Fusion Engineering and Design 63 /64 (2002) 153 /163

www.elsevier.com/locate/fusengdes

A review of dust in fusion devices: Implications for safety and

operational performance
J.P. Sharpe a,
, D.A. Petti a, H.-W. Bartels b
a

Fusion Safety Program, Idaho National Engineering and Environmental Laboratory, PO Box 1625, MS 3860, Idaho Falls, ID 83415
3860, USA
b
ITER International Team, IPP-Garching, Garching, Germany

Abstract
Dust is produced in fusion devices by energetic plasma /surface interactions. As the amount of dust increases,
potential safety and operational concerns arise. The dust may contain tritium, may be radioactive from activation
products, and may be chemically reactive and/or toxic. Possible accidents in large fusion reactors could mobilize the
dust and threaten public safety. Dust also poses potential problems to device operation. For example, plasma startup
could be impeded, particulate injected from flaking deposits may disrupt the fusion plasma, and tritium retention in
dust will affect fuel recovery systems. The current understanding of dusts role in fusion devices is reviewed in this paper
by discussing mechanisms of dust production, considering ways dust impacts device safety and operation, and
comparing characteristics of dust collected from existing fusion plasma research devices.
# 2002 Elsevier Science B.V. All rights reserved.
Keywords: Tokamak dust; Safety analysis; Fusion aerosols; Plasma /surface interaction

1. Introduction
By the very nature of its operation, a fusion
device generates aerosol particulate, broken flakes,
globules, chunks, and other debris, that may
ultimately affect its safety and operational performance. Particulate matter of this type, commonly
referred to as dust, does not strongly adhere to
surfaces and is capable of being mobilized. Evidence of the existence and impact of dust is readily

Corresponding author. Tel.: /1-208-526-9830; fax: /1208-526-2930

E-mail address: sharjp@inel.gov (J.P. Sharpe).

found in current fusion plasma research devices.

Streaking particles are often observed on images of
tokamak plasma discharges, particularly during
startup and periods of intense plasma /surface
interaction. Break-up of flakes or particles from
arcing are injected into the plasma and rocket
along field lines. Films of re-deposited material are
found in the proximity of intense plasma/surface
interaction. Disruptions and other off-normal
events cause significant erosion, and a sizeable
portion of the eroded material generates dust.
Several different mechanisms produce dust of
various shapes and characteristic sizes, as demonstrated in Fig. 1. Dust generated from the material
may be copious, radioactive, chemically reactive,

0920-3796/02/$ - see front matter # 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 0 - 3 7 9 6 ( 0 2 ) 0 0 1 9 1 - 6

154

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

Fig. 1. Representative SEM micrographs demonstrating various particulate shapes of dust collected from fusion devices.

and/or chemically toxic, thereby, posing significant

safety hazards should the dust be mobilized in an
accident or during maintenance. Shape, size, and
number (mass) are key factors in determining
dusts role in safety and its impact on machine
operation. Even for normal operation, a dirty,
dusty device is undesirable. Deposits of dust may,
for example, accumulate in the grooves of firstwall components, diminishing its effectiveness and
possibly shortening its lifetime. The effectiveness
of wall conditioning may be compromised by the
presence of foreign particulate matter. Off-normal
events, such as vertical displacement events in
magnetic fusion systems of or low-yield shots in
inertial fusion systems, may generate dust concentrations locally that are mobilized and adversely
affect the overall system. The impact of dust in the
chamber has recently been recognized as an issue
that must be addressed for future fusion devices.
This paper reviews our current knowledge about
the role of dust in fusion devices. Various mechanisms responsible for dust production are addressed. Safety and operational concerns are
discussed for next step fusion devices. Results of
recent collection activities are compared to show
similarities and differences in dust characteristics.
Dust from a variety of fusion research devices has
been analyzed: the tokamaks of DIII-D, TFTR,
ALCATOR-Cmod, Tore Supra, and ASDEXUpgrade, the LHD heliotron stellarator, and the
Nova laser chamber representative of an IFE
system. Finally, directions of future dust research
are considered.

