* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Three-dimensional electron magnetohydrodynamic reconnection. I
Plasma (physics) wikipedia , lookup
Magnetic monopole wikipedia , lookup
Maxwell's equations wikipedia , lookup
Electromagnetism wikipedia , lookup
Condensed matter physics wikipedia , lookup
Magnetic field wikipedia , lookup
History of quantum field theory wikipedia , lookup
Lorentz force wikipedia , lookup
Mathematical formulation of the Standard Model wikipedia , lookup
Aharonov–Bohm effect wikipedia , lookup
Field (physics) wikipedia , lookup
PHYSICS OF PLASMAS VOLUME 10, NUMBER 7 JULY 2003 Three-dimensional electron magnetohydrodynamic reconnection. I. Fields, currents, and flows R. L. Stenzel, M. C. Griskey, J. M. Urrutia, and K. D. Strohmaier Department of Physics and Astronomy, University of California, Los Angeles, California 90095-1547 共Received 28 August 2002; accepted 7 April 2003兲 In a large laboratory plasma, reconnection of three-dimensional 共3-D兲 magnetic fields is studied in the parameter regime of electron magnetohydrodynamics. A reversed magnetic field topology with two 3-D null points and a two-dimensional 共2-D兲 null line is established, and its free relaxation is studied experimentally. Major new findings include the absence of tilting instabilities in an unbounded plasma, relaxation times fast compared to classical diffusion times, dominance of field line annihilation at the 2-D current sheet versus reconnection at 3-D null points, conversion of magnetic energy into electron thermal energy, and excitation of various microinstabilities. This first of four companion papers focuses on the magnetic field topology and dynamics. © 2003 American Institute of Physics. 关DOI: 10.1063/1.1578998兴 of microinstabilities 共ion sound, whistlers, plasma waves兲 have been observed but their role in the dissipation process is still under investigation. The material is presented in four companion papers. The present, Part I, describes the experimental setup, the magnetic field topology, currents and electric fields, reconnection, tearing, annihilation and convection of magnetic fields. The second paper, Part II,13 deals with the stability of a freely relaxing FRC to translation, rotation, and precession. The third paper, Part III,14 deals with the energy flow from the external source into stored magnetic energy and its release into the particles. The fourth paper, Part IV,15 describes microinstabilities and their effect on the reconnection rate and electron heating. The present paper is organized as follows. In Sec. II, the plasma device and the measurement techniques are described. The experimental results, divided into various subsections, are presented in Sec. III. The conclusion, Sec. IV, points out the new findings and implications to other related research. I. INTRODUCTION Magnetic reconnection describes dynamic processes in plasmas that produce changes in field topologies, associated plasma transport, and energization of particles at the expense of magnetic field energy. These processes play an important role in space and laboratory plasma physics. In spite of a long history of research,1 reconnection remains at the forefront of interest in many areas.2 The topic is complicated because of coupled, multiscale processes with different physics near magnetic null points. For example, an outer region described by single-fluid magnetohydrodynamics 共MHD兲 is coupled to an intermediate region of electron magnetohydrodynamics 共EMHD兲,3 which in turn is coupled to the unmagnetized region of the null point. The field of reconnection has evolved from steady-state pictures for two-dimensional 共2-D兲 geometries4 to three-dimensional 共3-D兲, timedependent reconnection models5,6 supported by many numerical simulations including Hall effects.7,8 In space and fusion plasmas, it is often difficult to observationally resolve both global and microscopic processes simultaneously. Hence, dedicated laboratory experiments have been performed aimed at studying reconnection processes under controlled conditions.9,10 The present work describes a new experiment that focuses on reconnection in the EMHD parameter regime. Compared to earlier work in this parameter regime,11 the present experiment has no boundary effects, reconnection is both driven 共turn-on phase兲 and nondriven 共spontaneous, turn-off phase兲 and the field topology is 3-D. Major findings include 共i兲 a topology change from a field-reversed configuration 共FRC兲 into a uniform field on a time scale fast compared to classical diffusion times, but slower than whistler transit times, 共ii兲 no disruptive tilting instability of a freely relaxing EMHD FRC, 共iii兲 conversion of magnetic field energy into electron thermal energy, and 共iv兲 energy conversion which does not take place by reconnection at 3-D null points, but by field line annihilation in a toroidal magnetic neutral/electron current sheet.12 A variety 1070-664X/2003/10(7)/2780/14/$20.00 II. EXPERIMENTAL ARRANGEMENT The experiments are performed in a large laboratory plasma device schematically shown in Fig. 1共a兲. A 1 m diam ⫻2.5 m long plasma column of density n e ⱗ 1012 cm⫺3 , electron temperature kT e ⯝3 eV, ion temperature kT i ⭐0.3 eV, argon gas pressure p n ⯝0.4 mTorr 共plus 1% hydrogen for cathode activation兲 is produced in a uniform axial magnetic field B0 ⯝⫺5 G ẑ 共away from the cathode兲 with a pulsed dc discharge (V dis⯝50 V, I dis⯝600 A, t pulse⯝5 ms, t rep⯝1 s兲 by a 1 m diam oxide-coated cathode. In the early, current-free afterglow plasma 关 t a⬇115 s, n/( 兩 n/ t 兩 )⬇1 ms兴, pulsed currents 共100 A, t pulse⯝47 s, t rise⯝5 s, t fall⭐3 s兲 are applied to two insulated magnetic loops forming a Helmholtz coil 共30 cm diam, 15 cm axial spacing, 4 turns each兲. The pulse circuit consists of a charged, floating capacitor 共300 2780 © 2003 American Institute of Physics Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp Phys. Plasmas, Vol. 10, No. 7, July 2003 Fields, currents, and flows 2781 FIG. 2. Calculated field lines in vacuum for a strong Helmholtz coil field 共25 G兲 in a weak uniform field 共5 G兲 which is 共a兲 opposing, and 共b兲 aligned with the coil field. FIG. 1. 共a兲 Experimental setup with basic plasma parameters. The poloidal magnetic field lines produced by the Helmholtz coil are shown. 共b兲 Waveform of the applied current to the Helmholtz coil. F兲, several high-power field effect transistors 共1200 V, 100 A兲 and Rogowski loops to measure the pulsed current. The voltage applied to the coil is measured with a shunt resistor. The waveform of the applied coil current, shown in Fig. 1共b兲, has a slow rise time governed by the coil inductance (L coil ⬇15 H兲 and circuit resistance (R circuit⬇2 ⍀). Its pulse length 共47 s兲 is sufficient for the magnetic field to penetrate into the plasma and to establish the vacuum-like field topology 共as evidenced by a zero induced voltage on the coil at t⯝40 s兲. The abrupt turn-off is made feasible by using fast field-effect transistors. The reason for the fast switch-off is to study the free relaxation of the stored poloidal magnetic field inside the plasma shown schematically in Fig. 1共a兲. Since there are no conducting boundaries nearby, the relaxation can proceed self-consistently as in free space. The Helmholtz coil magnetic field employed in these experiments is typically stronger (B H⫽25 G兲 than the uniform field (B 0 ⫽5 G兲, resulting in a net field as shown in Fig. 2共a兲. The field in vacuum, calculated from Biot-Savart’s law and experimentally verified, has two cusp-type 3-D null points on axis, two O-type null lines and one X-type null line, all azimuthal 共or toroidal兲, at nearly the radius of the coil. Under identical conditions, the Helmholtz coil field can be reversed with a relay switch resulting in a net field as shown in Fig. 2共b兲. In such case, there are no 3-D null points on axis but two X- and two O-type null lines, all toroidal, in the vicinity of the individual coils. Comparison between these two configurations is made to point out the difference in the relaxation mechanism and electron energization with and without 