A REVIEW OF RF HEATING by Miklos Porkolab Preprint PFC/JA-78-3 Plasma Research Report PRR 78/19 May 1978 Invited lecture presented at ''International School of Plasma Physics, Course on Theory of Magnetically Confined Plasmas", Varenna, Italy, Sept. 1-10, 1977. ABSTRACT The funadmental aspects of RF heating theory is reviewed and a summary of recent experimental results are given. All frequency regimes are covered, from very low frequency magnetic pumping to electron cyclotron resonance heating. I A Review of RF Heating P + P n1 a + AP $ - P (1) loss Miklos Porkolab where P represents ohmic heating, P represents a-particle heating, APs Department of Physics and Plasma Fusion Center Massachusetts Institute of Technology represents supplementary heating, and Cambridge, Massachusetts 02139, USA Ploss is due to synchrotron radiation, We review briefly the present bremsstrahlung, and heat transport to status of the theory and the experi- the walls and the limiter. mental results of heating toroidal that in order to contain a-particles, plasmas (mainly tokamaks) by radio- we need approximately 3.5 Mamperes of frequency (rf) power. ohmic-heating current. We shall dis- We note Assuming that cuss a '-lmber of different frequency this is provided, in most devices regimes beginning with the electron with Z ff cyclotron frequency (i.e., f- 100 tary heating will be necessary to GHz) and ending with very-low achieve ignition conditions. frequencies (i.e. f w.1-10 KHz). at present the forerunner is neutral Before we discuss the details of rf beam heating, in-reactor-size plasmas heating, let us briefly point out beam energies of the range c the importance of supplementary 500 key will be necessary. .heating of fusion plasmas in order sent such beams do not exists, and to reach ignition temperatures. In 1 some form of supplemenWhile 300- 'At pre- the low efficiency of neutralization tokamaks initially ohmic heating will above 150 keV dictates the 'use (and provide a relatively hot (T e Ti 2 keV) target plasma. However,- as development) of negative ion beams. An alternative approach would be to the temperature rises, the resisti-3/2 In addivity decreases as T/. learn how to use the large amounts of tion, losses due to radiation in- high temperatures. crease as z n2T . Hence, at some true due to the following additional point radiative losses overcome reasons: RV power available to heat plasmas to This is especially (1) With the exception of ohmic heating, and the attainment of a few clean tokamaks, in most devices ignition cbnditions by ohmic heating alone is doubtful. Furthermore, the Z > 1. However, since radiative losses are especially sensitive to ohmic current cannot be increased by Zeff, impurities will have to be arbitrary amounts or MUD stability will be violated. I, greatly reduced in reactor grade A simple relation plasmas so that Zeff < 2, (and between typical losses and power Z input is given by the following However, the effectiveness of ohmic relationship: heating in. "clean" plasmas 1 is highly desirable) [1). 2 above T w 2 keV is greatly reduced. "conventional" large (low-density) The effectiveness of supplementary device,. an intermediate device, and a heating can be domonstrated by the high density, compact device ("igni- simple formula (valid for At > TE) tor"). The expected magnetic fields and machine parameters are also shown; AP- V AKnT) 2 TE (2) including the necessary temperature where IE is the energy confinement changes which the "supplementary" time, AP is the supplementary heating heating system must provide in order power, V is the plasma volume, and to reach ignition temperatures. Here A(nT) is the additional rise Inplasmawe assumed that eventually the contri2 energy. If we take T - C 0na , where bution of ohmic heating becomes neglic0 is a constant (as is observed in gible as compared with the supplemen- present day. experiments) then tary heating power (with the exception of the last case AP c (AT)R C , where C1 . "ignitor" [3)). While an intermediate the to l'arge system will require 55-125 MW of net additional heating power, a Thus, we see small system can "ignite" with only a 2 37r /C0 , and R is machine major radius. (3) which is called that the smaller the major radius, few MW of supplementary power. Of the less supplementary power one needs course, in a smaller device materials to attain ignition temperatures. damage due to neutron bombardment may Implicit in these considerations is be unacceptably high for reactor pur- that C 0 is independent of T. Although poses. Nevertheless, "ignitors" at present this appears to be consis- should be considered as relatively tent with experimental observations, "cheap" research devices where a-par- in future, hotter plasmas where trap- ticle heating (for example) and ped particles will be dominant this reactor-grade plasmas could be studied. may no longer be so, and a deteriora- It should be remembered that these tion of TE .may be expected. powers are the power absorbed by the Let us now examine in a few i plasma, and thus the source power will cases of interest how effective likely be twice as much as that shown supplementary heating is. in Table 1. In parti- Since the total heating cular,.we shall consider three dif- pulse length is a.few seconds, for all ferent sized machines and in each practical purposes we must consider case we shall assume that the energy CW power. 