Magnetic confinement experiments: plasma–material interactions, divertors, limiters, scrape-off layer (EX/D), stability (EX/S), wave–plasma interactions, current drive, heating, energetic particles (EX/W)

This paper summarizes the results presented at the 24th IAEA Fusion Energy Conference in the categories of plasma–material interactions, divertors, limiters, scrape-off layer (EX/D), stability (EX/S), wave–plasma interactions, current drive, heating, energetic particles (EX/W) in magnetic confinement experiments. In total, 149 papers including post-deadline papers have contributed to these categories. Several closely related papers, which are actually categorized in confinement (EX/C), have also been included. The understanding of experimental results has progressed remarkably, in particular, in the topics of resonant magnetic perturbation and ITER-like wall, which are the highlight of this conference. At the same time, identification of the bridging mechanism between the actuator and the consequence still requires further dedicated efforts so as to provide more accurate and reliable extrapolations to ITER and DEMO.


Introduction
The International Programme Committee identified the Terms of Reference to summary speakers as follows. (1) Capture the essence of the progress made. (2) What has the fusion community achieved over the past two years? (3) Where does fusion R&D stand right now? (4) Which critical issues, next steps and/or major challenges definitely need focused attention in the immediate future or medium term in order to ensure avoiding gaps and unnecessary delays/surprises on the way towards fusion demonstration? Drawing attention to the progress since the last 23rd IAEA-FEC in 2010 has led to the fulfilment of these Terms of Reference. The excellent summary of the same categories in the last FEC2010 was made by Dr Jean Jaquinot [1]. It describes the particular attention that had been given to issues in the critical path of ITER construction. This situation has been accelerated in these last two years. Urgent issues impacting the critical path of the ITER construction are becoming much clearer and the pressure of the demand for more distinct results, which would enable us to choose one among two or more options, is growing. For two examples, one can refer to the in-vessel edge localized mode (ELM) coils and the strategy of a full tungsten divertor in ITER. The last summary for FEC2010 also posed that the experimental results can be qualitatively understood but the extrapolation to ITER requires a more accurate knowledge of the plasma. This summary for FEC2012 confirms the progress in these last two years.
The presented contributions are summarized along with the highlighted topics; (1) pedestal stability and control which is primarily focused on ELM mitigation and suppression by resonant magnetic perturbations (RMPs), (2) reduction of heat load, (3) plasma-wall interaction (PWI) where the ITER-like wall (ILW) experiment on JET is a hot topic. And also (4) plasma-facing materials, (5) disruptions, (6) effect of rotation on magnetohydrodynamics (MHD), (7) spherical tokamak (ST) start-up, (8) plasma scenario and finally (9) waves and energetic particles (EPs). field emerges. In the last FEC2010, DIII-D [3], JET [4] MAST [5] and NSTX [6] reported related studies. The last summary for FEC2010 [1] pointed out the need to describe accurately the response of the plasma to an RMP and requested a more accurate knowledge of the plasma screening by the RMP. New results from more machines have been reported in the FEC2012. RMPs by in-vessel coils have become newly available in ASDEX-Upgrade [7] and KSTAR [8]. ELM mitigation by an RMP also has been demonstrated for stellartor/heliotron plasma in the Large Helical Device (LHD) [9]. In parallel with the demonstration of ELM mitigation and suppression, the remarkable progress in experimental observation of 3D magnetic fields in these devices should be emphasized.
DIII-D has reported RMP ELM suppression which can be extended to the ITER baseline scenario [10]. The n = 3 perturbation has demonstrated ELM suppression in a regime which matches the ITER target with the same plasma shape and I p /aB, and β N = 1.8 at ν * = 0.12 as shown in figure 1. The ELM suppression also has been demonstrated in the plasma with a significant fraction of helium (n He /n e = 0.25). They have identified a change in topology due to an RMP in the experiment by means of X-point soft x-ray imaging (see figure 2).
ASDEX-Upgrade has reported ELM mitigation with new saddle coils which generate the n = 1,2,4 perturbation [7]. Type-I ELMs were replaced by frequent small ELMs, which led to significant reduction in the peak divertor heat load and no penalty in energy confinement. In this demonstration in ASDEX-Upgrade, it should be pointed out that the ELM mitigation was insensitive to the resonance condition.
