The physics of tokamak start-up

D. Mueller

Citation: Phys. Plasmas 20, 058101 (2013); doi: 10.1063/1.4804416

View online: http://dx.doi.org/10.1063/1.4804416
View Table of Contents: http://pop.aip.org/resource/1/PHPAEN/v20/i5
Published by the American Institute of Physics.

Additional information on Phys. Plasmas

Journal Homepage: http://pop.aip.org/
Journal Information: http://pop.aip.org/about/about_the_journal
Top downloads: http://pop.aip.org/features/most_downloaded
Information for Authors: http://pop.aip.org/authors

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
 PHYSICS OF PLASMAS 20, 058101 (2013)

The physics of tokamak start-upa)

D. Muellerb)
Princeton Plasma Physics Laboratory, P.O. Box 451 Princeton, New Jersey 08543, USA

(Received 12 December 2012; accepted 14 March 2013; published online 10 May 2013)
Tokamak start-up on present-day devices usually relies on inductively induced voltage from a central
solenoid. In some cases, inductive startup is assisted with auxiliary power from electron cyclotron
radio frequency heating. International Thermonuclear Experimental Reactor, the National Spherical
Torus Experiment Upgrade and JT60, now under construction, will make use of the understanding
gained from present-day devices to ensure successful start-up. Design of a spherical tokamak (ST)
with DT capability for nuclear component testing would require an alternative to a central solenoid
because the small central column in an ST has insufficient space to provide shielding for the
insulators in the solenoid. Alternative start-up techniques such as induction using outer poloidal field
coils, electron Bernstein wave start-up, coaxial helicity injection, and point source helicity injection
have been used with success, but require demonstration of scaling to higher plasma current. V C 2013

AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4804416]

I. INTRODUCTION supplies, assisted by a resistive voltage in the coil and the

external circuit. Vcoil ¼ Vps  IcoilRcoilþ, where Vps denotes
This paper describes the physics of the start-up phase of
the applied power supply voltage, Vcoil is the voltage on the
tokamak operation. Recent experience on the start-up of the
coil, Icoil is the current in the coil, and Rcoilþ is the resist-
Experimental Advanced Superconducting Tokamak (EAST),1
ance of the coil, leads and any additional resistance that is
the Korean Superconducting Tokamak Advanced Research
inserted into the circuit. The one-turn loop voltage applied
(KSTAR),2 and start-up experiments in the Joint European
is given by Vloop ¼ VcoilM/L, where M is the mutual induct-
Torus (JET)3 with the International Thermonuclear
ance between the coil and the plasma and L is the induct-
Experimental Reactor (ITER)-like-wall (ILW)4 provides a
ance of the coil. The electric field at the radius R is
perspective for planning for the start-up of the National
E ¼ Vloop/2pR.
Spherical Torus Experiment Upgrade (NSTX-U),5 the Japan
Torus 60-Super Advanced (JT-60SA),6 and the ITER,7 which A. Breakdown and avalanche
will be carried out in the coming years. These devices will
rely upon a central solenoid for start-up. However for long Some free electrons are almost always present in a toka-
pulse spherical tokamaks (STs), non-central solenoidal start- mak chamber, but can be supplemented by radiation, heated
up is very important and it is essential for an ST reactor. The filaments, or RF waves. These are accelerated by the electric
leading techniques for start-up without a central solenoid in an field. If an electron gains over 13.6 eV before suffering a col-
ST include outer poloidal field coil start-up8,9 Electron lision with a neutral atom, it can ionize the neutral atom and
Bernstein Wave start-up10 and helicity injection form Coaxial leave two electrons, which can be accelerated by the electric
Helicity Injection (CHI)11 or from local helicity injection12 by field and produce more subsequent electrons. Figure 1 shows
means of plasma guns. the ionization cross-section of neutral hydrogen by electrons
as a function of electron energy from Gryzinski’s classical
II. INDUCTIVE START-UP model.13 Note the cross-section vanishes below 13.6 eV and
peaks at about 50 eV and falls at higher energy. This process
In present day tokamaks, the main technique to initiate
is known as the Townsend avalanche and is named after
breakdown and drive a toroidal current is use of a central
John Sealy Townsend.14 If an electron produces a electrons
solenoid that supplies magnetic flux and induces a toroidal
per meter, then dne ¼ a ne dx, where ne is the density of elec-
electric field. Central solenoid start-up can be divided into
trons and x is the distance along the electric field direction.15
three phases, the breakdown or avalanche phase, the impu-
An exponential growth in ne occurs ne ¼ ne(0) eax, where a is
rity burn-through phase, and the controlled plasma current
called the first Townsend coefficient. The Paschen curve that
ramp-up, which will now be described. Typically, before
describes the breakdown voltage of a gas between parallel
start-up, hydrogen or deuterium gas is injected into the vac-
plates for hydrogen is shown in Figure 2.16 In order to mini-
uum vessel and the solenoid is precharged with a current in
mize the loop voltage required and thereby reduce the hard-
the desired direction of the plasma current. The solenoid
ware demands, there is an optimal product of pressure times
current is then driven toward zero by the action of power
distance of a few Torr-cm for breakdown in hydrogen. For a
tokamak, the voltage is Vloop, the distance is 2pR, and the
a)
Paper UT3 1, Bull. Am. Phys. Soc. 57, 340 (2012). pressure, p, can be chosen to be near the minimum in the
b)
Invited speaker. Paschen curve to minimize hardware demands. The first

1070-664X/2013/20(5)/058101/10/$30.00 20, 058101-1 C 2013 AIP Publishing LLC

V

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-2 D. Mueller Phys. Plasmas 20, 058101 (2013)

FIG. 3. The number of new electrons per unit length of path for an electron in
a gas, the first Townsend coefficient, a is not a simple function E/p, but a/p is.

proceed. If the electrons are lost faster than the avalanche

proceeds the avalanche will fail.

