Table of contents

Because plasma-boundary physics encompasses some of the most important unresolved issues for both the International Thermonuclear Experimental Reactor (ITER) project and future fusion power reactors, there is a strong interest in the fusion community for better understanding and characterization of plasma–wall interactions. Chemical and physical sputtering cause the erosion of the limiters/divertor plates and vacuum vessel walls (made of C, Be and W, for example) and degrade fusion performance by diluting the fusion fuel and excessively cooling the core, while carbon redeposition could produce long-term in-vessel tritium retention, degrading the superior thermo-mechanical properties of the carbon materials. Mixed plasma-facing materials are proposed, requiring optimization for different power and particle flux characteristics. Knowledge of material properties as well as characteristics of the plasma–material interaction are prerequisites for such optimizations. Computational power will soon reach hundreds of teraflops, so that theoretical and plasma science expertise can be matched with new experimental capabilities in order to mount a strong response to these challenges.

To begin to address such questions, a Workshop on New Directions for Advanced Computer Simulations and Experiments in Fusion-Related Plasma–Surface Interactions for Fusion (PSIF) was held at the Oak Ridge National Laboratory from 21 to 23 March, 2005. The purpose of the workshop was to bring together researchers in fusion related plasma–wall interactions in order to address these topics and to identify the most needed and promising directions for study, to exchange opinions on the present depth of knowledge of surface properties for the main fusion-related materials, e.g., C, Be and W, especially for sputtering, reflection, and deuterium (tritium) retention properties. The goal was to suggest the most important next steps needed for such basic computational and experimental work to be facilitated by researchers in fusion, material, and physical sciences. Representatives from many fusion research laboratories attended, and 25 talks were given, the majority of them making up the content of these Workshop proceedings. The presentations of all talks and further information on the Workshop are available at http://www-cfadc.phy.ornl.gov/psif/home.html. The workshop talks dealt with identification of needs from the perspective of integrated fusion simulation and ITER design, recent developments and perspectives on computation of plasma-facing surface properties using the current and expected new generation of computation capability, and with the status of dedicated laboratory experiments which characterize the underlying processes of PSIF. The Workshop summary and conclusions are being published in Nuclear Fusion45 (2005).

We are indebted to Lynda Saddiq and Fay Ownby, secretaries in the Physics Division of ORNL, whose special efforts, devotion, and expertise made possible both the Workshop and these Proceedings.

J T Hogan, P S Krstic and F W Meyer Physics Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6372, USA

Designing the interface between a thermonuclear plasma and the solid material environment is arguably one of the highest technical challenges of ITER and of the successful development of future fusion power reactors. Carbon-based materials are recognized to have superior thermomechanical properties and do not melt, but they retain high levels of tritium by co-deposition with eroded carbon that could severely constrain plasma operations. A carbon-fibre-composite (CFC) divertor target in ITER would survive long enough only with methods of mitigation/suppression of large edge localised modes (ELMs). Metallic materials, such as tungsten, would avoid the tritium retention issue, but melt layer losses due to large ELMs and disruptions may lead to severe damage and unacceptably short lifetimes. Maintaining plasma purity with high-Z materials remains a concern. A mix of several different plasma-facing materials is proposed in ITER to optimize the requirements of areas with different power and particle flux characteristics (i.e., Be for the first wall, CFC for the divertor strike point tiles and W elsewhere in the divertor). However, this will inevitably lead to cross-material contamination and the formation of material mixtures, whose behaviour remains uncertain and requires further investigation. Some of the most outstanding plasma–wall interaction problems that are still at issue in the design of ITER are briefly discussed in this paper, together with prospects for their resolution, by either R&D or design changes and/or dedicated operation provisions. They include: (i) plasma thermal loads and material damage during type I ELMs and disruptions, and prospects for mitigation, (ii) control of co-deposited tritium inventory in the carbon films, (iii) effects of mixed-materials and (iv) plasma operation with tungsten plasma-facing components. Directions and priorities of future research are proposed to narrow the remaining uncertainties and improve confidence in solutions in these areas.

