EPJ Web of Conferences 246, 00013 (2020) [Link]

1051/epjconf/202024600013
Joint EPS-SIF International School on Energy 2019

Fusion energy
Francesco Romanelli
University of Rome “Tor Vergata” - via del Politecnico 1, 00133 Roma, Italy

ENEA - via Enrico Fermi 65, 00044 Frascati (RM), Italy

Summary. — This paper presents an overview of the main challenges that fusion
research is facing on the road to a demonstration power plant. The focus is on mag-
netic confinement fusion. Most of the challenges are being addressed in the context
of the ITER construction and exploitation. These include the demonstration of high
fusion gain regimes of operation, the management of high heat and particle loads
and the integration of the main technologies of a fusion power plant. In preparation
of DEMO, reliable solutions for the breeding blanket and neutron resistant materials
have to be developed.

1. – Introduction. Fusion in the context of energy technologies

In the quest for an energy system that meets simultaneously the needs of sustainabil-
ity, security of supply and economic competitiveness Europe is pursuing several technolo-
gies since a robust energy strategy has to rely on a portfolio of different energy sources
that include nuclear, solar, wind and hydropower [1].
In this context, fusion can play an important role since it has a number of advantages:

– It is virtually unlimited. For the deuterium-tritium reaction the fuel in sea-water

could supply the Earth (at the present consumption rate) for several million years.

– It does not produce greenhouse gases.

– It is intrinsically safe.
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

Table I. – Fusion reactions.

D+T → 4 He (3.5 MeV) + n (14.1 MeV)

D+D → T (1.01 MeV) + p (3.02 MeV) 50%

3
He (0.82 MeV) + n (2.45 MeV) 50%

D + 3 He → 4 He (3.6 MeV) + p (14.7 MeV)

T+T → 4 He + 2n + 11.3 MeV

3
He + T → 4 He + p + n + 12.1 MeV 51%
→ He (4.8 MeV) + D (9.5 MeV)
4
43%
→ He (2.4 MeV) + p (11.9 MeV)
5
6%(1 )

p + 6 Li → 4 He (1.7 MeV) + 3 He (2.3 MeV)

p + 7 Li → 24 He + 17.3 MeV 20%

→ 7 Be + n − 1.6 MeV 80%

D + 6 Li → 24 He + 22.4 MeV

p + 11 B → 34 He + 8.7 MeV

– It is environmentally responsible. The primary reaction does not produce radioac-

tive materials. It does produce neutrons that activate the reaction chamber. How-
ever, with a proper choice of materials, radioactivity decays in a few tens of years
and after a century they can be recycled in a new reactor.

Several fusion reactions can in principle be exploited (see table I). The cross section
σ and the reactivity σv (the product of the cross section and the particles relative
velocity averaged over the distribution functions of the reacting species) are shown in
fig. 1 for the DT, D3 He and DD reactions.
The reaction with the largest cross section is the reaction between two hydrogen
isotopes: deuterium and tritium. The DT reaction produces a 3.5 MeV alpha particle
and a 14 MeV neutron. To make fusion energy with the DT reaction we need to burn
deuterium and lithium. There is plenty of deuterium in sea-water (about 35 mg per
litre [2, 3] —making fusion with deuterium alone would supply the Earth, at the present
consumption rate, for several billions years!). On the contrary, tritium does not exist
on Earth. It is mildly radioactive (it undergoes beta decay releasing an electron with
a maximum energy of 18.6 keV) with a half-life of 12.3 years. Therefore, it has to be

(0 ) The 5 He isotope decays in 7 × 10−22 s in a neutron plus an alpha particle.

EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

100

10–27
Fusion cross-sections
a) Fusion Reaction Rate b)

10–28
D –T

<σν> (10-22 m3 s-1)

10

10–29
σ (m2)

D–D D-T D-He3

10–30
1
D-D
10–31
D–He3

JG95.597/13s

JG03.366-10c
10–32 0
1 10 100 1000 1 10 100 1000
Deuteron energy (keV) Ion Temperature (keV)

Fig. 1. – Fusion cross sections and fusion reactivity for the three main fusion reactions. The cross
section is plotted as a function of the energy of the deuteron impinging on the target at rest.

produced inside the reactor and this is where lithium comes into play. The reaction
between a neutron and a lithium nucleus produces a He nucleus together with a T
nucleus that can be extracted and re-circulated in the reactor.
In order for D and T to come at sufficiently close distance for fusion to occur they
need to overcome their mutual Coulomb repulsion. This can be achieved by heating the
DT gas to sufficiently high temperature (∼ 20 keV or 200 millions of ◦ C) in such a way
that the ions in the tail of the Maxwellian distribution function provide a significant
number of reactions. Matter at these temperatures is in the plasma state. The electrons
are no longer bound to the nuclei and the system becomes the superposition of two gases
of negatively charged electrons and of positively charged ions. To confine plasmas at
these temperatures two main methods have been successfully employed (see fig. 2).

– The first method is magnetic confinement that will be mostly covered in this lec-
ture. In this case we can take advantage of the charged nature of the plasma
constituents to confine them using intense magnetic fields (∼ 100000 times larger
than the average Earth magnetic field). In the presence of a magnetic field par-
ticles move along field lines as a train on a rail and are therefore confined in the
plane perpendicular to the magnetic field. In order to confine plasmas also in the
third direction the magnetic field lines are wound in such a way as to form a set of
nested toroidal surfaces called magnetic surfaces, with each line lying on a magnetic
surface. The combination of intense magnetic fields (∼ 5 T) and toroidal geometry
enables plasma confinement.