ELM mitigation has been established on MAST using an RMP with n = 3,4,6 [11]. The ELM frequency increased by up to a factor of 5 with a similar reduction in the energy released by a single ELM. The observation of the X-point lobe has indicated that the vacuum magnetic field gives an overestimate of the penetration. And one could get a good description of  the lobe structures if appropriate ideal screening exists at the edge (see figure 3). KSTAR has newly demonstrated ELM suppression by an RMP [8]. ELMs were suppressed at n = 1 for the first time. In this case, the density was pumped out initially by 10% and the stored energy dropped by 8%. They have also reported no mitigation with non-resonant anti-phase. In their study, electron cyclotron emission (ECE) imaging was highlighted as a powerful tool to identify what happens in edge plasmas, Figure 4. ECE imaging shows no edge filamentary structure when an ELM is suppressed at 4.3 s in KSTAR [12]. which showed that a filamentary structure disappeared when the ELM was suppressed [12] (see figure 4).
From JET, the progress of ELM mitigation by perturbation coils located outside of the vacuum vessel [13] has been reported. The type-I ELM changed to small and highfrequency (a few hundreds of hertz) ELMs in the high collisionality of ν * e = 2.0 at the pedestal while the increase in the ELM frequency was limited to 80 Hz in a low collisionality of ν * e = 0.8, which suggested saturation of the effect. ELM mitigation using external RMP coils has also been reported from LHD [9], where the interchange modes drove ELMs instead of peeling-ballooning modes. In this case, it should be pointed out that the resonance was located in the intrinsically stochastic edge layer which surrounds the nested flux surfaces.
Available experimental observations with an external nonaxisymmetric magnetic perturbation, which includes 'R'MP, are summarized in table 1. The inevitability of 'resonance' was not convincing and an overview of the results has suggested that the onset of island-formation/magnetic braiding may not be essential for the ELM control by 'R'MP coils. The mechanism of ELM suppression/mitigation by means of nonaxisymmetric magnetic perturbation awaits further physical clarification.
Quantitative understanding of the plasma response to these non-axisymmetric magnetic perturbations has been progressing, in particular, due to fine diagnostics and comparison with numerical computations, which showed clearly that plasma modifies the vacuum magnetic field. These results suggest an upgrade of the design base (e.g. in ITER) in terms of the Chirikov parameter [14].
With regard to ELMs, it should be noted that the first divertor ion temperature measurement has become available in MAST [15] and ASDEX-Upgrade [16], which improved the evaluation of the heat load in the presence of ELMs.
The importance of the plasma response to 3D magnetic fields is growing and the 3D plasma response to magnetic structure has been discussed in LHD [17,18] which is intrinsically 3D because of its stellartor/heliotron configuration. Comparison between the radial electric field E r gradient and the 3D MHD equilibrium modelling by the HINT2 [19] has indicated that the transition point to a strong E r gradient is a potential good index of the last closed flux surface [17] (see figure 5). LHD has also reported the experimental observation which suggests that RMP penetration depends on stability criteria such as magnetic shear and/or magnetic well/hill [18] (see figure 6).
Other interesting effects of non-axisymmetric fields on plasma have been reported from JET [20] and EXTRAP T2R [21] and the effect on fast-ion redistribution and loss has been discussed in a multi-machine study of ASDEX-Upgrade, DIII-D and KSTAR [22]. Analysis of the 3D magnetic field based on detailed diagnostics has been conducted on a highbeta tokamak [23] and RFPs [24,25].
In addition to the RMP, NSTX has reported the effect of pre-discharge lithium on ELMs [26]. With an increase in lithium deposition, global parameters generally improve and the ELM frequency declined to zero, and eventually ELMs were stabilized. This was explained by stability improvement due to edge plasma modification by wall conditioning.

Reduction of heat load
With regard to the reduction of the heat load on the in-vessel components, in particular, on divertor targets, remarkable progress has been seen in advanced divertors, novel efforts to reduce heat load, studies towards accurate evaluation   of heat load and suppression of impurity contamination. Innovative divertor concepts: super-X and snowflake have been investigated in MAST [15], NSTX [27], TCV [28] and DIII-D [29]. The initial experiment of the closed helical divertor has been reported from LHD [30]. Demonstrations of approaches to reduce heat load by means of impurity seeding [31][32][33] and geometrical modification [34] including application of non-axisymmetric 3D fields [35,36] have been reported. The progress of quantitative assessment towards an accurate evaluation of the heat load has also been remarkable [15,16,[37][38][39][40].