2 confinement time scales as TE - na , A typical rf heating system is .with the constant C obtained from -represented schematically in Fig. 1. the best present date results, namely It consists of an rf source, a transAlcator A [2]. Table 1 summarizes mission line, a vacuum (and dc) break, the expected confinement times for a and some sort of antenna structure 3 either inside or fiush mounted in a port. Finally, the radiated power comments regarding their main features. This is given in.Table 3, in order of must be transported toward the plasma' decreasing frequencies, all the way or electrostatic waves, and finally from electron-cyclotron resonance heating to very low frequency magnetic this power must be dissipated in the pumping. interior by means of electromagnetic form of thermal energy. Of course, if too much power is dissipated near We should note that while ECRH is the most recent experimental entry into tokamak rf-heating (at the plasma periphery, it will quickly B = 20 kG) there are some be lost and the efficiency of bulk new ideas for very low-frequency heating will be low. Thus, assuming that the total efficiency of absorp- heating [4). Let us now discuss the main theoretical aspects and the tion is at. best 70% and that at least status of the experiments in each of We note that 20% of the power is lost in the these frequency regimes. transmission line at the source we in the theoretical discussions we would need at least double the power shall present only linear mechanisms. shown in Table 1. It should also be In many cases nonlinear effects may realized that we cannot expect more become important at high pump powers, than 35-40% electric efficiency for the total rf source, and hence the and nonlinear effects such as trap- ping, stochasticity, and parametric required "plug" power may be four or instabilities may have to be taken A recent review of -five times that shown in Table 1. One important aspect of plasma parametric phenomena has been given rf interaction, which is often not by the present author elsewhere and mentioned, is that in addition to we shall not discuss them here into account. [51. actual bulk heating, rf power may be used for controlling the plasma properties. Some of these roles are listed in Table 2, and since most are self-explanatory (and some are speculative), we shall not discuss them here any further. Clearly, 3) Electron Cyclotron Resonance Heating A. Theoretical Considerations The linear theory bf electron- cyclotron resonance heating applicable to tokamaks has been examined recently The considerably more research work is by a number of authors [6-8]. needed to determine which of these renewed interest in this high fre- techniques hnvc potential for prac- quency tegime stems from new developments in the field of high-frequency tical applications. technology,namely the development of nterest in RF Heating gyrotrons [9,10). Thus, it appears Let us now list various frequen- that cw sources with power levels of cies of interest, and also make some the order of P = 0.1-0.2 MW, f - 28 Megimes of 4 GHz, and pulsed sources with P v 70 kW, f = 65 Gilz are available now [11] mode of propagation is useful and cw sources with frequencies of absorption coefficient at wu f = 100 CHz and P = 100-200 kW expec- is given by (using v2 ted to be available within three R 2 B W only when w < w pe years [12]. ce . The T/m) v2 2 . (6) The linear absorption mechanisms in this frequency regime are due to For example, for f - fce either finite values of k,, and cyclo- GHz one finds that 2nn > 1 if tron damping, or due to the inhomo- Te > 0.4 keV. geneous magnetic field. is efficient only in hot plasmas. summarize Let us now theoretical. predictions obtained by this mode of propagation is not Defining effective in heating the plasma. Extra-Ordinary Mode of Propaga- b. T - exp[-2inr) Thus, absorption It is also found that at w = 2wce the main results of the Antonsen and Mnnheimer [6]. 100 tion (3) In this case E0 B, k0 B, and we as the transmission coefficient, are interested in absorption A = (1-T) is the absorption coeffi- near cient. Furthermore, the magnetic = we 2 2 w ce' WUH (where , +w2 is the upper- field gradient scale length is de- hybrid frequency). fined as accessibility is prevented by a Direct cut-off layer in front of -the VR (4) B/RD upper-hybrid layer when the wave propagates from the low magnetic where the spatial derivative is field side. taken along the major radius, R. is given by The cut-off layer These relationships predict efficient absorption of the microwave beam in one passage when . W w W+ 27m > 1. (5) ce- 1 2 + +7 4W2 211 C m2 ce The wave may propagate to the upper-hybrid layer by tunneling The results of the theoretical predictions