– The second method is inertial confinement (see ref. [4]). In this case a sphere of solid
fuel is irradiated with electromagnetic waves or energetic particles. The surface
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

Fig. 2. – (a) Magnetic confinement geometry illustrated for the tokamak configuration —see text
(courtesy EUROfusion). (b) 1: Inertial confinement geometry. The shell target (blue circle) is
irradiated with the driver beams (blue arrows) and a spherical cloud of ablated plasma is formed.
2: The shell implodes as a consequence of ablation. 3: A hot spot of compressed fuel is formed
where fusion reactions take place. 4: The fuel is burned and an explosion occurs.

layer is ablated and its particles are ejected in the radial direction. This produces
a radial compression that, if sufficiently large, can burn a large fraction of the fuel.

It is important to stress that with both methods plasmas of thermonuclear interest

can be already routinely produced. In addition, a significant amount of controlled fusion
reactions have been already obtained. In a dedicated campaign on the JET machine in
1997 [5] up to 16 MW of fusion power has been obtained in transient conditions to be
compared with about 25 MW of power injected in the reaction chamber. Nevertheless,
making a fusion power plant requires the solution to a number of other challenges that
will be discussed in the following.
In 2012 the European Commission requested a Roadmap to fusion Energy in order to
understand if fusion can start playing a role in the electricity production on the time scale
of 2050 used for the Energy Roadmap of the European Union. The Fusion Roadmap [6]
(an update has been recently produced by EUROfusion [7]) was based on the following
assumptions:

– A program focussed around the priorities and organized in eight missions.

– A pragmatic approach —build on the shortest possible time scale a device capable
of breeding tritium and produce electricity.

– Early involvement of industry.

– Full exploitation of the opportunities arising from international collaborations.

EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

The Roadmap has been the reference document for the EURATOM activities in fusion
for Horizon 2020.
The Roadmap foresees two main facilities, ITER and DEMO:

– ITER will demonstrate burning plasma operation, i.e. operation with dominant
plasma heating produced by the fusion generated alpha particles. ITER is also
developing a large part of the technologies needed in a fusion power plant. However,
ITER will neither produce electricity nor will breed the tritium it uses.

– DEMO will be the first demonstration fusion power plant. In addition to what
ITER will accomplish, it will produce a net electricity output and will be self-
sufficient for the production of tritium.

Both ITER and DEMO are machines based on a specific magnetic confinement con-
figuration called tokamak [8] and schematically shown in fig. 2(a). The plasma is char-
acterized by a doughnut shape (torus) with major radius Ro and minor radius a. The
magnetic field is the superposition of a toroidal component Bf (mostly generated by a set
of external coils) and a poloidal component Bp (mostly produced by a current flowing in
the plasma itself in the toroidal direction). The combination of a toroidal and a poloidal
magnetic field produces magnetic field lines that lay on toroidal surfaces called magnetic
surfaces. The winding pitch of the magnetic field line on each magnetic surface is de-
scribed by the rotational transform (the number of turns made in the poloidal direction
for a single turn in the toroidal direction). The inverse of the rotational transform is
the safety factor. The existence of a rotational transform is essential for plasma confine-
ment. Without a rotational transform ions and electrons, under the effect of magnetic
field curvature and spatial in-homogeneity, would drift vertically in opposite directions
producing an electric field that, combined with the toroidal magnetic field, would push
the plasma outwards in the radial direction in a very short time.
After a confined plasma is established, it is heated up to temperatures of 10–20 keV by
auxiliary heating systems until the fusion generated alpha particles become the dominant
heating source. The heat generated in this way is transported from the center of the
plasma to the edge (see sect. 2) and removed from the reaction chamber. Similarly,
the He ashes produced after the fusion alphas have transferred all their energy to the
plasma need to be continuously removed to avoid poisoning the fuel (dilution). Plasma
exhaust takes place at a special location, called divertor, in the reaction chamber (usually
a niche at the bottom) sufficiently remote from the hot plasma such that the heat and
particle removal can take place without disturbing the dynamics of the plasma inside the
magnetic separatrix. The separatrix defines a sharp boundary between the hot plasma
region (inside) and the plasma edge region in contact with the chamber wall and the
divertor and therefore at much lower temperature.
In the following we review the main challenges of fusion research. The focus will be
on magnetic confinement although a number of challenges are common also to inertial
confinement.
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

2. – Demonstration of regimes with a high energy gain

The first challenge is the demonstration of regimes in which the amount of fusion
power absorbed in the plasma is much larger than the external heating power (burning
plasma conditions). The alpha particles produced in fusion reactions are charged and
therefore can be confined by the same magnetic field that confines the plasma. They
release their energy to the plasma via collisions. In burning plasma conditions the alpha
particle heating can almost entirely maintain the high plasma temperatures. In ITER
the target is a plasma in which alpha heating (100 MW) is twice the external heating
(50 MW). Since the 14 MeV neutrons carry four times more energy than the 3.5 MeV
alpha particles the fusion power produced in the form of neutrons in ITER is 400 MW.
Thus, the total fusion power is 100 MW + 400 MW = 500 MW and the fusion gain Q
(given by the ratio between the fusion power and the external power injected in the
reaction chamber) in ITER is Q = 500 MW/50 MW = 10. For comparison, the JET
achievement quoted above (16 MW of fusion power with 25 MW of externally injected
power [5]) corresponds to Q ∼ 0.67.
Why high gain has not been achieved so far? The problem is associated with the con-
duction losses due to the small-scale turbulence destabilized by the free-energy sources
always present in a confined plasma (the gradients of temperature and density). Radi-
ation losses are also present although usually much smaller. Losses tend to cool down
the plasma and must be balanced by the heating power. What we know from theory
and experiments is that, at fixed density, temperature and magnetic field, the power lost
via conduction is at most linearly increasing with the machine radius R [9] whereas the
fusion power increases as R3 . Therefore the solution that has been pursued has been to
make the machine size larger.
The argument above can be put on a quantitative basis as follows. For the sake of
simplicity we will assume that density and temperature are constant in space. The power
generated by fusion reactions is given by