The snowflake divertor for mitigation of the divertor heat flux has been reported from NSTX [27] and TCV [28]. Both experiments have succeeded in the redistribution of divertor heat flux and demonstrated significant reduction in divertor heat load. The divertor heat flux width strongly decreased as I p increased in NSTX, DIII-D and C-Mod [41]. The snowflake divertor experiment in NSTX with P NBI = 4 MW, P SOL = 3 MW has demonstrated significant reduction in divertor heat flux (from 3-7 to 0.5-1 MW m −2 ) [27]. In TCV, the inverted AXUV radiation imaging has clearly shown the power redistribution to secondary strike points as the shape transitioned from a single-null configuration to a snowflake configuration. Eventually, the peak ELM heat load on the primary strike point was reduced by a factor of 3.5 [28]. DIII-D has also demonstrated that the snowflake divertor configuration strongly reduced the peak divertor heat flux and increased radiative volume. It should be noted that ELMs are mitigated in the snowflake divertor configuration in DIII-D [29] (see figure 7).
The influence of perturbation coils on divertor heat load has been reported from LHD [35] and ASDEX-Upgrade [32]. In LHD, an RMP stabilized the radiative divertor in high density operation while the case without an RMP underwent radiation collapse. In the case with an RMP, radiation increased by a factor of 3 without degradation of core confinement and divertor heat load was reduced by a factor of 3-10. The limited zone around the X-point of the remnant m = 1/n = 1 island was selectively cooled [35] (see figure 8). In ASDEX-Upgrade, an RMP reduced the peak power load and makes the strike line split up [32]. In both experiments, validation of the 3D simulation of edge plasmas has been promoted, which facilitates understanding of the plasma response to an RMP. A toroidally non-axisymmetric 2D heat flux profile has been investigated for natural ELMs and ELMs triggered by the application of a non-axisymmetric magnetic perturbation in NSTX. The rate of degree of asymmetry seemed to be related to the phase locking of the heat flux profile to the external 3D field [36].

Plasma-wall interaction: ILW
This topic can be rephrased as the ILW in this FEC2012. The in-vessel components in JET have been converted to those like ITER, i.e. beryllium wall and tungsten divertor, while  In the case with an RMP, radiation is stabilized against the density. (b) Radiation profile on the poloidal cross-section with an RMP simulated by using EMC3 [42] and EIRENE [43].
keeping the previous geometrical configuration when carbon wall and divertor were used. This careful modification enabled assessment of the effect of the ILW by comparing the data with the carbon wall and divertor [44].  JET has reported very encouraging results in terms of fuel retention. Retention with the ILW was reduced by one order of magnitude from that with CFC [45] (see figure 9). The main mechanism was attributed to a change in co-deposition. Belayers have a lower fuel content than a C-layer. This reduction of retention is in line with the predictions for ITER from CFC to Be/W walls [46] and could secure 1250 DT discharge (400s) in ITER before cleaning.
The multi-machine scaling of the retention rate has been derived through the efforts of the ITPA, which suggested that the retention rate during a shot is expected to range from ∼1 g of T h −1 in full tungsten up to ∼100 g of T h −1 in full carbon for a nominal ITER shot [47].
Characterization of the initial breakdowns of plasma discharges has been carried out for operations with carbon and the ILW in JET [48]. Non-assisted breakdown has been demonstrated to an ITER value as low as 0.35 V m −1 . Resultant lower radiation levels at higher density have been achieved making the breakdown more robust. Unlike the Cwall, no de-conditioning event following disruptions has been observed with the ILW. Neither glow discharge cleaning nor Be evaporation during operation are needed any longer. These preferable consequences were attributed to the drop of carbon by a factor of 20 from CFC to ILW [49] (see figure 10).
The investigation of plasma heating under an ILW will be discussed in section 9.

Plasma-facing material: wall conditioning and basic research of materials
Although plasma-facing material issues used to be out of the scope of the preceding series of this conference, they have become a critical subject, which involves wall conditioning and basic research of materials.