B. Impact of error fields

We consider electron losses parallel and perpendicular
FIG. 1. The neutral hydrogen total ionization cross-section versus electron to the magnetic field separately. Under the influence of the
energy. Note that r vanishes below 13.6 eV peaks near 30 eV and then induced toroidal electric field, electrons drift along the mag-
decreases. netic field that is predominantly in the toroidal direction. The
electrons are subject transverse drifts due to the curvature of
the field, its radial gradient and perpendicular fields gener-
ated by the poloidal field system and by eddy currents
induced in surrounding conducting structures. Although the
startup phase of each tokamak is designed to produce a local-
ized, transient null in the poloidal field at the desired location
of breakdown at the time of breakdown, in practice, there are
always transverse fields surrounding the field null which
evolve in time, and these generally dominate the transverse
drifts of the electrons during startup so that after many tran-
sits, the electrons will impinge on the wall. For an average
stray field hdBi, the connection length to the wall, L is about
h•BT/hdBi where BT is the toroidal field and h is the typical
transverse distance to the wall of the device. For NSTX,
hdBi is 2.5 to 5 G so L is about 3000 m, much longer than
FIG. 2. The Paschen curve of breakdown voltage, V, between parallel plates the 200 m ionization length under typical inductive startup
separated by a distance d at pressure p for hydrogen. Note that there is a conditions.
minimum pd for which breakdown occurs and above that minimum, V During the avalanche phase, the average electron drift
increases approximately linearly such that for fixed separation E/p is approx-
imately constant. velocity vde parallel to the field is approximately 35 E/p (m/s)
so the time for electrons to drift to the wall is about 6 ms. For
ions, vdi  0.9E/p (m/s) and the time to drift to the wall is
Townsend coefficient, a, is not a simple function of E/p, but 150 ms, so secondary emission is unimportant during the
a/p is as is shown in Figure 3. As an example, for NSTX, avalanche. Lloyd17 estimates the time to complete the ava-
p  5  105 Torr and Vloop  2 V/turn, a  102/m, so the lanche process to be 41/vde (a-L1), which is 7 ms for NSTX
path length for which electrons must be confined before parameters.
being lost must be >100 m, i.e., many toroidal transits. If the For E/p > 5  103 Vm1Torr1, the electron temperature,
pressure is too high, the electrons will not gain enough Te, is high enough that thermal ionization is important. Energy
energy to ionize the neutrals before elastic scattering changes loss of the electrons in the ionization process limits Te to below
their direction so that the electric field slows them. If the 10 eV until ionization of the initial gas is nearly complete18
pressure is too low, the density of neutrals will not be suffi- so vD  4 to 40 m/s and the loss time is 25 to 250 ms greater
cient to provide electrons for the avalanche to proceed. than the avalanche time for NSTX. Over a wide range of devi-
Typically, if the pressure is within about a factor of 2 from ces, Vloop ¼ 2 to 30 V/turn, E ¼ 0.3 to 2 V/m, p ¼ 105 to
the optimum (in either direction) value, the avalanche will 104 Torr, and E/p ¼ 0.4 to 3  104 Vm1 Torr1. For JET,

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-3 D. Mueller Phys. Plasmas 20, 058101 (2013)

Tanga found for VloopXBT/hdBi > 103 V/m that the error various application techniques reduces oxygen impurities.23
fields were small enough and the loop voltage was high Lithium coatings can reduce the influx of C, O, and H/D.24
enough that the avalanche could proceed.19 The avalanche The use of metal walls can reduce the source of low-Z
proceeds until electron-ion collisions dominate the process impurities compared to graphite walls.25 Auxiliary heating
compared to electron-neutral collisions. The electron-ion and can be used to increase the power available to burn through
electron-neutral collision rates are equal when ne  0.1 n0. the low-Z radiation.
The current density, j ¼ c n0 e vde, where c is the H or D ioni- Recent modeling of the avalanche and burn-through
zation fraction is 15 to 40 kA/m2, which corresponds to a phases by H-T Kim26 of JET start-up agrees well with
plasma current, Ip, of 5 to 10 kA for NSTX and about 20 kA experiment. This model uses deuterium confinement time
for JET. For Ip ¼ 10 kA and a plasma minor radius of 0.5 m, sD from 1/sD ¼ 1/sD,|| þ 1/sD,? where sD,|| and sD,? are the
the poloidal field at the edge of the plasma is about 40 G, confinement times due to parallel and perpendicular losses.
comparable to the stray fields. At the end of the avalanche The parallel distance traveled before loss, L(t), is a function
phase, c ¼ 0.5, Coulomb collisions dominate but until ioniza- of time since as the Ip(t) increases the plasma’s poloidal
tion is nearly complete, Te is limited to below 10 eV. field becomes larger than the stray poloidal field. Kim
uses L(t) ¼ 0.25•a(t)•(BT/ h dBz(t) i )•exp(Ip(t)/Iref) with Iref
¼ 100 kA for JET. The confinement time due to parallel loss
C. Burn-through
sD,|| ¼ L(t)/Cs, where Cs is the sound speed ((Te þ Ti)/mD)1/2.
The burn-through phase begins at the end of the ava- The confinement time due to perpendicular losses sD,?
lanche. In this phase, low-Z impurities, usually originating ¼ a(t)/vBohm(t) where vBohm(t) ¼ 2DBohm(t)/a(t) and DBohm(t)
from the walls surrounding the plasma column, radiate and ¼ Te(eV)/16BT. A dynamic recycling coefficient is used for
can limit the temperature and the current ramp-rate so that deuterium while physical sputtering and a simple chemical
the discharge fails. The radiated power density, PRad, is given sputtering yield are used for C and O. This self-consistent
by neRnZf(Z,Te), where nZ is the impurity density, f(Z,Te) is model matches the experimental time history of the start-up
the cooling rate from impurities,20 and the sum is over the well, particularly for Ip, radiated power, carbon impurity
impurity species. Figure 4 shows the steady state cooling radiation emission, and Te.
rate due to impurity radiation from Be, C, and O as a func- Experimental results from JET with the ITER-like
tion of Te. Note that the cooling rate for Be is 10 times less wall, ILW, permit comparison of start-up conditions with
than for C or O. Also it is important to note that the peak graphite walls compared to the new Be and W surfaces.27
cooling occurs at about 8 and 20 eV for C and O, respec- The results indicate that the density behavior is different
tively. The power available to support the radiation is limited for the ILW and the graphite wall at the time of burn-
by the power supplies to E2/g at low Te and by j2g at high Te, through. For the ILW, the density scales linearly with the
where g is the plasma resistivity. There must be some power prefill gas pressure, whereas for the graphite wall, the
left over to increase Ip and heat the plasma or the discharge density varies with the prefill pressure, but with some addi-
will cool and collapse. The lower sputtering yield for high-Z tional amount due to recycling from the carbon wall. Also,
materials at low plasma temperature make them less impor- the radiated power at the time of burn-through is a steep
tant at start-up. In order to facilitate burn-through, various function of density for the carbon wall but weakly depend-
techniques have been employed. Wall conditioning can ent on density for the ILW. The latter point is likely due to
reduce the influx of low-Z materials. High temperature bake- the much reduced radiative cooling from Be compared to C
out removes hydrocarbons and water from graphite.21 Higher as seen in Fig. 4 as well as chemical sputtering of C from
surface temperatures accelerate this removal and tempera- the graphite wall. Furthermore, on JET with the ILW, there
tures of at least 300  C are usually required for effective were no failures of the start-up during the burn-through
bakeout. Helium Glow Discharge Cleaning (HeGDC) phase or failures due to deconditioning events, such as dis-
removes hydrogen and deuterium and water from the graph- ruptions or excessive gas puffing on the previous shot,
ite surface.22 Boronization or other surface coatings using unlike with the graphite wall.

