4.17

Tungsten as a Plasma-Facing Material

G. Pintsuk
Forschungszentrum Ju¨lich, Ju¨lich, Germany

ß 2012 Elsevier Ltd. All rights reserved.

4.17.1
4.17.2
4.17.3
4.17.3.1
4.17.3.2
4.17.3.2.1
4.17.3.2.2
4.17.3.2.3
4.17.3.2.4
4.17.3.2.5
4.17.3.2.6
4.17.3.2.7
4.17.3.3
4.17.4
4.17.4.1
4.17.4.1.1
4.17.4.1.2
4.17.4.1.3
4.17.4.1.4
4.17.4.1.5
4.17.4.1.6
4.17.4.2

4.17.4.2.1
4.17.4.2.2
4.17.4.3
4.17.4.3.1
4.17.4.3.2
4.17.4.3.3
4.17.4.3.4
4.17.4.4
4.17.4.4.1
4.17.4.4.2
4.17.4.4.3
4.17.5
References

Introduction
Functional Requirements
Material Selection
Fabrication and Microstructure
Advantages and Limitations for Fusion Application
High atomic number: material erosion/melting
Recrystallization
Machinability, mechanical properties, and DBTT
Component fabrication: CTE mismatch with heat sink
Neutron embrittlement
Neutron activation and radiological hazards
Material availability
Tungsten Grades
Influence of In-Service Conditions
Thermal Shock Resistance
Microstructure, composition, and mechanical properties

Power density and pulse duration
Base temperature
Repetition rate
Thermal shock during off-normal events: disruptions
Thermal shock during normal operation: ELMs
Thermal Fatigue Resistance
ITER
Prototype and commercial reactors
Neutron Irradiation
Thermophysical properties and swelling
Mechanical properties
Thermal shock on irradiated W
Thermal fatigue on irradiated W components
Ion Irradiation and Retention
He-irradiation
Hydrogen-irradiation and retention
Combined loading conditions
Conclusion

Abbreviations
APS
AUG
CFC
CTE
CVD
DBTT
DEMO

Atmospheric plasma spraying
ASDEX-upgrade

Carbon fiber composite
Coefficient of thermal expansion
Chemical vapor deposition
Ductile to brittle transition temperature
Demonstration fusion reactor

ECAE
ECAP
ELMs
fpy
FTU
ICRH
IFE

552
553
554
554
555
556
556
556
557
557
558
558
559
561
561
561

562
562
563
563
564
566
566
567
568
568
569
569
570
570
570
571
575
576
576

Equal-channel angular extrusion
Equal-channel angular pressure
Edge localized modes
Full power years
Frascati tokamak upgrade
(Frascati, Italy)
Ion cyclotron resonance heating
Inertial Fusion Experiment

551



552

Tungsten as a Plasma-Facing Material

IFMIF

International Fusion Materials Irradiation
Facility
ITER
Tokamak, Latin for ‘the way’
JET
Joint European Torus (Culham, UK)
LPPS
Low-pressure plasma spraying
MIM
Metal injection molding
NIF
National Ignition Facility (Livermore,
CA, USA)
PFC
Plasma-facing component
PFM
Plasma-facing material
PS
Plasma spraying
PVD
Physical vapor deposition
SC

Single crystal
SPS
Spark plasma sintering
TEXTOR Tokamak EXperiment for Technology
Oriented Research (Ju¨lich, Germany)
TZM
Ti–Zr–Mo
VPS
Vacuum plasma spraying

Symbols
cp
Tm
l
r

The specific heat
Melting temperature
Thermal conductivity
Density

4.17.1 Introduction
Until the mid-1990s, only few fusion devices used
high-Z elements in plasma-facing materials (PFMs).1
These devices either operated at high plasma currents and high plasma densities such as Alcator
C-Mod2 and Frascati tokamak upgrade (FTU)3,4 or
used high-Z materials only as test limiters such as
Tokamak EXperiment for Technology Oriented
Research (TEXTOR).5–9
Since then, high Z refractory metals have been

attracting growing interest as candidates for PFMs
because of their resistance against erosion and the
need for low erosion and stability against neutron
irradiation.10 Considerable effort has been made to
study the behavior of high Z impurities in the core
and edge plasmas, erosion/redeposition processes
at the limiter/divertor surfaces, hydrogen isotope
retention, and on material development and testing.
In particular, the modification of ASDEX-upgrade
(AUG) into a fully tungsten machine,11–17 which was
achieved in 2007, provided positive answers to critical

questions on the reliability of tokamak operation with
high-Z plasma-facing components (PFCs) and the
compatibility with standard and advanced H-mode
scenarios and with the available heating methods.10
Among the challenges, for tokamak devices, that still
remain are the strong increase of the W source and
W concentration resulting from ion cyclotron resonance heating (ICRH) and the need for rigorous modeling to support the extrapolation of current results to
ITER conditions. Clearly, not all questions posed
by ITER can be answered by AUG only. For example,
the effects of material mixing with Be, the melt behavior under transients, or the change of the hydrogen
retention due to damage by high-energy neutron irradiation18 cannot be addressed in AUG. Answers to
some of these issues may be provided by the ITERlike wall project in Joint European Torus (JET), which
is installing a bulk tungsten component for the strike
point and physical vapor deposition (PVD)-W-coated
carbon fiber composite (CFC) tiles for the remaining
parts of the divertor.19–21 The remaining questions
have to be answered by dedicated experiments in
other plasma devices or can only be assessed by modeling. However, the results obtained so far do not

exclude the use of W in ITER as a standard PFM.10
Further investigations related to future fusion power
plants such as demonstration fusion reactor (DEMO)
have to focus on the minimization of plasma heat loads
to the PFCs to increase their lifetime. In particular,
transient heat loads caused by instabilities significantly
decrease the operation domain of PFCs, due to thermal
stresses and consequent enhanced erosion.22 Therefore, it is also important to mitigate all instabilities,
such as edge localized modes (ELMs), that cause significant plasma transient heat losses.23 Plasma scenarios need to be developed, such that the conditions for
achieving the required fusion yield are maintained in
steady state, while at the same time sustaining tolerable
heat loads on the PFCs. The above-mentioned
upgrades to the JET24 and AUG15 will allow further
optimization of the plasma scenarios under these conditions, in particular with DEMO relevant tungsten
PFCs.25 These investigations will show how the identified deficiencies of W can be overcome or how they
have to be dealt with.
In addition to the application of tungsten in ITER
and in potential future tokamak devices such as
DEMO,26–29 tungsten also became an interesting alternative for the divertor of stellarators, for example,
Advanced Reactor Innovation Evaluation Studies –
Compact Stellerator (ARIES-CS),30 and as a first


Tungsten as a Plasma-Facing Material

wall material for inertial fusion devices.31 Due to
similar demands on the PFMs during the operation
of all these devices, similar problems have to be solved
for each application.


4.17.2 Functional Requirements
In the current design of the ITER divertor32–34 for
the start-up phase, tungsten has been selected as
armor for the divertor dome and the upper part of
the divertor vertical targets. In addition, due to excessive co-deposition of tritium in CFC raising regulatory concerns related to tritium inventory limits,
a full tungsten divertor will be installed before the
D–T phase of operation.32
The PFC design for ITER consists of bulk
W bonded to an actively pressurized water-cooled
Cu alloy heat sink. Here W has no primary structural
function. However, due to the operating conditions
listed in Table 1, the PFMs face large mechanical
loads particularly at the interface to the heat sink
material during cyclic steady state heat loads (see
Section 4.17.4.2) and at the plasma-loaded surface
during transient thermal events (see Section 4.17.4.1).
Furthermore, the material response to these loads is
influenced by the material damage or degradation
due to neutron irradiation (see Table 1, Sections
4.17.4.3.3 and 4.17.4.3.4).
Table 1

553

Along with thermally induced loads, the interaction of the PFM with the plasma, that is, the hydrogen isotopes D and T as well as the fusion product
He, is of importance (see Section 4.17.4.4) because
they have an influence on material erosion and nearsurface material degradation.
The further development of the ITER design
led to four conceptual designs for the DEMO
divertor.25,35 These designs include either water

(inlet 140  C/outlet 170  C) or, due to the higher
achievable efficiency, more probably He-cooling
(inlet 540  C/outlet 700  C). In all cases bulk W is
foreseen as the armor material that will have to face
peak steady state heat loads of 15 MW mÀ2 in case
of the water-cooled design and 10 MW mÀ2 for the
He-cooled designs. In contrast to ITER, off-normal
events such as disruptions have to be avoided
completely and transient thermal events during normal operation, for example, ELMs, have to be mitigated below the damage threshold of the material
(see Section 4.17.4.1). This may be particularly
important considering the expected neutron damage
that will amount up to 40–60 dpa during the planned
operation of the fusion reactor35 leading to a significant amount of transmutation products.36 However,
the main limiting factor is expected to be the material’s erosion leading to a maximum lifetime of 2 years
for the divertor armor.35

Operating conditions assumed for the design of the ITER PFCs during D–T operation

Material
Number of replacements
Baking temperature ( C)
Normal operation
Lifetime (number of cycles)
Peak surface heat flux (MW mÀ2)
Peak particle flux (1023 mÀ2 sÀ1)
ELM energy density (MJ mÀ2) controlled/uncontrolled
ELM duration (ms)
ELM frequency (Hz) controlled/uncontrolled
Maximum radiation damage (dpa)
Operation temperature design window during normal operation ( C)

Off normal operation: disruptions
Peak surface heat load (MJ mÀ2)
Duration (ms) rise time/decay time
Frequency (%)

Divertor target

Divertor baffle/dome

CFC/W
3
240

W
3
240

3000–10 000
$10a
$10
0.3–0.5/6–10
0.25–0.5
20–40/1–2
0.7b
200–1000

3000–10 000
3
<0.1
–

–

$4–40
1.5–3/1.5–6
<10

–
–

0.6b
200–600

Source: Federici, G.; Wuerz, H.; Janeschitz, G.; Tivey, R. Fusion Eng. Des. 2002, 61–62, 81–94; Loarte, A.; Saibene, G.; Sartori, R.; et al.
In Proceedings of the 22nd IAEA Fusion Energy Conference, Geneva, Switzerland, Oct 13–18, 2008; IT/P6-13; Raffray, A. R.; Nygren, R.;
Whyte, D. G.; et al. Fusion Eng. Des. 2010, 85, 93–108.
a
Slow transients lasting 10 s up to 20 MW mÀ2 (10%).
b
Without replacement.


554

Tungsten as a Plasma-Facing Material

In comparison to tokamaks, calculations for a
device such as ARIES-CS predict steady state heat
loads between 5 and 18 MW mÀ2.30,37 Similar to
DEMO, a He-cooled W divertor is anticipated with
a maximum heat removal capability of 10 MW mÀ2.

The design limits for neutron irradiation at the shield
of ARIES-CS are up to 200 dpa at 40 fpy (full power
years).38 The component lifetime limits are similarly
dictated by the material’s expected erosion.
Finally, tungsten or more specifically tungsten
coatings find their application also in the dry wall
concept for inertial fusion devices, for example, the
National Ignition Facility (NIF). In future, inertial
confinement devices, thermal loads will occur only in
the form of transient thermal loads (P ¼ 0.1 MJ mÀ2,
t ¼ 1–3 ms, f ¼ 5–15 Hz, Tbase ! 500  C).31 These are
similar to those expected during ELMs and almost
identical to those occurring in an X-ray anode39 and,
therefore, affect a thin surface layer only.

