- Email: [email protected]
Beam divergence and power loading on the beamline components of the negative-ion based NBI system for JT-60U
Fusion Engineering and Design 54 (2001) 321 – 330 www.elsevier.com/locate/fusengdes
Beam divergence and power loading on the beamline components of the negative-ion based NBI system for JT-60U Liqun Hu a,*, M. Kuriyama b, N. Akino b, N. Ebisawa b, A. Honda b, T. Itoh b, M. Kawai b, M. Kazawa b, M. Kusaka b, K. Mogaki b, T. Ohga b, K. Ohmori b, K. Ohshima b, Y. Okumura b, H. Oohara b, F. Satoh b, H. Seki b, Y. Tanai b, Y. Toyokawa b, K. Usui b, K. Watanabe b, M. Yamaguchi b, H. Yamazaki b, T. Yamazaki b a
Institute of Plasma Physics, Chinese Academy of Science, Shushanhu Road 350, Hefei 230031, People’s Republic of China b Japan Atomic Energy Research Institute, Naka Fusion Research Establishment, Mukouyama, Naka-machi, Naka-gun, Ibaraki-ken 311 -0193, Japan Received 7 August 2000; accepted 7 September 2000
Abstract In JT-60U, a high-energy neutral beam injection (NBI) program using negative-ion based NBI (N-NBI) for non-inductive current drive and core plasma heating studies in high-density plasma has progressed. The target performance of the N-NBI is a neutral beam injection power of 10 MW for 10 s at 500 keV. A neutral beam power of 5.2 MW at 350 keV with deuterium has already been injected into JT-60U plasma. The beam divergence and power loading onto the beamline components are two important items to evaluate beam performance. The beam divergence, estimated roughly from heat loads at the beam drift duct in beam injection experiments into JT-60U plasma, was around 4 mrad for the horizontal direction and 6 mrad for the vertical direction, which were close to the design values of 5 mrad. The power loading on beamline components were reasonable as compared with the design value. © 2001 Elsevier Science B.V. All rights reserved. Keywords: JT-60U; NBI; Negative-ion; Power loading; Beam divergence; Beamline
1. Introduction
* Corresponding author. Fax: +86-551-5591310. E-mail address: [email protected] (L. Hu).
High energy neutral beam injection (NBI) has been considered to play an essential role for noninductive current drive and core plasma heating in magnetically confined thermonuclear fusion re-
0920-3796/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 0 - 3 7 9 6 ( 0 0 ) 0 0 5 4 4 - 5
322
L. Hu et al. / Fusion Engineering and Design 54 (2001) 321–330
search. Moreover, it can be used for suppressing plasma instability and improving plasma confinement by delivering the momentum to the plasma in the toroidal direction, and for controlling the spatial distribution of the plasma current. For these purposes, high power NBIs that can deliver tens of MW neutral beams at 100 keV-class energies had been developed with positive-ion based NBI (P-NBI). The large plasma confinement devices such as JT-60U, joint European torus (JET) and TFTR have already attained high performance plasma conditions with D or D–T operation by injecting tens of MW neutral beams at an energy of about 100 keV. In particular, JT-60U and JET have already reached the breakeven conditions, i.e. fusion thermal output is equal to plasma heating power. For the reactor-like fusion devices such as international thermonuclear experimental reactors (ITER), and even the present large tokamak JT60U, a higher energy (0.5 MeV for JT-60U, 1 MeV for ITER) and a long pulse neutral beam is required for effective, central noninductive current drive and plasma heating at a high plasma density. Ion beam neutralization efficiency for the conventional P-NBI system decreases sharply with increasing the ion beam energy over 100 keV as shown in Fig. 1[1]. Therefore, it is necessary to utilize a negative ion beam, which
Fig. 1. Neutralization efficiency of both negative and positive ion beams (curves show a theoretical calculation, solid symbols are experimental result).
has neutralization efficiency as high as 60% when using a gas cell neutralizer. The negative ion source used in the NBI device is required to accelerate negative ions of a few tens of amperes at higher beam energy. Research on increasing negative ion source current has been carried out at many laboratories in the world for more than 30 years. The negative ion current achieved, however, was only 100 mA until 15 years ago. At JAERI, negative ion source R&D aiming at a higher beam current, more than 10 A, and a higher beam energy, over 500 keV, has been in progress since the mid 1980’s. Concerning higher currents, an H− current of 10 A at 50 keV was achieved in 1990 using a multi-cusp ion source with Cs seeding [2]. In 1994, a negative hydrogen beam of 180 mA at 400 keV was accelerated [3]. In addition to ion source R&D, the high voltage dc power supply for ion source and beamline components has been developed [4]. A several-year design study for a 500 keV negative-ion based NBI system was also completed. Based on these results, no technical problems were anticipated for the 500 keV negative-ion based NBI (N-NBI) system [5], shown in Figs. 2 and 3, and construction was begun. The performance goals were a neutral beam injection power of 10 MW for 10 s at 500 keV. Since the first operation of the N-NBI in 1996, the negative ion beam power per source has reached 14.3 A at 380 keV (5.4 MW) with deuterium and 18.5 A at 360 keV (6.6 MW) with hydrogen, and the neutral beam power injected into JT-60 has reached 5.2 MW at 350 keV with deuterium and 4.2 MW at 360 keV with hydrogen [1,6]. The neutralization efficiency of the negative ion beam has been confirmed to be about the 60% in the beam energy range of 250–385 keV, coinciding with theoretically predicted value as shown in Fig. 1. The estimation of the beam divergence and power loading on the beamline components is very important for ameliorating negative ion beam performance to achieve higher beam power. In this report, the power loading on the beamline components, estimation of neutral beam divergence, time evolution of the beam parameters for the beam injection experiments of 1998, and some issues to limit the N-NBI system to achieve de-
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
323
Fig. 2. Plan view of the 500 keV negative-ion based NBE system in the JT-60U torus hall.
signed performance goal are discussed. The layout of the N-NBI beamline is described briefly as well.