The SOLPS suite of codes is used extensively to model the edge plasma of existing tokamaks, and for design studies of new tokamaks. At its core is the combination of a fluid plasma code, B2, and a Monte Carlo neutrals code, Eirene. Usually, the interaction with the wall is handled by Eirene, but B2 also has the capability of treating neutrals with a fluid model, and so both codes need to deal with the interaction of the plasma and neutrals with material surfaces. A number of recent enhancements to the plasma–wall models, particularly on the B2 side, have been made, and are described. In particular, the plasma–wall interaction model has been extended to include the modelling of mixed materials (such as is proposed for ITER).

We review the mechanisms of convective cross-field plasma transport in fusion related devices. We discuss the models of convective transport caused by the effective plasma gravity, as well as by the ∇ _⊥T_e and parallel shear of E × B drift instabilities.

Tritium retention by codeposition with carbon on the divertor target plate is predicted to limit ITER's DT burning plasma operations (e.g. to about 100 pulses for the worst conditions) before the in-vessel tritium inventory limit, currently set at ∼350 g, is reached. At this point, ITER will only be able to continue its burning plasma program if technology is available that is capable of rapidly removing large quantities of tritium from the vessel with over 90% efficiency. The removal rate required is four orders of magnitude faster than that demonstrated in current tokamaks. 18 years after the observation of co-deposition on JET and TFTR, such technology is nowhere in sight. The inexorable conclusion is that either a major initiative in tritium removal should be funded or that research priorities for ITER should focus on metal alternatives at least for DT operations.

The components of the neutral- and plasma-surface interaction model used in the Monte Carlo neutral transport code DEGAS 2 are reviewed. The idealized surfaces and processes handled by that model are inadequate for accurately simulating neutral transport behaviour in present day and future fusion devices. We identify some of the physical processes missing from the model, such as mixed materials and implanted hydrogen, and make some suggestions for improving the model.

New results are presented on the erosion rates of DIII-D and JET codeposits as a function of oxygen pressure (0.21–21 kPa) at three oxidation temperatures (523, 573 and 623 K). The initial D removal rate during oxidation strongly depends on temperature (rate increases with increasing temperature). A strong dependence is also seen on pressure (again, increasing rate with increasing pressure) at 523 and 573 K, but at 623 K the pressure dependence tends to saturate for pressures ⩾2 kPa. The final D content after prolonged oxidation, however, strongly depends on both temperature and pressure in all cases—D content decreases with increasing temperature and increasing pressure. The highest erosion rates observed were ∼2.5  μm h⁻¹ for the DIII-D and ∼3.5  μm h⁻¹ for the JET specimens.

Simultaneous interaction of low-energy ions and atomic hydrogen—a process named chemical sputtering—causes erosion of C:H surfaces which is significantly higher than the sum of the individual processes—chemical erosion due to atomic hydrogen alone and physical sputtering due to ions. Above that, this process occurs also at energies below the threshold for physical sputtering. A microscopic reaction mechanism for chemical sputtering was recently suggested. A mathematical model based on this reaction mechanism is able to quantitatively reproduce the energy dependence of chemical sputtering by atomic hydrogen and argon using just one free parameter. The model is briefly reviewed and the underlying physical processes are discussed. We present new predictions for the chemical sputtering of carbon by atomic hydrogen and different ions relevant to plasma–wall interaction in nuclear fusion devices (He⁺, H⁺, D⁺, T⁺ and Ne⁺) and new experimental data for the chemical sputtering due to hydrogen and Ne⁺ ions.

Chemical erosion of carbon based plasma-facing materials and the re-deposition of tritium containing carbon layers are critical processes for the material choice in ITER. At the occasion of the workshop on plasma-surface interaction related to fusion (PSIF) the present status of knowledge was defined, suggestions for improved measurements discussed and the importance of atomistic modelling of individual processes underlined. In the present paper, the status of knowledge for these processes is outlined and important unresolved issues are identified: the extrapolation of measured erosion yields at high-particle fluxes towards low energies as expected in detached divertor plasmas and the transport and deposition of hydrocarbon molecules to the divertor plate and walls.