FIG. 4. The cooling rate due to impurity

radiation, assuming coronal equilibrium
is plotted as a function of electron tem-
perature peaks below 20 eV for low Z
impurities.

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-4 D. Mueller Phys. Plasmas 20, 058101 (2013)

D. Additional requirement for tokamaks with where nD, n0, and ne are the deuterium, neutral atom, and
superconducting coils electron density, respectively, S is the ionization rate, sp is
the particle confinement time, and (Vn/Vp) is a factor to take
Fully superconducting tokamaks have limited loop volt-
into account the fraction of the plasma volume that is acces-
age due to power supply cost, eddy-current heating in the
sible to neutrals. The 0-D model handles impurities by
coils and the need to limit induced currents in the cryostat
assuming they are a fixed fraction of nD and uses a deuterium
and thick vacuum vessel. For ITER, the design value for the
recycling coefficient of R ¼ 1.01, that is, for each 100 deuter-
maximum electric field at start-up, E  0.3 V/m, is at the
ons escaping the plasma to the walls, 101 will return to the
lower limit of successful breakdown observed with purely in-
plasma as a result of dislodging bound deuterium from the
ductive start-up in existing tokamaks. For example in EAST,
surface. The results of this modeling indicate that burn-
the maximum electric field that can be applied using the
through with 2% Be minority should be possible in ITER for
power supplies alone is about 0.2 V/m, below the value
low fill densities of 1.5  1017/m3 (2  106 Torr) and a post
needed for successful inductive start-up. Both EAST and
avalanche density of <1.5  1018/m3, However, for 5% Be
KSTAR employ circuits that allow resistors to be switched
or higher fill pressure, failure is likely. If 2 MW of ECRH is
in to each of the poloidal field coil circuits for a short time,
used, 5% Be with a post avalanche density of 5  1018/m3
in effect raising Rcoil discussed in the beginning of Sec. II to
can be successful, but not with 2% C. At the same density
produce a higher voltage for breakdown after the coils are
and 5% C, 5 MW of ECRH is required for robust start-up.
charged to their initial magnetization states. Note that the
ITER has plans for several MW of ECRH power and use of
voltage drop across the resistors increases the applied volt-
some of that during start-up should be adequate to ensure
age in the direction to decrease the magnitude of the coil cur-
success.
rents. Both of the operating fully superconducting tokamaks
For normal aspect ratio tokamaks with toroidal fields in
have less inductive power to heat the plasma and ramp the
the range of about 1 to 4 T, fundamental O-mode (E||B) and
plasma current compared to conventional tokamaks due to
2nd harmonic X-mode (E?B) can access the plasma from
their lower coil power supply voltage capabilities. The use of
low-field-side launch at the appropriate densities as has been
electron cyclotron radiofrequency heating (ECRH) has been
reported by Refs. 17, 29, and 30 and references cited therein.
demonstrated on a variety of tokamaks to lower the electric
The time evolution of ECRH assisted start-up is shown in
field required for breakdown by about a factor of about two.17
Figures 5 and 6, which are from 2nd harmonic X-Mode
Furthermore, ECRH can provide power to the plasma during
injection on DIII-D.31 The camera images in Figure 5 show
burn-through when the plasma current is low and other heat-
CIII emission during the discharge, which forms just inside
ing methods, particularly inductive and energetic neutral
of the 2nd harmonic resonance layer (a), expands radially (b)
beam heating are inefficient, and when ion cyclotron radiofre-
and (c), driven by the EXB drift, fills the vessel as the loop
quency heating can be difficult to apply due to coupling.
voltage is applied and the plasma current increases to form
Lloyd has used a zero-dimensional model to assess the
closed flux surfaces (d), the plasma limits on the inner wall
need for additional power during start-up for ITER.28 The
(e) and finally is moved to its preprogrammed position,
electron power balance in this model is given by Eq. (1). POH
limited on the low field side after 20 ms (f). Figure 6 shows
and PRF are the ohmic and RF input power, (PDion þ PDrad),
the progression of the plasma during ECRH start-up on
3d DIII-D.32 As PECH is increased, the first phase is collisionless
ðne KTe Þ ¼ POH þ PRF  ðPDion þ PDrad Þ  Pei
2 dt heating where the electrons do not gain sufficient energy to
 Pecon  Pbrem ionize the gas, as the power is increased the avalanche occurs
X and the plasma expands with low Ip. When the toroidal elec-
 ðPion þ Pline þ PRRE þ PDRE Þ (1)
I
tric field is applied, Ip increases and at about 20 kA, closed
flux surfaces form as evidenced by the rapid increase of
is the power lost to ionization and radiation from deuterium, TECE, the electron temperature measured by electron cyclo-
Pei is the power flow from the electrons to the ions, Pecon is tron emission, at 3 ms. Burn-through follows with addi-
the heat loss from the electrons due to confinement, Pbrem is tional heating from ECRH.
the radiated power due to bremsstrahlung, and the sum over
the impurity ions of (Pion þ Pline þ PRRE þ PDRE) is the power
E. Examples from EAST and KSTAR
lost due to ionization, line radiation, radiative recombination,
and dielectronic recombination of the impurities. The ion In general, the start-up phase of discharges gets attention
power balance is given by only when there is a failure. Sometimes the failure can be
3d traced rather quickly to some hardware issue but at other
ðni KTi Þ ¼ Pei  PCX  Picon ; (2) times, the cause is not obvious. A couple examples from the
2 dt
start-up of new devices are instructive. During the initial
where PCX is the loss due to charge-exchange and Picon is the attempts to start-up the EAST tokamak, there were repeated
heat loss from the ions due to confinement. The particle bal- failures with the plasma current never exceeding 35 kA and
ance is given by the discharge ending at 70 to 100 ms. At this early stage of
EAST’s operation, many of the diagnostics were unavailable.
dnD Vn nD The images of the plasma during the attempted start-up were
¼ Sn0 ne  ; (3)
dt Vp sp difficult to interpret because the interior surfaces of EAST

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-5 D. Mueller Phys. Plasmas 20, 058101 (2013)

FIG. 5. Fast-framing camera images show

the CIII emission at various times during the
start-up of DIII-D assisted by 2nd harmonic
X-Mode ECRH. The caption below each
frame shows the time in ms, Ip in kA, and
the loop voltage at that time. Reprinted with
permission from G. L. Jackson et al., Phys.
Plasmas 17, 056116 (2010). Copyright
2010, American Institute of Physics.