2. Layout of the 500 keV N-NBI beamline [7,8] The schematic drawings for the 500 keV N-NBI system are shown in Figs. 2 and 3. It is composed of one beamline, two ion sources, a power supply system, a control system and an auxiliary sub-system including water cooling, helium refrigeration and auxiliary vacuum pumping systems. The negative ion source is composed of a negative ion generator, an extractor and an accelerator. The negative ion generator is a cesium-seeded multicusp type. A small amount of cesium vapor is injected into the negative ion generator through a transport pipe from an external cesium oven to substantially enhance the negative ion yield by a factor of about 2 – 5 higher than that of the pure volume production. Furthermore, it considerably
decreases the optimum operating pressure without deterioration of negative ion production. The cesium effect strongly depends on the temperature of plasma grid. The optimum temperature is experimentally confirmed to be about 250–300°C. The extractor has three grids; a plasma grid, an extraction grid and an electron suppression grid. The extractor is a multi-aperture type and its aperture size is 14 min in diameter. The extraction area is 110 × 45 cm2, which is divided into five segments. The accelerator has also three grids; the first and second acceleration grids, and a grounded grid. The beams are geometrically directed to the focal point of 24 m from the ion source with a steering mechanism. Each beamline is also focussed using an aperture displacement technique. The beamline consists of an ion source tank, a neutralizer cell, an ion dump tank and a neutral beam drift duct. The ion source tank contains a set of cryopumps for minimizing the stripping loss
324
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
of negative ions in the ion source accelerator. The tank is covered with magnetic shielding to screen the stray magnetic field from JT-60. The neutralizer tank has two 10-m long neutralizer cells corresponding to the two ion sources. The ion dump tank contains a pair of bending coils, ion dumps for D− and D+, a retractable calorimeter, a beam scraper and cryopumps. The calorimeter placed after the ion bending coils is composed of arrays of a composite material of molybdenum and copper, and it is used for measuring the neutral beam power. It also serves as the beam target for the neutral beams when doing ion source conditioning. The neutral beam drift duct is composed of an isolation gate valve, a bellows and a duct containing a series of beam scrapers.
3. Beam characteristics of 1998 operations
3.1. Estimation method of the beam di6ergence The estimation of the beam divergence [9] is very important for ameliorating the negative ion beam performance to achieve higher beam power. The procedure for estimating beam divergence is shown in Fig. 4. The beam divergence in the horizontal and vertical directions, vx and vy,
were determined by comparing the heat loads measured on the beam limiters in the drift duct with the computational results. A three-dimensional beam trajectory calculation code ‘BEMPROF’ [10] is used for calculation of beam fractions deposited on the scrapers.
3.2. Time e6olution of the beam parameters during the neutral beam injection into JT-60 The time evolutions of the neutral beam parameters and ion source performance for the injection shots in 1998 are illustrated in Figs. 5 and 6. In these figures, open symbols show operation results with two ion sources and solid symbols show results with only the upper ion source. The beam species is deuterium before October and hydrogen in October. In the beginning of September, a cooling water leak occurred at a filament feedthrough inside the arc chamber of the lower ion source during the beam pulse. One month was needed for recovery. In the beam injection experiments, the beam energy Eb was around 350 keV and the maximum value was nearly 380 keV. The beam energy was kept a moderate level, around 350 keV, because the first priority in this experimental period was to extend the beam pulse duration. After the water
Fig. 3. A side view of the N-NBI beamline for JT-60U.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
325
Fig. 4. Estimation procedure for the beam divergence.
leak, the beam energy was limited to 300 –330 keV in spite of cleaning the inside of the ion source. This voltage deterioration seems to have been caused by micro-contamination on the surface of the accelerator grids due to the water leak. The integrated injection power Einj.m, i.e. the beam injection energy, defined as the product of beam power Pinj.m and beam pulse duration T, was around 3 MJ and the maximum value was 6.7 MJ. Although the beam pulse duration was around 1 s at the beginning of 1998, the pulse length was extended gradually with the operation time, and reached 2.2 s. The extension of the beam pulse benefited from a stabilization of the arc discharge by taking a longer pre-arc discharge time before beam acceleration in the negative ion generator, which will be discussed later in Section 4. The beam divergence estimated in the beam injection experiments has been confirmed around 4 mrad in the horizontal direction (vx ), and around 6 mrad in the vertical direction (vy ). These divergences roughly agree with the design values of 5 mrad. The beam characteristics of beam energy Eb, the beam injection energy Einj.m and beam pulse duration kept increasing gradually before October, while the beam divergence kept almost constant. After the water leak in September, these parameters became bad owing to the contamination in the ion source by the water leak. The other parameters of acceleration current Iacc, extraction current Iext and arc efficiency harc
are shown in Fig. 6, where harc is defined by the ratio of acceleration current Iacc and arc power Parc. The acceleration current with two ion sources follows the same trace as the arc efficiency harc. In the operation during March and April, the acceleration current Iacc with two ion sources reached a level close to 40 A. The acceleration current in the operation after July, however, did not reach the level of operation in March and April. This seems to be correlated with an instantaneous air leak observed in the beamline during the beam acceleration, though we could not identify the point where the air leak broke out. The extraction current did not change much during the operation between March and September with deuterium. However, the extraction current decreased greatly after changing the gas species in October. This means that the electron component in the extraction current with the hydrogen beam is smaller than that of the deuterium beam. The arc efficiency of the ion source, which means the efficiency of negative ion production, was kept at a higher rate of around 0.15 in March and April. In the operation after exposure to air for ion source maintenance in May, the arc efficiency did not reach the level of March and April, though it had recovered gradually with operation time owing to cleaning of the arc chamber wall. This means that phenomena such as a tiny air leak in the ion source would be bad for negative ion production.