(1) Pf us = 17.6 MeV nD nT σvV,

where nD is the particle density of the deuterium nuclei (number of deuterium nuclei
per unit volume), nT is the particle density of the tritium nuclei, σv is the Maxwellian
fusion reactivity and V is the plasma volume. The Maxwellian reactivity is a function of
the temperature of the reactants [10].
The power Pcond lost by the plasma through conduction is due to small-scale turbu-
lence and can be quantified in terms of the energy confinement time τE

(2) Pcond = W/τE ,

where W = (3/2) (ne Te + Σj nj Tj ) V is the internal energy of the plasma with the sum
extended to all the ion species and Te (Ti ) the electron (ion) temperature. The energy
confinement time is the characteristic time for the plasma to cool down once the heating
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

sources are switched off. It has nothing to do with the time plasma is confined. In ITER
τE is a few seconds whereas the plasma can be confined for hundreds of seconds.
The power lost by the plasma through radiation is mostly associated with
Bremsstrahlung and is given by

(3) Pbrem (MW) = 1.69 × 10−4 Zef f ne (1020 m−3 )2 Te (eV)1/2 V (m3 )

with ne the particle density of the electrons and Zef f ≡ Σj Zj2 (nj /ne ) the average
charge of the plasma ions.
Since the plasma must be locally neutral the electron density ne must be equal to the
sum of the ion densities nj times their charge Zj

(4) ne = Σj Zj nj .

The power balance equation in stationary conditions dictates that the fusion power re-
leased in the form of alpha particles (3.5 MeV/17.6 MeV ∼ 20% of the total) plus the
power Paux injected from external sources must be equal to the power lost by conduction
and radiation

(5) (3.5/17.6)Pf us + Paux = Pcond + Pbrem .

The fusion gain Q = Pf us /Paux becomes infinite for Paux = 0, meaning that the fusion
reactions are self-sustained. This is the so-called ignition condition.
For a pure deuterium and tritium plasma (nD = nT = ne /2) with equal electron and
ion temperatures, the ignition condition can be written as
 
(6) ne (1020 m−3 )τE (s) = 3.42 × 10−3 T (keV) σv(10−20 m3 /s) − KT (keV)1/2

with K = 3.82 × 10−4 m3 s−1 keV−1/2 . This is the Lawson criterion (fig. 3(a)). It
expresses the requirement in terms of the product of density and confinement time to
achieve ignition as a function of the plasma temperature.
The right-hand side of the Lawson criterion is a function of the temperature only
with a minimum corresponding to ne τE ≈ 1.6 × 1020 m−3 s at about 25 keV. For tem-
peratures approaching 4.3 keV the ne τE product tends to infinity. This is the so-called
ideal ignition temperature and represents the minimum temperature below which the
alpha particle power is smaller than Bremsstrahlung. Around the optimal tempera-
ture Bremsstrahlung can be neglected and we can approximate the fusion reactivity as
σv ∼ 10−24 T (keV)2 m3 /s. Upon substituting in eq. (6) we can determine a condition
on the triple product ne τE T ∼ 3.42 × 1021 m−3 s keV.
The fulfillment of the Lawson criterion is the main goal of fusion experiments and is
mostly related with the achievement of high confinement times. Fusion plasmas exhibit
the transition to self-organized states in which turbulence is quenched by the formation of
sheared flows and a “transport barrier” is locally formed. The most common (and easily
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

a) b)
10
ASDEX AUG START
14,00 nτE(1020m-3 s) D3D JET COMPASS
PBXM PDX JT60U
TDEV CMOD TCV
12,00 JFT2M ITER-FEAT
1

Measured Confinement Time

10,00

8,00 0.1

6,00

4,00 0.01

World Wide Data Base

2,00 (13 Devices)
T(keV)

JG00.335/1c
0.001
0,00
0 50 100 150 200 0.001 0.01 0.1 1 10
Predicted Confinement Time

Fig. 3. – (a) Lawson criterion nτE vs. temperature for a pure homogeneous DT plasma; and
(b) measured energy confinement time vs. the prediction of the ITER98y2 scaling (from ref. [9],
by courtesy of EUROfusion). ITER-FEAT was the name of ITER at the time of ref. [9].

reproducible) of these self-organized states is the so-called H-mode and is characterized

by a transport barrier located at the magnetic separatrix. Unfortunately, the prediction
of τE from first principle models of plasma turbulence still bears significant uncertainties.
Thus τE is extrapolated from the experimental values obtained in present day machines
that cover a range of three orders of magnitudes (from ∼ms values obtained on START
to ∼s values obtained on JET). A careful analysis of the experimental data has produced
the following scaling law for the energy confinement time in H-mode [9]:

(7) τIT ER98(y,2) (s) =

0.0562Ip (MA)0.93 B (T)0.15 P (MW)−0.69 n (1019 m−3 )0.41 M 0.19 Ro (m)1.97 0.58 κ0.78 ,

where Ip is the plasma current, B the toroidal magnetic field, P the heating power, n the
line averaged plasma density, M the average ion mass,  ≡ a/R the inverse plasma aspect
ratio and κ the plasma elongation. The extrapolation to ITER gives a value of 4.3 s, about
three times larger than the upper limit of τE in JET. The estimated value of the triple
product on ITER (with n = 1020 m−3 and T = 20 keV) is ne τE T ∼ 8.6 × 1021 m−3 s keV
in line with the requirements to achieve a fusion gain Q = 10. It should be noted that
although H-mode conditions can be easily reproduced in all the tokamaks with a magnetic
separatrix, the H-mode barrier formation requires a minimum power to be transported
across the separatrix by conductive losses that is at the moment difficult to estimate
both theoretically and experimentally.
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

