Neutral particle behavior in divertor simulator MAP-II

cpp header will be provided by the publisher Neutral particle behavior in divertor simulator MAP-II H. Matsuura 1, S. Kado 2, and Y. Kuwahara 3 1 Graduate School of Engineering, Osaka Prefecture University, Sakai Naka-ku, 599-8531 Osaka, Japan 2 High Temperature Plasma center, The University of Tokyo, Kashiwa, 277-8568 Chiba, Japan 3 School of Engineering, The University of Tokyo, Bunkyo-Ku, 113-8656 Tokyo, Japan Received 22 May 27, revised 15 September 25, accepted 2 December 23 Published online 3 December 23 Key words MAP-II, DEGAS, H-alpha spectra, neutral recycling, detached plasma. PACS 52.55.Rk;52.65.Pp Neutral particle behavior in the linear divertor simulator material and plasma(map)-ii is studied with Monte Calro simulation code DEGAS. Various model cases are simulated to compare the results with the attached and detached plasma condition. Neutral particle distribution depends upon the location of particle sources/sinks. H alpha spectra shape depends also plasma temperature and density and there exist a dominant pathway to produce exited H atoms. 1 Introduction Development of heat flux reduction method on divertor plate is an important issue for the future fusion reactor design. One of most promising scenario is to establish so-called detached plasma, whose electron temperature is kept below 1[eV] by powerful gas puffing. Linear divertor simulator material and plasma(map)-ii in the University of Tokyo has two independent vacuum chamber ( plasma source chamber and target chamber ). Detached plasma is simulated by increasing back ground gas pressure of the target chamber.[1] In this situation, neutral particles recycled from the target plate and those provided from gas puffing are expected to show different behavior, since their species and initial energy are different. In this paper, these behavior are studied with Monte Carlo simulation code DEGAS.[3] In [4], Monte Carlo simulation results of old MAP device with DEGAS2 code was already reported. But, neutral transport was simulated only in plasma column and large vacuum region was treated as uniform neutral gas source. Axial distribution of plasma parameters was also assumed to be uniform. So this study was onedimensional. In this paper, we construct 2D simulation model geometry including vacuum region in order to study particle balance of neutrals. Axial distribution of plasma parameters is also considered in detached condition. In section 2, we briefly explain MAP-II device and its operation to simulate detached plasma. Two dimensional model geometry is construct to compare with experimental results. In section 3, neutral hydrogen distributions are studied with DEGAS ver.63 code, which has already been applied even to three dimension model geometry. In MAP-II plasma, dissociation of hydrogen molecules desorbed from target plate or puffed from the wall plays important rolls in H alpha emission. In section 4, the contribution of various pathways to produce exited hydrogen atoms are compared. 2 MAP-II device and simulation model MAP II is a dual-chamber linear divertor simulator.[1] Steady state plasma is produced by PIG arc discharge, passed through source chamber and terminated at target plate in target chamber. By turning off the differential pumping of the source chamber, electron temperature at target chamber entrance decrease. ( (For example, Corresponding author: e-mail: matsu@me.osakafu-u.ac.jp, Phone: +81 72 254 9226, Fax: +81 72 254 994

4 H. Matsuura, S. Kado, and Y. Kuwahara: Neutral particle behavior in divertor simulator MAP-II Table 1 Simulation cases. n e and T e is plasma density and temperature at the center of plasma column. In detach case, n e decreases to 1 1 [cm 3 ] as approaching to plate. CASE particle source n e [cm 3 ] T e [ev] REF1 plate(arb.) 2.5 1 11 15 attach with pumping DET1 plate(arb.) 1.2 1 12 3.5 detach DET2 vol.recomb. 1.2 1 12 3.5 detach WALL1 plate+wall(1%) 1.2 1 12 3.5 detach WALL1 plate+wall(1%) 1.2 1 12 3.5 detach WALL1 plate+wall(1%) 1.2 1 12 3.5 detach WALL plate 1.2 1 12 3.5 detach ATT1 plate 1.2 1 12 3.5 attach Fig. 1 Schematic view of the model geometry of MAP-II target chamber. 15 3 [ev]. ) Then plasma entering target plate becomes detached gradually with the increase of target chamber neutral pressure. (For example, 1 1 [mtorr]. ) Neutral pressure in the target chamber is controlled with gas puff intensity from the chamber wall and the pumping speed. Radial (r) profiles of plasma parameter are given by the double probe measurement near target plate. [2] In detached plasma, the recombination front is formed in z direction and plasma parameters change there. This front moves with chamber gas pressure. [5] Axial (z) profile of plasma parameter can be deduced from the data in various pressure condition. Fig.1 shows the simulation model for this work. The target chamber of MAP-II is modeled with a simple cylinder of 5 [cm] diameter and 5 [cm] length, which is divided to 3 mash radially and 2 mesh axially. Since radius plasma column is very small (3 5[cm]) compared with that of the chamber (25[cm]), radial separation of mesh varies with radial position. Particle sinks are a plasma inlet from the source chamber and the pumping duct. If test particles reach these sinks, they are counted as being lost from the system. Various particle sources such as recycling at target plate, volume recombination in the chamber, gas puffing at the wall, and so on are considered and their contributions are compared. Simulation cases concerned to this work are summarized in Table.1. REF1 corresponds to the previous work simulation.[4] In this case, two chambers pressure of MAP II are kept low with strong pumping. Electron temperature is high and plasma is attached. In other cases, wall pumping ( see. Fig.1 ) is closed and lower