3-D null points. Other configurations such as two opposing coil fields, a single coil, and coil fields oblique to B0 have also been investigated, but results are preliminary. The time-varying magnetic fields produced by the pulsed current in the coil and the induced currents in the plasma are measured with a single vector magnetic probe, recording (B H,x ,B H,y ,B H,z ) versus time at a given position with digital oscilloscopes 共100 MHz, 8 bit兲. By repeating the highly reproducible discharges and moving the probe to many positions in a 3-D volume, the vector field BH(r,t) may be obtained. The total magnetic field is obtained by adding B 0 ⫽⫺5 G to the measured B H,z component, B(r,t)⫽BH(r,t) ⫹B0 . Typically, at a given position, an ensemble average over 10 repeated shots is formed, which increases the digitization accuracy from 8 bit to approximately 11 bit. The spatial resolution of the B-field measurement is given by the probe size 共0.75 cm兲, and is sufficient to resolve magnetic structures on the electron inertial scale length (c/ pe ⬇1 cm兲. The temporal resolution is on the order of the electron cyclotron period (2 / ce ⬇70 ns兲, and improved by temporal deconvolution using measured waveforms in vacuum and the coil current waveform. Integration of the probe signal (⬀ B/ t) is performed numerically, and the absolute probe calibration is done with the known field inside the Helmholtz coil. From the stored traces of BH(r,t), the spatial field topology can be displayed at any instant of time. The current density is calculated from Ampère’s law, J⫽ⵜ⫻B/ 0 共the displacement current is neglected because the magnetic field evolves on time scales long compared to ⫺1 pe ). The measurement accuracy is confirmed by checking that ⵜ•B⯝0 when 兩 B 兩 / 兩 ⵜB 兩 is greater than the probe spacing. Plasma parameters are obtained from a small Langmuir probe ( r 2 ⯝2.6 mm2 ). Density fluctuations are detected with a cylindrical probe 共0.5 mm diameter, 2 mm length兲 connected to a 50 ⍀ coaxial cable. Microwave emissions near the plasma frequency are done with a superheterodyne receiver. Light emission from the plasma is detected with a photomultiplier tube and resolved with a McPherson scanning monochromator. The diagnostic tools for microwave and light measurements will be described in more detail in Part IV. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp 2782 Stenzel et al. Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 3. Measured vector fields (B y ,B z )(0,y,z,t ⬘ ) in plasma at different times after switch-off of the coil current (t ⬘ ⫽0). 共a兲 Initial field topology when BH 储 ⫺B0 ; 共b兲 topology when I coil⫽0; 共c兲 field-reversed configuration at t ⬘ ⫽8 s; 共d兲–共f兲 corresponding data for BH 储 B0 . The steady state field configuration (t ⬘ ⫽0) rapidly relaxes to the uniform background field (B0 ⬇5 G兲. III. EXPERIMENTAL RESULTS A. Magnetic field topologies The change in the magnetic field topology during the free relaxation of the field induced in plasma by the Helmholtz coil is displayed in three snapshots of measured vector fields (B y ,B z ) in the central y–z plane (x⫽0) in Figs. 3共a兲– 3共c兲 for BH 储 ⫺B0 and Figs. 3共d兲–3共f兲 for BH 储 B0 . For the former case, as the coil current starts to decrease (t ⬘ ⫽0), the field topology is as in vacuum 关Fig. 3共a兲兴: an FRC. After the coil current has vanished 关 t ⬘ ⯝3 s, Fig. 3共b兲兴, the field near the individual coils has decayed, but no significant change has taken place in the center of the Helmholtz coil. Because the bulk field is now similar to the vacuum field, this implies that plasma currents comparable to the coil current (I coil⫻8 turns ⬇800 A兲 have been induced. Since the field near the individual coils is lower than before, the plasma currents must be distributed over a region (⯝4 cm兲 larger than the thin Helmholtz coil windings. As this is an EMHD plasma, the currents are carried by electrons. It is shown in the following that the current is predominantly a Hall current, driven by inductive and space-charge electric fields associated with the decaying magnetic fields. There are no conducting boundaries with induced currents nearby. The openloop coils have no influence on the distribution of plasma currents. The relaxation of the magnetic field proceeds selfconsistently as in free space. The currents are able to redistribute themselves freely and are not localized to a thin current sheet in the magnetic neutral layer where EMHD breaks down. Observation shows that from t ⬘ ⫽3 s to t ⬘ ⫽8 s, the FRC contracts radially and expands axially leading to the configuration shown in Fig. 3共c兲. The axial elongation and radial contraction are explained in the following as a result of magnetic field lines being frozen into electron fluid flow. The toroidal component of the magnetic field forms during the axial expansion. Although many aspects of the magnetic Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp Phys. Plasmas, Vol. 10, No. 7, July 2003 evolution are explained by EMHD, the rapid disappearance of flux near the wire indicates a breakdown of EMHD. Since the electron fluid is frozen to the magnetic flux in EMHD, the disappearance of the flux would require the electrons to flow into the wire violating ⵜ•J⫽0. For t ⬘ ⬎8 s, the FRC maintains a similar structure to Fig. 3共c兲 but decays in amplitude and shrinks axially, eventually relaxing to the uniform background field. The decay time scale ( ⬇15 s兲 is long compared to a whistler transit time across the magnetic structure ( w ⬇3 s for v whistler ⬇10 cm/ s and L B⬇30 cm兲 but short compared to the classical diffusion time scale ( diff⫽ 0 SpitzerL B2⬇1.5 ms for Spitzer⬇150 ⍀ ⫺1 cm⫺1 at kT e ⯝4 eV兲.16 The 15 s time scale corresponds to a conductivity of * ⫽15 s/( 0 L B2) ⯝1.3 ⍀ ⫺1 cm⫺1 . This anomalously low conductivity is thought to be due to ion turbulence observed during the decay of the FRC as shown in Part IV. The magnetic moment of the FRC oscillates slightly around the z axis but does not exhibit a major tilting instability. The oscillation is due to the FRC magnetic field being frozen to toroidal Hall electron flows17 and is discussed in Part II.13 For purpose of comparison, Figs. 3共d兲–3共f兲 show the topology of the total field for the case when the Helmholtz coil field is along the uniform field, BH 储 B0 . Again, the initial field is as in vacuum 关 t ⬘ ⯝0, Fig. 3共d兲兴. In contrast to the case where BH 储 ⫺B0 , the field in the center of the Helmholtz coil begins to decay as soon as the coil current begins its switch-off (0⭐t ⬘ ⭐3 s兲. After the coil current has vanished 关 t ⬘ ⯝3 s, Fig. 3共e兲兴, little magnetic field energy is left to relax. Thus, the relaxation proceeds much faster than in the case of the FRC and an almost uniform field is restored at t ⬘ ⯝8 s 关Fig. 3共f兲兴. As shown in Part III,14 the magnetic energy is axially convected away from the Helmholtz coil at the whistler speed when BH 储 B0 . This case is similar to the well-documented formation of a whistler wave packet by a dipole loop antenna.18 The propagation is predicted by EMHD theory, but just like the BH 储 ⫺B0 case, EMHD breaks down near the wire as evidenced by the rapid decay of the field in that region. It should be noted that EMHD does not predict propagation in the BH 储 ⫺B0 case because the two 3-D null points on the axis of the Helmholtz coil 关 z⫽⫾28 cm in Fig. 3共a兲兴 stagnate the wave propagation causing dissipation to dominate. Much less electron heating and its associated light emission is observed in the propagating case 共see Parts III and IV兲. In order to describe the field topology in 3-D space, we show in Fig. 4 the field properties in an orthogonal plane, the x–y plane in the center of the Helmholtz coil (z⫽0). Figure 4共a兲 displays contours of B z (x,y,t ⬘ ⫽7 s兲 for the FRC containing a null layer, and Fig. 4共b兲 shows B z (x,y,t ⬘ ⫽3 s兲 for the case BH 储 B0 without a null layer. For both polarities, the field shows good azimuthal symmetry. In the FRC case, the magnetic null layer 共where EMHD breaks down兲 has no significant influence on the stability of the field distribution. Note that at t ⬘ ⫽7 s the field strength at the center of the FRC is nearly 10 G, which means the field created by the plasma currents 共which opposes the 5 G background field兲 is about 15 G. Four microseconds earlier, the field created by the plasma currents has a magnitude of about 10 G in the Fields, currents, and flows 2783 FIG. 4. Axial field component B z in the central x–y plane of the Helmholtz coil (z⫽0). 