al. had reported
similar results of H energy distributions observed in their
surface production type H ion sources as described in
Sec. V B.
V. SURFACE PRODUCTION H2 ION SOURCES
A low work function metal surface immersed in plasma

can generate a flux of negative ions in the direction perpendicular to the surface. Thus, H ion beam can be formed
with the plasma source geometry shown in Fig. 41. In the
FIG. 40. Relative amounts of H current obtained by integrating velocity

spectrum of negative ions plotted as functions of work function for three
converter voltages. Reprinted with permission from J. Appl. Phys. 67, 6334
(1990). Copyright 1990 AIP Publishing LLC.
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-23
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 41. Surface production type H

ion source developed at LBL.
Reproduced with permission from
Rev. Sci. Instrum. 57, 831 (1986).
Copyright 1986 AIP Publishing LLC.
figure, a Mo made H production electrode or the converter

is installed inside of a magnetic multicusp plasma container
to confine a hydrogen plasma. The current density jR of
H/D ions from the surface due to reflection of hydrogen
can be calculated by
jR eRjHn RNn gn ;
(5)
where e is the unit charge, and jHn, RNn, and gn signify incident hydrogen flux, particle reflection coefficient, and negative ion yield for incoming hydrogen of species n,
respectively. The sum has to be calculated for incident species of kind n that include H, H2, H3, and atomic hydrogen H0. Ion induced desorption process also forms the
surface produced H current density jD of H/D ions
jD eRjn Yn gn :
(6)
Here, jn include all species of positive ions, like Cs. The

ion induced desorption yield or the sputtering yield Yn for
adsorbed hydrogen is affected by the surface density of
hydrogen atoms adsorbed on the surface. The negative ion
yield gn is a function of the H escape velocity and work
function, while the particle reflection coefficient RN and
hydrogen adsorption density are affected by the material
chosen as the converter. The effect of converter material
upon the produced H current extracted from the ion source
was investigated by Leung and Ehlers.202
The concave converter electrode geometrically focuses
the surface produced H ion beam, and a proper electrode
structure realizes a large current density H ion beam without drawing too much electron current from the ion source.
Constant introduction of Cs vapor into the ion source maintains the work function of the Mo converter surface low
enough to realize high negative ionization efficiency at the
converter surface. Thin metal liners cover the cold wall
surfaces to keep better Cs recycling through avoiding condensation of Cs inside of the source.
A. Surface sources for fusion applications
Plasma heating by negative ion based neutral beam

injection demands large current H ion beam. It also requires
quasi steady state operation for continuous fusion burning.

Right after the report from Novosibirsk about their success
of generating a large negative ion current out of their compact ion source, application of the similar source to fusion
plasma heating was attempted.203 The high power loading to
the cathode was recognized as the severe problem for continuous operation for fusion applications, and the ion source
design required larger surface area for high current H ion
production. Homogeneous illumination over a wide area of
the converter avoids power concentration and thus reduces
the possibility of unwanted direct arc discharge between the
converter and the surrounding wall. Ehlers and Leung had
initiated the surface production type, or so-called surface
conversion type H ion source development using a magnetic multicusp source at Berkeley.204 Based upon the design
of Ref. 18 to prove the operation principle, the original
source had a 8 cm wide and 25 cm long rectangular converter
with the cylindrically concave shape to focus the surface produced H ions at the extractor. With this shape of the converter, positive ions strike the converter surface by the
acceleration across the sheath, and the produced H ions are
accelerated back across the sheath toward the extractor.
Namely, the H ions are self-extracted by the electrical
bias induced to the converter. The converter size was
enlarged up to 15 cm 35 cm in the succeeding generation,205 which is shown in Fig. 41. The geometry of the large
source served as the base for designing beam extractiontransport system adaptive to a fusion reactor environment.206
The plasma volume does not necessarily have to be
large to irradiate the converter surface. Or, the H production surface does not have to be placed at the center of the
plasma either. Based on these concepts, the surface of a
grooved Mo converter of 5 cm 5 cm size was irradiated
with a thin dense hydrogen plasma maintained by an intense
hollow cathode discharge.207 An array of Ta vane shape converters was placed at the plasma expansion region of a
duoPIGatron ion source to form a broad H ion beam in
another configuration.208 The concept by van Bommel et al.
further separated the source plasma from the converter
surface, and positive ions were transported onto the surface
of external converter after being extracted from a remote ion
source.209 Proper source operation parameters were
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-24
discussed, as more information became available from

fundamental experiments and theory. Ion source models to
predict the optimum H ion current output were formulated
based upon fundamental process data,210 and the control of
Cs coverage was found important but difficult in an actual
ion source operation conditions.211
The extractable amount of H ion current increases by
enlarging the beam exit area of the surface H source, but the
enlarged extraction area also increases the amount of Cs to
drain out of the source. Drained Cs penetrates to the downstream accelerator and causes high-voltage holding problems.
Besides, direct Cs injection into the source increases the Cs inventory in the source, and only a part of Cs in the source is
utilized to reduce the work function of the converter. Layers
of Cs formed on other parts of the ion source wall alter hydrogen recycling in the source and decrease efficiency of the
source to produce ions. Attempts were made to reduce alkali
or alkaline-earth element leakage from and inventory in the
source. These include usage of porous tungsten material to
supply Cs from the back side of the converter212 and the
source design with Ba instead of Cs evaporated onto Mo surface as the converter material.213 Both schemes do not require
Cs evaporator, but Ba showed more stable operation in an
intense hydrogen or deuterium ion flux,214 and can form H
ions through the processes of recoil and direct reflection.215
Carbon materials are tested for their capability to realize
Cs-free surface H production in a hydrogen plasma, but
with low discharge power.216 Comparison in performance
was made for deuterium plasma,217 while the effect of surface temperature is investigated for highly oriented pyrolytic
graphite (HOPG) and boron doped diamond (BDD).218 The
process of H ion production at a carbon material surface is
now understood well enough to predict the energy distribution functions of negative ions.219
B. Surface sources for accelerator applications
Negative ionization at low work function surface has

been utilized for variety of negative ion species in the accelerator physics. An old review by Middleton had reported that
a Cs sputter ion source produced sizable amount of H ion
beam current.220 Higher current density beam of H ions
realizes storage of proton beam in a ring accelerator,221
which had triggered the rapid development of compact H
ion source for accelerator applications.222 This article covers
only a brief introduction to the operation of small size SPS
(which usually specifies compact and magnetized surface
production type arc discharge source originally developed at
Novosibirsk), because readers can refer to several good
review papers on these sources.223225
The scale in Fig. 42 shows how small an SPS is. The figure introduces an SPS of Penning discharge ion source structure,226 while the original success of high intensity H ion
beam production had been realized with the magnetron
discharge geometry.25 The magnetron SPS utilizes the selfextraction mechanism and many sources have been designed
with grooves at the cathode for geometrical focusing of H
ions.88 These sources are operated with discharge current
density often exceeding several A/cm2, and thus the H
Appl. Phys. Rev. 2, 021305 (2015)
FIG. 42. Structure of Penning type SPS. Reproduced from Rev. Sci.
Instrum. 83, 02A728 (2012). Copyright 2012 AIP Publishing LLC.
current density at the extraction aperture is also high. As a

consequence of both large angular and energy spreads
caused by self-extraction and intense space charge, the emittance of magnetron sources is large. Reduction of emittance
of the produced H ion beam is always essential for accelerator applications, and an ion source with the Penning discharge geometry of Fig. 42227 had been invented.222 In the
Penning discharge geometry, any H ions produced at the
surface of the cathode cannot be extracted directly from the
extraction aperture. Those H ions produced at the anode
surface and those created in the plasma volume by charge
exchange processes are extracted through the exit aperture
with a low effective ion temperature, and a small emittance
H ion beam can be obtained. The ISIS Penning SPS delivers
stable beam at Rutherford Appleton Laboratory,228 and the
performance of Penning SPS as well as that of magnetron
SPS were investigated as the candidate ion source for Linac4
accelerator at CERN.229
The design concept of multicusp surface conversion
source developed for fusion was implemented to accelerator
application at Los Alamos National Laboratory.230 Later,
KEK (Kou Enerugi-butsurigaku Kenkyusho: High Energy
Physics Laboratory, Tsukuba, Japan) utilized the design for
their accelerator231 and showed that the source can generate a
wide variety of negative ion species by changing the converter
material.232 Contamination of the accelerator column by Cs
due to the large consumption has to be avoided, and a Ba converter based ion source was developed for high energy accelerator application.233 Porous Mo converter to supply Cs from
the back side was also tested, and a performance better than a
solid Mo converter coupled to Cs injection was achieved.234
The surface production type ion source is still a viable option
to produce H ion beam with fair stability and intensity.
VI. VOLUME SOURCE OPERATION WITH CAESIUM
Historically, the addition of caesium into an ion source

