* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Flow Measurements on Single and Merging Spheromaks atSSX
Magnetic field wikipedia , lookup
Quantum vacuum thruster wikipedia , lookup
Neutron magnetic moment wikipedia , lookup
Time in physics wikipedia , lookup
Condensed matter physics wikipedia , lookup
Magnetic monopole wikipedia , lookup
Aharonov–Bohm effect wikipedia , lookup
Lorentz force wikipedia , lookup
Woodward effect wikipedia , lookup
Electromagnet wikipedia , lookup
Superconductivity wikipedia , lookup
Flow Measurements on Single and Merging Spheromaks atSSX Jason Horwitz Swarthmore College Department of Physics & Astronomy March 19, 2007 2 Abstract Results are presented from measurements of flow on spheromaks at the Swarthmore Spheromak Experiment. Mach probes of two different sizes, corresponding to two different methods of analysis, are used to make these flow measurements. Flows of as high as 70 km Is are detected in the single spheromak, as well as subsequent reversals of flow toward the edges of the spheromak. There is little flow perpendicular to the magnetic fields in a single spheromak, confttming that SSX spheromaks exhibit the property of frozen-in flux. Measurements during counter-helicity merging of two axially aligned spheromaks tentatively confttm the presence of bidirectional jets due to reconnection in the azimuthal direction. The azimuthal component of this outflow reaches speeds of 10 kmI s during the expected time frame for reconnection. 3 4 Contents 1 Introduction 1.1 Previous Lab Results in Flow and Stability 1.2 Previous Lab Results in Flow and Magnetic Reconnection 7 8 10 2 Plasma Theory 2.1 Frozen-in Magnetic Flux 2.2 The Induction Equation 2.3 The Alfven Speed 2.4 Magnetic Reconnection Basics 2.5 Sweet-Parker Reconnection 2.6 The Spheromak 13 15 17 19 20 21 24 3 Mach Probe Theory 3.1 The Hudis-Lidsky Model 3.2 Problems with the Hudis-Lidsky Model 3.3 Magnetized Mach Probe Theory 27 4 The Experiment 4.1 Making a Spheromak 4.2 Mach Probe Specifics 39 41 43 5 Results 5.1 5.2 5.3 5.4 47 47 51 54 57 Calibrating the Mach Probe Visualizing Mach Probe Data Single Spheromak Results Counter-Helicity Merging Results 30 33 35 6 Interpretation 6.1 Single Spheromak Flow 6.2 Flow During Counter-Helicity Merging 61 62 62 A Glossary B Argument for Flow in MHD Equilibria 65 67 References Acknowledgements 69 71 5 6 Chapter 1 Introduction Plasmas are the most prominent state of visible matter in our universe. A gas becomes ionized and becomes a plasma when it has enough energy such that electrons have been stripped off of ions in the gas. The plasma is a complex dynamic state, in which charged particles are immersed in, moving under the influence of, and therefore constantly regenerating a macroscopic electromagnetic field determined by the positions and velocities of other charged particles in the plasma. Key parameters, such as density, ion temperature, electron temperature, pressure, current, and drift are related in ways that, at a macroscopic level, are often complex and heavily dependent on the geometry of the plasma or the boundary conditions restraining the plasma. Nevertheless, as those interested in alternative energy sources look forward to a functional fusion reactor, scientists the world over are making efforts to create and better understand stable plasma formations in the laboratory. One essential component for understanding the stability and general dynamics of a plasma is flow, or average drift velocity. Specifically, there are two respects in which plasma flow is particularly significant for the current trajectory of plasma research. The first is the case of stability. It has been shown that there is a relationship between stability and the shear of flow in certain plasma formations. These relations have been used to make marked progress in the creation of hotter, longer-lasting plasmas. Perhaps this is most apparent in the case of H-mode tokamaks. The characteristics of this mode include higher plasma energy and longer particle confmement times. The transition to this higher confmement mode is accompanied by higher rotation of the plasma at its edge, along with some changes in electricity and density [1]. Secondly, there is the case of magnetic reconnection. Magnetic reconnection, to be described in greater detail later, is a phenomenon in which two areas of plasma with opposing magnetic field lines collide, resulting in 7 the annihilation of the field lines along with bursts of particle acceleration and heating. Currendy accepted models of magnetic reconnection predict high velocity bursts of outflow from the reconnection region, and the ability to quantify this flow has profound consequences for the interpretation of energy transport during this dynamic process. 1.1 Previous Lab Results in Flow and Stability Over the past decade, plasma theory has taken a turn towards a magnetohydrodynamic analysis of two intertwined fluids, corresponding to differing behavior between a general electron fluid and a general ion fluid. Equilibrium analysis of flows in a two-fluid plasma yields the result that plasma in a stable state requires the presence of significant sheared flows [2]. The Helically Symmetric Experiment (HSX) at the University of Wisconsin at Madison provided one of the first verifications for the plausibility of this two-fluid analysis. Using two separate Mach probes with a Gundestrup design (the same diagnostic that will be used for the majority of results in this thesis), the scientists at HSX verified that flows in their stellarator, in the presence of a rising current, evolved on two separate time scales, one at 300 ~s and the other at 20 ~s or less [3]. These separate time scales agree with neoclassical predictions dependent on two-fluid motion. e <:I .o:z. rE> e e 0 Figure 1.1 A schematic for the z-pinch at two separate times. Plasma is forced out from the sides of an electrode by magnetic pressure to eventually converge into a straight, pinched line of flowing hot plasma. Figure from http://www.aa.washington.edu/ AERP IZaP One particular plasma formulation in which shear flows are relevant is called the z-pinch, shown in Figure 1.1. The Zap Flow Z-Pinch lab at the University of Washington has just such a setup, and they have performed experiments to determine flow by measuring the Doppler shift of impurity lines along several axial chords. During the brief period of stability in the z-pinch, they found that flows at outer radii reach a maximum of approximately 110 8 km Is> while flows toward the center of the machine are closer to 50 106 Sl. I km s . The axial velocity shear dVr I dr during this period is 5 x After this quiescent period, the shear notably drops to 3 X 106 Sl and becomes unstable [4]. Another plasma configuration is the centrifugally confined plasma, such as that produced by the recently-built apparatus at the Maryland Centrifugal Experiment (MCX) at the University of Maryland. A centrifugally confined plasma maintains stability by rapid rotation in the presence of a constant axial magnetic field and radial electric field. This is an example of flows due to the presence of a non-zero E x B component in the plasma. Measurements were made of the rotational velocity of the plasma along 5 different radial chords. Rotational flow was found to be 20 the center of the chamber. At r=15 cm, rotational flow grows to as high as 70 back to 20 km Isnear the outside of the machine. km km Isnear Is> only to drop MCX also found that axial velocity shear was as high as 8 x 105 s\ concluding that the areas of high rotational flow contain a stabilized plasma [5]. These data imply that the bulk of centrifugally confined plasma is in the region of higher rotational flow with little exchange with the other regions of the chamber (due to such high shear). Figure 1.2. A diagram showing the experimental setup at the Maryland Centrifugal Experiment (MCX). Note that the E x B force is coming out of the page here, and therefore we expect the plasma to flow in this direction, resulting in a rotation. Figure from [6]. A configuration similar to the stellarator and tokamak described earlier, the reversed field pinch (RFP), consists of a toroid of plasma containing both toroidal and poloidal magnetic fields, driven by an exterior coil. The EXTRAP T2R device in Stockholm is capable of producing such a configuration, and measurements were performed on an RFP using the same diagnostic as in the zpinch and MCX, Doppler spectroscopy of impurity ions. In the case of RFPs, fast rotation is necessary to avoid loss of energy to the surrounding wall. This lab found that, towards the center of a stable RFP, ions have a rotational velocity of up to 40 velocity reduces to a mere 10 km I s [7]. 