共a兲 B z (x,y,t ⬘ ⫽7 s兲 for BH 储 ⫺B0 showing the magnetic null layer of the FRC. 共b兲 B z (x,y,t ⬘ ⫽3 s兲 for BH 储 B0 showing a strong net field on axis. The field has a good azimuthal symmetry for both polarities. BH 储 B0 case. This is another indication of how the field strength in the two cases evolves on different time scales. B. Moving field lines The concept of moving field lines, often used to demonstrate magnetic reconnection, is meaningful when the field lines are tied or frozen into the moving fluid.19 This requires that magnetic field lines be defined and followed as the experiment progresses. Experimentally, the measured vector magnetic field is used to trace the field lines, and their motion is determined from v⫽⫺J/(ne). However, this method is not easy to implement and often leads to erroneous results. For systems with cylindrical symmetry, flux conservation allows an alternate method: magnetic field lines are identical to contours of constant poloidal flux, ⌽(r,z)⫽ 兰 r0 B(r ⬘ ,z) •â 2 r ⬘ dr ⬘ ⫽const where â is identical to a vector parallel to the inner part of the separatrix. Following such contours as time advances is then a very simple process. We showed earlier that the magnetic field configuration is nearly axially symmetric ( / ⬇0). We can therefore calculate ⌽(r,z) once we identify the axis of symmetry. This is easiest for the steady-state field configuration 关see Figs. 2共a兲 and 2共b兲 of Part II13兴. The symmetry axis is easily identified as â⫽ẑ. We calculate ⌽(r,z) for both halves of the plane separately to account for the small differences in symmetry between the upper and the lower half planes. The calculated lines of constant poloidal flux are tangential to the vector field, confirming that they are field lines. The value of the Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp 2784 Stenzel et al. Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 5. Contours of constant poloidal magnetic flux 共magnetic field lines兲 for BH 储 ⫺B0 in the central y–z plane at different times after turn-on 共a兲–共c兲 and turn-off 共d兲–共f兲 of the coil current. 共a兲 At early times, field lines are open but distorted near the coils. 共b兲 With increasing coil current (t⫽5.8 s兲, closed field lines encircle each coil, developing X-type magnetic nulls lines that converge to the axis and form 3-D null points 共c兲. 共d兲 During the decay of the coil current, closed field lines around the coils vanish while the rest of the FRC remains relatively unchanged. 共e兲 In the absence of coil currents, the FRC is maintained by electron currents, which form a long neutral sheet 共f兲. flux contour has no relation to the local field strength but identifies a particular field line. The zero contour defines the separatrix, the boundary that separates open field lines connect to B0 from closed field lines within the FRC. This is clearly understood as follows: initially, the flux is zero at the symmetry axis. Since the field lines are in a single direction above and below the axis, the flux monotonically increases above/below the axis. If the field lines are closed, the flux reaches a maximum and decreases until it changes sign. The integral has accounted for all the field lines enclosed by the separatrix at that point. Hence, the ⌽(r,z)⫽0 contour defines the separatrix for axially symmetric magnetic field configurations. The time evolution of field lines is shown in Fig. 5 for BH 储 ⫺B0 and in Fig. 6 for BH 储 B0 . As the coil current begins to grow (t⫽1.6 s兲, Fig. 5共a兲 shows still open field lines associated with B0 , but distorted near the Helmholtz coil, which intersects the y–z plane at four points (y⯝⫾8 cm, z⯝⫾8 cm兲. The first magnetic null lines form just in- side each of the individual coils as the current increases. These are X-type nulls in the y–z plane with a toroidal separator. With increasing coil current, these null lines move radially toward the z axis. The motion of the null lines causes the field lines between the coils to become highly stretched and to reconnect forming an X- and O-type null lines in the center plane (z⫽0, r⬍r coil) as shown in Fig. 5共b兲 (t⫽5.8 s兲. The O-type null is bounded by a separatrix through the 3 X-type nulls. In vacuum 共not shown兲, two X-type null lines are also formed in the coil planes but no nulls in the center plane. O-type nulls cannot form in vacuum since currents flow only in the coils. With rising coil current the closed flux around the coils increases, which pushes the two X nulls in the coil planes toward the axis. The X null in the mid-plane remains almost stationary. Although reconnection at the X nulls feeds flux into the O null it does not grow, presumably due to field line annihilation. When the field lines trapped within the separatrix near the axis are all reconnected the two X nulls and O Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 6. Contours of constant poloidal magnetic flux 共magnetic field lines兲 for BH 储 B0 in the central y–z plane slightly after turn-on 共a兲, steady state 共b兲, and after turn-off 共c兲 and 共d兲 of the coil current. 共a兲 At early times field lines are open but distorted inward near the coils. 共b兲 Field lines are radially compressed as the coil current increases, leading to formation of null lines at steady state. 共c兲 After switch-off of the coil currents (t ⬘ ⲏ 3 s兲, the field lines straighten out rapidly due to convection of magnetic energy along ⫾B0 . 共d兲 The emission of two whistler wave packets causes field line compression (z⫽⫾25 cm兲 and rarefaction (z⯝0). null reach the axis. The O null vanishes and the two X nulls turn into 3-D null points 关 t⬇8.1 s, Fig. 5共c兲兴. The null point at z⯝8 cm is characterized by opposing field lines approaching the point along the z axis 共this axis is generally referred to as the ‘‘spine’’兲 and field lines flaring out from the null point along two conical surfaces that intersect at the null 共these surfaces are referred to as the ‘‘fan’’ planes兲. The null Fields, currents, and flows 2785 point at z⯝⫺8 cm is essentially identical, but with oppositely directed magnetic fields. A detailed classification of these types of null point structures is given in Ref. 20, where these nulls are termed ‘‘regular.’’ At this time, there are two nested separatrix surfaces, one, where ⌽⫽0, separating closed field lines around both coils from open field lines, and the second one, separating closed field lines around each coil from closed field lines encircling both coils. In contrast to the first separatrix, this second separatrix cannot be assigned any particular value because it separates closed field lines from other closed field lines. Further increase in the coil current pushes the 3-D null points axially outward until the steadystate condition is reached 共see Fig. 2 of Ref. 17兲. The topological changes during the rise of B H are produced by reconnection driven by the Helmholtz coil current. Flux originating from the coils reconnects at the axially symmetric X line to form common flux around both coils, as discussed in Sec. III C. However, as shown in Sec. III D, no flux is transferred across the outer separatrix surface and any additional flux simply increases the volume bounded by the outer separatrix surface. When the coil current is switched off, the field-reversed topology is maintained by induced electron currents 关Fig. 5共d兲兴. These are not confined to the location of the coils but flow in an axially elongated layer. The two original O points merge into a single O-type null. During the relaxation, the 3-D null points slowly move away along the z axis 关Fig. 5共e兲兴, while the O-type null layer moves radially inward 关Fig. 5共f兲兴, thereby elongating the FRC. The elongation is explained in the