Vous êtes sur la page 1sur 8

Gundestrup: A Langmuir/Mach probe array for measuring flows in the scrapeoff

layer of TdeV
Cyrus S. MacLatchy, Claude Boucher, Deborah A. Poirier, and James Gunn
Citation: Review of Scientific Instruments 63, 3923 (1992); doi: 10.1063/1.1143239
View online: http://dx.doi.org/10.1063/1.1143239
View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/63/8?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Multidirectional plasma flow measurement by Gundestrup Probe in scrape-off layer of ADITYA tokamak
Phys. Plasmas 22, 112501 (2015); 10.1063/1.4935292
Swinging reciprocating Mach probes for the high field side scrape-off layer in DIII-Da)
Rev. Sci. Instrum. 83, 10D723 (2012); 10.1063/1.4733571
A magnetically driven reciprocating probe for tokamak scrape-off layer measurements
Rev. Sci. Instrum. 82, 123505 (2011); 10.1063/1.3661128
High density Langmuir probe array for NSTX scrape-off layer measurements under lithiated divertor
conditionsa)
Rev. Sci. Instrum. 81, 10E117 (2010); 10.1063/1.3494381
A Langmuir/Mach probe array for edge plasma turbulence and flow
Rev. Sci. Instrum. 68, 377 (1997); 10.1063/1.1147834

Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 192.101.166.237 On: Thu,
07 Jul 2016 13:50:39

Gundestrup: A Langmuir/Mach probe array for measuring


in the scrape-off layer of TdeV

flows

Cyrus S. MacLatchy,a) Claude Boucher,b) Deborah A. Poirier,b) and James Gunna)


Centre Canadien de Fusion Magtktique,

Varennes, Quhbec, Canada

(Received 31 January 1992; accepted for publication

10 April 1992)

Gundestrup is a Langmuir/Mach probe array which measures the flow velocity in the scrape-off
layer of Tokamak de Varennes (TdeV). It is based on the concept of a Mach probe where
presheaths extending upstream and downstream from the probe, parallel to the magnetic field,
attract charge to a circular array of collecting pins. The polar distribution of ion saturation
currents to the circular array is used to compute the components of flow velocity in the plasma.
With Gundestrup, there is an assumed flow perpendicular to the magnetic field as well as
parallel to it. Equations representing the collection of charge by individual pins on the probe are
presented and sample flow patterns from the scrape-off layer are shown.

I. INTRODUCTION
The determination of flow velocity in the scrape-off
layer (SOL) and edge of tokamak plasmas has become of
prime importance given its possible role in confinement
and the L-H mode transition.lm3 Usually, this velocity is
calculated indirectly from other parameters. Diagnostics
such as ion beams or Langmuir probes, capable of measuring the plasma potential and its distribution, are used to
calculate the electric field. These measurements provide an
indirect method of calculating the flow velocity, assuming
that the flow is dominated by EXB drifts. If density and
temperature are measured or inferred, then the diamagnetic contribution to flow can be estimated. Using these
techniques, the value obtained for the velocity is at best an
estimate. If it is assumed that impurities in a hydrogen
plasma drift at the same or comparable velocity as the bulk
ions, then the plasma flow can be determined by measuring
the Doppler shift of impurity line radiation.4 Unfortunately, this basic assumption may not be correct.5 Several
experimenters have also used Mach probes to measure velocities.67 With these probes, two collectors aligned with
the magnetic field and shadowed from one another collect
current from the upstream and downstream directions.
The ratio between the upstream and downstream ion saturation currents is used to compute the Mach number of
the ~?ow.~This approach is limited to the determination of
the velocity parallel to the magnetic field.
The purpose of the present article is to describe a new
Langmuir/Mach
probe array which we call Gundestrup.
When used as a Langmuir probe, Gundestrup is capable of
measuring the density, temperature and potential in the
edge plasma of Tokamak de Varennes (TdeV). In addition, the ion saturation currents collected by the Gundestrup pins can be used to infer the speed and direction of a
flowing plasma. To achieve this, Gundestrup uses two separate probe arrays mounted on a 45mm-diam cylindrical
head made of graphite. As shown in Fig. 1, the simpler
array consists of two probes; one is a proud pin which
Acadia University, Wolfville, NS, BOP 1X0, Canada.
INRS-Iknergie, C. P. 1020, Varennes, Quhbec J3X lS2, Canada.