without a converter has been known to enhance the extracted
H current by a factor ranging from 2 to 8, as well as the
additional advantages of reducing both the co-extracted electron current and the operating pressure of the ion source. In
the vast majority of the literature on the subject, the
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-25
enhancement in extracted negative ion current due to the

addition of caesium into the ion source is ascribed to direct
production of negative ions on caesiated surfaces by hydrogen/Cs ions and hydrogen atom impacts. Surface produced
negative ions can reach plasma volume as the electrical
potential in the ion source plasma is usually positive with
respect to the wall of the ion source. However, direct surface
production is not the only reason to change the performance
of the ion source seeded with Cs. Addition of Cs into the ion
source reduces the surface work function of the ion source
wall, and not only the production of H ions is enhanced but
also the secondary electron emission coefficient of the source
wall is enlarged. This should change electron energy distribution functions in negative ion sources. Ion source wall
increases the adsorption probability of the plasma particles
on the surface by Cs introduction, and the decreased concentration of atomic hydrogen by surface adsorption should
reduce H ion loss in the plasma volume.
A. Experimental observations
The enhancement in H ion beam current due to Cs

injection was studied with a very small (2.5 cm diameter
5 cm long) multicusp ion source.235 Similar effect was also
observed by putting in Ba into a hydrogen discharge.236 As
Xe injection into a pure hydrogen discharge is also known to
enhance the H ion beam current,237 part of the reason for
the H current increase due to Cs injection was attributed to
the mechanism similar to the one caused by Xe injection.
Motions of heavy Cs and Xe ions are unaffected by the filter
magnetic field and can reach the extraction region from the
driver region. Electrons can be supplied to the extraction
region from the driver region, and/or the source wall of the
extraction region. A larger electron density in the extraction
region should increase H density near the extractor due to
frequent dissociative electron attachment. Enhanced secondary electron emission coefficient with the larger electron
reflection coefficient of the Cs coated plasma electrode
surface should also increase the local electron density in the
extraction region.238 However, the ratio of negative ion density to electron density in the extraction region increased
from 40% to 80% by the addition of Cs into a small ion
source.239 This observation indicates that there is little contribution from electron attachment to the increase on H ions
when Cs is added; the mechanism to enhance H ion current
by Cs addition should differ from that by Xe addition.
Enhancement in H ion beam current by Cs introduction
was more pronounced than that by Xe introduction. Two
characteristics of Cs: the lower ionization potential (3.9 eV
compared to 12.1 eV of Xe) leading to increased electron
density in the extraction region for dissociative electron
attachment and the smaller electro-negativity leading to surface negative ionization at the lower work function surface
of plasma electrode were considered responsible for the
larger enhancement due to Cs than that due to Xe. The latter
characteristic was proved to surpass the former as the plasma
potential of a small multicusp H source was measured by
operating the source seeded with Ba.240 When the plasma
potential was made negative with respect to the chamber
Appl. Phys. Rev. 2, 021305 (2015)
walls by injecting low energy electrons, the H ion current

signal decreased by a factor of 2.4, while the electron density
increased by a factor of 2. The reason for the observed reduction of H ion current despite higher electron density was
considered due to the forbidden transport of surface produced H ions originating at the chamber wall surface to the
plasma volume.
It is always difficult to separate the atomic and molecular processes in plasma from surface reaction. Steen and
Graham studied how plasma parameters and VUV to visible
light emission from a plasma changed by Cs introduction
into their ion source.241 They pointed out that reduction in
atomic hydrogen density in the plasma reduces H destruction in the plasma volume and can increase the H ion density in the plasma. The direct correlation between the
extracted H ion current and the work function of the plasma
grid has been confirmed by photoelectric method,242 and the
comparison of work function to plasma grid temperature
indicated that the low work function plasma grid surface can
be attained at high grid temperature.243
B. Cs seeded source operation for fusion experiments
and accelerators
Soon after the report of H ion current enhancement by

Cs injection into small sources, the Cs seeding was tested for
a large ion source developed for nuclear fusion experiments.244 The large input power to sustain an arc discharge
elevated the temperature of the plasma grid and increased
the H ion current due to the higher temperature of Cs
coated plasma grid. The grid temperature control allowed
obtaining H ion beam current as large as 3.3 A from a Cs
seeded ion source with 25 cm 25 cm extraction area.245
Meanwhile, Cs seeded operation was confirmed effective for
operation of a large negative ion source (34 cm diameter and
104 cm long).246 Unlike the case of surface production
source with a converter, leakage of Cs to the downstream
accelerator was not large for Cs seeded volume sources, and
800 keV, 0.32 A H ion beam was successfully extracted
from a multi aperture arc discharge source.247 By properly
controlling the plasma grid temperature, stable extraction of
H ion beam as long as 100 s was achieved.248 However,
tungsten evaporation from the cathodes of an arc discharge
source form Cs containing deposits adhering to the wall surface,249,250 which eventually requires frequent maintenance
of the machines for a thermonuclear fusion system.251
Therefore, large RF discharge driven H ion sources
equipped with a Cs oven are employed for research and
development for ITER program to determine if the system
requires less frequent maintenance than arc discharge
sources.252
Introduction of Cs seeded operation of H ion sources
for accelerators started earlier than fusion experiments.253
By the time, Leung et al. proposed the prototype ion source
for the super conducting super collider facility,140 the source
contained all major components constituting the present H
ion source for accelerators.142 The source was operated with
RF power and an optimized magnetic filter field with a collar, which was found effective to suppress electron current
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-26
Appl. Phys. Rev. 2, 021305 (2015)
for non Cs operation.254 The reduction in the amount of Cs

injected into the discharge was obtained by coating the surface of the collar.255 Currently, only 3 mg of Cs is injected
into this type of source to produce 50 mA current of H ion
beam.143
C. Optimization of Cs seeded operation of H2 sources
Large H ion sources for fusion experiments require