9 km Is> while towards the edge, the rotational 1.2 Previous Lab Results in Flow and Magnetic Reconnection In magnetic reconnection, as briefly described earlier, we see the dynamic result of sudden annihilation of magnetic fields in plasma. In a process that will be described in detail later, during magnetic reconnection, particles are accelerated outward perpendicular to the incoming fields , as in Figure 1.3. « Figure 1.3 A simplified 2-D depiction of magnetic reconnection. Opposed magnetic fields enter the figure from the top and bottom. As they break at the X point in the center, plasma is sent out the sides in the same plane as the original magnetic fields. A simplified theory, called the Sweet-Parker theory, predicts that this outflow will be equivalent to a velocity dependent on the magnetic field, referred to as the Alfven velocity. The specifics of this theory will be explained in more detail later. Observations at the Princeton Physics Plasma Laboratory in the Magnetic Reconnection Experiment (M~'0 showed that this was not a satisfactory model. Measurements on local flow near the reconnection region by a Mach probe showed equivalent opposing outflows of a mere 8 km Is> compared to a calculated Alfven velocity of 39 km Is. The PPPL attributed this difference to the presence of a large amount of electron pressure just outside the reconnection region, which inhibited the outflow [8]. Finally, there is a plasma formation used to induce reconnection in the laboratory that does not exhibit this buildup of electron pressure: the spheromak. Flows on the order of the predicted Alfvenic outflow velocity were measured during the reconnection of two spheromaks at the University of Tokyo. Ion Doppler spectroscopy data showed outflow velocities of 12 the Alfvenic speed was determined to be as low as 15 km Is[9]. km Is> whereas It will become clear shortly what a spheromak is and how the Swarthmore Spheromak Experiment (SSX) replicates reconnection in the laboratory. 10 N ow that the importance of flow measurements in the study of magnetically confmed plasmas has been contextualized, the next few sections of this thesis will introduce the reader to some basic concepts about plasma and to some of the more complex theory behind local measurements of flow. After these theory sections, I describe the specific plasma chamber at SSX and the relevant diagnostics used during the period of time in which data was collected for this thesis. Finally, results are presented concerning flow measurements taken on single spheromaks as well as merging spheromaks, and the implications of these results are interpreted and discussed. With the awareness that plasma physics tends to have a vocabulary all its own, I have included a glossary as an appendix following the fmal sections in case the reader is unfamiliar with some of the terms presented in this thesis. Another appendix summarizes a technical argument not covered in the body of this thesis. 11 12 Chapter 2 Plasma Theory Determining the equation of motion for anyone particle in a plasma is a simple task. Gravitational forces can be ignored in the presence of strong electromagnetic fields, so Newton's law and the Lorentz force law yield dv - - m- = q(E+vxB). dt (2.1) Together with Maxwell's equations, which determine how the electric and magnetic fields change according to the motion of charged particles, the motion of each particle is completely determined. Unfortunately, these equations would have to be calculated for every particle-particle interaction to microscopically analyze plasma motion, a task that even the fastest of computers cannot accomplish with any appreciable amount of particles. Rather, plasma theory has resorted to the use of models to approximate the behavior of plasmas assuming certain conditions. A relevant model for predicting flows is one referred to as magnetohydrodynamics (MHD). This model treats plasma as an electrically conducting fluid. Instead of analyzing the motions of individual particles, the MHD model describes plasma dynamics with parameters such as pressure, density, or flow velocity. Most modern formulations of MHD actually describe the plasma as two separate fluids, an ion fluid and an electron fluid. The following derivation was inspired by Falk [10] and Chen [11]. It is reasonable to assume that fluids consisting of these two different types of particles will behave differently, not only because they have different charges, but also because there is a significant 13 difference between their masses. When a conducting fluid of protons is in motion relative to a fluid of electrons, there is an effective current density. If we defme the flow velocity iT = Vi - ve , then - = enV,- where e is the charge of a proton and n=ne=n; is the proton, or this current density is J electron, density. In this context, the Lorentz force law still applies to each of the fluids, but we can consider it to be a force on a section of charged fluid immersed in an electromagnetic field. The charge q of this section of fluid will be proportional to the product of the density of particles in the section n and the charge of those particles e, so the force on the fluid is - FLorentz = en(E- + V- X B). (2.2) There is a second force, according to resistive MHD, that works against this Lorentz force. For simplicity, assume that the bulk of resistance to acceleration in the plasma is due to collisions between particles. Electron-electron and ion-ion collisions can be ignored because they conserve momentum, so only electron-ion collisions contribute to this resistivity. The collisional coupling will be proportional to e2 and n 2 because it is happening between two charged particles. It will also be proportional to the relative velocity of ions and electrons defmed earlier. The resistive force on a section of fluid in a plasma is F',esistive 2 = 1]e n 2V (2.3) Assuming that the only significant forces on a stable, charged fluid are the two opposing ones shown in (2.2) and (2.3) and adopting the defmition of current density written above, we gain the following equation of motion: - - - - E+ VxB=1]J, (2.4) often referred to as the resistive Ohm's law. The approximation that the only effective forces on a plasma are the Lorentz force and a resistive force is called the resistive MHD approximation. The proportionality constant that determines the extent to which ion-electron collisions contribute to the resistive force is the essential resistivity of the plasma, 11. There can be an even further approximation made under the name ideal MHD, which assumes that the fluid is perfectly conducting, or that 11=0. In this case, (2.4) reduces to the ideal Ohm's law: - - - E+VxB=O. 14 (2.5) Note that the flow velocity in directions perpendicular to the field are immediately determined by the ideal Ohm's law: V..l ExB (2.6) =--2- B 2.1 Frozen-in Magnetic Flux One of the most appealing results from ideal MHD is the idea of frozen-in magnetic flux. This approximation implies that the magnetic field lines move along with the plasma itself, or, some would prefer, the plasma sticks to the magnetic field lines as it moves. A proof attributed to Seshadri [12] begins by integrating the divergence of the magnetic field over a volume of plasma V bounded by the closed surface S*. According to Maxwell's equations, the divergence of a magnetic field is always zero, so, invoking the divergence theorem yields the equation fV.B(r,t)dV= f B(r,t)·dS*=O v (2.7) S* N ow, introduce a closed curve C which bounds an open surface S in a spatially and temporally dependent magnetic field B(r,t). Assume also that this curve moves with uniform velocity v(r). It is not required that all parts of the curve move with the same uniform velocity, implying that as it moves, the curve can get warped or undergo rotation and translation. Now defme the curve C and surface S as C 1 and Sl' respectively, at time t and as C2 and S2 at time t +~t, where ~t is an infmitesimal time interval. Defme the volume V in (2.7) as the infmitesimal volume bounded by the surfaces Sl' S2' and a cylindrical surface traced by the curve over the time ~t that can be labeled Se. As is apparent in Figure 2.1, the infmitesimal surface element dSc = (v(r)~t) X dR. The outwardly directed magnetic flux through the entire cylindrical surface Se is then f B(r,t)· dSc = f B(r,t) ·{v(r) X dR}~t = - f {vCr) xB(r,t)}· dR~t (2.8) where the last operation is a simplification from a vector formula. Now, recalling that the volume V is the volume bounded by the three surfaces in Figure 2.1, it is clear that the closed surface S* from (2.7) is actually S*=Sl+S2+Se. The total magnetic flux through this surface should equal zero, giving f B(r,t)· dS - f B(r,t)· dS - f {vCr) X B(r,t)}· dR~t = 0 S2 15 (2.9) where there is a negative sign in front of the second term because the surface vectors of Sz are antiparallel to the outgoing vector of the closed surface S*. Figure 2.1 Geometry of the infinitesimal volume enclosed by the surfaces Sj, Sz, and Sc. The definition of the rate of change of the magnetic flux through a surface is !£[I B(r,t)· dS] = lim ~[I B(r,t + l1t)· dS - IB(r,t). dS]. dt S ~HO l1t S2 S, (2.10) Expanding B(r,t + l1t) in a Taylor series gives !£[IB(r,t). dS] = lim[I~ B(r,t)· dS + ~{I B(r,t)· dS - IB(r,t). dS}]. dt S ~HO l1t S2 S2 S, at (2.11) Plugging (2.9) into (2.11) and evaluating at the limit reduces this to a- - - I -B(r,t)·dS- I [v(r)xB(r,t)]·df. -d [ IB(r,t)·dS]= &S s~ c (2.12) Applying Stokes' theorem, I- - I a- -- -d [ B(r,t)·dS]= [-B(r,t)-Vx{v(r)xB(r,t)}]·dS. dt s s at (2.13) Now recall the ideal Ohm's law (2.5) and Faraday's law, - aB at VxE=--. 16 (2.14) Combining these, we obtain the relation dB -- - - Vx{VxB}= O. (2.15) dt Thus, the integrand on the right side of (2.13) is equal to zero when ideal MHD is applied and, consequently, f- - d [ B(r,t)·dS]=O, dt s (2.16) implying that the magnetic flux through any surface flowing with a plasma which satisfies the ideal magnetohydrodynamic model is constant. This is exactly what it means for the magnetic field lines to be frozen in to the plasma. Frozen-in flux can be useful in the laboratory (though ideal MHD, and therefore this simplification, does not always apply) because it reflects how closely flow and magnetic fields are tied together. In the right conditions, well-placed magnetic fields can be used to induce desired flows in a plasma. 