TFR GROUP Association Euratom-CEA sur la Fusion, D~partement de Physique du Plasma et de la Fusion Contr61~e, Centre d'Etudes Nucl~aires, Bofte Postale no. 6, 92260 Fontenay-aux-Roses, France
Received 7 August 1978
The TFR program is a series of tokamak devices working in a wide range of toroldal magnetic fields (B T ~ 60 kG). This field is produced by a set of oil-cooled Bitter coils powered by an alternator coupled to a fly-wheel. The experimental results obtained on TFR 400 the first version (plasma current 400 kA) are given, the scaling laws for the main parameters of the discharge and the power balance are discussed. The previous results obtained on TFR 600 an upgraded version (plas- ma current 600 kA) are also given and the TFR 604 (plasma current 600 kA, 4 MW neutral injection heating) is presented.
I. Introduction
After the success obtained by soviet physicists in thermonuclear fusion research with magnetic config- urations of the tokamak type, it was decided to start the T F R experiment (Tokamak de Fontenay aux Roses) around the beginning of 1970. A tokamak is an axisymmetrical toro~dal system in which a hot and rather dense plasma is confined within a magnetic
field created by a current flowing through it (pololdal field 1 - 4 kG). The plasma column remains stable against magnetohydrodynamic motions provided that a strong toroTdal magnetic field (some tens of kilo- gausses) exist inside the plasma. The magnetic lines formed by the superposition of the pololdal and torotdal fields are helices which revolve around a torordal axis (see fig. 1). The toroldal field is pro- duced by external coils surrounding the plasma col-
Co~s produc~ the toro~ld, nn~'~W~
Bz fietd
B / Fig. 1. Magnetic configuration in a tokamak device.
332
TFR Group / Plasma confinement in TFR tokamak 333
umn whereas the plasma current is produced by an inductive coupling with coils which have the same axis as the torus.
At first, the TFR project was designed to study the physical properties of a hot plasma (i.e., with an electron temperature Te ~ 2 KeV, an ion temperature T i ~ 1 KeV and a mean density in the range of ne ~ 10 ~ a _ 10x4 cm-a). The technical parameters of the first version (TFR 400) which was in use from April 1973 to August 1976, were a plasma current, Iv = 400 kA, and a maximum toro~dal magnetic field, BT = 60 kG; the major radius of the toroidal vacuum chamber being R = 98 cm and the plasma radius a = 20 cm. The plasma heating, which is primarily ohmic heating due to the current flowing in the plasma (ohmic power P a <~ 800 kW) was supplemented in 1975-1976 by a fast neutral injection to increase the ion temperature. This method consists of injecting high-energy neutral atoms (E ~ 35 KeV) which are ionized in the plasma and confined, and which subse- quently transfer their energy to the bulk plasma (the maximum neutral power injected in TFR was PN = 570 kW). A physical study of the plasma in TFR was done using a number of diagnostic techniques which are reviewed in ref. [1 ]. The scaling laws for the main parameters of the plasma (temperatures, density and energy confinement time) were derived and the power balance without or with neutral injection was estab- lished.
During the period from August 1976 to September 1977, the TFR 400 was disassembled and replaced by a new version, the TFR 600. The main difference with the previous version, is a maximum plasma radius of a = 24 cm, allowing a plasma current o f I v = 600 kA. The toroldal field system is unchanged. The access to the plasma is improved in the new modification. 3 Types of additional heating methods were prepared for this machine: neutral injection as in TFR 400 (a doubling of the neutral power injected is expected), ion cyclotron resonance heating (ICRI-1) and cluster injection. In the ICRH method a wave is launched into the plasma at twice the l_armor frequency of the ions and interacts with them, so as to transfer energy to the plasma. A 500 kW generator is used for this heating experiment. In the last method hydrogen clus- ters composed of some tens or hundreds of atoms with an energy of 600 keV will be injected into the TFR 600. The injected power will be 100 kW. The
TFR 600 experiment is built to extend the physical study of the confinement and heating of a plasma in a tokamak device after TFR 400. The experiment started in the fall of 1977 and is now in progress.
While experiments on TFR 600 continue, a new project is proposed, namely TFR 604, which still uses the same toro~dal magnetic field system. The name of the project gives the parameters 600 kA (i.e. the same machine as TFR 600) and 4 MW of injected fast neu- trals. The neutral injection power can be increased thanks to a modification of some toro[dal magnetic field coils which leads to a better access of the neutral beam. This experiment is scheduled for 1980-1982.