continuous feeding of Cs into the source.6 The role of Cs
seeding for discharge operation of a large source for fusion
experiments can be different from that of a small source for
accelerator application, where tuning of the collar surface
work function seems to play the decisive role.256 The relative
importance of the direct surface production can be examined
by changing the bias to the plasma grid. Bacal et al.
reviewed the causes of the extracted current enhancement
for different setting of the plasma grid bias (Vb).257 Results
of caesiated operation were compared with operation in pure
hydrogen for filament driven sources from JAERI244,258
(designated JAERI 10 Amp source, Figure 43) and Ecole
Polytechnique259 (designated Camembert III, Figure 44).
Obviously in Camembert III, as in JAERI 10 Amp source,
the negative ion current in caesiated operation is enhanced
compared to the current found in pure hydrogen operation.
The amount of this enhancement was identical for both sources and was about 2.5 times the value without Cs when PE
was biased below the plasma potential. In the case of caesiated operation of Camembert III (Figure 44), the extracted
H current decreased by 50% when the PE bias voltage was
raised above the plasma potential. These data indicate a clear
threshold of the surface component, i.e., a sudden decrease
in the extracted H ion current near 0.5 V.
The bias voltage to the plasma electrode (or plasma
grid) affects the amount of extracted H current depending
upon the ion source size and/or structure. The possible mechanism of enhancing the H current can be summarized
FIG. 43. Extracted H ion current from JAERI 10 Amp source, in caesiated
and pure hydrogen operation regimes. Reproduced with permission from
Rev. Sci. Instrum. 85, 02B103 (2014). Copyright 2014 AIP Publishing
LLC.265
FIG. 44. Variation of the extracted H ion current with the plasma electrode
bias voltage at two values of the extraction voltage, in pure hydrogen and
caesium seeded operation. The hydrogen pressure was 0.4 Pa. Reproduced
with permission from Rev. Sci. Instrum. 69, 932934 (1998). Copyright
depending upon the bias voltage to the plasma grid (see also
Refs. 103 and 257):
(a)
(b)
(c)
In the whole bias range. The H ion current enhancement due to caesium can take place in the whole range
of bias voltage due to gettering of atomic hydrogen by
caesium. The effect of gettering is the reduction in H
destruction rate by associative and non-associative
detachment by atomic hydrogen.
Bias voltage below plasma potential. The enhancement
of extracted current at bias voltage negative with
respect to Vp (Vb < Vp) is usually ascribed to the direct
production of H ions by positive ions and atoms striking the low work function caesium covered surface.
Surface produced H ions leave the surface, traverse
across the sheath, migrate into plasma, and increase
H ion density in the source plasma.
Bias voltage above plasma potential. In pure hydrogen
discharge at positive bias voltage with respect to
plasma potential (Vb > Vp), volume produced H ions
form flow to the extraction electrodes. Flow of H ions
in the same bias voltage range but with the injection of
Cs into the source was confirmed in the experiments
conducted at the National Institute for Fusion Science
(Toki, Japan). Using cavity ring down spectroscopy for
measuring the negative ion density260 and a group of
Langmuir probes for estimating the flow of ions in the
source plasma,261 they have shown the result that a
substantial part of the H ions forming the extracted
beam current measured at Vb > Vp seems to originate in
the plasma beyond the magnetic filter toward the driver
region. Indeed, the results of these experiments show
that at Vb > Vp the induction of the electric field to
extract negative ions did not affect local negative ion
density near the plasma grid, which was reduced by the
electric field formed by the plasma grid bias. Negative
ions produced at the caesiated plasma grid cannot be
extracted due to the electric field to return surface
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-27
Appl. Phys. Rev. 2, 021305 (2015)
produced H ions back to the grid, except in the region

where beam extraction field penetrates up to the plasma
side of the grid. This part of H beam current should
be small when extractor structure is not properly
designed for surface produced H extraction. The density of H ions in the plasma volume in this bias voltage range may be higher than the corresponding
operation condition with Cs due to reduced hydrogen
atom concentration by adsorption effect at the source
wall surface, and/or surface H production at the ion
source wall other than the plasma grid.
The volume size of the fusion experiment ion sources
makes the effect due to volume process more apparent. It
may be interesting to note that the ratio of extraction surface
area to volume is larger for fusion experiment sources than
accelerator volume sources operated with Cs seeding.
The caesium operation for an RF excited large source
was studied at IPP Garching and reported in 2006 by Speth
et al.129 As in the smaller sources, the most important effects
of caesium were a substantial increase in the negative ion
current density compared with a Cs-free source, coupled
with a reduction in electron-to-negative ion ratio. The stability and reproducibility of the RF source were obtained by
two major changes in the source operation, as pointed out by
Franzen et al.:262
All source temperatures have been controlled. The temperature of the source body was kept at 3040 C; the plasma
grid temperature was in the range of 110150 C. In contrast to filament driven arc sources, the power of the rf
plasma onto the plasma grid is not sufficient for obtaining
the required temperatures so that the plasma grid has to be
actively heated. This, however, has the advantage that the
plasma grid temperature can be kept constant during and
also between the plasma pulses.
When the source is well caesiated, optimum performance
of the source was obtained when the plasma grid was biased above the floating potential. In addition, it was found
possible to realize more reliable operation by controlling
the bias current instead of the bias voltage.
Current densities of 330 A m22 and 230 A m22 have

been achieved for hydrogen and deuterium, respectively, at
a pressure of 0.3 Pa and an electron/H ion ratio of 1 for
a small extraction area (7.0 1023 m2) and short pulses
(<4 s).
In 2009, Christ-Koch et al.263 reported the results of a
laser-photodetachment study of the negative ion density produced in the short pulse and standard size IPP test facility
BATMAN. The variation of the density of negative deuterium ions versus the bias applied to the plasma electrode
shown in Figure 45 for a well caesiated source confirmed
the variation reported by Franzen et al.;262 the negative ion
current decreased by increasing plasma electrode bias with
respect to the plasma potential. The density ratio of negative
hydrogen ions to electrons was in the range of n(H)/
ne 0.33, demonstrating that the negative ions are the
dominant negative charge species in these types of ion
sources.
FIG. 45. Negative ion density and extracted ion current for different bias
voltages applied to the plasma electrode in deuterium discharges at 53 kW
and 0.45 Pa. Shows also the density to current ratio. Reprinted with permission from Christ-Koch et al., Plasma Sources Sci. Technol. 18, 025003
(2009). Copyright 2009 Institute of Physics.
VII. CONCLUSION
Two H production mechanisms, the process in the

plasma volume and the process at the plasma surface boundary, were independently studied in early days of negative ion
source development for fusion and accelerator applications.
Both processes have been carefully examined through theoretical predictions and results obtained from fundamental
experiments. Improvements in experimental techniques have
greatly contributed to deepen the understanding how H ions
are formed through each reaction occurring both in plasma
volume and at plasma-source wall interface. As caesiation of
the volume type source demonstrated an increase of extractable H ion current by several folds, present H source performance is discussed with plasma-wall interaction together
with atomic molecular processes in plasma volume.
Contemporary H ion source structures satisfy the
requirements to enlarge production and reduce destruction of
H ions to realize efficient extraction. Formation of low temperature high density plasma separated from high power density plasma by magnetic filter field and the preparation of
low work function surface are the two key factors to realize
an efficient H/D ion source. With the correct understanding of both volume and surface processes, state of the art H
ion sources for fusion experiments can deliver H ion current with enough intensity and duration. There still exist
some engineering problems associated with the source operation stability and life, but the ion source design base is complete and ready for both fusion experiments and accelerator
ion sources.
1
R. McAdams, Rev. Sci. Instrum. 85, 02B319 (2014).

P. Franzen et al., Nucl. Fusion 55, 053005 (2015).
V. Antoni et al., Rev. Sci. Instrum. 85, 02B128 (2014).
4
M. Kashiwagi et al., Rev. Sci. Instrum. 85, 02B320 (2014).
5
N. Umeda et al., Rev. Sci. Instrum. 85, 02B304 (2014).
6
Y. Takeiri et al., AIP Conf. Proc. 1515, 139 (2013).
7
R. S. Hemsworth and T. Inoue, IEEE Trans. Plasma Sci. 33(6), 1799
(2005).
8
L. W. Alvarez, Rev. Sci. Instrum. 22, 705 (1951).
9
K. W. Ehlers, Nucl. Instrum. Methods 32(2) 309316 (1965).
10
G. I. Budker and G. I. Dimov, in Proceedings of the International
Conference on High Energy Accelerators, Dubna (1963), p. 993.
11
R. L. Martin, IEEE Trans. Nucl. Sci. NS-18, 953 (1971).
2
3
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-28
12
G. I. Dimov, Rev. Sci. Instrum. 67(10) 3393 (1996).

