Towards a self-aiming microwave antenna to stabilise fusion plasmas

Over the past decades research into controlled nuclear fusion as a source of energy has ma- tured to a level where engineering aspects of a reactor are being addressed. The principal reactor design currently is a tokamak, a torus shaped vessel in which the fusion fuel - a hot plasma of hydrogen isotopes - is contained by magnetic fields. The goal of ITER, the large international fusion reactor presently under construction in France, is to generate 500 MW of fusion power, with a tenfold power gain. It has been known for a long time that the plasma confined by magnetic fields in a tokamak can develop structures, known as "tearing modes" or "magnetic islands". These deterio- rate the quality of the confinement of the plasma. This is sometimes wanted - for instance to °ush the ash out of the fusion reactor - but mostly undesirable. For fusion reactors, con- trol of tearing modes is therefore of great importance as a tool to optimize the performance. Tearing mode control requires two ingredients: detection and actuation. The actuator of choice is Electron Cyclotron Resonance Heating (ECRH), which allows precise localisation of power by the resonant absorption of a focused beam of high-power millimetre waves. The beam must be steered to the heart of the magnetic island with a steerable mirror, which is established technology. This is also the technology that will be applied in ITER. The problem addressed in this thesis is the detection and localisation of the island. In principle, any method that can detect and accurately localise a magnetic island could be suited for the purpose. There are, however, severe complicating factors. The most important is the fact that the plasma is a dispersive medium, in which the path of the ECRH beam is curved and depends on the local gradients of pressure and density. To convert a measurement of the position of a magnetic island at one place in the tokamak to the correct launching angles and phasing of the ECRH installation requires accurate measurements of a number of plasma parameters and major computational processing. For this reason, this method of steering the ECRH beam is susceptible to error. In this thesis a detection system is described that radically deals with the problem of abso- lute island localisation - by avoiding it. A receiver is developed that detects the island via the same sight line as used by the ECRH. In this scheme the island is detected by looking for the characteristic °uctuations on the Electron Cyclotron Emission (ECE) caused by an island. In the receiver, the ECE channels are selected around the ECRH frequency. In this way it is possible to steer the ECRH beam using the ECE signals as input, without having to localise the island in absolute coordinates. The minimum island size to which the ECRH beam can be steered only depends on the sensitivity of the detector (i.e., the minimum °uctuations the line-of- sight receiver can still detect) and the degree to which the launcher can be fine-tuned. There is a price to pay for the elegance of the scheme and that is the huge difference in power between the actuator (ECRH) and the sensor (ECE) sharing the transmission line. The ECRH power is of the order of 1 MW, while the ECE power in the transmission line is of the order of 1 nW, i.e. a di®erence of 15 orders of magnitude! First, there is the obvious problem of ECRH stray radiation coupling into - and destroying - the receiver. Secondly, even if a way is found to suppress the ECRH stray radiation at the gyrotron frequency, the gyrotron may produce spurious modes that cause problems; Thirdly, the element that separates the ECE from the ECRH must be capable of passing the 1 MW forward power. And finally, there is the possibility that the application of ECRH leads to emission from the plasma at frequencies shifted with respect to the ECRH, thereby disturbing the ECE spectrum. In this thesis, an in-sightline ECE radiometer was developed and applied at the TEXTOR tokamak at the Forschungszentrum JÄulich. In TEXTOR, with large radius 1.75 m and small radius 0.46 m, islands can be generated by external coils and reach a size of some 10 cm. The islands in general rotate in the plasma with a typical frequency of several kHz. The ECRH system uses a gyrotron with a maximum power of 850 kW at 140 GHz, with a 10 s pulse duration. The challenge of separating the MW ECRH power from the nW ECE signal is taken on by developing a Fabry-Perot type filter based on a resonant dielectric plate. This plate acts as a Frequency Selective Coupling element and is placed in the ECRH transmission line. In the final construction two of such dielectric plates are used, giving 45 dB suppression of the ECRH component with respect to the ECE power. The beam containing the ECE spectrum is coupled to a 80 dB notch filter in waveguide. The ECRH power that travels in the same direction as the ECE signal is estimated to be several hundreds of Watts at maximum. The total reduction of 125 dB reduces this to tens of pW, which is well below the ECE power levels. The notch filter at the gyrotron frequency has a spectral width of 100 MHz, which under normal operating conditions is wide enough not to cause problems due to drift of the gyrotron frequency. After the notch filter the signal is fed to a 6-channel heterodyne radiometer. The channel spacing of this radiometer is 3 GHz, the channel width is 500 MHz, the video bandwidth is 10 kHz, and the in-vessel receiver noise temperature is 100 eV. With this instrument, °uctuations caused by a 1 cm rotating island can still be detected. The limit on the ECE channel spacing is imposed by the dielectric losses in plate, as the channel spacing decreases with increasing plate thickness. At a channel spacing of 3 GHz, the ECRH transmission losses are 5%. The temperature rise of the plate limits operation to 3 s at 800 kW. In the final instrument specification the 3 GHz ECE bandwidth and the 3 s ECRH pulse length has been used. This instrument has been designed, built and tested at TEXTOR, and was found to per- form according to its design values. In the absence of ECRH power, it was shown that the in-vessel receiver noise temperature, noticeably due to the high-sensitivity radiometer, does not limit the minimum island size that can be detected. The ECRH stray radiation was measured to be a factor 10 below the ECE levels, which ensures that ECRH stray radiation cannot lead to a non-linear response of the instrument at the ECE frequencies. In the presence of 400 kW ECRH power, rotating islands have been detected and local- ized. Fluctuations in electron temperature have been detected below 10 eV, while the requirement for a 1 cm island is 20 eV. In short, the instrument performs according to its design values and in particular has given proof-of-principle of the in-sightline radiometer concept, with 15 orders of magnitude separation of forward and backward power. In some plasma conditions, in particular in the presence of magnetic islands, strong per- turbations are observed on the ECE channels: bursts of emission with a typical duration of several 100 ¹s. A closer inspection shows that these are caused by powerful frequency components that are shifted several 100 MHz, up to a few GHz, from the ECRH frequency. It is also noted that in the case of rotating islands the perturbations are correlated to the island rotation period. The correlation with a particular phase in the island supports the idea of a conversion process in the plasma with the high power ECRH beam. These anomalous bursts of ECE emission, however interesting from a plasma physics point of view, perturb the signals that are used for the island detection. However, a method was devised that filters the bursts from the signal, thus allowing the island localisation also in the presence of spurious ECE emission. With that, the proof-of-principle of the in-sightline ECE radiometer as the basis for a feedback control of the ECRH island suppression system is given: Towards a self-aiming microwave antenna to stabilise fusion plasmas! The next step is to use the line-of-sight ECE data to control the ECRH launcher and ECRH power. This work is outside the scope of this thesis, but considerable progress towards this goal has already been made in a follow-up project. In parallel to this, work has started on an in-sightline scheme with an in-waveguide coupling element that can handle continuous ECRH. This will open the way to application in ITER, and is presently being developed for application at the Asdex Upgrade tokamak at IPP-Garching.