4.17.3 Material Selection
4.17.3.1

Fabrication and Microstructure

Tungsten and tungsten alloys are commercially
available in many forms, for example, as bulk rods,
plates and discs, or thin coatings on various kinds of
substrates. For each of these tungsten products, optimized production routes exist involving mainly powder metallurgical techniques for bulk materials and
PVD and chemical vapor deposition (CVD) as well as
plasma spraying (PS) for coatings. Each of these
processes has its own advantages and disadvantages
as well as an individual influence on the material’s
microstructure and subsequently the material properties. In addition to the fabrication method, the raw
materials, the alloying elements and dopants/impurities, pre- and postthermomechanical treatments,

and the final shape/geometry have a strong impact
on the achieved microstructure.
Focusing on the powder metallurgy fabrication
route, tungsten powder is obtained from ammonium
paratungstate ((NH4)2WO4), tungsten oxide (WO3),
and tungsten blue oxide (WO3Àx) by hydrogen reduction at temperatures in the range of 700–1100  C. Various grain sizes can be produced depending on the
reduction temperature and the hydrogen dew-point.
The purity of the metal powder obtained is >99.97%.
In the manufacture of doped or alloyed tungsten products, the dopants or alloying elements are either
introduced into the raw materials before reduction or
they can be added to the metal powder after reduction.

Following the reduction stage, the powder is
sieved and homogenized. The initial densification
of the powder in various plate and rod geometries
takes place predominantly through die pressing and
cold isostatic pressing. The pressed compacts are
subsequently sintered at temperatures between 2000
and 2500  C (2273–2773 K), mostly using furnaces
with hydrogen flow. This increases the density and
the strength of the pressed blanks.40
After sintering, the products have a rather low
density of about 80% of the theoretical value and
poor mechanical properties. To increase density and
improve mechanical properties, the sintered products
are subject to a mechanical treatment such as rolling,
forging, or swaging at temperatures up to 1600  C.
Intermediate annealing, leading to recovery and
recrystallization, is necessary to maintain sufficient
workability. The working temperature can be

reduced as the degree of deformation increases. In
this way, forged parts such as rods and discs as well as
sheets and foils are produced.40
The final step, that is, the mechanical treatment,
changes the microstructure from isotropic with grain
sizes determined by the initially used powder size
into anisotropic. Depending on the deformation
method, the grains may show either:
 an elongated, needle-like structure along the
deformation direction for radially forged rods
and uniaxially rolled plates (see Figure 1(a)), or
(a)

200 mm
(b)

200 mm
Figure 1 Light microscopy images of etched
cross-sections of (a) a deformed rod and (b) a rolled plate.


Tungsten as a Plasma-Facing Material

 a flat disc-shaped structure for axially forged discs
or blanks and cross-rolled plates (see Figure 1(b)).
In addition to bulk materials, research and development is also directed on tungsten coatings. One possibility would be the plasma-spraying process, in which
powders are injected into a plasma flame, melted, and
accelerated toward the (heated) substrate. The deposited layers are splat-cooled, leading to a flat discshaped microstructure. Depending on the atmospheric
conditions, the result may be layers with high porosity
and oxygen content (water stabilized and atmospheric

plasma spraying, APS, see Figure 2(a))41,42 or low
porosity and good thermal contact (low-pressure or
vacuum plasma spraying, LPPS/VPS).26,43–47
In contrast, PVD and CVD coatings show a
columnar structure perpendicular to the coated substrate with grain sizes in the range of the coating
thickness (see Figure 2(b)). PVD coatings, which
are also used as thin intermediate layers below a
plasma-sprayed tungsten top layer,48 are deposits of
tungsten vapor on the substrate surface, which is in the
source’s line of sight.43,49 CVD coatings are reactions
of a W-containing gaseous phase and have the ability
to coat complex geometries.6,50–52 In both cases, a high
density ($100%) of the coatings is achieved.
The coated substrate can be graphite as used for
AUG (PS),12,13 CFC as used for the ITER-like wall

(a)

project in JET (PVD),21,53–55 or copper and steel as it
might be used for first wall applications in future
fusion devices (PS, PVD, CVD).44,49,50,56–60
4.17.3.2 Advantages and Limitations for
Fusion Application
For fusion plasma-facing applications, the most
essential properties are thermal conductivity,
strength and ductility, thermal shock and thermal
fatigue resistance, structural stability at elevated temperature, and stability of the properties under neutron irradiation. The advantages and disadvantages of
tungsten for these conditions are manifold and
opposed to each other as shown in Table 2. While
the advantages of the material are mainly related

to its high temperature-handling capability, the
limitations are associated with manufacturing and
handling at low temperatures (below ductile to brittle
transition temperature, DBTT61–63), plasma compatibility including neutron irradiation, and radiological issues.
However, with regard to other potential PFMs, for
example, Be (see Chapter 4.19, Beryllium as a
Plasma-Facing Material for Near-Term Fusion
Devices), CFC (see Chapter 4.18, Carbon as a
Fusion Plasma-Facing Material), and Mo, tungsten
is still the most promising, offering an advantageous
combination of physical properties and, therefore,
has become the material of choice for ITER and
DEMO. Since this decision was made, R&D efforts
for investigating newly developed tungsten grades

Table 2

50 mm
(b)

Disadvantages

 High melting point
 Low erosion (high

 High Z (low allowed






Figure 2 Light microscopy images of etched crosssections of (a) atmospheric plasma spraying W and (b)
chemical vapor deposition W on a graphite substrate.

Features of W armor materials

Advantages



300 mm

555

energy threshold for
sputtering)
High thermal stress
resistance
High thermal
conductivity
Low swelling
Low tritium retention

concentration in plasma)

 Potential loss of melt layer











during transient events
Recrystallization
Poor machinability
High DBTT
High CTE mismatch with Cu or
stainless steel heat sink
Neutron embrittlement
Irradiation-induced
transmutation
High radioactivity (short-term
waste, decay afterheat)
Explosion dust potential
Limited resistance to grain
growth


556

Tungsten as a Plasma-Facing Material

and alloys that are able to overcome or at least mitigate some of the above-mentioned disadvantages
have significantly increased.
4.17.3.2.1 High atomic number: material
erosion/melting


As the high atomic number is an intrinsic material
property that cannot be changed, the only possibility
to avoid plasma contamination by tungsten is to adapt
to the loading realities, that is, thermal loads and
plasma wall interaction conditions, and the energy of
the incident plasma particles. In particular, surface crack
formation, loosening of particles, and particle ejection or
melting are addressed (see Section 4.17.4). Concerning
the latter, the addition of suitable alloying elements
or dispersoids (see Section 4.17.3.3) reduces the
material’s thermal conductivity causing a reduction
of allowed applied heat fluxes. From this point of view
only low-alloyed grades should be considered and the
best grade is tungsten of high purity.
4.17.3.2.2 Recrystallization

Recrystallization is a thermally activated process.
Therefore, it is expected that the activation energy
of nucleation is dominated by small angle grain
boundaries. The activation energy of grain growth is
dominated by large angle grain boundaries.64 The
temperature of recrystallization depends mainly on
the deformation history, that is, the higher the degree
of deformation, the lower the recrystallization temperature,65,66 and the chemical purity. When heated
above the recrystallization temperature, the structure
of tungsten is altered due to grain growth causing an
increase in DBTT and reducing other mechanical
properties, that is, strength and hardness.67
There are several possibilities for increasing the

recrystallization temperature. Particle reinforcement
and controlled formation of porosity are the best and
most investigated options.68 For example, the higher
recrystallization temperature of dispersion strengthened alloys results from the interaction between
dispersoids and dislocations during hot-working;
the higher the amount of hot-work, the finer are the
dispersoid particles and the higher is the recrystallization temperature. During recrystallization, these
particles prevent secondary grain growth and consequently, the recrystallization temperature of dispersion strengthened alloys may increase compared to
pure W.67 Another example is highly creep-resistant
doped/nonsag materials with aligned porosity acting
as obstacles for dislocation movement as they are
used in the lighting industry.69

Experience shows that incomplete recrystallization often helps to achieve the desired balance in
material properties. If the operating temperature
is well known, controlled recrystallization during
application might be feasible as well.67 However,
for operational conditions in nuclear fusion devices,
it is expected that the very high thermal strain rates
experienced in the thin layer heated by plasma disruption or any other transient thermal event will
significantly affect the material’s microstructure and
properties.
4.17.3.2.3 Machinability, mechanical
properties, and DBTT

Mechanical properties of W strongly depend on variables such as production history, alloying elements,
impurity level, thermomechanical treatment, and
form of material. Depending on the production history and heat treatment, W and W-alloys could have
anisotropic mechanical properties. This is expressed
by showing significantly better properties in the

direction of elongated grains (by rolling, forging, or
due to deposition processes for coatings) but poorer
properties in other directions.70 While reported data
on single crystals (SCs) (e.g., Gumbsch62) and for
isotropic materials (e.g., Kurishita et al.71) give a
clear indication of the material’s performance, typically the reported data refer to the best orientation of
the material as shown for fusion relevant tungsten
grades in numerous publications.44,51,57,72–81 The properties in other directions, particularly the DBTT, could
significantly differ.76 This will affect the operational
performance, which is reflected by the orientationdependent thermal shock response.82
Tungsten is a body-centered cubic (bcc) refractory metal, with a comparatively low fracture
toughness,61,83 high DBTT, and poor machinability,
which is directly correlated to the material’s low
ductility and low grain boundary strength.67 However, DBTT is an ill-defined property and depends
strongly on purity, alloying elements, thermomechanical treatment, and, most essentially, the
testing/loading conditions due to its deformation
rate dependence.62,63 The obtained values vary
over a broad temperature range from room temperature (RT) to several hundreds of degrees Celsius.
The exact value depends on the stress state, for
example, a three-dimensional state of stress in the
sample leads to a lower DBTT.
Although many other parameters influence the
fracture of bcc metals, the DBTT is usually associated with the thermal activation of dislocation


Tungsten as a Plasma-Facing Material

kink pairs. Below this characteristic temperature
the separation of a screw dislocation into three partial
dislocations (which cannot easily recombine and

are therefore more or less immobile) is responsible
for the brittle behavior. Increasing temperature leads
to thermal activation of the kink mechanism and
increased ductility due to shielding of the crack
tip.84 There is an empirical correlation between temperature and activation energy for brittle-to-ductile
transitions in single-phase materials suggesting that
the ratio between the activation energy and the
DBTT gives approximately a value of 25.63
Another factor is the occurrence of interstitial
solute elements, such as oxygen, carbon, and nitrogen, which even in very small amounts tend to
segregate at grain boundaries, promoting intergranular brittleness and increasing the DBTT. Two ways
can be used to get rid of or mitigate the negative
effects of interstitial impurities: either a reduction of
the grain size,84 to dilute their effect on a larger
grain boundary surface, or the complete elimination
of grain boundaries, as in SCs. The development
of W-alloys essentially follows the first route, as
the SC technique, although effective, is too costly.
The conventional method to decrease the grain
size of tungsten or tungsten alloys is to deform the
material at an intermediate temperature, above
the DBTT and below the recrystallization temperature.81,84–86 The formation of oxides and carbides
of the alloy constituents helps to stabilize the grain
boundaries and to dispersion strengthen the matrix
at high temperature. Recently, mechanical alloying
followed by powder densification has been applied
to refractory alloys. Materials with a stabilized
fine-grained structure and with the grain boundary
strengthened by even finer dispersoids of TiC
improve the low-temperature impact toughness

of refractory alloys, leading to increased ductility
even down to RT and create superplasticity at high
temperatures.71,87–89
Another reliable method to increase the ductility
at low temperatures and therefore reduce the DBTT
is to alloy tungsten with the rather expensive element
rhenium, which is a substitutional solute in the
W lattice.67,83
As mentioned before, both material deformation
and heat treatment influence the DBTT. A heat
treatment slightly below the recrystallization temperature is able to significantly reduce the DBTT.
In contrast, annealing above the recrystallization
temperature reduces strength and hardness and
increases the DBTT.67