The technical data of the TFR 400 device are described in the section 2. The results of the physical exploitation of the machine (scaling laws and power balance) in ohmic heating conditions are given as well as in neutral heating conditions. The third section is devoted to the TFR 600 version for which the first results are given. Finally the TFR 604 project is briefly discussed in section 4.
2. The TFR 400 tokamak
2.1. Description o f the machine
2.1.1. General features [2] A general view of TFR is given in fig. 2. A primary
circuit induces a current Ip in the plasma through a magnetic circuit as in a transformer. The magnetic flux variation is 2 V • s. This primary circuit is fed by a condenser bank (800 kJ) plus a 3 MW rectifier power supply. The maximum plasma current is 400 kA. The plasma is contained in a toro~dal vacuum chamber composed of eight inconel corrugated bellows (liner) separated by observation ports. The chamber is pumped by two turbomolecuiar pumps in series (600 and 70 l/s) supplemented by a roughing pump. Usual background pressures are in the range 10 -a Torr. In order to decrease impurity desorption from the wall (oxygen, CO, CO2, H 2 0 . . . ), the liner can be baked up to 500°C and a "discharge cleaning" procedure is possible (bombardment of the wall by relatively low temperature and low density plasmas with a low rep- etition rate "-0.2 Hz). A molybdenum limiter (massive annular ring) defines the boundary of the plasma and protects the chamber from direct particle impacts. A
334
equilibrium coits
inductive colts
observation port
inconet - - I finer
TFR Group / Plasma confinement in TFR tokamak
poloida[ field circuit
netisation
l field
shet~
Fig. 2. General view of TFR.
thick copper shell surrounding the liner prevents fast plasma motions thanks to the eddy currents they induce in the shell. Unfortunately these currents are damped due to the finite resistivity of the copper (time constant ~50 ms) and for a long time-equilibri- um a vertical field produced by the primary circuit interacts with the plasma current to bring back the plasma to a central position within the chamber. The vertical field is driven by a feed-back system which reacts to plasma displacements measured by magnetic pick-up coils.
2.1.2. Torordal IieM magnet [3] To reach the maximum value of the plasma current
(400 kA) a torordal magnetic field of 60 kG is needed to satisfy the magnetohydrodynamic stability criterion (see section 2.2). The field is produced by a series of 24 oil-cooled Bitter type coils. Each coil consists of 35 turns of full hard copper. Each turn is insulated with polyimide sheets (see fig. 3a, b). A turn utilizes two copper plates and two insulator sheets stacked to form a double copper helix and a double insulator helix. Each coil is clamped between two high resis- tance aluminium alloy plates (damping pressure ~60 atm). A coil prototype at ~ scale has been tested at
a
CooLing slits
, ° , ° ,
• ~ °
, * ° * • ° i
lOOO
b InsuLation
~ / ~ o p p e r
Fig. 3. (a) Plate of copper used for the toroldal magnet, (b) stackkng mode for one turn.
TFR Group / Plasma confinement in TFR tokamak 335
[ j (10¢A)
4
,(u)
2 U
1
t(s) I I I I I
0.5 1 1.5 2 2.5
Fig. 4. Voltage and current in the magnet (60 kG field).
the magnet laboratory at Cambridge. The model coil has been energized with the pulsed power generator of the laboratory, under severe test conditions: the mean current density was 3.5 times higher than on TFR a n d the adiabatic temperature rise was more than 200°C. The cooling of the coils is provided by a dielectric fluid (pyralene) which is preferred to demineralised water. Indeed, since TFR is not easily disassembled a long life system is needed which cannot be achieved with water cooling, due to infiltrations between the copper plates which slowly destroy the insulators. The coolant is pumped in a closed loop which contains 1.5 m 3 of fluid, a heat exchanger transfers the heat to water in a primary circuit.
5.SKy
EXCITATION THYRISTOR m, hl BRIDGES o -,~ x i,-
u,. o AUTO.EXCITATION RECTIFIER
2.1.3. Power supply for the magnet [4] The magnet operates in a pulsed regime and is ener-
gized by a bipolar alternator coupled to a fly-wheel through a short torsible shaft. Due to the working conditions (1 s long pulses, with a pulse every 4 mn at full power) this solution allows a significant reduc- tion of the power generator cost. The shapes of the current and voltage delivered by the rectifier which follows the starer of the alternator are shown in fig. 4. The main characteristics of the power supply are given in table 1, and the electrical diagram of the system is displayed in fig. 5.
_ STATOR WINOIN6S
CIRCUIT B I E A ~
0o,,c,o. lit
LOA O R" 70 m.tt I.= 63mH
Fig. 5. Electrical diagram of the toroi'dal field generator.