With regard to wall conditioning, international research on ion cyclotron wall conditioning coordinated by ITPA has been promoted, in particular, for the application of ICRF since glow discharge cleaning is not applicable to steady-state superconducting machines such as ITER. The database has been integrated into the ITER baseline using the planned ICRF heating system, and into experimental and modelling efforts coordinated by ITPA [50].
Lithium coating has been investigated in many experiments such as T-11M [51], NSTX [26,52,53], TJ-II [54], RFX [55] and EAST [56] to improve plasma performance and to evaluate the prospects of an application to DEMO. The importance of wall conditioning and the wall material to plasma performance has been suggested for a few decades. For example, the characteristic differences of H-mode with carbon and metal walls have been investigated in ASDEX [57] and a theoretical model to explain these differences has been proposed [58]. The extrapolation to ITER requires a more accurate knowledge of this plasma-wall issue, and the comprehensive assessment of new results from the ILW on JET, these lithium experiments and the other metal wall experiments including the past experiments could lead to the sought resolution.
In this topic, the importance of small basic experiments which were probably out of the scope of this FEC should be emphasized in two approaches. The first point is the basic characterization of materials. Since tungsten forms a FUZZ structure with helium, the consequences such as dust production are a concern. Basic research provides confidence that key growth parameters, from smaller devices [59][60][61][62][63][64], can be used for prediction in future devices. For example, tungsten nano-tendril growth has been investigated in Alcator C-Mod [59] (see figure 11) and arcing on carbon and tungsten materials has been compared on JT-60U, LHD and NAGDIS-II [60]. The second point is the development of in situ measurements of the PWI. Postmortem assessment apparently is not sufficient for understanding PWI processes. For example, qualitative local fuel retention is evaluated by the combination of laser-induced fuel desorption by spot laser heating and spectroscopic detection in edge plasmas in TEXTOR [65]. In Alactor C-Mod, a 1 MeV deuteron ion beam has been installed for the time-resolved, in situ measurement of boron coating and fuel retention [66]. Substantial use of these in situ methodologies in large magnetic confinement experiments is expected to bridge the basic understanding and characterization in reactor-relevant environment.
It would be said that dust [61] and fuel removal [67] remain areas where considerable effort is still needed for the future success of ITER and DEMO.
With regard to runaway electrons, for example, DIII-D experiments provided the physics basis for runaway electron control in ITER. Radial stability provided time to dissipate runaways, which enable high-Z gas injection to increase dissipation of runaway electrons [68] (see figure 12). In Tore Supra, a new concept of neon gas injection by means of a bursting disk cartridge injector has been demonstrated [69].
In JET, the dynamics of disruptions were very different with the ILW due to higher plasma purity and consequent lower radiation during disruption. Massive gas injection (MGI) as a disruption mitigation tool is now mandatory for JET. With the mitigation by MGI, the forces and power loads resulting from disruptions are returned to the levels with a C-wall [74] (see figure 13). It should also be noted that no runaway electron formation has been observed with the ILW [70].
In NSTX, low disruptivity at a relatively high β N as high as 6 with low l i has been found from the vast database. This empirical observation was consistent with specific disruption control experiments. The operational envelope on the plane of resonant field amplification (RFA) and β N /l i showed a peak around β N /l i of 10 [75] (see figure 14). The disruption warning algorithm showed a high probability of success. ∼98% of disruptions can be flagged with at least a 10 ms warning [78].

Effect of flows on MHD instability
Several MHD studies have discussed the effect of flows on instabilities such as neo-classical tearing modes (NTMs) and sawteeth [81][82][83][84][85][86]. Plasma rotation alteration and MHD stability have been investigated in KSTAR. The H-mode operation of KSTAR has been expanded towards higher β N and lower l i and 3/2 and 2/1 tearing modes were characterized with   their dependence on plasma rotation (see figure 15). Alteration of rotation profile has been demonstrated by the application of n = 1,2 non-axisymmetric fields [81]. In Alcator C-Mod, NTMs have been observed in ICRF heated high-performance Imode plasmas.The NTMs slowed down plasma rotation, which is intrinsic rotation driven by ICRF mode conversion, and also affected plasma performance [82].  [81]. Plasma rotation shear is found to have a stabilizing effect on the 2/1 mode.