acts as a standard Langmuir probe while the other is a


flush-mounted probe which is used to estimate the end
effect for the other probes. The two-probe array is mounted
on the flat end of the cylindrical head which faces into the
plasma. The second array is a group of 12 pins which are
positioned at equal 30 intervals around the circumference
of the head. These pins are plane surfaced and each faces
radially outward. The Gundestrup assembly is mounted on
a long shaft which can be moved in and out of the plasma
by a stepping motor. The current to each of the 14 pins is
measured as a function of the biasing potential which is
applied as a 50-Hz sawtooth. Standard procedures are used
to compute the density, temperature, and potential from
the Langmuir probe characteristic of the proud pin. The
distribution of ion saturation currents collected by the 12pin array is used to calculate the magnitude and direction
of the flow in the plasma.
The edge plasma of a tokamak is characterized by intense magnetic fields which guide the motion of electrons
and ions. In addition, flow fields may also contribute to the
anisotropy of the current collected by a Langmuir probe.
Polar diagrams like those published by Ohtsuka et al.
provide strong evidence for this anisotropy. These factors
must be taken into account when the probe signal is used to
measure plasma parameters. Until only a few years ago,
probe signals in such environments were not well understood but a number of papers in recent years have addressed these various difficulties, l2 such that researchers
now use Langmuir/Mach
probes to measure not only the
usual plasma parameters but the flow parallel to the magnetic field as we11.13However, whereas a Langmuir/Mach
probe measures only parallel flows, the Gundestrup probe
measures flow perpendicular to the magnetic field as well.
II. THE SIMPLE MACH PROBE
Several authors have proposed theories concerning the
collection of ions by a Langmuir probe in a flowing, magnetized plasma.4-20 Because of the importance of these
models for the interpretation of results from tokamak edge
plasmas, experiments have been performed where the ion
saturation current has been measured under conditions of

@I 1992 AmericanDownload
Institute to IP:
of Physics
3923 On: Thu,
00346746/92/063923-07$02.00
Rev. Sci.
Instrum.
63 (6).toAugust
1992
Reuse of
AIP Publishing
content
is subject
the terms
at: https://publishing.aip.org/authors/rights-and-permissions.
192.101.166.237
3923
07 Jul 2016 13:50:39

proud

Notice that we use W, for the tip probe because it


faces both upstream and downstream. It is clear that the
sum of the upstream and downstream currents must be
equal to the current collected by the tip probe. Therefore,
x= (a+P)/2.

If-a
-Y
Irp;
;
I I
>izL (

lP

FIG. 1. The Gundestrup head, which is about 44.5 m m in diameter, is


constructed from graphite. The pins are tungsten. When operating, the
flat surface with the proud and flush collectors points into the tokamak.
The inset shows a detail of a flat-faced Gundestrup pin. The dimensions of
the pin are listed in Table I.

controlled magnetization and fiow.2-23 The experiments


support the theoretical approach taken by Hutchinson in
which a one-dimensional model is used to analyze a collisional plasma in a magnetic field with parallel flow. Our
analysis is based on this model.
It is convenient to introduce the Gundestrup equations
by first discussing a three-tip probe array similar to that
described by Boucher et al. l3 The pins are biased at sufficiently negative potential to ensure that the ion current is
saturated. We assume that there is a strong magnetic field
and that the plasma is constrained to flow parallel to B. A
single, cylindrical tip probe sticks into the plasma and collects current from all directions. Because of the strong
magnetic field, the collection is predominantly parallel to
B. The two additional probes are shadowed so that one
faces upstream and the other faces downstream. If the ions
are singly charged, then the currents collected by the
upstream-facing probe, the downstream-facing probe, and
the tip probe are, respectively,

(5)

