In fusion, ELMs are not trees but edge-localized modes. The name itself suggests that they are not understood, not unlike the term assigned to the Irritable Bowel Syndrome. The name has even spawned an adjective, ELMy, and a participle, ELMing, which should give philologists conniptions. ELMs occur at the pedestal in H-mode plasmas (Chap. 7). Recall that in this high-confinement mode, a transport barrier, shown earlier in Fig. 7.25, is formed at the edge of the plasma.

This thin layer holds back the plasma because it quenches all instabilities with strong electric field shear. But it can’t do that forever. If the plasma escaped at the classical diffusion rate due to collisions alone, the plasma pressure in the interior would rise so high that the barrier would break down. This breakdown occurs in short bursts, called ELMs, so that there is a steady release of plasma to the outside. Actually, this is a good thing because the “ash” of the DT reaction has to be taken out. This ash is the cleanest ash ever — pure helium — but it has to be removed because otherwise the expensive magnetic field would be used up in confining the ash rather than the fuel.

The H-mode occurs only when the heating power exceeds a certain threshold value. ELMs occur when the power is just above this threshold and are really localized near the plasma edge. Recall that the “edge” of the plasma is defined by the divertor, like the one at the bottom of Fig. 8.13. The plasma edge is defined by the last closed magnetic surface, the one at the X made by the field lines just above the divertor. Plasma venturing beyond that is led into the divertor, where it strikes high- temperature materials with heroic cooling to dissipate the heat. Also shown in the figure is the layer where the H-mode barrier exists and, inside that, the core plasma. The problem with ELMs is that the heat comes in short bursts — less than 1 ms — occurring a few times a second, and divertors cannot handle a heat flow that is not steady. A single ELM, while it lasts, can carry 20 GW of power, an energy flow

Blanket and

first wall

Region I

Core plasma

Region ll

Plasma edge and

H-mode

confinement barrier

Region ill

Scrape-off layer

Divertor plasma

Divertor chamber

Fig. 8.13 Cross-section of a tokamak with a single-null divertor, showing the scrape-off layer [16]
comparable to that of the Three Gorges Dam in China [17]. There are thus three tasks: measuring what ELMs do, explaining what causes them, and devising a way to suppress them.

It’s hard to measure what goes on inside the thin barrier layer during the unpre­dictable time when a burst occurs, but there is a large data base on the different types of ELMs and the conditions before and after they occur [18]. Three types of ELMs have been observed. As the heating power is increased past the H-mode threshold, Type 3 ELMs first occur. These occur rapidly, each with a small energy release. They come after a magnetic precursor signal can be detected. As the power is raised, the ELM frequency decreases until there are no ELMs at all. Then Type 2 ELMs, called “grassy” ELMs, occur; they are very small, rapid bursts whose time traces resemble grass. Further increase in power produces Type 1 ELMs. These occur in most H-mode tokamaks and release energy in rather regular bursts. Each pulse occurs when the density and temperature at the top of the pedestal reach critical values, and these drop when an ELM occurs. Density and temperature then recover slowly until the next burst is triggered. Although ELM-free discharges can be produced, they cause the temperature and density at the top of the pedestal to be rather low, and these control the quality of the fusion plasma in the main volume. It is found that the best fusion conditions can be produced by ELMy H-mode plasmas, in which the plasma is allowed to escape in regular Type 1 ELMs.

Many theorists [19] have worked on the ELM problem, and the consensus is that ELMs are a magnetic instability called a “peeling-ballooning” instability. Computations can predict the temperature and density values in the pedestal that can trigger an ELM, but they are far from explaining all the features that have been observed. And, as usual, there is no guarantee that another theory can’t also explain the ELM threshold. There is, however, good news. The DIII-D team at General Atomics have figured out a way to suppress ELMs without degrading the quality of the core plasma [20]. They apply “resonant magnetic perturbations” with an array of small coils just outside the plasma edge. These produce small magnetic islands in the edge region which work some kind of magic. Experimental results are promising enough that such coils are being considered and designed to be added to ITER.5