2. The origin of dust in fusion devices

Many mechanisms are responsible for generation of particulate in fusion devices, and different
mechanisms can dominate in various circumstances. The effectiveness of each dust generation
mechanism is characterized by the amount of
energy available for mobilization. Possible mechanisms in magnetic fusion systems include
blistering and fracturing of deposited layers, generation of reactive species in edge plasmas, arcing,
explosive ejection and brittle destruction of surface
imperfections, and nucleation of vaporized materials. Similar mechanisms are expected in inertial
fusion systems, since large heat loads will be
incident on surrounding surfaces, and additional
mechanisms are being identified for various wall
protection schemes. Many more magnetic fusion
devices are presently available to study dust;
hence, we will focus on the origin of dust in these
devices.
Materials exposed to a flux of energetic plasma
particles erode by physical and chemical sputtering
[1], and erosion rates can approach 1 m per year
for carbon machines [2]. Most of the eroded
material is re-deposited at or close by its origin,
although some material is transported and deposited in layers at more distant, and generally cooler,
locations. These layers are generally stratified from
deposition during cycles of plasma operation,
weakly adhered to the underlying surface, and
subject to break-up by mechanical and thermal
stresses. Properties of re-deposited layers in carbon

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

machines strongly depend on local plasma and

wall conditions; the coatings may be brittle and
contain mostly carbon atoms, or softer polymeric
coatings of less adhesion. Metals or other materials will deposit along with carbon, so that flakes
are often composed of mixed materials. Thin films
are also applied for the purpose of wall conditioning. Spallation and flaking of the layers of
deposited material produce dust particles whose
size depends on the stress mechanisms and local
deposit structure. Dust produced by flake fracturing generally does not display indications of
particle agglomeration or growth.
Dust growth may occur in the edge plasma,
where conditions resemble those of reactive plasmas [3]. Significant quantities of carbon, oxygen,
and impurities from wall conditioning are located
in the edge plasma, and hydrocarbons are present
from chemical erosion of graphite walls. Low
electron temperature (B/5 eV), low plasma density
(/1017 m 3), and high neutral concentrations
lead to possible growth of macromolecules by
multiple ion-molecule interactions, leading to
nucleation and growth of particles up to 1 mm in
size. Negative ions formed by electron attachment
to the particles are confined in the plasma by the
sheath at the wall [3]. This process is balanced by
photo-detachment from ultraviolet photons to
establish a local, geometry-dependent equilibrium.
Helium discharge cleaning and plasmas used for
thin-film wall conditioning produce a similar
environment for dust production. The amounts
and characteristics of dust formed by this mechanism can be investigated with dedicated diagnostics
in current tokamaks.
Unipolar arcs, often generated during the
plasma startup or rapidly varying plasma currents
in tokamaks, locally deposit significant energy
onto a material surface (e.g. first wall). The
interaction results in melting and vaporization of
the material, a process that liberates large particles
and molten drops. Field interactions move the arc
along the surface, creating tracks of damaged
material. Dust produced from arcing is more likely
to be spherical and composed only of material in
the source because the impact of mixed materials
on growth for this mechanism is negligible.

155

Off-normal plasma events (e.g. edge localized

modes, vertical displacement events, or full disruptions) that deposit enormous amounts of
plasma energy on a surface are known to generate
particulate. Heat fluxes of several GW m 2 for
100 ms have been observed from edge localized
modes in Joint European Torus (JET), and the
The International Thermonuclear Experimental
Reactor (ITER) divertor is designed to withstand
100 GW m2 for 10 ms during disruptions. Rapid,
intense heating of exposed material results in
vaporization and melting. Dust particles may be
created by in-flight condensation of the vaporized
material [4], pressure-driven ejection of melt layer
material [5], and explosive brittle destruction by
heating of gas pockets near the materials surface
[5]. Significant amounts of small dust particles (ca.
B/1 mm) are produced from the violent plasma /
surface interactions, and these particles can agglomerate into larger, more complex structures [4].
The design of next step devices must limit the
number off-normal events to maintain dust inventories at reasonable levels, and/or develop insitu techniques to remove the dust.