FIG. 7. Fast camera image of plasma in EAST just after breakdown. The
FIG. 6. Phases of plasma evolution with ECRH assisted plasma start-up.
bright plasma and metallic surfaces make interpretation of the plasma loca-
The first frame shows the ECH power (red), applied toroidal electric field
tion uncertain. Reprinted with permission from J. A. Leuer et al., Fusion
(green), and Ip (black) versus time. The 2nd and 3rd frames show the Da
Sci. Technol. 57, 48 (2010). Copyright 2010 The American Nuclear Society.
emission and intensity of the visible bremsstrahlung, respectively. The 4th
frame has the line average density of vertical views at 1.48 m (black),
1.94 m (red), and 2.1 m (green). The bottom frame is the electron tempera-
ture measured by electron cyclotron emission. Reprinted with permission later in time. Equation (4) provides an approximate value the
from G.L. Jackson et al., Fusion Sci. Technol. 57, 27 (2010). Copyright vertical field required for radial position control.
2010 The American Nuclear Society.
   
l Ip 8R0 li 3
Bz ¼  0 ln þ þ bp 
4pR0 a 2 ð 2
2 (4)
were mostly stainless steel, so there were multiple reflections 2
where bp  0:1 and li ¼ 2 2 Bp dV  1:
and no obvious single plasma contact point could be seen as l0 RI
the CCD image in Figure 7 from Ref. 33 illustrates. The
obvious candidate for the problem was failure to burn- When the breakdown resistor time was shortened to 50 ms,
through low Z impurities. The first attempts at breakdown the discharge survived and ramped up to about 150 kA on
used an insertable breakdown resistor in all the poloidal field the first attempt. The plasma current did decrease at
coil circuits for 100 ms to provide additional voltage from about 50 ms before the successful ramp-up as can be seen in
the IR drop in the coils that were precharged to positive cur- Figure 8 that shows the first successful EAST plasmas.
rent. However, the coil currents that were observed did not Further shortening of the resistor time did not result in suc-
match that from modeling done prior to operation. In particu- cessful plasmas. This experience emphasizes the need for
lar, the outer poloidal field (PF) coils which provide the verti- good modeling of the plasma circuit and power supplies.
cal field for plasma radial position control were more positive Since then, the power supplies on EAST have been upgraded
than the model indicated and differed further from the model to higher voltages for better control.

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-6 D. Mueller Phys. Plasmas 20, 058101 (2013)

the ILW,27 this increase of ne with Ip in the absence of addi-

tional fueling is not observed. The density behavior during
start-up for typical discharges in JET with the ILW is differ-
ent than that for JET with its graphite wall. For the graphite
wall, without gas fueling, the density rises with Ip, while for
the ILW, the density falls to very low levels unless additional
fueling is provided by means of gas puffing. Gas puffing
with the ILW can increase the density to values above those
typical of the graphite wall without gas puffing.

G. Current density during ramp-up

The current density profile evolution during ramp-up is
influenced in part by Te that in turn is determined partly by
FIG. 8. The first 3 successful discharges on EAST. Note that previous the density evolution. As Te rises, the rate at which the
attempts had all resulted in Ip falling from about 35 kA to zero before 0.1 s. plasma current is able to penetrate the plasma decreases and
Reprinted with permission from J. A. Leuer et al., Fusion Sci. Technol. 57,
48 (2010). Copyright 2010 The American Nuclear Society.
this can result in tearing mode instabilities during plasma
ramp-up37 at low li. Growing the plasma’s aperture size as Ip
is increased to keep the q(a)  constant during much of the
As the following example from KSTAR illustrates, it is
ramp-up allows the current to penetrate and can avoid insta-
not enough to
bilities.19 Figure 9 shows a comparison of NSTX discharges
Bz grown at about constant q with a large bore throughout the
Br ¼ n ½Z  Z0 ; (5) current ramp. The large bore plasma has lower li that indi-
R0
cates a broader current profile. The low loop voltage avail-
n ¼ ðR=Bz Þð@Bz =@RÞ ) 0 < n < 3=2; (6) able on ITER necessitates a low Ip ramp rate and allows
current to penetrate so that if a constant growth technique
supply the proper vertical field at a single location. The were to be used, early sawteeth would be facilitated.
shape of the field must provide stability against motion. The Furthermore, the discharge would be in contact with the lim-
field index, n, given in Eq. (6) must be greater than zero to iter up to about 15 MA before diverting and this could cause
ensure vertical stability and less that 3/2 to ensure radial excessive heating of the limiter. Experiments carried out on
stability. KSTAR has ferromagnetic material in its coil DIIII-D using a large bore scenario demonstrated less heat-
jackets,34 which was chosen for its thermal expansion com- ing of the limiter and li closer to the projected ITER target.38
patibility. During the first year of KSTAR operation, there
was a significant fraction of the discharges that would move
onto the inner wall and terminate early and it was not possi-
ble to produce a discharge without the use of ECRH assist.
An analysis of the field pattern around the time of breakdown
by Kim,35 which included the effects of the ferromagnetic
material, indicated that the vertical field near the central col-
umn was larger than was indicated by calculations that did
not take the ferromagnetic material effect into account.
There was a region inside some major radius where n > 3/2
and the plasma was radially unstable. This explained the
start-up failures. If the initial current channel was formed at
too small R, then the plasma was unstable. Without ECRH,
the initial plasma had lower current and lower b, so it
was formed at a smaller R where n > 3/2. When the initial
field coil currents were adjusted to compensate for the ferro-
magnetic effect, it was possible to initiate plasmas without
the use of ECRH and the random start-up failures largely
disappeared.

F. Electron density during ramp-up

The density during start-up depends upon recycling of
gas from the wall. It is a common feature in graphite walls
that the density increases with Ip during start-up when the FIG. 9. Comparison of evolution of a discharge initiated with a large-bore
(red) with one grown from a smaller aperture (blue) such that q95 reaches its
plasma is limited on the graphite surface even without addi- flattop value early and approximately constant after 0.05 s. Note that the fac-
tional gas fueling. In devices with all metal or Be surfaces tor of two difference in internal inductance (li) at the start of plasma current
such as Alcator-C-MOD,36 ASDEX-U,25 and JET with flattop.

