Edge Plasma Control by Ergodic Layers on Large Helical Device

A. Komori, T. Morisaki, M. Kobayashi, R. Sakamoto, S. Masuzaki, H. Tsuchiya,
M. Shoji, N. Ohyabu, O. Motojima and the LHD Experimental Group
National Institute for Fusion Science, 322-6 Oroshi-cho, Toki, Gifu 509-5292, Japan
Abstract. The ergodic layer, which is located around the natural helical separatrix of Large Helical Device, was
demonstrated experimentally to have an ability to control plasma density remarkably by changing its thickness, in
addition to a function of the prevention of impurity penetration into the core plasma. This reason is that neutral particles
are ionized in the ergodic layer, and flow along the field lines to the wall. The thickness of the ergodic layer is an
increasing function of a resonant perturbation field, generated by an additional coil system.
Keywords: LHD, Ergodic layer, Density control
PACS: 52.55.Hc, 52.25.Vy
INTRODUCTION
The Large Helical Device (LHD) is a superconducting heliotron-type device at the National Institute for Fusion
Science at Toki, Japan.1,2 One of the key research issues in the LHD program is to control heat and particle fluxes
to the wall and to enhance core plasma confinement.3 This control of the LHD edge plasma will primarily be done
with a closed full helical divertor, which utilizes a natural separatrix in the edge region.4 The heliotron-type
magnetic configuration has an ergodized layer around the separatrix. In the previous studies, functions of the
ergodic layer were studied, and it was clearly shown that the radiation power emitted from the plasma Prad decreases
with an increase in the thickness of the ergodic layer.5 This demonstrates that the ergodic layer reduces the influx of
impurities, and is also considered to be true for fueling particles. Generally speaking, it is difficult to keep the
same experimental conditions for a long time, for example, in order to study the dependence of plasma performance
on one parameter, but such a study can be easily performed in LHD by using long-pulse discharges. By changing
the thickness of the ergodic layer during a longer shot than ~35 sec, the shielding effect of the ergodic layer on
impurities was found to work better when the ergodic layer was thick and over ~15.5 cm wide near the X-point,
located inside the torus at the midplane of the horizontally elongated cross section.5
The LHD magnetic configuration can be changed widely by the external helical and poloidal coils. Also, LHD
can have various magnetic configurations by using external perturbation coils located above and below the torus.6
These perturbation coils are used for controlling magnetic islands, and, at the same time, is available for changing
the thickness of the ergodic layer. A local island divertor (LID), which utilizes an m/n = 1/1 magnetic island
generated by the perturbation coils, is a powerful tool for the control of edge plasmas,7 and its results have been
demonstrated in the recent LHD experiments.7-9 In the LID, a divertor head is inserted inside the O-point of the
m/n = 1/1 magnetic island generated by the perturbation coils, and the present experiments were performed without
insetting the divertor head.
In this paper we intend to describe the behavior of electron density by changing the thickness of the ergodic layer.
In addition to the prevention of impurity penetration into the core plasma, the active control of plasma density is of
importance in realizing steady-state stable fusion reactors.
EXPERIMENTAL RESULTS AND DISCUSSION
The effect of the ergodic layer on plasma performance was studied using hydrogen-puffing
neutral-beam-injection (NBI) discharges at magnetic axis positions, Rax‘s, of 3.6 m and 3.75 m, and magnetic fields,
Bt‘s, of 2.75 T and 2.64 T, respectively, with the NBI power of ~4 MW. The thickness of the ergodic layer is
controlled by changing ILID, which is a current of the perturbation coil system.5 The current ILID of -400 ~ -300 A is
required at these magnetic fields for eliminating the intrinsic islands, generated by error field. Figure 1 shows
magnetic surfaces of the horizontally elongated cross section, after the intrinsic islands are eliminated. There are
nested magnetic surfaces inside the last closed flux surface (LCFS) for core plasma confinement, and the thick
ergodic layer is found to exist around the separatrix outside the LCFS. In Fig. 1, the connection length, Lc, is also
depicted, which starts from the point on the bold line, drawn on the equatorial plane of the horizontally elongated
cross section, and ends on the vacuum vessel.