following as due to the frozen-in condition. As the 3-D null points migrate outward, the axis of symmetry, â, is observed to tilt, e.g., by about 6° at t ⬘ ⫽3 s in Fig. 5共e兲. The symmetry axis is identified from the unit vectors of (0,B y ,B z ) at each time step, and the flux integral is performed along a line perpendicular to this axis. The tilting of the FRC is discussed in detail in Part II.13 As time progresses, closed field lines disappear without reconnecting. They annihilate as they converge into the neutral sheet. This is discussed in more detail in Sec. III E. The 3-D null points and the null line collapse near the center when the closed flux has eroded. However, when B⫽0, there are still strong plasma currents flowing which produce B⫽5 G opposite to B0 . In the absence of a null layer, these are electron Hall currents. When the Helmholtz coil field is reversed (BH 储 B0 ), the field line motion is as shown in Fig. 6. The initial field perturbation near the coils 关Fig. 6共a兲兴 penetrates rapidly into the plasma. Closed field lines around the coils reconnect at X-type null lines that lie outside the coils and, in this case, the measurement volume 共see Fig. 2兲. These carry small electron currents such that vacuum-like reconnection allows fast field penetration within t⫽5 s. The addition of the coil field compresses the magnetic field inside the coil in steady state 关Fig. 6共b兲兴. When the coil current is switched off, the closed field lines around the coils vanish, and whistler wave packets are emitted, as explained in Sec. III F. This reduces the compression and the field lines in the center of the coil straighten out 关Fig. 6共c兲兴. As the packets propagate, the field Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp 2786 Stenzel et al. Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 8. Absence of reconnection at the 3-D null points on axis. Flux contours ⌽(y,z)⫽const 共field lines兲 are shown at four consecutive time steps in the vicinity of the left 3-D null point during the early phase of the FRC relaxation. Field lines do not move through the null point but all lines convect together in the ⫺z direction. FIG. 7. Driven reconnection at the X-type null point located between the coils (z⫽0, r⯝r coil). The motion of selected field lines (⌽⫽const) is shown in four consecutive time steps. 共a兲 Closed and open field lines are separated by a crossing field line, the separatrix (⌽⫽1 kG cm2 ). 共b兲 Increasing flux forces the former separatrix into the open field line region, with a different field line (⌽⫽1.5 kG cm2 ) becoming the separatrix, a process that is repeated and 共c兲 and 共d兲. lines are compressed and develop the opposite curvature 关 z ⫽⫾25 cm, Fig. 6共d兲兴. Thus, magnetic energy is rapidly convected rather than dissipated. This case is similar to the extensively studied generation of a whistler wave packet by a single loop antenna.18 C. Driven reconnection During the rise of the coil current, opposing magnetic field lines move against one another at the X-type null line between the coils 关 r⯝r coil , z⫽0, Fig. 5共b兲兴. The motion of selected field lines is shown in Fig. 7 by snapshots of flux contours, ⌽(y,z)⫽const, at four consecutive time steps in a y–z plane at x⫽0. At t⫽7.8 s 关Fig. 7共a兲兴 after turn-on, the X-type line is clearly displayed. Note that the X line is not as in vacuum, because during the rise time of the coil current there is an additional component, B z,p , due to inductively driven plasma currents. At this selected time, the ⌽⫽1 kG cm2 contour defines the separatrix between the field lines that close around each coil (⌽⬎1 kG cm2 ) from those that close around both coils (⌽⬍1 kG cm2 ). At a later time 关 t ⫽8.8 s, Fig. 7共b兲兴, the field line defined by the ⌽⫽1 kG cm2 contour has reconnected and moved outside the separatrix now defined by ⌽⫽1.5 kG cm2 . At t⫽9.8 s 关Fig. 7共c兲兴, the separatrix is now defined by the ⌽⫽2 kG cm2 contour, which has moved into the null line and reconnected. Reconnection at the X line continues until steady state (dI coil /dt⫽0) is achieved at t⯝39 s, when the X-line location is identical to that of the vacuum case. For the time range displayed in Fig. 7, the flux at the X line (r⯝13 cm兲 changes by 2.5 kG cm2 , generating an in- ductive electric field of E ⯝⫺60 mV/cm along the X line. If the electrons were fully unmagnetized in the vicinity of the X line, this electric field would drive a J ⯝ 储 E ⯝1.5 A/cm2 . However, the total B field is strong enough to keep the electrons magnetized throughout the region. Calculation of J x , equivalent to J in this plane, from ⵜ⫻B 共not shown兲 indicates that the current is not localized at the X line, and that it flows across the field lines as a Hall current. Since it is spread over a large area, the current density at the X line is only J ⯝0.5 A/cm2 . This is not surprising since the separatrix field lines intersect at approximately 84°, which implies a nearly zero current density at the X line. If the field lines intersected at 90°, Maxwell’s equations (ⵜ•B⫽0) require that the current density be identically zero. Thus, the plasma currents reduce the plasma reconnection rate by 10% of the vacuum rate,21 and the field line dynamics are predominantly driven by the Helmholtz coil current. Driven reconnection, however, is not the main focus of the present work as it previously has been studied by many authors.11,22–26 D. Absence of reconnection at 3-D null points After the FRC has been established and a steady-state achieved, the Helmholtz coil is turned off and the relaxation of the FRC is observed. Because considerable interest has been shown in reconnection processes near 3-D null points in space plasma physics,27,28 where three types of reconnection processes have been identified, we turn our attention to the dynamics of the magnetic field lines near the cusp-type null points at (0,0,⫾24 cm兲 that form the axial boundary of the FRC. Figure 8 displays snapshots of flux contours ⌽(y,z) ⫽const at four consecutive time steps in the vicinity of the 3-D null point located at z⫽⫺24 cm immediately after the Helmholtz coil current begins its turn-off (t ⬘ ⫽0). Three key observations are 共1兲 the axis of the FRC exhibits a slight tilt, 共2兲 the null point and separatrix structure move together in the ⫺z direction, and 共3兲 magnetic field lines do not move through the null point. The FRC tilts and Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp Phys. Plasmas, Vol. 10, No. 7, July 2003 Fields, currents, and flows 2787 eventually processes because toroidal electron drifts cause the field lines to rotate, as discussed in Part II.13 The motion of the separatrix structure in the ⫺z direction is due to poloidal electron flows, v z , which carry the frozen-in magnetic fields axially away from the center of the FRC, but never toward it. This flow is examined later on in Sec. III G. Because no field lines move through the null point, there is no flux transfer across the separatrix, hence no reconnection. Furthermore, because the system is axially symmetric, the calculation of the flux contours demands that field lines inside the separatrix always be associated with positive flux contours while those outside are always negative 共see Sec. III F兲. It is not possible to have a positive flux contour pass through the separatrix, because there cannot be a positive contour outside. The situation here is not the same as the driven reconnection discussed in the previous section where flux inside and outside the inner separatrix are of one sign. Since the enclosed flux does erode in time 关i.e., the peak value of ⌽(y,z) inside the FRC decreases兴, it must be removed by a different process, namely annihilation. E. Magnetic annihilation at an elongated O-type null layer Magnetic reconnection of opposing antiparallel field lines has been termed field line ‘‘annihilation.’’4,29 This onedimensional limit may approximately apply to a very elongated 2-D X, Y, or O-type neutral sheet. The present FRC develops an oblate neutral sheet and the field lines are observed to be annihilated in it. Figure 9 displays snapshots of