557

4.17.3.2.4 Component fabrication:
CTE mismatch with heat sink

A mismatch between the coefficients of thermal
expansion (CTEs) can lead to thermal stresses at
the interface, which are detrimental to the component lifetime. This can occur with either Cu-based
alloys or steels (steel is more likely to be used in case
of coatings) such as that used for water-cooled
designs, or to W and W-alloys in the He-cooled
design. In particular W and W alloys, in the coldworked and stress-relieved condition, tend to delaminate in the direction parallel to the direction of
deformation. Such delamination can occur during
machining or during operation. To avoid failure due
to delamination, the orientation of the texture has to

be perpendicular to the surface of the joints,90 raising
the question of the suitability of plasma-sprayed
W coatings. Two possible options are recommended
to mitigate the thermal stresses, that is, reducing the
joint interface by introducing castellations or using
smaller tiles,91–93 or introducing soft and chemically
stable interlayers94,95 or graded layers.96–101
Despite the fact that surface finish has no direct
effect on the performance of ITER-related components,94 it is recommended to avoid possible crack
initiators in the armor design, such as castellations
ending in the tile and to ensure accurate surface
finishing.102–104 Designs that have been proven to
reduce the tile and interface thermal stresses and to
extend the component lifetime beyond the design
limits are the macrobrush or the monoblock. The
latter is the reference design for ITER105 because it
provides the most reliable attachment and therefore a
reduced risk of catastrophic cascade failure.106
Finally, the thermal treatment of W during joining
manufacturing cycles might have an influence on
the material’s properties. While the process temperatures during joining of W and Cu do not lead to any
significant change of the W properties, in the case of
high-temperature brazing of W to W alloys for the
He-cooled divertor design,102 the recrystallization
temperature of W has to be taken into account.
4.17.3.2.5 Neutron embrittlement

There are little data available for irradiated tungsten
(see Chapter 1.04, Effect of Radiation on Strength
and Ductility of Metals and Alloys). Based on

results for other refractory alloys and limited data
on tungsten, one would expect neutron irradiation
to increase the strength and decrease the ductility of
the tungsten armor largely through increases in the
DBTT. To minimize the neutron-induced material


558

Tungsten as a Plasma-Facing Material

degradation, it is reasonable to limit the operational
conditions for components in a neutron environment
to temperatures above $900  C where recovery of
tungsten takes place,107 as the ductility loss is more
pronounced below about 0.3 Tm. This is possible in the
region close to the plasma-facing surface, but it is
impossible in the heat sink region as tungsten will be
in contact with materials that cannot operate at this
temperature and stress concentrations in these ‘cold’
areas have to be avoided.108 In the case of ITER, Cu
will be employed in the heat sink while steel is more
likely to be used in DEMO, which has a higher
operating temperature.29 Hence, a greater understanding of the irradiation response of tungsten at
temperatures between 700 and 1000  C is
needed.109,110 The effect of embrittlement is alleviated when operating above 250  C, although in the
presence of He (produced by transmutation reactions)
somewhat higher temperatures may be required.83
Although at intermediate temperatures (0.3–0.6 Tm),
void swelling and irradiation creep are the dominant

effects of irradiation, the amount of volumetric
swelling associated with void formation in refractory
alloys is generally within engineering design limits
(<5%) even for high neutron fluences ()10 dpa).
Very little experimental data exist on irradiation
creep of refractory alloys, but data for other bcc alloys
suggest that the irradiation creep will produce negligible deformation for near-term reactor applications.110
4.17.3.2.6 Neutron activation and
radiological hazards

The activation and transmutation of tungsten as
a PFM is a critical issue, particularly concerning
long-term storage and recycling times. Different
studies on activation issues have been performed.
These comprise the analysis of cross-sections for
high-energy neutrons,111,112 studies on the heliumcooled lithium lead divertor for DEMO,113 the inertial
fusion devices,109 other benchmark experiments,114,115
and modeling issues, for example, on the self-shielding
ability of tungsten.116,117
Furthermore, it was shown that the long-term
activation behavior is dominated by activation products
of the assumed material impurities while the shortterm behavior is due to the activation of the stable
W isotopes.113 For a short period of a few weeks, the
latter causes a huge amount of decay-induced afterheat
that has to be removed by continued active cooling.67
On the other hand, the accumulation of the highly
radioactive transmutation product 186mRe was determined to be most critical, limiting the component

lifetime to a maximum of 5 fpy when using pure W or
to 2 fpy when using Re-doped W before the limits for

storage by shallow land burial could be exceeded.109
The dose rate limits for recycling after different
applications are expected to be reached within
5 years115 to 50 years113 of storage or up to 75 years
after end of plant life.118 Fischer et al.113 take a limit of
100 mSv hÀ1 for remote handling into account, which
might be a problem at the times when maintenance
operations would be in progress. Taylor and
Pampin118 give a value of 20 mSv hÀ1 as the limit for
allowing tungsten to be categorized as a recyclable
material. The hands-on limit for tungsten should be
achieved after about 200 years.115
Besides waste management, tungsten has also been
investigated and evaluated according to characteristic
radiological hazards that might occur when using it as
PFM in tokamak fusion reactors. It was found that the
tritium permeation into tungsten does not, in contrast
to CFC, appear to be a major problem. However, due
to neutron activation, the mobilization of activation
products, for example, by forming volatile oxide species in the presence of steam and air, has to be limited
by establishing shutdown requirements to avoid
melting of tungsten in case of an accident. The potential exposure from mobilized activation products
from the tungsten divertor may be modified by varying the operating conditions of fusion power and
change-out time as well as the thickness of the
divertor armor. The dose can be reduced by selecting
shorter change-out times. However, the total lifecycle waste volume will be increased accordingly.
A thinner divertor will produce less mobilized activation products while suffering a more restrictive
shutdown requirement.119
4.17.3.2.7 Material availability


The quantity of W needed for the PFCs in a fusion
device such as ITER or DEMO represents only a
small fraction of the yearly production and the
world’s reserves120 and its production can be easily
satisfied by existing industrial capabilities. The same
point is valid for stellarators and even more for inertial fusion devices, which only work with thin coatings. However, the issue of component lifetime has to
be taken into account. Depending on the component
lifetime, the recycling rate, and the storage time until
a hands on level is achieved (see Section 4.17.3.2.6),
the operation of numerous power plants may require
an amount of tungsten that exceeds what is currently
available from the market.


Tungsten as a Plasma-Facing Material

4.17.3.3

Tungsten Grades

Within current R&D programs for the selection
and characterization of candidate grades of W and
W alloys for fusion applications, many materials produced according to the schemes outlined above were
investigated. These are discussed in the following section, which introduces some of their characteristics.
The manifold production processes described below
for pure W are also applicable to W alloys.
Pure tungsten (undoped)
 Sintered W is the most readily available and
cheapest grade with a grain size that depends
on the initially used W powder. However,

it is characterized by high porosity, low recrystallization temperature (1000–1200  C), and
low strength at elevated temperature.96 The
option of improving the sinterability by adding small amounts of activators (Ni, Fe)121
increases the radiological hazard due to additional activation products that have to be taken
into account.119
 Forged or swaged W offers an increased density
and a refined microstructure compared to sintered material, resulting in higher ductility and
mechanical strength. Forging and swaging are
therefore the industrial production processes
that are typically applied not only for pure
tungsten but also for most kinds of tungsten
alloys (see below). This grade of W is manufactured in block shape or more commonly
in the form of rods with different diameters
( 90 mm)40 showing an anisotropic microstructure122 with elongated grains along the
axial direction and an increasing grain size
and porosity with increasing rod diameter.
Thus, increasing rod diameter leads to a
decrease in mechanical strength and ductility.
For the production of monoblock tiles, such as
those planned for ITER, rods with a minimum
diameter of 30–35 mm are necessary.
 Rolled W is applied in the form of plates or foils
with thicknesses from 0.02 to 20 mm.40,123,124
It offers a densified but layered microstructure
that is strongly anisotropic, with flat disc-shaped
grains parallel to the rolled surface affecting the
mechanical properties (see Section 4.17.3.2.3)
and resulting in the risk of delamination.
 Double-forged W is in the form of blanks with a
diameter of 140 mm and a height of 45 mm. The

double-forging process, first in the radial and
then in the axial direction, provides a more











559

isotropic microstructure than it is generated by
single forging. This material should act as a
reference grade for establishing a reliable materials database for finite element calculations.82
SC W provides higher ductility than polycrystalline W, higher thermal conductivity, lower
neutron embrittlement, higher thermal fatigue
resistance, and a more stable structure at elevated temperatures. The disadvantages are high
cost and low industrial availability.96,125,126
Metal injection molded (MIM)-W127–129 provides
a dense and isotropic microstructure with
grain sizes on the order of the powder particle
sizes used. A final densification by hot isostatic
pressing (HIP) at temperatures >2000  C leads
to an improvement of the mechanical properties; recrystallization and grain growth do not
play a role. Furthermore, the production process offers the possibility of net shaping.
Spark plasma sintered (SPS)-W and resistance sintering under ultra-high pressure.130–132 The material is characterized by a short fabrication time

of only a few minutes keeping the initial fine
microstructure determined by the powders
used. The finer the grain size, the higher the
microhardness and the bending strength but also
the lower the achievable density. The application of alternatively uni-, two-, or threedirectional orthogonally applied forces for the
material’s densification during the process leads
to internal stresses, which have an influence on
the recrystallization behavior. Recrystallization
and grain growth occur at $1500  C. Depending on the amount of porosity, the finer the
initial grain size of tungsten, the smaller is
the grain growth.
Severe plastically deformed W (and W alloys, see
below) with ultra-fine grains in the nm range
are produced by either high-pressure torsion
at 400  C84,133 or by the equal-channel angular
extrusion or pressure (ECAE or ECAP) process at high temperatures (1000–1200  C).134
The material shows stable, that is, deformationindependent, recrystallization temperatures and
exhibits considerably enhanced ductility and
fracture toughness.61,85,86,135,136
Plasma-sprayed W involves, in general, application
of VPS, more precisely also called low-pressure
plasma spraying (LPPS), which provides a
significantly reduced oxygen content and
improved thermophysical properties compared to atmospheric (APS) or water-stabilized