In the case of inertial fusion a criterion similar to the Lawson criterion can be derived.
In this case the fuel is confined by its own inertia and we can define a confinement time
τICF ∼ Rf uel /Cs , where Rf uel is the radius of the compressed fuel and Cs the sound
velocity. The request that the confinement time is longer than the fusion reaction time
τf us = 1/(nσv) yields a condition on the product ρRf uel , (with ρ the fuel mass density)
that is analogous to the Lawson parameter nτE . For typical ICF parameters the product
ρRf uel must be larger than 2–3 g/cm2 [4].
Part of the challenge of demonstrating regimes with a high energy gain is represented
by the need of controlling macroscopic plasma instabilities [11]. Broadly speaking we can
distinguish two classes of instabilities:

– Instabilities that lead to a sudden loss of confinement (disruptions). These must

be avoided (operating far from the stability threshold), prevented (by deploying
appropriate control systems) or, in case they occur, mitigated (e.g. by the injection
of highly radiating elements) to avoid localized thermal loads on the plasma facing
components and the electromagnetic loads on the conductive structural components
inside the vacuum chamber.

– Instabilities that produce a minor redistribution of the plasma and can be benefi-
cial in facilitating the ejection of impurities. Instabilities belonging to this class are
benign provided their amplitude is kept sufficiently small through appropriate con-
trol methods. Examples of instabilities to be controlled are edge localized modes
(ELM) or sawtooth oscillations. In both cases the instability takes the form of a pe-
riodic relaxation that induces plasma redistribution (at the edge and at the centre,
respectively). Several methods have been developed to control these instabilities.
Magnetic perturbation coils producing 3D fields or injection of frozen deuterium
pellets have been shown to suppress or mitigate ELMs. Millimetre wave injection
is a reliable method to control neoclassical tearing modes. These control methods
have been demonstrated in present experiments and now await a final confirmation
in ITER.

In addition ITER will break new ground in the investigation of collective alpha par-
ticle effects, namely the destabilization of modes driven by a population of energetic
particles [12]. Our theoretical understanding shows that these modes should lead at
most to minor plasma redistribution for the operational scenarios investigated during
the first ITER phase but their role for the operations in fully steady-state scenarios is
expected to be substantial.
Finally, a specific challenge is related with the achievement of a fully steady state. This
is in principle the target of any energy technology that has to provide electricity to the
grid. An axisymmetric equilibrium requires a net plasma current to flow in the toroidal
direction in order to produce a magnetic field with a rotational transform. The plasma
current is produced via magnetic induction by a variable current flowing in a cylindrical
solenoid (CS) inserted into the central hole of the torus. The CS acts as the primary
winding of a conventional transformer in which the plasma is the secondary winding.
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

Since the plasma current has to flow always in the same direction (no alternate current
operation are foreseen) the current on the primary has to monotonically change starting
from the maximum current the central solenoid can bear. If we assume that this current is
in the positive direction, the central solenoid current first crosses the zero value and then
flows in the negative direction until the maximum current (in the opposite direction) is
again reached. Beyond this value it is no longer possible to induce current in the plasma.
For axisymmetric configurations there are two solutions to this problem:

– We abandon the idea of continuous plasma operation and, after the central solenoid
has completed the flux swing, we stop the plasma and recharge the transformer.
This requires some energy storage to be available. Pulsed operation could produce
fusion power for a few hours with a recharging time of about 20 minutes. This
solution is simple to implement from the point of view of plasma operation but
may induce cyclic fatigue in some component.

– The plasma current is generated through a combination of the current self generated
by the plasma (the so called bootstrap current) [8] and external current drive
systems [13]. These regimes have been demonstrated but are not fully qualified
and require a sophisticated control capability of the radial profiles of pressure and
current density. ITER has the goal of demonstrating their viability for a reactor.

There is a third possibility: abandon axisymmetric equilibria. In this case the ro-
tational transform can be produced through a set of external helical coils. This would
avoid entirely the need of inducing the plasma current (and so also avoid instabilities
associated to it). We will come back in sect. 9 to this possibility.

3. – The challenge of the heat and particle load

The heat that crosses the separatrix due to conduction losses flows along the magnetic
field lines (fig. 4) in a layer a few mm thick and is eventually deposited on the divertor [14].
Since all the heat is localized in a narrow layer, the heat load on the divertor in DEMO can
reach values up to 60 MW/m2 , comparable with the heat load at the surface of the Sun!
In addition, the continuous flow of particles impinging on the divertor surface may
produce large erosion. A simple estimate of erosion can be done as follows. In ITER
with a volume of 800 m3 and a density of 1020 m−3 there are about 8 × 1022 ions. If
the rate at which they are lost through the separatrix is similar to the estimated energy
confinement time (∼ 4 s) the number of particle arriving on the divertor per unit time is
2 × 1022 s−1 . The exposed surface is of the order of 2 m2 . If the probability of extracting
an atom from the divertor surface (the so-called sputtering yield) is 10−4 , the number
of atoms extracted in a year of operation is 3 × 1025 atoms per square meter, or taking
into account that solid tungsten has a density of 6.4 × 1028 m−3 , about 0.5 mm would be
eroded every year.
To keep the sputtering yield at low values requires that the temperature in front of
the divertor plates must be in the range of few eV. This can be understood as follows.
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

Fig. 4. – Divertor geometry. The orange line is the magnetic separatrix.