a single flux contour, ⌽(y,z)⫽600 G cm2 , at five consecutive time steps in the vicinity of the axially symmetric O-type null layer. Because of the symmetry, only the upper half plane of the y–z plane needs to be shown. This single flux contour 共or field line兲 is chosen for clarity and is characteristic of the other adjacent field lines which behave similarly. After the start of the free relaxation 关Fig. 9共a兲, t ⬘ ⫽3 s兴, the closed field line begins to stretch axially and contract radially resulting in the configuration of Fig. 9共b兲, t ⬘ ⫽6 s. By t ⬘ ⫽7 s, the field line assumes a very elongated loop 关Fig. 9共c兲兴, with adjacent sides of the loop forming opposing field lines. At t ⬘ ⫽7.5 s, the opposing lines begin to pinch inward at z⯝⫾10 cm and form three small islands, shown in Fig. 9共e兲 at t ⬘ ⫽8 s, and finally decay. The FRC topology changes by closed field lines annihilating at the elongated neutral sheet, not by reconnecting with the open field lines at the 3-D null points. The field lines evolve in this way because they are frozen into the electron fluid. The flows 共as shown in Sec. III G兲 are consistent with the predictions of EMHD theory. F. Convection of magnetic field Although the same initial magnetic field strength, B H , is applied to the plasma for both polarities, BH 储 ⫾B0 , the field relaxation rates and mechanisms appear to be quite different, judging from the magnetic topology observed. To investigate this difference, the poloidal magnetic flux due to magnetic field lines created by plasma currents alone is calculated. The contribution of the background field, B0 , and the Helmholtz FIG. 9. Magnetic field line tearing and annihilation at the elongated O-type null layer. A single flux contour ⌽(y,z)⫽600 G cm2 is followed in time in the upper y–z plane. 共a兲,共b兲 The closed field line stretches axially and contracts radially during the free relaxation. 共c兲,共d兲 The elongated field line moves into the null layer and 共e兲 tears into three small islands, which eventually vanish. All other flux contours behave similarly. coil field, BH , are ignored because they simply introduce offsets when added to that of the propagating plasma currents, ⌽ p (z,t)⫽ 兰 关 B z (r,z,t)⫺B 0 ⫺B H,z (r,z,t) 兴 2 r dr. The limit of the integral is forced to be the radial location where the z component of the plasma magnetic field changes sign. Choosing this limit prevents counting the return flux which, if fully included, would lead to a value of zero. Figure 10 shows the poloidal magnetic flux calculated in this fashion for both directions of B0 . For the BH 储 ⫺B0 case 关Fig. 10共a兲兴, the contours indicate that the flux is constrained to decay within the volume bounded by the slowly propagating separatrix. The constraint arises because the relaxing wave packet, which consists of whistlers, approaches the 3-D null points with decreasing velocity. The change in velocity is due to the dependence of the plane wave whistler group speed on the square root of the magnetic field, v g ⯝2(c/ pe )( ce ) 1/2. Thus the field stagnates at the null points. Because the expanding separatrix leaves the measurement volume, it is difficult to determine how much flux does get transmitted through it. However, it has been shown elsewhere that flux transmission through a stationary separatrix is extremely inefficient.30 On the other hand, the magnetic field residing outside the separatrix is free to travel away, and its reflection from the chamber end wall is observed at 10 s ⬍t ⬘⬍15 s. In contrast, the flux rapidly propagates away, v w⬇18 cm/ s, from the center of the Helmholtz coil for the BH 储 B0 case 关Fig. 10共b兲兴. This is consistent with the propagation first Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp 2788 Stenzel et al. Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 10. Poloidal magnetic flux, ⌽ p (z,t ⬘ ), in the z–t ⬘ plane showing the difference between convection and dissipation. 共a兲 For BH 储 ⫺B0 , the flux diffuses axially and decays mainly within the separatrix 共dashed line兲, while for the other case 共b兲, it propagates away from the end of the Helmholtz coil (z⯝⫺8 cm兲 and reflects at the chamber wall (z⯝⫺90 cm兲. 共c兲 Flux spatial decay in the presence of null points (BH 储 ⫺B0 ) is much stronger than for propagation in the whistler mode along open field lines (BH 储 B0 ). observed in the net field lines of Fig. 6. For the given parameters 关 B tot⬇20 G, n e ⬇1⫻1012 cm⫺3 , f ⬇1/(⌬t)⫽1/(3 s兲兴, the speed is in good agreement with that predicted from the whistler dispersion relation, v g ⬇30 cm/ s. Because the field seen by the packet decays to the ambient field, the speed of the reflected wave packet 共10 s ⬍t ⬘ ⬍15 s兲 is reduced accordingly. Additionally, reflection at the end wall disperses the wave packet changing its frequency spectrum and, therefore, its central group velocity. The magnitude of the contours indicates that the field created by the plasma is smaller for this case than when BH 储 ⫺B0 . The difference is understood when one examines the response of the plasma while fulfilling Faraday’s law: in the former case the changing magnetic field cannot propagate away and thus is maintained past the end of the coil current, while in the latter the field carries away the source currents as it travels. Figure 10共c兲 shows the spatial decay characteristics of the flux. It is obtained by plotting the peak value of ⌽ p (z,t) at each z position. To make a comparison between the two Helmholtz coil polarities easier, each case is normalized to the peak flux value at z⫽0, ⌽ p (0). For BH 储 ⫺B0 , the flux decays over a much smaller spatial scale length with an e-folding distance of 26 cm. The concentrated flux loss results in significant local electron heating and corresponding light emission that is not observed in the BH 储 B0 case.14 The BH 储 B0 case has a much longer decay distance of 62 cm, consistent with collisional damping seen in single-loopantenna generated whistler waves.18 FIG. 11. Schematic of fluid flows and frozen-in field open lines of a relaxing EMHD FRC. 共a兲 A toroidal electron flow, v tor , in a cylinder of finite axial length, rotates an axial field line B, equivalent to producing toroidal field components ⫾B tor where v tor / z⫽0. 共b兲 The source for the toroidal fields are two poloidal electron flows, ⫾ v pol , which convect the magnetic field lines axially and radially elongating and compressing the FRC. G. Field line twisting and stretching by fluid flows The observed magnetic field line dynamics can be interpreted as convections due to field lines frozen into electron flows. In EMHD, the electron flows and corresponding field line convection are coupled in ways that are not analogous to MHD. To provide a theoretical backdrop, a qualitative picture of what EMHD predicts for this experiment is presented, and the experimental results that quantitatively confirm the EMHD predictions are given. This is an important result because the parameter regime of the plasma gives a calculated classical electron mean free path that is similar in size to the magnetic field structures. This would seem to indicate that fluid theory is not applicable to this experiment. However, ion-acoustic turbulence, discussed in Part IV, decreases the electron mean free path permitting fluid theory to apply. The rise and fall of the coil current induces a plasma current that is primarily toroidally directed, consistent with Lenz’s law. Because all currents are carried by electron drifts in EMHD, the corresponding electron fluid flow speed is given by v ⫽⫺J/ne. The frozen-in condition of EMHD predicts that this flow sweeps the axial magnetic field in the toroidal direction. This is shown schematically in Fig. 11共a兲 for an open field line in the FRC case, i.e., a line outside the separatrix surface. Since the toroidal drift exists only near the Helmholtz coil, the field lines are only rotated in the center ( 兩 z 兩 ⬍ axial FRC length兲 but not elsewhere. This implies that near each end of the Helmholtz coil, toroidal field components of opposite signs are generated, one to produce the twist, the other to undo it. No net twist or helicity, H ⫽ 兰 A"B dV, where A is the magnetic vector potential, is generated. The toroidal field must be generated by an axially directed current which must close radially, i.e., a local poloidal current system, (J r , J z ), is formed. This current system Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp Phys. Plasmas, Vol. 10, No. 7, July 2003 Fields, currents, and flows 2789 FIG. 12. Measurement of electron flows and magnetic field during the free relaxation of the FRC. 