2.2. Experimental results in ohmic heating conditions [51
2. 2.1. Scaling laws A typical (300 kA, 50 kG) discharge is shown in
fig. 6. The main parameters of the discharge are plot- ted as functions of time. When the plasma current and toroldal field conditions are varied all over the avail- able range (up to 400 kA, 60 kG), one obtains the scaling laws for the main parameters of the plasma. This has been done on TFR for a large number of conditions.
In fig. 7 are plotted the variations of the main parameters of the plasma as functions of the plasma current (for stationary conditions):
(i) maximum central and mean densities achiev- able for a given current: ne(0), no;
(ii) central and mean electron temperatures Te(0), To;
0il) central ion temperature Ti(0);
336 TFR Group /Plasma confinement in TFR tokamak
Table 1 Main parameters of the toroidal field generator
Rotor diameter 50 cm Rotor length 220 cm Stator outer diameter 110 cm Rotor weight 3800 kg Stator weight 11400 kg Weight of the fly wheel 13500 kg Number of turns of the rotor winding 132 Rotor current 4000 A Rotor input voltage 900 V Rotor speed 6000 rpm Number of turns per phase 6 Number of phases 6 Peak voltage per phase 3500 V Peak current delivered by the rectifiers
connected in parallel 35 kA Peak power fed to the rotor 3.5 MW Peak power delivered by the stator 120 MW Energy stored in the fly wheel 400 MJ Energy delivered to the load in normal
operation 150 MJ Repetition rate at full power 1 pulse/4 mn
(iv) energy containment time of both ions and elec- trons rE deffmed as the ratio of the kinetic energy con- tained in the plasma (neTe + niTi) to the ohmic input
Ip
= / f - 100
i I I . V
volts (~) 3
1 I I I
xlO TM (~) cnn "3 3
2 1
T, (0) t t t keV
2 ,, .¢2t- ...... "4 . . . . . . . . . . . . "+-
I L I . T, (o)
keY I (~)
o,s /
o 2;o Fig. 6. Character ist ics o f a t yp ica l deu te r ium tokamak dLs- charge in TFR a = 20 cm, B T = 50 kG, Ip = 300 kA. From top to bottom: (a) plasma current Ip, (b) loop voltage V, (c) mean electron density Ee, (d) central electron temperature Te(0), (e) central ion temperature Ti(0).
power (Pn = V X Ip where V is the resistive loop vol-
tage). As can be seen in fig. 7a the maximum density
which can be achieved varies linearly with the plasma current. The limitation is due to "disruptions" of the current (i.e. break-down of the current) on a time scale of some milliseconds. The central electron tem- perature plotted in fig. 7b saturates for plasma cur- rents larger than 200 kA whereas the mean tempera- ture increases continuously. This indicates a broaden- ing of the temperature profile which is explained in section 2.2.2. In fig. 7c, the experimental scaling for the central ion temperature is compared to the classi- cal scaling I~13/x/A which takes place when binary coulomb collisions are responsible both for the heat
10
5
1
3'
2
1.0
08 0.6
0.4
(12
Hi, n,(O) 10 ~3/ / / :cm -3 //~/ /
'o ; ,
T__, (0) k,v @ T, • t T, (0)
~ I L I L _
T i (0) keV @ ~ ,
f ~ o D ptosmo
. - o . 0 . o0 ,oo .
f t I I " I _
20
15 10
5 ]p (kA) I I I J -
100 200 300 400
Fig. 7. Scaling of the main parameters of the TFR plasma versus plasma current from top to bottom: (a) mean and peak electron density, (b) mean and peak electron tempera- ture, (c) central ion temperature, (d) global energy contain- ment time.
TFR Group/Plasma confinement in TFR tokamak 337
transfer from electrons to ions and for heat conduc- tion loss of ions (,4 is the mass of the ions). This law seems to be satisfied except for hydrogen plasmas at plasma currents larger than 200 kA. Finally the energy containment time TE which scaled as rE ~ a/p [6] in soviet tokamaks is plotted in fig. 7d. It is obvious that this law is not verified for plasma currents above 200 kA. For these values the saturation of TE is due to the saturation of Te.