3. Safety and operational performance issues

The existence of dust in current fusion plasma
devices is well known and accepted as a matter of
little concern. Only when a few machines ran D-T
experiments did the role of dust become important
because of its ability to retain tritium. It is now
recognized that in next step devices, dust will play
an important role in determining their safety and
operational performance. Much greater plasma
energies and subsequent material erosion rates will
produce quantities of dust much greater than what
is found in machines today. Realization of the
possible public safety consequences from accidental dust mobilization prompted greater attention
to dust in the safety analyses of high energy
density machines (e.g. ITER, FIRE, NIF). Dusts
impact on the operation of such machines remains
uncertain.
Safety concerns of dust in future machines
include radiological hazard, chemical toxicity,
and chemical reactivity [6]. The public and per-

156

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

sonnel radiological hazards derive largely from

activation of structural material by fusion neutrons and retention of the tritium fusion fuel.
Various materials needed in future fusion reactors
contribute to activation, tritium retention, and
chemical reactivity in varying degrees. For example, ITER is proposing to use differing amounts of
tungsten, carbon, and beryllium in different regions as plasma facing components. With tungsten
dust, the greatest concern is its radiotoxicity
because of the high activation of tungsten. For
carbon dust, retention of large quantities of
tritium is problematic. For beryllium dust, its
chemical reactivity with steam leading to the
production of hydrogen is of primary importance,
followed by the chemical toxicity of beryllium
oxide. The specific radiological hazard is determined by the amount of dust mobilized and
transported from the facility in a given accident
scenario. Mobilization and transport of the dust
are determined by properties such as size and
shape. Therefore, accurate safety analyses must
reliably consider dust quantities and morphologies.
Dust is also chemically reactive if combustible
gas is generated during interaction with coolants
or air from a vacuum leak. Reaction rates are
dependent on dust amount, size, exposed surface
area, and temperature. Dust generation mechanisms in tokamaks give rise to particles of large
surface area. High temperatures associated with
some postulated accidents in water-cooled fusion
devices can cause significant hydrogen generation
rates, leading to explosive concentrations. Fig. 2
shows the relative generation rate of hydrogen
versus temperature for beryllium, graphite, and
tungsten dust reacting with steam. Significant rates
are achieved by beryllium at moderate temperatures possible in some accidents. Another chemical
reactivity concern is rapid dust oxidation upon
exposure to air, similar to dust explosions that can
occur in carbonaceous dust in mines and agricultural dust in grain silos. Availability of large
reactive surface areas enhances this reaction.
Determination of dust inventory limits for future
fusion reactors will require careful consideration
of these issues.

Fig. 2. Hydrogen generation rate per gram of 0.5 mm dust

particles reacting with steam at various temperatures likely in
an accident. Be-1 and Be-2 represent two different fits to data.
Adapted from [6].

ITER design project was the first time that

safety concerns of a larger fusion reactor were
addressed at levels appropriate for regulatory
review. Knowing dust will exist in the ITER
vacuum vessel, and recognizing the associated
radiological and chemical hazards, safety analysts
took an approach to ensure public safety by
developing strategies to confine the dust and limit
chemical reactions by limiting the total dust
inventory [7]. The safety approach developed for
ITER is based on assigning administrative guidelines for the maximum tolerable amount of dust
mass at locations inside the vacuum vessel. Two
inventory limits have been defined, the total
amount of dust in the vacuum vessel available
for accidental release is limited due to radiological
and toxicological concerns and the quantity of
dust residing on hot surfaces is limited due to
steam chemical reactivity concerns. Success of this
safety strategy requires development of reliable
methods for dust monitoring and removal. The
administrative guidelines are compared with estimates of the dust production rates for material
type, location, plasma pulse length, and disruption
frequency. Historically, there was a trend established during the ITER Conceptual Design Activity to describe the dust hazard by the cumulative
amount of dust mass in the machine. The initial