Both here and with the Gundestrup probe, we use the ratio
of i,/Id to determine the Mach number of the flow. This
enables us to determine x which is used in (3) to compute
the electron density. A complete characteristic is required
to evaluate the electron temperature so that c, can be evaluated from (4) for use in ( 3). In our conceptual picture of
the probes, a thin electrical sheath of a few Debye lengths
thickness is formed near the surface of the collecting pin.
The electric field in the sheath is sufficiently strong so that
any ion entering the sheath is immediately collected, Outside of the sheath there is a long transitional region, the
presheath, which joins the sheath edge to the unperturbed
plasma. Since the ions are constrained to move along magnetic field lines, the presheath extends outward from the
pin in the form of a tube with cross section equal to the
projected area of the pin. In the presheath, the ions are
accelerated toward the sheath, reaching it at or near the
sound speed. Ions move into the presheath region by diffusing across the magnetic field surface surrounding the
presheath flux tube. Depending on the direction in which
the probes face, the density of the ions reaching the sheath
edge is an, or /3n,. It is the values of a and 0 as a function
of the Mach number which are predicted by the Hutchinson modelI and it is the Hutchinson model which is used
to analyze the collection of current by the Gundestrup
pins.
The perturbation of the plasma caused by the probe
extends up and downstream along the magnetic field lines.
The length of this perturbed region is determined by the
rate at which ions diffuse across the magnetic field surface
into the flux tube. If a is the radius or typical dimension of
a probe, then the collection length for a typical probe can
be approximated by

Lp a2cs/Dl.

(6)

Here, D, is the cross-field diffusion coefficient usually taken


to be the Bohm value. Normally the presheath collection
length is much greater than the size of the probe.

I, = czApnpp

(1)

III. GUNDESTRUP

~d=&$h%
It=x2Qvcs

(2)

The Gundestrup analysis is based on the Mach probe


model described in the previous section. However, in addition to the streaming of plasma parallel to the B field, it
is necessary to allow for the possibility of flow normal to
the magnetic field. This flow might be associated with E
XB forces and would be represented by a drift in the guiding centers of the particles as they follow Larmor-type
gyrations through the plasma. In addition to the main
source of diffusion from the central plasma in the radial
direction, we also allow for diffusion into the flux tube
from the sides in the poloidal direction. Our objective here
is to assume a suitable model for the collection of ions by
the individual pins. We then use our model to predict the

(3)

where A,, n,, e, and c, are, respectively, the area of the


probe normal to the magnetic field, the electron density in
the unperturbed plasma, the charge on the electron, and
the ion sound speed. For singly ionized species, the sound
speed is defined by
c,=

k( T,+ Ti)/mi,

(4)

where T, and Ti are the electron. and ion temperatures. In


this work, we assume them to be equal. The ion mass is mi
and k is the Boltzmann constant.

3924
Rev. Sci. Instrum., Vol. 63, No. 8, August 1992
Langmuir/Mach
probe array
3924
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 192.101.166.237 On: Thu,
07 Jul 2016 13:50:39

polar distribution of current to the array. Finally, we parametrize the parallel and perpendicular currents and use a
chi-square analysis to get the best fitting polar diagram of
our modeled currents to the experimental values.
A major difficulty in this analysis is the determination
of the projected pin area. We examined three approaches
before adopting a final method: ( 1) a purely geometrical
calculation, (2) a method involving the ion Larmor radius,
and (3) a calculation based on the ion dynamics. As represented in Fig. 1, each pin is a machined half-cylinder
pointing radially out from the graphite housing. The cavity
surrounding the pin is twice the radius of the pin. The flat
area of the pin facing the plasma is Ap=2r& while that of
the cavity is 2A,
The purely geometrical picture must account for the
angle of the magnetic field with the collecting surface, the
projected area of the half-cylinder and the possibility of
shadowing by the housing when the magnetic field angle is
large. This approach results in an effective area for each pin
but generates large discrepancies for the center upstreamfacing pin when we apply the chi-square fit to the saturation currents.
The Larmor technique is a best-guess method. Here,
we assume that any ion passing closer to the housing than
the ion Larmor radius will be scraped off by the housing.
The large electric field between the housing and the pin is
assumed to be sufficient to guide all other ions into the pin.
For the typical ion Larmor radius of our plasma, this
method results in an effective area
AE cos e=: lSA, cos 6,
where 8 is the angle between the normal to A, and the
magnetic field.
Finally, the dynamic method assumes that the ions
either strike the pin directly, or enter the region between
the pin and the housing with a kinetic energy given by

i m,u=$ kT,+3kT,+eV,

(7)