560

Tungsten as a Plasma-Facing Material


plasma spraying.42 However, LPPS-W is typically characterized by a lower thermal conductivity (up to 60% of bulk tungsten is
reported67) and a lower strength than bulk
W particularly when deposited on large surfaces. The recrystallization temperature is
similar to pure W.48,137 Although the thickness of the plasma-sprayed coatings required
for fusion applications are flexible, coatings with 200 mm or thicker are commonly
produced.26,43,67,138,139 Furthermore, PS is the
only production method that offers the possibility to produce and repair W components.57,60,96
 CVD W provides a microstructure with a
columnar grain structure parallel to the surface, high thermal conductivity similar to bulk
W, and a very high density and purity.6,140,141
Thicknesses up to 10 mm were produced,67 but
its high cost is a significant drawback for practical applications.52,96
 PVD W provides a featureless structure that is
extremely dense and pore free. In contrast to
plasma sprayed and similar to CVD-coatings,
the deposition rates are low. Economic and
process-related restrictions generally limit the
deposited W thickness to 10–50 mm.13,54,55,67,142
 W foam for Inertial Fusion Experiment (IFE)
applications provides structural flexibility during quasivolumetric loading. The material is
microengineered with a relative density of
$21% and can be simultaneously optimized
for stiffness, strength, thermal conductivity,
and active surface area.143
Tungsten alloys
 Oxide dispersion strengthened W alloys such as
W–La2O3, W–Y2O3, and W–CeO2 with oxide
additions 2% are processed by powder metallurgy methods similar to pure W.40 The insoluble dispersoids, which are influenced in shape
and distribution by the thermomechanical
treatments during the production process,73,144

improve the grain boundary strength and
machinability and play an important role in
controlling recrystallization and the morphology of the recrystallized grains.68 This results
in a higher recrystallization temperature by
100–350 K by suppression of secondary grain
growth (i.e., grain boundary migration), lower
grain size, higher strength after recrystallization, and better machinability than sintered
W even at RT. This permits fabrication at











lower costs.67 The size of the dispersoids in
commercially available alloys is $10 mm; however, research on mechanically alloyed materials using submicron dispersoids is currently
being performed.145 However, the presence of
oxide particles with a melting temperature
below those of tungsten has a negative effect
on the erosion resistance.146,147
W–3–5% Re is, compared to sintered pure W,
characterized by a higher recrystallization temperature and strength even after recrystallization,148 better machinability, and improved
ductility at low temperatures.67 The addition
of Re, which has a high solubility in W, however, reduces thermal conductivity, increases
embrittlement after neutron irradiation, and

significantly increases the cost and safety concerns because of the high Re activation under
neutron irradiation.96
W–1–2% Mo (þY and Ti) cast alloy. The addition of Mo and the reactive elements Y and Ti,
which reduce the amount of free oxygen and
carbon and form obstacles to grain growth,
improves the mechanical properties compared
to large grained pure cast W.67,73
W–TiC produced by mechanical alloying and
slow deformation techniques provides, similar
to all other W alloys, higher strength and recrystallization temperature, better machinability,
and improved ductility compared to pure W
with superplastic behavior at temperatures of
1400–1700  C.89 The addition of Ti-carbide particles stabilizes the grains during the material’s
production process. This generates an isotropic
grain structure and has the additional effect
of keeping a fine grain structure even in the
recrystallized condition, but the alloy is more
expensive. After recrystallization, the finer
dispersoids of TiC particles improve the lowtemperature impact toughness of refractory
alloys following low-dose neutron irradiation.71,87–89,149–154 Other carbides, for example,
ZrC155,156 or HfC (in combination with Re and
Mo),96 can be used instead of TiC.
K-doped W is a nonsag material that contains a
maximum of 40 ppm of potassium.40 Originally
known from the lighting industry, it provides high
creep strength due to its aligned pore structure,
high recrystallization temperature >1600  C,
and good machinability.68,77,78,157
W–Si–Cr as a ternary or even by the addition
of another element as a quarternary alloy is a



Tungsten as a Plasma-Facing Material

newly developed and not yet optimized material that is being investigated as a wall protection material due to its favorable oxidation
resistance, preventing excessive material erosion in case of accidental air ingress.158,159
 Severe plastically deformed W alloys offer, similar
to pure tungsten (see above), significantly
improved fracture toughness and ductility.61,84
The addition of alloying elements to the starting
material (any developmental or commercial
produced W alloy), such as Re or dispersoids,
leads to an increasing stability of the grains and
therefore a higher recrystallization temperature and less grain growth.133
Any of the bulk materials mentioned above could be
used and are being investigated in its cold-worked,
stress-relieved, or recrystallized state. The latter is
of particular interest due to in situ recrystallization of
surface near regions during operation.108
In spite of the fact that a large variety of tungsten
grades and alloys already exist, the attempts to further optimize these materials are ongoing. The fabrication and successful testing of He-cooled divertor
mock-ups for DEMO and ARIES-CS102,160 under
a heat flux of 10 MW mÀ2 are important driving forces
for the present development of W alloys with improved
performance in the fusion environment.25 However,
R&D has to address many different issues related
to the performance of the material when exposed to
thermal loads, neutron irradiation, and the plasma;
these will be discussed in the following section.


4.17.4 Influence of In-Service
Conditions
4.17.4.1

Thermal Shock Resistance

Tungsten-armored PFCs will be subjected to different types of heat fluxes dependent on their field
of application (see Section 4.17.2). Among others,
this includes thermal transient loads (e.g., ELMs
and disruptions). The behavior of the material
under these conditions, that is, the combination of
cyclic steady state and transient heat loads, is a key
factor that has to be considered for the selection of a
suitable grade of W.
The machines simulating these operational
conditions are electron and ion beam facilities, quasistationary plasma accelerators, plasma guns, and
high-energy lasers. A most critical issue is the comparability of such simulations. Therefore, a round robin

561

test involving some representative facilities was made
for investigating the influence of the different time
regimes and different power density levels. The results
showed that when compared on the basis of a heat flux
parameter P (MW mÀ2 s1/2), which is directly proportional to the temperature increase, the cracking and
melting thresholds are almost identical. This permits
a direct transfer of the qualitative results obtained
in any of these facilities.161 In contrast, quantitative
results representative of the operational conditions in
large fusion devices can only be obtained when the

loads are applied in the desired time range. The reason
for this is the heat penetration depth and the related
stress field that is produced, which influences crack
and melting depth.
There are several parameters influencing the
thermal shock behavior of tungsten that will be discussed in the following sections for the different
materials under disruption and ELM-like loads.
4.17.4.1.1 Microstructure, composition, and
mechanical properties

During thermal shock loads, steep temperature
gradients of hundreds to several thousand degrees
Celsius on a length scale of millimeter or even
micrometer (depending on the pulse length) are
formed, influencing only a limited volume near the
loaded surface. While the heat load is applied, due to
thermal expansion and the decreasing strength of the
material at the surface compared to the bulk material,
compressive stresses are formed in the surface plane.
These stresses can lead to permanent plastic deformation that might, during cool down, generate tensile
stresses sufficiently high to initiate crack formation
perpendicular to the surface and thereby cause stress
relaxation at the surface.
Depending on the mechanical properties in the
surface plane, the amount and starting point of
crack formation can be influenced. Based on this
and on the fact that the mechanical properties are
strongly dependent on the material’s microstructure
(see Section 4.17.3.2.3), a grain orientation parallel
to the surface and therefore high strength in the

surface plane might be preferred.162 However, grains
oriented parallel to the surface, such as in rolled
materials or plasma-sprayed coatings, might result in
delamination (see Figure 3(a)), which causes overheating and subsequently surface melting if they have
a lower strength in the depth direction and exhibit
preferential cracking along the weak grain boundaries.
Therefore, a grain orientation perpendicular to
the surface and parallel to the direction of the heat


562

Tungsten as a Plasma-Facing Material

(a)

200 mm

(b)

100 mm

500 mm

Figure 3 Light microscopy images of the etched
cross-sections of thermal shock–loaded specimens with
grains oriented (a) parallel and (b) perpendicular to the
loaded surface; cracks follow the grain orientation/
deformation direction.


Figure 4 Light microscopy images of the etched
cross-sections of thermal shock–loaded metal
injection molding tungsten with isotropic grain
structure.

flow is recommended.90 This will cause cracks to
form along the grain boundaries toward the cooling
structure (see Figure 3(b)) causing no degradation or
only a negligible degradation of the material’s thermal transfer capabilities. Due to the lower mechanical properties in the surface plane, more or larger
cracks will form during thermal shocks, running perpendicular to the surface and following the grain
orientation.
In contrast to deformed materials, crack formation
and crack orientation in materials with isotropic or
almost isotropic grain structures, for example, MIM-W
or recrystallized W, is rather unstable and is strongly
enhanced for the weakened recrystallized material.
Depending on the applied power densities, the formed
temperature gradient, and the resultant stress fields
within the material, cracks initially running perpendicular to the surface might deflect at zones with
compressive stresses and keep running parallel to
the surface (see Figure 4).

operation) of the material, the damage and cracking
threshold are determined mainly by the material’s
mechanical properties. Damage here means that the
material’s surface has undergone a visible and measurable modification, for example, by surface roughening,
recrystallization, or pore/void formation.

4.17.4.1.2 Power density and pulse duration


The material’s response is strongly related to the
applied temperature fields and by this to the absorbed
power density and the pulse duration. This results in a
material-related surface temperature increase and
heat penetration depth.163 A classification of the
impact of the temperature field is made by establishing
three parameters: the damage, the cracking, and the
melting threshold. While the latter depends on the thermal conductivity and the melting temperature (for
alloys or mixed materials formed during tokamak

4.17.4.1.3 Base temperature

The base temperature influences the thermal shock
behavior in two ways. First, a higher base temperature influences the damage, cracking, and melting
threshold. All of them are essential because they
limit the operational conditions and when exceeded
cause enhanced material degradation. Therefore, lifetime estimates based on RT data will yield unrealistic
conclusions.
Second, crack formation strongly depends on
the plastic deformation at high temperatures and
even more on the stress developed during cool
down. To understand the influence of a higher base
temperature, one has to be aware of the typical shape
of the yield and tensile strength curve for W or a
W alloy.105,157 While the decrease in strength is rather
high at low temperatures, the curve flattens at high
temperatures despite a drop in strength when exceeding the recrystallization temperature. As a result, the
high temperature plastic deformation induced by the
combination of a heated surface and ‘cool’ base material
can be significantly reduced by a small increase in

base temperature. Combining this effect with the
increased ductility of W at the given base temperature,
brittle crack formation can be avoided when heating


Tungsten as a Plasma-Facing Material

the material above a certain threshold.82,157,164,165
This temperature threshold is related to the DBTT
but is not necessarily identical to it.
4.17.4.1.4 Repetition rate

In addition to the parameters mentioned above, the
damage, cracking, and melting thresholds are determined by the number of load repetitions, because of
continuing material degradation such as hardening
and recrystallization. This is of particular interest
for short transient events with a high repetition
rate in magnetic (ELMs) and inertial fusion devices.
Up to now the simulation of submillisecond events
(ELMs, IFE) has been performed only up to a relatively low number of cycles; large numbers of pulses
(e.g., >106 ELM pulses during the life-time of the
ITER divertor) are not feasible in the majority of
the above-mentioned test facilities.
4.17.4.1.5 Thermal shock during off-normal
events: disruptions

Disruptions still occur frequently in operating
tokamaks, and therefore they are also expected for
ITER with an anticipated occurrence in 10% of
the ITER pulses (3000 pulses per expected component lifetime). During a disruption in which the

plasma undergoes a partial or full thermal quench,
most of the plasma thermal energy will be dumped
on the divertor plates.166 Taking into account the
resultant loading conditions (see Section 4.17.2),
significant material loss from the tungsten plasmafacing surface should occur by melting and evaporation particularly in the dome area.167,168 In simulating
these events, the amount of melting, the melt motion
and subsequent roughening of the surface, the material erosion by droplet emission, the resolidification
behavior, and finally, the crack formation occurring
in the loaded area or at the boundary between melted
and unmelted zone are the most important parameters to be determined.
The underlying mechanisms for the abovementioned material degradation are well described
(see Figure 5).169 Thermal loading of tungsten and
metals, in general, at ‘moderate’ energy densities (up to
a few MJ mÀ2) will result in a homogeneous, localized
melting of the sample surface. When higher energy
densities are applied, surface evaporation occurs; the
momentum transfer due to evaporating atoms from the
surface generates an effective pressure on the melt
layer, which finally results in the formation of a melting
ridge. Increasing the incident energy density even further, the material’s response is characterized by intense

