PFC/JA-83-33 MARFES: AN EDGE PLASMA PHENOMENON B. Lipschultz, B. LaBombard, E. S. Marmar, M. M. Pickrell J. L. Terry, R. Watterson, S. M. Wolfe Plasma Fusion Center Massachusetts Institute of Technology Cambridge, MA 02139 October 1983 This work was supported by the U.S. Department of Energy Contract No. DE-AC02-78ET51013. Reproduction, translation, publication, use and disposal, in whole or in part by or for the United States government is permitted. By acceptance of this article, the publisher and/or recipient acGovernment's right to retain a non-exclusive, knowledges the U.S. royalty-free license in and to any copyright covering this paper. 1 MARFES: AN EDGE PLASMA PHENOMENON B. Lipschultz, B. LaBombard, E. S. Marmar, M. J. L. Terry, R. Watterson, S. M. Wolfe M. Pickrell* Plasma Fusion Center Massachusetts Institute of Technology Cambridge, Massachusetts 02139 USA Abstract A tokamak phenomenon, edge phenomenon, dubbed the 'marfe', is described. This observed in medium to high density Alcator C discharges, is characterized by greatly increased radiation, density, and density fluctuations, and decreased temperature in a inner major radius edge of the plasma. relatively small volume at the The marfe appears to be confined to minor radii greater than or on the order of that of the limiter. The affected region is typically above the midplane, extending poloidally for about 30 degrees and for 360 degrees toroidally. The temperature and density of the core plasma are unaffected by the marfe. port model is used thermal instability, mechanism out of onset is observed. current and to show that with impurity the marfe volume. the marfe is the manifestation of a radiation being the main energy loss A density threshold, nm is found to be an increasing a decreasing A simple trans- function of intrinsic Detailed observations from spectroscopy, function of plasma plasma bolometry, nm for marfe impurity levels. Langmuir probe meas- urements, interferometry and CO2 scattering are described. *Present address: Mexico. Los Alamos Scientific Laboratory, Los Alamos, New 2 I. Introduction In Alcator C tokamak discharges a phenomenon is frequently observed which dominates the physics of the edge plasma [1,2,3]. [4] has been coined to describe this occurre,. ze. The marfe is charac- terized by a region of cold dense plasma localized radius edge of the plasma. The name 'marfe' to the smaller major Similar observations have been reported on other devices [5,6,7]. It is apparent that the marfe thermal instability [8,9,10], is the manifestation in which radiation supplants conduction to the limiters as the principal heat loss mechanism The conditions for its of a saturated for the edge region. occurrence and location are not as well understood. There is evidence from tokamaks with poloidal divertors that heat flows from the main [11,12]. This, we feel, is linked to the marfe occurrence on the opposite plasma predominantly to the The precise location side of the plasma. well as the conditions for marfe onset, larger major radius at the plasma inner are determined edge edge, as by the geometry and magnitude of conduction flows and radiation losses [2]. The marfe is a phenomenon of the main and edge plasmas. involving transport We observe, and atomic physics using Langmuir probes [131, that electron density and temperature are not constant on a flux surface during a marfe. Large amounts of poloidally asymmetric radiation emission are measured. Since the input and radiated powers for the main plasma are not affected by the marfe we can conclude that the total power incident on the limiters involved in has these been reduced. inhomogeneities An understanding could be used to of the control mechanisms the power flowing into the scrape-off layer. If that power could be dissipated over 3 a large area at and radiation the plasma edge through charge-exchange neutral losses then the remaining sputtering and heat loads on the [14.], limiter/divertor plate surfaces would be greatly reduced. Similar effects are obtained when the high-recycling conditioi analogous to a marfe, is created in a divertor chamber. A general The effect of characterization the marfe of the marfe is given on the density interferometer in Section II. is discussed in Section III. Specific observations regarding Langmuir probe measurements at the marfe location data related to the power radiated by different The relationship of are presented a marfe to a in Section thermal IV. Section regions instability of the plasma. is discussed Section VI and a discussion on the marfe location is related VII. V reviews in in section Sections VIII and IX summarize this paper and detail current inves- tigations respectively. II. General characterization Alcator C has major and minor radii of 64 and 16.5 cm respectively. Typical operating parameters 2.5 x 101 4 cm- 3 with central are ion Signals from various diagnostics, rence at 