where V is the potential at the position where the ion enters the space between the pin and the housing. Since our
simple model assumes that the ion Larmor radius is comparable to this gap, most of the ions are scraped off before
penetrating deeply into the gap and the potential is just the
free-space potential between two concentric cylinders. The
energy terms on the right-hand side of (7) account for the
thermal energy and the energy gained by the ion when it
passes through the plasma sheath into the gap region. Although the ions are assumed to enter the sheath with an
initial guiding center velocity and a velocity of gyration,
the main impact on an ions velocity is due to the kinetic
energy it gains in passing through the sheath. With this
initial velocity, the trajectory of the ion is then followed to
see whether it strikes the pin or not. By using a Monte
Carlo technique with ions of varying impact parameters,
we are able to evaluate the effective area of the collecting
pin. Although we do not reproduce the calculations here,
the model does give us some confidence that the effective
area of the pins, shown in Fig. 2, is a weak function of the

0 (deg.)
FIG. 2. The effective collecting area of the Gundestrup pins is shown as
a function of the angle between the normal to the pin surface and the
magnetic field. A, (0) is the effective area from the dynamic model and
A, (dashed line) is the actual area of a pin.

magnetic field angle. When the angle with the magnetic


field is included, we get an average value of
AE cos 6= 1.4/t, cos 8,
a result not much different from the Larmor approach.
This final method was adopted because it appears to give a
more realistic model for the actual Gundestrup pins.
In Fig. 3, a typical upstream-facing probe is illustrated
with its presheath extending upstream parallel to the magnetic field. The density in the presheath near the collecting
pin is an,. Since the housing surrounding a pin is at the
floating potential, its presheath density is also an,. Analogous to the definitions used in Eqs. ( 1 )-( 3), we now define
parameters that can be used in the Gundestrup analysis. In
the equations below, JI is the current density perpendicular
to the magnetic field. To find the current density parallel to
the magnetic field at the surface of a collecting pin, J,, must
be multiplied by the factor a or fi depending on whether
the pin faces up or down stream.

FIG. 3. The collecting surface of a Gundestrup pin is shown with ion


currents flowing toward the pin as discussed in the text. The shaded
region represents the effective collection area of a Gundestrup pin, A,

Rev. Sci.content
Instrum.,
Vol. 63,
No. terms
6, August
1992
Langmuir/Mach
probe array
3925 On: Thu,
Reuse3925
of AIP Publishing
is subject
to the
at: https://publishing.aip.org/authors/rights-and-permissions.
Download to IP: 192.101.166.237
07 Jul 2016 13:50:39

Here, vI is the perpendicular component of the velocity.


We compute the parallel component from

where M is the Mach number taken from the ratio of the


upstream and downstream currents in the Hutchinson
model. If we use a to account for the density in the
presheath and assume that thle perpendicular current is
flowing in the positive y direction, then an up-facing pin
like that in Fig. 3 will collect

I,*=aJl, AEcos 8-aJLAEsin

8.

(8)

A down-facing pin gains charge due to JL and collects

Ip*= a;ill A, cos 8 + aJL AE sin 8.

?Z,---ni

eAE sin 8.

(10)

If we include a and the ion sound speed, this equation


becomes

(11)
Or, with the definitions for J,, and Lp, we have
i&r

a (1-a)JII AEsin 8.

=aAE(JII cos 6&J, sin 0) +K( 1--a)J,, AEsin 6.


For a downstream-facing

(12)

The ratio of a/&, cannot be precisely determined, nor can


we claim that (12) represents a rigorous theory for the
additional flux of ions into the presheath. Therefore, a/L,

(13)

pin, the equation is

Zp=~AE(J~~cosBiJLsin 8)+K(l--fi)JIIAEsin&

(141

These equations, which are written for each of the Gundestrup pins, contain three adjustable parameters, J,, , JI f
and K.
IV. MEASUREMENT

(9)

For pins which face downstrea:m, the equations are identical except that a is replaced with /3.
The model has so far assumed that the density in the
flux tube formed by the Gund,estrup head is constant for
any cross section normal to the B-field and equal to alt, or
fin, at the probe surface. In fact, the structure of the
presheath is more complicated than this simple model implies. It has been shown both theoretically226 and experimentally2 that the density in the flux tube is a function of
not only the distance from the collecting surface as in the
one-dimensional ( 1D) models but is a function of the other
spatial coordinates perpendicular to the magnetic field as
well. Thus, all along the presheath formed by the Gundestrup head, the density is nonuniform and there is a varying flux of ions into the presheath which is purely diffusive
and which can be collected by a pin facing outward.
In our simplified model for the diffusive component,
we assume that the pin region receives diffusion-drive ions
from both the radial and poloidal directions. Since all the
pins are at the edge of the inward-facing probe head, the
radial flux to each pin should be the same. On the other
hand, the poloidal flux influences those pins near the
boundary between the presheath and the unperturbed
plasma more strongly than those closer to the middle of the
presheath. This additional diffusion-driven current can be
estimated in the following way. If the density just outside
the presheath is n, and inside, n, then the current due to
diffusion is