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-7 D. Mueller Phys. Plasmas 20, 058101 (2013)

H. Plasma start-up without a solenoid

Use of a central solenoid both works reliably and is well
understood, but there are reasons to consider other start-up
strategies: In particular, elimination of the central solenoid is
required to achieve low aspect ratio at small device size for
fusion nuclear applications. If non-inductive current drive
could support a steady state reactor, an alternative start-up
technique would allow elimination of the solenoid, and with
a reduced size or no central solenoid, the cost and size of a
conventional tokamak reactor could be reduced. Inductive
techniques that do not rely upon a central solenoid must
address all the same physics issues, but the emphasis may
differ. For example, use of an iron core in place if an air-
cored solenoid can avoid engineering issues caused by neu-
tron damage to insulation in the central solenoid of a reactor,
but requires engineering design of the system to be carried
out.39 The outer poloidal field coils can be used to provide FIG. 10. Results of ray-tracing done for EBW on MAST. Note that because
both loop voltage and flux to start-up the plasma and has the incoming O-Mode ECH beam is launched from below the midplane
been demonstrated on JT-608 and DIII-D9. Providing a good towards the polarizing reflector at the midplane, the outgoing X-Mode ECH
is mostly above the device’s midplane. This vertical imbalance is exploited
field null and a stabilizing poloidal field while providing flux to produce co-current drive as described in the text.
by ramping the coils to finally provide a diverted plasma
shape at high plasma current capable of being sustained by
above or below the midplane changes. MAST has exploited
other means must address all the physics issues of inductive
that by using the radial field to shift the plasma’s midplane
breakdown through controlled plasma current ramp-up.40
above the machine’s midplane such that the EBW mostly
propagates below the plasma’s midplane and thus drives
III. START-UP USING RF WAVES
mostly co-directed current before closed flux forms and then
Lower Hybrid Current Drive (LHCD) start-up to 100 kA shifting the plasma down when closed flux forms and the
was demonstrated on PLT,41 and ECRH was used to provide direction of n|| changes. The result of making such a well-
start-up to low current levels on DIII-D,42 TS2,43 and orchestrated timing of the radial field to produce a vertical
LATE.44 For the ST, however, Electron-Bernstein Wave shift when the formation of closed flux surfaces appears is
(EBW) seems most promising.45 This technique has pro- illustrated in Figure 11 for two different vertical field pro-
duced Ip of 33 kA with only 100 kW of ECRH on MAST.10,46 grams. Extrapolation of these results from MAST indicates
The EBW is an electrostatic wave that can exist only in a start-up current per launched power of 0.33 MA/MW.
plasma so it cannot be launched directly by an antenna out-
side the plasma, but it can be produced by mode-conversion IV. START-UP USING HELICITY INJECTION
at the upper hybrid resonance (UHR) layer of X-Mode
The
Ð concept of magnetic helicity, which is given by
ECRH waves launched from the high-field side (HFS).47 The ~ •Bdv
~ where A ~ is the vector potential, B
~ is the mag-
K¼ A
UHR frequency is xUH ¼ ðx2p þ x2ce Þ1=2 where xce ¼ eB=me
netic field, and v is the plasma volume can be used to
and xp ¼ ðne e2 =e0 me Þ1=2 . Because access to the HFS is diffi-
describe current drive in a tokamak.48 Plasma start-up by
cult in an ST, the MAST solution is to launch O-Mode
helicity injection as well as coupling to inductive sustain-
ECRH at 28 GHz from the low field side. The O-Mode wave
ment has been successfully demonstrated by both transient
is not strongly damped below the density cutoff of 1  1019/
coaxial helicity injection (CHI)49,50 and by point source hel-
m3. A grooved polarizing mirror cut into the central column
icity injection from plasma guns.51
converts the O-Mode to X-Mode. The geometry and results
of ray-tracing modeling for MAST are shown in Figure 10.
A. Helicity injection from plasma guns
The O-Mode is launched from below the midplane so the
beam reflected at the midplane is largely above the midplane Plasma guns have been employed on PEGASUS to
as it propagates outward in the plasma. As the outgoing X- inject helicity and provide start-up plasmas that can be
Mode wave approaches the UHR, its index of refraction, n, coupled to inductive ramp-up.52 The gun location is flexible,
increases, it slows, is mode converted to an EBW reflected and the guns could be withdrawn after start-up and do not
perpendicular to the UHR. The sign of the refractive index in provide an obvious impurity source to the plasma. All these
the direction parallel to the magnetic field, n||, is determined are potential advantages of the use of the guns for start-up.
by the direction of the projection of wave vector k along the The scaling of Ip from helicity injection is not simple since it
poloidal field and it is the sign of n|| that determines if the arises from considering two limits.47 One is the relaxation
EBW will accelerate electrons parallel or anti-parallel to the limit that arises from consideration of the Taylor minimum
toroidal field when it is absorbed at the electron cyclotron energy state. The scalar representing the average inverse
resonance. Because the poloidal field changes sign when scale length of the helical magnetic field is kT ¼ l0 Ip =WT
closed flux forms, the direction of the EBW current driven and kinj ¼ l0 Iinj =Wedge in the plasma volume and in the

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-8 D. Mueller Phys. Plasmas 20, 058101 (2013)

