APS DPP 2005 – Denver, CO Impurity Source Studies in TCS J.A. Grossnickle, H.Y. Guo, G.C. Vlases Redmond Plasma Physics Laboratory, University of Washington Abstract The original TCS experiment demonstrated the robust ability to form and sustain FRCs in steady-state using Rotating Magnetic Fields (RMF). RMF parameters set the plasma density, but the temperature was severely limited by radiation, which was determined to be the dominant source of power loss for these plasmas. The total radiated power was strongly correlated with the Oxygen line radiation. This suggests Oxygen is the dominant radiating species. Determining the source of the impurities is an important question that must be answered for the TCS upgrade. Indications are that the primary sources of Oxygen are the stainless steel end cones. A Ti gettering system was installed, however, removal of Oxygen was accompanied by an increase in neutral Hydrogen, which also severely limited performance. These findings factored heavily into the design of the vacuum system and cleaning techniques for the TCS upgrade. The DIVIMP impurity code was modified to run on FRC geometries and showed that core impurity contamination is strongly dependent on electron density and radial transport. Thus, impurity source control is more crucial in TCS than high-density theta pinch formed FRCs. Power Balance Radiated power vs. RMF (input) power is shown at right. Radiated power is subject to a bolometer calibration factor. Radiation losses are a significant fraction of the input power, typically between 0.6 – 1.0. Pre-Ti Gettering Impurity Studies Oxygen III and Silicon III line intensity show a strong correlation with total radiated power. Carbon III line intensity shows a weak, if any, correlation with total radiated power. Since there is relatively little Silicon, this indicates Oxygen is the primary radiating impurity. Additionally, energy confinement time dropped with increasing Oxygen radiation. Doping with 25% CO2 (red) increased the CIII signal over the non-doped plasma (black), but did nothing to the OIII signal, implying there is much more Oxygen than Carbon. Impurity Sources The TCS main chamber first wall is primarily quartz. The end cones on both sides are stainless steel. The North side (right in the figure above) has a Tantalum shield covering approximately half of the cone, and joins a 1 meter long stainless end chamber. The South side cone joins the quartz acceleration section. The general conclusion is that the bulk of the impurities (Oxygen, Carbon, and Silicon) are coming from the stainless steel end-cones and plug section (Oxygen and Carbon), and the quartz acceleration section (Silicon), where most of the scrape-off layer field lines terminate, not from the quartz wall of the main TCS chamber. The Titanium gettering campaign and DIVIMP simulations bear out this conclusion. Iron from the end cones may also be present and would greatly increase impurity radiation. Plot of field lines for normal operation, using data from 28 excluded flux arrays. Note on the South end, where the quartz acceleration section is, all the SOL field lines exit through the cone or the quartz. On the North end most of the SOL field lines exit far down the SS end chamber. 2.5 m 40 cm 27 cm LSX/mod 80 cm TCS Schematic of the TCS device. End Coils Mirror Coils Plug Coil Cone Coils Tantalum Shield Acceleration Section

APS DPP 2005 – Denver, CO Ti-Gettering Experiments The principal objective of the TCS upgrade, now under design, is to clean up the machine (reduce impurity radiation, reduce resistivity, increase temperature, etc.). To this end Titanium gettering experiments were carried out on TCS. There were several objectives in these experiments. First, we wanted to gain experience with Ti gettering in general. We needed to know what Ti would do in our system, i.e. how would it be sputtered, would it be redeposited and where, how much do we need for effective impurity control, etc. Some of these issues, particularly redeposition, have affected our design of the upgrade. Second, we wanted to know where the impurities were coming from. There was general consensus that the bulk of the impurities in TCS (Oxygen, Carbon, and Silicon) come from the end-cones, acceleration section, and plug section, where most of the scrape-off layer field lines terminate. Covering these areas with Ti could give us an idea if this analysis was correct, and also shed light on to how “dirty” a quartz first wall in the main chamber would be. Last, there was some indication that Oxygen was the majority impurity and was clamping the plasma temperature. Since Ti gettering is particularly good at Oxygen control, it could help shed light on this issue. P rad decreases very slightly over the campaign. Phase I: pre-Ti gettering reference shots. Phase II: 8 hours H 2 GDC followed by 2 hours He GDC. Ti gettered South cone and acceleration tube only. Phase III: He GDC over several days, no Ti gettering. Phase IV: Several days He GDC. Ti gettered both cones. Phase V: He GDC overnight. Ti gettered both cones, then 2 hours of He GDC. Phase VI: Ti gettered both cones then He GDC overnight. Phase VII: shots , Ti gettered north cone only. At no time was Ti applied to the main quartz chamber. Oxygen results - In the shots immediately following the application of Titanium the Oxygen concentration was significantly reduced. As more and more shots were fired, however, the Oxygen levels returned to near pre- gettering levels for both the 105 and 74 kHz experiments. One interesting result was that even though Oxygen concentration was reduced, FRC performance was poor following application of Ti, and performance increased as Oxygen returned to reference pulse levels. This was due to increased D  and Carbon. Decreases in radiated power from Oxygen were more than compensated for by increases in power lost from Carbon and D , which actually led to decreased performance. Silicon results - Ti gettering on a small section of the North end of the acceleration section (approximately 10 cm) nearly eliminated the Silicon. It is reasonable to conclude from this that the majority of Silicon impurity in the plasma was coming from the quartz in the end of the acceleration section where field lines were terminating on the wall and not from the quartz wall in the main TCS chamber. Silicon doping experiments were run with a 1% Silene – 99% Deuterium mixture. These experiments confirmed that Silicon is dramatically reduced, and that the majority of Silicon impurity is coming from the acceleration section not the main TCS chamber. Silicon levels remained low for the remainder of the lifetime of TCS indicating once again that covering the small area of quartz where the field lines terminated eliminated the Silicon impurity. Typical plasma parameters for pre-Ti gettering reference (black), just after Ti is applied (red), and after many shots (green). Several hours of He GDC accomplishes similar results to firing many shots. Oxygen concentration is reduced after applying Ti. Carbon results - The process of applying the Ti caused a marked increase in Carbon III line radiation. Levels were approximately a factor of 2, or more, higher than reference levels. Carbon levels gradually decreased over many shots. Glow discharge cleaning also reduced this significantly. The increased Carbon, and D , is likely due to heating of the O-rings, O-ring grease, and/or the stainless steel during the deposition of Ti. D  results - High D  leads to poor performance due to increased losses from ionization and charge exchange. Losses from charge exchange and ionization more than make up for the decrease in radiated power after Ti. D  was substantially higher after Ti gettering impacting performance for many shots until it was reduced. D  did drop back to reference levels much faster than Carbon or Oxygen.