IdzDL --g-

is replaced with K and the total current to an upstreamfacing pin is finally written as

OF FLOW

In a typical measurement of flow, the ion saturation


current to each of the 12 Gundestrup pins is measured.
The current to two of the pins is not used in the analysis
because the magnetic field is at grazing incidence to the pin
surface rendering the collection area of the pins indeterminate. This is a difficult problem that has been discussed
elsewhere28 and remains unresolved. For each of the remaining 10 pins, Eqs. ( 12 ) or ( 13 ) , depending on whether
the pin faces up or downstream, is used to determine the
theoretical ion saturation current. Both Jr1and J4 are varied
to minimize the value of chi-squared defined by

x2=c (4n-Ip)2,
where I, is the measured saturation current and 1, is the
calculated value. The value of K (usually in the range
0.05 < K < 0.1) has been determined by obtaining a best tit
for the full range of flows encountered in this experiment.
It was not used to minimize x2 for individual flow patterns.
For the flows presented here, K=0.075.
The Gundestrup probe is mounted on the outboard
midplane of TdeV and is moved radially inward and outward relative to the last closed flux surface (LCFS) of the
plasma. TdeV is a medium-sized tokamak having a major
radius of 865 mm and a minor radius of 270 mm for the
experiments described here. Other relevant operating parameters are listed in Table I. TdeV is ohmically heated
and is operated in divertor mode for these experiments.
The divertor plates of TdeV are electrically insulated from
the vacuum vessel, enabling the plates to be biased at potentials varying from - 150 to + 150 V relative to the vacuum vessel. This feature allows the radial electric field in
the SOL to be modified. In the descriptions which follow,
data have been collected as a function of the distance from
the LCFS and of the radial electric field.
Figures 4-6 show representative polar diagrams of the
measured and calculated ion saturation currents for three
different flows in the SOL of TdeV. As mentioned above,
the two pins located on the y axis are not used in the
analysis because of the grazing incidence of the magnetic
field. In these diagrams, we are looking into the tokamak
from outside on the midplane. The magnetic field is sloped
at 8 and points downward toward the left. The current in
the main plasma points toward the right and is horizontal,
Positive velocities point either to the right or up as in a

Rev. Sci. Instrum., Vol. 63, No. 8, August 1992


Langmuir/Mach
probe array
3926
3926
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 192.101.166.237 On: Thu,
07 Jul 2016 13:50:39

TABLE

I. Experimental parameters (divertor mode, ohmic heating).


Plasma parameters for velocity plots

IP
II, (line avg.)
n,(SOL)
T:
CT
;jr
R
a

210 kA
2.5 X lOI mm3
2 X IO* me3
20 eV
5.7 X lo4 m/s
1.14 T
0.16 T
865 mm (approx.)
270 mm (approx.)
Gundestrup parameters

t
Head diameter

0.795 mm
1.87 mm
44.5 mm

Approximate value, varies with position.


b0= 8, estimated values at the probe position in the SOL.

standard Cartesian system. The operating parameters for


TdeV, included in Table I, are maintained at fixed conditions except that the biasing potential on the divertor plates
is varied during the experiment. Varying the biasing potential on the plates modifies the electric field in the SOL and
causes a change in the plasma flow which is easily observed
in the polar diagrams. As indicated in the figure captions,
the parallel and perpendicular components of velocity have
both been modified by the changes in the field. In Figs. 7
and 8, we present velocity profiles in the SOL for several

FIG. 4. This is a Gundestrup polar plot of the ion saturation current in


mA as a function of the angular position of the pins in the array for a
biasing potential of - 100 V on the divertor plates. We are looking into
the outboard plasma on the midplane of the tokamak. The radial electric
field in the SOL is about 1 kV m- and points inward while the sloped
line represents the orientation of the magnetic field which points to the
left. The current in the core plasma points to the right. Since the ion
saturation currents are larger on the right-hand side of the plot, one
deduces that there is a large velocity directed toward the left. Applying
the Gundestrup analysis, we find that the Mach number is 0.2, u,,=7.6
km s- and u,=O.2 km s- directed perpendicular to B and up. The
dashed line represents the measured currents and the solid line, the best fit
to the data. The open circles (0 ) represent data which are not used in the
calculation.