cpp header will be provided by the publisher 5 electron temperature and high density are assumed. Neutral particles are provides as hydrogen molecules except for DET2. Source strength at target plate in WALL1 ATT1 are estimated from plasma influx with the value of n e and T e, which is also same value of gas puff source intensity in WALL1. Source strength of wall gas puff in WALL1 is 1 times larger than WALL1. In attached case ( REF1 and ATT1 ), electron density and temperature is assumed to be constant along z-direction. In other cases, however, electron density at each r position decays with approaching to the plate ( z direction ). As for electron temperature, its full width of half maximum (FWHM) is assumed to become small near the plate with keeping its peak value constant. 3 Neutral particle distribution Average neutral density[cm 3 ] 1 16 3[mtorr] H 2 1[mtorr] 1 14 H 1 12 1 19 1 2 1 21 1 22 Total source intensity[s 1 ] Fig. 2 Vertical axis is neutral density averaged over whole simulation geometry.( Triangles are Hydrogen molecular density and circles are atomic density.) Horizontal axis is the total intensity of neutral sources. Fig.2 shows the almost linear relation between source intensity and average neutral density. For case ATT1, where only the recycling at plate is considered as neutral source, hydrogen molecular density is 1 14 [cm 3 ]. This value corresponds to neutral gas pressure of 3[mtorr], which is measured in MAP II for the attached plasma without the gas-puffing. On the other hand, typical pressure in detached plasma is 1 1[mtorr]. So this condition could be simulated as WALL1 ( Source intensity is 1 22 [s 1 ]) or WALL1 ( 1 21 [s 1 ]). If only volume recombiantion is considered as neutral source, total intensity is negligibly small. (1 13 [s 1 ]) Since the calculation of volume recombiantion in DEGAS does not contain molecular assisted recombination ( MAR ) effect, volume recombiantion intensity could be 1 3 times larger. As reported in [6], recombination rate constant for MAR is about 1 3 times larger than electron ion recombination (EIR) for T e = 3[eV]. Nevertheless, as shown in Fig.2, this recombination source is not so important in neutral particle balance. It is also shown that neutral gas influx of 1 1 times of plasma particle influx is necessary to reach the detached state. Since electron density of MAP-II plasma is smaller than real divertor plasma, low energy hydrogen molecules puffed from chamber wall can enter plasma deeply. About 1% molecules are lost with ionization and about 1% dissociate to hydrogen atoms and reach to the sinks. On the other hand, hydrogen atoms recycled at target plate easily escape from plasma, form hydrogen molecules, and are lost at sinks. Fig.3 compares the molecular hydrogen density profile for different neutral source intensity. Horizontal axis is radial position (r = 25[cm]) and vertical axis is axial position (z = 5[cm]). Though large vacuum region was treated as uniform neutral gas source in Tanaka s paper [4], hydrogen molecular profile depends upon the location of particle source( plate ) and sink ( pumping duct ). In WALL1 case ( Fig.3a) ), neutral density has very localized peak ( n H2 5 1 14 [cm 3 ] ) near target plate, since the radius of plasma column is smaller than the chamber and recycling at plate is very localized. As there is a particle sink ( a duct connected to source chamber ) at the left corner of the top boundary, neutral density shows the minimum value there. In the case with active pumping like REF1 and [4], the pump duct acts as another sink and neutral density profile shows another dip. As the contribution of wall gas puffing increases, neutral density increases ( n H2 2 1 16 [cm 3 ] ) and