APS DPP 2005 – Denver, CO Summary Oxygen is believed to be the primary radiator in TCS. Impurities are coming from the quartz acceleration section and stainless steel end cones where open field lines terminate. This was confirmed during the Ti gettering experiments. Silicon impurity was all but eliminated by application of Titanium on a short section of the quartz acceleration tube. Ti gettering eliminated Silicon and reduced Oxygen concentration, but increased Carbon and D  emission which limited performance. As Oxygen concentration rose and Carbon and D  dropped, performance increased to pre-Ti levels. It is believed that heating the stainless during the deposition of Ti liberated C and H from the stainless surfaces. When only the North cone was coated with Ti, increases in C were much smaller, most likely due to the Tantalum protecting the stainless cone. At the same time, O did not decrease since few field lines terminated on the Ti coated portion of the cone. The plasma was found to be controlled by wall recycling after 0.5ms for all cases. Even for cases when the fill gas was 100% He, Ar, or Ne, the plasma was nearly all D after 0.5ms. DIVIMP predicts larger impurity leakage rates than in tokamaks or theta pinch FRCs due to worse transport in TCS FRCs. Thus, impurity control is critical. DIVIMP  DIVIMP is a widely used Tokamak Monte Carlo impurity transport code []. A 2D background plasma for DIVIMP is either generated using ‘onion- skin’ models, based on the measurements of Langmuir probes at the divertor target plates to define boundary conditions, or taken from 2D solutions of a fluid code such as UEDGE or EDGE2D. In our simulations, the background deuterium plasmas are simply specified as follows: - Temperatures are constant in the SOL: T e = T i = 15 eV; - Densities are specified by n e,sep =1-3  m -3, and n = 2 cm. Carbon atoms are launched from different locations assuming either physical sputtering with a Thompson velocity distribution, or chemical sputtering with a cosine angular distribution and an initial energy of 1 eV, as would be expected from simple molecular breakup of hydrocarbons. Chemical yields are computed using the data from Toronto group (Haasz 1997). Cross-field transport coefficient for impurities: D  = 100 m 2 /s for RMF formed FRCs, and D  = 1.0 m 2 /s for comparison with typical SOL conditions in Tokamaks. DIVIMP was modified to run on the TCS geometry by David Elder and Dr. Peter Stangeby of the University of Toronto. To supply a grid for DIVIMP, the UEDGE mesh generator was modified to fit the FRC geometry by Dr. Marvin Rensink of LLNL, using experimentally determined flux surfaces.  In TCS DIVIMP calculations Carbon atoms are launched from various locations in TCS corresponding to the most likely sources of impurity.  A mid-plane launch corresponds to sputtering from the main quartz chamber.  A near cone launch corresponds to sputtering from the cone sections where there is no field-line penetration of the wall.  A divertor launch corresponds to sputtering from either the cone or end- chamber sections where the field lines exit through the wall.  (a) Mid-plane launch – chemical sputtering with an initial energy of 1 eV and cosine angular distribution.  (b) Near cone launch – chemical sputtering with an initial energy of 1 eV and cosine angular distribution.  (c) Divertor target launch– physical sputtering with a Thompson velocity distribution.  If the DIVIMP results for a mid-plane launch of Carbon are indicative of what happens with Silicon impurity, this gives further indication that the Silicon in TCS comes primarily from the quartz acceleration section and not the main quartz chamber.  DIVIMP predicts that poor transport in FRCs greatly enhances core leakage.  Impurity source control is even more crucial in TCS than in mainline tokamaks.  Impurity control is also different from standard theta pinch formed FRCs because n e is 100  lower. n e = m -3 T e =T i = 15 eV D  = 100 m 2 /s  Leakage rate is very poor for a midplane source, ~100% for TCS FRC conditions.  It is absolutely necessary to prevent plasma from contacting the quartz wall at the midplane.  Tantalum coated Aluminum rings (limiters) will be present in the TCS Upgrade to minimize wall contact.