FIG. 5. Here, the biasing potential on the plates is zero. However, the
radial electric field is about 1.2 kV m- and points outward, opposite to
that in Fig. 4. It is clear from this polar plot that the flow velocity has
changed direction from that shown previously. The Mach number is 0.3,
uII=8.6 km s- and points right while u,=3.9 km s- and points downward.

biasing conditions. We see changes in both the parallel


velocity and the perpendicular velocity as the biasing potential is varied from - 150 to + 150 V. Finally, in Figs. 9
and 10, the velocity components are plotted as a function
of the radial electric field in the SOL. We note that the
slopes of these curves can be derived from the magnetic
field. The slope of the uI line is the inverse of the total
magnetic field and the slope of the uII line is the reciprocal
of the poloidal magnetic field. These results indicate that
the velocity is due to the Lorentz force. We defer a more
complete discussion of the origin of the fluid motion to a
future paper, limiting the scope of the present article to the
description of the Gundestrup probe.

FIG. 6. The biasing potential is + 150 V, producing a radial electric field


of about 4.5 kV m- pointing outward. The flow has increased substantially with a Mach number of 0.8, a,,=43 km s- to the right and u,=5.4
km s- pointing downward.

Reuse 3927
of AIP Publishing
is subject
to the
at: https://publishing.aip.org/authors/rights-and-permissions.
Download to IP: 192.101.166.237
Rev. Scl.content
Instrum.,
Vol. 63,
No. terms
8, August
1992
Langmuirf Mach probe array
3927 On: Thu,
07 Jul 2016 13:50:39

-A
f -Jj

VI
xi

40

zPrl

20 I

20

40
60
DLCFS ( mm )

40

8(

.g

20

;ij
P
;

-I

-20

-2000

2000

4000

Radial Electric Field ( V/m )


FIG. 7. The parallel component of velocity is shown as a function of the
DLCFS for a series of biasing potentials. Here, V= -+ I50 V, X= + 100 V,
A=+5OV,O=OV,
+=-5OV,
O:=-lfKJV,andCl=-15OV.

V. DlSCUSSlON
At the best of times, the interpretation of probe results
is fraught with difficulties. The Gundestrup results are no
different because, in addition to the usual Langmuir probe
difficulties, we must also be concerned about the viability
of our assumptions concerning the flow of charge to the
Gundestrup pins. In the discussion which follows, we will
address the various assumptions used to model the fiow of
charge.
The effective collecting area for the pins is a critical
element in the computations. Rather than taking a purely
geometrical interpretation, we have chosen to estimate the
pin area by following the trajectories of ions into the space
between the pins and the graphite housing. Since the minimum area of the pins is A,, the effective area of the pins is
expected to lie in the range A, < A,< 1.4A,. Within this
range, the uncertainty in AE is about *20%. Since AE
appears to the first power in each term of Eqs. ( 13) and
(14), this introduces an identical uncertainty in both uIl
and ul.
The Hutchinson model has been used to calculate the
Mach number and hence uII . Experiments have been performed in which the Mach number computed from the
ratio of the upstream to downstream saturation currents
was compared to the flow velocity computed from the

FIG. 9. The average parallel component of velocity is shown as a function


of the radial electric field in the SOL. This average is taken over the range
from 10 to 30 mm from the LCFS. A positive value of the field indicates
that the field points outward from the plasma. The solid line represents
the least-squares fit to the experimental data, which is shown as an open
square (Cl). The error bars are estimated from the scatter in the velocity
values computed from the Gundestrup analysis.

Doppler shift in laser-induced fluorescence measurements.


Results from these experiments support the use of the
Hutchinson model but additional work is required to determine the accuracy of velocities computed from the up/
down asymmetry in ion collection.
The Gundestrup model itself is based on three basic
assumptions. There is a flow of charge to the pins parallel
to the magnetic field which is governed by the Hutchinson
model. A perpendicular flow of charge exists whose origin
is not considered in our model but which contributes to the
asymmetry in the polar plots. In order to fill in the side
pins, it has been necessary to assume that diffusion into the
presheath of the Gundestrup housing is stronger at the
sides than on its axis of symmetry. There is evidence of this
type of behavior on other large probing objects in the SOL
where investigators report variations in density profiles
taken across the presheath. 24,25The final verification of this
model will depend on experience with known flows and
comparison of the Gundestrup results with other diagnostics.

0
-2
-4
~

40
60
20
DLCFS ( mm )

80

FIG. 8. The perpendicular component of velocity is shown as a function


of the DLCFS. We are using the same symbol conventions as for Fig. 7.

-6;

Electric Field ( V/m >

FIG. 10. The average perpendicular component of the velocity is shown


as a function of the radial electric field. The solid line shows the leastsquares fit to the data.

Rev. Sci. Instrum., Vol. 63, No. 8, August 1992


3928
Langmuir/Mach
probe array
3928
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 192.101.166.237 On: Thu,
07 Jul 2016 13:50:39

We must also be concerned with the connection length


for both the probe head and the pins. For the probe head,
L, is approximately 10 m and it is likely connected to the
divertor plates. On the other hand, the pins are in the
boundary of the head presheath and the calculated value of
L, is approximately 0.1 m. We would argue that the pins
are not connected and there is experimental evidence to
support this view. When we use the Hutchinson model for
an unconnected probe to compute the electron density, we
get very good agreement with lithium beam ablation measurements in the SOL of TdeV.29 Also, if we compare the
proud probe signal with those from the up and
downstream-facing pins in the Gundestrup array, we find
that all tests which we apply indicate that the proud probe
and the Gundestrup pins are in the same plasma. For example, if we use Eq. (3) and the signal from the proud pin
to compute n, ( 1) and the signal from the upstream-facing
pin, and finally, (2) and the signal from the downstreamfacing probe, we find that the densities thus computed are
essentially identical. Also, within experimental error, we
find that I,= I, + I, as one would expect if the presheaths
of the three pins are being generated by the same plasma
conditions. Since the proud pin samples the unperturbed
plasma, it appears that the Gundestrup pins are also sampling an essentially unperturbed plasma. This view is consistent with a transitional region from the undisturbed
plasma to the core of the presheath which is large compared to the Gundestrup pins.
If we look at the shot-to-shot reproducibility of the
velocities or the variation of the velocities within a sequence of 10 or 12 consecutive characteristics taken over a
time period of 200 ms, we get the following picture. The
parallel flow is quite reproducible and varies by less than
10% from characteristic to characteristic. On the other
hand, the nature of the fitting procedure ensures that any
change in the computed parallel velocity produces a similar change in the magnitude of the perpendicular velocity
(Au I~ -AU,,). Thus, the percentage errors generated in
the perpendicular component can be quite large. For perpendicular velocities of several km/s, the variations are
typically 30%, while for velocities less than 500 m s-l, the
error can be as great as the measured quantity itself. Thus
the Gundestrup results are unreliable for small perpendicular velocities, while for large parallel flows they are quite
reproducible.
As a new tool for tokamak research, Gundestrup will
require operating experience and careful analysis of measurements in simple flow regimes but there is clearly significant potential for using a Gundestrup probe in more
complicated flows such as might exist near X points and
divertor plates or at the critical transitional region between
the SOL and the interior plasma. In future applications, it
will be important to design collecting pins whose areas are
more reliably computed than those in the present device.
As well, collection lengths must be minimized to maintain
spatial resolution. Present plans are to mount Gundestrup
on a fast scanning system so that we can penetrate the last
closed flux surface of TdeV.

ACKNOWLEDGMENTS
We would like to thank Barry Stansfield and Sid Gulick for many helpful discussions, Michel Leblanc for his
excellent technical assistance, and the entire TdeV team for
keeping runaway electrons to a minimum. D.A.P. is grateful for support from the Canadian Fusion Fuels Technology Project and J. G. acknowledges the Natural Sciences
and Engineering Research Council of Canada for financial
support. The Centre Canadien de Fusion Magnetique is a
joint venture of Hydro-Quebec, Atomic Energy of Canada,
Ltd., and INRS-Energie.

R. J. Groebner, K. H. Burrell, and R.P. Seraydarian, Phys. Rev. Lett.


64, 3015 (1990).
*S. -I. Itoh and K. Itoh, Phys. Rev. Lett. 60, 2276 (1988).
3K. C. Shaing and E. C. Crume, Jr., Phys. Rev. Lett. 63, 2369 (1989).
4M. Bitter, K. L. Wong, S. Scott, H. Hsuan, B. Grek, D. Johnson, and
G. Tait, Phys. Fluids B 2, 1503 (1990).
Y. B. Kim, P. H. Diamond, and R. J. Groebner, Phys. Fluids B 3,205O
(1991).
6P. J. Harbour and G. Proudfoot, J. Nucl. Mater. 121, 222 (1984).
G. Proudfoot, P. J. Harbour, J. Allen, and A. Lewis, J. Nucl. Mater.
128-129, 180 (1984).
*P. C. Stangeby, Phys. Fluids 27, 2699 (1984).
Gundestrup is the site of a Danish peat bog where a cult cauldron
dating back to the first or second century B.C. was discovered in 1891.
It is decorated with the faces of numerous deities which look radially
outward from the cauldrons sides. P. Mac Cana, Celtic Mythology, 2nd
ed. (Newnes, Middlesex, GB, 1984), p. 8.
B. Lipshcultz, I. Hutchinson, B. LaBombard, and A. Wan, J. Vat. Sci.
Technol. A 4, 1810 (1986).
H. Ohtsuka, H. Kimura, S. Shimomura, H. Maeda, S. Yamamoto, M.
Nagami, N. Ueda, A. Kitsunezaki, and T. Nagashima, Plasma Phys. 20,
749 (1978).
*P. C. Stangeby and G. McCracken, Nucl. Fusion 30, 1225 (1990).
13C. Boucher, C. S. MacLatchy, G. Le Clair, J. L. Lachambre, and M.
St-Onge, J. Nucl. Mater. 176-177, 1050 (1990).
14P. C. Stangeby, Phys. Fluids 27, 682 ( 1984).
P. C. Stangeby, J. Nucl. Mater. 121, 36 (1984).
161. H. Hutchinson, Phys. Fluids 30, 3777 (1987).
P. C. Stangeby, Phys. Fluids 31, 2726 (1988).
*I. H. Hutchinson, Phys. Fluids 31, 2728 ( 1988).
19K-S. Chung and I. H. Hutchinson, Phys. Fluids B 3, 3053 (1991).
2oI. H. Hutchinson, Phys. Fluids B 3, 847 (1991).
*K-S. Chung, I. H. Hutchinson, B. LaBombard, and R. W. Conn, Phys.
Fluids B 1, 2229 ( 1989).
22S. L. Gulick, B. L. Stansfield, Z. Abou-Assaleh, C. Boucher, J. P.
Matte, T. W. Johnston, and R. Marchand, J. Nucl. Mater. 176-177,
1059 (1990).
23D. A. Poirier, S. L. Gulick, C. Boucher, B. L. Stansfield, and C. S.
MacLatchy, Bull. Am. Phys. Sot. 36, 2423 (1991).
24L. Koltai, D. Hildebrandt, J. S. Bakos, and P. Bachmann, J. Nucl.
Mater. 176-177, 1044 (1990).
25H. X. Vu and R. W. Gould, Phys. Fluids B 1, 68 ( 1989).
261. H. Hutchinson, Phys. Rev. A 37, 4358 (1988).
27G. F. Matthews, P. C. Stangeby, and P. Sewell, J. Nucl. Mater. 145-147,
220 (1987).
28G. F. Matthews, S. J. Fielding, G. M. McCracken, C. S. Pitcher, P. C.
Stangeby, and M. Ulrickson, Plasma Phys. Contr. Fusion 32, 1301
(1990).
2VD. Michaud, G. G. Ross, F. Schiettekatte, B. Stansfield, P. Couture and
C. Boucher, J. Nucl. Mater. (to be published, April 1992).

Langmuir/Mach
probe array
3929 On: Thu,
Reuse of
AIP Publishing
content
is subject
at: https://publishing.aip.org/authors/rights-and-permissions.
Download to IP: 192.101.166.237
3929
Rev. Sci.
Instrum.,
Vol. to
63,the
No.terms
8, August
1992
07 Jul 2016 13:50:39

Vous aimerez peut-être aussi