All these results have been presented, in the usual way, as functions of the plasma current. They can also be plotted against toroMal magnetic field. Indeed the main r61e of the toroMal magnetic field is to insure that the magnetohydrodynamic stability criterium
BT ~ BT q(O) Rj(O) Rip > 1 (1)
is fulfilled [frO) is the central current density]. In most experiments on TFR, the central q value (safety fac-
n¢ 1o) ,(1013 ¢m_1
10 * % (o ) ~ • • 8 ~ " "e
6 • ° • • pox x e
..L
Te(°) keY
2 x Te ( ° ) ~ ~ x
~ 1 I I I I - t
T F mS •
20 • •
e l * - -
1C , ,
, , B ~ ( ~
10 20 30 ~0 50 60
F ig 8 Scaling of the main parameters of the TFR plasma Ver sus toro~'dal magnetic field. (a) Mean and peak electron den- sity (maximum), (b) central ion and electron temperature, (c) energy containment time.
tor) is of the order of 1 so that the larger the toro~dal field, the larger the permissible plasma current (i.e. without disruptions). The maximum density, electron and ion central temperatures and the energy contain- ment time are plotted versus the toroMal magnetic field in fig. 8. (The corresponding plasma current is close to the maximum possible value for each mag- netic field value). The increase of the density and tem- peratures when B T increases is due to the possibility to have a larger current whereas the energy contain- ment time seems to saturate. This last point is due to the fact that the ohmic input power increases roughly as fast as the kinetic pressure (neTe + niTi).
General scaling laws are derived from the analysis of the TFR results, and given hereafter. If one assumes that the internal disruptions are responsible for the losses, the energy containment time can be written as
Te ~ he(O) ~ a2R . (2) Another scaling law is derived from the considera-
tion that in good experimental conditions, the loop voltage is nearly constant (~2 V); in that case:
T E ~ R a B q ( a ) s/4 . (3)
Actually, due to the connexion which exists between the different parameters of the plasma these 2 laws are not significantly different.
2 . 2 2 Power balance in ohmic heating[5] The different effects which must be accounted for
_ t I = 1 ions - P=i
/P~, IP~. %,. VPDi VPI) . "Z'~ P R
Fig. 9. Power balance diagram for the ions and the electzons of the plasma. Pf~ = ohmic power; PE = extracted power from the sources; PN = neutral a t o m s power; PNi = part o f P N trans- ferred to the ions (neutral injection); PNe = part o f P N trans- ferred to the electrons (neutral injection); PR = radiated power (line radiation essentially); Pcx = charge exchange losses; Pie and Pei = transfer between electrons and ions;
PKi and PDi = conduction and diffusion losses of ions; PKe and PDe = conduction and diffusion losses of electrons.
338 TFR Group / Plasma confinement in TFR tokamak
in the ion and electron energy balance are displayed in fig, 9. The energy balance is estimated on the basis of the density and temperature profiles measurements in the stationnary state (Ip = const.). This study leads to a division of the plasma into three regions.
2.2. 2.1. A central core (0 < r ~ 5 cm) In this region the profiles are flat and the density
and temperatures are maximum. The soft X-rays emit- ted by the plasma (intensity proportional to no 2 T~e with 1.7 < 5 ~< 4) show that a sawtooth modulation occurs when the central electron temperature Te(0) reaches its maximum value (see fig. 10). This fluctua- tion corresponds to a 10-15% sawtooth oscillation of the electron temperature and a 1-2% sawtooth oscillation of the electron density as shown by other measurements. These sawtooth relaxations correspond
i {o)101Zcm -3 140kA 2SkG~Jl Sl
' r , 0
' 't++/~o. ~ - I - . . . . . . , ~ ' ° " I
I , ( * ) W
t m s J I I I I
50 10o 1SO 200 2SO
2.10 3
Fig. 10. (a) Electron temperature and density versus time, (b) oscillogram of the soft X-ray emission exhibiting a saw- tooth fluctuation, (c) enlargement of the soft X-ray signal
to a heat transfer from the central core where q becomes less than 1 [see eq. (1)] to the outside (where q is larger than I). This can be explained as follows: since the ohmic power is proportional to T 312e , if the loss term varies more slowly with Te, the larger T e the more it increases which is the case at the center. The current profile (~T3e 12) peaks and the stability con- dition [eq. (1)] fails. Consequently MHD instabilities occur which destroy the local magnetic lines hence a broadening of the current profile and a heat transfer take place. When the broadening of the prof'fle is suffi- cient to provide again the condition q > 1, the process starts again. In consequence, when the torordal mag- netic field is increased for discharges with constant plasma current, the condition q(0) > 1 remains ful- filled for more peaked currents and the central elec- tron temperature increases. This is shown in fig. 11. In addition to that turbulent process, a part of the power is transferred to the outside by impurity radia- tion. Indeed in the case of the plasma (Te ~> 1 keV) the light impurities (oxygen, c a r b o n . . . ) are com- pletely ionized and thus do not radiate lines (which is
Tt kiV
Tp= 140kA q (a) _ ~ BT=+ 50 k C, OU$$ 7
x 40k Gous$ 5,7 2 " ~ e 33kOouss /,,7
+~ • 25kGouss 3,6
1,5 -
I
0 , 5
I I
5 10 15 r cm 20
Fig. 11. Elect~on temperature profile for a constant plasma current (140 kA) and various toroldal magnetic fields: B T = 25, 33, 40, 50 kG.
TFR Group / Plasma confinement in TFR tokamak 339
the dominant radiation loss in the keV range of tem- perature). In contrast the heavy impurities such as Me ~°+, Me 31+ which have been identified emit line radiation (excitation) which has been estimated to represent a 60 kW loss, i.e. 15% of the ohmic power at the center.
The power transferred by collisions from the elec- trons to the ions which can be calculated from the temperature and density profiles is not balanced by the power loss associated with the different mecha- nisms which have been accounted for. These are the- charge exchange, particle diffusion and classical heat conduction. If one assumes that the discrepancy is due to an anomalous heat transfer, the anomalous conduc- tion coefficient Kia can be compared to the classical one Kic (collisions). One can see in fig. 12a that the anomaly reaches a maximum in the intermediate region region (see hereafter) and in fig. 12b that the anomaly increases with the plasma current (i.e. with the ion temperature since T i ~ 1313 ) which is in contradiction
Kio+Kic
a i I , , , , I , , , , I ,
5 10 15 =V(cm}
3'Kt=÷.K=o i,= 2°ml' I I ' ~ i ¢ ;
1 It" lr
b rplkA I I ! • '
0 100 200 300 /,OO Fig. 12. (a) Ratio of the experimental to the classical ion heat conduction coefficient versus the plasma radius, (b) the same ratio as a function of the plasma current.
with classical diffusion theory. Up to now no physical effect has been clearly identified which could explain such a result.
2 . 2 2 2 An intermediate region (5 ~ r ~ 15 cm) This region corresponds to the maximum gradient
of the density and temperature profiles. In conse- quence the power balance is dominated by DVn or KVT effects (when D and K are diffusion and con- duction coefficients respectively). In this region, radia- tion loss and electron-ion transfer are responsible for 20% and 10% respectively in the electron power balance. The remaining 70% are due to diffusion and conduction but they cannot be accounted for by clas- sical diffusion alone (binary collisions). Turbulent pro- cesses must be considered to explain the experimental results. A possible candidate is the dissipative trapped electron instability [7]. Indeed the diffusion and con. duction coefficients associated with this instability have been evaluated for different experimental condi. tions [8] and it has been shown that the main electron temperature Te scales with plasma current as predicted by the theory. Nevertheless, direct measurements of the turbulent wave spectra by microwave scattering have been carried out (determination of the frequency and wave lengths of the turbulent waves) leading to the density fluctuation. Theoretical calculations using this result and assuming that the turbulence is actually due to instable driftwaves give a thermal conduction coefficient somewhat lower than needed to explain the experimental electron balance [9]. However, the measurement was not carried out in the region where the turbulence is expected to reach a maximum and new experiments on TFR 600 must shed some light on that point.
The behaviour of the ions in this region is similar to that of the central core, the anomaly being maxi- mum (see fig. 12).
2.2. 2.3. An outer region ( 15 cm ~ r ~ 20 cm) Here the ohmic power is much less than the power
brought in by the conduction and diffusion from the inner part of the plasma. The radiation losses essen- tially due to oxygen line radiation are responsible for ! of the power lost whereas the ] remaining are lost 3 by conduction and diffusion. This region is experi- mentally unexplored (except by spectroscopy) and the power balance is not very reliable. This is cape-
340 TFR Group / Plasma confinement in TFR tokamak
500
400
300
200
100
200
100
a P.w
%
% b
/__ 5
I ---------FI PKc 10 15 r(cm)
I t
2 0
Fig. 13. (a) Electron power balance as a function of the plas- ma radius (see caption of fig. 9 for symbols), (b) ion power balance as a function of the plasma radius (see caption of fig. 9 for symbols, PKc is the power transferred by classical ion conduction.
cially true for the ions (the ion temperature could be larger than the electron temperature and consequently the power transfer would be from the ions to the elec- trons) and we do not discuss this point here.
The power balance in ohmic heating is illustrated in fig. 13a-b for the electrons and for the ions respec- tively.