The function of the ergodic layer is considered as follows. Since the ergodic layer connects with the vacuum
vessel or the carbon plates through the natural helical separatrix, the particles ionized in the ergodic layer flow along
the field lines to the vacuum vessel, as shown in Fig.2, and hence, ionized fueling particles cannot penetrate into the
core plasma. The thickness of the ergodic layer is considered to be important in this ionization process, because
neutral particles passing through the thick ergodic layer has much more probability of ionization in the ergodic layer,
compared with that in the thin ergodic layer. Thus the plasma density in the core plasma and, especially, in the
edge plasma should be controlled with the thickness of ergodic layer. This function is coherent with the highly
efficient pumping and core fueling, because the removal of ionized particles by the ergodic layer promotes the
efficient pumping, and the reduction, especially, of the density in the edge plasma causes a deep penetration of the
pellet into the core plasma. On the contrary, there is a possibility that the particles ionized near the vacuum vessel
flow into the core plasma through the ergodic layer. The amount of these particles is very smal1, compared with
the outward particles, because the volume of the ergodic layer surrounding the core plasma is much larger than that
between the vacuum vessel and the ergodic layer surrounding the core plasma, that is, the volume of the ergodic
layer along the mustache like separatrix. However, the amount of the inward particles is expected to become large,
if the ergodic layer surrounding the core plasma touches the vacuum vessel directly.
This shielding effect is also true for all impurities such as iron, oxygen, and nitrogen. In fact, the radiation
power, Prad, measured by a bolometer and normalized by the averaged electron density squared, n−
e
2, demonstrated
clearly in the previous experiments5 to decrease with an increase in ILID, that is, the thickness of the ergodic layer,
indicating that the ergodic layer prevents impurities from penetrating into the core plasma.

The experiments were performed with the constant NBI power, and the amount of gas puffing was also kept
constant through a series of shots, in which only ILID was changed. In Fig. 3, drerg is the thickness of the ergodic
layer near the X-point, located outside the torus at the midplane of the horizontally elongated cross section. The
ergodic layer at Rax = 3.75 m is about 2 times as thick as that at Rax = 3.6 m, and there is a tendency for the thickness
of the ergodic layer of the magnetic configuration, whose Rax is shifted on the inward side of the torus, to become
thin. It is clearly shown in Fig. 3(a) that n

e decreases with drerg from ~3.5 × 1019 m-3 to ~2 × 1019 m-3. The thicker
ergodic layer causes the lower n−e, as expected, and this indicates the shielding effect of the ergodic layer. In the
case of Rax = 3.6 m, that is, in Fig. 3(b), an increase in drerg leads to the reduction of n−
e a little when n−
e ~ 3.5 × 1019
m-3, while n−
e is, however, independent of drerg at a low n−
e of ~1 × 1019 m-3. The gentle decrease in n

e with drerg at
Rax = 3.6 m is probably attributed to the thin ergodic layer. The reason why n−
e is constant, independent of drerg, at
Rax = 3.6 m is considered as follows. It is well known that the electron temperature, Te, is an important factor in the
ionization process, because the cross section of ionization is a function of Te. The electron temperature, Te, profiles
were measured along the major radius, R, by the Thomson scattering in the present experiments. When n

e is low,
the Te profile is found to be higher than that with the high n

e, and this is because the heating power is the same for
both high and low n

e discharges with the different amounts of gas puffing. In the case of the low n−
e and high Te
profile, it is expected that the ionization occur mainly on the outer side of the ergodic layer, so that the thickness of
(a) (b)
the ergodic layer does not affect the amount of ionized particles in the ergodic layer, leading to the constant n

e.
Taking into account the thickness of the ergodic layer, the electron temperature profile, the amount of gas puffing,
and so on, the precise n

e behavior will be shown by the calculation using the EMC3-EIRENE code.10
SUMMARY
The shielding effect of the ergodic layer on fueling particles and impurities were experimentally confirmed on
LHD. The core plasma density can successfully be controlled by changing the thickness of the ergodic layer,
which is an increasing function of a resonant perturbation field, formed by an external perturbation coil system. Its
thickness changes in the experiments at a magnetic axis position of 3.75 m from ~42 cm to ~48 cm and at 3.6 m
from ~18 cm to ~23 cm near the X-point, located outside the torus at the midplane of the horizontally elongated
cross section. The precise behavior of plasma density will be shown by the calculation using the EMC3-EIRENE
code, taking into account the thickness of the ergodic layer, the electron temperature profile, the amount of gas
puffing, and so on.
ACKNOWLEDGMENTS
The authors would like to thank all members of device engineering group for their support and operation of the
machine. This research is partially supported by the Grant-Aid for Scientific Research from MEXT of Japanese
government.
REFERENCES
1. A. Iiyoshi et al., Nucl. Fusion 39, 1245 (1999).
2. O. Motojima et al., Phys. Plasmas 6, 1843 (1999).
3. A. Komori et al., in Plasma Physics and Controlled Nuclear Fusion Research 1994 (Proc. 15th Int. Conf. Seville, 1994), vol. 2,
p. 773.
4. T. Morisaki et al., J. Nucl. Material. 337-339, 154-160 (2005).
5. A. Komori et al., J. Nucl. Material. 313-316, 1267-1271 (2003).
6. A. Komori et al., Fusion Eng. Des. 39-40, 241 (1998).
7. A. Komori et al., Fus. Sci. & Tech. 46, 167-174 (2004).
8. A. Komori et al., Nucl. Fusion 45, 837-842 (2005).
9. T. Morisaki et al. (this coference).
10. Y. Feng et al., Plasma Phys. Control. Fus. 44, 611 (2002).

Опубликовано в рубрике Documents