are as follows; a. Ordinary mode of propagation In this case E0 | . IB, k2.tB, where IR the external magnetic field. Accessibility requires that b0 > W , and hence this from- the low magnetic field side *2 U W 2 . Since in toka- maks we = wpC. accessibility exists when the wave propagates from the high magnetic field side [13] or when w = 2wce Thus, we see that even if the 5 wave is launched from the outside in use w = 2w . In addition, one may ce the equitorial plane, accessibility want to beam the microwaves at some may be achieved by scattering around critical angle to optimize the k,, the torus (between the chamber wall spectrum for the 0-mode of propagation and the plasma column) and penetra- (although a spread in the k,, spectrum ting from the high magnetic field will automatically be introduced due side. to the finite launching horn-size). i. W- W C~e One finds then that if k,, strong absorption by cyclotron damping takes place if w On date on ECR! heating of tokamaks have been obtained in the TM-3 tokamak In these experiments, the [11). freqincies used were in the range + 0. f w - 2wce 35-65 CHz, and the power was injected by open ended waveguides. Absorption can be efficient even if k, Experimental Results The only experimental results to Wce does not take place since IE ii. wc . B. 0, 0 efficient the other hand, if k,, absorption at w # 0 and 21m > 1 where T1 = Rg W At low densities (n < 10 13 -3 cm ) most of the energy ended up in a high energy tail. vte (8) (n c c2 C At high densities 1013 cm ) efficient bulk electron heating was observed at both Again, this absorption mechanism is efficient if the plasma is relatively hot (i.e., T ; 0.5 keV). iii. W w - w and at w m w w = 2w . To date heating has not been observed. Typical pulse lengths were At Wu = 1 msec and maximum power levels at 70 GHz uh In this case heating occurs by were P = 60 kW, which raised the absorption of the Bernstein wave peak electron temperature in the cen- which is a consequence of linear ter of the discharge by as much as mode-conversion of the extraordinary ATe = 150 eV. mode at a) Wuh [14). The maximum absorption Alternatively, efficiency for bulk electron heating Bernstein waves may be generated by in these experiments was estimated to parametric decay processes also [15). be 30%. There has been experimental Accessibility is best by propagation evidence that the electron energy from the high magnetic field side, confinement increased with electron or by tunneling if W2 < temperature, at least in the range pe , or by ceo the so-called z-hole effect [16]. From these considerations one T u 100 to 500 eV. Whether these trends would remain at temperatures may conclude that the best strategy above a few keV, remains to be seen. is to use the 0-mode in tokamaks In summary, while the mechanism of with WPC < Wce, and when w absorption was not determined, the > we 6 initial. ECRII results are encouraging, W - th (0) and k_ + and we expect to see a big increase effects are included, in the research activities in this regime. Whcthcr enough CW power can If hot plasma the slow-wave .converts into an ion plasma wave before the resonance layer, i.e. where be developed to heat large tokamaks to ignition by the use of ECRH alone remains to be seen. (11) w = 3k Lyti' Nevertheless, one can consider ECRH heating to and this ion wave then propagates provide power for radialprofile outward. modification in order to improve MID absorbed by "ion-landau" damping in stability. the inhomogeneous magnetic field The ion wave is either [24,25) or is soon converted once 4. Lower-II brid Range of Frequencies more into an ion-Bernstein wave which A. Theory then propagates back toward the center in the lower-hybrid range of and is absorbed by harmonic-ion-cy- frequencies we may consider heating clotron damping at the next exact at the lower-hybrid resonance fre- cyclotron harmonic layer [13). The quency (or slightly above if tempera- accessibility conditions is roughly ture effects are included), Wh mwh pg(l+w2pe 1w2ce)-1 [17) 2 n i> 1 + (9) 22 2 W/ W pe ce 'Res (12) ck,,/w is the index of by the process of mode-conversion where n,, into ion plasma (Bernstein) waves refraction, and the right-hand-side [13,17] or we may consider heating by is to be evaluated near the resonant electron Landau damping if w > wth [18,19,20). In both cases one layer. assumes that the so-called slow-wave accessibility condition must be When such a layer is not present in the plasma, a more exact satisfied, namely ( ++ k-2 I k (10) cJ W/w =wn,,y 2 2 /l +nil(y - 1) (13) is excited at the surface by a slowwave structure, and the wave propa- where y 2 gates inward along "resonance cones" (which are trajectories of the group is especially important when one velocity in the 2, r plane) [21-23). w /wcewci. Equation (13) considers an inhomogeneous plasma, or when one desires to heat by elec- Since the k., spectrum is fixed by tron Landau damping. the slow-wave structure, kL = k r increases as the wave propagates case, one relies on the fact that if radially inward until