FIG. 12. The main components of the NSTX CHI system discussed in the
text. The fast, color camera images on the right show the plasma growing
FIG. 11. The radial field in MAST is used to move the plasma centroid up
into the vessel in time. The green color is due to Li emission.
before closed flux surfaces are formed so that so the majority of the X-Mode
ECH and EBW are below the plasma midplane and produces co-current
drive. Moving the plasma down as Ip increases to form closed flux puts the
EBW above the plasma midplane and produces co-current drive. The red- is applied from a variable (5 to 50 mF) capacitor bank across
dashed curve with constant vertical field demonstrates that the current is not the insulating gap. Breakdown proceeds, again via a
driven by flux from the vertical field. The blue curves indicate a case with
the vertical field increased as Ip is increased to maintain better position con- Townsend avalanche along the helical field connecting inner
trol. Reprinted with permission from V. F. Shevchenko et al., Nucl. Fusion and outer vessel. Since the toroidal field is much stronger
50, 022004 (2010). than the initial poloidal field, each field line wraps many
times around the major axis while connecting the inner and
injector regions, respectively, where WT ¼ BT Ap and outer electrodes so the toroidal current can be tens to hun-
Wedge ¼ 2pRedge wBz;edge . Bz;edge is due to both Ip and the vac- dreds of times the injector current (Iinj) between the electro-
uum field. Relaxation drives current from higher to lower k des. The direction of the JpolX BT is up into the vacuum
and this results in vessel. When Iinj > 2 winj2 / (l02d2ITF), where winj is the flux
  connecting the inner and outer vessels, d is the separation of
eAp Iinj ITF 1=2 the flux footprint, and ITF is the total toroidal current in the
Ip  fGeom ; (7)
2pRedge w center column,53 the plasma rapidly expands to fill the vessel
as can be seen by the three fast camera images at 1, 1.4, and
where 1 < fGeom < 3 is a factor that depends upon geometry. 2.5 ms. The capacitor voltage driving the injected current is
The Ip limit in Eq. (7) indicates Ip scales with injector prop- then removed by a fast acting crowbar switch. This forces
erties like the square root of the gun current and inversely as reconnection of the field lines and when Iinj ¼ 0, all the toroi-
the square root of the source width. A second scaling arises dal current is flowing on closed flux surfaces. Similar to
from the helicity input rate and implies that Ip driven by hel- Eq. (7), it can be shown, using the inverse scale lengths for
icity from the injector scales directly with the area of the gun the helical magnetic field and the equation for the injector
source and the gun bias voltage. Thus, the scaling of Ip will current that the CHI produced plasma current is directly pro-
depend upon the toroidal field, the gun impedance, and the portional to the injector flux that connects the lower divertor
geometry. Results from PEGASUS have demonstrated the plates.
ramp-up of an 80 kA plasma initiated by plasma guns to
150 kA inductively.50 IP  2wT winj =ðl20 d2 ITF Þ: (8)

Since it begins as an electrode discharge, potential difficul-

B. Coaxial helicity injection (CHI)
ties with CHI are that impurities from the electrode surfaces,
Implementation of transient CHI on NSTX is accom- low Te, or unsuitably high ne may render the CHI formed
plished through series of actions described below. The sketch plasma unsuitable for start-up of a tokamak discharge. These
in Figure 12 shows the important components for transient considerations have been addressed experimentally54,55 and
CHI. The vacuum vessel is separated electrically by insulat- as can be seen in Figure 13, discharges that are initiated with
ing gaps at the top and bottom of the machine. The toroidal transient CHI can be coupled to inductive ramp-up. In this
field is applied and the lower divertor coils are used to pro- example, the CHI initiated plasmas have Ip > 300 kA greater
duce a poloidal field that connects the inner and outer than those formed with inductive start-up using the same in-
vessels. Deuterium gas is injected at the bottom and 1.65 kV ductive flux.

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-9 D. Mueller Phys. Plasmas 20, 058101 (2013)