326
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
Since the upper ion source did not suffer serious damage as did the lower one through the water leak in September, the beam performance with only the upper source (solid dots) was a little better than that of the operation with two ion sources including the lower source.
3.3. Power flow on the beamline components Power loading on the beamline components was measured with water calorimetry. A diagram of power flow onto the beamline components is illustrated in Fig. 7. In the extractor, both negative ions and electrons are extracted, but only negative ions go into accelerator, because electrons are dumped on the extractor grid due to the
Fig. 6. Time evolution of the ion source performance (solid dots show the operation with only upper ion source).
Fig. 5. Time evolution of the neutral beam parameters for injection shots in 1998 (solid dots show the operation with only upper ion source).
action of permanent magnets buried in the extraction grid and a bias voltage between the arc chamber and the plasma grid. During the acceleration, 30–40% of the accelerated ions are lost due to imperfect beam optics, though the design value of the loss is less than 10%. The negative ions accelerated in the ion source enter the neutralizer cell just after passing through the first beam limiter. The beam losses in the first beam limiter and neutralizer cell are 3 and 7%, respectively. After passing through the neutralizer cell, the residual negative and positive ion beams, around 10–15% each, are reflected to horizontal directions through a stray magnetic field from JT-60U, and are finally captured in the ion dumps. About 5% of the neutral beam is lost in the drift duct owing to re-ionization loss ( 2%) and geometrical loss (3%). The remaining neutral beam, 2832%, is injected into JT-60U plasma. Corresponding power flows on the beamline components in percentage are shown in Table 1.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
Figs. 8 and 9 compare the performance of the lower and the upper ion sources as a function of arc power and extraction voltage, where the measured current ratio is the ratio of beam current impinging on each component against acceleration drain current Iacc, i.e. slow D− shown in Fig. 7. The lower ion source has a higher calorimeter current ratio (ratio of neutral beam power to the acceleration input power) and higher arc efficiency. These do not appear to saturate with increasing extraction voltage. On the other hand, the current ratio for each accelerator grid in the upper ion source is higher than that of the lower source. In the accelerator, much of the heat loads were deposited on the second acceleration grid (A2G) and grounded grid (GRG) in the JT-60 negative ion source. Here the current ratio on the first acceleration grid (A1G) was not measured, and was estimated from other measured current ratios. Comparing operation with hydrogen and deuterium, it is obvious that the operation with hydrogen has higher performance than with deuterium, as shown in Fig. 10. With a higher arc power and a higher extraction voltage, there are striking differences between hydrogen and deuterium. The calorimeter current ratio with hydrogen is roughly 10% higher than that with
327
deuterium at same arc power, even with a slightly lower extraction voltage. The arc efficiency with hydrogen is also 30–40% higher than with deuterium. Arc power Parc of more than 140 kW, in general, brings about the deterioration of both calorimeter current ratio and arc efficiency.
4. Discussion for the approach to the designated injection power Since its first operation in 1996, performance of the N-NBI was improved substantially by solving many problems involving in the ion source, power supplies, and beamline [1,6,11]. However, some difficulties continue to slow progress towards the design goal of a NBI power of 10 MW for 10 s at 500 keV. These are as follows.
4.1. Stabilizing arc discharge In a negative ion source, it is not easy to stabilize the arc discharge because of the presence of introducing cesium vapor in the arc chamber to enhance negative ion production rate. Most of the cesium deposits on the inner surface of the arc chamber where the surface temperature is kept at room temperature. The cesium flows into source
Fig. 7. Schematic diagram of power flow in the beamline.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
328
Table 1 Power loads on the beamline componentsa Ion Source
Fraction (%)
1BL
AC1G
AC2G
GRG
10
10–15
10–15
3
NC
7
ID
10–15
CM DD
INJ
5 3035
28–32
a
All current ratios above are that of beam current impinging on each component against acceleration drain current Iacc, i.e. slow D (AC1G and AC2G, first and second acceleration grid; GRG, grounded grid; 1BL, first beam limiter; NC, neutralizer cell; ID, each of ion dumps for D− and D+; DD, the drift duct; CM, the ratio of the neutral beam power entering the drift duct; INJ, the ratio of the injected neutral power to JT-60U plasma). −
plasma by evaporating during arc discharge, and hence the arc characteristic changes with a time constant of several seconds. Since beam perveance changes with changes in the arc discharge, secondary electrons produced on the grid surface cause the heat load on the grid increase due to increasing negative ion bombardments onto the accelerator grid, hence a breakdown occurs easily. Therefore, the arc discharge of the negative ion source has to be stabilized before beam acceleration. A pre-arc discharge with a duration of several seconds is necessary for stabilizing the arc discharge before beam acceleration.
4.2. Minimizing electron acceleration In a negative ion source, electrons are accelerated with the negative ions. Most of the electrons extracted with the negative ions are trapped in the extractor by applying a magnetic field, and less than a few percent leak into the accelerator. Electrons are generated in the accelerator by chargeexchange collisions with slow gases and by secondary electron emission due to impingement of negative ions onto the grids. These electrons accelerate with the negative ions and collide easily with the grids because of different trajectories than the negative ions. These phenomena easily cause accelerator breakdown. To minimize the electron acceleration, leakage electrons from the extractor have to be minimized by optimizing an extractor structure, and electrons produced in the accelerator have to be minimized by operating under as low a pressure as possible.