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

expectation was that the mass might be limited to

about the kilogram level. This was proven to be
too optimistic, since in a single disruption several
kg of dust may be produced [7]. Even a 10 kg dust
limit seemed an unreasonable constraint on ITER
operation. Therefore, the 100 kg level was introduced early in the ITER Engineering Design
Activity as the dust limit. Distributed over an
area of 200 m2, or roughly the area below the
divertor, this amount of dust would result in a
layer of thickness (assuming 50% dust density
compared with the bulk density of the original
material): tungsten dust, /50 mm; beryllium and
graphite, /500 mm. Layers of this thickness are
generally observable with optical systems. A
comprehensive safety analysis has shown that the
ITER confinement design keeps environmental
releases well below established release guidelines,
e.g. 50 mSv dose limit, used to address hypothetical accidents in the assessment of the ultimate
safety margins. The ITER confinement releases no
more than 50 mSv dose at the site boundary for a
total dust inventory of 100 kg. In addition, an
assessment for 350 kg of tungsten dust (of
significant radiological hazard in ITER) has been
performed to add margin to the uncertainties
associated with the 100 kg limit. The safety reports
of ITER are now based on this higher assessment
value of 350 kg, and therefore, add conservatism
and margin for operation of ITER. Smaller
inventory limits have also been placed on hot
surfaces where hydrogen generation could become
a concern. For example, only 6 kg of beryllium
dust may be on a hot surface such that the
hydrogen combustion limit is not exceeded. These
values are based on observed dust size distributions and specific surface areas measured from
existing tokamaks and plasma gun experiments,
and the extensive chemical reactivity databases for
beryllium, carbon, and tungsten (see, for example,
[6]). These dust limits are based on dust with a
count median diameter (CMD) of 0.5 mm, a
geometric standard deviation (GSD) of 2, and a
specific surface area of 4 m2 g1. Safety factors are
incorporated into the limits to account for uncertainties in the current understanding of dust in
fusion machines.

157

In addition to safety issues, dust may present

operational difficulties in future fusion reactors.
Retention of tritium is not only a safety concern
but also a concern for machine operation because
fuel captured in the walls and in dust is not
utilized. Tritium recovery/clean-up systems have
historically not considered tritiated particulate;
therefore, on-site tritium inventories are greater
than previously estimated.
Another issue arises from dust collecting in gaps
and crevices on an engineered surface, strongly
affecting the performance of the surface. Efficiency of heat transfer from the surface of an
actively cooled component may be greatly reduced
when layers of dust with poor thermal contact are
present. Dust present on windows or view ports
may diminish the effectiveness of optical diagnostics, valve binding is problematic as dust deposits
on valve seats [8], and fouling of cyropumps may
occur since dust is not removed during pump
regeneration.
Particles flaking off of upper tiles during a
plasma discharge could induce a disruption. Experiments have been performed [9] to study the
effect of carbon particles dropped from above
onto a plasma discharge. Although roughly 106
particles with diameters B/2 mm did not affect a
fully developed discharge, particles present during
plasma startup increased impurity radiation. Most
dust particles fall to the bottom of the device
following the plasma discharge [2]. Smaller particles may be re-injected into the plasma by electric
and magnetic forces, creating difficulties for
plasma breakdown and burn through. As the
dust particles are vaporized, the partially ionized
atoms radiate power and significantly increase
plasma resistivity. Greater resistivity requires a
larger loop voltage for startup; the superconducting coils likely to be used on next step machines
may be incapable of the increased voltages [2].
Plasma initiation by the decay of tritium retained in dust particles is another possibility [10],
an effect that complicates controlled plasma
startup. Investigations of these and other issues
of dusts impact on fusion reactor operation are
currently underway on dedicated experiments;
there is much to be gained in developing a detailed
understanding of dust interactions on fusion

158

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

plasmas. Given the potential for large quantities of

dust in next step devices, methods for in-vessel
clearing of dust are also being examined.