G. I. Dimov and G. V. Roslyakov, Prib. Tekh. Eksp. No. 1, 29 (1974).
14
J. A. Fasolo, IEEE Trans. Nucl. Sci. NS-24, 1597 (1977).
15
E. B. Hooper, Jr., O. A. Anderson, T. J. Orzechowski, and P. Poulsen, in
Proc. Symp. on the Production and Neutralization of Negative Hydrogen
Ions and Beams, BNL Report 50727, 2630 September 1977, edited by
K. Prelec, p. 163.
16
N. N. Semashko, V. V. Kusnetsov, and A. I. Krylov, in Proc. Symp. on
the Production and Neutralization of Negative Hydrogen Ions and
Beams, BNL Report 50727, 2630 September 1977, edited by K. Prelec,
p. 170.
17
Yu. I. Belchenko, G. I. Dimov, and V. G. Dudnikov, Izv. Akad. Nauk
SSSR, Ser. Fiz. 37, 2573 (1973).
18
K. W. Ehlers and K. N. Leung, Rev. Sci. Instrum. 51, 721 (1980).
19
E. Nicolopoulou, M. Bacal, and H. J. Doucet, J. Phys. (Paris) 38,
13991404 (1977); , in Proc. Symp. on the Production and Neutralization
of Negative Hydrogen Ions and Beams, BNL Report 50727 (1977), edited
by K. Prelec, p. 26.
20
M. Bacal and G. Hamilton, Phys. Rev. Lett. 42, 1538 (1979).
21
K. Prelec and T. Sluyters, Rev. Sci. Instrum. 44, 1451 (1973).
22
C. D. Moak, H. E. Banta, J. T. Thurston, J. W. Johnson, and R. F. King,
Rev. Sci. Instrum. 30, 694 (1959).
23
P. Lawrence, K. E. Beauchamp, and J. L. McKibben, Nucl. Instrum.
Methods 32, 357 (1965).
24
Yu. I. Belchenko, G. I. Dimov, and V. G. Dudnikov, Sov. Phys.-Tech.
Phys. 18, 1083 (1974).
25
Yu. I. Belchenko, G. I. Dimov, V. G. Dudnikov, and A. S. Kuprianov,
Rev. Phys. Appl. 23, 1847 (1988).
26
J. M. Wadehra and J. N. Bardsley, Phys. Rev. Lett. 41, 1795 (1978);
Phys. Rev. A 20, 1398 (1979).
27
M. Allan and S. F. Wong, Phys. Rev. Lett. 41, 1791 (1978).
28
J. N. Bardsley and J. S. Cohen, J. Phys. B: At. Mol. Phys. 11, 3645
(1978).
29
C. Bottcher and B. D. Buckley, J. Phys. B: At. Mol. Phys. 12, L497
(1979).
30
J. N. Bardsley, J. Phys. B: At. Mol. Phys. 1, 349 (1968).
31
T. F. OMalley, Phys. Rev. 150, 14 (1966).
32
G. J. Schulz, Rev. Mod. Phys. 45, 423 (1973).
33
J. M. Wadehra, Phys. Rev. A 29(1), 106 (1984).
34
J. M. Wadehra, Appl. Phys. Lett. 35, 917 (1979).
35
J. P. Gauyacq, J. Phys. B: At. Mol. Phys. 18, 1859 (1985).
36
I. I. Fabrikant, J. M. Wadehra, and Y. Xu, Phys. Scr. T96, 45 (2002).
37
W. Domcke, Phys. Rep. 208, 97 (1991).
38
^
M. Cizek,
J. Horacek, and W. Domcke, J. Phys. B: At., Mol. Opt. Phys.
31, 2571 (1998).
39
^
J. Horacek, M. Cizek,
K. Houfek, P. Kolorene, and W. Domcke, Phys.
Rev. A 70, 052712 (2004).
40
M. Bacal, Nucl. Fusion 46, S250 (2006).
41
W. B. Kunkel, private communication (1980).
42
J. R. Hiskes, J. Appl. Phys. 51, 4592 (1980).
43
J. R. Hiskes, J. Appl. Phys. 70, 3409 (1991).
44
J. Marx, A. Lebehot, and R. Campargue, J. Phys. (Paris) 46, 1667 (1985).
45
M. Cacciatore, M. Capitelli, R. Celiberto, P. Civet, and C. Gorse, AIP
Conf. Proc. 210, 74 (1990).
46
W. G. Graham, J. Phys. D: Appl. Phys. 17, 2225 (1984).
47
M. Pealat, J.-P. E. Taran, M. Bacal, and A. M. Bruneteau, J. Chem. Phys.
82, 4943 (1985).
48
R. I. Hall, I. Cadez, M. Landau, F. Pichou, and C. Schermann, Phys. Rev.
Lett. 60(4), 337 (1988).
49
C. Schermann, R. I. Hall, M. Landau, F. Pichou, and I. Cadez, AIP Conf.
Proc. 210, 159 (1991).
50
A. T. Young, G. C. Stutzin, K. N. Leung, and W. B. Kunkel, AIP Conf.
Proc. 210, 450 (1991).
51
G. C. Stutzin, A. T. Young, A. S. Schlachter, J. W. Stearns, K. N. Leung,
W. B. Kunkel, G. T. Worth, and R. R. Stevens, Rev. Sci. Instrum. 59,
1479 (1988).
52
G. C. Stutzin, A. T. Young, A. S. Schlachter, K. N. Leung, and W. B.
Kunkel, Chem. Phys. Lett. 155, 475 (1989).
53
J. R. Hiskes and A. M. Karo, Appl. Phys. Lett. 54, 508 (1989).
54
T. Mosbach, Plasma Sources Sci. Technol. 14, 610 (2005).
55
M. B. Hopkins and K. N. Mellon, Phys. Rev. Lett. 67(4), 449 (1991).
56
K. N. Leung, K. W. Ehlers, and R. V. Pyle, Appl. Phys. Lett. 47, 227
(1985).
57
O. Fukumasa and S. Saeki, J. Phys. D: Appl. Phys. 20, 237 (1987).
13
Appl. Phys. Rev. 2, 021305 (2015)

58
T. Inoue, Y. Masuda, Y. Ohara, Y. Okumura, M. Bacal, and P.

Berlemont, Plasma Sources Sci. Technol. 1, 75 (1992).
59
T. Kuo, D. Yuan, K. Jayamanna, M. McDonald, R. Baartman, P. Schmor,
and G. Duto, Rev. Sci. Instrum. 67, 1314 (1996).
60
J. Peters, Rev. Sci. Instrum. 71, 1069 (2000).
61
D. Spence and K. R. Lykke, in Proceedings of XIX International Linear
Accelerator Conference, Argonne National Lab. (1998), Report ANL-98/
28, pp. 508510.
62
M. Bacal, A. A. Ivanov, Jr., M. Glass-Maujean, Y. Matsumoto, M.
Nishiura, M. Sasao, and M. Wada, Rev. Sci. Instrum. 75, 1699 (2004).
63
M. Bacal and D. A. Skinner, Comments At. Mol. Phys. 23, 283 (1990).
64
M. Bacal, in Nonequilibrium Processes in Partially Ionized Gases, Proc.
NATO ASI Series, Series B, edited by M. Capitelli and J. N. Bardsley
(Plenum, New York, 1990), Vol. 220, p. 213.
65
A. M. Karo, J. R. Hiskes, and R. Hardy, J. Vac. Sci. Technol., A 3, 1222
(1985).
66
P. Berlemont, D. A. Skinner, and M. Bacal, Chem. Phys. Lett. 183, 397
(1991).
67
J. M. Wadehra, AIP. Conf. Proc. 111, 46 (1984).
68
G. J. Schulz and R. K. Asundi, Phys. Rev. 158, 25 (1967).
69
D. Rapp, T. E. Sharp, and D. D. Briglia, Phys. Rev. Lett. 14, 533 (1965).
70
Yu. N. Demkov, Phys. Lett. 15, 235 (1965).
71
Yu. N. Demkov, Zh. Eksp. Teor. Fiz. 46, 1126 (1964), [Sov. Phys. JETP
19, 762 (1964)].
72
C. M
undel, M. Berman, and W. Domcke, Phys. Rev. A 32, 181 (1985).
73
M. Bacal, A. Hatayama, and J. Peters, IEEE Trans. Plasma Sci. 33, 1845
(2005).
74
M. Bacal, Chem. Phys. 396, 3 (2012).
75
K. N. Leung, K. W. Ehlers, and M. Bacal, Rev. Sci. Instrum. 54, 56 (1983).
76
M. Bacal, A. M. Bruneteau, H. J. Doucet, W. G. Graham, and G. W.
Hamilton, H and D production in plasma, in Proceedings of the
Second International Symposium on the Production and Neutralization of
Negative Hydrogen Ions and Beams, BNL Report. 51304 (1980), edited
by T. Sluyters.
77
T. S. Green, Development of negative ion sources, in Proceedings of
11th Symp. Fusion Eng., Austin, TX, November 1985, pp. 103107.
78
H. Horiike, Y. Ohara, Y. Okumura, T. Shibata, and S. Tanaka,
Conceptual design of negative-ion-based 500 keV 20 MW neutral beam
injector, Japan Atomic Energy Research Institute, Report No. JAERIM86-064, April 1986.
79
Y. Okumura, H. Horiike, T. Inoue, T. Kurashima, S. Matsuda, Y. Ohara,
and S. Tanaka, AIP Conf. Proc. 158, 309 (1986).
80
K. W. Ehlers, B. F. Gavin, and E. L. Habbard, Nucl. Instrum. Methods
22, 87 (1963).
81
G. I. Dimov et al., Prib. Tekh. Eksp. No. 4, 30 (1967).
82
H. Zhang, Ion Sources (Science Press, Springer Verlag, Beijin, 1999).
83
K. Jimbo, Phys. Fluids 27, 2752 (1984).
84
K. Jimbo, K. W. Ehlers, K. N. Leung, and R. V. Pyle, Nucl. Instrum.
Methods Phys. Res., Sect. A 248(23), 282 (1986).
85
V. P. Goretsky, A. V. Riabtsev, I. A. Soloshenko, A. F. Tarasenko, and
A. I. Schedrin, Rev. Sci. Instrum. 67, 1622 (1996).
86
C. C. Tsai, W. K. Dagenhart, W. L. Stirling, G. C. Barber et al., AIP
Conf. Proc. 158, 194 (1987).
87
Yu. I. Belchenko, G. I. Dimov, and V. G. Dudnikov, Nucl. Fusion 14,
113 (1974).
88
Yu. I. Belchenko and A. S. Kupriyanov, Rev. Phys. Appl. (Paris) 23,
1889 (1988).
89
K. N. Leung, G. J. DeVries, K. W. Ehlers, L. T. Jackson, J. W. Stearns,
M. D. Williams, M. G. McHarg, D. P. Ball, W. T. Lewis, and P. W.
Allison, Rev. Sci. Instrum. 58, 235 (1987); AIP Conf. Proc. 158, 356
(1987).
90
M. Bacal and K. N. Leung, AIP Conf. Proc. 111, 105 (1983).
91
K. N. Leung and M. Bacal, Rev. Sci. Instrum. 55(3), 338 (1984).
92
M. Bacal, A. M. Bruneteau, and M. Nachman, J. Appl. Phys. 55, 15
(1984).
93
C. Courteille, J. Bruneteau, F. P. G. Valckx, Z. Sledziewski, and M.
Bacal, Rev. Sci. Instrum. 64, 3265 (1993).
94
C. Courteille, A. M. Bruneteau, and M. Bacal, Rev. Sci. Instrum. 66,
2533 (1995).
95
A. A. Ivanov, Jr., C. Rouille, M. Bacal, Y. Arnal, S. Bechu, and J.
Pelletier, Rev. Sci. Instrum. 75, 1750 (2004).
96
R. Gobin, P. Auvray, M. Bacal, J. Breton, O. Delferrie`re, F. Harrault,
A. A. Ivanov, Jr., P. Svarnas, and O. Tuske, Nucl. Fusion 46, S281
(2006).
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-29
97
P. Svarnas, J. Breton, M. Bacal, and R. Faulkner, IEEE Trans. Plasma