4.3. Impro6ing source plasma uniformity Though the issue of the non-uniformity of the source plasma is not described in this article, it is also one of the important issues. In the largescale arc chamber of the negative ion source, making the source plasma uniformly is not easy because of the magnetic field configuration necessary for producing negative ions. The magnetic fields affect the source plasma drift. The non-uniformity of the source plasma makes uniform beam acceleration difficult over a wide acceleration area because of mis-matching of beam perveance. Due to increasing negative ion bombardments onto the accelerator grid, secondary electrons produced on the grid surface cause the heat load on the accelerator increase greatly, and hence accelerator breakdown occurs easily.
4.4. Keeping good 6acuum condition in the ion source In the negative ion source, the cesium deposition surface is easily contaminated by oxygen. A very tiny air leak, which is not a problem at all for a positive-ion source, affects the negative ion production rate and decreases the negative ion current. Hence, the beam perveance in the accelerator changes and electrons stripped from negative ions through charge exchange increase the heat load on the accelerator grid, hence breakdown occurs easily.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
329
5. Summary Although some malfunctions such as a water leak in the lower ion source took place during the operation in 1998, much progress was made. After the water leak, a moderate neutral beam power could be injected into JT-60U. Some results from the whole year’s operation data of the N-NBI are, 1. beam divergence of the negative ion beam has been confirmed to be around 4 mrad in horizontal direction and 6 mrad in vertical direction. These roughly agree with the design values; 2. power loading measured on the beam components was reasonable. The upper and lower
Fig. 9. Comparison between the lower and upper ion source as a function of extraction voltage Vext (solid symbol, lower ion source; open symbol, upper ion source). Beam parameters: deuterium, Vacc =370 kV; Iacc =19 – 22 A; Parc =150 kW; Ipg =4.5 kA, Vb =4.6 V. Ipg is a current flowing through the plasma grid to separate the source plasma into a high temperature plasma and a low temperature one. Vb is a bias voltage applied between the arc chamber and the plasma grid to suppress electron currents coextracted with the negative ions. Fig. 8. Comparison between the lower and upper ion source as a function of arc power Parc at the same ion source operation condition (solid symbol, lower ion source; open symbol, upper ion source). A2G, second acceleration grid; GRG, grounded grid; 1BL, first beam limiter; NC, neutralizer cell; ID (D−) and ID (D+), ion dumps for D− and D+; CM, calorimeter, i.e. the neutral beam power entering the drift duct; D− (total), total D− current out of the accelerator.
ion sources had nearly the same performance. The operation with hydrogen had a higher performance than with deuterium; 3. beam performance has kept improving gradually. The best experimental results in 1998, not all achieved simultaneously, were, a maximum
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
330
References
Fig. 10. Comparison between different ion species in the same ion source and nearly same conditions (solid symbol, hydrogen beam; open symbol, deuterium beam).
beam energy of 380 keV; a maximum beam injection power of 5.2 MW and energy of 7.6 MJ; a longest pulse duration of 2.2 s; and a smallest beam divergence of 3.5 mrad in the horizontal direction and 5 mrad in the vertical direction. .
Acknowledgements The authors would like to thank the members of JT-60 group. They are also grateful to Dr M. Shimizu, Dr T. Kimura, Dr M. Ohta and Dr H. Kishimoto for their continuous encouragement. One of the authors, Liqun Hu, would like to thank Chinese Academy of Science, and JAERI to give him this opportunity to work at JAERI.
[1] M. Kuriyama, N. Akino, N. Ebisawa, L. Grisham, H. Liquen, A. Honda, et al., Development of negative-ion based NBI system for JT-60, J. Nucl. Sci. Technol. 35 (11) (1998) 739 – 749. [2] Y. Okumura, M. Hanada, et al., A 10 amperes negative ion source for high energy neutral beam injector, in: Proceedings of 16th SOFT, 1990, pp. 1026 – 1030. [3] K. Watanabe, M. Hanada, T. Inoue, M. Mizuno, Y. Ohara, Y. Okumura et al., High energy acceleration of H− ions, IAEA TCM on Negative Ion Based Neutral Beam Injectors, Naka, Japan, 1991, pp. 184 – 186. [4] M. Mizuno, M. Hanada, T. Inoue, J. Koizumi, S. Kunienda, M. Kuriyama et al., High energy acceleration of H− ions, IAEA TCM on Negative Ion Based Neutral Beam Injectors, Naka, Japan, 1991, pp. 259 – 263. [5] M. Kuriyama, N. Akino, M. Araki, N. Ebisawa, M. Hanada, T. Inoue, et al., Construction of a 500 keV/ negative-ion-based NBI system for JT-60U, Proceedings of 15th SOFE, Hyaniss, USA, 1993, pp. 470 – 473. [6] M. Kuriyama, N. Akino, T. Aoyagi, N. Ebisawa, N. Isozaki, A. Honda, et al., Operation of the negative-ion based NBI for JT-60U, Fusion Eng. Des. 39 – 40 (1998) 115 – 121. [7] M. Mizuno, M. Akiba, N. Akino, M. Araki, N. Ebisawa, M. Hanada, et al., Design of a 500 keV negative-ion-based NBI system for JT-60U, Plasma Devices Operat. 3 (1994) 199 – 210. [8] M. Kuriyama, N. Akino, M. Araki, N. Ebisawa, M. Hanada, T. Inoue, et al., High energy negative-ion based neutral beam injection system for JT-60U, Fusion Eng. Des. 26 (1995) 445 – 453. [9] L. Hu, NBI group, Investigation of power loading onto beamline components and time evolution of negativeion based NBI system for JT-60U tokamak, in: Proceedings of 15th Japan Annual Meeting of Plasma Physics and Nuclear Fusion, Tokyo, Japan, December 14, 1998, Oral, 3PC10, p. 332. [10] Y. Ohara, M. Kuriyama, A computer code for computing the beam profiles in the NBI line ‘BEMPROF’, JAERI-M82-066, July 1982. [11] M. Kuriyama, N. Akino, N. Ebisawa, et al., Power flow in the negative-ion based neutral beam injection for JT-60, Rev. Sci. Instrum. 71 (2) (2000) 751– 754.