4. Comparison of dust from various fusion machines

Dust has been collected and characterized in a
number of fusion research devices, including:
DIII-D [11 /13], TFTR [13,14], Alcator-Cmod
[13,15], JET [16], TEXTOR [2,3], Tore Supra
[17,18], ASDEX-Upgrade [19], LHD [20], and
the NOVA [21] laser facility. Collection opportunities occurred during periods of scheduled maintenance, when the vacuum chamber is vented and
personnel may gain access. In addition, a few
dedicated experiments to investigate dust in fusion
systems have been carried out [22], and several
more are being planned.
Systematic collection routines were used to
sample various locations within these devices,
with specific attention to the lower regions of their
chambers. Experience has shown these areas
collect the greatest amount of particulate during
plasma operation. Positions were selected in all
devices based on the expected mass concentration
due to gravitation settling, proximity to important
structures (e.g. antenna armor or NBI beam
dumps), and significant thermal, magnetic, or
power flux gradients in the vicinity of the surface.
Comparison of particle size distribution, surface
mass concentration distribution, composition and
shape, and specific surface from various machines
area are discussed in this section.

cles that are generally flat (e.g. small flakes) or

fibrous would produce errors in transport calculations. Most dust found in present fusion devices,
however, does not strongly deviate from spherical
surface-to-volume ratio (see Fig. 1). Agglomerates
are the exception, although generally their constituents are also nearly spherical and they do not
constitute a large fraction of total aerosol mass.
A size distribution generated by optical microscopy and count analysis of typical dust from a
fusion machine appears in Fig. 3. This distribution
was obtained with dust collected at the lower
divertor region of ASDEX-Upgrade. The figure
displays measured size frequency data along with
the fitted lognormal distribution; agreement of the
analytical fit is acceptable (linear correlation
coefficient R2 /0.99132). Nearly 5700 individual
particles were sized and counted to build this
distribution, providing a statistically accurate
representation of the underlying distribution of
particles collected on the filter substrate. The large
majority of dust samples analyzed with this
technique yielded nearly lognormal distributions,
although some samples showed what could be
interpreted as bi-modal distributions (two resolvable peaks in size frequency).
Comparison of average dust sizes from different
regions of the various fusion devices is given in

4.1. Particle size distributions

Distribution of the sizes of dust particles captured by filtered vacuum collection from different
fusion devices is generally obtained by optical
microscopy imaging. Sizing and counting a population of particles with the optical microscope
provides a size distribution of particles projected
area equivalent diameters, and moments (CMD
and GSD) of this distribution are reported. Countbased distributions based on projected area measurements do not, however, include the effect of
particle shape. A distribution derived from parti-

Fig. 3. Typical count-based size distribution of dust from a

fusion device. This particular distribution was obtained from
dust collected at the very bottom of the ASDEX-Upgrade
vessel, behind the lower divertor structure.

Machine

Collected
mass (mg)
DIII-D
TFTR
AlcatorCmod
JET
TEXTORc
Tore Supra
ASDEX-Upgrade
LHD
NOVA

Middle regionsa

Lower regions

3.34
32.0
40.2

CMD (mm)9/ Surface mass den- Collected

GSD
sity (mg m 2)
mass (mg)

Upper regions

CMD (mm)9/ Surface mass

Collected
GSD
density (mg m 2) mass (mg)

CMD (mm)9/ Surface mass

GSD
density (mg m 2)

18.5b
67.0
1.73

0.609/2.35
1.609/2.33
1.539/2.80

896b
/
87.0

4.20
/
0.48

0.899/2.92
/
1.229/2.03

8.40
/
66.5

1300
/
595
1300

/
/
1.97
2.54

/
/
2.989/2.94
3.699/2.81

/
/
31.9
55.7

/
/
0.55
1.25

/
/
3.329/2.94
3.599/3.08

/
/
5.33
28.3

58.7
24 800

0.76
3.4

6.319/2.39
0.769/2.03

168
36.8

1.07
4.45

8.739/2.09
0.909/1.93

0.669/2.82
0.889/2.63
1.589/2.80

23.5
/
5470

/
/
14.3
116

279/( /)
5.209/( /)
2.689/2.89
2.219/2.93

0.80
1165

8.599/2.67
1.129/1.90

Data are grouped according to relative poloidal location.

a
Middle regions include inside surfaces of ports */a location where debris can easily settle.
b
This value resulted from shards of glass being collected from a diagnostic port where a plasma probe had broken.
c
Observation of much smaller particulate occurred, but analysis details are not given in [3].