Incident beam

Cracking
roughening

Homogeneous
melting

563


Incident beam

Melt ejection

Boiling and
droplet formation

Increasing energy density

Figure 5 Performance of tungsten and metals in general
under transient thermal loads.

boiling and convection of the melt layer resulting in
droplet formation and ejection.170–172 Open pores in
the recrystallized material have a strong impact on the
thermophysical properties.
The melting threshold and subsequently the
amount of melt formation depend on the material’s
thermal conductivity, which is lower for porous
materials such as plasma-sprayed tungsten, and for
tungsten alloys. In particular, it has to be taken into
account that dispersoids such as La2O3 (Tm ¼ 2578 K)
have a lower melting temperature than tungsten. This
may result in early melting and increased evaporation
causing the formation of a porous and depleted surface layer, which becomes even more important when
applying loads below the melting threshold (see
below and Section 4.17.4.1.2). On the other hand,
the melting threshold is correlated with the base
temperature of the PFM. When the base temperature

increases, the melting threshold energy decreases
and the amount of melt formation, the obtained crater depth, and the evaporation losses for the same
applied loading conditions increase significantly.169
As it cools, the material resolidifies in a recrystallized
state providing a columnar grain structure typical of
PVD or CVD coatings. With further cooling, depending
on the base temperature of the material/component
(see ‘Base Temperature’ in Section 4.17.4.1.3), brittle
crack formation will not take place above a certain
threshold temperature. However, with fast cooling
after loading below this temperature, the material
will undergo severe cracking with crack lengths that
can reach the order of millimeters.169
When the qualification of different W grades and
alloys108,141,147 is done in combination with thermal
fatigue loading,90 materials with high thermal conductivity in combination with superior mechanical
properties, that is, with high ductility, performed
best with regard to melt material loss and crack
formation. This comprises low-alloyed W materials
with increased ductility such as W–Re or W–Ta, or
fine-grained pure W or W alloys.


564

Tungsten as a Plasma-Facing Material

Disruption simulation experiments on bulk
tungsten and tungsten coatings have also been
described in the literature. These were performed

not only to investigate the melting behavior but
also for the purpose of characterizing the cracking
behavior.26,42,60,101,122,130,131,162,165,173–176 Although
these experiments are more related to those on the
characterization of ELM conditions (see Section
4.17.4.1.2) and were often performed only at RT,
the results indicate that the use of highly ductile
SC materials is preferred.90,177 Alternately, in case
of cheaper polycrystalline materials, it is necessary
for the material to have the proper microstructure
orientation as described above, that is, the grain orientation perpendicular to the loaded surface. The
reason for this is that crack formation occurs mainly
along the grain boundaries and follows the orientation of the deformed microstructure. The crack depth
is, in general, related to the applied loading conditions
and therefore the pulse length, which determines the
heat penetration depth and the temperature and
stress gradient induced during loading. The temperature gradient also determines the recrystallization
zone, which is generated below the loaded area as a
function of temperature (However, the quantification of the applied conditions and by this a comparison of the materials
response is often not straight forward as each testing
facility has its own characteristics. Most of the time the
cited incident power density, for example, in Hirooka
et al.,174 Linke et al.,178 and Makhankov et al.,90 does not
correspond to the absorbed power density. For example, with an electron beam at 10 keV, Pabs % 0.62
Pinc179 – with the ratio slightly decreasing at higher
acceleration voltages. In a plasma accelerator, Pabs
depends on incident angle and for a perpendicular
impact might only reach 0.1 Pinc.180 For a rough estimate of the temperature impact, the given conditions
can be compared to the heat flux parameter introduced

above. For a base temperature of RT, this amounts to a
melting threshold of $60 MW mÀ2 sÀ1/2 for pure and
fully dense tungsten. Due to the fact that this parameter163 is also directly proportional to the thermal conductivity l, the specific heat cp , and the density r:
pffiffi DT qffiffiffiffiffiffiffiffiffiffiffi
½1Š
plcp r
P t¼
2
a decrease in thermophysical properties consequently
reduces the heat flux parameter and the melting
threshold.
As all performed investigations indicated that
melting will cause increased material degradation

and the continuous erosion of the PFM, it will significantly limit the lifetime of the PFCs. Therefore, the
safe and economic operation of a future fusion reactor requires that scenarios causing melt formation
have to be limited to a minimum.
4.17.4.1.6 Thermal shock during normal
operation: ELMs

In contrast to disruptions, ELMs occur during normal operation in the H-mode and are characterized
as instabilities caused by the steep temperature and
density gradients at the plasma edge, which deposit a
significant amount of energy at a high repetition
rate.181,182 In particular, it is the expected high repetition rate for ELMs during the lifetime of the PFC
(>1 million of events at a frequency of 1–25 Hz183)
that, although yet unexplored, will impose high
demands on the PFMs.
While it is the desire of plasma physicists to operate in H-mode regimes with high-energy ELM deposition (!1 MJ mÀ2), the response of bulk tungsten,
tungsten coatings, and tungsten alloys to such loading

conditions, that is, surface melting, melt motion,
material erosion, and vaporization,170,171,184–189 is
detrimental. To obtain further insight into material
behavior under these conditions, modeling of experimental conditions was carried out.9,167,168,190–195 It
has been shown that with regard to melt motion/
erosion, the results of the different facilities cannot
be directly compared196 and none of the testing facilities used provides identical conditions to those that
will occur in a tokamak. However, mitigation techniques have been explored for reducing the applied
ELM energy, which, in general, can only be done at
the expense of a higher repetition rate.183 The extent
to which the ELMs have to be mitigated depends on
the melt formation at tile edges due to the shallow
plasma impact, which was experimentally found to be
between 0.4 and 0.6 MJ mÀ2 for pure forged tungsten.189,197 On the other hand, the effect of crack
formation during ELMs on the lifetime behavior of
the PFCs has to be taken into account. As mentioned
before, this behavior is yet unexplored at high repetition rates.
Typical investigations on various grades of
W,82,187,189,198 coatings21,54,186,199 and alloys146,157,187
were in the range of 10–100 repetitions. In a few cases
up to 1000 repetitions and in single experiments even
on the order of tens of thousands of repetitions have
been obtained, depending on the testing facility used.
As the repetition rate is still rather low compared to
the expected millions of events, the main interest of


Tungsten as a Plasma-Facing Material

Cracks

Surface modification

No damage

40
35
30

threshold

Heat flux parameter (MW m–2 s–1/2)

45

25
20
Cracking

15
10

Heat flux

Damage threshold

5
0
0

200

400
Temperature (ЊC)

600

200

400
Temperature (ЊC)

600

45
40
Cracking threshold

Heat flux parameter (MW m–2 s–1/2)

(a)

35
30
25

Heat flux

20
15

10
5
0
0

(b)

Figure 6 Thermal shock testing results of double forged
W as a function of temperature and the heat flux parameter;
grain orientation (a) perpendicular and (b) ‘parallel’
(one direction still perpendicular, indicated by the
orientation of the large cracks) to the heat flux.

these investigations was the qualification of different
W grades and alloys (see Section 4.17.3.3) with
regard to their damage and cracking thresholds. The
characterization was done as a function of the main
parameters described in Section 4.17.4.1, that is,
microstructure, power density, and base temperature.
The results obtained so far showed that crack
formation200 vanishes above a certain base temperature (see Figure 6).82,157,198 This temperature
decreases with increasing material ductility, indicating that the use of W alloys or fine-grained W is
preferred. In the case of an anisotropic microstructure,
this effect strongly depends on the material’s orientation. Better results are obtained for grain orientations
parallel to the loaded surface (see Section 4.17.4.1),
yielding differences in the threshold temperature
compared to the orthogonal direction of up to several
hundred K (cf. Figure 6(a) and 6(b)). Recrystallization
leads to a slight homogenization of the material’s

565

microstructure and therefore the mechanical properties; however, there is no full convergency of the
orientation-dependent thresholds.82
Despite the fact that for the currently limited
number of applied pulses no crack formation was
observed above a material and orientation-dependent
temperature, the material is still damaged by plastic
deformation and surface roughening. The evolution
of this plastic deformation and of the related material
hardening as a function of the applied number
of loads is still unclear and has to be investigated.
However, there are also heat load levels (at least
up to Tbase 800  C), at which no visual material
degradation could be determined and the future
goal will be to investigate if these damage thresholds
are still valid for high repetition rates, at higher base
temperatures, and particularly in combination with
neutron irradiation (see Section 4.17.4.3) and plasma
wall interaction (see Section 4.17.4.4).
All the information given above on the effect of
ELMs is also directly transferable to the short transient events expected for inertial fusion applications
and has been verified by IFE-related tests on different W-based materials.201–204 There are coating
parameters of high interest besides those mentioned
above; these include the manufacturing-induced
residual stresses at the surface, which are dependent
on the used substrate, and the coating thickness. As
mentioned in Section 4.17.4.1.1, the applied loading
conditions and therefore the pulse length determine
the heat penetration depth.163 As a result, the temperature and stress gradient induced under IFE

applications should be similar to those in X-ray
anodes (see Section 4.17.2). In case of thin coatings,
residual and induced stresses might affect the coating
to substrate interface and could lead to interfacial
crack formation and delamination. This leads to
minimum requirements for coating thicknesses that
depend on the applied loading conditions.54 For
example, in industrially produced X-ray anodes,
W–Re coatings are typically used with a thickness
of 200–700 mm205,206 to provide better mechanical
and thermal-shock properties compared to pure W.204
However, the first experience on the influence of
ELMs on coatings under real plasma operational conditions will be gained in the ITER-like wall project
in JET, which involves testing relatively thin PVDtungsten coatings (14–20 mm) on a CFC substrate that
provides a strong and anisotropic CTE difference.19,142
The behavior of this material under the above
outlined transient heat loads is of course a key factor
for the lifetime assessment of PFCs. However, the


566

Tungsten as a Plasma-Facing Material

results obtained for pure thermal shock testing might
underestimate the material damage and by this overestimate its lifetime. Only a combination of thermal
shock, thermal fatigue (see Section 4.17.4.2), neutron irradiation (see Section 4.17.4.3), and plasma
wall interaction (see Section 4.17.4.4) will be able to
give appropriate answers for the selection of suitable
grades of W.