S. H. Park, K. H. Kim, J. Hong, S. H. Baek, M. K. Kim, T. G. Lee, S. I. Lee,

Y. S. Bae, H. Yonekawa, J. H. Choi, I. S. Hwang, Y. J. Kim, K. W. Cho, Y.
M. Park, J. Y. Kim, J. H. Lee, J. S. Bak, M. Kwon, G. S. Lee, J. G. Kwak,
H. S. Ahn, M. L. Walker, D. A. Humphreys, J. A. Leuer, A. Hyatt, G.
Jackson, D. Mueller, and D. P. Ivanov, Fusion Eng. Des. 84, 344 (2009).
3
P. H. Rebut, R. J. Bickerton, and B. E. Keen, Nucl. Fusion 25, 1011
(1985).
4
G. F. Matthews, P. Coad, H. Greuner, M. Hill, T. Hirai, J. Likonen, H.
Maier, M. Mayer, R. Neu, V. Philipps, R. Pitts, and V. Riccardo, J. Nucl.
Mater. 390, 934 (2009).
5
J. E. Menard, S. Gerhardt, M. Bell, J. Bialek, A. Brooks, J. Canik,
J. Chrzanowski, M. Denault, L. Dudek, D. A. Gates, N. Gorelenkov,
W. Guttenfelder, R. Hatcher, J. Hosea, R. Kaita, S. Kaye, C. Kessel,
E. Kolemen, H. Kugel, R. Maingi, M. Mardenfeld, D. Mueller, B. Nelson,
C. Neumeyer, M. Ono, E. Perry, R. Ramakrishnan, R. Raman, Y. Ren,
S. Sabbagh, M. Smith, V. Soukhanovskii, T. Stevenson, R. Strykowsky,
D. Stutman, G. Taylor, P. Titus, K. Tresemer, K. Tritz, M. Viola,
M. Williams, R. Woolley, H. Yuh, H. Zhang, Y. Zhai, A. Zolfaghari, and
NSTX Team, Nucl. Fusion 52, 083015 (2012).
6
N. Hosogane, JT-60SA Design Team, and Japan-Europe Satellite
Tokamak Working Group, Fusion Sci. Technol. 52, 375 (2007).
7
M. Shimada, D. J. Campbell, V. Mukhovatov, M. Fujiwara, N. Kirneva,
K. Lackner, M. Nagami, V. D. Pustovitov, N. Uckan, J. Wesley, N.
FIG. 13. The plasma current in kA for a discharge initiated with CHI and Asakura, A. E. Costley, A. J. H. Donne, E. J. Doyle, A. Fasoli, C.
further ramped with induction (blue) is compared to a reference inductive- Gormezano, Y. Gribov, O. Gruber, T. C. Hender, W. Houlberg, S. Ide, Y.
only discharge from the NSTX 10YR data base that reached 1 MA in a Kamada, A. Leonard, B. Lipschultz, A. Loarte, K. Miyamoto, V.
shorter time than other discharges (black). At 120 ms, the solenoid flux used Mukhovatov, T. H. Osborne, A. Polevoi, and A. C. C. Sips, Nucl. Fusion
by both discharges is the same. 47, S1 (2007).
8
M. Ushigome, S. Ide1, S. Itoh, E. Jotaki, O. Mitarai, S. Shiraiwa, T.
Suzuki, Y. Takase, S. Tanaka, T. Fujita, P. Gohil, Y. Kamada1, L. Lao, T.
V. SUMMARY Luce, Y. Miura, O. Naito, T. Ozeki, P. Politzer, Y. Sakamoto, and JT-60
Team, Nucl. Fusion 46, 207 (2006).
Inductive start-up is well understood. The Townsend 9
J. A. Leuer, G. Cunningham, D. Mueller, N. H. Brooks, N. W. Eidietis, D.
avalanche and impurity burn-through have been modeled self- A. Humphreys, A. W. Hyatt, G. L. Jackson, J. Lohr, P. A. Politzer, R. I.
consistently. With ECRH assist, inductive startup should serve to Pinsker, R. Prater, P. L. Taylor, M. L. Walker, R. V. Budny, D. A. Gates,
produce high current plasmas in ITER and JT-60SA. However, A. Nagy, S.-H. Hahn, Y.-K. Oh, S.-W. Yoon, J. H. Yu, M. Murakami, J.
M. Park, and A. C. Sontag, Nucl. Fusion 51, 063038 (2011).
future spherical tokamaks with nuclear capability will have little 10
V. F. Shevchenko1, M. R. O’Brien1, D. Taylor et al., Nucl. Fusion 50,
space for a central solenoid, which makes the usual inductive 022004 (2010).
11
technique problematic, although the use of an unshielded iron B. A. Nelson, T. R. Jarboe, A. K. Martin, D. J. Orvis, J. Xie, C. Zhang,
core could provide some initial current. Outer PF induction, and L. Zhou, Phys. Plasmas 2, 2337 (1995).
12
N. W. Eidietis, R. J. Fonck, G. D. Garstka, E. A. Unterberg, and G. R.
EBW, CHI, and point source helicity injection are possible alter- Winz, J. Fusion Energy 26, 43 (2007).
natives for start-up on ST reactors. Each of these techniques 13
M. Gryzinski, Phys. Rev. 138, A305, A322, A336 (1965).
14
demonstrated start-up to significant current. Experiments on pres- J. S. Townsend, Electricity in Gases (Clarendon Press, Oxford, 1915).
15
ent devices will provide scaling to higher plasma current and S. C. Brown, Introduction to Electrical Discharges in Gases (John Wiley
and Sons, 1966).
refine the choices available for future ST reactors. 16
F. Paschen, Ann. Phys. 273(5), 69 (1989); M. A. Lieberman and A. J.
Lichtenberg, Principles of Plasma Discharges and Materials Processing
(John Wiley & Sons Inc., Hoboken, NJ, 2005).
ACKNOWLEDGMENTS 17
B. Lloyd, G. L. Jackson, T. S. Taylor, E. A. Lazarus, T. C. Luce, and R.
Prater, Nucl. Fusion 31, 2031 (1991).
18
I would like to thank Gary Jackson, Peter deVries, R. Papoular, Nucl. Fusion 16, 37 (1976).
19
Vladimir Schevchenko, Gary Taylor, Michael Bell, Roger A. Tanga, P. R. Thomas, J. G. Cordey, J. P. Christiansen, S. Ejima, A.
Kellman, E. Lazzaro, P. J. Lomas, P. Morgan, F. Nave, P. Noll, and F. C.
Raman, Devon Battaglia, Walter Guttenfelder, David Gates, Schuller, in Tokamak Start-up, edited by H. Knoepfel (Plenum Press, NY,
and Jon Menard for valuable discussions and for sharing the 1985).
20
results of their research with me. This work was supported D. E. Post, R. V. Jensen, C. B. Tarter, W. H. Grasberger, and W. A.
by U.S. DOE Contract No. DE-AC02-09CH11466. Lokke, At. Data Nucl. Data Tables 20, 397 (1977).
21
A. E. Pontau and D. H. Morse, J. Nucl. Mater. 141–143, 124 (1986).
This manuscript has been authored by Princeton University 22
G. L. Jackson, T. S. Taylor, and P. L. Taylor, Nucl. Fusion 30, 2305
and collaborators under Contract No. DEAC0209CH11466 with (1990).
23
the U.S. Department of Energy. The publisher, by accepting the J. Winter, H. G. Esser, L. K€onen, V. Philipps, H. Reimer, J. v. Seggern, J.
Schl€uter, E. Vietzke, F. Waelbroeck, P. Wienhold, T. Banno, D. Ringer,
article for publication, acknowledges that the United States
and S. Veprek, J. Nucl. Mater. 162–164, 713 (1989).
Government retains a nonexclusive, paidup, irrevocable, world- 24
H. W. Kugel, M. G. Bell, J.-W. Ahn, J. P. Allain, R. Bell, J. Boedo, C.
wide license to publish or reproduce the published form of this Bush, D. Gates, T. Gray, S. Kaye, R. Kaita, B. LeBlanc, R. Maingi, R.
manuscript, allow others to do so, for United States Government Majeski, D. Mansfield, J. Menard, D. Mueller, M. Ono, S. Paul, R. Raman,
A. L. Roquemore, P. W. Ross, S. Sabbagh, H. Schneider, C. H. Skinner,
purposes. V. Soukhanovskii, T. Stevenson, J. Timberlake, W. R. Wampler, and L.
Zakharov, Phys. Plasmas 15, 056118 (2008).
1 25
B. Wan, Plasma Sci. Technol. 9, 125 (2007). R. L. Neu, IEEE Trans. Plasma Sci. 38(3), 454 (2010).
2 26
Y.-K. Oh, W. C. Kim, K. R. Park, M. K. Park, H. L. Yang, Y. S. Kim, Y. H.-T. Kim, W. Fundamenski, A. C. C. Sips, and EFDA-JET Contributors,
Chu, Y. O. Kim, J. G. Bak, E. N. Baang, S. W. Yoon, S. H. Hahn, H. J. Lee, Nucl. Fusion 52, 103016 (2012).

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions
058101-10 D. Mueller Phys. Plasmas 20, 058101 (2013)