at r - 0, in the outer layers of the plasma In the latter the kl, spectrum is "tailored" so that 7 (14a) c/vte >> 3p,, B. Summary of Lower-Hybrid Experimental Results and near the center In recent ypars there have been several large-scale lower hybrid (14b) C/vte experiments performed in tokamaks, and some of these investigations -are then absorption occurs near the cen2 ter (where v2 still going on at the multi - 100 KW (T /m )). I* fact, level [30-32]. in a hot plasma (i.e. Te The great interest 10 keV) it in this regime arises from the parti- may be difficult to satisfy Eqs. 13 cularly advantageous technological and 14a simultaneously, and at these aspects of LURF heating, namely that temperatures we may have to abandon one can use phased-arrays of wave- the slow wave as a useful mean s to guides mounted in ports as sources of deliver the rf energy to the core of the plasma column. However, one the microwave power [33,34). It has been shown recently that such a could then use the whistler mode "grill" structure behaves as a slow- (fast wave) and rely on electr on wave structure [35], and the theore- Landau damping for absorption [25). tical work of Brambilla [34) allows Alternatively, sufficiently strong one to design grills for different rf fields could flatten the distri- machine parameters (the "Brambilla bution function and allow pene tration code"). [19,20). sources are available, or can be Because the path of wave- pene- tration is long, (typically L = (mi/m e) a, where a is t In addition, ample power built upon demand, with frequencies up to 10 GEz. e In most cases one would use f v 1-5 GHz for the frequency. plasma radius and Lz is the path The most recent experiments at the length along the magnetic fIel d) lower hybrid frequency were carried nonlinear effects in the outer plasma out on the Princeton ATC tokamak at lyers are particularly dangerc us, f especially parametric instabil ities, 0,11, were used as a phase-array solition formation, and scattc ring system [30,31]. by drift wave fluctuations [2 F,28]. Wega experiment loops phased appropriately are being used to inject Again, some of these phenomen i were = 800 MHz where 2 waveguides phased At Grenoble in the paper and we shall not discus S them the rf power at 500 MHz [32). In preparation at present is the general here. We note, however, that in Atomic Doublet 11 experiment where a addition to parametric effect s slotted slow wave structure will be [27,28) stochastic rf fields could used to launch waves at w > w also play an important role in non- heating by Landau damping will be linear ion.heating [29). attempted. At MIT, preparations are discussed in our recent reviec , and 8 under way for a 100 kW split wave- represent T guide experiment on Alcator A. From the past experiments the showed little or no heating, the OVII on CV lines showed following conclusions may be drawn: good heating, although typi- a. -The phased array grill works cally less than the perpen- as predicted by the Bram- dicular charge-exchange billa theory; namely, while measurements indicated (i.e. AT w 50-100 eV). in a single waveguide most of the power is reflected, in at least 2 waveguides driven out of phase (0,180) in Wega in some cases a the reflection was reduced density increase of 50% (or to about ten percent, as more) was observed. In f. The total energy absorbed particular, no tuning elements were needed, which is by the bulk of ions was difficult to estimate; how- an important technological ever under reasonable advantage of this type of assumptions 15-20% could be coupling system. accounted for by bulk-ion In the future four (or more) waveguide grills will be used. heating. g. In ATC a definite threshold Perpendicular charge-ex- power and density for heating change results are similar was observed [31]. in both Wega and ATC; namely, upon application of was decreased, the threshold P power increased from P = 10 100 kW, typically ATi c. e. While in ATC the density increase did not exceed 10%, predicted by theory. b. near the edge) 100-150 eV increase In particular, as the density kW to P > 100 kW, until was observed. neither parametric decay While in ATC parallel (monitored by a probe at charge-exchange measurements the plasma edge) nor plasma were inconclusive, in Wega heating were observed. gqod agreement with the a theoretical analysis it perpendicular measurements was concluded that in ATC were obtained. the heating took place at d. Doppler-broadening measure- (r/a) v 0.3-0.5 when ments of OVIT, CV and CIV 1 Z wo/w impurities showed similar heating was studied only when 1 Z Wo/wlh Z 1.3. results in both ATC and Wega. In particular, while the CIV measurements (which < 2. In Wega h. More recently, in Wega heating of electrons was From 9 also observed [36]. It is 2 not clear, however, whether such an electron heating was due to impurities or 2A 2 N (M - (15) cil where direct heating by the pump wave, or by the parametric VA - AB/[47rn(m e4n ici /W p)c p1 (W decay waves. We also note that in addition to bulk is the Alfven speed, R is the major ion