4.17.4.2

Thermal Fatigue Resistance

The thermal fatigue resistance of tungsten is strongly
related to its performance as part of current inertially,
and future actively cooled components for application in magnetic fusion devices. The functional
requirements these components have to fulfill are
listed in Section 4.17.2.
State of the art inertially cooled components
include W coatings on graphite, CFC, and
TZM.12,13,46,47,54,140,207,208 These concepts are used
or are going to be used in the large operating tokamaks, for example, AUG and JET. During thermal
loading, they mainly suffer from the problem of high
interfacial stresses as a result of the CTE difference
between the W coating and the substrate. Furthermore, interfacial reaction products and their potential reduced power handling capability have to be
taken into account.
Besides coatings, recent development of an inertially cooled bulk tungsten divertor for JET20,123
showed that under thermal fatigue loads the W quality
is of minor importance for the integrity of the component. The major issue for the tungsten PFM was found
to be the necessary shadowing of the plasma-loaded
surface to avoid overheating and melting at tile edges
as a result of the shallow angle of the incident plasma.
This was realized by surface shaping.209
In the design of actively cooled components,
tungsten is joined to a water-cooled Cu-based
heat sink (ITER) or He-cooled steel or W-based heat
sinks (e.g., DEMO, ARIES-CS). Direct cooling of
the tungsten armor should be avoided, particularly
without castellation, as the induced stresses might

cause catastrophic material failure with subsequent
water or He-leakage.124 Therefore, the only performance requirements are a sufficiently good surface
quality to reduce possible crack initiation points
and therefore suitable fabrication and surface finishing technologies,103,210,211 the chemical compatibility
with the heat sink and, if present, the joining interface
material, and the cyclic stability of the joint(s).
The latter is influenced by the temperature gradient

applied during steady state heat loads, the difference
of the CTEs, the quality of the joining process and,
perhaps most important for reducing induced stresses, the tile size, or the dimensions of the castellated
segments (see Section 4.17.3.2.4).
Smaller tile sizes significantly improve the stress
situation at the interface, and also at the top surface of
the PFM. This has to be taken into account when
comparing the thermal fatigue results of various
kinds of components and the response of different
grades and alloys of W, as shown by Makhankov
et al.,90 where smaller tile sizes resulted in little or
no crack formation. Furthermore, variations in the
size of the component investigated can often explain
the contradictory results presented in the literature
that show good behavior of a material in one test while
it fails in another. However, there are limitations to
the minimum size of tiles and a compromise between
operational and economical needs has to be made.
Despite the fact that design and manufacturing
technique seems to be more important than the
mechanical properties and the microstructure of the
particular W grade or alloy, the latter should still

be considered. Similar to thermal shock results, the
risk of delamination parallel to the loaded surface
at the interface90 or anywhere in the bulk material
has to be minimized. Therefore, the grain orientation
of the PFM microstructure should be perpendicular
to the loaded surface, although this still bears the risk
of crack formation toward the cooling structure.108
To avoid subsequent water or He-leakage in case
of crack propagation into the heat sink material,
particularly in the He-cooled divertor design (see
Section 4.17.4.2.2), suitable material and design
solutions still have to be found. Furthermore, shadowing of adjacent tiles similar to the JET bulk
W divertor has not yet been included in the design
of the actively cooled components.
4.17.4.2.1 ITER

The performance of bulk W for ITER has been
investigated using water-cooled divertor designs, that
is, flat tile, macrobrush, and, most relevant, monoblock
options. One important factor in these design solutions
is the maximum allowable distance between the front
surface and coolant to accommodate the heat without
melting96 and, if possible, to avoid recrystallization
during normal operating conditions. The ability to
estimate this parameter requires not only the thermal
conductivity of the materials but also the amount of
allowable damage at the interface. This requires
knowing not only the damage produced during

Tungsten as a Plasma-Facing Material

operation but also understanding the manufacturing
accuracy and reproducibility because tens of
thousands of armor/heat sink joints will be produced.
Studies on this issue have shown that the current
W monoblock design with a defect extension up to
50 appears to be suitable for the upper part of the
vertical target (P ¼ 10 MW mÀ2), but is not well
adapted to a heat flux of 20 MW mÀ2, which is necessary for application at the strike point of the vertical
target, as systematic defect propagation was observed.
A tungsten flat tile design with 6-mm long defects
in the material interface was studied and proved to
be compatible with fluxes of 5 MW mÀ2 but was
unable to sustain cyclic fluxes of 10 MW mÀ2.212
These results confirm that the monoblock geometry generally proves to have superior behavior under
high heat flux testing when compared with flat tile
geometry. However, it is worthwhile to continue the
investigation of the flat W tile design for low-flux
regions despite the hazard of cascade failure of the
flat tiles106 for two reasons: cost and weight.
Besides this characterization, a number of high heat
flux tests have been carried out on mock-ups and
prototypes without artificial defects representing the
different design options to assess the ‘fitness for purpose’ of the developed technologies.33,90,161,213–218
The results obtained for small test mock-ups of the
flat-tile and monoblock design can be transferred to
large-scale prototypes for the divertor vertical target.
Independent of the type of pure Wor W–La2O3 armor
material used in these prototypes, the W parts survived

in the nonneutron-irradiated condition up to 1000
cycles at 20 MW mÀ2 in the monoblock design213,219
and up to 1000 cycles at 18 MW mÀ2 in the flat tile

567

design (see Figure 7).213 This is far beyond the design
requirements for use in the upper part of the vertical
target (P ¼ 5 MW mÀ2) and, in case of the monoblock
design, even meets the design requirements for the
strike point area of the vertical target.
Alternative concepts such as explosive bonding
of tungsten to a heat sink material,220 PS on a
Cu-alloy216 or on EUROFER steel44 could probably
be of use in the divertor but even more for first wall
applications for fusion machines beyond ITER. However, these concepts often suffer from high interfacial
stresses as a result of the CTE difference between the
W coating and the substrate.
4.17.4.2.2 Prototype and commercial reactors

There are many design proposals for a He-cooled
first wall and divertor concept for DEMO and
ARIES-CS.37,160,221 Among these, the He-cooled
modular design with jet cooling (HEMJ)102 is the
most developed and qualified in terms of microstructural response,103 having survived at reduced coolant
temperatures of 450–550  C at least 100 cycles at
11 MW mÀ2 without failure. In contrast to the results
obtained for water-cooled components for ITER, no
influence of grain orientation on the components
performance was observed.102 This might be a result

of the higher temperature, which was always above
the DBTT.
Nevertheless, some difficulties in the design still
have to be resolved. First, there are problems related
to temperature with a desired coolant temperature of
!600  C; these include material recrystallization at
the top surface and the necessary high temperature
joining to the W-based heat sink material. Second,

W macrobrush
0 dpa:

1000 cycles at 18 MW m−2

0.6 dpa: 1000 cycles at 10 MW m−2
(increasing of Tsurf)

W monoblock
0 dpa:

1000 cycles at 20 MW m−2

0.6 dpa: 1000 cycles at 18 MW m−2
(no degradation)

Figure 7 Thermal fatigue testing results of W macrobrush and W monoblock mock-ups before and after neutron irradiation.


Tungsten as a Plasma-Facing Material

issues related to the material’s mechanical properties
must be solved, in particular for the ductility of
W-based structural material and its neutron irradiation resistance (see Section 4.17.4.3). Finally, the
manufacturing reproducibility has to be at a high
level because of the large number of small units
(1 unit % 3 Â 10À4 m2) necessary for cladding the
DEMO divertor.
4.17.4.3

Neutron Irradiation

The irradiation of tungsten and tungsten alloys with
energetic neutrons (14 MeV) resulting from the D–T
reaction causes radiological hazards that were already
discussed in Section 4.17.3.2.6. In addition, the neutron irradiation affects the material composition by
transmutation of tungsten to Re and subsequently
osmium (transmutation of W isotopes to Ta and Hf
are negligible222). The amount of transmutation
strongly depends on the applied neutron wall load
and neutron spectrum223 and for the W to Re transmutation reaction reaches values between 0.3 and 5 at.%
per dpa.222 The subsequent transmutation of Re to Os
is expected to occur faster than the production of Re
from W resulting in a steadily proceeding burnup of
Re. The neutron fluence on the first wall varies
strongly with location. For the full lifetime of ITER
a maximum of $0.3 MWa mÀ2 is achieved224
(%1.35 dpa in tungsten225). As the divertor PFCs will
be exchanged 3 times and only the last three will
operate in a D–T environment, a neutron fluence
of $0.1 MW mÀ2 is expected during the lifetime of

each PFC. For DEMO, an average neutron wall load
of 2 MW mÀ2 is assumed for the main chamber, which
would result in $45 dpa after 5 full power operation
years. These conditions yield a transmutation of
100% W into 75% W, 12% Re, and 13% Os.36 For
geometrical reasons, that is, larger surface to angular
extension ratio, it will be roughly a factor 2 less in
the divertor region.
Furthermore, neutron irradiation damages the
material properties by the formation of vacancies and
interstitials (see Chapter 1.03, Radiation-Induced
Effects on Microstructure). Their behavior including
analysis of displacement cross-sections,226,227 diffusion, mutual recombination, and clustering are being
assessed by atomistic modeling.228–231
Both transmutation and defect generation influences the material properties and subsequently the
material response to steady state and transient
thermal loads.

4.17.4.3.1 Thermophysical properties
and swelling

The influence of neutron irradiation on the thermophysical properties is related to the irradiation temperature and the number of defects generated in
the crystal structure. At temperatures <1000  C, the
electrical232–234 and thermal conductivity217 of tungsten and tungsten alloys decrease with increasing irradiation dose. However, at elevated temperatures such
as those occurring in a fusion environment, the effect
of neutron irradiation is strongly mitigated by annealing.107 Complete recovery of defect-induced material
degradation should occur at temperatures >1200  C
(see Figure 8).
In addition to defect generation, material degradation is also related to the formation of transmutation
products such as Re and Os, which in general exhibit

poorer thermophysical properties. Transmutationinduced degradation increases with increasing temperature and irradiation dose, which makes it the
most relevant process for the degradation of material
properties for future fusion reactors such as DEMO.
Despite the potential for full recovery of the material defects mentioned above, void-induced swelling
occurs. The results235,236 of tungsten and tungsten
alloys show that the material’s volume increases with
increasing irradiation temperature ( 1050  C).237
W–Re alloys exhibit significantly improved swelling
behavior compared to pure W, with a local maximum at $750  C. However, the swelling only
amounts to 1.7% at 9.5 dpa.237 Therefore, a negligible effect of swelling can be expected for the operation of ITER. Experimental values do not exist at
temperatures >1050  C as expected for the operation
of DEMO.

70
Thermal diffusivity (mm2 s–1)

568

65

Nonirradiated
0.6 dpa at 200 ЊC

60
55
50
45
40
35
30

0

200

400

600

800

1000 1200 1400 1600

Temperature (ЊC)

Figure 8 Thermal diffusivity of W–1% La2O3 in
nonirradiated and irradiated condition.


Tungsten as a Plasma-Facing Material

4.17.4.3.2 Mechanical properties

Data in the literature on mechanical properties of
neutron-irradiated tungsten are very limited.234,238,239
However, in combination with experimental results
obtained for other refractory metals, it has been
shown that in metals with a bcc lattice structure,
irradiation hardening causes a steep increase in
yield stress and a decrease in ductility.110 Consequently, the material fails by brittle cleavage fracture as soon as the yield stress exceeds the cleavage
strength. Therefore, the increase of the DBTT

depends on the neutron fluence, the neutron spectrum (will be addressed by the International Fusion
Materials Irradiation Facility, IFMIF), and the
irradiation temperature. The radiation hardening
in bcc alloys at low temperatures (<0.3 Tm) occurs
even for doses as low as $0.15–0.6 dpa (irradiation
of plasma facing materials for ITER and DEMO,
PARIDE campaigns217), which corresponds to the
expected ITER conditions. Therefore, operation
of tungsten at temperatures >1000  C would be
preferred as full or at least partial recovery of
defect-induced material degradation is achieved by
annealing at 1200  C.234 This implies that the nearsurface part of a W component will retain its
ductility, which has a beneficial effect on the crack
resistance at the plasma loaded surface. However,
such temperatures are not feasible at the interface
to the heat sink where tungsten will be in contact
with copper (ITER) or steel (DEMO), which are
limited to significantly lower operational temperatures. Hence, better understanding of the irradiation
effects on tungsten at temperatures between 700 and
1000  C is needed, particularly related to reactor
application in DEMO.109,110,240
In addition to the influencing factors on the DBTT
mentioned above, that is, neutron fluence, neutron
spectrum, and irradiation temperature, the material’s
composition also plays an important role. While the
addition of Re has a beneficial effect on the material’s
ductility in the nonirradiated state, under neutron
irradiation it results in more rapid and severe embrittlement than it is observed for pure W.239 Similarly,
less mechanical strength and an increased loss of
ductility compared to pure W is found for particlestrengthened W alloys (e.g., W–1% La2O3) when

tested up to 700  C. The only exception among all
explored tungsten alloys might be mechanically
alloyed W–TiC (see Section 4.17.3.3) that showed
no irradiation hardening as measured by Vickers hardness at 600  C.87