For the sake of exactness, it must be stressed tha t analysis of the ion power balance was made from the experimental knowledge of density and temperature profiles as mentioned earlier, deducing from them the experimental conduction coefficient K i = Kia + Kie. So an anomalous conduction coefficient Kia for the ions was introduced. However, a numerical simulation of the TFR plasma [10] carried out at Fontenay using roughly the classical ion conduction coefficient yields the experimental profiles, within the accuracy of the
measurements. This shows that a variation of the pro- files, even small, creates a large variation of the heat conduction coefficient, thus jeopardizing the conclu- sions of the ion energy balance. Nevertheless, the fact that the ion conduction coefficient increases with Ti indicates a non classical behaviour of the ions.
2.3. Neutral injection experiments
2.3.1. Description o f the injectors • The neutral injector is composed of two injection
lines located at diametrically opposed positions around the torus. One of them is presented in fig. 14. An injec- tion line is composed of five duopigatron sources emit. ring deuterium or hydrogen ions which are subse- quently neutralized before they enter the vacuum chamber where they are captured by the plasma. The maximum power extracted in real plasma experiments was PE = 1.62 MW and the injected neutral power in such conditions was PN = 570 kW, (50 A, 32.5 kV for the ten sources). In fig. 15 are plotted the main char- acteristics of the plasma versus time for a typical neu- tral injection experiment. The ion temperature increase
, i
¥ I II
L!
Fig. 14. Schematic design of a neutral injection on TFR I, II, Ill, IV, V: duopigatron ion sources (0 80 ram); (2) gas con for neutralization of the ion beam, (3) pumping chamber (of the neutraliT~tion gas), (4) toroTdal chamber.
Fig. 15. Characteristics of a neutral injection experiment versus time (deuterium plasma), a = 20 cm, B T = 53.6 kG, lp = 292 kA, PN = 470 kW. From top to bottom: (a) plasma current lp, (b) central ion temperature, (c) mean electron den- sity.
is 600 eV whereas the increase of the plasma density is attributed to the neutral gas influx from the sources. The increase of Ti takes place all over the plasma cross- section (see fig. 16). The electron temperature remains unchanged during injection although the power trans-
k.
x x x
"" " ' . , . , _
0 g •0 l's ,-,:.,. 2o-
Fig. 16. Ion temperature profile during and before a neutral injection experiment (same conditions as on fig. 15).
- , , , , , , , ,
5 0 0 ~T, ir~O) I ! . . . . . . . .
,00 = .v D ~/~/ 3o0 ~ [I 2OO
tO0
110 ~1
lO 20
Fig. 17. Ion temperature increase versus neutral power txans- mitted to the ions PNi normalized for a 1013 cm -s density.
mitted to electrons is expected to be of the order of the power transmitted to the ions. The explanation could be either an enhancement of the radiation loss (high Z impurities influx is also observed during injec- tion) or a strong dependence of the electron heat conduction (Ke ~ T~e/~ for the trapped electron modes) forbidding any increase of Te. The main positive point is that the ion temperature increase during injection scales proportionally to the neutral power transmitted to the ions PNi; no saturation takes place, which is very encouraging for the future (see fig. 17).
The study of the power balance leads to the same kind of conclusions as with ohmic heating alone (see section 2.2.2.),
3. The TFR 600 tokamak [ 11 ]
3.1. General design
As was said in section 1, the TFR 600 version is similar to TFR 400, except that the thick copper shell has been removed allowing a plasma radius of a = 24 cm. The stabilizing effect of the copper shell (against macroscopic motions of the discharge) is replaced by a dynamic feed-back stabilization system (~1 kHz) composed of coils producing a vertical field acting with the plasma current to provide equilibrium. Since the experiments started on TFR 600 (Autumn 1977) a large amount of work was devoted to this problem. Many technical problems are still to be solved and it was only possible to get discharges with moderate performances (plasma current less than 200 kA and plasma duration less than 200 ms). The main
342 TFR Group / Plasma confinement in TFR tokamak
reasons for these limitations are: (i) the equilibrium is very sensitive to the break-
down condition, (ii) the vacuum chamber is thin (skin time ~100
/as), (iii) the equilibrium coils are far from the plasma, (iv) the iron core of the transformer has a huge
effect. The last three points are very unfavorable in the
absence of copper shell. Work on the feedback is car- fled out in alternance with other experiments, which are described below.