heating an energetic perpendicu- radius, a is the minor radius, and lar ion tail was also obsekved, with a life-time of 50-100 psec. It was V a P + jM/21 concluded that this ion tail was produced on the surface of the plasma and where p, M, and N are integers. Similarly, an- increase in soft and In paz,.icular, v is the "perpendi- hard x-rays were also observed. cular" wave number consisting of In summary, while the initial the .radial mode number p and the results are encouraging a clear poloidal mode-number M, and N/R = k,, understanding of the physics of LHRH is the toroidal mode number. The is not yet available. In particular, important prediction of Eq. 15 is the fate of the rf power-flow is not that fast-wave propagation is understood, and clearly more experi- possible only if the density and mental work is required. radius is such that V i , N 0, and > Eq. 15 becomes 5. Ion Cyclotron Range of Frequencies (CRF) na2> 2 2 2 B F A. Theory 5xl m 2 si e 2(16) Heating near the ion-cyclotron frequency may be done via the slow wave near w L wei or the fast wave near w V tci [37-40). Thus, fast wave propagation demands sufficiently large and dense devices, In tokamaks especially if w < w i is desirable. the most important regime is w = 2 Wei The heating rates are given by the and heating is achieved by the small following expressions [42): left-hand component of the fast wave a. W (magneio-acoustic wave). The slow- wave is expected to mode-convert near the tokamak surface, and it may W0 i 1 4w 2 i 0 cc2 a E+ 2 (17a) be used mainly in heating stellara- tors. The dispersion relation for fast waves is given by the following expreenion [40,41): We notice that since in a single ionspecies plasma E + 0 at w - ci , the power absorbed, P 0. However, in a two-ion species plasma this is no * ii 10 longer so. the RF voltage (namely, RM depends on b. w w 2w Q, and Imax is limited by voltage breakdown). P i2 R I1+1 2 ..R+E |2 16iw r where E+ Vr 0 are excited it is desirable to "lock onto" one mode. and "stay with it" E+ is'the gradient during the rf pulse. Ilowever, since 81nT /B2 is the ion of E+, and 0 "beta". (17b) Thus, if cavity modes (19) Q - (wo/Aw) - 2no/An We note that while Eq. 17b predicts weak absorption in present tokamaks, in future high density, hot where (An/no) is the fractional den- tokamaks the absorption is expected sity change, we see that as the den- to be efficient. sity changes in time one jumps from c. one mod(. to the next, and a feedback Ileatin1, jylectron-Landau Dapng system may be necessary to track the This has been calculated by This is a rather difficult Stix [40,41), and again, in hot toka- mode. maks it is expected to be an effi- technical task if many modes are cient absorption process, while in present in the system. machines with T Z 1 kev it is a a reactor size plasma hundreds of An advantage of rather weak process. In fact, in modes may be excited as it will no this process is that it allows opera- longer be feasible to track modes tion at w < w i, where it can be [41). cofisidered a type of "Alfven wave the effects of the rotational trans- One may have to consider then form also [43]. heating" (rather than ICRF). e. Impurities d. Cavity Modes Recently a number of authors Once the fast wave propagates, It is possible to set-up standing discussed the effects of even a few waves around the major circumference percent of impurities present in the of the torus, the condition being plasma upon the fast wave propagation kJZ/271 integer. This allows set- [38,44,45]. For example, a few per- ting up modes with relatively large cent of hydrogen present in a deu- values of the "quality factor", Q. terium plasma presents an 11 Since the power absorbed depends on the coil impedance 'Ri, and is given funda- mental resonance (w = wci(H+)) when w 2w ciD +) which may provide a mechanism of efficient power absorp- by In terms of the current I, by tion by the H+ minority rather than SRH (18) the D majority species. Absorption by H+ leads to a large and fast H it turns out that relatively large -tail which will have to be confined values of Q allow effipient power and its energy transmitted to D+ by absorption for acceptable values of slow collisions. Additional problems 11 may be created by setting up an ion- w L 2wci(D ) heating experi- ion hybrid resonance layer near the plasma edge which may lead to mode ment was carried out [47]. Coupling experiments carried c. conversion and large damping near the out ot low powers in the edge. French TFR device showed Additional enhanced absorption near the ion-ion hybrid layer may lower-Q values (by one to also be due to large gradients in E two orders of magnitude) between the cyclotron and the hybrid than predicted by theories layers [45). which ignored impurity effects [48]. Again, the importance of small amounts B. Ex erimental Results Experimental results on ICRF of impurities were blamed heating of tokamaks have been for the results, especially obtained .n the past in a number of in D2 plasmas. devices, and further experiments are d. The Princeton ATC experi- planned on the TFR device in France ments showed the best ion and the PLT device in Princeton, USA. heating to date using ICRF In particular, the following is a power near w brief summary of the main features In this experiment good bulk of these experiments. ion heating was obtained. a. On the Princeton ST device In addition, a significant the ICRF experiment showed ion tail was also produced. ion heating, but there were However, it was not deter- also impurity problems mined experimentally whether present [46]. this ion tail consisted of In particu- lar, the impurity level deuterium or hydrogen. increased with injected maximum input energy was P Z 2 kJ which was compar- energy. When the total external rf energy exceeded the plasma thermal energy, namely b. 2w ci(D+) [49]. tPRF > nkT : 200 The able to the thermal energy content of the plasma. At higher energies the m = 2 Joules, disruption occurred MHD mode was intensified and [46J. no experiments were carried These results were blamed on bad banana orbits out. and fast particles hitting efficiency was about 40%, the wall.. and the energy absorbed by On the Soviet TM-l-VCHI device efficient heating of an ions was comparable to that energetic H tail wan detec- heating. red (rather than heating of anisms in these experiments D ) when an The 'maximum heating obtained by neutral-beam The heating mech- 2 were not determined, and Q-measure- where 0 = C 8'n T /B is the total ee ± ments and their comparison with beta of electrons, theory have not yet been published. E In summary, the ATC ICRF experiments (We note that Eq. (21) may be used are most encouraging and further both at w < wci and wi< improvements can be expected in the Since the damping is weak, high Q- forthcoming PLT experiments. values may be established for one or ~0 2T /m and is the electric field intensity. < 2 wcV) two cavity modes which may then be 6. Alfven Wavelleating tracked by a feedback system. A. particular, by lowering the frequency Theory This is the regime of low the number of modes in reactor-size frequencies, such that w < w , typically 300 KIz - 10 MHz in toka- maks. In plasmas can be lowered by orders of magnitude from the ICRF regime. Two types of modes are of In the case of the shear Alfven interest, namely the compressional Alfven wave (e k vA) and the shear Alfven wave (w k,.vA). wave one would use a coil structuresuch that the resonance The com- pressional Alfven wave with a small w = k.,(r )v !r B +kZB /4np (22) component of k,, may be described by and it Eq. .(15), may be used to heat would be achieved at some radial electrons by Landau and transit-time position, r. Then a mode conversion damping in reactor size plasmas [41). into short wavelength electroqtatic The frequency to be used is expected waves may take place [50) or MHD to be w rw ci/10, so that direct multi-spectral absorption may take acceleration of 08+ impurities may place The critical density be avoided. [511, both leading to absorption of the energy. In particular, in the mode-conversion model [50) the con- may be set at verted wave would have to travel na2 ; 5x10 7 cm-. (20) around the torus once or twice (similarly to the lower-hybrid wave) For example, a plasma column with and absorption by collisional, 3 n ; 5x10. , a = 100 cm would satisfy Landau, or parametric effects may this condition (such as the forth- coming TFTR device at PPL). The damping rate and power absorption is given by Stix as follows [40,41'): 16F [ J I 2 '60c 2 2 xe ;-1 take place. k.tej (21) B. Experimental Results There are to date two experiments where shear-Alfven wave heating have been studied: (a) In proto-cleo at Wisconsin a doubling of T and T was observed when a power of P m 10 kW was applied in an initial 13 low density plasma with temperature T v10 eV [52]. of f 100 Klz are contemplated. However, after the heating strong "pump-out" was. obser- ved in this device. B. (b) In the Experiment The only results to date have Heliotron-D torsatron at Kyoto the *been obtained by using ion TTMP on following results were obtained [53]: the Petula tokamak at Grenoble [55]. applying a power of P = 400 kW, This experiment confirms the preTe changed from 150 eV to 300 eV and T dictions of theory, namely good bulk changed from 20 eV to 5Q eV. A maxi- ion heating has been obtained mum heating efficiency of about 30% was observed. The physical mechanisi responsible for the heating remains unclear. (AT - AT,, - T /2) with (AE/B) 0.015 without major disruption or impurity problems. In particular, the results were in agreement with theoretical predictions, and scaling 7. Transit Time Magnetic Pumping A. Theory should improve with larger machine size. This technique has been studied years ago extensively by Canobbio at 8. Very Low Frequency Heating Grenoble [54]. The electron TTMP is similar to the compressional fast This case was studied originally by .Schlhter in the 1950's and the wave heating proposed by Stix more main absorption mechanism considered recently, namely toroidal eigen-modes was due to ion-ion collisions [56]. are set up and absorption takes placc The absorption-rate obtained is given by modulation of the magnetic field. by The power absorption is given by v P/vol = 3 2 V n-It Q Vh -8B J - nT . (23) W (24) 22 VIi where (At/B) is the modulation depth. where R is the major radius, and B The advantage of this heating is the modulated magnetic field technique is due to the possibility intensity. of using a few kHz for the frequency Electrons are preferen- tially heated since w/k, = vte is so that the rf coils could be placed set up by the coils. outside the vacuum chamber. Frequencies However, contemplated are typically f = 1-10 since we expect that w Z Vi, we see Mlz. that in order to obtain absorption On the other hand, ion TTMP relies on forced (evanescent) rf rates comparable to the ion collision fields such that w/k, v vti, and frequency one needs AB/B * 1, which since vti is energetically not practical. scent. heated, VA the mode is evane- Here ions are preferentially More recently Canobbio proposed and frequencies of the order that the efficiency of this type of 14 magnetic pumping could be improved by pursued. coupling to a low-frequency resonance, for example that due to the Acknowledftements VB drift, and in that case consider- This research was supported by able improvement over Eq. 24 could be the US Department of Energy. obtained [4]. This work is still in progress and preliminary indications are that significant heating (i.e. References AT/T = 1) may be obtained by modula- 1. D. Meade, Nucl. Fusion 14, 289 tion depths of AB/B = .1-0.2. (1974). 2. 9. Summnry and Comments M. Gaudreau et al., Phys. Rev. Lett. 39, 1266 (1977) It is clear that the physics of rf heating is complicated, and 3. B. Coppi, Comments Plasma Phys. Cont. Fusion, 3, 47 (1977) there are still too few experimental results available. However, it appears that in the available experi- ments efficiencies of 20-40% have 4. E. Canobbio, Proc. Conf. on been obtained (where by efficiency Plasma Physics and Controlled we mean ri = A(nT)/P(incident). Nuclear Fusion Research, Berch- Furthermore, in many cases improved tesgaden, efficiency is expected with larger Vol. 3, p. 19 W. Germany, Oct. 1976, 5. M. Porkolab, Nucl. Fusion 18, 367 machine size and hotter target plasma, and theoretically efficien- (1978) cies of n = 60-70% appear achievable. 6. T.M. Antonsen, Jr. and W.M. ManThe efficiency of rf power generation heimer, "Absorption and Trans- is good, typically 30-70%, and thus formation of Electromagnetic Waves technologically speaking rf heating in the Vicinity of Electron Cyclotron Harmonics", NRL preprint, to' is attractive for the purposes of reactor-grade plasma heating. More experiments are clearly needed in be published in Phys. Fluids 7. 0. Eldridge, W. Namkung and order to learn more about the physics A.C. England, Oak Ridge National of wave penetration, heating and Laboratory, ORNL/TM-6052, Novem- impurity problems. ber, 1977 Finally, more materials research will be necessary 8. I. Fidone, G. Granata, R.L. to solve the problem of coil Meyer and C. Ramponi, Association shielding at low frequencies. Since Euratom Report EUR-CEA-FC-912 at present no clear "winner" is (1977) apparent, experiments in all froquency regimen should be vigorously 9. N.I. Zaytsev, T.B. Pankratova, M.I. Petelin, and V.A. Flyagin, Radio Eng. Electron Phys. 19 -- 15 103 (1974) 10. 21. J.L. Hirshfield and V.L. Cranastien, IEEE Transactions on Phys. Fluids 14, 857 (1971). 22. Microwave Theory and Techniques, MIT-25, 513 (1977). 11. 23. 13. 24. 15. Heating in Toroidal Devices", H. Jory, Varian Associates, Varenna, Italy, September 1976 privat communication (Editrice Compositori, Bologna, T.H. Stix, Phys. Rev. Lett 15, Italy, 1976, p. 92). 26. I. Fidone and G. Granata, Phys. Symposium on Plasma R.L. Berger, F.W. Perkins, . and F. Trojon, Princeton Plasma Fluids 18, 1685 (1973) Physics Laboratory Report PPPL- M. Porkolab, Nucl. Fusion 12, 1366 (August, 1977) 27. V. Cinzburg, The Propagation of M. Porkolab, Phys. Fluids 20, 2058 (1977) 28. R.L. Berger, L. Chen, P.K. Kaw, mas (Pergamon, New York, 1970) and F.W. Perkins, Phys. Fluids. p. 385 20, 1864 (1977) V.E. Golant, Zh. Tekh. Fiz. 41,' 29. 2492 (197]) [Sov. Phys. Tech. C.F.F. Karney and A. 16,1980 (1972)) 18. P.M. Bellan and M. Porkolab, Phys. RF Power, Ph.D. Thesis, E.E. Department, MIT, May, 30. of Technology Report PRR 76/23, July, 1976. Rev. Lett. 39, 550 (1977) . Heating of Ions in a Tokamak by 19. A. Bers, F.W. Chambers and N.J. Fisch, Massachusetts Institute Bers, Phys. Also, C.F.F. Karney, Stochastic Phys. Fluids 19, 995 (1976) 1977. S. Bernabei, et al., in "Third Symposium on Plasma Heating in Toroidal Devices", Varenna, Also, A. Bers in "Third Symposium on Plasma Italy, 1976 [Editrice Composi- Heating in Toroidal Devices", tori, Bologna, Italy, p. 68, Varenna, Italy, Sept. 1976 1976) [Editrice Compositori, Bologna, 20. S. Bernabei and I. Fidone, in Plasma Phys. 2, 212 (1976) klectromagnetic Waves in Plas- 17. (1976) 'Third 329 (1972) 16. I. Fidone, Phys. Fluids 19, 334 Also, Alikaev et al., Sov. Phys. 78 (1965) 14. P.M. Bellan and M. Porkolab, Phys. Fluids 17, 1592 (1974). Fusion and Plasma P1aysics, Lausanne, September 1975, p. 144 25. 