Beam divergence and power loading on the beamline components of the negative-ion based NBI system for JT-60U Liqun Hu a,*, M. Kuriyama b, N. Akino b, N. Ebisawa b, A. Honda b, T. Itoh b, M. Kawai b, M. Kazawa b, M. Kusaka b, K. Mogaki b, T. Ohga b, K. Ohmori b, K. Ohshima b, Y. Okumura b, H. Oohara b, F. Satoh b, H. Seki b, Y. Tanai b, Y. Toyokawa b, K. Usui b, K. Watanabe b, M. Yamaguchi b, H. Yamazaki b, T. Yamazaki b a
Institute of Plasma Physics, Chinese Academy of Science, Shushanhu Road 350, Hefei 230031, People’s Republic of China b Japan Atomic Energy Research Institute, Naka Fusion Research Establishment, Mukouyama, Naka-machi, Naka-gun, Ibaraki-ken 311 -0193, Japan Received 7 August 2000; accepted 7 September 2000
Abstract In JT-60U, a high-energy neutral beam injection (NBI) program using negative-ion based NBI (N-NBI) for non-inductive current drive and core plasma heating studies in high-density plasma has progressed. The target performance of the N-NBI is a neutral beam injection power of 10 MW for 10 s at 500 keV. A neutral beam power of 5.2 MW at 350 keV with deuterium has already been injected into JT-60U plasma. The beam divergence and power loading onto the beamline components are two important items to evaluate beam performance. The beam divergence, estimated roughly from heat loads at the beam drift duct in beam injection experiments into JT-60U plasma, was around 4 mrad for the horizontal direction and 6 mrad for the vertical direction, which were close to the design values of 5 mrad. The power loading on beamline components were reasonable as compared with the design value. © 2001 Elsevier Science B.V. All rights reserved. Keywords: JT-60U; NBI; Negative-ion; Power loading; Beam divergence; Beamline
1. Introduction
* Corresponding author. Fax: +86-551-5591310. E-mail address: [email protected] (L. Hu).
High energy neutral beam injection (NBI) has been considered to play an essential role for noninductive current drive and core plasma heating in magnetically confined thermonuclear fusion re-
0920-3796/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 0 - 3 7 9 6 ( 0 0 ) 0 0 5 4 4 - 5
322
L. Hu et al. / Fusion Engineering and Design 54 (2001) 321–330
search. Moreover, it can be used for suppressing plasma instability and improving plasma confinement by delivering the momentum to the plasma in the toroidal direction, and for controlling the spatial distribution of the plasma current. For these purposes, high power NBIs that can deliver tens of MW neutral beams at 100 keV-class energies had been developed with positive-ion based NBI (P-NBI). The large plasma confinement devices such as JT-60U, joint European torus (JET) and TFTR have already attained high performance plasma conditions with D or D–T operation by injecting tens of MW neutral beams at an energy of about 100 keV. In particular, JT-60U and JET have already reached the breakeven conditions, i.e. fusion thermal output is equal to plasma heating power. For the reactor-like fusion devices such as international thermonuclear experimental reactors (ITER), and even the present large tokamak JT60U, a higher energy (0.5 MeV for JT-60U, 1 MeV for ITER) and a long pulse neutral beam is required for effective, central noninductive current drive and plasma heating at a high plasma density. Ion beam neutralization efficiency for the conventional P-NBI system decreases sharply with increasing the ion beam energy over 100 keV as shown in Fig. 1[1]. Therefore, it is necessary to utilize a negative ion beam, which
Fig. 1. Neutralization efficiency of both negative and positive ion beams (curves show a theoretical calculation, solid symbols are experimental result).
has neutralization efficiency as high as 60% when using a gas cell neutralizer. The negative ion source used in the NBI device is required to accelerate negative ions of a few tens of amperes at higher beam energy. Research on increasing negative ion source current has been carried out at many laboratories in the world for more than 30 years. The negative ion current achieved, however, was only 100 mA until 15 years ago. At JAERI, negative ion source R&D aiming at a higher beam current, more than 10 A, and a higher beam energy, over 500 keV, has been in progress since the mid 1980’s. Concerning higher currents, an H− current of 10 A at 50 keV was achieved in 1990 using a multi-cusp ion source with Cs seeding [2]. In 1994, a negative hydrogen beam of 180 mA at 400 keV was accelerated [3]. In addition to ion source R&D, the high voltage dc power supply for ion source and beamline components has been developed [4]. A several-year design study for a 500 keV negative-ion based NBI system was also completed. Based on these results, no technical problems were anticipated for the 500 keV negative-ion based NBI (N-NBI) system [5], shown in Figs. 2 and 3, and construction was begun. The performance goals were a neutral beam injection power of 10 MW for 10 s at 500 keV. Since the first operation of the N-NBI in 1996, the negative ion beam power per source has reached 14.3 A at 380 keV (5.4 MW) with deuterium and 18.5 A at 360 keV (6.6 MW) with hydrogen, and the neutral beam power injected into JT-60 has reached 5.2 MW at 350 keV with deuterium and 4.2 MW at 360 keV with hydrogen [1,6]. The neutralization efficiency of the negative ion beam has been confirmed to be about the 60% in the beam energy range of 250–385 keV, coinciding with theoretically predicted value as shown in Fig. 1. The estimation of the beam divergence and power loading on the beamline components is very important for ameliorating negative ion beam performance to achieve higher beam power. In this report, the power loading on the beamline components, estimation of neutral beam divergence, time evolution of the beam parameters for the beam injection experiments of 1998, and some issues to limit the N-NBI system to achieve de-
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
323
Fig. 2. Plan view of the 500 keV negative-ion based NBE system in the JT-60U torus hall.
signed performance goal are discussed. The layout of the N-NBI beamline is described briefly as well.
2. Layout of the 500 keV N-NBI beamline [7,8] The schematic drawings for the 500 keV N-NBI system are shown in Figs. 2 and 3. It is composed of one beamline, two ion sources, a power supply system, a control system and an auxiliary sub-system including water cooling, helium refrigeration and auxiliary vacuum pumping systems. The negative ion source is composed of a negative ion generator, an extractor and an accelerator. The negative ion generator is a cesium-seeded multicusp type. A small amount of cesium vapor is injected into the negative ion generator through a transport pipe from an external cesium oven to substantially enhance the negative ion yield by a factor of about 2 – 5 higher than that of the pure volume production. Furthermore, it considerably
decreases the optimum operating pressure without deterioration of negative ion production. The cesium effect strongly depends on the temperature of plasma grid. The optimum temperature is experimentally confirmed to be about 250–300°C. The extractor has three grids; a plasma grid, an extraction grid and an electron suppression grid. The extractor is a multi-aperture type and its aperture size is 14 min in diameter. The extraction area is 110 × 45 cm2, which is divided into five segments. The accelerator has also three grids; the first and second acceleration grids, and a grounded grid. The beams are geometrically directed to the focal point of 24 m from the ion source with a steering mechanism. Each beamline is also focussed using an aperture displacement technique. The beamline consists of an ion source tank, a neutralizer cell, an ion dump tank and a neutral beam drift duct. The ion source tank contains a set of cryopumps for minimizing the stripping loss
324
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
of negative ions in the ion source accelerator. The tank is covered with magnetic shielding to screen the stray magnetic field from JT-60. The neutralizer tank has two 10-m long neutralizer cells corresponding to the two ion sources. The ion dump tank contains a pair of bending coils, ion dumps for D− and D+, a retractable calorimeter, a beam scraper and cryopumps. The calorimeter placed after the ion bending coils is composed of arrays of a composite material of molybdenum and copper, and it is used for measuring the neutral beam power. It also serves as the beam target for the neutral beams when doing ion source conditioning. The neutral beam drift duct is composed of an isolation gate valve, a bellows and a duct containing a series of beam scrapers.
3. Beam characteristics of 1998 operations
3.1. Estimation method of the beam di6ergence The estimation of the beam divergence [9] is very important for ameliorating the negative ion beam performance to achieve higher beam power. The procedure for estimating beam divergence is shown in Fig. 4. The beam divergence in the horizontal and vertical directions, vx and vy,
were determined by comparing the heat loads measured on the beam limiters in the drift duct with the computational results. A three-dimensional beam trajectory calculation code ‘BEMPROF’ [10] is used for calculation of beam fractions deposited on the scrapers.
3.2. Time e6olution of the beam parameters during the neutral beam injection into JT-60 The time evolutions of the neutral beam parameters and ion source performance for the injection shots in 1998 are illustrated in Figs. 5 and 6. In these figures, open symbols show operation results with two ion sources and solid symbols show results with only the upper ion source. The beam species is deuterium before October and hydrogen in October. In the beginning of September, a cooling water leak occurred at a filament feedthrough inside the arc chamber of the lower ion source during the beam pulse. One month was needed for recovery. In the beam injection experiments, the beam energy Eb was around 350 keV and the maximum value was nearly 380 keV. The beam energy was kept a moderate level, around 350 keV, because the first priority in this experimental period was to extend the beam pulse duration. After the water
Fig. 3. A side view of the N-NBI beamline for JT-60U.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
325
Fig. 4. Estimation procedure for the beam divergence.
leak, the beam energy was limited to 300 –330 keV in spite of cleaning the inside of the ion source. This voltage deterioration seems to have been caused by micro-contamination on the surface of the accelerator grids due to the water leak. The integrated injection power Einj.m, i.e. the beam injection energy, defined as the product of beam power Pinj.m and beam pulse duration T, was around 3 MJ and the maximum value was 6.7 MJ. Although the beam pulse duration was around 1 s at the beginning of 1998, the pulse length was extended gradually with the operation time, and reached 2.2 s. The extension of the beam pulse benefited from a stabilization of the arc discharge by taking a longer pre-arc discharge time before beam acceleration in the negative ion generator, which will be discussed later in Section 4. The beam divergence estimated in the beam injection experiments has been confirmed around 4 mrad in the horizontal direction (vx ), and around 6 mrad in the vertical direction (vy ). These divergences roughly agree with the design values of 5 mrad. The beam characteristics of beam energy Eb, the beam injection energy Einj.m and beam pulse duration kept increasing gradually before October, while the beam divergence kept almost constant. After the water leak in September, these parameters became bad owing to the contamination in the ion source by the water leak. The other parameters of acceleration current Iacc, extraction current Iext and arc efficiency harc
are shown in Fig. 6, where harc is defined by the ratio of acceleration current Iacc and arc power Parc. The acceleration current with two ion sources follows the same trace as the arc efficiency harc. In the operation during March and April, the acceleration current Iacc with two ion sources reached a level close to 40 A. The acceleration current in the operation after July, however, did not reach the level of operation in March and April. This seems to be correlated with an instantaneous air leak observed in the beamline during the beam acceleration, though we could not identify the point where the air leak broke out. The extraction current did not change much during the operation between March and September with deuterium. However, the extraction current decreased greatly after changing the gas species in October. This means that the electron component in the extraction current with the hydrogen beam is smaller than that of the deuterium beam. The arc efficiency of the ion source, which means the efficiency of negative ion production, was kept at a higher rate of around 0.15 in March and April. In the operation after exposure to air for ion source maintenance in May, the arc efficiency did not reach the level of March and April, though it had recovered gradually with operation time owing to cleaning of the arc chamber wall. This means that phenomena such as a tiny air leak in the ion source would be bad for negative ion production.