120 ms, are shown t B tor ' 8T$ and electron displaying in Figure 1. pm) chord (1a) tered out view of its appearence of the field of noise on the phase shift display. effect on the central vertical chord (1b) litude, high frequency, fluctuations. = keV. occur- the vertical is refracted detector, ne j 1 of a marfe At marfe onset, (119 kA, temperatures evidence inside density interferometer of plasma > 400 resulting In contrast, or scatin the the only is the appearance of low amp- Significant effects are also to be 4 found in such traces detector, along a (ld); and C III as: chord power at emission the same (4651A ) horizontal (1c), major chordal measured radius as by a bolometer la; emission (le). Ha emission Line emis- sion from higher ionization states of carbon and other impurities show no This leads us to conclude that the marfe is an edge pheno- such effect. menon. Confirmation is supplied by two simultaneous perpendicular views of the marfe in addition to measurements ,of parameters from plasma within the limiter radius. Characterization of the marfe onset involves several factors: plasma density (F ), plasma current, fixed plasma current, the marfe occurs for He above a critical density, nm. and discharge As the plasma current is increased, dependence is shown in Figure low-z gaseous impurities. into a non-marfing 2. to a value discharge. integrated comparable plasma (out to r = a). on the signal to nm has been lowered A plot of this by injection of The over N VI emission (3a) and total power open circles) both rise by a factor = 2. the marfe the integral of volume (3b, triangles) rises radiated power over the bulk This observation, combined with effects observed from the -12 indicate that a marfe nm also increases. For a Figure 3 shows the effect of nitrogen injection radiated from the bulk plasma (3b, Radiated power cleanliness. cm chord of the density interferometer(3c), has occurred. The total ohmic input power rose from 1.0 to 1.5 MW during the N 2 injection. The location in major and minor radii of full poloidal limiters has been varied for confinement studies in Alcator C [15]. plasmas have been produced at three major radii: Marfes occurred at the inside edge of the plasma, 10 cm minor radius 57.5, 64 and 70.5 cm. similar to 16.5 cm 5 plasmas, for the first second. One first and third major radial positions, obvious difference between and third were located close to, surface of the vessel (tangent still in place) while three repectively, to the the second these 16.5 cm. case was but not for the cases is that the the inner and outer radius limiter which was centrally located, with the vessel walls far these data: the marfe location is not a result of the flux surface geom- from the plasma. etry specific to 16.5 cm One can draw two radius plasmas; conclusions from the wall proximity may affect edge plasma parameters such as electron densi.ty and temperature, although there is no direct experimental evidence to support this conjecture. Impurity levels for the three cases, as monitored through Zeff, showed no significant differences. We can further characterize the profile information measurements from the location of the marfe by examining supplied by various diagnostics. several ports, and correlation of Multichord Ha separate diagnos- tics, indicate that the marfe location is toroidally symmetric, that is, not following field lines. Figure 4 shows two such profiles derived from a vertically viewing bolometer visible continuum array (4b). array above. Horizontal widths of the marfe In Alcator C, fluctuations as center varies brightness profiles we have observed, changes in both the magnitude measured by a horizontally viewing slightly above the midplane. profile measurements the marfe and These measurements indicate that the marfe occurred at the plasma's inside edge, precise position of (4a) from the midplane to 8 cm indicate typical to be in the vertical halfrange with the onset of a marfe, and spatial distribution of the CO2 laser The scattering 5-8 cm. dramatic edge density technique [16]. 6 The scattered laser frequencies between power, which is 20 and 1000 khz, proportional to -n2 and is shown in Figure 5a. include Concurrent with the onset of a marfe at 150 ms, a large increase in scattered signal is recorded. A similar effect visible continuum detector array is found horizontal on (Figure 5b). The scattering ments are averages along a vertical chord through the plasma. of over two orders of magnitude in scattered of chords the measureIncreases power are often observed when the scattering volume intersects the marfe region. The spatial distribution of the edge density fluctuations was studied correlation method using the crossed-beam derived by performing [16]. a cross-correlation of Spatial information is two C0 2 laser scattering signals which have chordal views through the plasma crossing at an angle of 2.9 degrees. The