共a兲 Toroidal electron flow, ( v x , v y )⫽⫺ⵜ⫻B/( 0 ne) in the central x – y plane. The electron flow inside the null layer (r⯝13 cm兲 is driven by the toroidal inductive electric field E ⬀⫺ B z / t, while the cross-field Hall drift is due to a radial space charge electric fields Er . 共b兲 Poloidal electron flow, ( v y , v z )(0,y,z). The axial flow convects the radial field component outward, the radial flow convects the axial field components radially inward. This produces a highly elongated FRC. 共c兲 Strong B x (0,y,z) produced by the toroidal twist at the ends of the FRC, B x ⬍0 for y⬎10 cm, z⯝⫺20 cm and B x ⬎0 for y⬍⫺10 cm, z⯝⫺25 cm. In the center of the FRC (z⯝0), B x is due to a tilt/precession 共Ref. 13兲. is centered at the radial position where B tor is formed, outside the separatrix. The corresponding poloidal flows, vpol , are sketched in Fig. 11共b兲. The effect of the axial flow is that it convects the radial fields of the FRC axially outward, while the radial flow convects the axial field lines radially inwards. This results in the observed elongation of the FRC shown in Figs. 5共d兲–5共f兲. Furthermore, the tearing into three islands 共Fig. 9兲 is likely caused by the radial inflows at two axial locations off center, z⯝⫾10 cm. Because the presented magnetic field measurement is an average (N⫽10) at each spatial location, it is clear that the position of the island formation is reproducible. The qualitative predictions of EMHD theory shown in Fig. 11 are supported by laboratory observations. The toroidal electron flow that carries the axial magnetic field lines in the toroidal direction is calculated from the current density, J⫽ⵜ⫻B/ 0 . Figure 12共a兲 shows vectors of vtor(x,y) in the z⫽0 plane. Outside the null layer (r null⯝13 cm兲, the flow is a Hall electron drift due to a radial space-charge electric field, vd ⫽Er ⫻zB z /B z2 , parallel to the inductive electric field E , while inside the null layer, the flow is antiparallel to E . Electron pressure gradient drifts, vdia ⫽ⵜ(nkT e )⫻B/(neB 2 ), cannot account for the observed drift, because they point in the opposite direction of the drift for B z ⬎0. The poloidal flow, vpol(y,z), that causes the axial expansion of the FRC and the island formation of the field lines is presented in Fig. 12共b兲 in the x⫽0 plane. The flows are peaked off-axis at the radius of the toroidal current layer. As predicted, the poloidal flow/current creates the out-ofplane toroidal field component shown in Fig. 12共c兲. Note that B ⫽⫺B x for y⬎0 and B ⫽B x for y⬍0. The contour plot shows that B x is largest near the outer edges of the y–z plane (z⬇⫾20 cm, y⬇⫾12 cm兲, i.e., near the ends of the FRC, and changes sign with y and z. Thus, the electron flow along the separator creates a similar quadrupole-like structure as theoretically predicted for a 2-D X-type null point.3,31,32 In the center of the FRC, the B x component is caused by a different effect, i.e., a tilt of the FRC.13 The toroidal field modifies the structure of the field lines that intersect 3-D null points. In the absence of a toroidal field, the null point is classified as a simple radial cusp-type null, otherwise it is classified as a spiral null point. Detailed definitions of these terms and how null points are quantitatively classified are given in Ref. 20. A few characteristic field lines are traced in Fig. 13 near the null point at z ⯝⫺30 cm. Closed field lines along the spine flare out in a fan to form part of the separatrix surface. Open field lines just outside the separatrix 共not shown for clarity兲 have the same fan lines but oppositely directed spine lines. The twist of the fan lines is direct evidence of the toroidal field, B , produced by field line rotation in the toroidal electron flow. Since the twist increases with radius, the fan lines are purely radially directed near the axis, but deviate from the radial direction with the toroidal displacement growing radially. The field line twist/helicity density changes from lefthanded/negative inside the separatrix to right-handed/ positive outside the separatrix. The same holds true for the other null point at z⯝30 cm, so the entire FRC has no net FIG. 13. Measured field topology of a 3-D spiral null point of the relaxing FRC. The field lines are traced from the inner spine to the left fan assuming / ⫽0. The twist of the fan lines is a manifestation of B and v z . Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp 2790 Stenzel et al. Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 14. 共a兲 Schematic picture of an axial magnetic field line frozen into the toroidal electron flow of the decaying field with BH 储 B0 . The field line rotation creates toroidal field components ⫾B where v tor / z⫽0. 共b兲 Measured toroidal field ⫾B x (0,y,z) and poloidal currents Jpol⫽(J y ,J z ) during the free relaxation of the field shown in Figs. 3共c兲 and 3共d兲. helicity. During the turn-on of the coil current, the toroidal and poloidal electron drifts are reversed from those during current turn-off. The axial inflow now creates a concave separatrix surface near the 3-D null points such that the FRC becomes oblate 关Fig. 5共c兲兴 compared to the prolate shape in steady state 关Fig. 5共d兲兴 or decay 关Fig. 5共f兲兴. When the Helmholtz coil field is parallel to the uniform field, the drifts are reversed from that of the FRC. Figure 14共a兲 presents a schematic picture of the field line twist and Fig. 14共b兲 shows data of the resulting out-of-plane toroidal field component B and poloidal currents (J y ,J z )(y,z). The toroidal electron drift rotates the field lines, a radial drift compresses the field lines, and an axial drift stretches them. Toroidal and poloidal drifts are self-consistently coupled resulting in 3-D field topologies in EMHD. During the current rise, the axial elongation and radial pinching of the field perturbation induces the emission of two propagating current vortices. As the coil current is turned off, a radial electron outflow is established near the center of the coil due to E ⫻B drifts.14 This outward flow is balanced by inward flows axially away from the coil. As the drift region expands axially a compression front moves away from the coil center, causing the field lines to bulge, as observed in Fig. 6共d兲. The result is a propagating displacement of the magnetic field lines transversely to the background field, a characteristic of propagating whistlers and Alfvén waves.19 A quantitative check of the frozen-in concept for both orientations of B is presented in Fig. 15. The toroidal displacement of the electron fluid in the axial center of the relaxing FRC (z⫽0) is determined from s v (t)⫽ 兰 v tor dt ⬘ FIG. 15. Comparison of the toroidal displacements s(t) of the electron fluid and the magnetic field lines to check the frozen-in condition. 共a兲 During the relaxation of the FRC, the fluid rotation and field line rotation assume the same values after current switch-off (t ⬘ ⬎3 s兲. 