Since electrons have a larger mobility than ions they tend to be more easily lost to the
divertor plates which become negatively charged with a potential that is a fraction of the
electron temperature at the plate divided by the electron charge. This negative potential
accelerates the plasma ions towards the plate up to velocities of the order of the local
plasma sound velocity. Such a flow of ions is beneficial in opposing the diffusion out
of the divertor region of the impurities released from the plates and of the He ashes
(thus avoiding plasma contamination and allowing efficient He pumping), however it also
increases the probability of impurity extraction. An ion is indeed accelerated to energies
proportional to the electron temperature at the plate and the probability of extracting
an impurity atom from the plate become significant unless the temperature is low.
The solution to this challenge is made of two recipes. The first recipe is the devel-
opment of plasma facing components that can withstand high heat and particle fluxes.
The solution foreseen for ITER is the so-called tungsten monoblock made by a tung-
sten dice with a cooling channel made of CuCrZr attached to tungsten through a Cu
interlayer. The W-monoblock has been shown to withstand more than 1000 cycles up to
20 MW/m2 . This however would not be enough even for ITER. The second recipe is to
produce semi-detached divertor conditions through the formation of a cloud of neutral
gas that absorbs energy and momentum of the incoming particles.
In this way it is possible to achieve temperatures in front of the plate of the order
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

a) b)
1
10 100
100 E(eV) 1000
E(eV) 1000
Sputtering
yield
0,1

0,01

0,001

0,0001

0,00001

Fig. 5. – Tungsten monoblock mock up (by courtesy of EUROfusion) and physical sputtering
yield vs. incident ion energy for tungsten (dashed lines) and carbon (continuous lines) under the
bombardment by tritium (black), deuterium (blue) and hydrogen (orange).

of few eV, sufficiently low to avoid substantial erosion in stationary conditions (or also
under the effect of slow transients) in ITER (see fig. 5(b) where the sputtering yields for
C and W under the bombardment of different hydrogen isotopes are plotted using the
fits of ref. [15]).
A specific issue is the effect of transient heat loads such as those generated by ELMs
and disruptions. Transient loads may drastically reduce the lifetime of plasma facing
components (see, e.g., fig. 7 of ref. [14]). Therefore appropriate disruption mitigation
systems must be in place and the ELM amplitude must be kept small such that the local
deposition is below 1 MJ/m2 per ELM.
The standard divertor solution is expected to be sufficient to cope with the ITER
heat loads. However it is unclear whether it can work also for the much larger heat loads
of DEMO. In principle, it can work provided a large fraction of the heating power is
radiated before crossing the separatrix. In this case the heat would be exhausted on the
large area of the main wall whereas the divertor would continue to play its role for the
plasma/impurity density control. However, plasma regimes with high radiation usually
exhibit a lower energy confinement time. The development of regimes that simultaneously
radiate a large fraction of the heating power from the region inside the separatrix and
maintain high confinement is one of the most important research lines for ITER and
DEMO.
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

Fig. 6. – (a) Snowflake divertor configuration tested on the TCV tokamak (from ref. [16]).
(b) Super-X configuration to be tested on the MAST-U tokamak (from ref. [17]).

What if the predicted DEMO heat load will be too large to be managed with the
conventional divertor solution? There is some room of maneuver with the present divertor
configuration if the bottom divertor is paired by a divertor at the top or if the magnetic
separatrix contact points are slowly swept up and down on the target plate in order to
spread the heat load. However, if this will be not enough radically new solutions will be
needed. These can be divided in two classes:
– The use of advanced magnetic field configuration with a larger divertor wetted
surface. These configurations (“snowflake”, super-X, etc. —see fig. 6) are presently
tested at the proof of principle level [16, 17].
– The use of liquid metals. In this case the surface exposed to the heat flux is
continuously re-formed and its damage is no longer an issue.

The debate about the compatibility of these alternative solutions with the constraints
of a reactor is still open. Nevertheless it is clear that the challenge of the exhaust problem
deserves a wide investigation. To this goal, the possibility of a Divertor Tokamak Test
facility was advocated in the Roadmap [6] and the realization of a dedicated experiment
is presently under way (see, e.g., [18]).

4. – Materials that can withstand high neutron fluxes

The use of the DT reaction makes fusion conditions easier to achieve but it has the
drawback that 80% of the fusion power is released in the form of 14 MeV neutrons. Part
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

of the fusion-generated neutrons does not react with the lithium in the blanket to pro-
duce tritium but is absorbed by the structural materials of the reactor. This produces a
degradation of structural properties and activation of wall materials. This degradation
is mostly localized in the first few tens of centimetres (from the plasma exposed surface)
of the plasma facing components. The vacuum vessel receives very low levels of neutron
irradiation and it maintains its structural properties throughout the life of the reac-
tor. On the contrary, the plasma facing components will need to be replaced every 3-4
years.
The damage suffered by structural materials can be quantified in terms of the displace-
ments per atom (dpa). The nucleus in the structural material lattice that has absorbed
the neutron energy (primary knock-on atom) releases its energy by displacing the sur-
rounding atoms and producing point defects and dislocations. This effect is measured
by the average number of displacements per atoms (dpa) in the lattice. Although dete-
rioration of structural properties due to neutron irradiation is encountered also in fission
reactors, the high energy of fusion neutrons (14 MeV compared with around 2 MeV for
neutrons produced in fission reactors) induces a substantial amount of reactions form-
ing H or He that accumulates in the lattice. These two effects produce embrittlement,
swelling and irradiation creep and become important beyond the level of ∼ 10 dpa [19,20].
As a result the temperature window for operation is limited between 350 ◦ C and 550 ◦ C.
Typical numbers for radiation damage are 100–150 dpa in a fusion reactor, 30–70 dpa
in DEMO and 2 dpa in ITER. Thus, there is no problem with material degradation in
ITER. However, appropriate materials must be developed and qualified for fusion reac-
tor application to avoid frequent replacement of the plasma facing components and the
blanket. For DEMO, it is possible to consider an initial exploitation with the presently
qualified materials, but the second phase will require significant advances in material
performance.
In view of the high-energy neutrons that characterize fusion, such a qualification
will require a dedicated facility, the International Fusion Material Irradiation Facility
(IFMIF). IFMIF will produce a neutron spectrum similar to that of a fusion reactor
through various stripping reactions between two 40 MeV/125 mA deuteron beams and
a liquid lithium target. The engineering validation and engineering design activity are
presently being finalized within the framework of a collaboration between the EU and
Japan [21].
Activation is the second issue of neutron irradiation. Since the fusion reaction does
not produce radioactive materials (the primary source of waste for fission), the issue of
activation is limited to the structural materials of the plasma facing components. Among
the structural materials so far investigated, reduced activation ferritic-martensitic steels
(RAFM, such as EUROFER [22]), appears to be the near-term solution. RAFM
steels differ from austenitic steels as molybdenum, nickel and niobium are replaced by
tungsten, tantalum, vanadium and/or titanium that have a better behaviour under
neutron irradiation. These alloys achieve a sufficiently low level of radioactivity in a
sufficiently short time (say 100 years after the end of the reactor operation) in such
a way that all the reactor materials can be easily recycled in a new reactor. Simple
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

a) b)
105
MANET-II
104 OPTIFER
EUROFER-97

Surface Gamma Dose Rate [Sv/h]

F82H-mod
103 EUROFER ref.
FE
102

101

100

10-1 High Level Waste

10-2 Medium Level Waste

10-3
Low Level Waste
10-4
Hands-on Level
10-5

10-6

10-7
10-3 10-2 10-1 100 101 102 103 104 105 106
Time after Irradiation [y]

Fig. 7. – (a) Picture of EUROFER components (by courtesy of EUROfusion) and; (b) contact
dose rate vs. time for different steels from ref. [23].

recycling techniques are expected to apply below contact dose rates ≤ 2 mSv/h, whereas
for a contact dose rate ≤ 20 mSv/h recycling is still considered possible although through
more complex remote handling systems. For comparison, the hands-on limit (i.e., the
contact dose rate that allows a maximum dose of 20 mSv/y for a radiation exposed
worker) corresponds to 10 μSv/h.
In its ideal composition, EUROFER (Cr 9% W 1.1% Mn 0.4% V 0.2% C 0.11% Ta
% Si 0.05% N 0.03% Ti 0.01% and Fe for the rest) would achieve a level of radiation
below 2 mSv/h in less than 100 years and the hands-on limit in about 400 years. The
EUROFER produced today (the so-called EUROFER97, see fig. 7a) still contains some
impurities but can already reach the simple recycling limit in ∼ 100 years after shut
down, see ref. [23] and fig. 7b. Thus, for fusion power no geological repository will be
necessary.
The development of advanced steels that would allow a wider operating temperature
window is an active area of research. Oxide dispersion strengthened steel is expected
to allow operation up to 650 ◦ C. A review of the present status of materials for reactor
applications can be found in ref. [24].

5. – Demonstration of tritium self-sufficiency

Tritium does not exist in nature. It must be produced inside the reactor. A 1.5 GWe
reactor will consume about 0.5 kg of tritium per day and will need to produce the same
amount every day. A tritium breeding-ratio around 1.1–1.15, is the target for the self-
sustainability of a reactor.
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

Tritium is produced through the reaction between neutron and lithium

n + 6 Li → 4 He (2.1 MeV) + T (2.7 MeV),

n + 7 Li → 4 He + T + n − 2.5 MeV.

The reaction involving 6 Li is exoergic and provides an additional contribution to the

thermal power of the reactor. Its cross section at low energy is large (> 100 b at 1 eV).
The reaction involving 7 Li is endoergic. It has a threshold at 2.5 MeV so only high-energy
neutrons can be used. Its maximum cross section is below 1b. Therefore the lithium of
a reactor must be enriched in 6 Li. The tritium production using these two reactions has
been tested at low 14 MeV neutron fluence [25]. These show good agreement between
the calculated and experimental production of tritium.
Some of the neutrons are lost due to various processes (absorption, streaming through
the ports, etc.). Thus, neutron multipliers must be used. The main candidates are
beryllium and lead. The (n, 2n) reaction of Be has a lower cross section (∼ 0.5 b) but also
a lower neutron energy threshold (1.7 MeV). In the case of Pb the cross section is larger
(∼ 2 b) but the neutron energy threshold is at 7.4 MeV. The use of Be as a multiplier is
made in connection to the solid breeder concept in which ceramic compounds (Li4 SiO4 ,
Li2 TiO3 or others) are produced in the form of pebbles. The breeder is separated from
the Be multiplier and the two are cooled by Helium (water is avoided due to the reactivity
with Be). The use of Pb is made in connection with the liquid breeder concept in which
the breeder and the multiplier form a liquid eutectic compound (Li17 Pb83 ) that flows at
low velocity and is cooled by either He or water. There are advantages and disadvantages
in both concepts and one of the ITER goals is to test blanket prototypes to verify the
modelling assumptions [26].

6. – The intrinsic safety features of fusion

Fusion has a feature that makes it attractive within the nuclear technologies: its
intrinsic safety. No chain reactions take place in a fusion power plant. The main difference
between fission and fusion power plants is that a fission power plant is like a pile (the
Fermi pile!): all the energy that will be released is stored in the fuel initially placed in
the reactor. A recovery action following a malfunctioning during operations may need
to cope with the release of a large amount of power. A fusion plant is like a normal
gas boiler, in order to run it has to be continuously fuelled. If something goes wrong
it is sufficient to stop fuel from entering the reaction chamber and the reactor stops.
In both cases there is always a residual decay heat that has to be exhausted to avoid
melting of the confinement structures (this was indeed the problem with the Fukushima
accident). However, in the case of fusion the decay heat can be exhausted using only
passive conduction. Indeed the incidental studies done so far show that the temperature
of the in-vessel components in a fusion power plant always remains below the melting
temperature [27].
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

7. – The integration of advanced technologies

Fusion power plants require the integration of different technologies. A large number
of these technologies have been developed in connection with the ITER R&D activities
such as those related with large superconducting magnets and heating and current drive
systems. DEMO will have to integrate the technologies for tritium breeding (tested at
the level of proof-of-principle in ITER) and for the production of electricity through
a balance-of-plant (BoP) system. The integration of different technologies motivates a
focused effort on the solutions that can be really implemented in a reactor.
The technology of low-temperature superconducting magnets is today well estab-
lished. The challenges of the ITER magnets [28] have been addressed within a dedicated
R&D programme during the Engineering Design Activity [29]. The design and con-
struction of the magnet system for DEMO requires limited extrapolations of the ITER
design and ensures that a technical solution already exists while fusion-relevant high-
temperature superconductors are developed in parallel. R&D activity is ongoing on
more advanced low-temperature superconducting cables in order to reduce degradation
under cyclic operation and cost. The integration of the magnet systems into DEMO is
expected to largely build on the ITER experience.
ITER will test three auxiliary heating and current drive systems: neutral beam in-
jection, electron cyclotron heating and current drive and ion cyclotron heating. Their
use in ITER and DEMO poses specific technical challenges such as the development of
high-power, continuous source (gyrotrons) for millimetre wave radiation (170 GHz) for
electron cyclotron heating and current drive and large negative ion sources and accel-
erator systems for negative ion injection (1 MeV/200 Am−2 ). The choice of the system
will have to be made mostly on the basis of the impact on tritium breeding (the open-
ing in the vacuum vessel for the power injection are lost to the breeding blanket), on
the degree of reliability and on the demonstration of sufficiently high efficiency to avoid
large amounts of re-circulating power (see ref. [30] for a recent review and references
therein).
A large impact on the design will come from the requirements of minimizing the plant
down time through an effective remote maintenance system. The scheme presently under
development foresees the use of large vertical sectors (vertical maintenance scheme) [31].
The integration of the DEMO components with the BoP has been investigated in the
last few years. The choice of the BoP has a number of consequences on the choice of
blanket coolant and materials [32].

8. – Electricity at low cost from fusion

The cost of fusion electricity has been the subject of several studies using the standard
levelised cost of electricity approach by the IEA that includes the future expenditure
(capital, operation and maintenance, replacements, fuel and decommissioning charges)
all discounted to present day. With reasonable assumptions on the technology learning
factor (based on the present experience with several energy technologies), on the plant
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

availability (75%) and lifetime (40 years) the cost of the fusion kWh is estimated to be
between 5 c¤/kWh and 10 c¤/kWh [27] in line with present market values. Most of the
cost depends on the capital investment whereas fuel costs have a negligible impact.
DEMO is expected to be a relatively small extrapolation with respect to ITER. Its
major radius should be at most 50% larger than ITER if the present physics and tech-
nological basis will be confirmed by future R&D. Therefore extrapolations to DEMO of
the ITER experience in many areas can be made with confidence.
Nevertheless the experience with the ITER costs and with the costs of several nuclear
power plants under construction shows that particular attention has to be devoted to
this challenge. DEMO is not expected to produce electricity at a competitive price but
should demonstrate that the capital costs can be contained to a level that makes fusion
a competitive energy source in the long term.

9. – Stellarators

Achieving steady-state operation in a tokamak is made difficult by the need of oper-

ational regimes of high plasma pressure that are more prone to plasma instabilities and
require a sophisticated control system. A radical solution is to abandon the tokamak line
and consider non-axisymmetric configurations such as the stellarator [33]. In these con-
figurations the rotational transform that is needed to confine the plasma is not produced
by inducing a net plasma current but through a set of external coils.
A non-axisymmetric configuration has always regions of stochastic magnetic field that
act as a thermal short circuit and tend to degrade energy confinement. However, the
configuration can be optimized to reduce the size of these regions and their impact on
global confinement.
The difficulty of manufacturing external coils with a complex structure is balanced by
the avoidance of the plasma current. In addition, the absence of a plasma current also
reduces one of the major free-energy source driving plasma disruptions.
Stellarator research has made significant progress. The energy confinement is well
characterized and the exploitation of the new W7X device in Germany has started.
If the excessive occurrence of disruptions will make the tokamak line not suited for
commercial reactor applications, the stellarator line will be a potentially valid alternative
that will make full use of the R&D achievements obtained during the construction and
operation of DEMO.

10. – Concluding remarks

Fusion has the potential to become a major source of electricity. The research carried
out in the last 50 years allows today to routinely produce plasmas at reactor relevant
density and temperature. Substantial progresses have been also made in addressing the
key technological challenges.
The demonstration of plasma operations with alpha particles being the dominant
heating mechanism will be achieved on ITER. On the basis of the present theoretical
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

and experimental evidence a fusion gain Q = 10 is expected. Methods to avoid, prevent