cross-correlation of these scattered signals provides of the relative a measure amplitude of the density fluctuations in the The shared volume of these shared volume of the two observation chords. two chords is approximately 6 cm in vertical extent and 3 mm wide in both Vertical transverse directions. the common scattering scans volume vertically of ff2 are performed from shot to shot. by moving Moving the entire arrangement in major radius permits comparison of density fluctuations at different major radius locations. A vertical shown in Figure scan of 6a. the crossed The major inside the magnetic axis. points in time before (open beams radius of through the marfe this vertical region scan was is 4 cm Correlation coefficient scans corresponding to circles) and during (closed circles) a marfe are nearly identical, even though the scattered laser power has increased over two orders of magnitude in the interval between the two measurements. 7 A single narrow region (less than the instrumental width) of fluctuations was found at the limiter radius above the midplane. A similar scan through a vertical chord 4 cm in major radius outside the magnetic axis appears in Figure 6b. In this symmetric profile with case, before the marfe the marfe, the circles) a peak at both upper and lower limiter Such a profile is typical of a non-marfe seen. (open scattered signal rises by a discharge factor of a radii is [17]. During 15 in a region localized to a narrow band at the upper limiter radius. We conclude that density fluctuations the marfe, exist at the marfe region before onset of a then increase dramatically in magnitude and spread across the top of the poloidal cross-section after marfe onset. density fluctuations The amplitude of these was reduced as the poloidal distance from the marfe along the edge plasma was increased. III. Interferometer effects As noted above, one of the signatures of the marfe is the "break-up" of the signal from the innermost chord of the far infrared interferometer. The normal output of this diagnostic is a display of the phase lag between a probe and reference beam. the plasma along the -12 During the marfe, the beam passing through cm chord no longer reaches the detector; as a result, the phase comparator circuit receives only wide-band preamplifier noise at one input, producing the "hash" shown in Fig. la. The most straightforward explanation of the interferometer signal loss is that the density in the marfe region exceeds the cut-off density for the laser frequency. However, the cut-off density corresponding to X = 0.119 mm is 7.9 x 1016 cm- 3 , more than two orders of magnitude larger 8 We therefore regard this explanation as that a typical central density. extremely improbable. A second, more likely mechanism, is refraction of the probe beam by a strong density gradient Deflection of in the vicinity of the marfe. the beam by an angle of ~ 20 mrad would be sufficient to cause the observed loss of signal. The refraction due to a uniform (or smoothly varying) density gradient, is approximately given by all Ym S - --- .(1) The refraction index where Xm is the path length in the marfe. = (1 - 5 /W2) 1/2 1 - /22, to the path results in refraction. and Taking 6 > 2 perpendicular gradient the only x 10-2, and the thickness of the marfe Xm - 2 cm we find 6n - 1.5 x 101 5 cm~ 4 (2) . 6x 3 4 Since the density in the marfe is typically a few 101 cm- , this gradient corresponds to a (perpendicular) scale length of the order of millimeters, which is not unreasonable. IV. Edge probe measurements A Langmuir probe [13] marfe location as determined ments (Figure 4). with larger same time, and vertically inserted into the from the above brightness profile measure- The ion density and electron temperature in that region are shown in Figure 7. marfe, was built The edge density (Figure 7a) increases occurring the electron temperature at (Figure increases during a larger minor radii. 7b) decreases. At the However, in 9 contrast to the density change, the temperature decrease is greatest just If these data are extrapolated to slightly outside the limiter radius. slightly increased (~ find the plasma we radius, within the limiter findings These 2). cooled observations similar to are the density and on the FT tokamak [7]. poloidal other at measurements Edge probe angles also have At 45 degrees away from the marfe center, obtained from Alcator C. been the density was observed to drop slightly, and the electron temperature stayed relatively constant degrees), as the marfe away poloidally Further began. (135 These data agree with there was normally no measurable effect. spectroscopic measurements of the marfe's poloidal extent. probe The Langmuir was amplitudes for frequencies also up to to used kHz. 10 measure (Figure 7c). cm prior to a marfe. amplitude was found to peak at 17.5 fluctuation density fluctuation The There was no radial dependence after marfe