Sci. 35, 11561162 (2007).
98
M. Bacal, J. Bruneteau, and P. Devynck, Rev. Sci. Instrum. 59, 2152
(1988).
99
F. El Balghiti-Sube, F. G. Baksht, and M. Bacal, Rev. Sci. Instrum. 67,
2221 (1996).
100
A. Hatayama, T. Sakurabayashi, Y. Ishi, K. Makino, M. Ogasawara, and
M. Bacal, Rev. Sci. Instrum. 73, 910 (2002).
101
T. Sakurabayashi, A. Hatayama, K. Miyamoto, M. Ogasawara, and M.
Bacal, Rev. Sci. Instrum. 73, 1048 (2002).
102
S. Kuppel, D. Matsushita, A. Hatayama, and M. Bacal, J. Appl. Phys.
109, 013305 (2011).
103
M. Bacal, M. Sasao, M. Wada, and R. McAdams, AIP Conf. Proc. 1655,
02001 (2015).
104
A. J. T. Holmes, G. Dammertz, and T. S. Green, Rev. Sci. Instrum. 56,
1697 (1985).
105
R. McAdams, A. J. T. Holmes, M. P. S. Nightingale, L. M. Lea, M. D.
Hinton, A. F. Newman, and T. S. Green, AIP Conf. Proc. 158, 298 (1987).
106
R. McAdams, R. F. King, and A. F. Newman, Rev. Sci. Instrum. 61, 2176
(1990).
107
P. J. Eenshuistra, M. Gochitashvilli, R. Becker, A. W. Kleyn, and H. J.
Hopman, J. Appl. Phys. 67, 85 (1990).
108
P. McNeely, D. Boilson, N. Curran, M. B. Hopkins, and D. Vender, AIP
Conf. Proc. 439, 113 (1998).
109
A. M. Bruneteau, C. Courteille, R. Leroy, and M. Bacal, Rev. Sci.
Instrum. 67, 3827 (1996).
110
P. Svarnas, J. Breton, M. Bacal, and T. Mosbach, Rev. Sci. Instrum. 77,
03A532 (2006).
111
T. S. Paunska, A. Shivarova, and K. H. Tarnev, J. Appl. Phys. 107,
083301 (2010).
112
Ts. Paunska, A. Shivarova, Kh. Tarnev, and Ts. Tsankov, Phys. Plasmas
18, 023503 (2011).
113
St. Lishev, A. Shivarova, D. Iordanov, D. Todorov, and A. Demerdzhiev,
AIP Conf. Proc. 1515, 270 (2012).
114
St. Lishev, Ts. Paunska, A. Shivarova, and Kh. Tarnev, Rev. Sci. Instrum.
83, 02A702 (2013).
115
St. Lishev, A. Shivarova, and Kh. Tarnev, Rev. Sci. Instrum. 83, 02A709
(2013).
116
St. Lishev, D. Yordanov, and A. Shivarova, Rev. Sci. Instrum. 85,
02B101 (2014).
117
A. Demerdzhiev, Kh. Tarnev, St. Lishev, and A. Shivarova, Rev. Sci.
Instrum. 85, 02B105 (2014).
118
J. R. Hiskes, A. M. Karo, and P. A. Willmann, J. Appl. Phys. 58, 1759
(1985).
119
K. N. Leung, K. W. Ehlers, and R. V. Pyle, Rev. Sci. Instrum. 56, 2097
(1985).
120
N. P. Curran, M. B. Hopkins, D. Vender, and B. W. James, Plasma
Sources Sci. Technol. 9, 169 (2000).
121
M. Bacal, M. Nishiura, M. Sasao, M. Hamabe, M. Wada, and H.
Yamaoka, Rev. Sci. Instrum. 73, 903 (2002).
122
A. A. Ivanov, Jr., M. Bacal, C. Rouille, M. Nishiura, and M. Sasao, Rev.
Sci. Instrum. 75, 1747 (2004).
123
E. Speth, M. Ciric, J. H. Feist et al., Fusion Eng. Des. 46, 383 (1999).
124
O. Vollmer, B. Heinemann, W. Kraus et al., Rev. Sci. Instrum. 71, 939
(2000).
125
O. Vollmer, B. Heinemann, W. Kraus et al., Fusion Eng. Des. 5657, 465
(2001).
126
W. Kraus, private communication (2015).
127
W. Kraus, P. McNeely, E. Speth, B. Heinemann, O. Vollmer, and R.
Wilhelm, Rev. Sci. Instrum. 73, 1096 (2002).
128
W. Kraus, P. Franzen, B. Heinemann, E. Speth, and O. Vollmer, Fusion
Eng. Des. 5657, 499 (2001).
129
E. Speth, H. D. Falter, P. Franzen et al., Nucl. Fusion 46, S220 (2006).
130
U. Fantz, H. Falter, P. Franzen et al., Nucl. Fusion 46, S297 (2006).
131
P. Franzen, L. Schiesko, M. Fr
oschle, D. W
underlich, U. Fantz et al.,
Plasma Phys. Controlled Fusion 53, 115006 (2011).
132
L. Schiesko, P. Franzen, and U. Fantz, Plasma Sources Sci. Technol. 21,
065007 (2012).
133
L. Schiesko, C. Hopf, P. Franzen, W. Kraus, R. Riedl, U. Fantz et al.,
Nucl. Fusion 51, 113021 (2011).
134
L. Schiesko, P. McNeely, P. Franzen, U. Fantz et al., Plasma Phys.
Controlled Fusion 54, 105002 (2012).
135
D. Wunderlich, U. Fantz, P. Franzen, R. Riedl, and F. Bonomo, Rev. Sci.
Instrum. 84, 093102 (2013).
Appl. Phys. Rev. 2, 021305 (2015)