27 39
P. C. de Vries et al., “Comparison of plasma breakdown with a carbon and D. A. Gates, C. Jun, I. Zatz, and A. Zolfaghari, Fusion Eng. Des. 86, 41
ITER-like wall,” Nucl. Fusion (submitted); P. C. de Vries, P. J. Lomas, F. (2011).
40
Maviglia, R. Albanese, I. Coffey, E. Joffrin, M. Lehnen, A. Manzanares, W. Choe, J. Kim, and M. Ono, Nucl. Fusion 45, 1463 (2005).
41
M. O’Mulane, I. Nunes, G. van Rooij, F. G. Rimini, M. F. Stamp, and F. Jobes, J. Stevens, R. Bell, S. Bernabei, A. Cavallo, T. K. Chu, S. Cohen,
JET-EFDA Contributors, in 25th IAEA, SanDiego, EXD4-2, 2012. B. Denne, P. Efthimion, E. Hinnov, W. Hooke, J. Hosea, E. Mazzucato, R.
28
B. Lloyd, P. G. Carolan, and C. D. Warric, Plasma Phys. Controlled McWilliams, R. Motley, S. Suckewer, G. Taylor, J. Timberlake, S. von
Fusion 38, 1627 (1996). Goeler, and R. Wilson, Phys. Rev. Lett. 52(12), 1005 (1984); C. F. F.
29
G. L. Jackson, J. S. deGrassie, C. P. Moeller, and R. Prater, Nucl. Fusion Karney, N. J. Fisch, and F. C. Jobes, Phys. Rev. A 32, 2554–2556 (1985);
47, 257 (2007). N. J. Fisch, Rev. Mod. Phys. 59, 175–234 (1987).
30 42
K. Kajiwara, Y. Ikeda, M. Seki, S. Moriyama, T. Oikawa, T. Fujii, and G. L. Jackson, D. A. Humphreys, A. W. Hyatt, J. M. Lohr, T. C. Luce, and
JT-60 Team, Nucl. Fusion 45, 694 (2005). J. H. Yu, Nucl. Fusion 51, 083015 (2011).
31 43
G. L. Jackson, P. A. Politzer, D. A. Humphreys, T. A. Casper, A. W. A. Ejiri, Y. Takase, H. Kasaharaet, T. Yamada, K. Hanada, K. N. Sato, H.
Hyatt, J. A. Leuer, J. Lohr, T. C. Luce, M. A. Van Zeeland, and J. H. Yu, Zushi, K. Nakamura, M. Sakamoto, H. Idei, M. Hasegawa, A. Iyomasa, N.
Phys. Plasmas 17, 056116 (2010). Imamura, K. Esaki, M. Kitaguchi, K. Sasaki, H. Hoshika, O. Mitarai, and
32
G. L. Jackson, M. E. Austin, J. S. de Grassie, A. W. Hyatt, J. M. Lohr, T. N. Nishino, Nucl. Fusion 46, 709 (2006).
44
C. Luce, R. Prater, and W. P. West, Fusion Sci. Technol. 57, 27 (2010). T. Yoshinaga, M. Uchida, H. Tanaka, and T. Maekawa, Phys. Rev. Lett.
33
J. A. Leuer, B. J. Xiao, D. A. Humphreys, M. L. Walker, A. W. Hyatt, G. 96, 125005 (2006).
45
L. Jackson, D. Mueller, B. G. Penaflor, D. A. Piglowski, R. D. Johnson, A. A. Montes and G. O. Ludwig, Plasma Phys. Controlled Fusion 28, 1765
S. Welander, Q. P. Yuan, H. Z. Wang, J. R. Luo, and EAST Team, Fusion (I986).
46
Sci. Technol. 57, 48 (2010). V. Shevchenko, in EC-15 Joint Workshop, Yosemite, CA, USA (2008).
34 47
H. Yonekawa, Y. O. Kim, H. J. Lee, S. W. Yoon, S. Hahn, K. S. Lee et al., H. P. Laqua, Plasma Phys. Controlled Fusion 49, R1 (2007).
48
IEEE Trans. Appl. Supercond. 19, 1573–1576 (2009). D. J. Battaglia, M. W. Bongard, R. J. Fonck, and A. J. Redd, Nucl. Fusion
35
J. Kim, S. W. Yoon, Y. M. Jeon, J. A. Leuer, N. W. Eidietis, D. Mueller, 51, 073029 (2011).
49
S. Park, Y. U. Nam, J. Chung, K. D. Lee, S. H. Hahn, Y. S. Bae, W. C. R. Raman, T. R. Jarboe, W. T. Hamp, A. J. Redd, B. A. Nelson, R. G.
Kim, Y. K. Oh, H. L. Yang, K. R. Park, H. K. Na, and KSTAR Team, O’Neill, P. E. Sieck, and R. J. Smith, Phys. Plasmas 14, 022504 (2007).
50
Nucl. Fusion 51, 083034 (2011). R. Raman, T. R. Jarboe1, D. Mueller, B. A. Nelson, M. G. Bell, R. Bell,
36
I. H. Hutchinson, R. Boivin, F. Bombarda, P. Bonoli, S. Fairfax, C. Fiore, D. Gates, S. Gerhardt, J. Hosea, R. Kaita, H. Kugel, B. LeBlanc,
J. Goetz, S. Golovato, R. Granetz, M. Greenwald, S. Horne, A. Hubbard, R. Maingi, R. Maqueda, J. Menard, M. Nagata, M. Ono, S. Paul,
J. Irby, B. LaBombard, B. Lipschultz, E. Marmar, G. McCracken, M. L. Roquemore, S. Sabbagh, V. Soukhanovskii, and G. Taylor, Nucl.
Porkolab, J. Rice, J. Snipes, Y. Takase, J. Terry, S. Wolfe, C. Christensen, Fusion 49, 065006 (2009).
51
D. Garnier, M. Graf, T. Hsu, T. Luke, M. May, A. Niemczewski, G. D. J. Battaglia, M. W. Bongard, R. J. Fonck, A. J. Redd, and A. C. Sontag,
Tinios, J. Schachter, and J. Urbahn, Phys. Plasmas 1, 1511 (1994). Phys. Rev. Lett. 102, 225003 (2009).
37 52
J. Wesson, R. D. Gill, M. Hugonet, J. A. Snipes, D. J. Ward, D. V. Bartlett, D. J. Battaglia, J. Fusion Energy 28, 140 (2009).
53
D. J. Campbell, P. A. Duperrex, A. W. Edwards, R. S. Granetz, N. A. O. T. R. Jarboe, Fusion Technol. 15, 7 (1989).
54
Gottardi, T. C. Hender, E. Lazzaro, P. J. Lomas, N. Lopes Cardozo, K. F. B. A. Nelson, T. R. Jarboe, D. Mueller, B. A. Nelson, M. G. Bell, M. Ono,
Mast, M. F. F. Nave, N. A. Salmon, P. Smeulders, P. R. Thomas, B. J. D. T. Bigelow, R. Kaita, B. LeBlanc, R. Maqueda, J. Menard, S. Paul, L.
Tubbing, M. F. Turner, and A. Weller, Nucl. Fusion 29, 641 (1989). Roquemore, and NSTX Research Team, Nucl. Fusion 51, 063008 (2011).
38 55
G. Jackson, T. A. Casper, T. C. Luce, D. A. Humphreys, J. R. Ferron, A. D. Mueller, M. G. Bell, R. Bell, B. P. Benoit, A. L. Roquemore, R.
W. Hyatt, E. A. Lazarus, R. A. Moyer, T. W. Petrie, D. L. Rudakov, and Raman, T. R. Jarboe, B. A. Nelson, S. A. Sabbagh, and V. Soukanovskii,
W. P. West, Nucl. Fusion 48, 125002 (2008). IEEE Trans. Plasma Sci. 38, 371 (2010).

Downloaded 03 Jun 2013 to 131.170.6.51. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://pop.aip.org/about/rights_and_permissions