共b兲 During the turn-on of the coil current with BH 储 B0 , the growing angular displacements of the fluid and the field lines agree when the whistler vortex is located within the measured volume. ⫽⫺ 兰 t0 J tor /(ne) dt ⬘ and plotted versus time in Fig. 15共a兲. At the chosen radius (r⯝12 cm兲, the fluid rotates by s v ,max ⯝15 cm or ⯝72°. The rotation is limited because the toroidal current moves in time radially inward. The net field line rotation at the same radius is determined by s B (t) z ⫽ 兰 ds⫽ 兰 0max(B tor /B z ) dz ⬘ . After the coil current has vanished (t ⬘ ⬎ 3 s兲, the field line is displaced by essentially the same distance as a fluid element. For the case of BH 储 B0 and during the turn-on of the coil current, the angular displacements of the fluid and the field lines grow in time 关Fig. 15共b兲兴. There is good agreement as long as the propagating whistler vortex remains within the measurement volume. After the propagating twist B tor leaves the region of observation, the field line displacement cannot be accounted for. Within these constraints, we conclude that the magnetic field is frozen into the electron fluid. H. Neutral sheet currents versus Hall currents Reconnection is associated with the formation of current sheets in regions of magnetic neutral sheets. The thickness of the current sheet is thought to be of order L, the scale length on which the magnetic Reynolds number is R m ⫽ 0 v AL ⯝1, where v A is the Alfvén speed in MHD plasmas. For the classical conductivity, Spitzer , this value leads to unphysically small values in collisionless plasmas. Anomalous resistivity, inertial, and Hall effects are thought to broaden the current sheet but the topic is still under investigation. Recent laboratory experiments have shown that current sheets in MHD plasmas assume a thickness comparable to the ion inertial scale, c/ pi . 24,33,34 Much less is known in EMHD Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 16. 共a兲 Toroidal electron current density, J x , in the center of the relaxing FRC vs radius y and time t ⬘ . Current layer shrinks radially in time like the magnetic null layer 共dashed black contour兲, but current does not exactly peak where B⫽0. 共b兲 Corresponding data for the case BH 储 B0 without null points. Hall current rapidly decays in the central plane (z⫽0) due to the emission of two whistler vortices in ⫾B0 direction. plasmas about the current sheet thickness. By analogy to MHD, one would expect for the present parameters a sheet thickness of either L⯝( 0 v ) ⫺1 ⭐ 1 mm, for v ⫽ v whistler ⯝10 cm/ s and ⫽ Spitzer⫽150 ⍀ ⫺1 cm⫺1 (kT e ⯝4 eV兲,16 or the electron inertial length c/ pe ⯝1 cm at n e ⯝5⫻1011 cm⫺3 . Neither of these values is observed in the present experiment. From the measured magnetic field, the current density is obtained via Ampère’s law. The dominant component is given by J x ⫽( B z / y⫺ B y / z)/ 0 . The peak of J x usually occurs at a smaller radius than that for B z ⫽0. Figure 16 presents contour plots of J x versus radius and time for both polarities of BH . For the FRC 关Fig. 16共a兲兴, the null layer 共dashed black contour兲 initially expands beyond the individual coils at t ⬘ ⯝2 s. The expansion of the null layer coincides with the rapid field decay near the coils seen in Fig. 3. In time, the peak current density and the null layer move radially inward and converge. For t ⬘ ⬍10 s, the peak current density remains inside the null layer during the convergence. Because the peak current flows in the magnetized region of the plasma 共not in the neutral layer兲 and it is perpendicular to the field lines, the current is predominantly a Hall current. This is the current mentioned in Sec. III A that satisfies the force-free condition of EMHD and accounts for the stability of the FRC structure. Despite this stability, the FRC’s magnetic moment oscillates around the z axis, causing the decrease and subsequent increase of J x observed at t ⬘ ⯝6 s in Fig. 16共a兲. The oscillation is discussed in Part II.13 For the opposite polarity (BH 储 B0 ) shown in Fig. 16共b兲, the toroidal current profile exhibits markedly different behavior. The currents form two whistler wave packets which propagate away from the Helmholtz coil region in the ⫾B0 direction. It was mentioned in Sec. III A that the whistler transit time is about w ⬇3 s. This time scale is consistent with the time it takes for the majority of J x to vanish in the figure. Fields, currents, and flows 2791 FIG. 17. 共a兲 Radial profiles of the axial field component B z (y) at x⫽z ⫽0, t ⬘ ⫽6 s for both polarities of BH . Note the field reversal for BH 储 ⫺B0 and the absence of null layer and faster relaxation for BH 储 B0 . 共b兲 Radial profiles of the toroidal current density, J x (y)⯝ ⫺1 0 B z / y. For the FRC, the current is not localized in the magnetic null layer but flows adjacent to it as a Hall current across field lines. The widths of the current ‘‘sheets’’ with or without the magnetic neutral layer are comparable, ⌬y ⯝7 cm⯝7c/ pe . Although the time evolution of the current profile appears quite different for the two cases, the fact that both current systems are Hall currents leads to similar spatial profiles in J x . To see this, radial profiles of the dominant magnetic field component B z (y) and J x (y) at x⫽z⫽0 and t ⬘ ⫽6 s are shown in Figs. 17共a兲 and 17共b兲, respectively. In the FRC case, the magnetic null layer is seen at y⫽⫾10 cm. For BH 储 B0 , there is no sign change in the magnetic field, but B H,z has decayed more rapidly due to the vortices convecting fields away from the Helmholtz coil region. Figure 17共b兲 shows the radial profiles of the current density, J x (y), scaled differently for the two configurations. In both cases, the current ‘‘sheet’’ has a full width at half maximum of ⌬y⯝7 cm, which is much broader than the electron inertial length c/ pe ⯝1 cm or the electron cyclotron radii, r ce ⯝1 cm (kT e ⫽3 eV, B⫽5 G兲. There is no enhancement of the current in the neutral layer for the FRC case. The bulk current, then, is a Hall current flowing perpendicular to the field lines. In the FRC case, the current density yields an electron drift velocity of v d,e ⫽J/n e e⯝107 cm/s, which is small compared to the electron thermal velocity ( v th,e ⯝108 cm/s兲 but large compared to the ion sound speed 关 c s ⯝(kT e /m i ) 1/2 ⯝3⫻105 cm/s兴. With T e ⰇT i , the electron drift excites current-driven ion sound instabilities. As mentioned previously, the Part IV companion paper15 presents measurements of the turbulence, where it is proposed as the mechanism for the anomalous conductivity and the observed decay time that is much shorter than the classical diffusion time. The current density, J x 共calculated from the curl of B), is integrated to find the total toroidal current, I tor⫽ 兰 J x dydz. Its time dependence is shown in Fig. 18 for both polarities of BH . The coil current and the plasma currents are both measured. During the quasi steady-state period of I coil (t⯝40 Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp 2792 Stenzel et al. Phys. Plasmas, Vol. 10, No. 7, July 2003 FIG. 18. Net toroidal current 共coil and plasma兲 vs time t. During turn-on, the coil current is opposed by induced electron currents, resulting in a slower current rise than in vacuum. After turn-off of the coil current, plasma currents of comparable strength are induced. Time variations of plasma currents are slower in the FRC configuration than for BH 储 B0 . s兲, the measured current agrees well with that in the individual coils. During turn-on (0⬍tⱗ40 s兲, the coil current is opposed by induced electron currents, resulting in a slower net current rise than in vacuum. After turn-off of the coil current, plasma currents of comparable strength are induced and decay in time as the field relaxes. Time variations of plasma currents are slower in the FRC case because the magnetic field is trapped between null points. For BH 储 B0 , the field propagates out of the volume, which accounts for the faster current decay. The secondary peak at t⯝57 s is caused by the reflection of the wave packet from the chamber wall, which is about ⌬z⯝45 cm away. This will be discussed in depth in Part III.14 I. Electric fields Electric fields in EMHD are both due to induction and space charges.35 The inductive electric field is due to the decay of the axial magnetic field, E ⬀⫺ B z / t. We have calculated E ⫽⫺(d⌽/dt)/(2 r) from ⌽(x,y) where r⫽ 关 (x⫺x c ) 2 ⫹(y ⫽ 兰 r0 dr ⬘ 兰 20 B z (r ⬘ , ⬘ ) d ⬘ , 2 1/2 ⫺y c ) )] and (x c ,y c ) is the center of the FRC, based on the observed symmetry ( / ⯝0). For the FRC case we find that the toroidal inductive electric field increase radially and peaks at a larger radius (r⯝15 cm兲 than the neutral sheet (r⯝10 cm兲 where the flux assumes a maximum. Due to