2.2 The Induction Equation Returning to the resistive MHD model, which postulates the resistive Ohm's law (2.4), the consequences of the extra resistivity term are not yet clear. One would assume that frozen-in flux does not apply because ideal MHD was necessary for its proof. This is indeed the case, for the following [13] will show that field lines in a resistive plasma undergo diffusion, causing changes in flux, in addition to convection. Begin by taking the curl of the resistive Ohm's law: v X (Ii + iT X B) = 1]V X J (2.17) if we assume that the resistivity remains constant throughout the plasma. Substituting relations from Ampere's law, fl) = V X 13, and Faraday's law (2.14), this becomes dB - - 1] - - - + V X (V X B) = - V X (V X B). dt flo (2.18) Invoking a vector field identity and the fact that the divergence of the magnetic field is zero, this equation reduces to - + _V2 1] -dB = V X (V- X B) B, dt flo (2.19) often referred to as the induction equation. This equation without the second term on the right is exactly the same as (2.15), which was derived using the ideal Ohm's law. Thus, the first term on the 17 right can be attributed to magnetic field convection, which is behavior of the magnetic field insofar as its flux is frozen in with the plasma flow. Along those same lines, if one were to eliminate the convection term from (2.19), he would be left with a diffusion equation djj = !Lv 2 jj dt 110 ' (2.20) implying that this term is responsible for the diffusion of magnetic fields. In addition, this relation suggests that the higher the resistivity 11, the quicker the magnetic field will diffuse, which is perfectly consistent with the idea that no diffusion occurs at 11=0. A measure of the extent to which a magnetic field in a plasma convects or diffuses would be a useful characteristic. It would essentially tell the experimenter to what extent ideal MHD holds to what extent the magnetic fields are frozen into the plasma. This measure is achieved by taking the approximate ratio of the convection term to the diffusion term from the induction equation. In order to generalize appropriately, the assumption is made that the vector operator V is approximately equal to 1/L> where L is the length scale of the plasma. The length scale is simply a measure of the approximate size of the plasma (most laboratory plasmas have length scales of 1-10 m). With the length scale approximation, the ratio is vB convection "" ~ diffusion 1']B l1oL2 (2.21) The result is a dimensionless quantity called the magnetic Reynolds number, RM : R _l1oLv M - 1'] (2.22) This relationship shows that the amount of convection of a field will go up compared to diffusion of the magnetic field in a plasma as the flow of the plasma increases, a gratifyingly intuitive result which explains why, in many plasma formations, faster plasmas lose magnetic energy more slowly and therefore are more stable to a point. Past a certain speed, however, turbulence not accounted for by resistive MHD begins to have a destabilizing effect. Also note that the Reynolds number is dependent on an approximately defmed length scale L. This length scale originated with the del operator, so when the magnetic field does not vary 18 considerably over small distances, L is larger, while if the spatial variation of the magnetic field is high, L is smaller. Therefore, the Reynolds number is relatively high in the case of most stable configurations of plasma. However, in the case of events such as magnetic reconnection, where the distance over which the magnetic field reverses direction is small, the Reynolds number will be low and the amount of magnetic field diffusion will be higher. 2.3 The Alfven Speed Before continuing to a theoretical discussion of magnetic reconnection, it is appropriate to discuss another parameter of plasmas: the Alfven speed. This is the speed that a portion of the plasma would reach if all of its magnetic energy were converted to kinetic energy. Some may refer to it as the MHD speed limit of a plasma because, in many plasmas, the only stored energy is in the form of magnetic fields. The derivation simply requires equating magnetic energy density to kinetic energy density: 1 2 1 2 - B =-pvA 2/10 (2.23) 2 where VA is the Alfven speed and B is the magnitude of the local magnetic field. The Alfven speed is (2.24) This speed goes up if the magnetic field is higher since there is more energy stored in the initial field and goes down if the density is higher because there is more mass over which to distribute the energy. The Alfven speed at SSX is approximately 100 km Is. It is worth pointing out that a common parameter used to describe plasmas is the Lundquist number, which is simply the magnetic Reynolds number with the velocity set to the Alfven velocity. Essentially, the Lundquist number reflects the highest possible Reynolds number for an MHD plasma: /1o LEfio - -LV - -S= -A TJ TJ - (2.25) P The Lundquist number at SSX is approximately one thousand, suggesting that the frozen-in flux condition is pretty well satisfied for plasmas not undergoing magnetic reconnection. 19 2.4 Magnetic Reconnection Basics During magnetic reconnection, two regions of plasma with opposing magnetic fields approach each other in a process in which their fields are annihilated, canceling each other out, and then reconnected at lower energy. Plasma is forced out of the region at high speeds corresponding to the leftover kinetic energy. Note that, for field lines to be broken, the frozen-in flux conditions specified by ideal MHD must be broken. This suggests that we cannot use ideal MHD to analyze magnetic reconnection, for if we did, we would predict that, as shown on the left in Figure 2.2, the field lines would continue moving towards each other until there were an infinite curl of B around the central reconnection region, resulting in an infinite current. Magnetic reconnection is a complicated, three-dimensional process. The theory presented shortly will be a 2-dimensional simplification that provides a rough approximation of what occurs during magnetic reconnection in a resistive MHD plasma. Most laboratories, including our own, attempt to induce magnetic reconnection in as predictable a way as possible, but ultimately, the breaking of magnetic field lines is a difficult physical process to understand. The magnetic fields undergoing reconnection are usually twisted and sheared, and the released energy actually contributes to a lot of different processes, including but not exclusive to, bulk plasma motion, electron and ion heating, and superthermal electron and ion acceleration [14]. y RESIST E MHO wi SWEET-PARKER MODEL IDEAL MHO Figure 2.2 Magnetic reconnection. On the left is a depiction of two incident opposed magnetic fields according to the ideal MHD formulation: reconnection does not occur. On the right is the schematic for the Sweet-Parker model, which makes predictions about magnetic reconnection based on resistive MHD. 20 2.5 Sweet-Parker Reconnection One of the fIrst models for magnetic reconnection, now most useful for its simplicity, was released by Sweet [1 S] and Parker [16], but here I will use a formulation by Priest [17]. It makes the simplifying assumption of a two-dimensional geometry, as illustrated in Figure 2.2. The model assumes a steady state, meaning that the inflow velocity and the outflow velocity remain constant. There will be no non-zero time derivatives. In my derivation, I will use a coordinate system in which plasma flows from positive and negative y into the reconnection region with magnetic fIelds pointing towards positive and negative x, respectively. The outflows occur towards positive and negative x with the outgoing fIelds aligned along positive and negative y, respectively. There is a further assumption that reconnection occurs in a very thin layer so that the incoming magnetic fIelds are essentially parallel to each other just outside the layer. The curvature of the fIeld lines determines the size of this reconnection region. For the Sweet-Parker model to apply, width of the layer 2L must be much greater than the thickness of the layer 2 i!. . As one might expect, our derivation begins with the resistive Ohm's law (2.4). Outside the reconnection zone, we assume that the incoming magnetic fIelds are approximately straight, implying that the curl of the magnetic fIeld in this region is zero and, according to Ampere's law, there is no current density in the region. The resistive Ohm's law then tells us that (2.26) and (2.27) at both positive and negative y outside the reconnection region. Here, v in is the inflow velocity of the plasma and Bin is the magnitude of the incoming magnetic fIeld. In the reconnection layer, we assume that the magnetic fIelds are annihilated so B=O. In this layer, Ohm's law gives us Ez = 1Jlz. (2.28) The current inside the reconnection layer is easily extracted using Ampere's law. Now imagine an Amperian loop running around the boundary of the reconnection layer. Assuming that the outgoing magnetic fIelds on the sides contribute a negligible amount to the line integral and again assuming that the magnetic fIelds are straight along the x-axis, we can set the current density within the layer equal to the line integral of the magnetic fIelds around the layer, providing the relation B;n2L + B;n2L = JIJz2L2i!. 21 (2.29) or (2.30) Recall the Sweet-Parker assumption that reconnection is a steady state, implying -dB = -v X E- = 0, according to Faraday's law. dt If there is no curl of the electric field, then the electric field outside the reconnection layer must be equal to the electric field inside the layer. Combining equations (2.27), (2.28), and (2.30) provides an expression for the incoming velocity: 17 J10 f v. =--. In (2.31) Finally, the steady-state condition requires that there be no buildup of mass in the reconnection layer. This tells us that the incoming mass flux of plasma must be equal to the outgoing mass flux of plasma, or (2.32) where V out is the outflow speed and we assume constant density. Substituting (2.32) into (2.31) yields (2.33) Now, we must invoke a new equation from MHD, the fluid equation of motion [9]: p[ d17 - at + (v . V)V] = J X B - VP . (2.34) The left side of the equation is from Newton's law: p is the mass density and the bracketed factor is the convective derivative of plasma velocity - acceleration. This convective derivative consists of two components. The first component is the acceleration of fluid within the structure independent of overall structural motion. The second component of the convective derivative comes into play when the structure itself is accelerating as a whole in addition to accelerations within the structure. This becomes particularly apparent when there is a velocity shear present. Even if a plasma is not accelerating while it moves past a point in space, the velocity measured at that point will change merely because of changes in velocity that have spatial dependence within the structure. The right side of the equation is a little more straightforward. The first component corresponds to the magnetic force on a current or, in this case, a flow of charged fluids. The [mal term predicts a force 22 based on plasma pressure. This is a common term in fluid equations, and it corresponds to the tendency for fluids to move to areas of lower pressure. In a steady state and assuming that pressure gradients are negligible, this equation becomes (2.35) Like when deriving the Reynolds number, we will approximate the gradient by 1/L. Applying (2.30), the solution to this equation in the outflow region is (2.36) Combining (2.26), (2.27), (2.32), and (2.36), the outflow speed reduces to V out 2 Bin 2 =-- (2.37) l10P which, when compared to (2.24) above, shows that the outflow velocity from the reconnection region according to the Sweet-Parker model is equal to the Alfven speed. This is not surprising, for the Sweet-Parker model does not account for the magnetic energy going into any process other than kinetic outflow. The initial assumption that i!. < <L presumes that the outflowing magnetic field would not be of significant magnitude compared to the inflowing magnetic field. With known outflow, the inflow is predictable from (2.33): V in 2 1J l10L = - - VA · (2.38) N ow we can defme a new plasma parameter called the reconnection rate M: the ratio between the inflow velocity during reconnection and the Alfven velocity. The Sweet-Parker reconnection rate is (2.39) The term in the radical is just the inverse of the Lundquist number derived earlier (2.25), so this reconnection rate reduces to I Msp = JS. 23 (2.40) 2.6 The Spheromak A spheromak is a stable configuration of plasma, which is notable for being, among other things, force-free. That is to say that one need not run a large sustaining current like in the case of a tokamak or centrifugal plasma in order to form a spheromak. It is essentially a donut of plasma containing both toroidal and poloidal fields. Plasma confined in magnetic field Polodial Magnetic Field Torodial Magnetic Field Figure 2.3 A rough image of a spheromak. The plasma is contained poloidal and toroidal magnetic fields. As a stable plasma, insofar as MHD is accurate, the spheromak should be in an equilibrium state. This suggests that the spheromak as a whole has no bulk motion. Seeing that we are interested in flows, this sounds odd, but that is only because we are referring to bulk flows rather than internal flows. In an equilibrium state, we merely require that the magnetic fields and velocities are constant in time. We do not require that the internal velocities of the plasma go to zero. A more extensive discussion about why we expect some flow can be found in Appendix A. Another consequence 0 f the spheromak being force- free is that it has low pressure [19]. Thus, returning to MHD equation (2.34) and making these simplifications, the equilibrium equation becomes JxB=O (2.41) (VxB)xB=O. (2.42) or, from Ampere's law, 24 Given no reason to generalize that either the magnetic field or the current density go to zero, it must follow from this equation that the curl of the magnetic field is perpendicular to the magnetic field itself. This implies that at equilibrium, the curl of the magnetic field is a multiple of the magnetic field: (2.43) - - - 'I ... ." I! :<f ."/ ~ ~ -- ~ - f " . ..: :, . f h , -h - ....r ,, ~ ;:::. ... T .;. I - ~ ~ D ')l .US· 62 Figure 2.4 Magnetic characteristics of a stable single spheromak inside the SSX chamber. The views from the side of the chamber exhibit poloidal field characteristics while the cross-sections below show the toroidal field. Figure from [18]. Let us introduce one final plasma parameter, the magnetic helicity K. This is a measure of how twisted the magnetic flux tubes in a given plasma are. It is defined as follows: K= fA.BdV, (2.44) v where A is the vector potential. It turns out that K is conserved in the approximation of constant magnetic energy. In other words, K is relatively constant in plasmas like spheromaks, where the magnetic energy is mosdy convective. It has been shown that solutions that minimize the energy for a given K also predict a constant value for A in (2.43), making this equilibrium equation solvable. The equilibrium solution for the magnetic field of a spheromak in a cylindrical container of height h and radius a is given by [20] as 25 = -B)d1{lr)cos(kz), Brp = B o IlJ1 {lr)sin(kz), B z = B)daC1r)sin(kz) Br where kh=1t, la=3.83, and /1 = (2.45) e: J J. 3 + (: This suggests that the magnetic field gnes like a half-period over the axis of the machine and like fIrst -order solutions of Bessel functions in the radial direction. The validity of this model has been verifIed in SSX spheromaks [21], [22]. 26 Chapter 3 Mach Probe Theory When making local flow measurements in a plasma, there are several difficulties to be considered. Primarily, it is necessary to differentiate between flow velocity and thermal velocity. Thermal velocity is a Gaussian distribution of velocities manifested by an ensemble of particles. If we are to determine the flow velocity by observing the velocity of individual microscopic particles, then we must account for this thermal velocity and recognize its occurrence independently of a macroscopic flow velocity in the plasma. Secondly, as is the case for all local measurements, caution should be taken to perturb the plasma as little as possible and to account for any perturbations that are made by the diagnostic in the plasma. Local flow measurements require that a diagnostic making the measurement be immersed in the plasma and, the diagnostic having finite dimensions, perturb the motion of the plasma that it is simultaneously trying to measure. This is particularly noticeable during magnetic reconnection, when the diagnostic itself is occupying the reconnection region, effecting the phenomenon in often unpredictable ways. A third difficulty follows from the charged nature of the particles in a plasma. Because of the magnetic fields present throughout the plasma, particles very rarely follow straight trajectories. Rather, they move throughout the plasma on orbits, the size of which are determined by the magnitude of the electric and magnetic fields and the velocity of the particle. This implies that, if a particle's velocity is measured at one point in time, it could actually be the reverse of the overall drift velocity of that particle. The particle could be moving backwards temporarily due to the orbital nature of its path. Any local flow measurements will have to take into account that charged particles in a magnetic field, rather than free particles moving in straight lines, are being observed. 27 The diagnostic used to make local flow measurements at SSX is commonly referred to as a Mach probe. The Mach probe is a variation on the Langmuir probe, a simple metal rod held at a bias voltage and immersed in the plasma. The ions collected by the Langmuir probe can be used to predict plasma density and temperature - this diagnostic has been used to measure these parameters in plasma chambers for decades, though we will not be using it for that purpose here. Now imagine that one side of the Langmuir probe is blocked and the probe is negatively charged, such that it collects ions from only one direction. Right next to this "directed" Langmuir probe, on the other side of the obstruction, is placed another Langmuir probe that collects ions from the other direction. The resulting instrument is the simplest orientation of a Mach probe, as shown in Figure 3.1. Its functionality is actually rather simple in principle. Let us imagine that we turn the Mach probe such that one of the ion collectors (as I will refer to the biased Langmuir probes from here on) is facing toward the direction of plasma flow, and the other ion collector is facing away from the plasma flow. As ions move through the plasma, they will be caught by the ion collectors upon which they are incident. The collector facing "upstream" (towards the flow) will collect more ions than the collector facing "downstream." It is by taking the ratio of the current collected by these two probes that we gain insight into the ion drift velocity in a plasma. barrier upstrea co llector FLOW :> Figure 3.1. The simplest possible Mach probe consists of two ion collectors divided by a barrier. The upstream collector collects more ions than the downstream collector, giving us the means to approximate the flow. 28 Clearly, if there is no net drift velocity, the ratio of the two currents will be one: each collector is gathering the same amount of current. However, when the plasma has a nonzero drift velocity, how does this ratio deviate from one? This analysis proves to be rather difficult, and, without making several assumptions, can become almost completely intractable. The theory behind Mach probe analysis can be separated into two distinct regimes. The "magnetized" regime corresponds to a larger probe, specifically one with a radius that is greater than the ion gyroradius. The "unmagnetized" regime is exactly the opposite - the radius of a probe in the unmagnetized regime is smaller than the ion gyroradius of the plasma. The results presented here include probes of both sizes and, therefore, invoke the analysis for both regimes. The theory for each of these models is relatively similar, though unequally well developed. For example, a rigorous treatment of the magnetized regime has been presented with success [23], [24], but the unmagnetized regime is still a controversial field, and analysis done thus far usually only extends to unique cases. It is also worth adding that many of these analyses have been done using powerful computer simulations and not with rigorous mathematical treatments. The electromagnetic forces determining the motion of a very large ensemble of plasma particles are often too difficult to resolve without simplifications that cripple the power of one's analysis. Before we take a closer look at some of the models used to develop Mach probe theory, I would like to clarify some crucial concepts. Each theory invokes the use of a Mach number rather than directly referring to a drift velocity. The Mach number M is actually the ratio of the drift velocity v to the sound speed in the plasma Cs : v M=.- (3.1) Cs or, put another way, the Mach number is the drift velocity normalized to the sound speed in the plasma. Sound speed in a plasma is a function of temperature that determines roughly how quickly particle waves move through the plasma. It is provided by the equation (3.2) where Te and Ti are the electron and ion temperatures respectively, y is the ratio of specific heats at constant pressure and volume, respectively, familiar from thermodynamics, and tIl; is the ion mass. I have absorbed Boltzmann's constant k into the statements of temperature here and will continue to 29 do so, as is the common practice in plasma physics, such that the "temperatures" listed are actually in units of electron volts. 3.1 The Hudis-Lidsky Model Until relatively recently, the Hudis-Lidsky model [25] had gone unchallenged as the method for Mach probes in the unmagnetized regime. It consists of a one-dimensional analysis of the differences in current between two oppositely directed probes. This model is only appropriate after adopting four major assumptions. The fIrst one is that the orientation of the probe with reference to the magnetic fIeld does not effect the results. This is basically the same as saying that we are working within the unmagnetized regime. The orbits of the particles are not a factor in this theory. The second assumption is a result of the fact that the Hudis-Lidsky model consists of a onedimensional analysis. That is, the geometry of the probe itself does not effect ion collection. There is no turbulence unique to a certain probe shape as the plasma goes by. Another assumption states that the plasma should be collisionless and high density. This assumption assures that we need not consider collisions with other particles when determining the energy and motion of particles near the probe. The [mal condition is similar to the second: there is no differentiation between the plasma flow into and away from the probe. The probe is not blocking the motion of the plasma, and therefore we can treat the velocity of the plasma as it approaches the upstream collector as if it were the negative of the velocity of the plasma as it moves away from the downstream collector. In more exact terms, we can sum up these assumptions in the form of relations between plasma and probe parameters: (3.3) or the Debye length is much less than the radius of the probe. The Debye length is an important parameter in plasmas. It determines the distance over which a potential immersed in the plasma will act. Because a plasma consists of flowing particles, when a charge is immersed in the plasma, its potential will be shielded since particles with opposite charges will be attracted to it while others are repelled. It is possible to determine from Poisson's equation V 2 l/> = - j( the rate at which the potential drops off with distance from a charged body in the plasma [11]. The electric potential decreases with distance x according to 30 Ixl/ l/J = l/Jo e - /AD (3.4) with the parameter An defmed as the Debye length: (3.5) where, again, Boltzmann's constant has been absorbed into the temperature Te. SSX has calculated a Debye length of 3 J.lm, given an electron temperature of 20 eV and a density of 1014 cm 3 . The assumption made in (3.3) is usually a safe one, considering that the Debye length in most laboratory plasmas is on the scale of microns, while the radius of most Mach probes is on the scale of centimeters. Another assumption in terms of plasma parameters is (3.6) or the ion gyroradius is less than the radius of the probe. The ion gyrodradius is the radius traversed by an ion in a magnetic field of magnitude B due to the Lorentz force. To approximate the characteristic ion gyroradius for a plasma, we assume that the ions are moving at the ion thermal velocity, giving an ion gyroradius of miv-L -J~mi a - - - -'--'--i IqlB - eB . (3.7) When the ion gyroradius is more than the radius of the probe, then we must take into account those particles which hit the downstream side because they have curled around the probe due to magnetic forces. Conversely, when the ion gyroradius is less than the probe, we do not have to account for the effect of the magnetic field on individual particles. This is why the relation in (3.6) characterizes the unmagnetized regime. The ion gyroradius at SSX is on the order of a centimeter. It is for this reason that we chose to analyze the flow with two separate Mach probes, one with a radius smaller than the ion gyroradius and the other with a radius larger than the ion gyroradius. A third condition is that (3.8) or the ion temperature is less than the electron temperature. This condition is arguably one of the most restrictive conditions for this model, considering that it is often not the case in laboratory 31 plasmas that ion temperatures are considerably less than electron temperatures. In fact, at SSX, the ion temperature is approximately 60 eV while the electron temperature is approximately 20 eV: this relationship does not hold. Finally, Vi > C s (at the sheath edge), (3.9) also known as the Bohm sheath criterion. The sheath is the region within one Debye length of a potential in the plasma. In this case, it is a small region adjacent to each ion collector in which there is a significant potential drop. The Bohm sheath criterion requires that the ions be moving at speeds higher than the sound speed before they enter this sheath. This implies, according to the HudisLidsky model, that there must exist some presheath in which particles are accelerated to this required speed before they enter the sheath to be collected by the ion collectors. The Hudis-Lidsky model assumes that, an infInite distance from the probe, ions start at the thermal velocity, (3.10) which is the velocity that an ion would travel if the average thermal energy (TJ were converted to kinetic energy. Accounting for the one-dimensional motion of particles moving through the presheath and sheath on either side of the probe, the model postulates the following formula for received current: (3.11) A is the area of the exposed ion collector and Vd is the drift velocity of the plasma. Hudis and Lidsky have determined a more specifIc value for C [25], but it is not necessary for the analysis performed here. The plus and minus signs correspond to the upstream and downstream currents, respectively. When we divide the upstream current by the downstream current, we fmd an exponential relationship between the current density ratio and the Mach number. (3.12) Note that I have divided through by the area of the ion collector so that we are actually considering the ratio of current densities rather than current. We would not want the size of our collectors to effect the determined Mach drift. Simplifying the equation above, we have 32 · 4vt *Vd ~=eSS (3.13) jdown where I have separated out the ratio of drift velocity to sound speed so that we can put the exponential in the form jup --=e kM (3.14) jdown where M is the Mach number and k is simply (3.15) 3.2 Problems with the Hudis-Lidsky Model The Hudis-Lidsky model was eventually satisfactorily debunked by Hutchinson [26] for several reasons. First, the one-dimensional approach does not provide a satisfactory description of particle behavior around a Mach probe. In the real conditions for a Mach probe, even at small ion temperatures, ions will be drawn to the downstream side of the probe by being accelerated transversely with the presheath potential, as shown in Figure 3.1. This common behavior could not be predicted with a one-dimensional model that assumes all of the particle's energy is concentrated into one-dimensional acceleration. The next problem lies in the assumption that the Bohm sheath edge criterion (3.9) applies. Based on the condition that ion density and electron density be equal at the edge of the sheath and (3.11), Hudis and Lidsky project an ion density n i at the sheath radius rs of _ (v, ±Vd )2 C, 2 (3.16) where, as before, the plus and minus signs refer to upstream and downstream densities. Hutchinson notes something odd about the implications of this relation. If the drift velocity is greater than the thermal velocity of the plasma, the downstream current will actually increase as the drift velocity increases. Not only is this not intuitive - it does not happen in the laboratory, implying a further restriction on this model Vd < Vt" At SSX, the drift velocity of ions is often greater than their thermal velocity. 33 \, robe", , • \ \ \ - I I ./ Do v stream ,.---- - ~-- -/ Velocity / \I / . I (' Figure 3.2 A two-dimensional depiction of particle behavior around a Mach probe in the unmagnetized regime. Image from [5]. Hutchinson used a particle-in-cell (PIC) computational method to analytically derive the ion distribution on a sphere moving through a plasma (relatively equivalent to a stationary sphere in a moving plasma) [27] with the intention of comparing the situation to a Mach probe in a flowing plasma. His findings agreed with the logarithmic relationship between the Mach number M and the ratio of upstream and downstream currents, shown in (3.14), but he did not agree with the proposed proportionality constant k shown in (3.15). His findings propose results for k corresponding to two different temperature extremes. For T j < 3Te , he finds that k is constant: kum T<3T , Ie = 1.34. (3.1 7) At the other extreme, for T j > 10Te , he finds a temperature-dependent k: (3.1 8) The ratio of ion and electron temperatures in SSX plasma lies close to the lower ion temperature extreme, given that our ion temperatures vary from 60 to 80 eV and our electron temperatures vary from 20 to 25 eV. 34 3.3 Magnetized Mach Probe Theory In magnetized Mach probes, the ion gyroradius is smaller than the probe radius, while in unmagnetized Mach probes, the opposite is true. The kinetic theory derived here can be attributed to Chung and Hutchinson [24]. Let us begin by considering the sheath and presheath of a Mach probe, as described earlier in reference to the Hudis-Lidsky theory. In the case of a magnetized probe, we will assume that the magnetic field is aligned with the direction of flow of the plasma (consistent with frozen-in magnetic flux), but allow for diffusion into and out of the pathway of flow towards and away from the probe, as shown in Figure 3.3. rro b·e IE) . Oi ffus ive Ion SOiJ r'ce ,-. .....,-tl- -U .....t(v.z)~ Wf... Par tild ,e Ex.cha.nge Figure 3.3 Schematic of the ion collection process assuming a one-dimensional analysis with magnetic field in the same direction as flow. Ions are accelerated toward the probe in the presheath region, and are given the freedom to diffuse in and out of this region. Figure from [24]. Calling f(z,vz>t) the one-dimensional ion distribution function, the Boltzmann equation for a fluid provides (3.19) where C f is the collision operator and Sf is a volume source of ions, and V z and az are the respective velocities and accelerations of the distributed particles, dependent on the Lorentz force. Assuming a 35 steady-state ion distribution (eliminating the time derivative) and a collisionless plasma, the Boltzmann equation becomes (3.20) where m, q, and <!> correspond to ion mass, ion charge, and electric potential. Assuming a constant total energy for a given ion E, consider the familiar energy relation I 2 "2mvz + qljJ(z) = E. (3.21) It is clear from this equation alone that the relation between potential and velocity for each particle will be a parabolic one given a position z. The rub, however, is to determine the spatial variation of potential that satisfies these equations. Assuming isothermal electrons, a Boltzmann relation implies that there is a spatially dependent background density of electrons against which the ions move: (3.22) where n= refers to the density of the plasma outside of the presheath and T e corresponds to the electron temperature. The electron and ion densities, insofar as they are responsible for the potential, must follow Poisson's equation: (3.23) Here, I have assumed that the ions are simply protons, or that the plasma originated from hydrogen gas, which is the case for SSx. Finally, we must account for the cross-field transport originating from diffusion in and out of the presheath region. Consider a frequency W(z,v) which dictates the rate at which particles move between the outer plasma and the presheath. The volume source of ions Sf specified in equation (3.20) must depend on this frequency: Sf = W(z, v)[f=(v)- fez, v)]. (3.24) The first term on the right-hand side corresponds to the ion distribution function of the exterior plasma as a function of velocity. When multiplied by the rate at which particles move between the outer plasma and presheath, it determines the relative amount of ions entering the flux tube from the outside. The second term on the right corresponds to those leaving the tube. We are making the assumption that a particle at position z with velocity v is just as likely to leave the tube as another 36 particle with the same position and velocity is to enter it; in other words, we are assuming that the rate is the same for both the process of leaving the tube and entering the tube. Once these equations have been established, they can be analyzed using numerical means. For more details on how Chung and Hutchinson went about this, see [24]. The only further steps taken before computational analysis were an assumption that the exchange rate between the exterior plasma and the presheath is independent of particle velocity (W(z,v)=W(z)) and a nondimensionalization of each of the plasma parameters in the above equations. In the end, it is affttmed that there is a logarithmic relationship between the upstream and downstream currents in a Mach probe and the flow of the exterior plasma. Assuming Ti = T e , the calibration constant k for a Mach probe immersed in a plasma that satisfies the magnetized condition is k m "T=Te = 1.7. 37 (3.25) 38 Chapter 4 The Experiment The Swarthmore Spheromak Experiment (SSX) lab is equipped with a chamber capable of producing single stable spheromak formations or merging two axially-aligned spheromaks. All measurements upon the plasma are made within a vacuum chamber, shown in Figure 4.1, consisting of a stainless steel cylinder 0.60 m in diameter and approximately 1 m in length. The chamber is kept at a vacuum of 2 x 10 7 torr by an oil-free cryogenic pump. Lining the inside of the chamber are two cylindrical copper containers called flux conservers with inner radii of 0.20 m. These flux conservers are placed side by side and separated by 2 cm at the midplane so that diagnostics may be inserted into the midplane region. The plasma is contained inside these flux conservers over the course of the shot, as they provide the high electrical conductivity necessary at the edge to allow for stable boundary conditions. Additionally, having perfectly conducting boundary conditions in the laboratory makes for a much simpler comparison to computer simulations. The vacuum chamber is flanked by two coaxial plasma guns. The energy provided to the guns during plasma formation comes from 2 large capacitor banks (the lab has a total of four). Each of these banks have a capacitance of 0.5 mF and are capable of being charged to a voltage of 10 kV, providing a maximum energy output of 25 kJ each. In practice, however, the capacitor banks are only charged to about 5 kV in the process of creating a spheromak. There is a second group of capacitors that drive current through coils surrounding each gun. These currents establish the magnetic field necessary for a "stuffing" flux in each gun. This effect will be described in more detail shortly. 39 Figure 4.1 The SSX plasma chamber. Several diagnostics are visible in this picture. The two small cylindrical steel casings protruding toward us from near the top of the chamber are two of the probes in the magnetic edge array. The aluminum bar on the far right of the picture holds the HeNe interferometer setup, which sends a laser through a window on the side of the chamber. Also, on the largest visible flange, both the soft x-ray and the IDS diagnostics are used to observe the plasma from the edge of the midplane. The large green structure to the left in the back is one of the main capacitor banks, and the steel cylinders protruding outward from either side of the chamber are the coaxial guns. Surrounding the midplane on the outside of the chamber, are two reconnection control coils (RCC's). These toroidal coils contain a current that drives a magnetic field in a poloidal direction through the center of the chamber. Originally installed to stabilize and hold apart spheromaks during magnetic reconnection, the currents in these coils are now often used in alignment with the current of a single spheromak in the chamber in order to stabilize it. The RCC's are powered by, of course, another set of capacitors. Over the course of a shot (approximately 100 /ls), they produce a steady magnetic field of 700 G at the central axis, in alignment with the poloidal fields of a spheromak in the chamber. In addition to the Mach probe, whose results are contained in this thesis, the SSX lab is currendy working with several diagnostics. Other diagnostics that protrude into the chamber are four magnetic probes, making up an array used to approximate the orientation of macroscopic fields 40 with high temporal resolution (10 MHz) at the edge of the chamber for minimal perturbation. Optical diagnostics used recently include a soft x-ray (SXR) detector and a vacuum ultraviolet (VUV) monochrometer. Analysis from these has been oriented towards determining timedependent electron temperature and model impurity levels within the plasma [28]. Results from a HeN e-quadrature interferometer used to measure plasma density along a chord directly in front of a plasma gun were presented by Gray [29]. Also, some interesting ion Doppler spectroscopy (IDS) results concerning ion temperature and flow during reconnection were presented last year by my colleague, Jerome Fung [13]. These were attained by an ion Doppler spectrometer with a CzernyTurner configuration focused along an adjustable chord through the midplane of the chamber. In the summer of 2005, with the help of Michael Schaffer of General Atomics, SSX set up a glow discharge cleaning (GDC) system to remove impurities from inside the chamber prior to taking a new set of data. The motivation for such a system stems from the hypothesis that colder impurities get stuck to the walls of the chamber over time and, when introduced into plasmas, have an adverse effect on their stability. This new cleaning system has led to lower densities and higher ion temperatures in our plasmas. The glow discharge is a plasma itself, consisting of partially ionized helium. During the course of a glow, the pressure of helium in the chamber is maintained at about 100 millitorr while a constant DC voltage of +300 V is applied to the inner electrodes of each of the plasma guns. At this voltage, helium plasma is forced out of the guns to scour any impurity buildup on the inside of the chamber. Also, if any diagnostics protruding into the plasma are set at a bias voltage (such as our Mach probe), they are cleaned as well, making GDC particularly important to use directly after a new diagnostic has been introduced into the system. When the voltage on the electrodes is reversed, such that the inner electrode is held at -800 V, the glow occurs inside the guns (plasma is not expelled into the chamber). Each of these settings runs for at least an hour for a thorough cleaning. Preliminary results suggest a marked increase in ion temperature to up to 100 eV and a lowering of the plasma density by an order of magnitude to 1014 cm 3 as a consequence of the new GDC system. 4.1 Making a Spheromak The coaxial guns consist of a hollow cylindrical outer electrode, which is grounded, and an inner, charged cylindrical electrode. Recall also that there is a coil surrounding the gun to induce a stuffing 41 flux as well as RCC coils surrounding the midplane of the chamber itself. There are plenty of electronics running even before the plasma is created. Throughout this paper, I refer to the time at which the main capacitor banks are discharged, or the time at which the voltage on the inner electrode quickly rises, as t=O. Because they take the longest to reach the necessary current, the RCC coils are always fired first, at approximately t=-50 ms, though this time is adjustable. Then, at t=-25 ms, the capacitor banks controlling the coil surrounding the gun fire, setting off the stuffIng flux. This stuffIng flux results in a magnetic field with a flux of around 1 mWb emanating out from the inner electrode and around to the outer electrode in the radial direction. These are both fired early because the amount of time it takes for their exterior magnetic fields to penetrate the metal chamber is on the order of 10 ms. (r d) Figure 4.2 The formation of a spheromak in a coaxial gun. First (a), hydrogen gas is puffed into the end of the gun while a stuffIng field is present at the opposite end. Then, when the voltage is applied to the inner electrode (b), current runs through the plasma to induce a toroidal field that both contributes to the toroidal field in the eventual spheromak and forces the spheromak out of the gun. (c) As the plasma passes by the stuffIng field, it stretches out the field lines and gains a poloidal magnetic field. Finally, after about 20 j.ls, the spheromak is emitted from the gun (d). Figure from [30]. Then, between 400 Jls and 700 Jls prior to the firing of the large capacitor banks, hydrogen gas is puffed into the gun through high-speed valves that surround the part of the gun furthest from the chamber. It has been projected that the amount of time delay between this puff of gas and the firing of the banks has a profound effect on the density of the plasma because it determines the overall amount of gas that enters the guns prior to ionization. Finally, the large capacitors are discharged, resulting in a voltage on the inner electrode of approximately -5 kV. This voltage is high enough to ionize the gas (it literally rips electrons off of protons) such that a plasma is formed. Because an ionized plasma contains mobile charge, a current immediately flows between the grounded outer electrode and the negatively charged inner electrode, illustrated in Figure 4.2b. As this current flows down the inner electrode, a corresponding azimuthal magnetic field develops, as 42 according to Ampere's law, with two essential consequences. Not only does this field become the toroidal field of the eventual spheromak, but it induces a force directly out of the gun, in the J x B direction (where J is the radial current density in the plasma and B is the toroidal magnetic field). Finally, as the spheromak is forced out of the gun, it encounters the stuffing flux, which is weak enough to allow the plasma to eventually break through into the chamber but also strong enough to induce a necessary component for the spheromak: a poloidal magnetic field. This formation process is complete and the spheromak enters the chamber by approximately t=20 ).ls. The magnetic fields in the plasma usually decay after another 80 J.ls, or when t = 100 J.ls. Two of the main modes of our experiment, co-helicity and counter-helicity merging, require that we have the ability to adjust the direction of fields in a spheromak. This is not easily accomplished with the toroidal fields because of the electrical setup of the coaxial gun, but the current in the stuffing coil, and therefore the stuffing flux, is easily reversed. This results in a change in "handedness" of the magnetic fields within the spheromak. One refers to a spheromak as righthanded if the poloidal fields curl around the toroidal fields in the same way that a magnetic field curls around current according to the right-hand rule. The spheromak is left-handed if the poloidal field is in the reverse orientation with respect to the toroidal field. Thus, it is clear that we can create either orientation of magnetic field in a spheromak simply by reversing the stuffing flux and, therefore, the poloidal component of the magnetic field. 4.2 Mach Probe Specifics There were two different Mach probes used to obtain the results presented here. They both consisted of six ion collectors in a cylindrical boron nitride turret, referred to as a "gundestrup" design [31]. The general schematic for the smaller Mach probe is presented in Figure 4.3. The smaller Mach probe was built by Falk [8]. Since the probe was first built, the slits have been elongated to increase the amount of received current. The entire probe, from its boron nitride tip to the box containing the wire leads, is 75.9 cm. The cylindrical boron nitride sheath has a diameter of 0.65 cm with slits 0.05 cm wide. The tungsten rods, or ion collectors, are protruding from a stainless steel rod of the same diameter as the sheath and are threaded into the sheath such that they are aligned with the slits. The boron nitride sheath is sealed to the end of the stainless steel shaft using TorrSeal. The central pin acts as a reference from which to bias the other ion collectors. It remains 43 close to the local voltage of the surrounding plasma. The tip of this reference pin extends from the end of the Mach probe by about 1 mm. Each pin is welded to their respective leads using a gold pin because of the difficulties involved in welding tungsten direcdy to copper. Note, in Figure 4.3, that the vacuum seal does not occur at the tip of the Mach probe, but at a point exterior to the machine, where the electrical leads are fed thru a vacuum weld. This means that, when the Mach probe is in the chamber, the interior of the entire stainless steel rod is under vacuum. referenoe pin - - - - - ------r4= -- - feed-thru oonflat fla ng e , bo ro n nitride sheath tungste n rods electrical feedthru (vacuum we ld) Figure 4.3 A basic schematic of the Mach probe. Tungsten rods are inserted into a boron nitride sheath at the end of the probe and the leads from these rods run down a stainless steel tube. The feed-thru con flat flange