12. R.J. Briggs and R.R. Parker, Phys. Rev. Lett. 29, 852 (1972). V.V. Allknev et al., 7--th European Conference on Controlled R.K. Fisher and R.W. Could, 31. M. Porkolab, S. Bernabei, W.M. Italy, 1976, p. 99) Hooke, R.W. Motley and T. Naga- F.J. Paoloni, R.W. Motley, W.M. shima, Phys. Rev. Lett. 38, 230 Hooke and S. Bernabei, Phys. Rev. Lett. 17, 1081 (1977). (1977) 32. P. Blanc, et al., Proc. of Conf. 16 on Plasma Physics and Controlled Devices, Varenna, Italy, 1976 Nuclear Fusion Research, Berch- (Editrice Compositori, Bologna, tesgaden, W. Germany, Oct. 1976, Italy, 1976, p. 50) Vol. III, p. 59. 43. C. Cattanci and R. Croci, Nucl. 33. P. Lallia, in Proceedings of the Second Topical Conference on 34. 35. 44. Physics Laboratory Report PPPL Texas, 1974 1336 (April, 1977) M. Brambilla, Nuci. Fusion 16,* 45. 38. Physics Laboratoy Report PPPL S. Bernabei, M.A. Heald, W.M. 1374 (October, 1977). 46. 5th Conf. on Plasma Physics and H. Pacher and the WEGA team, Cont. Nuclear Fusion Res., Tokyo, T.H. Stix, The Theory of Plasma 42. p. 65 V.L. Vdovin, et al., in Third Int. meeting on "Theoretical 1962) and Experimental Aspects of J. Adam, et al., Euratom Reports Heating of Toroidal Plasmas", EUR-CEA-FC 579 (1971 and EUR- Grenoble, July, 1976, 48. T.F.R. Group, in Third p. 349 Int. F.W. Perkins, Symposium on meeting on "Theoretical and Plasma Heating and Injection, Experimental Aspects of Heating Varenna, 1972 (Editrice Composi- of Toroidal Plasmas", Grenoble, tori, Bologna, Italy, 1973, July, 1976, p. 87. 49. T.H. Stix, Nuci. Fusion 15, 737 (1975) 41. 1974, Vol. II, 47. Waves (McGraw-Hill, New York, p. 20). 40. J. Adam, et al., Proc. of the Rev. Lett. 34, 866 (1975) CEA-FC 711 (1973) 39. H. Takahashi, Princeton Plasma 47 (1976) privat communication 37. F.W. Perkins, Princeton Plasma rf Plasma Heating, Lubbock, looke, and F.J. Paoloni, Phys. 36. Fusion 17, 239 (1977) H. Takahashi, Rev. Lett. 50. et al., Phys. 39, 31 (1977). A. Hasegawa and L. Chen, Phys. T.H. Stix, in Third Symposium Fluids 17, 1399 (1974); on Plasma Heating in Toroidal ibid, Phys. Rev. Devices, Varenna, Italy, 1976 (1975); (Editrice Compositori, Bologna, ibid, Phys. Fluids 19, Italy; 1976, p. 156) (1976); J. Adam, in Symposium on Plasma ibid, Phys.. Rev. Lett. Heating and Injection, Varenna, (1976). Italy, 1972 (Editrice Composi- 51. Lett. }5, 370 1924 36, 1362 J.A. Tataronis and W. Grossmann, tori, Bologna, Italy, 1973, Z. Phys. 261, 203 (1973); p. 83) ibid, 217 Also, in "Third Symposium on Also in Proc. Sec. Conf. on rf Planma heating in Toroidal Heating of Plasmas, Lubbock, 17 Texas, 1974; 55. also, Nucl. Fusion 1_6, 4 (1976) 52. S.N. Colovato, J.L. Shohet and J.A. Tataronis, Phys. Rev. Lett. 53. published in 1978 56. A. Schl'tter, Z. Naturforschung 12a, 822 (1957). 37, 1272 (1976) T. Obiki, et al., Phys. Rev. Lett. 39, 812 (1977) 54. Petula Group, Grenoble, to be FIGURE CAPTION E. Canobbio, Nuclear Fusion 12, 561 (1972) and Course on the Fig. 1) A schematic representa- Toroidal Reactors, Erice, Italy, tion of an rf heating system. p. 71, (1972) Also, in "Symposium on Plasma Heating and Injection", Varenna, Italy, P. 14 (1972). 18 ANT ENNA R.F TRNSsMISSION SOURCC NETWOXtK MATCHWu NETWOR11 WAVES VACUUM WIN DOW Fig. 1 19 h (0) (Te;n) n a R xE (cm) (i) (W) (sec) 8 2 4 1xIO'4 6 4x10 14 15 1xlO15 2 AT (keV) 10 V AP cm OVW) 6.4x10 125 &6 1 4 2 8 8x107 54 0.2 0.5 0.2 5 4x105 4 Table 1 Examples of Ilypothetical Tokamak Parameters Near Ignition Conditions B - toroidal magnetic field; n - average density; a = minor radius; R - major radius; xE . energy confinement time; AT AT AT tempera- ture increase needed to achieve ignition (assuming that the ohmic heating contribution will be negligible); V - plasma volume; All power absorbed to achieve the AT indicated. 1. Bulk plasma heating 2. Tail-Heating for a 2-component tokawak 3. Radial temperature profile modification 4. Measuresient of the transport coefficients by local heating 5. Feedback control 6. De-trapping of trapped particles 7. Enhanced dc resistivity 8. DC currents due to quasi-linear effects 9. RF plugging of mirrors 10. Density build-up 1. RF confinement. Table 2 Possible Roles for RF lcating 20 Nature (Type) Electron Cyclotron Resonance (ECIU) Coupler Sburces Waveguide or horn Gyrotron Characteristic Frequency f e,2fc' fuh -100 Gilz Comments (CW sources) Presently available: P ' 200 kW, f 28-65 CII z Next 3 years P 200 IJI, f ~ 100 GIIz Lower Hybrid Range of Frequencier, (LIURF) fLh pi Waveguide Array (grill) 1-8 Glz Klystrons Presently available, 0.25-0.5 M%'/ tube, 1 1M possi- ble, CW Coils (Ridged waveguides may be possible) Tubes Presently available, 0.5-1 11 f < fci 1-10 MIZ Coi]p Tubes W-S Transit time mapnetic pnrnpinp (TTIP) electron or ion 0.1-1 MHz Coils Tubes MW-S Very low frequency (VLF) 1-10 klz Coils outside chamber Generator MW-S Ion-Cyclotron Range of frequencies (ICRF) f ,i2f 5-100clIz Alfven wave: shear or com-' pressional . Table 3