326
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
Since the upper ion source did not suffer serious damage as did the lower one through the water leak in September, the beam performance with only the upper source (solid dots) was a little better than that of the operation with two ion sources including the lower source.
3.3. Power flow on the beamline components Power loading on the beamline components was measured with water calorimetry. A diagram of power flow onto the beamline components is illustrated in Fig. 7. In the extractor, both negative ions and electrons are extracted, but only negative ions go into accelerator, because electrons are dumped on the extractor grid due to the
Fig. 6. Time evolution of the ion source performance (solid dots show the operation with only upper ion source).
Fig. 5. Time evolution of the neutral beam parameters for injection shots in 1998 (solid dots show the operation with only upper ion source).
action of permanent magnets buried in the extraction grid and a bias voltage between the arc chamber and the plasma grid. During the acceleration, 30–40% of the accelerated ions are lost due to imperfect beam optics, though the design value of the loss is less than 10%. The negative ions accelerated in the ion source enter the neutralizer cell just after passing through the first beam limiter. The beam losses in the first beam limiter and neutralizer cell are 3 and 7%, respectively. After passing through the neutralizer cell, the residual negative and positive ion beams, around 10–15% each, are reflected to horizontal directions through a stray magnetic field from JT-60U, and are finally captured in the ion dumps. About 5% of the neutral beam is lost in the drift duct owing to re-ionization loss ( 2%) and geometrical loss (3%). The remaining neutral beam, 2832%, is injected into JT-60U plasma. Corresponding power flows on the beamline components in percentage are shown in Table 1.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
Figs. 8 and 9 compare the performance of the lower and the upper ion sources as a function of arc power and extraction voltage, where the measured current ratio is the ratio of beam current impinging on each component against acceleration drain current Iacc, i.e. slow D− shown in Fig. 7. The lower ion source has a higher calorimeter current ratio (ratio of neutral beam power to the acceleration input power) and higher arc efficiency. These do not appear to saturate with increasing extraction voltage. On the other hand, the current ratio for each accelerator grid in the upper ion source is higher than that of the lower source. In the accelerator, much of the heat loads were deposited on the second acceleration grid (A2G) and grounded grid (GRG) in the JT-60 negative ion source. Here the current ratio on the first acceleration grid (A1G) was not measured, and was estimated from other measured current ratios. Comparing operation with hydrogen and deuterium, it is obvious that the operation with hydrogen has higher performance than with deuterium, as shown in Fig. 10. With a higher arc power and a higher extraction voltage, there are striking differences between hydrogen and deuterium. The calorimeter current ratio with hydrogen is roughly 10% higher than that with
327
deuterium at same arc power, even with a slightly lower extraction voltage. The arc efficiency with hydrogen is also 30–40% higher than with deuterium. Arc power Parc of more than 140 kW, in general, brings about the deterioration of both calorimeter current ratio and arc efficiency.
4. Discussion for the approach to the designated injection power Since its first operation in 1996, performance of the N-NBI was improved substantially by solving many problems involving in the ion source, power supplies, and beamline [1,6,11]. However, some difficulties continue to slow progress towards the design goal of a NBI power of 10 MW for 10 s at 500 keV. These are as follows.
4.1. Stabilizing arc discharge In a negative ion source, it is not easy to stabilize the arc discharge because of the presence of introducing cesium vapor in the arc chamber to enhance negative ion production rate. Most of the cesium deposits on the inner surface of the arc chamber where the surface temperature is kept at room temperature. The cesium flows into source
Fig. 7. Schematic diagram of power flow in the beamline.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
328
Table 1 Power loads on the beamline componentsa Ion Source
Fraction (%)
1BL
AC1G
AC2G
GRG
10
10–15
10–15
3
NC
7
ID
10–15
CM DD
INJ
5 3035
28–32
a
All current ratios above are that of beam current impinging on each component against acceleration drain current Iacc, i.e. slow D (AC1G and AC2G, first and second acceleration grid; GRG, grounded grid; 1BL, first beam limiter; NC, neutralizer cell; ID, each of ion dumps for D− and D+; DD, the drift duct; CM, the ratio of the neutral beam power entering the drift duct; INJ, the ratio of the injected neutral power to JT-60U plasma). −
plasma by evaporating during arc discharge, and hence the arc characteristic changes with a time constant of several seconds. Since beam perveance changes with changes in the arc discharge, secondary electrons produced on the grid surface cause the heat load on the grid increase due to increasing negative ion bombardments onto the accelerator grid, hence a breakdown occurs easily. Therefore, the arc discharge of the negative ion source has to be stabilized before beam acceleration. A pre-arc discharge with a duration of several seconds is necessary for stabilizing the arc discharge before beam acceleration.
4.2. Minimizing electron acceleration In a negative ion source, electrons are accelerated with the negative ions. Most of the electrons extracted with the negative ions are trapped in the extractor by applying a magnetic field, and less than a few percent leak into the accelerator. Electrons are generated in the accelerator by chargeexchange collisions with slow gases and by secondary electron emission due to impingement of negative ions onto the grids. These electrons accelerate with the negative ions and collide easily with the grids because of different trajectories than the negative ions. These phenomena easily cause accelerator breakdown. To minimize the electron acceleration, leakage electrons from the extractor have to be minimized by optimizing an extractor structure, and electrons produced in the accelerator have to be minimized by operating under as low a pressure as possible.
4.3. Impro6ing source plasma uniformity Though the issue of the non-uniformity of the source plasma is not described in this article, it is also one of the important issues. In the largescale arc chamber of the negative ion source, making the source plasma uniformly is not easy because of the magnetic field configuration necessary for producing negative ions. The magnetic fields affect the source plasma drift. The non-uniformity of the source plasma makes uniform beam acceleration difficult over a wide acceleration area because of mis-matching of beam perveance. Due to increasing negative ion bombardments onto the accelerator grid, secondary electrons produced on the grid surface cause the heat load on the accelerator increase greatly, and hence accelerator breakdown occurs easily.
4.4. Keeping good 6acuum condition in the ion source In the negative ion source, the cesium deposition surface is easily contaminated by oxygen. A very tiny air leak, which is not a problem at all for a positive-ion source, affects the negative ion production rate and decreases the negative ion current. Hence, the beam perveance in the accelerator changes and electrons stripped from negative ions through charge exchange increase the heat load on the accelerator grid, hence breakdown occurs easily.
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
329
5. Summary Although some malfunctions such as a water leak in the lower ion source took place during the operation in 1998, much progress was made. After the water leak, a moderate neutral beam power could be injected into JT-60U. Some results from the whole year’s operation data of the N-NBI are, 1. beam divergence of the negative ion beam has been confirmed to be around 4 mrad in horizontal direction and 6 mrad in vertical direction. These roughly agree with the design values; 2. power loading measured on the beam components was reasonable. The upper and lower
Fig. 9. Comparison between the lower and upper ion source as a function of extraction voltage Vext (solid symbol, lower ion source; open symbol, upper ion source). Beam parameters: deuterium, Vacc =370 kV; Iacc =19 – 22 A; Parc =150 kW; Ipg =4.5 kA, Vb =4.6 V. Ipg is a current flowing through the plasma grid to separate the source plasma into a high temperature plasma and a low temperature one. Vb is a bias voltage applied between the arc chamber and the plasma grid to suppress electron currents coextracted with the negative ions. Fig. 8. Comparison between the lower and upper ion source as a function of arc power Parc at the same ion source operation condition (solid symbol, lower ion source; open symbol, upper ion source). A2G, second acceleration grid; GRG, grounded grid; 1BL, first beam limiter; NC, neutralizer cell; ID (D−) and ID (D+), ion dumps for D− and D+; CM, calorimeter, i.e. the neutral beam power entering the drift duct; D− (total), total D− current out of the accelerator.
ion sources had nearly the same performance. The operation with hydrogen had a higher performance than with deuterium; 3. beam performance has kept improving gradually. The best experimental results in 1998, not all achieved simultaneously, were, a maximum
L. Hu et al. / Fusion Engineering and Design 54 (2001) 323–330
330
References
Fig. 10. Comparison between different ion species in the same ion source and nearly same conditions (solid symbol, hydrogen beam; open symbol, deuterium beam).
beam energy of 380 keV; a maximum beam injection power of 5.2 MW and energy of 7.6 MJ; a longest pulse duration of 2.2 s; and a smallest beam divergence of 3.5 mrad in the horizontal direction and 5 mrad in the vertical direction. .
Acknowledgements The authors would like to thank the members of JT-60 group. They are also grateful to Dr M. Shimizu, Dr T. Kimura, Dr M. Ohta and Dr H. Kishimoto for their continuous encouragement. One of the authors, Liqun Hu, would like to thank Chinese Academy of Science, and JAERI to give him this opportunity to work at JAERI.
[1] M. Kuriyama, N. Akino, N. Ebisawa, L. Grisham, H. Liquen, A. Honda, et al., Development of negative-ion based NBI system for JT-60, J. Nucl. Sci. Technol. 35 (11) (1998) 739 – 749. [2] Y. Okumura, M. Hanada, et al., A 10 amperes negative ion source for high energy neutral beam injector, in: Proceedings of 16th SOFT, 1990, pp. 1026 – 1030. [3] K. Watanabe, M. Hanada, T. Inoue, M. Mizuno, Y. Ohara, Y. Okumura et al., High energy acceleration of H− ions, IAEA TCM on Negative Ion Based Neutral Beam Injectors, Naka, Japan, 1991, pp. 184 – 186. [4] M. Mizuno, M. Hanada, T. Inoue, J. Koizumi, S. Kunienda, M. Kuriyama et al., High energy acceleration of H− ions, IAEA TCM on Negative Ion Based Neutral Beam Injectors, Naka, Japan, 1991, pp. 259 – 263. [5] M. Kuriyama, N. Akino, M. Araki, N. Ebisawa, M. Hanada, T. Inoue, et al., Construction of a 500 keV/ negative-ion-based NBI system for JT-60U, Proceedings of 15th SOFE, Hyaniss, USA, 1993, pp. 470 – 473. [6] M. Kuriyama, N. Akino, T. Aoyagi, N. Ebisawa, N. Isozaki, A. Honda, et al., Operation of the negative-ion based NBI for JT-60U, Fusion Eng. Des. 39 – 40 (1998) 115 – 121. [7] M. Mizuno, M. Akiba, N. Akino, M. Araki, N. Ebisawa, M. Hanada, et al., Design of a 500 keV negative-ion-based NBI system for JT-60U, Plasma Devices Operat. 3 (1994) 199 – 210. [8] M. Kuriyama, N. Akino, M. Araki, N. Ebisawa, M. Hanada, T. Inoue, et al., High energy negative-ion based neutral beam injection system for JT-60U, Fusion Eng. Des. 26 (1995) 445 – 453. [9] L. Hu, NBI group, Investigation of power loading onto beamline components and time evolution of negativeion based NBI system for JT-60U tokamak, in: Proceedings of 15th Japan Annual Meeting of Plasma Physics and Nuclear Fusion, Tokyo, Japan, December 14, 1998, Oral, 3PC10, p. 332. [10] Y. Ohara, M. Kuriyama, A computer code for computing the beam profiles in the NBI line ‘BEMPROF’, JAERI-M82-066, July 1982. [11] M. Kuriyama, N. Akino, N. Ebisawa, et al., Power flow in the negative-ion based neutral beam injection for JT-60, Rev. Sci. Instrum. 71 (2) (2000) 751– 754.