ST start-up
Efficient start-up of a ST is a prerequisite for the next generation ST, therefore, non-inductive start-up and current drive using lower hybrid waves [87,88], electron cyclotron waves [89], electron Bernstein waves [90] and helicity injection [91,92] have been intensively studied.
The development of non-inductive current drive by LHCD has been implemented in Globus M and this scenario will be the focal point of the next program Globus-M2 [87]. NSTX has demonstrated that L-mode discharge ramping to 1MA required 35% less inductive flux when coaxial helicity injection (CHI) was used [91] (see figure 16).
With regard to the heating of ST, ST merging produced bi-directional outflows, causing reconnection heating with high power up to 5 MW. It has been demonstrated that a low-β ST was transformed to a high-β ST with β of 40% with an absolute min-B profile [93]. This proof of principle promotes the application of this scheme to a larger ST, MAST.
Intelligent real time control has been reported from several experiments. In particular, TCV has demonstrated that electron cyclotron resonance heating (ECRH) triggers individual ELMs independently of the preceding ones. For a given trigger sequence, the resulting ELM sequences were highly reproducible (see figure 17). Localization of ECRH power was a key for high efficiency control with lower power. Failsafe NTM prevention also has been demonstrated by sawtooth pacing by a combination of pre-emptive low-power ECRH on island and backup power for NTM stabilization [94]. DIII-D has reported that real-time control of EC power and mirror steering provided complete stabilization of m/n = 2/1 NTM. Here it should be noted that motional Stark effect (MSE) detection is used for the first time, which has the potential to save required leading time and power compared with other diagnostics [95]. A real-time ECRH/electron cyclotron current drive (ECCD) system to control NTMs as well as sawtooth pacing with pulsed ECH has also been reported by FTU [96,97]. Safe application of heating power and high-performance scenario development for a metallic wall is a critical issue in the plasma scenario. Achievements of high heating power have been reported from JET [98] and ASDEX-Upgrade [99]. In JET, following the neutral beam injection (NBI) system upgrade, power levels of 25.7 MW, and pulse length of 15s (using a single injection box) were obtained with the ILW. Although all private and surrounding limiters of ICH have been changed to beryllium, 4 MW could be coupled in H-mode. Although good heating performance was observed, higher bulk radiation in the ICH plasma than NBI heated plasmas was observed due to higher tungsten and nickel levels. The source of higher tungsten concentration observed with ICRH has not yet been identified [98]. ASDEX-Upgrade has demonstrated high-performance discharges with P /R = 14 MW m −1 . Nitrogen puff to control divertor radiation has enabled heating power of 23 MW with a total radiation of 20 MW including divertor radiation of 9 MW. Eventually, the divertor heat load was reduced to less than 5 MW m −2 at the H factor of around 1 and β N = 2.8. ICRH of 4 MW is compatible with the tungsten wall [99]. In LHD, metallic impurities have been suppressed to low levels, i.e. n Fe /n e < 10 −5 , even though the first wall is made of stainless steel. Impurity screening in the stochastic layer surrounding nested flux surfaces worked to suppress the penetration of metallic impurities and this screening effect was more effective for heavier impurities due to their outer ionization front [100].
Steady-state tokamak operation requires broad current and pressure distributions to enable high β N . Current distributed off axis is favourable for steady-state, MHD stability and confinement. DIII-D used off-axis NBI to develop and test the scenario for high β N steady-state operation. An elaborate NBI facility allowing vertical steering has demonstrated development of a quasi-stationary scenario [101]. Off-axis NBI current drive (NBCD) has produced broad current and pressure profiles with sustained q min > 2 for a higher β N (∼4) stability limit [102] (see figure 18). It has also demonstrated a quasi-stationary scenario with performance that projects to Q = 5 in ITER with q min ∼ 1.5 and a current drive fraction due to NBI of 70% for twice the resistive diffusion time. Off-axis Figure 18. Off-axis NBI produces broad current and pressure profiles with sustained q min > 2 for higher β N stability limits on DIII-D [102].
NBCD in MAST has indicated that the fast-ion distribution is close to the classical picture [108].