569

Finally, the mechanical properties are influenced
by neutron-induced He-generation and the transmutation of tungsten. While He generation in W is, compared to CFC and Be, very small ($0.7 appm He per
dpa) and its influence on the mechanical properties
of W negligible,73,83,224 the transmutation of W into
Re and subsequently Os significantly alters the material structure and its properties. The generation
of significant amounts of ternary a and subsequently
s-phases results in extreme material embrittlement
and will cause shrinkage. In combination with thermally induced strains, this might produce high
tensile stresses causing the extremely brittle material
to extensively crack and perhaps even crumble to
powder.36
4.17.4.3.3 Thermal shock on irradiated W

The simulation of disruptions (<20 MJ mÀ2, t ¼ 5 ms)
on VPS-W and W alloys irradiated to a dose of
0.2 dpa at 350 and 750  C resulted in heavy melting
of the material241 but yielded no measurable degradation by neutron irradiation. This is understandable
because the decrease in thermal conductivity, which
is the most important material parameter for melting
experiments, is almost negligible for the given irradiation conditions.217 Because of the continuous modification of the material composition by transmutation
described above, with increasing levels of Re and Os,
the thermal conductivity and the related melting
threshold power density is expected to decrease

steadily.
In further investigations applying ELM-like loads
on pure W and W–La2O3 irradiated to 0.6 dpa at
200  C, the crack pattern generated after irradiation
exhibited an increased crack density in combination
with a smaller crack width. Furthermore, in W–0.2%
Re SC that was exposed to the same neutron irradiation conditions and exhibited no crack formation
in the nonirradiated state, cracks were formed along
the crystallographic planes (see Figure 9).177 The
effect was more obvious in the results for the SC
material, but in both cases the observed degradation
was a result of mechanical property changes and a
rise of the DBTT in particular. This would indicate
a rise in the threshold temperature for crack formation (see also Section 4.17.4.3.3), which has not
been verified yet. For the evaluation of the material’s
performance in DEMO, both higher transmutation
rates and significantly higher temperatures that are
expected to stimulate defect recovery have to be
taken into account.


570

Tungsten as a Plasma-Facing Material

4.17.4.4

(a)

500 mm

(b)

500 mm
Figure 9 Thermal shock response of W–0.2% Re
(HF ¼ 41 MW mÀ2 s1/2, P ¼ 1.31 GW mÀ2, t ¼ 1 ms, n ¼ 10);
(a) before and (b) after neutron irradiation (0.6 dpa at 200  C).

4.17.4.3.4 Thermal fatigue on irradiated
W components

Information on the thermal fatigue resistance of
W components is limited to the experience obtained
in two irradiation campaigns for ITER which reached
neutron doses of 0.15 and 0.6 dpa at 200  C. Reference and irradiated actively cooled mock-ups with
W–1% La2O3 as the PFM were exposed to 1000
cyclic steady state heat loads at power densities up
to 18 MW mÀ2.213,216,217
The results obtained indicate that at these neutron
fluences the material changes occurring in tungsten do
not have any significant influence on the component’s
performance. However, mock-ups based on the macrobrush design experience a degradation of the maximum
achievable power density from 18 to 10 MW mÀ2,
which is related to neutron embrittlement and
subsequent cracking failure of the Cu-heat sink material. In contrast, monoblock mock-ups show identical
high-level performance before and after irradiation,
which makes it the favored design for ITER.
Despite these positive results, based on the
irradiation-induced mechanical property changes
outlined above, the use of tungsten in any highly
stressed component at low temperatures <500  C

has to be avoided.108

Ion Irradiation and Retention

In addition to the impact of thermal loads and neutrons, plasma wall interaction comprises the contact
of the PFMs with hydrogen isotopes, the helium ash
and impurities originating from eroded surfaces. The
interaction of these different particles with the PFM
leads to near-surface material modification and degradation in the nm and low-mm range. The extent of
the damage depends on the energy of the particles,
their fluence, the surface temperature of the PFM
and the temperature gradients within the PFC during
steady state and thermal shock loading. Furthermore,
the material’s microstructure, composition, and predamage, for example, cracking, have an influence on
the material’s performance. Knowledge of all these
parameters, particularly with regard to operational
conditions, is essential to determine the material’s
lifetime due to erosion, the amount of dust formation,
which influences the plasma performance, and the
safety hazards due to H and He-retention.
4.17.4.4.1 He-irradiation

The incident energy of He ions impinging on the wall
of fusion devices is expected to vary between eV
and MeV. Energies in the eV range are representative
for the majority of particles in a magnetic fusion
device that have lost most of their energy due to
the interaction with the plasma. In contrast, MeV
He-ions, carrying more or less all of their initial
energy, are typical for inertial confinement facilities.

This broad range of energetic particles results in
varying amounts of sputtered material (erosion rate
1.8 times higher than for deuterium242) and further
interaction of the ions with the tungsten wall leads to
additional material degradation. This starts with the
formation of vacancies along the path of the ions and
continues via vacancy clustering, bubble formation,
blistering, and the formation of spongy structures.
However, there exists a lower energy limit for
He-penetration, which is related to the surface barrier potential that was theoretically calculated to be
about 6 eV,243 and with slight variations this value
was verified by experiments.149,244 For faster ions
penetrating into the material, the generation of particular defects depends on the ion energy, ion
fluence, and temperature.26,245,246 The correlation
between these parameters is summarized as follows:
4.17.4.4.1.1

Influence of ion energy and fluence

The ion energy determines the initial penetration
depth of the particle and the vacancy concentration


Tungsten as a Plasma-Facing Material

varies as a function of the implantation fluence.247
The higher the ion energy, the lower the fluence
and the temperature required to create material
damage beyond vacancies and vacancy clusters. For
example, very small blisters were observed for

8 keV ions at RT and a fluence of 4 Â 1021 HeþÁmÀ2
and above.247 In contrast, for low energies (<30 eV),
temperatures >1300 K and fluences of about 1026
Heþ mÀ2 and more are necessary to form bubbles
and surface holes.149,248 The reason for the lack
of blistering at low temperature and low ion
energy is assumed to be the trapping of He-ions
at defects/vacancies in the very near-surface range.
With increasing temperature, the defects and
He-atoms debond and the He-atoms diffuse toward
the bulk, agglomerate, and result in blistering.249
Similarly, with increasing energy, the penetration
depth of the He-atoms increases from nm for
eV-ions up to 1.7 mm for 1.3 MeV He-ions and the
probability of blister formation correspondingly
rises. In both cases, whether the process is driven
by He-diffusion or high penetration depth, blistering and exfoliation are expected to occur when the
amount of He locally reaches 4 at.% and 20–40 at.%,
respectively.250
4.17.4.4.1.2 Influence of temperature

Vacancy mobility is dependent on temperature and
starts at 523–573 K.251,252 As the mobility of vacancies
and the formation of thermal vacancies are driving
forces for the formation of bubbles, holes, and blisters, an increase in temperature increases the size and
decreases the density of material damage.149,253 However, it is not only the temperature during ion irradiation but also the annealing temperature during
experiments such as thermal desorption measurements, which can influence the damage
characteristics.247 The formation of holes and
porous structures observed after thermal treatments,254 particularly for temperatures above the
material-dependent recrystallization temperature, is

related to the movement of vacancies, accelerating
the expansion and coalescence of He bubbles, their
migration to the surface,253 and subsequently the
release of He. The latter is also a function of temperature, showing several release peaks between 400
and 1600 K related to different trapping sites247
and determines the amount of He retained as a function of the incident fluence.242,247,255 However,
helium retention may be mitigated by cyclic
He-implantation and high temperature heating, for
example, flash heating to 2000  C, because He flows

571

away before critical amounts accumulate and form
complex He-vacancy clusters with higher binding
energy.250
4.17.4.4.1.3 Influence of material’s
microstructure

The impact of the material’s microstructure is related
to the amount of intrinsic defects at which He can be
trapped and therefore determines the amount of
He-retention.250 SC tungsten contains fewer defects
than powder metallurgically (PM) produced tungsten
(grain boundaries) and plasma-sprayed tungsten (grain
boundaries and porosity), which is directly related to
the thickness of porous/spongy structures (porosity
about 90%,256 see Figure 10) that form depending
on energy and temperature.257,258 However, investigations at 1650 K have shown that at such high temperatures there exists no difference between SC-W and
PM-W, even for ion energies as low as 25 eV. The main
trap sites at such high temperatures are thermal vacancies while intrinsic defects play a minor role.149,251

With regard to the material’s lifetime under
He-exposure, the migration of He-bubbles toward
the surface and the formation of pores and porous/
spongy structures seem to prevent the rupture and
exfoliation that can accompany blistering. This is
important as the exfoliation of blisters creates dust,199
which limits the plasma performance. For an anticipated flux of 2 Â 1018 Heþ mÀ2 sÀ1 at 850  C in inertial confinement devices, this may lead to a removal of
20 mm yearÀ1 from the wall.109 Therefore, one way
to increase the material’s lifetime might be to operate
it at higher temperature.253 Another approach would
be to develop advanced microengineered materials
that have typical feature sizes less than the classical
helium migration distance (20 nm).109 However,
bubbles, holes, and porous/spongy structures significantly influence the material’s performance by
reducing its thermal conductivity in the near-surface
layer. This might play an important role when determining the erosion and melt formation under combined He-irradiation and transient thermal loads,
which will be shortly addressed in Section 4.17.4.4.3.
4.17.4.4.2 Hydrogen-irradiation and retention

Besides He, hydrogen isotopes, particularly the fuel
elements deuterium and tritium, are the main incident ion species contacting PFMs and PFCs. The
energy of these particles corresponds with the plasma
temperature at the edge, which is in the range of
some eV, but also includes highly energetic particles
( 10 keV) escaping from the inner core of the


572

Tungsten as a Plasma-Facing Material

SR W

(a)

W-Re(5% wt)

(d)

W-Re(10% wt)

(g)

SC W (100)

(b)

WLa2O3(1% wt)

ASTM B760
(ITER)

(c)

W-TiC(1.5% wt)

(e)

(f)

VPS-W
(EAST)

RC-W

(h)

(i)

Figure 10 Cross-sectional scanning electron microscopic images for nine different grades of W relevant to fusion
engineering practice. All target specimens were exposed to consistent pure He plasmas at 1120 K for 1 h. The Heþ
impact energy was $40 eV; (a) PLANSEE stress-relieved W, (b) single crystal h100i W, (c) ITER ASTM B760 compliant W,
(d) PLANSEE W–5% Re, (e) PLANSEE W–1% La2O3, (f) ultrafine-grained W–1.5% TiC, (g) ULTRAMET CVD W–10% Re,
(h) VPS W (EAST), and (i) recrystallized W. Reproduced from Baldwin, M. J.; Doerner, R. P. J. Nucl. Mater. 2010, 404, 165–173,
with permission from Elsevier.