3. 2. Wall conditioning [12]
The impurity content in TFR 400 was significant (Zef f = Y.nzZ~[ne ~ 5 - 8 , n z is the partial density of an impurity with a charge number Z): the oxygen content was some 1012 cm -3 and the molybdenum content some 1010 cm-a. In order to decrease undesir- able impurity effects on the plasma, the molybdenum limiter (Z = 42) was replaced by an inconel limiter (Z ~ 28, 58% Ni) and the wall was conditioned to eliminate the adsorbed oxygen and to reduce the me- tallic oxides. The method which has been used is a discharge cleaning of the chamber at a temperature of 300°C, in a way similar to what is proposed in ref. [13]. This discharge cleaning consists of relatively high density (~3 X 10 is cm -3) and low electron tem- perature (~<5 eV) tokamak discharges of 12 ms dura- tion with a repetition rate of 0.3 Hz. The working gas is hydrogen which is injected at a background pressure of 5 X 1013 cm -s. The loop voltage is ~40 V and the plasma current is Ip = 8 kA. The principle is to trans- form the metallic oxydes in the wall (MO) into metal (M) by the two step process:
M O + H ~ M O H , M O H + H - ~ M + O H ~ . (4)
This explains the necessity to have a high density hydrogen plasma. The low temperature avoids the dis- sociation of OH2 which is pumped as water vapor. The wall is kept at a high temperature so that it does not pump the OH2 molecules again. The best results are obtained when the water vapor production rate mea- sured at the pumping port (by a quadrupolar mass spectrometer) is maximum. In such conditions, it has been possible to reduce the oxygen content of the wall from 1 monolayer (2 X 10 Is at/cm 2) down to a
Table 2
Plasma characteristics: Clean cham- Contaminated t -- 100 ms ber 0.01 mo- chamber
nolayer 1 monolayer
Plasma density on axis 1014 5 x 101 z meanvalue (cm -3) 5.5 X 1013 2.5 X 1013
Electron temperature on 780 1720 axis mean value (eV) 320 600
Ion temperature on axis 650 800 mean value (eV) 300 450
Zef f (resistivity) 1.15 3.5
Global energy confine- ment time, ZE (ms) 20 10
Radiation losses on axis (W/cm 3) <0.1 1
Total power radiation losses (KW) 110 210
Total Joule power (kW) 380 460
Oxygen desorption during
one shot, Po 0.006 0.035 PHo
Carbon desorption during
one shot, P._~c 0.013 0.015 PHo
0.0006 monolayer. Comparison of the plasma param- eters for 200 kA, 40 kG discharges with an oxygen cover of a 0.01 and 1 monolayer respectively is given on table 2.
3.3. ICRH experiments [14]
A physical study of the magnetosonic wave damp- ing had been made in the plasma of TFR 400 at a low power level. This wave was launched into the plasma by an antenna fed by a power generator. A heating experiment is now carried out on TFR 600 with a 500 kW generator in the same conditions as on TFR 400. The physical mechanisms of wave absorption in the plasma are not discussed here. It must be men- tioned simply that the theory predicts that in a deu-
TFR Group/Plasma confinement in TFR tokamak 343
terium plasma which contains a hydrogen component (however small) the power o f the wave is transferred mainly to the low density protons and to the elec- trons. Subsequently the power is transferred to the deuterons via the charged particle collisions. The results which are reported here are extremely prelim- inary since they have been obtained after two weeks of work on TFR. Although they must be considered with the usual caution relative to preliminary results, the situation seems encouraging: during a 100 kW, 20 ms RF pulse, the electron density increased by less than 10%, the loop voltage is not affected and the enhancement of impuri ty radiation remains small. Simultaneously charge exchange measurements indi- cate a 150 to 200 eV increase of the ion temperature without any departure from maxwellian distr ibution of the ions up to 5 keV. Soft X.ray measurements indi- cate a comparable increase of Te in the central core of the plasma. These results refer to a discharge with the following characteristics:
lp = 130-150 kA ;BT = 4 0 - 4 2 kG ;a = 16 cm ; V = 3 V ; discharge durat ion : 150 ms ; ne(0) = 4 X 1013 c m -3 ; Te(0 ) = 1500 eV ; Ti(0 ) = 5 0 0 - 600 eV ; filling gas D2 ; amount of H2 "~ 5%.
The RF frequency was 64.4 MHz and the pulse duration 5 - 2 7 ms, usually 100 ms after plasma initia. tion. The maximum power delivered by the generator was 120 kW limited by breakdown in the antenna or transmission line.
4. T h e T F R 6 0 4 t o k a m a k [ 1 5 ]
The neutral injection experiments on T F R 600 have not yet started and it is difficult to envisage results on TFR 604. We simply summarize here the characteris- tics of the whole neutral injection program on TFR 400, TFR 600 and TFR 604, which are given in table 3. The main advantage o f the TFR 604 compared to the two previous versions is that a modificat ion o f 6 toroidal field coils allow injection of neutral beams at 43 ° with respect to the magnetic axis (on TFR 400 and 600 the injection is 82 ° and 75 ° respectively, to the magnetic axis which is rather unfavorable). This offers the possibility of working in a wider range o f conditions than before (larger range of densi ty for example and consequently o f ion temperature since Ti ~ PN/ne).