136
U. Fantz, P. Franzen, B. Heinemann, and D. W

underlich, Rev. Sci.
Instrum. 85, 02B305 (2014).
137
U. Fantz, L. Schiesko, and D. W
underlich, Plasma Sources Sci. Technol.
23, 044002 (2014).
138
P. Franzen, D. W
underlich, U. Fantz et al., Plasma Phys. Controlled
Fusion 56, 025007 (2014).
139
K. N. Leung, K. W. Ehlers, C. A. Hauck, W. B. Kunkel, and A. F.
Lietzke, Rev. Sci. Instrum. 59, 453 (1988).
140
K. N. Leung, G. J. DeVries, W. F. DiVergilio, R. W. Hamm, C. A.
Hauck, W. B. Kunkel, D. S. McDonald, and M. B. Williams, Rev. Sci.
Instrum. 62, 100 (1991).
141
K. N. Leung, Rev. Sci. Instrum. 65(4), 1165 (1994).
142
M. P. Stockli, B. X. Han, T. W. Hardek et al., Rev. Sci. Instrum. 83,
02A732 (2012).
143
M. P. Stockli, K. D. Ewald, B. X. Han et al., Rev. Sci. Instrum. 85,
02B137 (2014).
144
145
J. Peters, Rev. Sci. Instrum. 79, 02A515 (2008).
146
D. P. Moehs, J. Peters, and J. J. Peters, Rev. Sci. Instrum. 69, 1991
(1998).
147
D. P. Moehs, J. Peters, and J. Sherman, IEEE Trans. Plasma Sci. 33, 1786
(2005).
148
D. K
uchler, T. Meinschad, J. Peters, and R. Scrivens, Rev. Sci. Instrum.
79, 02A504 (2008).
149
C. Schmitzer, M. Kronberger, J. Lettry, J. Sanchez-Arias, and H. St
ori,
Rev. Sci. Instrum. 83, 02A715 (2012).
150
M. Kronberger, D. Faircloth, J. Lettry, M. Paoluzzi, H. Pereira, J.
Sanchez Arias, C. Schmitzer, and R. Scrivens, Rev. Sci. Instrum. 83,
02A703 (2012).
151
A. Ueno, K. Ikegami, and Y. Kondo, Rev. Sci. Instrum. 75, 1714 (2004).
152
A. Ueno, H. Oguri, K. Ikegami, Y. Namekawa, and K. Ohkoshi, Rev. Sci.
Instrum. 81, 02A720 (2010).
153
Y. Hwa An, B. Ki Jung, and Y. S. Hwang, Rev. Sci. Instrum. 81, 02A702
(2010).
154
Y. Hwa An, W. Hwi Cho, K.-J. Chung, K. Lee, S. Jang, S.-G. Lee, and Y.
S. Hwang, Rev. Sci. Instrum. 83, 02A727 (2012).
155
K.-J. Chong, B.-K. Jung, Y. Hwa An, J.-J. Dang, and Y. S. Hwang, Rev.
Sci. Instrum. 85, 02B119 (2014).
156
A. Lacoste, T. Lagarde, S. Bechu, Y. Arnal, and J. Pelletier, Plasma
157
P. Svarnas, M. Bacal, P. Auvray, S. Bechu, and J. Pelletier, Rev. Sci.
Instrum. 77, 03A512 (2006).
158
P. Svarnas, B. M. Annaratone, S. Bechu, J. Pelletier, and M. Bacal,
Plasma Sources Sci. Technol. 18, 045010 (2009).
159
S. Bechu, A. Soum-Glaude, A. Be`s, P. Svarnas, S. Aleiferis, A. A.
Ivanov, Jr., and M. Bacal, Phys. Plasmas 20, 101601 (2013).
160
S. Aleiferis, P. Svarnas, I. Tsiroudis, S. Bechu, M. Bacal, and A. Lacoste,
IEEE Trans. Plasma Sci. 42, 2828 (2014).
161
K. S. Woodcock, Phys. Rev. 38, 1696 (1931).
162
R. A. Baragiola, E. R. Salvatelli, and E. Alonso, Nucl. Instrum. Methods
110, 507 (1973).
163
J. A. Philips, Phys. Rev. 97, 404 (1955).
164
W. Gruebler, P. Schwandlt, T. J. Yule, and W. Haeberli, Nucl. Instrum.
Methods 41, 245 (1996).
165
H. Verbeek, W. Eckstein, and S. Datz, J. Appl. Phys. 47, 1785 (1976).
166
M. E. Kishinevskiy, Novosibirsk Report No. 11673 (1973); Sov. J.
Tech. Phys. 48, 773 (1978).
167
R. K. Janev, Surf. Sci. 45, 609 (1974).
168
J. R. Hiskes, J. Phys. Colloques 40, C7179 (1979).
169
B. Rasser, J. N. M. van Wunnik, and J. Los, Surf. Sci. 118, 697 (1982).
170
A. Blandin, A. Nourtier, and D. Hone, J. Phys. 37, 369 (1976).
171
H. L. Cui, J. Vac. Sci. Technol., A 9, 1823 (1991).
172
W. R. Koppers, B. Berenbak, D. Vlachos, U. van Slooten, and A. W.
Kleyn, Phys. Rev. B 57, 13246 (1998).
173
R. Zimny, Surf. Sci. 292, 325 (1993).
174
F. Wyptta, R. Zimny, and H. Winter, Nucl. Instrum. Methods Phys. Res.,
Sect. B 58, 379 (1991).
175
M. Maazouz, R. Baragiola, A. Borisov, V. A. Esaulov, S. Lacombe, J.
P. Gauyacq, L. Guillemot, and D. Teillet-Billy, Surf. Sci. 364, L568
(1996).
176
A. G. Borisov and H. Winter, Nucl. Instrum Methods Phys. Res., Sect. B
115, 142 (1996).
177
A. A. Magunov, I. K. Gainullin, and I. F. Urazgildin, Bull. Russ. Acad.
Sci.: Phys. 71, 1486 (2007).
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-30
178
E. R. Amanbaev, D. K. Sestakov, and I. K. Gainullin, J. Surf. Invest. 3,