axial symmetry there are no space charge electric fields or pressure gradients in the toroidal direction. In the neutral sheet E drives the toroidal current J . Outside the neutral sheet the inductive field produces a radial electron Hall drift, v r ⫽E /B z , directed into the neutral sheet, which carries frozen-in field lines into it. The essentially incompressible electrons (ⵜ•J⯝0) cannot stagnate in the null layer, and they flow axially outward at the ends of the null layer forming two poloidal current loops (J r ,J z ). A rotational electron flow and a diverging energy flow 共Poynting’s vector is parallel to the radial E⫻B drift兲 imply that the frozen-in condition is broken near the null layer. The radial electric field can be calculated from the EMHD force law, ⫺neE ⫹J⫻BÀⵜp e ⫽0, where p e ⫽nkT e is the electron pressure. It will be shown in Part III14 that there is a radial electron pressure gradient due to a density maximum on axis. The electron temperature peaks in the current sheet but ⵜp e remains radially inward. Since J flows in the direction and B reverses sign across the neutral sheet, the radial space charge electric field must change sign. The peak radial electric fields and pressure gradients at z⫽0 are of order E r ⬇ 兩 ⵜ p e /ne 兩 ⱗ0.4 V/cm. For purpose of comparison, we have also calculated E for the case BH 储 B0 . In the absence of a magnetic null the electrons remain magnetized everywhere. Since BH is now reversed, the direction of E is also reversed but not the radial electron E ⫻Bz drift. The latter gives rise to the poloidal current loop of a whistler vortex. The radial spacecharge electric field points outward and is of similar magnitude as the inductive field. There are no significant radial pressure gradients. The toroidal and poloidal Hall currents are found to be of comparable magnitude. Finally, it should be pointed out that especially in the case BH 储 ⫺B0 , there must be a space-charge electric field E z along the nonuniform axial magnetic field B z (z), so as to balance the magnetic force on the electrons, ⫺ e ⵜB⫺eE ⯝0, where e ⫽m e v⬜2 /2B is the electron magnetic moment. The potential difference between the null points and the center of the FRC should be of order kT e /e⯝4 V. The unmagnetized ions will respond to this electric field, but their displacement and energy gain are small due to the short time scale involved, ⌬z⫽(eE/2m i )(⌬t) 2 ⯝0.5 cm and eE⌬z ⯝0.25 eV, for E⯝0.5 V/cm, ⌬t⯝10 s. IV. CONCLUSIONS This first part of a series of articles on EMHD reconnection describes the magnetic field topologies of a time-varying Helmholtz coil field inside a uniformly magnetized background plasma. Of particular interest is the period of free relaxation after the rapid switch-off of the coil currents. Two configurations are compared: First, with the dipole moment opposite to the weaker background field, an EMHD FRC is produced. Second, with the dipole moment along the background field, a large amplitude whistler vortex is generated. The relaxation of the fields in these two cases is different. The FRC fields stagnate and dissipate while the vortex fields propagate. The motion of magnetic field lines in the presence of electron fluid flows is discussed. Frozen-in, axisymmetric field lines are defined by contours of constant poloidal flux. Driven reconnection near an X-type null line, absence of reconnection near a 3-D spiral null point, and annihilation of field lines near an elongated O-type null line are deduced from the motion of field lines near magnetic null points. Annihilation is responsible for the disappearance of closed field lines in a freely decaying FRC. Field line annihilation in O-type nulls implies a breakdown of the frozen-in condition since the highly incompressible electrons cannot have a divergent flow, just like the field lines. The frozen-in concept has been verified experimentally in regions of magnetized electrons. For example, the poloidal field in a relaxing EMHD FRC is associated with a toroidal electron drift which twists the frozen-in field lines and creates toroidal fields or poloidal currents/drifts. These convect the poloidal fields, produce an elongated FRC and drive electron tearing modes. The coupling of toroidal and poloidal electron flows produces, in general, 3-D magnetic fields with helicity density Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp Phys. Plasmas, Vol. 10, No. 7, July 2003 but not necessarily net helicity. The toroidal electron current in an FRC is not localized within the magnetic neutral sheet but broadened by Hall currents outside the null layer. This is an important difference to 2D numerical simulations which predict electron current sheets of width c/ pe . 5 The forcefree nature of EMHD fields may account for the observed stability of the FRC to tilting. ACKNOWLEDGMENT The authors gratefully acknowledge support for this work from NSF Grant No. 0076065. R. G. Giovanelli, Mon. Not. R. Astron. Soc. 107, 338 共1947兲. Magnetic Reconnection in Space and Laboratory Plasmas, edited by M. Hoshino, R. L. Stenzel, and K. Shibata, University of Tokyo 共Terra, Tokyo, Japan, 2001兲, special issue of Earth, Planets and Space 53 共2001兲. 3 A. S. Kingsep, K. V. Chukbar, and V. V. Yan’kov, in Reviews of Plasma Physics 共Consultants Bureau, New York, 1990兲, Vol. 16, p. 243. 4 V. M. Vasyliunas, Rev. Geophys. Space Phys. 13, 303 共1975兲. 5 J. Birn, J. F. Drake, M. A. Shay, B. N. Rogers, R. E. Denton, M. Hesse, M. Kuznetsova, Z. W. Ma, A. Bhattacharjee, A. Otto et al., J. Geophys. Res. 106, 3715 共2001兲. 6 E. R. Priest, Phys. Plasmas 4, 1945 共1997兲. 7 J. Drake, D. Biskamp, and A. Zeiler, Geophys. Res. Lett. 24, 2921 共1997兲. 8 D. Biskamp, E. Schwarz, A. Zeiler, A. Celani, and J. F. Drake, Phys. Plasmas 6, 751 共1999兲. 9 M. R. Brown, Phys. Plasmas 6, 1717 共1999兲. 10 M. Yamada, Earth, Planets and Space 53, 509 共2001兲. 11 R. L. Stenzel and W. Gekelman, J. Geophys. Res. 86, 649 共1981兲. 12 M. F. Heyn and M. I. Pudovkin, J. Plasma Phys. 49, 17 共1993兲. 13 M. C. Griskey, R. L. Stenzel, J. M. Urrutia, and K. D. Strohmaier, Phys. Plasmas 10, 2794 共2003兲, Part II. 1 2 Fields, currents, and flows 14 2793 J. M. Urrutia, R. L. Stenzel, M. C. Griskey, and K. D. Strohmaier, Phys. Plasmas 10, 2801 共2003兲, Part III. 15 R. L. Stenzel, M. C. Griskey, J. M. Urrutia, and K. D. Strohmaier, Phys. Plasmas 10, 2810 共2003兲, Part IV. 16 NRL Plasma Formulary, edited by J. D. Huba 共Naval Research Laboratory, Washington, DC, 1990兲, p. 29, NRL Publ. No. 6790-98-358. 17 R. L. Stenzel, M. C. Griskey, J. M. Urrutia, and K. D. Strohmaier, Phys. Rev. Lett. 88, 185004 共2002兲. 18 C. L. Rousculp, R. L. Stenzel, and J. M. Urrutia, Phys. Plasmas 2, 4083 共1995兲. 19 H. Alfvén, Ark. Mat., Astron. Fys. 29B, 2 共1942兲. 20 C. E. Parnell, J. M. Smith, T. Neukirch, and E. R. Priest, Phys. Plasmas 3, 759 共1996兲. 21 M. C. Griskey, Ph.D. thesis, University of California, Los Angeles, 2002. 22 A. G. Frank, Plasma Phys. Controlled Fusion 41, A687 共1999兲. 23 P. J. Baum and A. Bratenahl, J. Plasma Phys. 18, 257 共1977兲. 24 Y. Ono, A. Morita Katsurai, and M. Yamada, Phys. Fluids B 5, 3691 共1993兲. 25 M. Yamada, J. Geophys. Res. 104, 14529 共1999兲. 26 J. Egedal and A. Fasoli, Phys. Rev. Lett. 86, 5047 共2001兲. 27 M. Hesse and J. Birn, IEEE Trans. Plasma Sci. 28, 1887 共2000兲. 28 D. S. Brown and E. R. Priest, Astron. Astrophys. 367, 339 共2001兲. 29 W. I. Axford, in Magnetic Reconnection in Space and Laboratory Plasmas, edited by J. E. W. Hones 共American Geophysical Union, Washington, DC, 1984兲, No. 30 in Geophys. Monograph, pp. 1– 8. 30 M. C. Griskey and R. L. Stenzel, Phys. Plasmas 8, 4810 共2001兲. 31 M. E. Mandt, R. E. Denton, and J. F. Drake, Geophys. Res. Lett. 21, 73 共1994兲. 32 P. L. Pritchett, J. Geophys. Res. 106, 3783 共2001兲. 33 M. Yamada, H. Ji, S. Hsu, T. Carter, R. Kulsrud, N. Bretz, F. Jobes, Y. Ono, and F. Perkins, Phys. Plasmas 4, 1936 共1997兲. 34 T. W. Kornack, P. K. Sollins, and M. R. Brown, Phys. Rev. E 58, R36 共1998兲. 35 C. L. Rousculp, R. L. Stenzel, and J. M. Urrutia, Phys. Rev. Lett. 72, 1658 共1994兲. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp