2.1. Fusion

Inspired by the production of energy in the stars, we look at nuclear fusion as an energy source for a safe and long-term solution that can definitively replace fossil energy sources. However, the goal of nuclear fusion for energy in the laboratory remains elusive, after great global efforts. In recent years, steps have been taken with important advances in knowledge and through the construction of promising new experimental facilities. The International Thermonuclear Experimental Reactor (ITER)^[ Reference Bigot^49], currently in construction in Cadarache, France, aims to realize a gigantic plant to demonstrate for the first time a net production of energy from magnetic confinement fusion. The scheme of the installation, based on the tokamak concept, takes advantage of the knowledge acquired in decades of research with experimental installations of reduced dimensions, such as the Frascati Tokamak Upgrade (FTU, Italy)^[ Reference Pucella, Alessi, Amicucci, Angelini, Apicella, Apruzzese, Artaserse, Belli, Bin and Boncagni^50] and the Culham Joint European Torus (JET, UK)^[ Reference Rebut, Bickerton and Keen^51].

Laser-driven inertial fusion energy (IFE)^[ Reference Ghoranneviss and Elahi^52– Reference Batani, Colaïtis, Consoli, Danson, Gizzi, Honrubia, Kühl, Le Pape, Miquel, Perlado, Scott, Tatarakis, Tikhonchuk and Volpe^55] is based on the concept of ignition of inertial confinement fusion (ICF) that was proposed in 1972^[ Reference Nuckolls, Wood, Thiessen and Zimmerman^56, Reference Basov, Krokhin and Sklizkov^57], followed by a worldwide effort to demonstrate ignition of inertial fusion in the laboratory. IFE is finally developing today after decades of experiments and thanks to important achievements in the knowledge of fundamental physical processes. Recent historical breakthroughs in ignition have been achieved at the National Ignition Facility (NIF, United States)^[ Reference Moses^58] and are briefly discussed for completeness in Section 9.3.

Governments in various countries are undertaking the modernization and development of their power generation infrastructure. Although still in the laboratory development phase, fusion power is starting to play a vital role in future energy generation plans, as demonstrated by the emerging national fusion roadmaps issued by leading countries, including the United States, the UK, France, Germany and Italy, and by the increasing public and private funding of fusion companies, now exceeding 7 billion euros^{[ 59]}. The global fusion power market chart predicts significant growth in the coming years. Several factors are at play here, propelling growth in the global fusion power market. One of the most prominent factors is that in the coming years the world will see a massive demand for electricity. Rapid urbanization and modernization are further compounding the situation. It is believed that the population will increase by 2 billion, reaching 9.7 billion by 2050, with a probable consequent increase in energy needs^[ Reference Romanello, Salvatores, Schwenk-Ferrero, Gabrielli, Vezzoni, Rineiski and Fazio^60]. Moreover, as environmental awareness increases along with mounting concerns about power generated through conventional means, the acceptance of fusion power and its popularity has further increased.

2.2. Fission

Following the recent geopolitical situation arising from the Russia–Ukraine war, a review is ongoing on the planned shutdown of some existing nuclear plants that still have a great potential for energy production. Japan has also reopened many of the suspended reactors after the Earthquake in 2011. At the same time, new nuclear power plants are being planned and built all over the world. However, the current nuclear fleet is approaching the end of its useful life due to the lack of investment in new nuclear installations over the last 30 years. On the other hand, large-scale commercial reactors ( $>1$ GW, 7 billion dollars per unit) feature proven generating technology with a global deployment history.

Recognizing the affordability challenge of large-scale plants, the nuclear industry has proposed small modular reactor (SMR) technologies (300 MW–1 GW, 3 billion dollars per hybrid prototype-commercial unit) as an alternative. According to a feasibility study conducted by the UK National Nuclear Laboratory^[ Reference Waddington^61], there is a very significant global market need for energy that cannot be met in all circumstances by large-scale nuclear reactors, which presents a real opportunity for SMRs. The size of the potential global SMR market will be approximately 65–85 GW by 2035, valued at 250 billion to 400 billion dollars. We should note that the current worldwide electrical power consumption is approximately 25,000 MWh and is expected to increase by approximately another 15,000 MWh by 2035, of which the SMR could cover up to approximately 1000 MWh.

There is ongoing research on new types of nuclear power plants, including thorium-based designs and models that clarify the problem of nuclear waste storage using reprocessed fuel rods^[ Reference Humphrey and Khandaker^62]. These are inherently safer than the old power plants. The GIF has selected seven reactor technologies^[ Reference Kelly^63] for further research and development (R&D), namely the high-temperature gas-cooled reactor (HTGR), the very-high-temperature reactor (VHTR), the MSR, the supercritical water-cooled reactor (SCWR), the gas-cooled fast reactor (GFR), the sodium-cooled fast reactor (SFR) and the lead-cooled fast reactor (LFR). The most developed GIF reactor design, the SFR, has received the greatest share of funding over the years with a number of demonstration facilities operated, as well as two commercial reactors, operating in Russia. One of these, the MSR, is considered to potentially have the highest inherent safety of all models^[ Reference Emblemsvåg^64]. After years of experimental operations at Oak Ridge National Laboratory (United States) and even a commercial unit in Hamm-Uentrop (Germany), an experimental thorium-based MSR is to start operation in Wuwei, China.

With the right support, appropriate investments, licensing and public acceptance, certain nuclear designs could be available in a reasonable time frame. This might be the only way to achieve net zero by 2050.

In addition to energy production, the ADS was also proposed as an effective way to transmute radioactive isotopes contained in the spent fuel of existing reactors^[ Reference Gokhale, Deokattey and Kumar^3, Reference Gulevich, Kalugin, Ponomarev, Seliverstov and Seregin^65], thus contributing to the management of nuclear waste disposal. However, the ADS is not officially listed among GIF reactors^[ Reference Abram and Ion^66], as it is not a critical device. In contrast, all other GIF reactors rely on critical fission reactions to function. This is a feature of ADSs that is potentially attractive for reducing the investment in safety and operating costs. The European Sustainable Nuclear Industrial Initiative (ESNII) funds an ADS subcritical reactor, the Multi-purpose hYbrid Research Reactor for High-tech Applications (MYRRHA)^[ Reference Abderrahim, Kupschus, Malambu, Benoit, Van Tichelen, Arien, Vermeersch, D’hondt, Jongen, Ternier and Vandeplassche^67]. A similar project is in development in Japan (J-PARC).

3.1. Accelerator overview

Looking at the past 75 years, particle accelerators can be classified according to their technologies and the particle beam characteristics they are producing, namely beam intensity and/or power, final energy and spatial dimensions.

Concerning proton and ion accelerators, three main branches have emerged, namely electrostatic accelerators, the iconic Van de Graaff accelerator, with the tandem design accelerating high voltage (HV) up to 25 MV, and cyclotrons and synchrotrons, with circular accelerators using a high magnetic field combined with RF acceleration. The size and particle energy of circular accelerators range, respectively, from 1 m diameter for 1 MeV to 27 km for 10 TeV, while LINACs based on successive accelerating structures driven by RF electromagnetic fields feature a size and final particle energy spanning from 1 m for MeV to km for 100 GeV. By interactions with the appropriate target, these machines can produce secondary beams of intense neutrons, pions, positrons, muons and other so-called rare isotope beams.

Accelerators are extremely diffused globally and can be very different in cost, from thousands to billions of euros. They are part of a worldwide multibillion euro industrial market, as illustrated in Table 1. Accelerator science and technology is central to the most important advancements in medicine^[ Reference Peach, Wilson and Jones^70], materials^[ Reference Staron, Schreyer, Clemens and Mayer^71], energy^[ Reference Heidet, Brown and Tahar^72], security^[ Reference Chao and Chou^73] and climatology^[ Reference Elmore and Phillips^74] that emanate from nuclear physics (see, for example, Ref. [Reference Balosso, Baroni, Bleicher, Brandenburg, Burigo, Colautti, Combs, Cuttone, Debus, De marzi, Dobeš, Durante, Evangelista, Fagotti, Gales, Georg, Graeff, Haberer, Habrand, Maj, Mammar, Mazal, Meyroneinc, Mishustin, Olko, Patriarca, Pshenichnov, Roellinghoff, Röhrich and Wessels75]). The economic impact is significant: the US Department of Energy (DOE) estimates 500 billion dollars from particle beam accelerators; the European Physical Society estimated a total turnover from physics activity in Europe of 3700 billion dollars in 2010.

Table 1 Worldwide inventory of accelerators detailed for 2000 and 2014. Figures in italic for 2014 are not updated^{[ 68]}. The current total number is up to 30,000 accelerators. For a full review see also Ref. [Reference Widodo, Bastianudin, Triatmoko, hardjo, Diah, Adabiah, Andriyanti, Mulyani and Wijaya69].

The answers to some important questions facing our planet will come from interdisciplinary efforts in medicine, energy, climate and the marketplace. Today, the vast majority of accelerators are not used for fundamental science but for industrial processes and for applications relevant to society. Among these, the most notable applications include electronics, cutting and welding of electron beams^[ Reference Gonzalez-Martinez, Bachmatiuk, Bezugly, Kunstmann, Gemming, Liu, Cuniberti and Rümmeli^76, Reference Sun, Karppi and Mater^77], hardening materials^[ Reference Wang, Lei, Liu, Ma, Liu, Xiao and Shi^78], medical diagnosis^[ Reference Balosso, Baroni, Bleicher, Brandenburg, Burigo, Colautti, Combs, Cuttone, Debus, De marzi, Dobeš, Durante, Evangelista, Fagotti, Gales, Georg, Graeff, Haberer, Habrand, Maj, Mammar, Mazal, Meyroneinc, Mishustin, Olko, Patriarca, Pshenichnov, Roellinghoff, Röhrich and Wessels^75, Reference Wang, Chen, Santos Augusto, Liang, Qin, Liu and Liu^79], cancer treatment^[ Reference Balosso, Baroni, Bleicher, Brandenburg, Burigo, Colautti, Combs, Cuttone, Debus, De marzi, Dobeš, Durante, Evangelista, Fagotti, Gales, Georg, Graeff, Haberer, Habrand, Maj, Mammar, Mazal, Meyroneinc, Mishustin, Olko, Patriarca, Pshenichnov, Roellinghoff, Röhrich and Wessels^75, Reference Lennox^80, Reference Kiyanagi, Sakurai, Kumada and Tanaka^81], also including highly innovative concepts^[ Reference Panaino, Piccinini, Andreassi, Bandini, Borghini, Borgia, Di Naro, Labate, Maggiulli, Portaluri and Gizzi^82], monitoring air pollution and climate change^[ Reference Chmielewski, Zimek, Iller, Tyminski and Licki^83], examination and isotope dating of artifacts and ancient objects^[ Reference Kutschera^84], sterilization of food and medical goods^[ Reference Pillai and Shayanfar^85] and cargo scanning^[ Reference Tang^86]. Possible future applications to alternative energy sources^[ Reference Hossain, Taher, Das and Nucl^87] are also being developed. More than 400 billion euros of end products are produced, sterilized or examined using industrial accelerators annually worldwide.

More than 26,000 accelerators have been built globally over the last 60 years to produce particle beams for use in industrial processes, which does not include the 11,000 particle accelerators that have been produced exclusively for medical therapy. More than 24,000 patients have been treated by the recently introduced hadron therapy in the EU and more than 75,000 worldwide. About 200 particle accelerators are used for research with an estimated consolidated running budget of 1 billion euros.

3.2. Evolution of accelerators, beam energy and intensity frontiers

Here we describe the state-of-the-art and the recent progress in modern accelerator technologies (1980–2020). The three branches of accelerators have continuously evolved, reaching higher energies^[ Reference Aberle, Alonso, Brüning, Fessia, Rossi, Tavian, Zerlauth, Adorisio, Adraktas and Ady^88] and higher intensity^[ Reference Ma, Wen, Zhang, Yu, Cheng, Yang, Huang, Wang, Zhu, Cai, Zhao, Mao, Yang, Zhou, Xu, Yuan, Xia, Zhao, Xiao and Zhan^89] and beam power^[ Reference Leeper, Alberts, Asay, Baca, Baker, Breeze, Chandler, Cook, Cooper, Deeney, Derzon, Douglas, Fehl, Gilliland, Hebron, Hurst, Jobe, Kellogg, Lash, Lazier, Matzen, McDaniel, McGurn, Mehlhorn, Moats, Mock, Muron, Nash, Olson, Porter, Quintenz, Reyes, Ruggles, Ruiz, Sanford, Schmidlapp, Seamen, Spielman, Stark, Struve, Stygar, Tibbetts-Russell, Torres, Vargas, Wagoner, Wakefield, Hammer, Ryutov, Tabak, Wilks, Bowers, McLenithan and Peterson^90]. The collider machines (e^– and e⁺ laser sources, electron cyclotron resonance ion sources for P and ions, superconducting coils and RF superconducting cavities for acceleration) had reached maturity by the beginning of 2000; examples include the Large Electron-Positron (LEP) at CERN^[ Reference Taylor, Treille and Ser^91] (100 GeV e⁺, e⁻ collider), Tevatron at Fermilab^[ Reference Pasquinelli, Drendel, Gollwitzer, Johnson, Lebedev, Leveling, Morgan, Nagaslaev, Peterson, Sondgeroth and Werkema^92] (TeV p-pbar collider) and RHIC at Brookhaven^[ Reference Harrison, Peggs and Roser^93] (100 GeV heavy-ion collider). The world record of 13 TeV for p-pbar collisions was reached at CERN with a complex combination of its accelerators (LINAC, PS, SPS, LHC)^[ Reference Fartoukh, Kostoglou, Camillocci, Arduini, Bartosik, Bracco, Brodzinski, Bruce, Buffat, Calviani, Cerutti, Efthymiopoulos, Goddard, Iadarola, Karastathis, Lechner, Metral, Mounet, Nuiry, Papadopoulou, Papaphilippou, Petersen, Persson, Redaelli, Rumolo, Salvant, Sterbini, Timko, Garcia and Wenninger^94]. However, the energy frontier of the technology of ‘classical’ accelerators, for example, the next generation of accelerators at 1000 TeV, would confirm Fermi’s vision (1954): ‘with conventional technology, the accelerator would girdle the Earth’.

From the point of view of intensity and beam power frontier, in the last two decades, thanks to the development of superconductive (SC) RF cavities and high-intensity electrons or ion sources, the International Linear Collider (ILC) project, an electron and positron superconducting LINAC longer than 30 km and with energies over 500 GeV, is being considered^[ Reference Bambade, Barklow, Behnke, Berggren, Brau, Burrows, Denisov, Faus-Golfe, Foster, Fujii, Fuster, Gaede, Grannis, Grojean, Hutton, List, List, Michizono, Miyamoto, Napoly, Peskin, Poeschl, Simon, Strube, Tian, Titov, Vos, White, Wilson, Yamamoto, Yamamoto and Yokoya^95]. Synchrotrons also find an extremely important application in the high-brightness light sources based on FEL schemes^[ Reference Pellegrini, Marinelli and Reiche^96]. Accelerator-based light sources have taken advantage of these technologies, giving rise to high flux of dense and highly collimated photons ( ${10}^9$ photons/s from eV to 10 keV, such as SOLEIL^[ Reference Nadolski, Abeillé, Abiven, Bouvet, Brunelle, Buteau, Béchu, Chado, Couprie, Delétoille, Gamelin, Herbeaux, Hubert, Lamarre, Leroux, Lestrade, Loulergue, Marchand, Marcouillé, Nadji, Nagaoka, Zéphir, Ribeiro, Schagene, Tavakoli and Tordeux^97] and ELETTRA^[ Reference Wagner^98]). The high-intensity frontier was the objective of few e^–, p and ions SC LINAC machine projects, such as the following.

• • European X-ray Free Electron Laser (European XFEL), Hamburg, Germany: The most powerful coherent X-ray laser source, SC LINAC 15 GeV, 3.4 km. Total cost 1.22 billion euros (2005), and operating at full design value^[ Reference Aghababyan, Bacher, Bartkiewicz, Böckmann, Bruns, Clausen, Delfs, Duval, Fröhlich, Gerhardt, Gindler, Hatje, Hensler, Jaeger, Kammering, Karstensen, Keller, Kocharyan, Korth, Labudda, Limberg, Meykopff, Moeller, Penning, Petrosyan, Petrosyan, Petrosyan, Petrosyan, Pototzki, Rehlich, Rettig-Labusga, Rickens, Schlesselmann, Schoeneburg, Sombrowski, Staack, Stechmann, Szczesny, Walla, Wilgen, Wilksen, Wu, Abeghyan, Beckmann, Boukhelef, Coppola, Esenov, Fernandes, Gessler, Giambartolomei, Hauf, Heisen, Karabekyan, Kumar, Maia, Parenti, Silenzi, Namin, Szuba, Teichmann, Tolkiehn, Weger, Wiggins, Wrona, Yakopov and Youngman^99].

• • European Spallation Source (ESS), Lund, Sweden: MW class, state-of-the-art proton SC RF accelerator. The most powerful neutron source worldwide, which is currently in its commissioning phase^[ Reference Garoby, Vergara, Danared, Alonso, Bargallo, Cheymol, Darve, Eshraqi, Hassanzadegan, Jansson, Kittelmann, Levinsen, Lindroos, Martins, Midttun, Miyamoto, Molloy, Phan, Ponton, Sargsyan, Shea, Sunesson, Tchelidze, Thomas, Jensen, Hees, Arnold, Juni-Ferreira, Jensen, Lundmark, Gazis, Weisend, Anthony, Pitcher, Coney, Göhran, Haines, Linander, Lyngh, Oden, Carling, Andersson, Birch, Cereijo, Friedrich, Korhonen, Laface, Mansouri-Sharifabad, Monera-Martinez, Nordt, Paulic, Piso, Regnell, Zaera-Sanz, Aberg, Breimer, Batkov, Lee, Zanini, Kickulies, Bessler, Ringnér, Jurns, Sadeghzadeh, Nilsson, Olsson, Presteng, Carlsson, Polato, Harborn, Sjögreen, Muhrer and Sordo^100, Reference Miyamoto, Eshraqi, Levinsen, Milas and Noll^101].

• • MYHRRA, Brussels, Belgium: The multi-disciplinary project is based on an SC proton LINAC, 600 MeV, 5 mA. It was funded by Belgium and is set to finish construction in 2035 with an estimated cost 1.5 billion euros. It includes ADS transmutation, and isotope separation online facilities. The Phase-1, a 100 MeV accelerator is currently under construction^[ Reference Abderrahim, Kupschus, Malambu, Benoit, Van Tichelen, Arien, Vermeersch, D’hondt, Jongen, Ternier and Vandeplassche^67, Reference Van Oost, Terentyev and Abderrahim^102].

• • Système de Production d’Ions Radioactifs en Ligne de 2e generation (SPIRAL2), Caen, France: LINAC, 33 MeV $\mathrm{H}^{+}$ , 40 MeV $\mathrm{D}^{+}$ (5 mA) and 14.5 MeV for heavy ions with a mass-to-charge ratio A/Q $\le$ 3 (1 mA). In operation since December 2019^[ Reference Gales^103, Reference Orduz, Uriot, Emmanuel, Pierre-Lagniel, Savalle, Normand, Jamet and Di Giacomo^104].

• • Facility for Rare Isotope Beams (FRIB), East Lansing, United States: SC LINAC for heavy ions with a primary beam power of 400 kW and beam energies $\ge 200$ MeV/u. The first beam was in spring 2022^[ Reference Wei, Ao, Arend, Beher, Bollen, Bultman, Casagrande, Chang, Choi and Cogan^105].

All of these examples plus the summary graph shown in Figure 2 demonstrate the maturity of high-power accelerator technologies.

Figure 2 Beam power frontier for ion beam science and applications.

4.1. Accelerator-driven energy production (fast thorium reactor)

The choice of thorium (Th-232) as nuclear fuel in ADSs is particularly attractive, partly because it is more abundant than uranium in Earth’s crust. Another advantage of thorium is that it reduces proliferation risks (the produced U-233 is often contaminated with U-232, which decays emitting strong gamma, notably from daughter thallium-208 making handling, storage and weaponization of U-233 extremely challenging) as well as the radioactive hazards posed by heavy MAs in the spent reactor fuel (although it was observed that some daughters in the decay chain in the long term can reduce this advantage).

The most prominent proposer of an ADS using thorium fuel and optimized for energy production is Carlo Rubbia, who referred to this concept as the energy amplifier^[ Reference Rubbia, Abderrahim, Björnberg, Carluec, Gherardi, Romero, Gudowski, Heusener, Leeb, von Lensa, Locatelli, Magill, Martinez-Val, Monti, Mueller, Napolitano, Pérez-Navarro, Salvatores, Carvalho Soares and Thomas^4]. This is a reactor with a fast neutron spectrum and a subcritical core that contains thorium and an initial quantity of fissile material, such as plutonium from a conventional light-water reactor (LWR). Molten lead is chosen as the coolant, both because of its thermodynamic properties and because it does not slow down neutrons. In addition to extracting heat from the core, the coolant also acts as a spallation target for neutron production. To drive the system, a proton accelerator with an energy of 1 GeV and a beam current of about 10 mA is required. To achieve such specifications, one can use accelerators with circular geometry (such as cyclotrons) or LINACs. Since the power output (with a nominal thermal power of about 1500 MW) is to be regulated by variations of the proton beam intensity, no control rods are necessary. Reprocessing of the spent fuel consists of thorium replenishment and removal of fission products. The MAs are reloaded in the reactor, as their high fission probability with fast neutrons allows for their more efficient incineration (i.e., destruction by fission).

However, several technical challenges need to be overcome before a fully operational ADS could be built. The main technical challenge for the ADS has to do with developing a reliable particle accelerator with sufficient power to drive the reactor (an average proton beam power of 10 MW is required for the energy amplifier). At present, proton accelerators with energy in the 1 GeV range are limited to about 1.5 MW in terms of beam power. Another important figure of merit is the beam availability, which should be improved from the current 85%–90% level to at least 95% if commercial ADS operation is to be achieved. In addition, mastering the reprocessing of Th-based fuel remains a challenge at the industrial level.

4.2. Accelerator-driven transmutation

One of the main concerns of the public when discussing nuclear energy is the safe storage of nuclear waste, especially spent nuclear fuel. Since the 1990s, transmutation has been known as a special technique with which the radiotoxicity level of the radioactive waste can be reduced to that of mined uranium, that is, 300–400 years (instead of 300,000 years). The core idea is^[ Reference Tommasi, Delpech, Grouiller and Zaetta^107] to convert the particularly long-lived radioactive components into short-lived ones through nuclear reactions using free neutrons or by spallation.

The nuclear fuel spent in LWRs, which represent the majority of nuclear power plants operating around the world today, is approximately made up of the following elements^[ Reference Pershagen and Bowen^108].

• • Some 94% is still uranium, with an enrichment of approximately 1% (so, still more enriched than the natural one), which can be recovered.

• • Approximately 1% is plutonium, which is the main contributor to the long-term radiotoxic inventory. Plutonium can be recycled (PUREX process) and used to manufacture new fuel and be safely burned in thermal and fast reactors^[ Reference Broeders, Kiefhaber and Wiese^109], as already done for decades.

• • Some 4%–5% of the initial fuel mass really ‘burned’, producing fission products: their radiotoxicity reduces to the level of uranium extracted from mines in approximately 400 years, so they are not considered dangerous in the repository from the radiotoxic point of view.

• • Some 0.1% of the initial fuel mass is made up of MAs, that is, neptunium, americium and curium, which are the main radiotoxic contributors after plutonium (their decay time is of the order of 10,000 years, compared to 100,000 years for plutonium). Transmutation technologies focus on these nuclides (see Figure 3).

Figure 3 Radiotoxicity components of discharged liquid water reactor fuel (60 GWd/t). Adapted from Ref. [Reference Salvatores and Palmiotti110].

The basis of transmutation is that the ratio between fission to the capture cross-section is sensibly higher for MAs at higher neutron energies. Hence, when fast neutrons (i.e., $E>1$ MeV) are used, fission becomes more likely than capture (Figure 4).

Figure 4 Am-241 radiative capture and fission cross-section comparison versus energy.

The main reasons for adopting a subcritical system are as follows.

• • The control of a nuclear reactor system is given by a delayed neutron fraction, called $\beta$ . This number is sensibly lower for plutonium and MAs with respect to uranium, so, if proper transmutation efficiencies have to be achieved, as a proper amount of MA must be loaded in the fuel ( $>10\%$ ), a subcritical system (i.e., the infinite multiplication factor of the six-factor formula satisfies ${k}_{\mathrm{inf}}<1$ ) is needed.

• • Doppler feedback is reduced with the increasing number of MAs.

• • Critical reactors can transmute, in fact, 2–4 kg of MA per TWh, while a subcritical system can transmute up to 42 kg/TWh.

Following the idea proposed in 1980 in the United States by Takahashi and Bowman, the European Roadmap for developing a new approach using an ADS for nuclear waste incineration was promoted by C. Rubbia et al. in 2001^[ Reference Rubbia, Abderrahim, Björnberg, Carluec, Gherardi, Romero, Gudowski, Heusener, Leeb, von Lensa, Locatelli, Magill, Martinez-Val, Monti, Mueller, Napolitano, Pérez-Navarro, Salvatores, Carvalho Soares and Thomas^4]. MYRRHA^[ Reference Van Oost, Terentyev and Abderrahim^102] is the first project that couples a nuclear reactor to a proton accelerator that is currently under design^[ Reference De Bruyn, Abderrahim, Baeten and Leysen^111– Reference Abderrahim, Baeten, Sneyers, Schyns, Schuurmans, Kochetkov, Van den Eynde and Biarrotte^113]. It is a fast, subcritical system under development by the Belgian Nuclear Research Center (SCK-CEN) with EU (and possibly international) funding. The total cost is estimated at 1.2 billion euros, with the facility projected to be fully operational in 2038. The main objective of the project is to demonstrate the ADS concept at a power level (100 MW), which would allow future extrapolation to an industrial-scale reactor and also to study MA transmutation techniques. A lead-bismuth eutectic is used as the coolant, while mixed oxide (MOX) fuel forms the core of the reactor. A critical mode is also available for reactor research and radioisotope production.

In the framework of the EUROTRANS project, which aims to demonstrate the feasibility of transmutation in an advanced 50–100 ${\mathrm{MW}}$ experimental transmuter accelerator-driven facility (XT-ADS), it was determined that to drive an industrial transmuter, an accelerator should provide protons up to 800 MeV, with a current intensity of 20 mA^[ Reference Artioli, Chen, Gabrielli, Glinatsis, Liu, Maschek, Petrovich, Rineiski, Sarotto and Schikorr^114, Reference Podlech, Barbanotti, Bechtold, Biarrotte, Busch, Bousson, Dziuba, Junquera, Klein, Luong, Mueller, Olry, Panzeri, Pierini, Ratzinger, Tiede and Zhang^115] (which would generate approximately 3 $\times$ 10 ${}^{18}$ n/s). Sudden and unpredictable interruptions of operation (the so-called trips) typical of such accelerators^[ Reference Burgazzi and Pierini^116] are also considered, as they can significantly damage the reactor structures and the spallation target, thus also reducing the availability of the plant and consequently its economical convenience. Despite the very high beam stability and availability of modern accelerators (90%), superconducting RF and magnetic technologies are well known to give many tens of sudden interruptions per year (in the millisecond to second range). The tolerable maximum duration of a beam trip is assumed to be 3 s, with the number of trips per year not higher than five. The stability of the beam on the target, in addition, would require an operating range of approximately 1% with respect to energy and current, while it is 10% concerning position and size. The foreseen power of an accelerator for an industrial transmutation machine is of the order of 15–20 MW. In fact, to obtain the required neutron supply of ${10}^{17}$ – ${10}^{18}$ n/s, which is necessary for compensating the subcritical core, beam intensities of up to tens of milliamps are required, corresponding to a power of tens of MW. According to estimates, such an accelerator cost would account for approximately one-quarter of the cost of a plant such as the XT-ADS.

4.3. Industrial applications of the ADS

In medicine, radiotracers^[ Reference Hansen and Bender^117] and radiopharmaceuticals^[ Reference Vermeulen, Vandamme, Bormans and Cleeren^118] are widely used in clinical practice as imaging tools for the diagnosis and follow-up of tumour patients. Actually, a key objective here is the integration of the preclinical development of new chemical components with the industrial production of new radiopharmaceuticals, promoting clinical studies, now limited to a small number of radiotracers (e.g., 18F-fluoro-deoxiglucose, FDG), in combination with innovative therapeutic approaches. The development of radiopharmaceuticals is normally established using locally available accelerators to ensure rapid supply and use compatible with short-lived isotopes. Clearly, this application requires small and cost-efficient accelerators to allow affordable and safe^[ Reference Poli, Quaglierini, Zega, Pardini, Telleschi, Iervasi and Guiducci^119] in situ production and delivery.

Radiation damage research^[ Reference Campajola and Di Capua^120] (fusion, space) is another field of industrial applications for accelerators. The Fusion Prototypical Neutron Source (FPNS)^[ Reference Wiffen^121] is a neutron source that would simulate the harsh deuterium–tritium (D-T) fusion environment of a burning fusion device. It will help to build the scientific foundation required to design next-step materials. Although the complete set of the key parameters is still not agreed upon, the neutron source must be capable of delivering 10 displacements per atom (dpa) per year. Some 10 dpa per year can be produced using a 14 MeV neutron with flux of ${10}^{15}$ n/(cm ${}^2\cdot$ s) run per year. The Rotating Target Neutron Source (RTNS-II)^[ Reference Petö and Pepelnik^122], which operated from 1979 to 1987, produced D-T neutrons and was used for structure material experiments with an estimated flux of $1.2\times {10}^{13}$ n/(cm ${}^2\cdot$ s). Another machine for radiation studies is the proposed Materials Test Station (MTS) at the LANSCE accelerator at LANL^[ Reference Pitcher^123] with a peak fast neutron flux greater than ${10}^{15}$ n/(cm ${}^2\cdot$ s). The International Fusion Materials Irradiation Facility – Demo Oriented NEutron Source (IFMIF-DONES) is an upcoming neutron production facility that utilizes a 125 mA deuteron beam, accelerated up to 40 MeV impinging on a liquid lithium target to qualify materials to be used in future fusion power reactors^[ Reference Ibarra, Arbeiter, Bernardi, Cappelli, García, Heidinger, Krolas, Fischer, Martin-Fuertes, Micciché, Muñoz, Nitti, Pérez, Pinna and Tian^124, Reference Cara, Beauvais, Brañas, Bredy, Carmona, Chel, Comunian, Desmons, Facco, Gastinel, Gex, Gobin, Gournay, Heidinger, Knaster, Maebara, Marroncle, Massaut and Matsumoto^125].

4.4. Energy and yield requirements for a neutron source

Fission releases 200 MeV energy and two to three neutrons along with fission products (fragments) and consumes one actinide. Thus, we can estimate the actinide mass consumed and thermal power released for a given neutron rate and a multiplication factor ${k}_{\mathrm{eff}}$ . A single pressure water reactor releases 12 kg of MAs as its waste, assuming 1 GWe year. To completely transmute this amount with a direct neutron (i.e., no multiplication), ${10}^{19}$ n/s is needed. In contrast, ${10}^{17}$  n/s is needed for a system designed with ${k}_{\mathrm{eff}}=0.98$ . Either approach is accompanied by the release of 75 ${\mathrm{MW}}$ of power. An analysis of the amount of thermal energy released as a function of the neutron rate is shown in Figure 5 for various multiplication factors.

Figure 5 Actinide mass consumed and thermal power released for a given neutron rate and criticality.

5.1. Evolution of laser science and technology

The invention of the laser (T. Maiman, 1960; C. H. Townes, Nobel Prize, 1964) has changed the world by controlling the coherence of light. In the following decades two major discoveries have guided the rise of laser science and technologies towards high intensity (petawatt, PW = 10¹⁵ W) and very short pulses (femtosecond, fs = 10^–15 s), as a new probe, from the infinitely small to infinitely large physical systems. Ultra-intense laser interaction with matter can generate a formidable wave in a plasma that particles can surf along, gaining energy according to the so-called laser wakefield acceleration (LWFA) mechanism proposed by Tajima and Dawson (1979)^[ Reference Tajima and Dawson^126]. Less than two decades later (1985), the CPA method^[ Reference Strickland and Mourou^8] was demonstrated by D. Strickland and G. Mourou, who received the Nobel Prize 2018 for this achievement. In CPA, stretching the laser pulses in time is carried out to reduce the peak power, then the pulse is amplified to the desired energy and is finally compressed in time. At the end of the process, when the pulse is focused, extreme light intensity can be achieved. These two discoveries led to the present roadmap of ‘extreme light’ probing matter from the eV atomic regime (1980) to the multi-PW domain, approaching a focused intensity of ${10}^{24}$ W/cm ${}^2$ . At the same time, laser technology and relevant laser–plasma physics are advancing rapidly. Outstanding progress has been made in high-power laser technology in the last decade, with laser peak powers exceeding the PW level^[ Reference Lureau, Matras, Chalus, Derycke, Morbieu, Radier, Casagrande, Laux, Ricaud and Rey^127– Reference Gan, Yu, Wang, Liu, Xu, Li, Li, Yu, Wang, Liu, Chen, Peng, Xu, Yao, Zhang, Chen, Tang, Wang, Yin, Liang, Leng, Li and Xu^131]. Also, high-repetition-rate systems, that is, 100 Hz and beyond, with moderate peak power, that is, tens of TW and above, are starting to become available, paving the way to new and compact particle accelerators and photon sources.

5.2. Laser electron acceleration and laser requirements

Since the ‘dream beam’ results^[ Reference Mangles, Murphy, Najmudin, Thomas, Collier, Dangor, Divall, Foster, Gallacher, Hooker, Jaroszynski, Langley, Mori, Norreys, Tsung, Viskup, Walton and Krushelnick^132– Reference Faure, Glinec, Pukhov, Kiselev, Gordienko, Lefebvre, Rousseau, Burgy and Malka^134] emerged after the first impressive exploitation of ultra-intense lasers, LPAs have been developing at an unprecedented pace, with the most recent achievement of record electron energy exceeding 10 GeV^[ Reference Gonsalves, Nakamura, Daniels, Benedetti, Pieronek, de Raadt, Steinke, Bin, Bulanov, van Tilborg, Geddes, Schroeder, Tóth, Esarey, Swanson, Fan-Chiang, Bagdasarov, Bobrova, Gasilov, Korn, Sasorov and Leemans^6], and demonstration of staging^[ Reference Foerster, Döpp, Haberstroh, Grafenstein, Campbell, Chang, Corde, Cabadağ, Debus, Gilljohann, Habib, Heinemann, Hidding, Irman, Irshad, Knetsch, Kononenko, de la Ossa, Nutter, Pausch, Schilling, Schletter, Schöbel, Schramm, Travac, Ufer and Karsch^135], where the concept of an acceleration module, required for scaling to high energy, was first presented. At the same time, an effort continues to be made in understanding the electron injection process^[ Reference Kuschel, Schwab, Yeung, Hollatz, Seidel, Ziegler, Sävert, Kaluza and Zepf^136], separated from the wakefield generation and the subsequent acceleration process^[ Reference Stragier^137]. In fact, a breakthrough in the quality of electron beams is expected from the control of the electron injection to minimize beam emittance and energy spread, two of the most important quality factors of the accelerated bunch. Here several milestones have been achieved using different physical processes, and injection schemes continue to evolve^[ Reference Cabadag, Pausch, Schöbel, Buss-mann, Chang, Corde, Debus, Ding, Döpp, Foerster, Gilljohann, Haberstroh, Heinemann, Hidding, Karsch, Koehler, Kononenko, Knetsch, Kurz, de la Ossa, Nutter, Raj, Steiniger, Schramm, Ufer and Irman^138, Reference Xu, Bae, Ezzat, Kim, Yang, Lee, Yoon, Sung, Lee, Ji, Shen and Nam^139], moving towards an ‘optical engineering’ approach of the process where laser and plasma parameters are fine-tuned using accurate three-dimensional numerical simulations. In view of this, the control over laser and plasma input parameters and their stability is of paramount importance.

In parallel, a major effort is being dedicated to the transport of electron beams^[ Reference Cook, Carlsson, Moeller, Nagler and Tzeferacos^140] and the generation of secondary radiation through several mechanisms, ranging from the impact of the target for Bremsstrahlung emission^[ Reference Pomerantz, McCary, Meadows, Arefiev, Bernstein, Chester, Cortez, Donovan, Dyer, Gaul, Hamilton, Kuk, Lestrade, Wang, Ditmire and Hegelich^13, Reference Giulietti, Bourgeois, Ceccotti, Davoine, Dobosz, D’Oliveira, Galimberti, Galy, Gamucci, Giulietti, Gizzi, Hamilton, Lefebvre, Labate, Marquès, Monot, Popescu, Réau, Sarri, Tomassini and Martin^141] and the generation of positrons^[ Reference Sugimoto, He, Iwata, Yeh, Tangtartharakul, Arefiev and Sentoku^142, Reference Martinez, Barbosa and Vranic^143], to Thomson scattering for $\gamma$ -ray emission^[ Reference Sarri, Corvan, Schumaker, Cole, Di Piazza, Ahmed, Harvey, Keitel, Krushelnick, Mangles, Najmudin, Symes, Thomas, Yeung, Zhao and Zepf^144], to the undulator for the generation of X-ray FELs^[ Reference Fuchs, Weingartner, Popp, Major, Becker, Osterhoff, Cortrie, Zeitler, Hörlein, Tsakiris, Schramm, Rowlands-Rees, Hooker, Habs, Krausz, Karsch and Grüner^145]. From the point of view of transport of accelerated electrons, plasma lenses^[ Reference Thaury, Guillaume, Döpp, Lehe, Lifschitz, Phuoc, Gautier, Goddet, Tafzi, Flacco, Tissandier, Sebban, Rousse and Malka^146, Reference Lehe, Thaury, Guillaume, Lifschitz and Malka^147] are being studied as powerful, tunable devices capable of cm-scale focal lengths for high-energy beams, thus allowing compactness of LPAs to be preserved, while effectively transporting the electrons from one stage to another.

In the classical picture of LWFA^[ Reference Tajima and Dawson^126, Reference Macchi^148], a longitudinal electron plasma wave is excited by the ponderomotive force associated with an ultra-short, ultra-intense laser pulse. The electron plasma wave is characterized by a longitudinal electric field and a phase velocity set by the group velocity of the laser pulse, ${v}_{\mathrm{g}}=c\sqrt{1-{\omega}_{\mathrm{p}}^2/{\omega}_{\mathrm{L}}^2}$ , where ${\omega}_{\mathrm{p}}=\sqrt{n_{\mathrm{e}}{e}^2/{\varepsilon}_0{m}_{\mathrm{e}}}$ is the electron plasma frequency, with ${n}_{\mathrm{e}}$ being the electron plasma density, $e$ , ${m}_{\mathrm{e}}$ and ${\varepsilon}_0$ the electron charge and mass and the dielectric constant, respectively, and ${\omega}_{\mathrm{L}}$ the laser frequency. Electrons in phase with the wave are accelerated until, travelling faster than the electron plasma wave, they overcome the accelerating field of the wave and start experiencing a decelerating field. At high laser intensity, this classical scenario is significantly modified and numerical simulations provide detailed description of plasma wave excitation and evolution as well as electron injection and acceleration^[ Reference Rosenzweig^149]. The most compact configuration to obtain high-energy electron bunches from laser–plasma interaction is based upon a gas jet of a few millimetres, working in the so-called blowout regime^[ Reference Pukhov and Vehn^150]. Recently, several mechanisms have been identified and implemented to control the injection of electrons in a well-formed wake wave with the objective to achieve a localized injection of electrons with a limited longitudinal spatial extent, to ensure the reduced energy spread of accelerated electrons. Wave breaking is certainly the most fundamental process leading to the injection of electrons in a plasma wave. While transverse wave breaking^[ Reference Bulanov, Pegoraro, Pukhov and Sakharov^151] suffers from a delocalized injection of electrons and consequently large energy spread, longitudinal wave breaking via down-ramp^[ Reference Tomassini, Galimberti, Giulietti, Giulietti, Gizzi, Labate and Pegoraro^152] density-transition certainly provides more localized injection and limited energy spread of electrons. Activation of such injection schemes requires accurate control of the shape and profile of electron distribution, which can be achieved using custom gas targets and plasma tailoring. Recent successful implementations of this principle yielding very localized injection have been demonstrated that rely on plasma lensing^[ Reference Gonsalves, Nakamura, Lin, Panasenko, Shiraishi, Sokollik, Benedetti, Schroeder, Geddes, van Tilborg, Osterhoff, Esarey, Toth and Leemans^153] and the shock-front in gas jets^[ Reference Buck, Wenz, Xu, Khrennikov, Schmid, Heigoldt, Mikhailova, Geissler, Shen, Krausz, Karsch and Veisz^154]. Ponderomotive injection^[ Reference Umstadter, Kim and Dodd^155] also enables a high degree of control on the exact location of injection, but requires significantly more complex experimental configurations with additional laser pulses. A conceptually simple technique to enhance electron injection is the so-called ionization injection^[ Reference Chen, Sheng, Ma and Zhang^156], in which field ionization properties of some gases are exploited to increase the electron density in the bubble. Recent advances of this scheme also enable control of the spatial distribution of the ionization injection and consequent smaller energy spread. In the two-colour ionization injection^[ Reference Yu, Esarey, Schroeder, Vay, Benedetti, Geddes, Chen and Leemans^157] two laser pulses are used. The main pulse has a long wavelength and a large normalized amplitude. The second pulse, the ionization pulse, is a shorter wavelength, typically a frequency doubled, 400 nm Ti:sapphire (Ti:Sa) pulse. While the main pulse cannot further ionize the electrons in the external shells of the large Z dopant due to its large wavelength, the electric field of the ionization pulse is large enough to generate newborn electrons that will be trapped in the bucket. This opens up the possibility of using gas species with relatively low ionization potentials, thus enabling separation of wake excitation from particle extraction and trapping. A significant development of this concept has been proposed recently^[ Reference Tomassini, De Nicola, Labate, Londrillo, Fedele, Terzani and Gizzi^158, Reference Tomassini, Terzani, Labate, Toci, Chance, Nghiem and Gizzi^159] in the context of the EuPRAXIA project, in which ionization injection occurs in a large-amplitude plasma wave driven by a train of resonant ultra-short pulses. In this so-called resonant multi-pulse ionization injection scheme, the main portion of a single ultra-short laser system pulse is temporally shaped as a sequence of resonant sub-pulses, while a minor portion acts as the ionizing pulse. Simulations show that high-quality electron bunches with normalized emittance as low as 0.08 mm $\cdot$ mrad and 0.65% energy spread can be obtained with a single present-day 100-TW-class Ti:Sa laser system.

In another form of intense laser acceleration of electrons, known as direct laser acceleration (DLA), plasmas of near-critical densities are used as targets. In this case, the plasma frequency is too high to form wakefield structures, and the electrons that are pushed outward by the ponderomotive force leave behind a channel of positively charged ions and quasistatic electric and magnetic fields. Although these fields are weaker than the laser field, they significantly affect the behaviour of electrons in the channel. The equation of motion of these electrons is analogous to that of a driven oscillator^[ Reference Arefiev, Khudik and Schollmeier^160], implying a resonant behaviour. The two frequencies to consider are that of electron oscillations under the quasistatic fields of the channel, and that of the electron motion under the laser field. At sufficiently high plasma densities, the frequencies become comparable, resulting in the channel field having a greater effect on the phase of the transverse oscillations and causing their amplitudes to increase. Under this resonance condition, the channel fields rotate the transverse momentum into longitudinal momentum in each oscillation of the laser field. This yields electron energies higher than those that an electron would acquire by interacting with the laser field in vacuum^[ Reference Arefiev, Khudik and Schollmeier^160– Reference Gong, Mackenroth, Wang, Yan, Toncian and Arefiev^163]. DLA of electrons has been observed experimentally for over a quarter of a century^[ Reference Pomerantz, McCary, Meadows, Arefiev, Bernstein, Chester, Cortez, Donovan, Dyer, Gaul, Hamilton, Kuk, Lestrade, Wang, Ditmire and Hegelich^13, Reference Malka, Lefebvre and Miquel^164– Reference Cohen, Meir, Tangtartharakul, Perelmutter, Elkind, Gershuni, Levanon, Arefiev and Pomerantz^172]. Plasmas of near-critical density appropriate for DLA may be formed using gas jet systems, which are simple to operate with up to kHz rates^[ Reference Powell, Payeur, Fourmaux, Ibrahim, Kieffer, MacLean and Légaré^171], but their material selection is limited, or by using low-density polymer foam targets^[ Reference Willingale, Arefiev, Williams, Chen, Dollar, Hazi, Maksimchuk, Manuel, Marley, Nazarov, Zhao and Zulick^173, Reference Willingale, Nagel, Thomas, Bellei, Clarke, Dangor, Heathcote, Kaluza, Kamperidis, Kneip, Krushelnick, Lopes, Mangles, Nazarov, Nilson and Najmudin^174]. Alternatively, plasma plumes may be formed by pre-exploding a solid foil using a pre-pulse, allowing an expansion time of the order of nanoseconds before irradiation with the main pulse^[ Reference Pomerantz, McCary, Meadows, Arefiev, Bernstein, Chester, Cortez, Donovan, Dyer, Gaul, Hamilton, Kuk, Lestrade, Wang, Ditmire and Hegelich^13, Reference Cohen, Meir, Tangtartharakul, Perelmutter, Elkind, Gershuni, Levanon, Arefiev and Pomerantz^172, Reference Gizzi, Giulietti, Giulietti, Audebert, Bastiani, Geindre and Mysyrowicz^175, Reference Giulietti, Galimberti, Giulietti, Gizzi, Borghesi, Balcou, Rousse and Rousseau^176]. The energy spectrum of DLA electrons typically follows a Boltzmann-like distribution with an effective temperature in the MeV range, which scales as the square root of the laser intensity^[ Reference Pukhov, Sheng and Meyer-ter-Vehn^177]. While DLA may be less attractive than LWFA for certain applications due to its continuous spectrum and lower energies, it does have a notable advantage in terms of extremely high conversion efficiency of the incident laser energy to relativistic electrons, which can be as high as 23%^[ Reference Rosmej, Gyrdymov, Günther, Andreev, Tavana, Neumayer, Zähter, Zahn, Popov, Borisenko, Kantsyrev, Skobliakov, Panyushkin, Bogdanov, Consoli, Shen and Pukhov^178], making it an ideal choice for photo-neutron generation.

In view of the fine control of electron acceleration highlighted above, a new perspective for compact, all-laser-driven X-ray and $\gamma$ -ray sources is emerging rapidly^[ Reference Yi, Pukhov, Luu-Thanh and Shen^179, Reference Feng, Qin, Geng, Yu, Wang, Wu, Yan, Ji and Shen^180], aiming at a brightness currently achievable only at large-scale facilities such as synchrotron radiation facilities. Bremsstrahlung or fluorescence emission driven by fast electron generation in laser interaction with solids was demonstrated to provide effective ultra-short X-ray and neutron emission with unique properties^[ Reference Pomerantz, McCary, Meadows, Arefiev, Bernstein, Chester, Cortez, Donovan, Dyer, Gaul, Hamilton, Kuk, Lestrade, Wang, Ditmire and Hegelich^13, Reference Lemos, Albert, Shaw, Papp, Polanek, King, Milder, Marsh, Pak, Pollock, Hegelich, Moody, Park, Tommasini, Williams, Chen and Joshi^181, Reference Gizzi, Koester, Labate and Levato^182]. On the other hand, laser–plasma electron acceleration is being considered in place of conventional RF electron accelerators for a variety of radiation emission mechanisms. Broadband radiation generation schemes, including betatron and Bremsstrahlung, are being developed while Thomson scattering by collision with a synchronized laser pulse is being proposed for the generation of narrower band radiation. Moreover, with the increasing quality of laser-driven electron bunches that led recently to the demonstration of FEL emission^[ Reference Wang, Feng, Ke, Yu, Xu, Qi, Chen, Qin, Zhang, Fang, Liu, Jiang, Wang, Wang, Yang, Wu, Leng, Liu and Li^183] and the expected availability of high-repetition-rate laser drivers^[ Reference Gizzi, Koester, Labate, Mathieu, Mazzotta, Toci and Vannini^184], X-ray FEL facilities are also being designed^[ Reference Assmann, Weikum, Akhter, Alesini, Alexandrova, Anania, Andreev, Andriyash, Artioli and Aschikhin^29, Reference Walker, Alesini, Alexandrova, Anania, Andreev, Andriyash, Aschikhin, Assmann, Audet and Bacci^185].

Many of these sources and beamlines have been demonstrated in the laboratory environment and are now progressing towards implementation for industrial and medical applications. In the context of LWFA, given the ever increasing level of control and reliability of available schemes, compact, laser-driven accelerators are being explored for radiotherapy and diagnostics applications in areas where electron beams with energy of several tens of MeV up to a few 100 MeV, in the so-called very high energy electron (VHEE) range, are normally used as primary beams^[ Reference Labate, Palla, Panetta, Avella, Baffigi, Brandi, Di Martino, Fulgentini, Giulietti, Köster, Terzani, Tomassini, Traino and Gizzi^186]. A laser-driven electron accelerator may have several advantages compared to conventional LINACs, ranging from the small size of the acceleration region, to the possibility of multiplexing the electron source using a single laser driver. From the point of view of the specific properties of laser-accelerated electrons, given the ultra-short duration of laser-accelerated bunches compared to conventional RF LINACs, a new regime of ultrafast radiation biology is emerging^[ Reference Gauduel^187], which recently culminated in the identification of the so-called FLASH effect^[ Reference Favaudon, Caplier, Monceau, Pouzoulet, Sayarath, Fouillade, Poupon, Brito, Hupé, Bourhis, Hall, Fontaine and Vozenin^188] and related radiotherapy applications^[ Reference Borghini, Labate, Piccinini, Panaino, Andreassi and Gizzi^189].

5.3. Ion accelerating schemes and laser requirements

Ion acceleration can be driven by the laser–plasma interaction at a relativistic focused laser intensity of the order of ${10}^{18}$ W/cm ${}^2$ or above. The most established mechanism for ion acceleration is so-called target normal sheath acceleration (TNSA)^[ Reference Snavely, Key, Hatchett, Cowan, Roth, Phillips, Stoyer, Henry, Sangster, Singh, Wilks, MacKinnon, Offenberger, Pennington, Yasuike, Langdon, Lasinski, Johnson, Perry and Campbell^190] in which the laser interacts with a thin foil target. The accelerated ions are emitted in the direction normal to the surface of the target regardless of the laser angle of incidence, because the accelerating fields are produced by the space charge separation within the target set by escaping fast electrons. Accelerated nucleons are dominated by contaminants on the surface of the target. The process leads to a broad ion spectrum and the efficiency of the process relative to the pulse intensity scaled as approximately $\sqrt{I}$ ^[ Reference Keppler, Elkina, Becker, Hein, Hornung, Mäusezahl, Rödel, Tamer, Zepf and Kaluza^191, Reference Wilks, Kruer, Tabak and Langdon^192]. The maximum proton cutoff energy achievable with the peak power of the PW scale is typically of the order of $50{-}75$ MeV^[ Reference Keppler, Elkina, Becker, Hein, Hornung, Mäusezahl, Rödel, Tamer, Zepf and Kaluza^191], but several techniques have been proposed concerning laser and target configurations to improve laser coupling^[ Reference Torrisi, Calcagno, Giulietti, Cutroneo, bone and Skala^193– Reference Gizzi, Boella, Labate, Baffigi, Bilbao, Brandi, Cristoforetti, Fazzi, Fulgentini, Giove, Koester, Palla and Tomassini^195], beam quality^[ Reference Huang, Albright, Yin, Wu, Bowers, Hegelich and Fernández^196] and/or ion numbers.

Another ion acceleration mechanism, so-called radiation pressure acceleration (RPA)^[ Reference Robinson, Zepf, Kar, Evans and Bellei^197], promises improved cutoff energies and higher laser-to-ion conversion efficiencies, but is less accessible experimentally due to more demanding laser and target requirements. The optimum target foil thickness is defined by the intensity of the laser and the density of the target material, but it is typically on the nm-scale. The high-intensity laser pulse is incident normal to the target foil with circular polarization and relies on the radiation pressure acting on the electrons of the target plasma; it pushes through the back surface and accelerates the irradiated film portion. The field uniformly accelerates the target ions, thus leading to a nearly mono-energetic ion spectrum in ideal simulation cases. In addition, the acceleration process is efficient and scales favourably with the energy of the laser pulse ( $\sim I$ ). To date, the maximum cutoff energy reported is 93 MeV^[ Reference Kim, Pae, Choi, Lee, Kim, Singhal, Sung, Lee, Lee, Nickles, Jeong, Kim and Nam^198], but simulations point towards the capability of reaching several hundred MeV energies^[ Reference Wu, Gong, Shou, Tang, Yu, Mourou and Yan^199, Reference Chou, Grassi, Glenzer and Fiuza^200] (see Figure 6(a)). The prediction for the ideal case with a pulse duration of a few cycles leads to maximum efficiency and reduced beam instabilities. While few experimental studies of RPA have been reported to date, it is expected that with the new generation of intense laser facilities coming online new results are likely to emerge in the near future. Owing to the nature of real-world laser conditions, primarily finite contrast level, but also limited polarizations, most RPA demonstrations are likely to be hybrid results with the TNSA mechanism producing lower-energy background ions^[ Reference Ter-Avetisyan, Varmazyar, Singh, Son, Fule, Bychenkov, Farkas, Nelissen, Mondal, Papp, Börzsönyi, Csontos, Lécz, Somoskői, Tóth, Tóth, Andriy, Margarone, Necas, Mourou, Szabó and Osvay^201], as shown in Figure 6(b), with RPA or similar mechanisms contributing the higher-energy component. In fact, recent measurements confirm this scenario with the observation of record energies up to 150 MeV^[ Reference Ziegler, Göthel, Assenbaum, Bernert, Brack, Cowan, Dover, Gaus, Kluge, Kraft, Kroll, Metzkes-Ng, Nishiuchi, Prencipe, Püschel, Rehwald, Reimold, Schlenvoigt, Umlandt, Vescovi, Schramm and Zeil^202].

Figure 6 Advanced numerical simulation results of highest-energy RPA proton acceleration against experimental results. (a) RPA simulations^[ Reference Chou, Grassi, Glenzer and Fiuza^200]. (b) Hybrid RPA results^[ Reference Higginson, Gray, King, Dance, Williamson, Butler, Wilson, Capdessus, Armstrong, Green, Hawkes, Martin, Wei, Mirfayzi, Yuan, Kar, Borghesi, Clarke, Neely and McKenna^203].

5.4. Targets for laser-driven proton acceleration

In order to efficiently accelerate ions with lasers, the laser pulse needs to interact with a high-density target material – practically solid-state^[ Reference Maksimchuk, Gu, Flippo, Umstadter and Bychenkov^204, Reference Morrison, Feister, Frische, Austin, Ngirmang, Murphy, Orban, Chowdhury and Roquemore^205] or ultrahigh-pressure gas^[ Reference Puyuelo-Valdes, Henares, Hannachi, Ceccotti, Domange, Ehret, d’Humieres, Lancia, Marquès, Ribeyre, Santos, Tikhonchuk and Tarisien^206, Reference Ospina-Bohórquez, Debayle, Santos, Volpe and Gremillet^207]. In single-shot mode, all options are viable and have been demonstrated. Moreover, various complex targets (multilayer, foam, nanotube, etc.)^[ Reference Yazdani, Sadighi-Bonabi, Afarideh, Yazdanpanah and Hora^208] have also been suggested and partially tested, to enhance laser-to-ion efficiency and/or shape the spectra of the accelerated ions. Beyond single-shot mode, however, the choice of targets is limited. The debris originating from the interaction with a solid-state material contaminates the surrounding optics. Although solutions exist to protect the optics (i.e., transparent foils), considerable engineering is needed to keep these protections transparent over several hours of operation at high repetition rate. Such foils may also degrade the focusability of the laser beam, making additional wave front correction necessary. When dealing with large-size optics and vacuum conditions, the challenge is significant.

There are targets that are inherently debris-free and offer multiple-shot/rep-rated operation. These are liquid jets, cryo-cooled water or heavy-water ribbons and ultrahigh-pressure gas jets. The operation of a liquid jet can degrade the vacuum level in the chamber, even at the recently accepted relatively higher value of ${10}^{-3}$ mbar^[ Reference Treffert, Glenn, Chou, Crissman, Curry, DePonte, Fiuza, Hartley, Ofori-Okai, Roth, Glenzer and Gauthier^209]. The liquid jet vapour can be efficiently removed with cold fingers^[ Reference Füle, Kovács, Gilinger, Karnok, Gaál, Figul, Marowsky and Osvay^210]. However, careful engineering is needed – again – to avoid icing while keeping the system operational. The demonstrated repetition rate is of the order of 1 kHz, defined by the thickness of the jet, the flow rate and the interaction area^[ Reference Morrison, Feister, Frische, Austin, Ngirmang, Murphy, Orban, Chowdhury and Roquemore^205, Reference Füle, Kovács, Gilinger, Karnok, Gaál, Figul, Marowsky and Osvay^210]. The use of cryo ribbons^[ Reference Obst, Göde, Rehwald, Brack, Branco, Bock, Bussmann, Cowan, Curry, Fiuza, Gauthier, Gebhardt, Helbig, Huebl, Hübner, Irman, Kazak, Kim, Kluge, Kraft, Loeser, Metzkes, Mishra, Rödel, Schlenvoigt, Siebold, Tiggesbäumker, Wolter, Ziegler, Schramm, Glenzer and Zeil^211] is similar, with the addition that the cryo technology itself is quite bulky, complicated and causes vibration that is difficult to isolate within the interaction chamber.

The repetition rate of high-pressure gas jets is limited to around 100 Hz, mainly for technological reasons. In order to keep the pressure high (i.e., ensure the required density) in the vacuum chamber, the laser needs to be focused close to the nozzle. This close proximity degrades the nozzle, which therefore needs to be changed frequently, assuming continuous operation. Unlike single-shot solid-state(-like) targets, no efforts are known to make a complex target from any of the above technologies that are capable of multi-shot operation. Hence, further technologies need to be explored to enhance the laser-to-ion conversion efficiencies, such as post-acceleration or multi-jet operation. Although fundamental principles do not limit this technology, a proper cost–benefit analysis is necessary before adopting any of these target options.

5.5. RF and laser accelerators

Based on the current state-of-the-art accelerators, we can summarize the following. The most advanced superconducting RF technology is potentially capable of reaching the required specifications but with inherent difficulty related to the long construction times, especially for the case of proton beams with respect to electron beams. At the same time, producing relativistic protons and heavy-ion beams by RF accelerator technology is very demanding in terms of accelerating structures, real estate and money.

Indeed, the LPA is a ‘disruptive technology’ that is enabling new physics at the interface of atoms^[ Reference Pikuz, Faenov, Colgan, Dance, Abdallah, Wagenaars, Booth, Culfa, Evans, Gray, Kaempfer, Lancaster, McKenna, Rossall, Skobelev, Schulze, Uschmann, Zhidkov and Woolsey^212], plasma, quantum electrodynamics (QED) at the extremes^[ Reference Jirka, Klimo, Vranic, Weber and Korn^213] and nuclear physics^[ Reference Zamfir^214] while becoming off the shelf for many societal applications^[ Reference Margarone, Cirrone, Cuttone, Amico, Andò, Borghesi, Bulanov, Bulanov, Chatain, Fajstavr, Giuffrida, Grepl, Kar, Krasa, Kramer, Larosa, Leanza, Levato, Maggiore, Manti, Milluzzo, Odlozilik, Olsovcova, Perin, Pipek, Psikal, Petringa, Ridky, Romano, Rus, Russo, Schillaci, Scuderi, Velyhan, Versaci, Wiste, Zakova and Korn^215]. With adequate focusing, PW lasers can generate peak electric fields of the order of ${10}^{12}$ V/cm with relatively efficient conversion to relativistic electrons and photons with energies in excess of tens of GeV and protons and ions to hundreds of MeV within a few micrometres. Given the potential for energetic laser-driven particles, we can now consider experiments in laser laboratories that so far have been possible only with conventional nuclear accelerators and reactors. Plus, the significant reduction in size (currently the SLAC 50 GeV LINAC is 3 km long), ground dimensions (now km ${}^2$ ) and cost (billions of euros for RF technologies) can potentially be achieved by order of magnitude in some cases, particularly for high-energy machines.

Although emerging laser-driven technologies are very promising in terms of cost, size and available parameter range, the vast majority of experiments have been performed in laboratories where the operation of the laser system was not designed for nuclear facility counterparts. The cross-over of the gap between the laboratory and facility-based experiments was identified in Europe as a major step forward. For this to happen, high-power laser drivers will have to overcome two main limitations, namely the wall-plug efficiency and the repetition rate at high energy per pulse. At the same time, progress is expected in the acceleration regimes, understanding the specificity of many regimes and optimizing target properties versus achievable laser architectures and specifications that have the potential for scalability.

6.1. Fusion (and neutron stripping/knock out)

The ions accelerated are mostly protons and deuterons. For protons, all charge exchange processes require proton energy well above a few MeV to generate fast neutrons with exemplary reactions Be(p,n), Li(p,n) having a threshold energy of 2 MeV (Figure 7(a)). Proton knockout (p, xn) processes have large cross-sections at much higher energies (tens of MeV), while the optimum of spallation neutrons is obtained from protons with more than 500 MeV. For deuterons, the exemplary reactions D(d,n), T(d,n) do not have a physical threshold. The T(d,n) reaction has the largest cross-section around 140 keV, while the D(d,n) reaction reaches similar efficiency (yield) for deuterons with a kinetic energy level around 5 MeV (Figure 7(b)). All the experimental results and most of the predictions reveal that the most efficient, ion-based neutron generation can be achieved with hundreds of femtosecond, high-energy lasers. The demonstrated highest conversion efficiency per shot is $6.9\times {10}^8$ n/J with a sub-ps, 100 TW laser^[ Reference Günther, Rosmej, Tavana, Gyrdymov, Skobliakov, Kantsyrev, Zähter, Borisenko, Pukhov and Andreev^169], while it is almost two magnitudes lower for a 40 fs, 30 TW laser^[ Reference Zulick, Dollar, Chvykov, Davis, Kalinchenko, Maksimchuk, Petrov, Raymond, Thomas, Willingale, Yanovsky and Krushelnick^217] (Table 3).

Figure 7 Neutron yield versus the initiating ion energy for stripping (a) and fusion (b) (D-D and D-T) interactions. Panel (a) is adapted from Ref. [Reference Zimmer218]. Results in (b) are calculated in the same manner as in Ref. [Reference Tajima, Necas, Mourou, Gales and Leroy26]. Note: D-D, deuterium–deuterium.

Table 3 State-of-the art of neutron generation with lasers for the three major processes, neutron stripping/knock on (first ten rows), photo-neutrons (next four rows) and spallation (last two rows). Only the best results are shown here per process, in terms of neutron number per laser pulse energy. For comparison, the required average power of lasers that would be necessary to achieve ${10}^{15}$ n/ s is also shown, assuming that all the targetry issues are solved. Remark: all the values are taken from peer-reviewed papers (references available). The laser energy on target was taken as the reported energy reaching the primary target, regardless of the focal spot qualities. Hence, the values of neutron number per laser energy are conservative by a factor of 2–5.

The prediction for the highest average power neutron source based on the demonstrated yield, taking into account the state-of-the-art parameters of the lasers, and assuming solutions for the high-repetition-rate target parts, is approximately $6\times {10}^{10}$ (n/s)/kW (kHz, few-cycle laser, using the T(d,n) reaction). The high efficiency of particle acceleration with a few-cycle laser is yet to be demonstrated. R&D activities are needed towards (i) generation of spectrally shaped (quasi-monochromatic) protons or deuterons and (ii) enhancement of the efficiency of ion generation (via novel interactions and/or target structures).

6.2. Photonuclear process

Photo-neutron generation is a two-step process: laser-accelerated electrons generate gamma rays that subsequently interact with a nucleus to eject neutrons. The phenomenon of laser acceleration of electrons has been studied both theoretically and experimentally for over four decades. However, only a few groups have studied the laser electron acceleration and subsequent generation of gamma rays and neutrons^[ Reference Giulietti, Bourgeois, Ceccotti, Davoine, Dobosz, D’Oliveira, Galimberti, Galy, Gamucci, Giulietti, Gizzi, Hamilton, Lefebvre, Labate, Marquès, Monot, Popescu, Réau, Sarri, Tomassini and Martin^141, Reference Günther, Rosmej, Tavana, Gyrdymov, Skobliakov, Kantsyrev, Zähter, Borisenko, Pukhov and Andreev^169, Reference Cohen, Meir, Tangtartharakul, Perelmutter, Elkind, Gershuni, Levanon, Arefiev and Pomerantz^172, Reference Rosmej, Gyrdymov, Günther, Andreev, Tavana, Neumayer, Zähter, Zahn, Popov, Borisenko, Kantsyrev, Skobliakov, Panyushkin, Bogdanov, Consoli, Shen and Pukhov^178, Reference Pomerantz, McCary, Meadows, Arefiev, Bernstein, Chester, Cortez, Donovan, Dyer, Gaul, Hamilton, Kuk, Lestrade, Wang, Ditmire and Hegelich^219]. Since photo-neutron generation is a resonant reaction, that is, the cross-section has a peak around 14 MeV of gamma photons, only electrons in excess of 20 MeV will generate gamma rays with sufficient energy (Figure 8(a)). According to numerical simulations, the neutron yield is linearly proportional to the electron energy above 50 MeV (Figure 8(b)). The demonstrated highest conversion efficiency per shot is approximately $1.5\times {10}^8$ n/J with a sub-ps 100 TW laser, while it is three orders of magnitude lower for a 45 fs 60 TW laser^[ Reference Li, Feng, Wang, Tan, Ge, Liu, Yan, Zhang, Fu and Chen^220]. It is worth mentioning that the so far only multiple-shot photo-neutron generation^[ Reference Feng, Fu, Li, Zhang, Wang, Li, Zhu, Tan, Mirzaie, Zhang and Chen^221] was demonstrated by this laser, too. We stress that the prediction for the highest average power neutron source based on the theoretical calculations, taking into account the state-of-the-art parameters of the lasers, and assuming solutions for the high-repetition-rate target, is $6\times {10}^{11}$ (n/s)/kW (kHz few-cycle laser). We observe here that the gap between the theoretical prediction and the so-far experimentally demonstrated yield per kW average power is five orders of magnitude. R&D directions to advance further are (i) experimental demonstration of the predicted high efficiency with few-cycle pulses; (ii) optimization of the laser-generated electron energy spectrum to the resonance.

Figure 8 Cross-section of neutron generation in two isotopes of Pb in the function of gamma photon energy (a) and simulation of neutron yield per electron for various electron energies (b). Panel (a) is adapted from Ref. [Reference Essabaa, Arianer, Ausset, Bajeat, Baronick, Clapier, Coacolo, Donzaud, Ducourtieux, Galès, Gardès, Grialou, Hosni, Guillemaud-Mueller, Ibrahim, Junquera, Lau, Le Blanc, Lefort, Le Scornet, Lesrel, Mueller, Obert, Perru, Potier, Proust, Pougheon, Roussière, Rouvière, Sauvage, Sorlin, Tkatchenko, Verney, Waast, Rinolfi, Rossat, Forkel-Wirth, Muller, Bienvenu, Bourdon, Garvey, Jacquemard and Omeich222].

6.3. Spallation

Spallation requires the minimum kinetic energy of the protons to exceed 250 MeV. It is an attractive neutron source with a very high neutron yield for protons with energy greater than 600 MeV (Figure 9). So far, no laser-driven experimental setup has been capable of reaching this energy level – as discussed above, the state-of-the-art of laser-driven proton acceleration energy is approximately $100$ MeV^[ Reference Higginson, Gray, King, Dance, Williamson, Butler, Wilson, Capdessus, Armstrong, Green, Hawkes, Martin, Wei, Mirfayzi, Yuan, Kar, Borghesi, Clarke, Neely and McKenna^223]. The predictions for the highest average power neutron source based on the theoretical calculations, taking into account the state-of-the-art parameters of the lasers, and assuming solutions for the high-repetition-rate target parts, are approximately $1.2\times {10}^{10}$ (n/s)/kW for a 10 Hz, half-PW laser and approximately $4.3\times {10}^{12}$ (n/s)/kW for a 10 PW (1 shot/min) laser. Major ways to advance R&D further include (i) enhancement of the energy of the laser-accelerated protons with higher peak intensity and/or multiple stage accelerators and/or post-acceleration; (ii) an experimental demonstration of a laser-based neutron spallation source.

Figure 9 Neutron yield versus proton energy (a) using Monte-Carlo N-particle simulation. (b) Experimental results from the Paul Scherrer Institute (PSI) cyclotron. The simulation agrees well with the experimental values up to 1 GeV proton energies. Adapted from Ref. [Reference Amirkhani, Hassanzadeh and Safari224].

6.4. Towards laser-driven neutron sources for an LDADS

Among the three major mechanisms of laser accelerator generated neutrons presented here, spallation provides the highest yield per inducing particle. This could be a multiplicity of ten or more, depending on the proton energy. However, the usual spectra of laser-generated protons decrease exponentially towards high proton energies. Hence, the – simulated – conversion efficiency is less than 1. Moreover, the required minimum proton energy is yet to be demonstrated, as the necessary ultrahigh peak intensity laser is still years away from being built. In contrast, both fusion (stripping) and photonuclear processes in various arrangements have been studied experimentally with existing lasers of various parameters, although they offer a multiplication factor well below 1. Fusion requires laser acceleration of deuterons only to 140 keV, resulting in a modest peak yield of ${10}^{-5}$ n/D. For stripping, protons need to be accelerated to 10s of MeV, with a yield up to 0.2 n/p. However, the photo-neutron yield can be as high as 0.05 n/e for 180 MeV electrons. Efficient ion acceleration with lasers requires high-density targets, such as solid-state, liquid or ultrahigh-pressure gas jets – the latter is still to be demonstrated. Upon interaction, a large amount of debris is generated, which contaminates the optics, the vacuum components, etc. No standard solution has yet been found to overcome this problem. Hence, the only viable option currently consists of the use of liquid jets (or ultrahigh-pressure gas jets). Moreover, the laser pulse has to be very clean temporally (very high contrast). The production of such a high temporal contrast laser pulse at the required repetition rate (average power) poses a challenge on its own. With respect to electron acceleration, it works very well in gas jets or gas cells at modest pressure. The requirements of the necessary laser pulse parameters are much more relaxed, too. However, systematic experimental studies are needed to validate the ambitious theoretical prediction of laser-based photo-neutron generation with a few-cycle laser. A summary of the main key features of state-of-the-art laser neutron generation follows.

• • Laser-driven spallation is the highest-gain and the highest-risk approach. Once multi-10 PW peak power, rep-rated laser technology is demonstrated, this would require a quarter of a MW average laser power for the production of ${10}^{15}$ neutrons in a second. Taking into account the main HW components needed and the maturity of the acceleration mechanism, this approach has a technology readiness level (TRL) ranging from 2 to 3 (model-based validation).

• • Fusion (or stripping) is the most studied and experimentally known procedure, offering fewer surprises. According to the already demonstrated production rates, the necessary average power of the laser systems would be a few MW. This approach has a TRL ranging from 4 to 5 (test in controlled environment).

• • Photo-neutron generation is the easiest route, from a target engineering point of view, but it is not free from experimental risks. Here, the required average power is in the region of 10 MW. This approach has a TRL ranging from 4 to 5 (test in controlled environment).

Further studies on the generation processes may decrease the necessary laser average power by an order of magnitude, either through the optimization of the particle spectrum that generates the neutrons or by increasing the coupling efficiency of the laser–matter interaction. However, it is clear that lasers are the primary enabling technology for all these approaches to neutron generation. In the following section an overview of the current status of high-power, ultra-short pulse laser technology will be given, with the aim to identify possible candidate systems and scalable architectures for future industrial development.

7.1. High-peak-power lasers

Ultra-short pulsed, intense lasers can deliver a high peak power of coherent light with a minimal amount of energy, to be used as a driver for high gradient acceleration technology through the creation of a plasma. In seeking higher peak intensity laser pulses, the tactic is to begin with a low-energy, ultra-short pulse from an oscillator and to increase its energy level through a series of amplifiers. These amplifiers are based on gain media that transfer energy from pump sources into the broad spectral frequencies of the ultra-short signal pulse to be amplified (i.e., through amplified stimulated emission or optical parametric amplification). The gain medium mostly used for the shortest pulsed laser systems is sapphire crystal doped with Ti ${}^{3+}$ ions (Ti:Sa), whose optical properties support energy transfer to a wide spectral bandwidth ( $\approx$ 300 nm) and set the typical laser operating wavelength in the near-infrared (NIR) region of the spectrum ( $\lambda \sim 800$ nm). The pulse duration of these systems is typically in the femtosecond regime (1 fs: ${10}^{-15}$ s). At this pulse duration even small energies, that is, microjoules, are sufficient to reach intensities that degrade the beam quality within the gain media and eventually reach the point of damaging the material. This historically prevented progress towards increasing laser intensities. The introduction of CPA^[ Reference Strickland and Mourou^8, Reference Mourou^236, Reference Strickland^237] allows overcoming this material limitation by stretching the spectral frequencies of the short pulse in time, the so-called chirp. Doing so in a controlled manner permits the individual frequencies to be amplified towards the energy limit and then recombined in time within a compressor, as illustrated in Figure 11. Here, laser systems are referred to by their pulse peak power level, which is defined as pulse energy divided by the pulse duration, that is, TW class (1 TW = ${10}^{12}$ W), which is readily available. Several PW-level lasers are available at facilities around the world with the record for the highest power laser system set by the 10 PW HPLS of ELI-NP facility, having roughly 250 J within a 25 fs pulse duration^[ Reference Dancus, Cojocaru, Schmelz, Matei, Vasescu, Nistor, Talposi, Iancu, Bleotu and Naziru^239]. Along with the continued development of the achievable amplified pulse energy comes the potential peak power delivered in the focused intensity of these laser facilities. This value has been increasing over the years with the highest reported intensity being ${10}^{23}$ W/cm ${}^2$ by the 4 PW laser at CoReLS, in Gwangju, Republic of Korea^[ Reference Yoon, Kim, Choi, Sung, Lee, Lee and Nam^129] (see Figure 12).

Figure 11 Schematic of the CPA technique in the optical domain for the generation of high-peak-power, ultra-short pulse laser light. Adapted from Ref. [238].

Figure 12 Evolution of focused laser intensity showing the main advances and leading to the achievement of the ultra-relativistic regime.

7.2. High-average-power intense lasers

As anticipated, most high-intensity lasers to date use indirect (laser-pumped-laser) CPA architectures based on high-energy Ti:Sa amplifiers pumped by frequency doubled, Q-switched nanosecond Nd lasers. The pulse duration of Ti:Sa CPA lasers is around 30 fs at a wavelength centred around 800 nm with average power ranging from a few watts ( $\sim 0.1$ Hz) to $\approx 300$ W (10 Hz). They have already demonstrated acceleration of high-energy particles in the GeV range and schemes for producing higher particle energy via multiple staging. An important next step for the context of this document is the demonstration of higher pulse repetition rate operation of the order of 100 Hz or more at high average power, beyond the current limit of approximately $100$ W. Figure 13 shows the average power of various laser facilities and technologies. The increase of the pulse repetition rate of the laser systems is limited by several factors, including the pump laser technology and, specifically, the ability to manage the corresponding increased thermal load on the gain medium (Ti:Sa crystals), beam propagation optics and compressor diffraction gratings. Most of the existing facilities use pump lasers based on inefficient flashlamp technology. One exception is the HAPLS^[ Reference Borneis, Laštovička, Sokol, Jeong, Condamine, Renner, Tikhonchuk, Bohlin, Fajstavr, Hernandez, Jourdain, Kumar, Modřanský, Pokorný, Wolf, Zhai, Korn and Weber^241] (ELI Beamlines L3) laser developed by Lawrence Livermore National Laboratory (LLNL) in the United States which is the first PW-class laser to take advantage of the improved efficiency provided by diode pump laser technology of the pump lasers of the Ti:Sa architecture. The HAPLS pump engine has a design pulse energy of around 70 J at 527 nm generated from two Nd:glass, multi-slab amplifiers operating at 1053 nm and pumped with 885 nm diodes at 10 Hz, both cooled by helium gas at room temperature^[ Reference Bayramian^242]. A similar architecture and cooling approach are adopted in the final Ti:Sa amplifier. The system has recently been commissioned at ELI Beamlines and has demonstrated 16 J pulses at 28 fs and 3.3 Hz (53 W average power)^[ Reference Wang, Wang, Rockwood, Luther, Hollinger, Curtis, Calvi, Menoni and Rocca^243].

Figure 13 Chart of laser systems and respective average power, that is, pulse repetition rate versus pulse energy^[ Reference Gaul and Tóth^240].

Progress in flashlamp-based pump lasers has also been achieved with the Amplitude Technologies P60 pump system at the ELI-ALPS^[ Reference Nagymihály, Falcoz, Bussiere, Bohus, Pajer, Lehotai, Ravet-Senkans, Roy, Calvez, Mollica, Branly, Paul, Börzsönyi, Varjú, Szabó and Kalashnikov^244] facility, demonstrating 53 J at 532 nm with over 3 million shots at up to 10 Hz^[ Reference Mason, Divoky, Ertel, Pilar, Butcher, Hanuš, Banerjee, Phillips, Smith, De Vido, Lucianetti, Hernandez-Gomez, Edwards, Mocek and Collier^245]. Here flashlamps are used to pump multiple ceramic neodymium-doped yttrium aluminium garnet (Nd:YAG) thin-disk amplifier modules in an active-mirror configuration, cooled by liquid at room temperature. The final Ti:Sa amplifier also uses a room-temperature liquid coolant.

Direct CPA architectures, which combine the advantages of direct diode pumping of a broad bandwidth gain medium, offering the potential of higher efficiency, are also being developed. The PENELOPE (HZDR, Germany) and POLARIS (HI-Jena, Germany) lasers use crystalline Yb:CaF₂ and a mixture of Yb:glass/Yb:CaF₂ gain media, respectively, both pumped with diodes at 940 nm. However, the smaller bandwidth of Yb:CaF₂ compared to Ti:Sa limits the shortest pulse duration to 150 fs. PENELOPE is designed to produce an average power of 150 W (150 J at 1 Hz) at 1030 nm^[ Reference Albach, Loeser, Siebold and Schramm^246].

Alternative optical parametric chirped pulse amplification (OPCPA) architectures exploiting optical parametric amplification within large-aperture lithium triborate (LBO) crystals are also under development. These include diode-pumped lasers at the ELI Beamlines facility, L1 ALLEGRA (100 mJ at 1 kHz) and L2 AMOS (100 TW, 2–5 J between 10 and 50 Hz)^[ Reference Rus, Bakule, Kramer, Naylon, Thoma, Fibrich, Green, Lagron, Antipenkov and Bartoníek^247] and the Shenguang II Multi-PW beamline (SIOM, China)^[ Reference Xie, Zhu, Yang, Kang, Zhu, Guo, Zhu and Gao^248]. The OPCPA pump laser at the heart of the AMOS system is based on DiPOLE diode-pumped ceramic Yb:YAG multi-slab amplifier technology, cooled by cryogenic helium gas, developed by the Central Laser Facility at the STFC Rutherford Appleton Laboratory in the UK. DiPOLE pump lasers have demonstrated efficient and stable operation with average powers of over 1 kW (100 J at 10 Hz) at 1030 nm, pumped by 940 nm diodes^[ Reference Mason, Divoky, Ertel, Pilar, Butcher, Hanuš, Banerjee, Phillips, Smith, De Vido, Lucianetti, Hernandez-Gomez, Edwards, Mocek and Collier^245]. The properties of these lasers are shown in Table 4.

Table 4 A sample list of ultra-intense lasers with repetition rates above 1 Hz, with diverse gain media and their main parameters ^a .

^a Asterisks denote parameters of the lasers which are currently under commissioning.

Although progress over the past decade has been significant, with kW pump lasers for Ti:Sa systems emerging (e.g., P60 and DiPOLE), laser driver technology for particle accelerators still requires a further increase in average power of at least one order of magnitude, to 10 kW, and an increase in pulse repetition rate to greater than 100 Hz. To meet these requirements, all measures leading to a gain in efficiency will be crucial, especially if scaling to high-energy accelerators is foreseen. For comparison, in RF accelerators there is a great deal of attention to the wall-plug efficiency of RF power coupling to the accelerator cavity that is typically above 60%. To aim at such values, all the losses of efficiency will have to be tackled and diode pumping in place of flashlamps is definitely the first step. The major challenge here will be to increase the lifetime of the diodes while reducing the cost, possibly exploring new approaches that natively support higher frequency operation and production techniques that can scale to larger numbers to reduce costs. Potential improvements in diode laser efficiency and peak power scaling can also be achieved by cooling diode bars to near-cryogenic temperatures. A recent R&D project Cryolaser at the FBH facility demonstrated 1.7 kW peak power (1.2 ms, 10 Hz) from a single bar at 940 nm with greater efficiency than 60% when cooled to 223 K^[ Reference Crump, Frevert, Hösler, Bugge, Knigge, Pittroff, Erbert and Tränkle^249]. Increasing the diode power density and therefore reducing the number of bars required offers another approach to cost reduction for higher-power pump systems.

The Extreme Photonics Applications Centre (EPAC) at the STFC^[ Reference Mason, Stuart, Phillips, Heathcote, Buck, Wojtusiak, Galimberti, de Faria Pinto, Hawkes, Tomlinson, Pattathil, Butcher, Hernandez-Gomez and Collier^250] facility in the UK, over the next 5 years, will provide a unique capability to study applications of 10 Hz PW laser-driven accelerators and test strategies for power and pulse rate scaling. This includes a pathway to exploit 100 Hz DiPOLE pump technology currently under development at STFC, which would push PW-class lasers to an average power of 3 kW at an output energy of 30 J. On a similar trajectory, Amplitude Technologies is working on pathways to extend the repetition rate of the P60 pump laser to 50 Hz, and possibly 100 Hz, with minimum investment and no impact on thermal management, by replacing the flashlamp cassette with a diode cassette. With these performances it is possible to imagine using up to six of these next-generation pump lasers, possibly exploiting pulse interleaving techniques, to pump a Ti:Sa amplifier system and deliver sub-100 fs pulses with 10 kW (100 J, 100 Hz) average power before compression. Similar performances have been envisaged for the EuPRAXIA project and conceptual design of the required beamlines has already been delivered^[ Reference Gizzi, Koester, Labate, Mathieu, Mazzotta, Toci and Vannini^184]. Intermediate steps towards these goals are underway in projects such as k-BELLA at LBNL (United States) and KALDERA at DESY (Germany) aiming to deliver 3 J in 30 fs at 1 kHz, a first attempt to hit the kHz barrier. The architecture of this laser has not yet been chosen and there are several possibilities, all based on indirect Ti:Sa amplifier schemes, either cryo-cooled to improve efficiency and thermal management and pumped by Yb:YAG based lasers, or as thin disks with an active-mirror geometry, pumped by incoherently combined fibre lasers cryo-cooled with gas jets.

Ti:Sa lasers have proven reliability and higher TRL compared to other technologies and are the most widely used systems for LPA research. With the aim to develop higher average power systems, both the optical-to-optical and the overall wall-plug efficiencies of the laser become a concern as the laser cooling capacity, and complexity of the laser design, scales with the amount of energy converted into heat in the laser gain medium. For indirect CPA architectures the efficiencies for each laser, wavelength converter and all optical transfer stages must be taken into consideration and will eventually become a limiting factor due to prohibitive electrical power requirements. For this reason, a baseline Ti:Sa design requires the most effective strategies to reduce power consumption to an acceptable level while providing a viable solution compatible with mid-term application goals. For the purpose of this paper, scaling the pulse repetition rate to the kHz level and beyond becomes mandatory and electrical power consumption can be significantly reduced by adopting a direct diode-pumped CPA architecture.

7.2.2. Thin-disk lasers

In the past decade a strong research and technological development effort has been directed towards laser systems that combine ultrafast (i.e., from a few picoseconds or sub-picoseconds) pulse duration with a relatively high (in the range from several 10 W to multiple 100 W) average power, obtained from the combination of relatively low pulse energy with a high pulse repetition rate (e.g., several MHz). Besides the scientific applications, this interest has been motivated mainly by the development of new, high-precision material processing techniques.

Indeed, machining of delicate and heat-sensitive materials benefits from the use of ultra-short pulse duration, enabling results that cannot be reached with conventional laser processing with continuous or nanosecond-pulsed lasers^[ Reference Sugioka and Cheng^255]. As conventional Ti:Sa ultrafast systems based on 1–10 kHz regenerative amplifiers are unable to provide a sufficient machining speed for industrial applications, novel ultra-short pulse generation and amplification concepts based on other laser media, mainly Yb doped, have emerged in recent years. Different laser concepts were applied, starting from the known rod to disk or slab designs, and extending to fibre lasers. These new developments quickly provided reliable laser sources with ultra-short pulse durations for high-quality laser material processing. Thin-disk lasers are therefore now established at an industrial level and can provide kW average power and sub-picosecond pulse duration. In thin-disk lasers, the disk is mounted on a heat sink and functions as a high-reflection (HR) mirror for the laser resonator and a bending mirror for the pump beam, as shown in Figure 14.

Figure 14 Working principle of a thin-disk amplifier (adapted from https://www.rp-photonics.com/thin_disk_lasers.html).

We underline here that specifications of thin-disk lasers are not, by themselves, adequate to drive a plasma accelerator, but given the availability of robust, industrial-grade systems at high average power, different approaches are being pursued to make this technology play a role in particle acceleration. One approach consists of post compression of spectrally broadened pulses, as in the case of the multi-pass technique using a gas cell^[ Reference Viotti, Seidel, Escoto, Rajhans, Leemans, Hartl and Heyl^256]. This approach is currently being implemented at higher and higher peak and average power levels and is providing a possible alternative path to light sources with tens to hundreds of TW peak power and multiple kW of average power.

In this context, a completely different approach^[ Reference Jakobsson, Hooker and Walczak^257] for LPAs based on thin-disk technology relies on a plasma modulation effect that can convert a picosecond pulse in a train of ultra-short pulses that can resonantly drive an electron plasma accelerator. Experimental demonstration of this scheme is in progress and the perspective is that industrial kW, ps laser systems may enable efficient generation of high-energy electron beams.

7.2.4. Diode-pumped OPCPA systems

OPCPA uses a different approach to amplification of broadband laser pulses that is developing fast and holds the promise of delivering revolutionary performances to drive plasma accelerators and other major applications of such high-intensity lasers. Unlike CPA (Figure 15(a)), OPCPA (Figure 15(b)) uses parametric gain within a single pass through a nonlinear crystal that can be very high, so that OPCPA systems require fewer amplification stages and usually do not involve complicated multi-pass geometries. With optimized phase-matching conditions, the gain bandwidth can be very large, allowing femtosecond high-energy pulses to be generated. Thermal effects in the amplifier crystal, such as thermal lensing, are much weaker than in a laser amplifier, since there is only a small amount of heating due to weak parasitic absorption. This together with the very high quantum efficiency allows for scaling to very high energy and peak power levels, and also to a high beam quality of the amplified pulses. As the parametric gain occurs only within the duration of the pump pulse, one avoids the problems of power losses by amplified spontaneous emission in high-gain laser amplifiers. Also, one can easily generate high-energy pulses with very high-intensity contrast, that is, with a very low level of power before the actual pulse. Based on this approach, several high-energy systems are in operation^[ Reference Li, Leng and Li^271] or are being developed, including the DUHA system at ELI Beamlines^[ Reference Green, Bartoníek, Indra, Fibrich, Eisenschreiber, Novák, Majer, Tykalewicz and Rus^272], the SYLOS lasers at ELI-ALPS^[ Reference Toth, Stanislauskas, Balciunas, Budriunas, Adamonis, Danilevicius, Viskontas, Lengvinas, Veitas, Gadonas, Varanavičius, Csontos, Somoskoi, Toth, Borzsonyi and Osvay^273] and Vulcan at CLF^[ Reference Buck, Oliveira, Angelides and Galimberti^274].

Figure 15 Operating principle of the CPA scheme used in common titanium sapphire laser amplifiers (a) and the comparable OPCPA scheme (b). Adapted from Ref. [Reference Heyl, Arnold, Couairon and L’Huillier270].

7.2.5. Technology readiness of high-average-power laser drivers

The above description of the range of ultrafast, high-average-power laser technologies and architectures can be summarized by the plot of Figure 16, where the most advanced approaches are compared with respect to their estimated TRL and ultimate foreseen wall-plug efficiency. Clearly, the current industry standard for PW-class lasers, namely titanium sapphire, is established for scientific proof-of-principle demonstrator operations and will continue to serve, also in the foreseen nuclear applications development phase, for prototyping of components and subsystems. In the medium term, more efficient technologies will develop, with some of them eventually becoming the new standard for high workload, 24/7 applications, and industrial uses.

Figure 16 TRL of main laser driver technologies versus wall-plug efficiency, ranging from established flashlamp-pumped Ti:Sa systems to fibre systems with coherent combination, to diode-pumped technology for Ti:Sa, OPCPA, Yb:YAG thin-disk and direct diode-pumped CPA systems.

Each of these schemes has its own bottlenecks and critical issues, as outlined above. In some cases these issues concern the availability of materials (crystals, ceramics), while in others the issues are thermal management, quality, stability and complexity. All of these issues are currently being tackled at different levels in an increasing number of laboratories and solutions are emerging at a fast pace, making the risk related to the laser driver development manageable. A paramount example of these developments is the engagement of large laser infrastructures (e.g., Eupraxia, ELI) in further laser development, where the target is indeed the enhancement of the average power and the repetition rate to soon hit the barrier of the kW–kHz range where most of the envisaged applications become viable.

8.1. Laser developments

For the laser driver, as the most innovative subsystem of the installation, a development phase is foreseen, followed by a down-selection of the final technology among those outlined above. A fundamental risk concerning the laser driver lies in the possibility to reach the required repetition rate and efficiency. Depending on the technology, overall wall-plug efficiency and related thermal management of the laser components are the priority.

The development phase should proceed with a stepwise approach, with increasing average power of the single laser unit, and engineering to reduce costs and increase reliability. If the planned development activities reveal any physical constraints in this respect, a redesign of the laser system based on alternative technologies would have to be considered. A staged integration and a whole range of testing and prototyping activities will then be carried out to demonstrate scalability and projected efficiency, prior to production of individual laser units. Depending on the technology, high-level risks need to be considered for backup solutions in case the technical design R&D raises fundamental performance limitations.

Laser stability and operation reliability requirements are of primary concern for a viable plasma accelerator operation. This lies intrinsically in the possibility of multiplexing basic unit systems, including a laser unit and plasma targets, to the number needed to reach the required neutron level. A mitigation strategy could foresee the design of an alternative basic unit plasma accelerator with evolutionary technology.

8.2. Target system developments

Plasma target specifications will depend on the neutron production mechanism that will be down-selected after the development phase. In general, plasma targets, either solid/liquid targets or gas targets, are already provided as standard industrial components and require a modest adaptation to the laser-driven accelerator setup. The installation foreseen here will add significant challenges to the design and fabrication of components by requiring components that can withstand significantly higher repetition rates than the usual case in current plasma acceleration operation. To overcome these issues for the target design, the options are being explored conceptually and will be extensively tested in the coming development stage. This strategy will also help address the risk of simulation errors described above. In addition to experimental testing, the development phase will focus on cross-checking the simulation results with multiple codes and producing complete tolerance studies.

The development of secondary targets, from which neutrons are generated, seems less risky. On the one hand, in a scheme where neutrons are generated in the cooling material of the subcritical reactor, one can benefit from the already ongoing developments of accelerator-driven subcritical reactors, such as MYRHHA. On the other hand, if a neutron target is needed separately from the subcritical reactor, then one can consider the rich experience in design and operation of similar target systems that have been used in accelerator-driven neutron sources for decades (such as ISIS^[ Reference Thomason^275], ESS^[ Reference Andersen, Argyriou, Jackson, Houston, Henry, Deen, Toft-Petersen, Beran, Strobl and Arnold^276] and more).

8.3. Radiation safety and environment protection

Although based on novel laser acceleration technologies, the proposed LDADS installation will share issues similar to other ADS projects with regard to its radiological effects, materials and water consumption, environmental emissions and most other environmental impact factors. However, full studies are needed on how to design and implement safety and control systems in a comprehensive way to limit the environmental effects of the facility. This will be carried out in close collaboration with dedicated technical groups and thus build on extensive experience from other operational or planned machines. A specific benefit of this installation compared to other ADS designs is the potential for a considerable reduction of the accelerator footprint due to the potentially higher gradients of LPAs.

8.4. Operational stability and safety

One of the major criticisms of accelerator-driven subcritical reactors is that the existing LINAC or cyclotron technology so far is not free from beam tripping, which causes a huge impact on operational risk. This operational risk is of an economic nature primarily, as beam trips may cause the cease of the fission processes in the subcritical reactor, causing a complete stop of energy production for a few days. It should be emphasized that such beam trips do not affect the inherent nuclear safety of the subcritical reactor. An LDADS promises, by construction, more reliable and safer neutron production. It is worth looking at the major components and what risk of operation they mean in the case of unexpected failure.

8.4.1. Laser systems

The reliability and robustness of an LDADS are based on the multiple number of identical individual laser systems. To show the numbers, we assume that an LDADS consists of N units of identical laser systems, each specified for a trouble-free operation of H hours. We assume that the LDADS has a normal maintenance period of W weeks. The probability that one laser system becomes unavailable in between two maintenance periods is as follows:

$$\begin{align*}P(l)=\sum \limits_{i=0}^l\left(\begin{array}{c}N\\ {}i\end{array}\right){\left(1-\frac{H}{168W}\right)}^{N-i}{\left(\frac{H}{168W}\right)}^i,\end{align*}$$

where we assumed that $168W>H$ , otherwise the probability is zero. In numbers, if the maintenance period is 50 weeks, the trouble-free operation of a laser system is two months ( ${\sim}1440$ hours), and assuming the LDADS consists of 100 laser units, the probability of one laser failing between two maintenance periods is ${10}^{-7}$ .

The above consideration is also a key element of optimizing the number of laser systems and their requirement of trouble-free operation. If lasers are more reliable, then – from this point of view – fewer laser systems would provide the same reliability and vice versa. For instance, 10 laser systems would be enough if their trouble-free operation is 10 months (7250 hours). Hence, a given reliability can be ensured with the variation of the number of laser systems and their trouble-free operation.

9.1. Comparison of RF- and laser-based neutron sources

To drive a 100 ${\mathrm{MW}}$ ADS for green energy production, any neutron source would become technologically feasible once it is able to provide at least ${10}^{15}$   $\mathrm{neutrons}/({\mathrm{cm}}^2\cdot\mathrm{s}$ ) that is reliable 24/7/365. The important parameter for maintaining the stability of a subcritical reactor is the repetition rate of the external neutron source. Of course, the higher the repetition rate, the better, and the most continuous is the best. To establish the requirement for a pulsed neutron source, we need to consider that an ADS can typically tolerate a beam trip that lasts no longer than a few seconds. That is, a neutron source at a 1 Hz repetition rate may be just about satisfactory, but a 10 Hz repetition would provide a safer margin.

RF accelerator-based neutron sources have already demonstrated the production of the necessary neutron yield. They are capable of running at the necessary repetition rate (e.g., ISIS: 50 Hz, ESS: 20 Hz, MYRHHA: 20 Hz) and, in principle, generate neutrons 24/7. In practice, due to beam trips and further technological challenges, no RF accelerator-driven neutron source has been known to run uninterruptedly for more than a few weeks so far.

Current laser-based sources that provide the highest neutron number per laser shot run in single-pulse mode. A few hours-long continuous pulsed operation with a significant neutron yield around ${10}^6$ neutrons/s has just been demonstrated at 0.5 Hz^[ Reference Lelièvre, Catrix, Vallières, Fourmaux, Allaoua, Anthonippillai, Antici, Ducasse and Fuchs^38] and 10 Hz^[ Reference Stuhl, Varmazyar, Elekes, Halász, Gilinger, Füle, Karnok, Buzás, Kovács, Nagy, Mohácsi, Bíró, Csedreki, Fenyvesi, Fülöp, Korkulu, Kuti, Csontos, Geetha, Tóth, Szabó and Osvay^40]. Currently, one PW-class laser of ELI-ERIC is aimed at operating at 10 Hz, too, while several multi-10 TW lasers with kW-range average power are in construction for operating at 100 Hz. With the development of target technology, there appears to be no principal obstacle that would, in a few years, prevent a laser-based neutron source from operating 24/7/365 with a yield exceeding ${10}^{10}\ \mathrm{neutrons}/\mathrm{s}$ . As tens of kW-class lasers suitable for pumping are already available, it is foreseen that the ultimate neutron yield would peak around ${{10}^{12}{-}{10}^{14}}$   $\mathrm{neutrons}/\mathrm{s}$ per single laser system. Along with the above performance reachable in 10 years, we also estimate the overall plug-in efficiency increasing to a few percent, but it would remain still a few times lower than that of RF accelerators.

Ultimately, green energy production by the LDADS will possibly use a modular approach based on these developments for plasma accelerators, combining many such systems to reach the required MW average power, in a similar fashion as for IFE, where a 10 MW laser average power is envisaged for the future reactor, obtained by combining (incoherently) many multi-kW beamlines^[ Reference Batani, Colaïtis, Consoli, Danson, Gizzi, Honrubia, Kühl, Le Pape, Miquel, Perlado, Scott, Tatarakis, Tikhonchuk and Volpe^55]. As the operational stability of an ADS increases with the number of independent neutron sources, the currently foreseen number of 20–100 laser systems necessary to drive an ADS is rather beneficial.

We think that the choice between LDADS and ADS technologies would be defined by the actual business case, the leverage between the running costs and assets. With regard to long-term running costs, the efficiency of the wall plug and the stability of the operation are the two key factors. In other words, what is affordable: a lower plug-in efficiency but reliable operation, or a higher operation cost without the fear of stopping the ADS, hence energy production/transmutation. As far as assets are concerned, the laser systems need a much smaller footprint and basically no radiation shielding, while the RF accelerator has a large footprint and serious radiation protection enclosing the entire installation. The size of the subcritical reactor and its radiation protection seem the same for both cases.

9.2. Nuclear fuel enrichment

Firstly, in the nuclear industry, an important application is fuel enrichment. Currently, according to the International Atomic Energy Agency (IAEA), 415 nuclear reactors are in operation worldwide^{[ 277]}, among which, approximately 90% of them exploit fuels with enriched uranium (i.e., a material with a fissile nuclide U-235 content higher than the natural value, 0.711%), up to 5%. Only Canadian reactors (CANDU type), using heavy water, can directly exploit natural uranium. The process with which natural uranium is enriched in its fissile component, which is the isotope 235, is referred to as ‘enrichment’, and represents a very complex – but also important – technology. Uranium can be enriched with various technologies, such as gas diffusion^[ Reference Villani^278] or ultracentrifugation^[ Reference Koh, Almughlliq, Vaswani, Peiris and Mitchell^279], which require different amounts of energy. Recently, a technology based on laser application was developed and is already operational in some plants (the SILEX plant in Sydney, Australia). The separation of isotopes by laser excitation (SILEX) process^[ Reference Snyder^280] is the only third-generation enrichment technology at an advanced stage of commercialization today; its separation efficiency is much higher compared to centrifuge technology. Laser technology in fact presents the following characteristics:

• • inherently higher efficiency resulting in lower enrichment costs;

• • smaller environmental footprint than centrifuge and diffusion plants;

• • greater flexibility in producing advanced fuels for advanced SMRs;

• • it is expected to have the lowest enrichment plant capital costs.

The technology is based on the physical phenomenon linked to selective ionization of isotope 235 with respect to 238 if special lasers are applied: ionized uranium is then collected by using an electric or magnetic field. The technique can also be used for the separation of other types of nuclides. If a centrifuge consumes 50 kWh/SWU (separative working unit, which represents somehow the ‘effort’ needed to separate the fertile part of the fuel), early estimates mention 15 kWh/SWU with laser separation technologies^[ Reference Rigny^281]. More advanced laser technologies could further improve this value and the economical convenience of these kinds of plants.

9.3. Laser inertial confinement fusion

In August 2021 the NIF of the LLNL (United States) announced a major advance in reaching ignition, with 71% (1.35 MJ) of the 1.9 MJ input laser energy converted into products of D-T fusion reactions, namely neutrons and alpha particles^[ Reference Abu-Shawareb, Acree, Adams, Adams, Addis, Aden, Adrian, Afeyan, Aggleton and Aghaian^20]. This achievement was soon followed by the first ignition demonstration on 5 December 2022^[ Reference Danson and Gizzi^19] with 2.05 MJ laser input energy and 3.15 MJ output fusion energy, with 153% gain. The highest current record was reported in July 2023 with 2.2 MJ laser input energy converted to 5.2 MJ output fusion energy with a gain 236% and a total neutron yield well in excess of ${10}^{17}$ neutrons. With this result, the ignition milestone^{[ 282]}, which requires the fusion energy yield to be equal to or greater than the input laser energy, has been surpassed five times to date^{[ 283]}, demonstrating unambiguously the validity and feasibility of the ICF concept. This achievement is the result of continuous efforts with a number of experimental campaigns and theoretical developments aimed at tuning some of the many parameters that characterize the NIF design.

It is worth highlighting that laser-driven IFE is relying on similar high-efficiency and high-average-power laser technology that is expected to be needed for the green energy applications considered here. Synergy with the ongoing development of laser drivers for IFE can therefore be established and could significantly reduce the time needed to reach the required TRL.