5. Conclusions
The TFR program which started five years ago appears to be a technological success, especially from the point of view of the torordal magnetic field pro- duction. The fly-wheel coupled to an al ternator is in that case a very reliable means of obtaining 60 kG - 1.5 s pulses o f magnetic field at 4 min intervals. More- over the oil-cooled Bitter coils have worked for more than 3 X 104 shots without any major failure. It is hoped that this situation will continue until 1982 so
Table 3 Characteristics of the neutral injection heating program on TFR (for explanation of symbols see caption of figs. 6 and 9)
Physical aims Parameters Plasma
TFR 400 Ion heating, scaling laws (1975-76) PN > Pei
TFR 600 Ion energy balance scaling laws ( 1 9 7 8 - 8 0 ) PN = P ~
TFR 604 ( 1 9 8 0 - 8 2 )
Equil~rium and stability of the plasma at high injected power, scaling laws, plasma wall inter- action PN >> P~2
400 kA, 60 kG a = 20 cm PN = 0.5 MW PN +P~ = 1.2 MW
600 kA, 60 kG a = 23 cm PN = 1.2-1.5 MW Ps2 + PN = 2.3 MW
600 kA, 60 kG a = 23 cm PN = 4 MW P N + P ~ =5 MW
~" -- 8 X 1013 c~-3 T e + T i= 1.6 keV rE = 15-20 ms
If rE = 15-20 ms and ~" = 8 X 1013 cm-3
then T e + T i = 2.1 keV
If r E = 15-20 ms and ~-= 8 × 1013 cm -3 then_T e + T i = 4.6 keV and Ti(0) = 4-5 keV
344 TFR Group / Plasma confinement in TFR tokamak
that the whole program TFR 400, TFR 600 and TFR 604 can be carried out.
The first version TFR 400 (400 kA, 60 kG) was studied from 1973 to 1976. The caracteristics of the plasma in ohmic heating conditions were ~e ~ 6 X 10 la cm -3, Te(0 ) ~ 2.5 keV, Ti(0 ) ~ 1 keV and r~ 20 ms. Neutral injection heating increased significantly the ion temperature (T i doubled in low density dis- charges). A study of the scaling of the main param. eters of the plasma showed that the density and the electron and ion temperatures increase with the plas- ma current as well as with the toroidal magnetic field whereas the energy containment time is appro.ximately constant or even decreases. The power balance has been established in a cross.section of the plasma. Three regions are distinguished:
- T h e central core where MHD instabilities (disrup- tions) and high Z impurity radiations are responsible for the electron losses while the ions are suspected to have a non classical behaviour (anomalous heat con- duction).
- A n intermediate region where turbulent heat con- duction occures for the electrons and where the ions behave as in the central core.
- T h e periphery where a part of the electron power is radiated by low Z impurities but where the mea- surements are not sufficient to allow a precise balance. Finally it must be emphasized that the linear growth of the ion temperature with the neutral injected power is very encouraging.
On the new version, TFR 600 (600 kA, 60 kG) the main problem remains the dynamic feedback control of the discharge. It was possible to achieve pure plas- ma conditions thanks to a new conditioning of the wall and to the use of a low Z limiter thus reducing the radiation power losses. Preliminary experiments
on ICRH are very promising since a 100 kW RF pulse seems to produce ions and electrons heating with a small change of density. On this machine a 1.2 MW neutral injection heating and a 100 kW cluster injec- tion will be studied in the next future.
Finally TFR 604 (600 kA, 60 kG + 4 MW neutral injection) which is scheduled for 1980-1982 could allow to obtain a central ion temperature in the range of 4 to 5 keV.
References
[1] Equipe TFR, NucL Fusion 18 (1978) 647. [2] Equipe TFR, in: Vlth Europ. Conf. Controlled Fusion
and Plasma Physics, Moscow, USSR (1973). [3] M. Huguet, P.H. Rebut and A. Torossian, in: 4th Int.
Conf. Magnet Technology, Brookhaven (1972) p. 60. [4] P.H. Rebut and A. Torossian, in: 4th Int. Conf. Magnet
Technology, Brookhaven (1972) p. 56. [5] TFR Group, in: 6th Conf. Plasma Physics and ControUed
Nuclear Fusion Research I (1976) p. 35. [6] L.A. Axtsimovitch, Nucl. Fusion 2 (1972) 215. [7] B.B. Kadomtsev and O.P. Pogutse, in: Reviews of Plasma