or mitigate plasma disruptions and to control benign plasma instabilities have been
developed and must now be tested in ITER. ITER will also show the viability of fully
steady-state operations for reactor application.
The challenge of coping with the heat exhaust will be addressed in ITER through
the use of high-heat flux components based on the tungsten monoblock technology and
the use of partially detached divertor operation. In order to mitigate the risk that this
solution cannot be extrapolated to DEMO, schemes based on the use of advanced divertor
configuration and liquid metals as plasma facing materials are being investigated.
The qualification of structural materials that can withstand the intense neutron flux
of a fusion reactor and have benign activation properties needs to be further pursued.
Possible materials (such as EUROFER) have already been produced. They could be
used with limited extrapolations at the beginning of DEMO operation since the level
of nuclear damage is expected to be below 20 dpa for the first phase of the DEMO
operation. For the second phase of DEMO operation new materials need to be qualified.
This requires a specific facility (IFMIF) with a neutron energy spectrum that simulates
that of a fusion reactor.
Efficient tritium breeding is mandatory in a fusion power plant since about 0.5 kg of
tritium are burned every day and an equal amount has to be produced in the blanket
and re-circulated in the reactor. Blanket technologies based either on the liquid eutectic
LiPb compound or the solid ceramic breeder with Be as neutron multiplier are under
development and need to be integrated into a coherent design that accounts for the
limited operating temperature window of the presently available materials, the need of
minimizing tritium permeation outside the fuel cycle and the requirements of the balance
of plant system to produce a net electrical power output.
ITER will demonstrate many of the critical technologies for a fusion power plant:
magnets based on the use of low-temperature superconductivity, construction of large
scale vacuum vessel with strict tolerances, actively cooled plasma facing components
under high heat loads, high efficiency auxiliary heating system, fuel cycle technologies
capable of managing large tritium inventories. DEMO will largely benefit of this experi-
ence but will have to ensure reliable breeding blanket technologies and efficient balance
of plant systems at sufficiently low cost.
Stellarator is a promising alternative to the tokamak line due to its intrinsically steady-
state operation and lack of disruptions. The ongoing R&D will have to develop into a
burning plasma stellarator experiment in due time to demonstrate its viability as a fusion
power plant.
The National Ignition Facility in operation at Livermore since 2009 will demonstrate
the potential of inertial confinement fusion. For the first time fusion energy larger than
the fuel energy at the burn time has been achieved with about 50% of the reactions due to
self-heating [34]. Although the process to produce fusion energy is totally different from
that of magnetic confinement, many of the technological challenges (such as neutron re-
sistant material and tritium breeding) are common to magnetic and inertial confinement.
A successful R&D focused on the program priorities and the first operational experi-
EPJ Web of Conferences 246, 00013 (2020) [Link]
Joint EPS-SIF International School on Energy 2019

ence on ITER may allow the construction of a DEMO reactor and the demonstration of
fusion electricity around the middle of this century.

REFERENCES

[1] A European strategic energy technology plan (SET-plan) - Towards a low carbon future
COM(2007)723.
[2] Vienna Standard Mean Ocean Water, Reference Sheet for International Measurement
Standards (IAEA Vienna) Dec 2006, [Link]
[Link].
[3] Rosman K. J. R. and Taylor P. D. P. (Editors) Pure Appl. Chem., 70 (1998) 217.
[4] Atzeni S. and Meyer-ter-Vehn J., The Physics of Inertial Fusion (Oxford University
Press) 2004.
[5] Keilhacker M. (on behalf of the JET Team), Nucl. Fusion, 39 (1999) 209.
[6] Romanelli F. et al., Fusion electricity - A Roadmap to the realization of fusion energy
ISBN: 978-3-00-040720-8, [Link]
2013/01/[Link].
[7] [Link]
[8] Wesson J., Tokamaks, third edition (Clarendon Press, Oxford) 2004.
[9] ITER Physics Expert Groups on Confinement and Transport and Confinement
Modelling and Database, ITER Physics Basis Editors and ITER EDA, Nucl.
Fusion, 39 (1999) 2175.
[10] Bosch H. S. and Hale G. M., Nucl. Fusion, 32 (1992) 611.
[11] Hender T. C. et al., Nucl. Fusion, 47 (2007) S128.
[12] Fasoli A. et al., Nucl. Fusion, 47 (2007) S264.
[13] Jacquinot J. et al., Nucl. Fusion, 38 (1998) 2495.
[14] Loarte A. et al., Nucl. Fusion, 47 (2007) S203.
[15] Matsunami N. et al., IPP-AM-32 report Nagoya (1983).
[16] Piras F. et al., Plasma Phys. Control. Fusion, 51 (2009) 055009.
[17] Morris W. et al., IEEE Trans. Plasma Sci., 46 (2018) 1217.
[18] Albanese R. et al., Fus. Eng. Des., 146 (2018) 194.
[19] Baluc N. et al., Nucl. Fusion, 47 (2007) S696.
[20] Baluc N., Phys. Scr. T, 138 (2009) 014004.
[21] Knaster J. et al., Nucl. Fusion, 57 (2017) 102016.
[22] van der Schaaf B. et al., J. Nucl. Mater., 283-287 (2000) 52.
[23] Lindau R. et al., Fus. Eng. Des., 75-79 (2005) 95.
[24] Stork D. et al., J. Nucl. Mater., 455 (2014) 277.
[25] Batistoni P. et al., Fus. Eng. Des., 82 (2007) 2095.
[26] Federici G. et al., Fus. Eng. Des., 141 (2019) 30.
[27] Maisonnier D. et al., Nucl. Fusion, 47 (2007) 1524.
[28] Mitchell N., Fus. Eng. Des., 66 (2003) 971.
[29] Mitchell N. et al., Supercond. Sci. Technol., 20 (2007) 25.
[30] Franke T. et al., IEEE Trans. Plasma Sci., 46 (2018) 1633.
[31] Keep J. et al., Fus. Eng. Des., 124 (2017) 420.
[32] Barucca L., Fus. Eng. Des., 136 (2018) 1557.
[33] Klinger T. et al., Nucl. Fusion, 59 (2019) 112004.
[34] Hurricane O. A. et al., Nature, 506 (2014) 343.