6 H. Matsuura, S. Kado, and Y. Kuwahara: Neutral particle behavior in divertor simulator MAP-II plasma becomes detached. Since gas puff is assumed to occur at whole wall in this work, neutral distribution becomes more homogeneous. Localized recycling contribution at plate becomes negligible. Even in this case, particle sink effect is still clear and neutral density at plasma boundary or inside plasma column is not constant along z direction. ( Fig.3b) ) This indicate that, in order to compare H alpha spectrum obtained by experiments and simulation, the precise orientation of detectors must be considered in simulation models. H_molcular_density (MAX=5.E14[1/cc]) H_molcular_density (MAX=3.E16[1/cc]). E 5. E 14 5[cm]. E 3. E 16 5[cm] 25[cm] a) 25[cm] b) Fig. 3 Hydrogen molecular density. a) is for case WALL1 and b) is for case WALL1. 4 H alpha spectrum H alpha line results from radiative decay of exited hydrogen atoms from the principal quantum state n = 3 to n = 2, and it is widely used as a monitoring tool of neutral hydrogen behavior. The line intensity is the measure of density of H(n = 3) and its wavelength spectra reflects the velocity distribution of hydrogen atoms. As there is many pathways to generate H(n = 3), there is many groups of H(n = 3) with different density and characteristic energy. DEGAS ver.63 considers 1 pathways. In [8], contribution of charge exchange, sputtering, reflection, and dissociation to H alpha spectra in TFTR was compared. In torus edge plasma, electron excitation of hydrogen atom entering into main plasma plays important rolls.[7] In MAP-II case, however, plasma density is lower and H alpha detector does not see the target plate directly. So dissociation of hydrogen molecules desorbed from target plate or puffed from the wall plays important rolls. Figure 4a) shows three dominant contribution in H alpha spectra calculated with DEGAS for case REF1 (detached). Narrow peak written as H2DE is the contribution of H(n = 3) directly produced in dissociation of hydrogen molecules (e + H 2 H(n = 3) + H + + 2e). Two broad profiles come from the contribution of hydrogen molecular ions. In DEGAS simulation, hydrogen molecular ions are produced by electron impact ionization of hydrogen molecules and assumed to be lost immediately by H2+DE( e + H 2 + 2H(n = 3) ) or H2+DI( e + H 2 + H(n = 3) + H+ + e ). In order to improve statistical accuracy of these H 2 + contribution, more test particle flights is necessary in the simulation. In ATT1 case (Fig.4b)), where plasma is still attached, the contribution of H2+DI disappears. Electron temperature in ATT1 is lower than REF1, so high energy electron which can react as H2+DI seems to decrease. On the other hand, electron density becomes larger and more H(n = 1) produced in H2DS(e + H 2 2H + e)

cpp header will be provided by the publisher 7 could be excited to n = 3 with electron impact excitation. In DEGAS, this excitation probability is calculated with collisional-radiative(cr) model and contribution of hydrogen atoms produced in ground state to H alpha emission spectra is also considered. In previous work with DEGAS2 [4, 9], it is pointed that the contribution to H alpha spectra of H(n = 2) becomes important at least for electron temperature of 1[eV], since it can be produced more easily than H(n = 3) and excited easily than H(n = 1). Unfortunately, present DEGAS does not consider this effect. So comparison with DEGAS2 simulation results will be interesting, which is left for future work. Comparison with spectroscopic measurement is also now under way. 4 [x1 14 ] Intensity[arb.] 2 H alpha specrum(ref1) H2DE 8 [x1 14 ] Intensity[arb.] 6 4 H alpha specrum(att1) H2DE H2+DE H2+DE H2+DI 2 H2DS 6562 6563 6564 Wave length[1 1 m] 6562 6563 6564 Wave length[1 1 m] Fig. 4 H alpha emission spectrum. a) is for case REF1 and b) is for case ATT1. a) b) 5 Conclusion Two dimensional mesh model geometry is constructed to study neutral particle behavior in divertor simulator MAP II. The results of this study are the following. Volume recombination source is not so important in neutral particle balance. Neutral gas influx of 1 1 times of plasma particle influx is necessary to reach the detached state. Dissociation of hydrogen molecules desorbed from target plate or puffed from the wall plays important rolls in H alpha spectra in MAP-II. H alpha spectra profile depends also plasma temperature and density and there exist a dominant pathway to produce exited H atoms. Acknowledgements One of Authors (H.M.) acknowledges Dr. M.Shoji of National Institute for Fusion Science and Dr. Y.Nakashima of Tsukuba University for their advice on DEGAS simulation. References [1] S. Kado et al.: J. Plasma Fusion Res. 81, 81(25). [2] S. Kado et al.: J.Nucl.Mater., 337-339, 166(25) [3] D. Heifetz et al.: J. Comp. Phys. 46, 39(1982). [4] S.Tanaka et al.: Plasma Phys.Control. Fusion, 42, 191(2). [5] F.Scotti et al.: Plasma and Fusion Research, 1, 54(26). [6] S.I.Krasheninikov er al.: Phys. Plasma, 4, 1638(1997). [7] H.Matsuura et al.: J. Plasma Fusion Res. SERIES7, 16(26). [8] D.P.Stotler et al.: Phys. Plasma, 3, 484(1996). [9] B.Xiao et al.: J.Nucl.Mater., 29-293, 793(21). [1] Y.Nakashima et al.: J.Nucl.Mater., 196-198, 493(1992).