| One of the interesting properties of magnetized plasmas is that at any given frequency, several different kinds of plasma waves with very different wavelengths can coexist. If the wave frequency is near the ion cyclotron frequency, then a fast, longwavelength wave launched into a non-uniform plasma can, over a short distance, couple to a short wavelength ion Bernstein wave (IBW) and an ion cyclotron wave (ICW) through a process known as mode conversion. The fast wave and the ion cyclotron wave are similar to light waves, in that their polarization is nearly perpendicular to their direction of propagation. The ion Bernstein wave is similar to a sound wave, in that its polarization is nearly parallel to its direction of propagation. Now, with the help of terascale computers and algorithms developed in our SciDAC project, we are able to solve the wave equations for plasmas with 2D spatial variations at sufficiently high resolution to study mode conversion. Figure 7 shows one component of the AORSA2D code's solution for a wave's electric field, following the propagation of fast waves launched from an antenna on the right (not shown) into the Alcator C-Mod tokamak at MIT. The large-scale structure in the left-hand panel is the fast wave. The curved blue line (seen more clearly in the magnified view to the right) is the mode conversion layer. The figure shows a conversion to one kind of short-wavelength mode, an IBW, propagating to the left near the horizontal mid-line. There is also some conversion to a completely different type of short-wavelength mode (slow ICW), however, which can be seen above and below the mid-line propagating to the right. It had previously been expected that the dominant conversion would be to an IBW wave, which would propagate only to the left of the conversion surface.
This process has been observed experimentally on the Alcator C-Mod tokamak, using an innovative new diagnostic technique called Phase Contrast Imaging (PCI). PCI measures the density fluctuations associated with mode-converted waves. The PCI measurements are integrated along a vertical chord through the plasma. These Alcator C-Mod measurements have been modeled extensively using the TORIC code, incorporating a "synthetic diagnostic" to predict the signals observed by PCI instruments. Figure 8 shows the in-phase and quadrature components of this prediction, as well as a comparison between the modulus of the line-integrated experimental density perturbation and the TORIC prediction. The results are in good agreement with respect to the spatial structure and spectrum of the wave.
The fusion program is actively developing techniques to make use of these mode-converted waves, which may be able to control the current and pressure profiles in fusion devices. Ion Bernstein waves are absorbed primarily by electrons, and are effective in driving the current. The slower ion cyclotron waves, on the other hand, are absorbed mainly by ions and should be more effective in driving plasma flow and improving confinement. Given these potential uses, our results are extremely encouraging; they increase our confidence in the capacity of our models to accurately simulate mode conversion in a tokamak. More generally, these simulations demonstrate the feasibility of solving a problem containing processes at disparate spatial scales. The techniques used by our simulations could be applied to other areas of physics, such as wave propagation in the Earth's magnetosphere.
If petascale computing resources were available to this project, it would also be possible to assess the feasibility of using mode-converted ICRF waves for plasma control in the planned ITER device. We would also be able to simulate 3D plasma confinement devices such as stellarators. |