865 (2009).
179
K. Tsumori, W. R. Koppers, R. M. A. Heeren, M. F. Kadodwala, J. H. M.
Beijersbergen, and A. W. Kleyn, J. Appl. Phys. 81, 6390 (1997).
180
F. Bonetto, E. A. Garcia, R. Vidal, J. Ferron, and E. C. Goldberg, Appl.
Surf. Sci. 254, 62 (2007).
181
B. Obreshkov and U. Thumm, Phys. Rev. A 76, 052902 (2007).
182
M. Maazouz, L. Guilleot, V. A. Esaulov, and D. J. OConnor, Surf. Sci.
398, 49 (1998).
183
J. R. Hiskes, A. Karo, and M. Gardner, J. Appl. Phys. 47, 3888 (1976).
184
P. J. Schneider, K. H. Berkner, W. G. Graham, R. V. Pyle, and J. W.
Stearns, Phys. Rev. B 23, 941 (1981).
185
J. N. M. van Wunnik, J. J. C. Geerlings, E. H. A. Granneman, and J. Los,
Surf. Sci. 131, 17 (1983).
186
J. R. Hiskes and P. J. Schneider, Phys. Rev. B 23, 949 (1981).
187
E. H. A. Granneman, J. J. C. Geerlings, J. N. M. Van Wunnik, P. J. van
Bommel, H. J. Hopman, and J. Los, AIP Conf. Proc. 111, 206 (1984).
188
P. J. M. van Bommel, J. J. C. Geerlings, J. N. M. van Wunnik, P.
Massmann, E. H. A. Granneman, and J. Los, J. Appl. Phys. 54, 5676
(1983).
189
J. J. C. Geerlings, P. W. van Amerstfoort, L. F. Tz. Kwakman, E. H. A.
Granneman, J. Los, and J. P. Gauyaco, Surf. Sci. 157, 151 (1985).
190
M. Seidl, H. L. Cui, J. D. Isenberg, H. J. Kwon, B. S. Lee, and S. T.
Melnychuk, J. Appl. Phys. 79, 2896 (1996).
191
P. W. van Amersfoort, J. J. C. Geerings, R. Rodink, E. H. A. Granneman,
and J. Los, J. Appl. Phys. 59, 241 (1986).
192
A. Pargellis and M. Seidel, Phys. Rev. B 25, 4356 (1982).
193
J. L. Lopes, J. A. Greer, and M. Seidl, J. Appl. Phys. 60, 17 (1986).
194
B. S. Lee and M. Seidl, Appl. Phys. Lett. 61, 2857 (1992).
195
T. Okuyama and Y. Mori, Rev. Sci. Instrum. 63, 2711 (1992).
196
M. L. Yu, Phys. Rev. Lett. 40, 574 (1978).
197
A. Hoffman and M. Petravic, Phys. Rev. B 53, 6996 (1996).
198
C. Goeden and G. Dollinger, Appl. Surf. Sci. 147, 107 (1999).
199
H. Kawano, Int. J. Mass Spectrom. Ion Processes 104, 23 (1991).
200
M. Wada, R. V. Pyle, and J. W. Stearns, J. Appl. Phys. 67, 6334 (1990).
201
J. R. Hiskes, A. M. Karo, E. Wimmer, A. J. Freeman, and S. R. Chubb,
J. Vac. Sci. Technol., A 2, 670 (1984).
202
K. N. Leung and K. W. Ehlers, J. Nucl. Mater. 128129, 694 (1984).
203
K. Prelec, Nucl. Instrum. Methods 144, 413 (1977).
204
K. N. Leung and K. W. Ehlers, Rev. Sci. Instrum. 53, 803 (1982).
205
J. W. Kwan, G. D. Ackerman, O. A. Anderson, C. F. Chan, W. S. Cooper,
G. J. DeVries, A. F. Lietzke, L. Soroka, and W. F. Steele, Rev. Sci.
Instrum. 57, 831 (1986).
206
O. A. Anderson, Large aperture D- accelerators, in Proceedings of the
Second International Symposium on the Production and Neutralization of
Negative Hydrogen Ions and Beams, edited by Th. Sluyters (1980), p.
355.
207
A. I. Hershcovitch, U. J. Kovarik, and K. Prelec, Rev. Sci. Instrum. 57,
827 (1986).
208
C. C. Tsai, R. R. Feezell, H. H. Haselton, P. M. Ryan, D. E. Shechter, W.
L. Stirling, and J. H. Whealton, Characteristics of a modified duoPIGatron
negative ion source, in Proceedings of the Second International
Symposium on the Production and Neutralization of Negative Hydrogen
Ions and Beams, edited by Th. Sluyters (1980), p. 225.
209
P. J. van Bommel, J. H. M. Bonnie, E. H. A. Granneman, P. Massmann,
and H. J. Hopman, J. Appl. Phys. 58, 2492 (1985).
210
Y. Chun Tong, P. W. van Amersfoort, E. H. A. Granneman, and H. J.
Hopman, Phys. Lett. A 110, 24 (1985).
211
P. W. van Amersfoort, Y. C. Tong, and E. H. A. Granneman, J. Appl.
Phys. 58, 2317 (1985).
212
C. F. A. van Os, E. H. A. Granneman, and P. W. van Amersfoort, J. Appl.
Phys. 61, 5000 (1987).
213
C. F. A. van Os, W. B. Kunkel, C. Leguijt, and J. Los, J. Appl. Phys. 70,
2575 (1991).
214
R. M. A. Heeren, D. Ciric, S. Yagura, H. J. Hopman, and A. W. Kleyn,
Nucl. Instrum. Methods Phys. Res., Sect. B 69, 389 (1992).
215
R. M. A. Heeren, M. J. de Graaf, D. Ciric, S. Yagura, H. J. Hopman, and
A. W. Kleyn, Appl. Surf. Sci. 70/71, 332 (1993).
216
L. Schiesko, M. Carrere, G. Cartry, and J. M. Layet, Plasma Sources Sci.
Technol. 17, 035023 (2008).
217
L. Schiesko, M. Carrere, J.-M. Layet, and G. Cartry, Plasma Sources Sci.
Technol. 19, 045016 (2010).
218
P. Kumar, A. Ahmad, C. Pardanaud, M. Carrere, J. M. Layet, G. Cartry, F.
Silva, A. Gicquel, and R. Engeln, J. Phys. D: Appl. Phys. 44, 372002 (2011).
Appl. Phys. Rev. 2, 021305 (2015)