247
48.0

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

Table 1
Comparison of average dust sizes and surface concentrations from various fusion devices

159

160

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

Table 1. The overall average particle size for this

data is 2.89/2.4 mm, however, because of the
lognormal nature of the size distribution dust
particles in fusion devices from 0.5 to 10 mm in
size (by count) may be expected. If mobilized in an
accident, these particles are easily transportable.
The similarity in dust size among these different
machines indicates similar processes are likely
involved in their production, primarily via condensation of material eroded during plasma /surface interactions.
4.2. Surface mass density
The total dust mass collected from individual
locations when divided by the sampled area gives
the surface mass density at that location, and the
product of surface mass density and the total
component area (e.g. vacuum vessel floor area,
port area, etc.) provides an estimate of the
quantity of dust existing at that component.
Summing these dust inventories from various
components roughly indicates the total amount
of dust in the fusion device.
Table 1 includes average values of collected
mass and the associated surface mass density for
several fusion machines. Comparatively large
quantities of dust were collected in the TFTR,
ASDEX-Upgrade and Alcator-Cmod chambers,
with nearly all mass collected in the lower region.
The collection of material from a broken probe
resulted in greater than expected mass concentrations in the middle region of the TFTR vessel.
With the exception of JET, TEXTOR and NOVA,
total sampled areas of the other devices were
similar; hence, the reported collected mass is a
direct indication of the relative amount of dust
among the different machines. These values do
not, however, reflect differences in operating
history, machine configuration, or cleaning efforts
in the devices.
Consider for example the measured dust surface
mass density from regions of ASDEX-Upgrade
[19]. Most of the dust mass was found at the very
bottom of the chamber. No apparent trends are
found for the middle and upper regions, however,
dust deposits in the pumping duct increase at
greater distances from the main chamber. Values

of the total areas associated with each region in

ASDEX-Upgrade are presently being obtained,
thus, an estimate of the dust inventory for this
device does not yet exist. Estimates based on
collection in Tore Supra and DIII-D are 27 and
90/120 g, respectively. Variation in the surface
mass density for several devices is shown in Fig. 4.
The plot shows how CMDs vary with increasing
surface mass density. No statistically significant
trend is apparent */indicating the likelihood that
dust collected at a given location is similar in size
to dust collected at all other locations, regardless
of the collected mass. Large data scatter from
LHD results from the very small amount of dust
collected.
4.3. Dust composition, shape, and tritium content
Dust found in fusion devices nominally shares
the composition of walls that interact with the
plasma. Many of the devices studied to date are
carbon machines, thus, carbon particulate is
pervasive. Metal machines such as Alcator-Cmod
produce metallic dust. Wall surface treatment also
affects dust composition. Experiments performed
following boronization of Alcator-Cmod walls
generated particulate containing significant
amounts of boron. Only a few machines have

Fig. 4. Comparison of dust sizes vs. surface mass density for

various fusion devices. A linear fit is provided from the data to
aid viewing the relative sizes. LHD and ASDEX-Upgrade have
comparatively large scatter in the size data.

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

operated with metal and carbon components (e.g.