onset. V. Radiated Power During a marfe, the radiated power at the plasma undergoes edge This is illustrated by bolometer array meas- considerable changes [2,3]. urements shown in Figure 8 for discharges bounded by a molybdenum limiter. At low ne, ions in the most of central the plasma. radiated power decreases scrapeoff layer, of the plasma, measured although As rapidly. radiation ne is raised, Meanwhile, concentrated is emitted by molybdenum the magnitude the power of radiated by the at the smaller major radius edge is only a small fraction of the total ohmic input. the marfe threshold is passed, this When radiated power in the rather small marfe 10 region increases to a value on the order of that which is radiated by all As ;e is in- range of 20% to 30% of the total ohmic input power (POH). creased still further, the power the This value is typically in of the plasma inside the limiter radius. radiated by both the marfe region and the bulk of the plasma also increase. Because such from the large amounts small relatively of volume, marfe 20 watts/cm3 . there are typically radiated We power local have are being radiation identified emitted emissivities four groups of processes which might contribute to this large value of local emissivity. They are atomic and molecular hydrogen radiation, charge-exchange neutral emission, and low-z impurity radiation. Measurements of molecular been obtained. line radiation have Although the intensity of these lines increases dramati- cally during a marfe, ments indicate and atomic hydrogen based on absolute brightness measure- calculations ~ 10% of that a maximum of from the power radiated the marfe region can be accounted for by molecular and atomic hydrogen processes. that The hypothesis neutrals charge-exchange greatly contribute to marfe power losses was nullified by examining data obtained from two thermistor detectors detector was placed mounted flush with so as to have a direct only a short expanse of edge plasma. the main plasma, "view" chamber wall: of the marfe One through The second "viewed" the marfe through which should reionize virtually all neutrals which have trajectories passing through it. plus edge the vacuum plasmas during The total power radiated by the central a marfe, as measured by these 271 steradian 11 viewing bolometers, well as to each was compared other. measurements were The equal to from the bolometer to that magnitudes within of all three experimental array, as radiated power uncertainties. This indicates that charge-exchange neutrals are not a significant energy loss channel from the marfe. The remaining process local radiated power in which could a marfe is Profile impurities located there. account for the large thru line emission cm < r < rwall) of from the low-z of this emission in the measurements visible and ultraviolet wavelength ranges have been made. from the plasma periphery (14 value Low-z radiation was found to be primarily emitted from the smaller major radius edge of the plasma, which is consistent with the bolometer array measurements discussed above. emission lines of C III, asymmetry increased with emission increases not the case though the for edge C IV, and the higher is the magnitude of this in-out C V, When a marfe occurs, the asymmetry in C III e.. with bolometer in accordance plasma ionization expected not derived from that calculate the approximate local measurements. to in coronal Even equilibrium we have utilized steady state equilibrium emissivity. be This is above. states discussed (because of particle transport processes), cooling rates For measured condition assumption to Utilizing the edge density and temperatures probe measurements described in Section III, and taking carbon as a representative low-z impurity, we find local edge Zeff's of >1.5 are required to explain the observed radiation emissivities from the marfe region. This includes some extrapolation to radii smaller than the limiter radius. The marfe typically has little or no direct effect on the parameters 12 of the central plasma. that the marfe density. Circumstantial effect an indirect may have evidence, however, does indicate central on the molybdenum Returning to Figure 8, one finds that the regimes of molybdenum- and marfe-dominated discharges are observation could be coincidental, mutually exclusive. Although this no discharge has been recorded with a simultaneous occurence of large amounts of both central and edge derived of both types of radiation occur signals detectors Individual shown: two chords viewing vertically chiefly through the marfe from in the same discharge but at different times. by edge where large amounts This is further illustrated by Figure 9, radiation. region. the bolometer array This edge are center and two through the plasma This discharge is initially dominated at approximately 200 ms from the start radiation which, discharge, disappears. in is radiation