219
A. Ahmad, J. Dubois, T. Pasquet, M. Carrere, M. M. Layet, J. B. Faure,

G. Cartry, P. Kumar, T. Minea, S. Mochalskyy, and A. Simonin, Plasma
220
R. Middleton, IEEE Trans. Nucl. Sci. NS-23, 1098 (1976).
221
G. I. Budker, G. I. Dimov, and V. G. Dudnikov, Sov. At. Energy 22, 441
(1967); At. Energ. 22, 348356 (1967).
222
Yu. I. Belchenko, G. E. Derevyankin, G. I. Dimov, and V. G. Dudnikov,
J. Appl. Mech. Tech. Phys. 28, 568 (1987).
223
V. G. Dudnikov, Rev. Sci. Instrum. 63, 2660 (1992).
224
V. G. Dudnikov, Rev. Sci. Instrum. 73, 992 (2002).
225
Yu. I. Belchenko, V. I. Davydenko, G. E. Derevyankin, G. I. Dimov, V.
G. Dudnikov, I. I. Morosov, G. V. Roslyakov, and A. L. Schabalin, Rev.
Sci. Instrum. 61, 378 (1990).
226
V. G. Dudnikov, G. E. Derevyankin, D. V. Kovalevsky, V. Ya. Savkin, E.
A. Sokolovsky, and S. K. Guharay, Rev. Sci. Instrum. 67, 1614 (1996).
227
D. Faircloth et al., AIP Conf. Proc. 1515, 359 (2013).
228
D. C. Faircloth, J. W. G. Thomason, and M. O. Whitehead, Rev. Sci.
Instrum. 75, 1735 (2004).
229
J. Lettry et al., Rev. Sci. Instrum. 83, 02A728 (2012).
230
R. L. York, R. R. Stevens, Jr., R. A. Dehaven, J. R. McConnell, E. P.
Chamberlin, and R. Kandarian, Nucl. Instrum. Methods Phys. Res., Sect.
B 1011, 891 (1985).
231
A. Takagi, Y. Mori, K. Ikegami, and S. Fukumoto, IEEE Trans. Nucl.
Sci. NS-32, 1782 (1985).
232
G. D. Alton, Y. Mori, A. Takagi, A. Ueno, and S. Fukumoto, Nucl.
Instrum. Methods Phys. Res., Sect. A 270, 194 (1988).
233
J. W. Kwan, G. D. Ackerman, C. F. Chan, M. E. Cooper, W. S. Cooper,
G. J. deVries, K. N. Leung, W. F. Steele, R. P. Wells, A. West, and M. D.
Williams, AIP Conf. Proc. 287, 295 (1994).
234
J. G. Alessi, A. Hershcovich, and Th. Sluyter, Rev. Sci. Instrum. 55, 8 (1984).
235
S. R. Walther, K. N. Leung, and W. B. Kunkel, J. Appl. Phys. 64, 3424
(1988).
236
S. R. Walther, K. N. Leung, and W. B. Kunkel, Appl. Phys. Lett. 54, 210
(1989).
237
K. W. Ehlers, K. N. Leung, R. V. Pyle, and W. B. Kunkel, AIP. Conf.
Proc. No. 158, 282 (1987).
238
H. J. Hopman, H. Zeijlemaker, and J. Verhoeven, Appl. Surf. Sci. 171,
197 (2001).
239
M. Nishiura, M. Sasao, and M. Wada, in Proceedings of the International
Conference on Ion Implantation Technology (1999), p. 318.
240
K. N. Leung, S. R. Walther, and W. B. Kunkel, Phys. Rev. Lett. 62, 764
(1989).
241
P. G. Steen and W. G. Graham, Appl. Phys. Lett. 75, 2737 (1999).
242
K. Shinto, Y. Okumura, T. Ando, M. Wada, H. Tsuda, T. Inoue, K.
Miyamaoto, and A. Nagase, Jpn. J. Appl. Phys., Part 1 35, 1894 (1996).
243
T. Morishita, M. Kashiwagi, M. Hanada, Y. Okumura, K. Watanabe, A.
Hatayama, and M. Ogasawara, Jpn. J. Appl. Phys., Part 1 40, 4709 (2001).
244
Y. Okumura, M. Hanada, T. Inoue, H. Kojima, Y. Matsuda, Y. Ohara, M.
Seki, and K. Watanabe, AIP Conf. Proc. 210, 169 (1990).
245
A. Ando, Y. Takeiri, K. Tsumori, O. Kaneko, Y. Oka, R. Akiyama, T.
Kawamoto, K. Mineo, T. Kurata, and T. Kuroda, Rev. Sci. Instrum. 63,
2683 (1992).
246
M. Hanada, T. Inoue, M. Mizuno, Y. Ohara, Y. Okumura, Y. Suzuki, H.
Tanaka, M. Tanaka, and K. Watanabe, Rev. Sci. Instrum. 63, 2699
(1992).
247
M. Taniguchi, T. Inoue, N. Umeda, M. Kashiwagi, K. Watanabe, H.
Tobari, M. Dairaku, and K. Sakamoto, Rev. Sci. Instrum. 79, 02C110
(2008).
248
A. Kojima, M. Hanada, M. Yoshida, H. Tobari, M. Kashiwagi, N.
Umeda, K. Watanabe, and L. R. Grisham, Rev. Sci. Instrum. 85, 02B312
(2014).
249
R. S. Hemsworth, D. Boilson, H. P. L. De Esch, A. Krylov, P. l.
Massmann, and L. Svensson, Nucl. Fusion 46, S239 (2006).
250
A. Krylov, D. Boilson, U. Fantz, R. S. Hemsworth, O. Provitina, S.
Pontremoli, and B. Zaniol, Nucl. Fusion 46, S324 (2006).
251
R. Hemsworth et al., Nucl. Fusion 49, 045006 (2009).
252
A. Rizzolo, M. Pavei, and N. Pomaro, Fusion Eng. Des. 88, 1007 (2013).
253
K. N. Leung, C. A. Hauck, W. B. Kunkel, and S. R. Walther, Rev. Sci.
Instrum. 60, 531 (1989).
254
K. N. Leung, C. A. Hauck, W. B. Kunkel, and S. R., Walther, Rev. Sci.
Instrum. 61, 1110 (1990).
255
K. N. Leung, D. A. Bachman, and D. S. McDonald, Rev. Sci. Instrum.
64, 970973 (1993).
256
M. Stockli, J. Phys.: Conf. Ser. 399, 012001 (2012).
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44
021305-31
257
M. Bacal, R. McAdams, and E. Surrey, Rev. Sci. Instrum. 86, 02B103

(2014).
258
M. Tanaka, M. Hanada, T. Inoue, and Y. Okumura, Report No. JAERI-M
93-132, Japan Atomic Energy Research Institute, 1993 (in Japanese).
259
M. Bacal, F. El Balghiti-Sube, L. I. Elizarov, and A. Y. Tontegode, Rev.
Sci. Instrum. 69, 932934 (1998).
260
H. Nakano, K. Tsumori, M. Kisaki, K. Ikeda, K. Nagaoka, M. Osakabe,
O. Kaneko, E. Asano, T. Kondo, M. Sato, M. Shibuya, S. Komada, H.
Sekiguchi, and Y. Takeiri, AIP Conf. Proc. 1515, 237 (2013).
Appl. Phys. Rev. 2, 021305 (2015)

261
K. Tsumori, H. Nakano, M. Kisaki, K. Ikeda, K. Nagaoka, M. Osakabe,

Y. Takeiri, O. Kaneko, M. Shibuya, E. Asano, T. Kondo, M. Sato, S.
Komada, and H. Sekiguchi, AIP Conf. Proc. 1515, 149 (2013).
262
P. Franzen, H. D. Falter, U. Fantz et al., Nucl. Fusion 47, 264
(2007).
263
S. Christ-Koch, U. Fantz, M. Berger et al., Plasma Sources Sci. Technol.
18, 025003 (2009).
264
265
M. Bacal et al., Rev. Sci. Instrum. 85, 02B103 (2014).
103.49.147.5 On: Wed, 21 Oct 2015 10:16:44

1 4921298 PDF

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

1 4921298 PDF

Caricato da

Copyright:

Formati disponibili

Negative hydrogen ion production mechanisms

M. Bacal and M. Wada

APPLIED PHYSICS REVIEWS 2, 021305 (2015)

APPLIED PHYSICS REVIEWS

Negative hydrogen ion production mechanisms

UPMC, LPP, Ecole Polytechnique, UMR CNRS 7648, Palaiseau, France

IV. SURFACE PRODUCTION MECHANISM . . . . .

A. Need of D2 negative ion beams for magnetic

This paper discusses the negative hydrogen/deuterium

C 2015 AIP Publishing LLC

M. Bacal and M. Wada

factor of Q 10 and should produce a self-sustaining fusion

FIG. 1. Neutralization efficiency of positive and negative ions versus their

Appl. Phys. Rev. 2, 021305 (2015)

optimum thickness remains acceptable at higher kinetic

B. Need of H2 negative ion beams for high energy

The use of negative ion sources began in the 1950s, e.g.,

M. Bacal and M. Wada

mechanisms. In the meantime, the embodiment of these

Appl. Phys. Rev. 2, 021305 (2015)

source. However, the negative ion currents were in the tens

II. THE VOLUME PRODUCTION MECHANISM

A review by Prelec and Sluyters21 describes the early

M. Bacal and M. Wada

Appl. Phys. Rev. 2, 021305 (2015)

The theoretical work on DA started in the 1960 s. The

or to the vibrational excitation of the molecule (denoted by

FIG. 3. Dissociative attachment cross section for various vibrational states

In 1979, theoretical26 and experimental27 work showed

FIG. 4. Dissociative attachment cross section for various rotational states of

M. Bacal and M. Wada

FIG. 5. Internal state dependence of threshold dissociative attachment cross

Appl. Phys. Rev. 2, 021305 (2015)

These rate coefficients can be as high as 108 cm3 s1 when

FIG. 6. Rate coefficients of dissociative electron attachment to all possible J

FIG. 7. Dissociative attachment cross sections calculated by Horacek et al.

M. Bacal and M. Wada

FIG. 8. Dissociative attachment cross sections calculated by Horacek et al.

Appl. Phys. Rev. 2, 021305 (2015)

H2 B1 Ru ; C1 Pu ! H2 X1 Rg ; v00 h:

This process is denoted by E-V singlet excitation.

M. Bacal and M. Wada

ions in a hydrogen discharge. The effective cross section for

Appl. Phys. Rev. 2, 021305 (2015)

FIG. 10. Vibrational populations as a function of vibrational energy for

experimental apparatus has been described in detail by

M. Bacal and M. Wada

Appl. Phys. Rev. 2, 021305 (2015)

Vibrational level distributions of hydrogen molecules

found that Al and Cu liners produced higher extracted H

M. Bacal and M. Wada

corresponding filaments, and also covered by caesium, as a

Figures 16 and 17 show the dissociative attachment

Appl. Phys. Rev. 2, 021305 (2015)

FIG. 16. Dissociative attachment cross sections for various rotationless

and Wadehra26,67 for various vibrational and rotational states

M. Bacal and M. Wada

Appl. Phys. Rev. 2, 021305 (2015)

FIG. 18. The energy dependence of the

isotope effect have been reported by Mundel, Berman, and

In 19631965 at Berkeley, the hot-cathode reflex-type

M. Bacal and M. Wada

ratio of 1 from an extraction area similar to positive-ion