We describe here an ORNL Physics Division research activity whose focus is the investigation of chemical sputtering of graphite by atomic and molecular D ions at very low energies that have so far been unexplored. Our initial experimental approach is based on the use of a quadrupole mass spectrometer (QMS) which samples the partial pressure of selected mass species produced in a scattering chamber when the incident ion beam strikes a graphite sample. While most of the ORNL results obtained to date are for D₂⁺ ions incident on ATJ graphite, preliminary results are shown for D⁺ projectiles incident at energies down to 10 eV D⁻¹, and for D₃⁺ ions incident at 10 and 4.5 eV D⁻¹. The possibility of obtaining complementary information using a time-of-flight approach is discussed as well.

Staff activities within the Atomic and Molecular Data Unit of the International Atomic Energy Agency are dedicated to the provision of comprehensive and up to date databases to quantify atomic, molecular and plasma–surface interactions for fusion-related research. Tritium production and subsequent behaviour are extremely important factors in the design of future fusion experiments. The Unit has sponsored several relevant projects, and has an active Co-ordinated Research Project (CRP) on 'tritium inventory in fusion reactors' focused on tritium retention issues and possible methods of removal. This CRP has already produced a number of significant results, and will conclude with a final Research Co-ordination Meeting (RCM) in 2006. Progress from the CRP is presented, along with the proposed work plan for the remainder of the project. On completion, comprehensive summaries of the results of the CRP will be published as a dedicated issue of the IAEA journal Atomic and Plasma-Material Interaction Data for Fusion.

Atomistic computer simulations, especially molecular dynamics, but also kinetic Monte Carlo simulations and electronic structure calculations, have proven to be a valuable tool for studying radiation effects in fusion reactor materials. In this paper, I will first review a few cases where these methods have given additional insights into the interaction between a fusion plasma and the first wall of a reactor. Then I will, in the spirit of the workshop theme of 'new directions in plasma–wall interactions' discuss some possible future avenues of research.

Simulations of amorphous carbon were performed at densities ranging from 2.0 to 3.0 g cm⁻³ with a reactive bond-order potential. The fraction of sp³ bonding increases with increasing density, as is observed experimentally, but with generally too much sp² content. Ring size distributions are calculated, with a number of large rings observed. It is suggested that structural quantities that are more directly related to physical properties—such as void volumes and coordination numbers—are more useful than ring size distributions in characterizing the structure of amorphous carbon. Void fractions and void volume distributions are calculated, indicating that a percolating void network exists at 2.0 g cm⁻³, large, non-percolating voids exist at intermediate density, and no voids are found larger than atomic volumes at 3.0 g cm⁻³.

The interaction of edge plasma with material surfaces raises key issues for the viability of the international thermonuclear reactor (ITER) and future fusion reactors, including heat-flux limits, net material erosion and impurity production. After exposure of the graphite divertor plates to the plasma in a fusion device, an amorphous C/H layer forms. This layer contains 20–30 atomic percent D/T bonded to C. Subsequent D/T impingement on this layer produces a variety of hydrocarbons that are sputtered back into the sheath region. We present molecular dynamics (MD) simulations of D/T impacts on amorphous carbon layer as a function of ion energy and orientation, using the AIREBO potential. In particular, energies are varied between 10 and 150 eV to transition from chemical to physical sputtering. These results are used to quantify yield, hydrocarbon composition and eventual plasma contamination.

A dynamic Monte Carlo simulation code, erosion and deposition based on a dynamic model (EDDY), which combines ion reflection and sputtering of plasma-facing materials with transport of released particles in plasma, is developed. The surface composition changes due to impurity deposition, collisional mixing, and thermal diffusion of the deposited impurities at elevated temperatures are taken into account. Ionization and dissociation of released atoms and molecules and their gyromotion in the plasma are calculated, so that prompt and local redepositions are simulated. This enables the simulation of erosion and redeposition patterns, and the impurity depth profiles of plasma-facing materials. Recent applications to analyses of plasma-wall interactions in existing fusion devices are presented.