replaced by of the molybdenum radiation from the central plasma. The response time of the detectors is approximately 20 ms. concluded It may be control the source, influx and/or retention main plasma somehow inhibit Studies of the transport impurities indicate that unaffected by the marfe. the marfe suppresses the of the the occurence artificially impurity edge that of molybdenum ions of a marfe or in the vice-versa. trace non-recycling injected transport which conditions confinement and are We are thus led to conclude that the onset of source of molybdenum. Presumably, decreased edge temperatures lead to a suppresion of the mechanism responsible for molybdenum evolution from the limiters. VI. Thermal instability and parallel transport The thermal stability of the edge plasma with tokamak density limit has been discussed elsewhere [8]. respect to the We consider, in a 13 similar fashion, the electron energy equation: aE 3 - = DnT 5 - 3T = 2 V - - (-KVT + -nTV) Qei + SEe - (3) 2 at For this simple discussion we reduce the above equation to one dimension along a field line and neglect in energy sources (SEe), duction, will and Qei. or increase in the divergence 3E/3T cause convective terms < 0. Following use the continuity equation to rewrite Eq. Any decrease of the heat con- Braginskii [18], we can 1 in terms of T rather than nT: 3 -n- 3T -3 2 at 2 5 nv - VT - nTV - v - - = V - (-KVT + -nTv) 2 Qei Again, we discard the velocity terms. as that of Eq. 3, (4) + SEe The right hand side is now the same but the time derivative is operating on Te. Thus, the imbalance of right hand side terms causes an explicit change in Te. If we require pressure to be ne, constant along a field line, of terms, of such an imbalance then the effect would be opposite to that on Te. on The energy loss term for electrons includes radiation: SEe = - -nenofo(Te) I ni is the impurity density the constituent neutral rate (watts-m3 ). instability to for the Ith impurity species. no refers to gas density and fi(Te) is the radiation cooling dfi(Te)/dTe < 0 is a necessary condition for a thermal occur in coronal equilibrium, edge plasma. (5) Z nenifi(Te) the edge this Any decrease plasma. For impurities of condition can be satisfied in Te for a fixed density Z < 8 over areas in of the or pressure will 14 cause the radiation Te, causing to increase. This heat loss will to the perturbed conducted unless power even more losses, further decrease region correspondingly increases. The thermal stability of this 1-D system can be judged by depressing the local temperature an amount 6T and by examining the new power balance. Should the radiation term increase faster than the conduction term, aT/at x would be negative and a thermal instability would occur. = L is far away to the location of the temperature perturbation while x is fixed. along the field line where the temperature power equilibrium scribing the - nInefi(Te) V balance (CK> prior 5 -- VT to 0 corresponds The equation deis perturbation the (K/T>( VT) 2 (6) 2 where one impurity is assumed to dominate and <K> is the averaged value conductivity of the thermal For a perturbation -6Te over L. at x - 0, an instability occurs if: 5 neni[fi(Te - 6Te) - > fi(Te)] (K/T>[(VTfinal) - 2 - VTinitial) 2 ] 2 VT > (7) 6Te + 5<(/T>L Rearranging equations, the instability criterion becomes: -5 dfj - < K/T > ( neniL dTe Using carbon [13], we have (8) VT cooling rates [19] and edge Langmuir probe calculated values for the left-(LHS) measurements and right-(RHS) hand 15 sides of VQT is Eq. 8 evaluated as a using the equilibrium the instability condition is first this radius, of condition a local edge carbon density, nc = .02 x ne, Eq. met at r=16.9 cm. This difference can be found to attributed 6. of Assuming this calculation predicts that In Figure 10, for we plot both sides of equation 8 versus Fse. density nm is higher than that The value and Re. radius of minor function The threshold = 2.5 x 10 14). experimentally ("e the crudeness of the model. The actual carbon abundance and cooling rate in the marfe region are unknown. In addition, no field line outside the limiter radius is continuous from the outside to the inside edge of the plasma. heat flow should be included; poloidal Pfirsch-Schlflter Some sort of perpendicular either diffusion out of the main plasma or bulk convective in the edge flows [201 would circumvent the limiters. Eq. 8 was also replacing evaluated carbon with oxygen as the dominant impurity. No unstable cases were found in the plasma edge simply because the slope of the cooling function fI is positive for most values of edge temperature. The implication is that an 'oxygen' marfe must start at a radius slightly less than rlim. At these radii, field lines can transport heat more readily; implying that this type of marfe will be more difficult to initiate. analyzes this situation The work of Ohyabu [8,91 and shows that it can lead to