JET and ASDEX-Upgrade), and the composition
of dust from mixed material machines has not yet
been studied. ITER will likely utilize various
materials exposed to plasma; experiments will
likely improve the understanding of dust from
mixed materials and provide validation of schemes
developed for monitoring and removing dust from
ITER.
Large variations in particle shape are not
generally observed. Although not necessarily spherical, dust shape is often described as granular or
globular, meaning a particles aspect ratio (maximum to minimum dimension) is relatively low. An
exception is the presence of large, thin flakes on
surfaces affected by co-deposition (usually cooler
regions away from plasma exposure). Upon mobilization (e.g. during collection), these flakes are
normally broken up into fragments with low
aspect ratios [22]. Fibers are rarely observed in
collected particulate, except possibly at locations
where known failures of electrical insulation
occurred. An important concern for transport of
dust with low aspect ratios is the influence of
particle shape on effective drag forces.
Tritium content of dust and flakes collected
from JET and TFTR was measured to estimate the
tritium inventory held in debris. Dust collected
from the inner wall of the TFTR vacuum vessel
contained tritium at a measured concentration of
4.0 /1010 Bq g1, and material (mostly flakes)
from a lower view port window held 2.1 /1010 Bq
g1. At the beginning of DT operation, JET dust
was measured [16] and had a concentration of
1.33 /106 Bq g1, whereas, the flakes had tritium
at a concentration of 8.8 /106 Bq g1. Different
operating histories of the two machines accounts
for the large difference in measured tritium
inventories. Similarity is evident in that tritium
concentrations are roughly the same (within a
factor of 10) in dust and flakes, suggesting that
breaking-up of flakes is a dominant source of dust
particles in these machines. Surface adsorption is
the main route for tritium uptake, and the exposed
area available for uptake in dust is nominally
greater than the area of flakes. If dust were not
predominately from flake break-up, the expected

161

tritium inventory in dust would be (/50 times)

greater than that for flakes.
4.4. Specific surface area
Measurement of the specific surface area (m2
g ) of dust from several fusion devices was
performed using the BET gas adsorption technique. Fig. 5 gives a comparison of the measurements, displaying specific surface area versus mean
volume-surface
diameter
(dMVS /
CMD exp[2.5 ln2(GSD)]). Curves for fully dense
carbon (graphite density of 1.760 g cm 3) and
molybdenum (density of 10.2 g cm 3) are shown
to illustrate that measured dust from carbon and
metal machines generally have some degree of
porosity. With the exception of TFTR, there is
little variation of specific surface area among
carbon machines. Also shown for comparison is
the measured specific surface area of ATJ graphite
dust generated in a plasma gun configured to
simulate disruption heat loads [23]. This data point
fits well within measurements of dust from most
operating fusion devices. Causes for the outlying
TFTR data are not presently known. One factor
that will affect variation in specific surface area
measurements is dust composition. For example,
the smaller measurement from Alcator-Cmod was
obtained with dust from non-boronized walls, and
approaches the value of fully dense molybdenum
1

Fig. 5. Specific surface area of dust and flakes plotted against

the dusts mean volume-surface diameter, dMVS.

162

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163

spheres. The other measurement, four times

greater, was taken following runs with boronized
walls. Such an effect was not found in other
devices, possibly because the effect is greater for
metal and non-metal mixtures (as with AlcatorCmod) or the other machines routinely condition
their walls so that unconditioned wall dust has not
been characterized. Future work that attempts to
characterize the metallic and non-metallic components of the dust would be of great benefit.

5. Directions of future dust work

There remain many questions and uncertainties
regarding dusts impact on future fusion devices.
Research must continue to focus on many aspects
of the role dust plays in the safety and operation of
these devices. Detailed studies on the origination
of the dust must be performed with dedicated
experiments on devices that simulate fusion
plasma conditions or with the existing tokamaks.
Characterizing the effects from mixtures of materials on dust generation requires significantly more
work. Understanding the dust source will improve
modeling for estimates of dust inventory in large
power reactors, reducing the uncertainties in safety
analysis and safety limits, as well as verifying
methods for monitoring and removal. In addition,
benchmark experiments must be performed to
study effects of dust re-suspension and transport
in accidents.
Practical issues of dealing with dust are being
addressed for ITER. In-vessel dust monitoring and
removal techniques are presently under evaluation.
Experiments and prototypes to test such systems
must be built before applying the designs in ITER.
Dust impact on the operation of other next step
devices (FIRE, IGNITOR) have not been considered, although it must be eventually addressed.

References
[1] G. Federici, et al., Plasma /material interactions in current
tokamaks and their implications for next-step fusion
reactors, External Report PPPL-3512, IPP-9/128, January,
2001.

[2] J. Winter, Dust: a new challenge in nuclear fusion research,

Phys. Plasmas 7 (2000) 3862 /3866.
[3] J. Winter, G. Gebauer, Dust in magnetic confinement
fusion devices and its impact on plasma operation, J. Nucl.
Mater. 266 /269 (1999) 228 /233.
[4] J.P. Sharpe, et al., Modeling of particulate production in
the SIRENS plasma disruption simulator, J. Nucl. Mater.
290 /293 (2001) 1128 /1133.
[5] A. Hassanein, I. Konkaskbaev, Macroscopic erosion of
plasma facing and nearby components during plasma
instabilites: the droplet shielding phenomena, J. Nucl.
Mater. 290 /293 (2001) 1074 /1078.
[6] K.A. McCarthy, et al., The safety implications of tokamak
dust size and surface area, Fusion Eng. Des. 42 (1998) 45 /
52.
[7] S.J. Piet, et al., ITER tokamak dust */limits, production,
removal, surveying, Proceedings of the 17th IEEE/NPSS
Symposium on Fusion Engineering, (1998) 167 /170.
[8] H. Tsujikawa, et al., Preliminary studies on large metalsealed gate valves for the International Thermonuclear
Experimental Reactor (ITER), Vacuum 47 (1996) 639 /
646.
[9] K. Narihara, et al., Nucl. Fusion 37 (1997) 1177 /1189.
[10] J. Winter, V.E. Fortov, A.P. Nefedov, Radioactive dust
levitation and its consequences for fusion devices, J. Nucl.
Mater. 290 /293 (2001) 509 /512.
[11] W.J. Carmack, et al., DIII-D dust particulate characterization, INEEL External Report INEEL/EXT-97-00702,
November 1997.
[12] W.J. Carmack, et al., DIII-D dust particulate characterization (June 1998 Vent), INEEL External Report INEEL/
EXT-99-00095, January 1999.
[13] W.J. Carmack, et al., Characterization and analysis of
dusts produced in three experimental tokamaks: TFTR,
DIII-D, and Alcator C-Mod, Fusion Eng. Des. 51 /52
(2000) 477 /484.
[14] W.J. Carmack, et al., Analysis and characterization of
TFTR tokamak dust, ITER Engineering Design File
ITER/US/98/TE/SA-8, April 1998.
[15] S.V. Gorman, et al., C-Mod dust particulate characterization, ITER Engineering Design File ITER/US/97/TE/SA17, August 1997.
[16] A.T. Peacock, et al., Dust and flakes in the JET MkIIa
divertor, analysis and results, J. Nucl. Mater. 266 /269
(1999) 423 /428.
[17] P. Chappuis, et al., Dust characterization and analysis in
Tore Supra, J. Nucl. Mater. 290 /293 (2001) 245 /249.
[18] J.P. Sharpe, P. Chappuis, D.A. Petti, Characterization of
tokamak dust collected from Tore Supra, Fusion Technol.
39 (2001) 1061 /1065.
[19] J.P. Sharpe, Collection and characterization of particulate
from the ASDEX-Upgrade tokamak (July 2000 Vent),
INEEL External Report INEEL/EXT-01-00544, April
2001.
[20] J.P. Sharpe, Collection and characterization of particulate
from the large helical device, INEEL External Report
INEEL/EXT-02-00301, February 2002.

J.P. Sharpe et al. / Fusion Engineering and Design 63 /64 (2002) 153 /163
[21] J.P. Sharpe, W.J. Carmack, D.A. Petti, Characterization
of dust and debris collected from the NOVA facility,
Fusion Technol. 39 (2001) 1066 /1070.
[22] M. Rubel, et al., Dust particles in controlled fusion
devices: morphology, observations in the plasma, and

163

influence on the plasma performance, Nucl. Fusion 41

(2001) 1087 /1099.
[23] J.P. Sharpe, M.A. Bourham, J.G. Gilligan, Generation
and characterization of carbon particulate in disruption
simulations, Fusion Technol. 34 (1998) 634 /639.