plasma. The impact of the energetic hydrogen ions is
influenced by the incident ion energy, the ion fluence, the temperature, and the material’s composition
and microstructure. The resulting damage, that is,
vacancy formation, vacancy clustering, bubble
formation, and blistering, determine not only the
amount of material degradation and erosion but also
the hydrogen (tritium) retention in the material. For
active control of the hydrogen retention, short-term
thermal treatments of the surface are being investigated. However, the short thermal load required to
effectively remove the deuterium and tritium may
also destroy the thin material layer (nm to low-mm
range) that is responsible for the majority of the
retention.259
4.17.4.4.2.1 Influence of ion energy, fluence, and

temperature

Independent of the ion energy, blistering (see
Figures 11 and 12) due to H occurs only during
irradiation at temperatures below 900–950 K and as
a function of the ion fluence199,260 at 500 K.261–263
This temperature dependence of blistering is attributed to the formation, movement, and agglomeration

of vacancies containing trapped hydrogen,264 which is
dominant at temperatures <500 K, while the detrapping of hydrogen from the defects is prevalent at
temperatures >500 K.263
The fluence threshold for blister formation
increases with decreasing ion energy and significantly increases to values >1025 Dþ mÀ2 at ion energies <20 eV. This is assumed to be the result of thin
(a few monolayers) which act as oxide diffusion barriers at the material’s surface.265 Furthermore, with
increasing fluence, the size and number of blisters
can increase up to a few 100 mm260 until saturation
is reached, which is assumed to be related to the
rupture of blisters.265–267 The rupture and related
dust formation is the result of hydrogen accumulation
and pressure build up, which is effectively released
by the failure of the blister cap. However, whether
the blister has vented or not, the thermal contact
with the substrate has been significantly reduced
(see Figure 11). This eventually results in melting or
vaporization of the thin blister cap, particularly during
transient thermal loading conditions as described in
Section 4.17.4.1, which contributes to further material’s erosion and probably plasma contamination.268


Tungsten as a Plasma-Facing Material

573

(a)

(a)

Large blisters

Small blisters

(b)

(b)

Large blisters

(c)
Crack/void along
grain boundary
(c)

Lids of small blisters partially or
fully removed by FIB fabrication

Figure 11 Scanning electron micrographs of tungsten
exposed to a hydrogen fluence of 1026 D mÀ2 at
480 K (45 tilt). (a) Overall image; (b) cross-sectional image
of a large blister; (c) internal image of small blisters.
Reproduced from Shu, W. M.; Kawasuso, A.; Yamanishi, T.

J. Nucl. Mater. 2009, 386–388, 356–359, with permission
from Elsevier.

In contrast, from the point of view of H retention,
blistering is favorable because tritium accumulates in
blisters, which act as a diffusion barrier for hydrogen,
and which can otherwise penetrate deep into the

Figure 12 Scanning electron micrographs of small
blisters appearing at tungsten exposed to a hydrogen
fluence of 1026 D mÀ2 at 480 K (45 tilt). (a) Initial
stage; (b) growing; (c) bursting. Reproduced from
Shu, W. M.; Kawasuso, A.; Yamanishi, T. J. Nucl.
Mater. 2009, 386–388, 356–359, with permission from
Elsevier.

material even at RT269–274 until it finally ends up in
the heat sink structure and the coolant. Accordingly,
as tritium retention is, in general, strongly correlated
with the generation of blisters,275 it shows a maximum at an irradiation temperature of about
500 K.45,261–263,266,268,276 However, the retention of
tritium and deuterium is also dependent on the
trapping sites existing in the material. These are,
in ascending order of their trapping potential, residual impurities, from which slow desorption occurs
even at RT,277 grain boundaries and dislocations,


574

Tungsten as a Plasma-Facing Material

radiation-induced vacancies and vacancy clusters,
and pores. Depending on the occurrence and dominance of particular sites, the temperature at which
the maximum hydrogen retention is observed varies
between 450 and 850 K.138,262,264,268,277–281
With increasing temperature (>1000 K), such as
that occurring at the strike point of the divertor,
and with the lack of blisters, continuously decreasing
hydrogen retention is observed.2,255,278,282,283 The
remaining amount of retained hydrogen might be
attributed to the presence of hydrogen as a solute,
which depends only slightly on the incident ion
energy, but scales with the implantation fluence and
which is assumed to be of the same order of magnitude as the trapped concentration. It decreases
only slightly with increasing temperature and at
1600 K still amounts to about 10% of the initial
hydrogen content retained at 300 K.274
In addition to blisters, the high amount of hydrogen
out-gassing at temperatures of 873 K and above results
in the formation of bubbles and pores.253,277,284,285
This effect depends on the ion energy and fluence,
which determines the amount and penetration depth
of trapped hydrogen. Even though a beneficial
smoothing effect on the surface quality is observed
in comparison to pure annealing without hydrogen
impact, at high temperatures up to 2500 K the surface
smoothening might be accompanied by detrimental
crack formation.284
4.17.4.4.2.2 Influence of material’s composition
and microstructure

As mentioned above, hydrogen retention depends on
the trapping sites available in the material and their
relative energies. Their existence and concentrations
are influenced not only by the impinging H-ions,
but also by the manufacturing process, thermal pretreatments, the material’s composition and microstructure, and the surface quality. Accordingly, the
retention increases with the amount of porosity
in the material, as it allows a deep penetration of
hydrogen and the voids and pores provide the highest
trapping energies264,275 with thermal desorption
occurring at temperatures >700 K.45,262
Another material parameter that increases hydrogen retention is the number of dislocations,266,286
particularly those introduced during deformation
processes used for material densification. However,
the recrystallization of the material removes not
only dislocations but also vacancies and vacancy clusters, which have been introduced by the impinging
H-ions286,287 and as grain boundaries. This effectively

reduces the trapping sites for hydrogen retention
and, consequently, the lowest retention is observed
for high-purity SC materials, particularly due to
the low diffusion rate compared to polycrystalline
tungsten.275,288,289 This low diffusion rate results in a
near-surface accumulation of hydrogen, which acts
as a diffusion barrier and leads to a saturation of
hydrogen retention with increasing fluence.290 Such
saturation is not observed for pure polycrystalline
tungsten due to the possible hydrogen migration
along grain boundaries.291
Finally, the hydrogen retention is influenced by

impurities277 and dopants. The addition of La2O3 and
TiC particles as well as the formation of pores, for
example, by potassium doping, not only introduce
traps and increase hydrogen retention,293 but also
decrease the diffusion rate.291 In contrast, alloying
with up to 10% Re has no measurable effect on the
H retention properties of the material,279 as it only
creates a slightly deformed crystal lattice structure
but no additional hydrogen traps.
In addition to hydrogen retention, material damage and particularly blistering is influenced by the
material’s microstructure. Blistering occurs preferentially when the crystal is oriented with the h111i
direction perpendicular to the surface292 and the
blisters develop in different shapes from low, large,
and spherical to high, small, and dome or coneshaped.45,262,267 The blisters in recrystallized materials are mainly plateau-shaped, often multilayer
structures, which indicate a step-wise build-up,
and in few cases also small blisters on top of large
ones are formed.263,293 However, it is significant that
the blister size is commonly limited by the grain
size45,294 indicating that the grain boundaries play
an important role in the formation of blisters.
Accordingly, SCs and nanostructured materials such
as W–TiC provide the strongest resistance against
blistering, although the particular reason is different.
For SCs, the hydrogen diffusion and accumulation is
limited and there is a fast desorption from lowenergy traps at elevated temperatures. In contrast,
for nanostructured materials the size of individual
grains is extremely small and so is the volume for
blister formation. Furthermore, the migration of
hydrogen is significantly increased by the large number of grain boundaries.292
Further material parameters that reduce blister

formation are open porosity and the surface finish,
particularly the number of random or artificially
introduced scratches that might act similar to grain
boundaries.45 In contrast, the introduction of


Tungsten as a Plasma-Facing Material

impurities and dopants in commercially available
grades of tungsten increases the number of blisters
and exfoliation in both their stress relieved and
recrystallized states.277,293
4.17.4.4.3 Combined loading conditions

As described above, the damage mechanisms of
hydrogen and He-irradiation are rather similar,
although they occur in different temperature ranges.
Accordingly, their mutual interaction is also
strongly influenced by the implantation temperature. Therefore, the testing sequence plays a role
in the behavior, as for He preirradiation followed by
hydrogen implantation, the implantation temperature of He determines the amount and kind of
produced damage and the He-retention, which subsequently influences the hydrogen uptake occurring as described in Section 4.17.4.4.2. For
example, He-implantation at RT either does not
change the retention or may increase it due to the
formation of additional trapping sites295–297 and
the lower diffusion rate of He compared to H.
With increasing He implantation temperature up
to 800 K, hydrogen retention significantly decreases
compared to pure hydrogen irradiation.261,292,296
This may be attributed to the occupation of trap

sites by He as a result of its increasing mobility.298
Potential trap sites are the numerous He-induced
nanosized bubbles acting as a diffusion barrier.292
A further increase in temperature to 1600 K does
create significant material damage by He due to
pore and bubble formation or even blistering. This
tremendously increases the number of trap sites in
the material and leads to He desorption during
implantation and accordingly increases the hydrogen retention.299
For simultaneous loading of He and hydrogen,
the fraction of He should reach at least 5 at.% to
observe significant changes in the material’s
response.261,292 Furthermore, for implantation temperatures below 900 K, results similar to those
described above are observed for sequential ion
beam loading.299 However, due to desorption of
hydrogen at high temperatures >1000 K, no hydrogen retention takes place and the damage mechanisms are dominated by the He-irradiation during
such temperature excursions.
Correlated with hydrogen retention, blister formation at temperatures <900 K decreases with
decreasing hydrogen retention. This is valid until
the number of voids and pores, which enhance hydrogen retention, start to form open porosity and thereby

575

generate small grain structures. These allow a fast
hydrogen diffusion through the material and limit
the agglomeration of hydrogen necessary to form blisters similar to the case of nano-structured materials
such as W–TiC described above (Section 4.17.4.4.2).
Investigations of the influence of radiation damage (highly energetic hydrogen, neutrons) and
impurity irradiation, for example, by carbon atoms,
resulted in the depth resolved and particle energy–

dependent formation of dislocations, dislocation
loops, and even small voids acting as effective
trapping sites for hydrogen and influence blister
formation.274,276,300–309 Upon annealing, the dislocations and dislocation loops were moved and/or
annihilated,310,311 which is positive news as it
would limit the tritium inventory,312 as long as no
He is present in the system. In contrast, with the
addition of He the dislocations, dislocation loops,
and helium bubbles do not vanish at identical
annealing conditions, which has a direct impact on
the mechanical and thermophysical performance
of tungsten. However, He positively influences
hydrogen retention in an intermediated temperature range as described above and inhibits the formation of a W carbide layer, which is typical for
combined hydrogen and carbon loading.311
Finally, the results obtained from the investigation
of the mutual influence of ion irradiation and thermal
loads are strongly correlated with the choice of the
heat source. In the literature, electron beam guns
were favored,313,314 which are characterized by
heat deposition in a depth range of several micrometers for W, depending on the acceleration voltage.
However, as the thickness of the ion-irradiationaffected surface layer is comparably thin, the
majority of the electrons pass through the modified
surface layer, which leads to most doubtful conclusions. In contrast, lasers are more reliable, as they
apply only surface heat loads. The combination
of He-irradiation and laser-induced thermal loads
(DT ¼ 1400 K, n ¼ 18 000) at high base temperatures
($1700 K), resulted in an affected layer thickness
(13 mm) about 10 times larger than that without
laser irradiation (1–2 mm), which might be attributed
to the steep temperature gradient supporting the

diffusion of He. This surface modification combined
with laser-induced surface roughening, as observed
in Section 4.17.4.1.2 for typical thermal shock loads,
leads to an enhanced degradation of the thermal
diffusivity of W, which further increases the surface
roughness and results in local or full melting of the
W surface.199