a major disruption. As the plasma He increases, initiating a marfe. This behavior experimental data reviewed above. raised, the edge ne(Te) increases (decreases), is evidenced Furthermore, by both Figure as the plasma 10 and current is one finds that the edge temperature rises due to increased heat flow into the edge (Figure 11). A rise in edge temperature would raise the 16 marfe threshold nm according to Eq. 8 and experiment (Figure 2). Perpendicular heat transport is inherently the mechanism that allows the marfe to saturate or reach steady state. the perpendicular Perhaps VT, increase. nificant. and parallel heat transport, more Eventually, an importantly, equilibrium As Te (x = 0) drops, which are proportional to convective is reached terms similar become to that sigof plasmas near limiter/divertor surfaces. The poloidal and toroidal extent of the marfe have been discussed above. In addition, the marfe not restricted to radii greater than that of the limiter. from two observations: region is This is inferred the marfe appears at all toroidal angles First, even though the edge plasma is divided toroidally into several separate Second, regions by full poloidal limiters. Figure la is derived the density signal shown in chord which passes between from an interferometer poloidal limiter rings separated by only 2 cm. VII. Discussion Both PDX heat flow is plasma. [11] and [121 ASDEX the smaller major a minimum near [211 Analytic modelling reported have of perpendicular that radius periphery of the Pfirsch-Schlilter flows plasma suggest that this in-out asymmetry in perpendicular due to bulk flow convection. the previous section assumed heat flow to the edge plasma The point where a marfe inside edge of the plasma. in the main flux is heat The thermal instability model described in some poloidal in the equilibrium occurs (x = 0) of asymmetry would perpendicular state before then be a marfe. located at the ASDEX [6] and FT [7] report marfe-like phenom- ena on the midplane at this location. In Alcator C the marfe is normally 17 located above the midplane at the inside edge of the plasma. bute this positional difference experiments to up-down asymmetries large to smaller major radius. asymmetry in flows: between marfes in heat There on Alcator We attriC and other flowing along the edge from could be several causes of this heat flows out of the main plasma may not be up-down symmetric; the edge plasma may rotate poloidally due to a radial electric field; and/or there may be radiation losses in the edge which lead to asymmetric poloidal temperature gradiepts and therefore, asymmetric poloidal heat flows. There exists little the first experimental evidence two possible causes stated above. to support or disqualify Fluctuations propagating in the ion diamagnetic drift direction have been detected, for Tie > 1.5 x 1014 using CO 2 laser scattering related to a poloidal Plasma convection experimental asymmetric heat flows These propagating fluctuations may be bulk flow. Pfirsch-Schlilter plasma There is more [171. suggested rotation as a modification flows is discussed evidence to low-Z impurity radiation spectroscopic measurements edge impurity that a C III shell from the edge we know that there increases with vertical asymmetry exists intensities (bottom/top) as cause 20. of profiles uti- This emission is typ- 12. which emission simultaneous support the third C III brightness above. lizing a horizontal view are shown in Figure ical of in reference of plasma. is a in-out ie. as Figure well. a function From vertical asymmetry in 12 The of ne is indicates ratio of given in Figure 13. Several points should be noted about these C III brightness profile data. First, the marfe does not occur at the lower inside edge of the 18 plasma (region 1), where the pre-marfe local radiation losses are highest. In addition, the radiation losses implication is that accordingly. the heat from region 1 increase with R e. conduction to that region has The increased In particular, at the marfe onset, region 1 is not affected, implying a very 'stiff' source of heat conduction. The second point to be noted is that the radiation losses from the edge of the Examining Eq. 8, upper inside changed. plasma the (region marfe 2) are at appears constant region as 2 not is Re because radiation losses have increased, but rather because the heat conducted to that region (RHS of Eq. 8) must have decreased. We might losses are have the expected marfe greatest before the marfe. ae where increasing no has Re is as explanation follows: effect on occur there. fixed flow increases to It there. follows region 1 statement in terms of Eq. 8, the right hand constant while and that decreases as region The C III to He 2, One possible in region 2 is relatively constant as a function of R e is relatively radiation occurs in region losses. radiation Te the where Instead it increased, C III emission resulting