The problem of plasma–wall interaction requires multiscale methods (molecular dynamics and kinetic Monte Carlo) for both the material and the plasma. One possible concept for such multiscale modelling is presented, including validation strategies.

The recombination of atomic hydrogen on graphite surface via the Eley–Rideal mechanism has been studied in the framework of the semiclassical time-dependent method and a potential energy surface assumed in our previous work. The results show that at the surface temperature, T_S=500 K, the probability for H₂ formation is large in a rather narrow range of collisional energies, from 0.01 up to 0.2 eV. At E_kin > 0.3 eV the recombination reaction is suppressed. Three other different mass combinations have been considered in the simulation: D+D_ad, D+H_ad and H+D_ad. A large isotope effect has been pointed out on the energy thresholds, the energy structure of the recombination probability as well as for the energy partitioning in the final states of the molecular species scattered in the gas phase.

We recently developed a framework for implementing a scaled self-interaction corrected density functional theory (DFT-SIC) into pseudo-potential plane-wave DFT. The technique implements the original method of Perdew and Zunger by direct minimization of the DFT-SIC total energy functional. By using maximally localized Wannier functions, DFT-SIC calculation can be carried out efficiently even for extended systems. Using this new development, the formation energies of defects in 3C–SiC (silicon carbide) were calculated and compared to more standard DFT calculations. Differences of up to 1 eV were seen between DFT and DFT-SIC calculations of the formation energies. When compared to DFT, DFT-SIC produced less-stable vacancies and silicon interstitials, and more stable antisites and carbon interstitials. The most favourable interstitials were found to be C interstitials in a C⁺–C⟨100⟩ dumbbell configuration, with the formation energy of 5.91 eV with DFT and 5.65 eV with DFT-SIC. Si interstitials were not as stable as C interstitials. The most favourable Si interstitial was found to be Si tetrahedral surrounded by four C atoms, with a formation energy of 7.65 eV with DFT and 8.71 eV with DFT-SIC.

The adsorption of hydrogen atoms on the (0001) basal plane of graphite is studied using periodic first principle density functional formalism. A model structure for the hydrogen-saturated surface is proposed according to the periodic boundary conditions. Quantum molecular dynamics using the Nosé–Hoover thermostat is carried on to study the saturated surface dehydrogenation by atomic oxygen at a temperature of 300 K.

The hydrocarbon fluxes released from a hydrogenated graphite surface due to thermal erosion and chemical sputtering become subject to a number of particle–surface interaction (PSI) processes in the near-surface region. These processes involve an electron transfer between the released particle and the surface and, thus, affect the charge state distribution of released hydrocarbons. The electron capture from the surface by a hydrocarbon ion to an excited dissociative state can lead to prompt dissociation of the formed molecule, producing two (or more) lighter fragments. These PSI processes (of resonant and Auger type) have so far not been included in the plasma–wall interaction studies. It is, however, obvious that they play a decisive role in determination of chemical composition and charge state distribution of hydrocarbon fluxes released from the surface and entering the edge/divertor plasma region. It is equally obvious that the same processes also take place when the hydrocarbon species approach the graphite surface from the interior of the edge/divertor plasma region, thus affecting their sticking to the surface.

We are focusing on developing a predictive theoretical, algorithmic and computational framework describing the many-body dynamics of the plasma–wall interactions by applying multiscale quantum-classical (QC) molecular dynamics (MD) and quantum chemistry approaches. Computer simulations of the dynamics of molecular systems often encounter serious limitations, of both phenomenological and quantitative nature, due to our inability to treat non-adiabatic transition dynamics and evolution beyond the ground Born–Oppenheimer electronic surface. This is, in particular, present due to charge transfer, collisional-cascade excitations and chemical sputtering of plasma ions impinging on a plasma-facing surface. A quantum mechanical treatment of the full non-adiabatic multibody dynamics is not currently feasible. To address existing deficiencies in knowledge of non-adiabatic MD, we propose excited state many-body dynamics by considering multiphysics described by time-dependent versions of the mean-field theories as well as QC Liouville equations, combined with multiresolution techniques for solving the quantum part of the problem.