Novel approaches of ICRF have also been reported from LHD [105] and Alcator C-Mod [109,110]. LHD has reported that the use of dipole phasing gave better heating performance than monopole phasing as predicted by the theory [105]. Alcator C-mode has reported observation of ICRF-induced radial electric field in the scrape-off layer (SOL), which can play a role in the increase of impurity source and penetration [110], and the successful result from field aligned ICRF antenna to reduce impurity contamination [109]. These observations have suggested the importance of the control of the RF field to suppress impurity contamination and to enhance heating efficiency. In this regard, NSTX has reported interesting evidence that RF edge losses suggest fast-wave propagation in SOL and its modelling which was meant to predict and minimize such edge losses in the ICRF heating regime [111].
The experiment of off-axis NBI in MAST showed that confinement of fast ions was close to classical and that off-axis NBI broadened fast-ion pressure profile, eventually mitigating the growth of fishbone modes and the redistribution of fast ions [108]. In the MST RFP, the analysis of the d-d neutron decay rate due to a tangential NBI with 25 keV has suggested that fast-ion confinement is close to classical even in the standard RFP with a stochastic magnetic field [112]. Also several bursty instabilities driven by energetic ions have been observed for the first time in an RFP [112].
The dynamics of EP-driven modes in wall-stabilized high-beta plasmas has been investigated by joint experiments of JT-60U and DIII-D. ELM triggering by EP-driven modes has been identified for the first time. ELM triggering by EP-driven modes occurred with strong distortion (n = 1) amplitude, and EP transport to the edge by EP-driven modes affected edge ELM stability [113] (see figure 19). Energetic-ion-driven MHD instabilities in stellarator/heliotron plasmas with low magnetic shear have been studied in joint experiments of Heliotron-J (low rotational transform) and TJ-II (high rotational transform) with different degree of rotational transforms. The scan of rotational transform has indicated that global Alfvén eigenmodes (GAEs) are mainly destabilized in low shear helical plasma and helicity AEs (HAEs) were destabilized in high rotational transform configuration [114]. Heliotron-J has also reported that EP-driven modes were suppressed when the magnetic shear exceeds the threshold for ECCD in the counter direction [115]. Tilting the NBI in DIII-D also allowed the variation of the AEs drive and tests of stability models [116].
HL-2A has reported multi-mode co-existence of an electron-beta-induced AE (BAE). A new mode is excited in the presence of both strong tearing modes and BAEs with a geodesic acoustic mode (GAM) in ohmic plasmas. Since the GAM was localized in the core, it was different from one excited by the drift-wave turbulence in the edge plasma [117]. LHD has reported an interesting observation of energy transfer between ions and the GAM. During the burst of the GAM, the energetic ions lost energy, and the ions in the lower energy range gained the energy (see figure 20). This observation suggested that the GAM can become an efficient energy channel from EPs to bulk ions [118].
Two new approaches using wave heating schemes have been proposed to control the edge plasma. One is an edge magnetic topology change induced by LHCD in EAST. Helical radiation belts led to the splitting of divertor strike points with effects similar to those with an RMP [131]. Another was ICRF in NSTX. Optimized higher harmonic fast waves can drive edge harmonic oscillations with larger displacements [132]. It is expected that the effect of both approaches on the control of ELMs will be assessed in the near future.

Concluding remarks
Compared with the achievements reported at the last FEC2010, the understanding of experimental results has progressed remarkably, in particular, in RMP and ILW, which are the Figure 20. Energy distribution function measured by the neutral particle analyser (NPA) during the GAM bursts in LHD [118]. Variations in the energy distribution function from the spectrum in the range between 60 and 175 keV. The energy of the NBI is 175 keV. The burst starts at 4.425 s. highlight of this FEC2012. It should be noted that PWI used to be out of the scope of this series of conference, but now it has become a super critical topic. Together with the recognition of remarkable progress, it was confirmed again that we still need tenacious efforts to get a more accurate and comprehensive knowledge to extrapolate to ITER and DEMO. Typical remaining issues are the significance of 'resonance' in ELM suppression/mitigation by RMP, effect of metal walls on the performance of plasma confinement, and documentation of disruptions.
The meaning of 'ITER relevant' should be assessed more seriously than before. Identification of correlation is not enough and only identification of the bridging mechanism between the actuator and the consequence, in other words, identification of causality could provide the most reliable extrapolation to ITER/DEMO. The fusion community has to prepare resolutions to these issues in time to lead ITER to success. It should be emphasized that no single experiment can resolve these issues. It is quite important to rally all kindred experiments to solve the problem.