in increased to 1 decreases, emission from indicating that Te the heat increases, 2. region Putting this we see that the left hand side is staying side drops in magnitude for region 2. Eventually, the instability condition is met there. VIII. Summary The marfe major radius is a cool high-density periphery of the plasma. region It located appears for at ne the smaller above some 19 The threshold density, nm.. value of nm is increased by raising the plasma current and decreases by increasing the impurity density. Experimentally, the edge region is characterized both before and during a marfe by poloidal variations radiation is marfe. both in-out The magnitude Before a marfe occurs, in a and number up-down of these of parameters. asymmetric, relative inside edge of the plasma. and during increases with Eie. increase in magnitude and af.ter the onset of a marfe, Also, a are localized to the upper- These fluctuations expand poloidally during a marfe. before asymmetries density fluctuations Edge impurity the local (marfe region) electron density increases and temperature decreases. Power radiated ~ .2-.3 from this edge region typically to increases levels The power radiated from this small fraction of the edge x POH. plasma is often equal (central) plasma. to, or greater that than, from the bulk emitted The marfe thus supplants the limiter as an edge energy sink. A simple thermal instability model is model, This discussed. combined with the hypothesis that perpendicular heat flow out of the main plasma is not poloidally uniform, correctly predicts the marfe characteristics: threshold density, nm; location in minor radius poloidal and Alcator C, exhibits verti- angle; and scaling of nm with plasma current. cal asymmeties in impurity radiation emission. This asymmetry in radiated power causes asymmetries vertical asymmetries one in edge power might inside midplane of the plasma. expect flows. the In the absence marfe to be of these located at the If vertical asymmetries in emitted power are present, the marfe location could shift off the midplane, as observed in Alcator C. 20 Further work is needed to from onset through saturation. understand the dynamics of the marfe The physics involved in this phenomenon is quite complicated, similar to that of the plasma located at limiter/divertor surfaces. IX. Continuing investigations Currently work on understanding the marfe is being pursued both experimentally and theoretically at M.I.T.. built to measure edge parameters, at one toroidal location, simultaneously A multiprobe array is being at a number of poloidal angles and minor radii. A one-dimensional numerical model for transport along a field line [22] is being utilized to understand the marfe. This model includes convective as well as conductive transport in a time-independent, ular source terms be implemented. single fluid treatment. Perpendic- of both uniform and nonuniform spatial dependence can The earlier analytic work mentioned [20] is being ex- tended to a more rigorous two-dimensional numerical model. Acknowledgements The authors would like to thank the rest outside collaborators U.S.D.0.E. for their assistance. contract # DE-ACO2-78ET51013. of the Alcator group and This work was supported by 21 References [11 J. L. Terry, et al., Hydrogen Line Phys. Soc., [2] B. "Strongly Emission in Some Enhanced Alcator Low-Energy Continuum C Discharges" Bull. and Amer. 26, 886(1981). Lipschultz, B. LaBombard, M.M. Pickrell, J.L. Terry "Marfes: Poloidally Asymmetric Edge Conditions in Alcator C" Bull. Amer. Phys. Soc., 27, 937(1982). [3] M. M. Pickrell, "The Role of Radiation on the Power Balance of the Alcator C Tokamak" M.I.T. Plasma Fusion Center Report, PFC/RR-82-30. [4] The name "marfe" is a concatenation Marmar and Wolfe. of the last names of authors Indications of a marfe occurence were first noted on signals from diagnostics operated by those persons. [5] D. R. [6] H. Niedermeyer, Baker, R.T. Snider, et al., M. Nagami, presented Nuclear Fusion 22, 807(1982). at the European Physics Society conference in Aachen, September, 1983. [7] F. Alladio et al., Physics Letters, 90A, 405(1982) [8] N. Ohyabu, Nuclear Fusion, 9, 1491(1979) [91 N. Ohyabu, Kakuyugo-Kenkyu, 43, 9(1981). [101 J. Neuhauser, "Characteristics of a Radiating Layer Near the Boundary of a Contaminated Plasma", Max-Planck-Institut ffr Plasmaphysik, IPP 1/182(1980). [11] D. K. Owens, et. al., Journal of Nuclear Materials, 93-94, 213 (1980). [12] H. Keilhacker, et al., Max-Planck-Institut IPP 111/72, (1982). fur Plasmaphysik, Report 22 LaBombard, [13] B. Lipschultz, B. of Probe Measurements the Pickrell, M. Edge Y. in Alcator Plasma "Langmuir Takase, C" Bull. Amer. Phys. Soc., 27, 1036(1982). [14] B. Lenhert, Nuclear Fusion, 19, 1319(1978) [15] B. Blackwell, et. al., 9th (IAEA International Conf. 1982) 1983, Baltimore, Physics and Controlled Nuclear Fusion, on Plasma IAEA- CN-41/C-4. [16] C.M. Surko and R.E. Slusher Phys. Fluids 23, 2425(1980). [17] R.L. C.M. Watterson, Surko and R.E.. Slusher, "Spectra, Spatial Distribution and Propagation Velocity of Low Frequency Fluctuations in Alcator C" Bull. [18] S.I. [19] D. E. Braginskii, Post, Amer. Reviews et al., Phys. Soc., 27, 937(1982). of Plasma Physics, Vol. 1, Atomic Data and Nuclear Tables, 205-311(1965). 20, 397(1977). [20] B. LaBombard, M.I.T., Plasma Fusion Center Report. [21] M. D. Rosen and J. M. Greene, Princeton Report PPPL-1315. [221 B. Lipschultz, M.I.T., Plasma Fusion Center Report # PFC/RR-83-25. 23 Figure Captions Figure 1: Diagnostic traces vs. time. The marfe occurs at 120 ms. a) The inside (-12 cm) vertical density interferometer chord; b) The central vertical density interferometer chord; c) The inside (-12 cm) bolometer channel; d) H-alpha emission which saturates after marfe onset; and e) C III line emission (4651A). Figure 2: Threshold density, nm, for marfe to occur vs. plasma current. Figure 3: The effects of N 2 injected into a non-marfing discharge: a) N VI emission (1897A); b) Power radiated from the marfe area at the plasma edge (closed triangles) and from the main plasma (open circles); c) Line-averaged electron density from inside (-12 cm) vertical interferometer chord. Figure 4: Brightness profiles before and during a marfe: a) Bolometer profiles as viewed vertically, and b) Visible continuum profiles viewed from the horizontal direction. Figure 5: Evidence of high-frequency edge density fluctuatons during a marfe: a) n~ 2 signal from CO 2 scattering; and b) Horizontal chord of the visible continuum array. The marfe begins at 150 ms. Figure 6: Two-beam CO2 laser correlation measurements before (open circles) and during (closed circles) a marfe. The two traces shown in a) and b) correspond to vertical profile measurements through the plasma for R-Ro = +4 cm. and R-Ro = -4cm. respectively. I 24 Figure 7: Langmuir probe measurements, before and during a marfe, versus minor radius outside the limiter radius: a) Ion density; b) Electron temperature; and c) Fluctuation amplitude. Figure 8: Fraction of the ohmic input power radiated from the marfe region (x-x-x) and from the remaining bulk of the plasma (closed circles) vs. -ne Figure 9: Brightness of four vertical bolometer chords vs. time: traces a) and b) principally view the inside edge of the plasma and are a measure of power radiated from the marfe region; traces c) and d) are central chords and thus are good measures of molybdenum radiation. Figure 10: The left-(LHS) and right-(RHS) hand sides of Eq. 6 plotted as a function of ne. Figure 11: Te, at r - 17.3 cm, versus plasma current for a fixed Figure 12: C III brightness profiles (horizontal view): e. before the marfe, the intensity of emission from the bottom inside edge of the plasma increase with ne (open circles); when the marfe occurs the intensity at the upper inside edge of the plasma increases (closed circles). Figure 13: Pre-marfe ratio of C III brightness from lower inside over upper inside edge of the plasma vs. ne. I I i a bl C d I 0 100 I 200 msec Figure 1 300 400 nm vs. PLASMA CURRENT 3.3 X x E X x x X X X E C 2.71X X C x 2. 1 X (D X 1.5 L 25 0 x I . 350 450 . 550 Plasma Current (kA) Figure 2 sv/~f1 Injection Time a Cn z- 800 -A- Power from Marfe 6001 -o-- Power from Bulk 4001 0 -J 2001 C,) C zd 0 100 200 300 400 500 Figure 3 MILLISECONDS 8447 BOLOMETER IC a] W 111r411~ 0 wr C MARFE 0d C -12.0 'I -8.0 -4.0 0 HORIZONTAL 4.0 8.0 12.0 POSITION (cm) VISIBLE CONTINUUM 16.0 b] MARFE + +1I 4 top -16 VERTICAL POSITION bottom PFC-&99 Figure 4 I I I i I I I 300 400 a 0 100 200 500 Time (msec) Figure 5 0.80 0- I 0.60 -I S0 0.40 / E-0 0.20 C CR/ Ol -20 -10 0 +10 +20 + +30 0 0.8010.600.40- 0.20 0 0 .Oo ' ZI -20 0-Oi -10 0 +10 +20 +30 Vertical Distance (cm) Figure 6 1 015_ / A o o/ 14 10 - / 4.1 mm 0 0 8/ 0 0 E 0 0 0 Am .0/I / 10 0 0 .0 / 0 Y 2.5 mm 0 z 0 0 0 10 0 / 0 0 - AVERAGE DENSITY BEFORE MARFE I / 10 -J 18.5 * -AVERAGE DENSITY DURING MARFE 0 0 n 18.0 m 17.5 17.0 165 16.0 RADIAL POSITION (cm) Figure 7 a 20-1 00 0 15- 0 D: LU 0 10- 03 z 0 0 C") LU 0 0 BEFORE MARFE: 0 0-PEAK Te 5- DURING MARFE: 7. *-AVERAGE Figure 7b 0 -I. 18.5 In ~m 18.0 I 17.5 17.0 RADIAL POSITION (cm) 16.5 1 16.0 Te 0.5- 0 0,4- 0.3. cr 0 S S 0 S 0 0 0 . 0 0.2- 0 0 8: . 0 0 8 0 0.1- 0 0 0 0 0 0 18.5 ,~ 00- I I 18.0 17.5 RADIAL v I-. BEFORE MARFE DURING MARFE 17.0 POSITION 16.5 16.0 (cm) Figure 7 c 0.8 K- 0.6 I 0 a- 4x 0 *0 0 a- x 0.4 0.2 x 0 0 x xxxx-X I 2 I 4 6 ne (1014 c m3) Figure 8 marfe a) -12.6cm 04 b) -Il I m - -c c) -2.5 cm d) Center I I I Time (100 I I I I ms/div) Figure 9 6f7Z o -R.H.S 0~x - L. H.S 0 -6 C a 10 100 10 10 14 >(10 c -3- 6.0 - 5.01- 0 4.0 I- 3.0 2.0 I 200 I 300 Plasma Current 400 (kA) Figure 11 9471 C II BRIGHTNESS PROFILE: MARFE AND PRE-MAREE 987 - 6 0 5 U z 1- 4 - CC 3 00 Top - -0 Midplane Bottom Figure 12 PFC - 820/ 4 I I C'T .5101.J. -F (014i3i3 LU-80 Fiur 13