seals to the outside of the chamber to maintain a vacuum. The mach probe can move in and out of the chamber as well as rotate in this flange. The vacuum seal is made at a plug at the furthest end of the stainless steel rod where the leads are passed through. The large Mach probe is somewhat different than the smaller one in schematic, though it has the same boron nitride sheath design and contains the same materials. It was originally engineered so that it could both fit on our chamber and on the chamber at the Madison Symmetric Torus (MST) lab in Wisconsin, but this collaboration has yet to be carried out. The probe stalk is contained in a larger cylinder mounted on a track, on which the probe glides in and out, and is allowed to rotate. It is this larger cylinder containing the stalk that is under vacuum when the probe is attached to the machine. The boron nitride of this probe has a diameter of 1.6 cm and slits 0.05 cm wide. The length of the entire probe is 105 cm. 44 Figure 4.4 The tip of the larger Mach probe at two stages of construction. On the left, you can see the empty boron nitride sheath and empty wire casing. On the right is the boron nitride sheath with all of the tungsten rods inserted and their wire leads protruding. Figure 4.5 The two fully assembled Mach probes. The probe on the left is visibly smaller than the one on the right. On the right, the large Mach probe is retracted almost completely into the larger cylinder containing its stalk. The Mach probes are used separately and inserted into the same portal on the chamber. Each is oriented along the radial axis of the chamber at the midplane. Before data is taken, six capacitor banks are charged to -40 V, corresponding to the ion saturation current. The ion saturation current occurs at diagnostics biased to a potential so negative that all negatively charged particles are effectively repelled from the diagnostic and only a positive current is collected. Once the capacitor banks are charged, they are isolated from anything other than the Mach probe. These capacitor banks serve to provide the bias between the ion collectors and the central pin. Finally, during the course of a shot, currents in each ion collector are measured by matched Pearson model 45 411 current transformers with an output of 0.1 V / A ± 1%. It is important that these transformers be matched so that we can properly compare the currents between separate ion collectors. 46 Chapter 5 Results 5.1 Calibrating the Mach Probe As the theory developed in the Mach probe section demonstrates, analysis of Mach probe data is reduced to flnding a calibration constant k that determines the logarithmic relationship between the ratio of opposing ion currents and the drift velocity of the plasma in which the probe is immersed. This constant is closely tied to the geometry of the probe and dimensions of the probe relative to the magnetic speciflcs of the surrounding plasma. It would be advantageous if, rather than attempting to derive this calibration constant using still controversial analytical means and merely approximate plasma parameters, we could calibrate the probe by comparing its output to a known plasma drift velocity. Mach probe West mag probe East mag probe PLASMA FLOW West gun <'lres) Figure 5.1 On the left is a diagram of the chamber when prepared for calibration. The Mach probe is centered between two magnetic probes. The picture on the right shows the setup in the laboratory. The large metal track in the middle supports the Mach probe while the two whitecapped cylindrical housings on either side contain wires for the magnetic probes. 47 This known plasma drift velocity can be determined by tracking the magnetic energy in the chamber after a single spheromak is launched into the chamber from one side. As shown in the figures on the previous page, we calibrate the Mach probe by taking data with the Mach probe at the center of the chamber while simultaneously taking data with magnetic probes placed on either side of the chamber. Recall that we are dealing with a plasma that has a Lundquist number of approximately 1000, implying that the plasma will "stick" to the magnetic field lines reliably. We can therefore assume that wherever an appreciable amount of magnetic energy is present within the chamber, there is also plasma present. The magnetic probes used in our labs have a high time resolution (less than 1 j.ls), allowing us to view the magnetic field with some precision. In this case, I will use the total amount of magnetic field squared at each of the three-dimensional probes (B,2 + B/ + B z) to obtain a value proportional to the magnetic energy present at the probe. The result for a single spheromak shot is shown below. Figure 5.2 The total magnetic "energy" at each of the magnetic probes for one single spheromak shot. The solid line corresponds to the West magnetic probe while the dashed line corresponds to the East magnetic probe. The quick rise in energy makes it relatively easy to determine the time of flight for the plasma between the two probes. 48 This figure makes it apparent that, when the plasma is initially present at either of the probes, the magnetic "energy" at the probe is on the order of 106 G 2 • Noise on the order of 105 G 2 prior to the sudden rise in the East probe suggests that the time of flight would be the difference between the time at which the East probe reads a magnetic "energy" of 5xl05 G 2 and when the West probe reads the same "energy." An average taken over twenty-five shots gives us a value for the initial drift velocity of a spheromak out of the gun for our usual setup of 5 kV charge and 1 mWb of stuffing flux: (5.1) Assuming an electron temperature of 25 eV [28] and an isothermal plasma, the sound speed in the plasma is C, = t~~' = 49. 0kmls· (5.2) Therefore, the initial Mach drift of the plasma at the center of the chamber should be Mi =~=1.74+0.15· (5.3) Cs For our flow experiments, we used two different Mach probes, one larger than the other. Since k is dependent on the relative proportions of the Mach probe and the magnetic characteristics of the plasma, these two probes should also have two different k values even though they exhibit the same general geometry. We took calibration data with two different orientations for each of the probes. In one orientation, one ion collector is facing directly upstream and another is facing directly downstream, such that they are lined up with the cylindrical axis of the chamber. In the other orientation, the Mach probe is rotated 180°, and we expect to see an inversion of the flows measured using the original orientation for all 3 pairs of probes. For calibration purposes, however, we are only concerned with the ratios of the upstream and downstream probes, shown in Figure 5.3 below. 49 Figure 5.3 The above two graphs show the logs of the ratio between the upstream and downstream current densities averaged over 15 shots in the two different calibration orientations. The error bars represent the standard deviation of the mean. The inversion reflects the fact that the probe predicts a negative velocity when rotated 1800 from its original orientation. 50 Using our data from time of flight, it is clear that the initial flow velocity should occur at the Mach probe around t = 30 J..ls. The log of the ratio for the large Mach probe at this time is 4.0 ± 0.5, while the log of the ratio for the small Mach probe at this time is 3.3 ± 0.7. Recalling the initial Mach drift of the plasma out of the guns, we now have enough information to determine the calibration constants. k1g = 2.30 ± 0.35 (5.4) = 1.90 ± 0.43 (5.5) k sm The value for k1g does not agree with Hutchinson's value for k for a magnetized probe (3.25) of 1.7. However, this value was found by setting ion temperature equal to electron temperature, which is not valid at SSx. It is harder to evaluate the value determined for k.m because it is in the unmagnetized realm. However, it is between the two values predicted for the two different temperature extremes by (3.17) and (3.18). Considering that the temperature of our plasma also lies between these extremes, unless we had some other way to approximate the value of the calibration constant, this level of agreement is the best that we could hope for. Note that, also, the condition of inversion has been satisfied within error in these two graphs. This was the case for each of the two other pairs of probes during calibration, as well. 5.2 Visualizing Mach Probe Data N ow that we know the calibration constants for each Mach probe, we can take data to determine the flows in a plasma. Assuming cylindrical symmetry for the probe, we can use the same calibration constant for each separate opposing pair of ion collectors. A look at the raw data in Figure 5.4 provides insight to the difficulties involved in analyzing the Mach probe. Notice how, at some points in the data, the current density in a given ion collector will suddenly drop to zero or rise from zero. Considering that the ion collectors are each facing nearly equal densities of plasma and that the sudden drops to zero would correspond to unrealistically high plasma flows (because the flows are determined by a ratio of current densities), it can be assumed that these portions of the data do not actually correspond to measurements of flow of the plasma. Rather, there must be some electronic source of these irregularities in the data acquisition process or in the interaction of the probe with the plasma. We can only speculate as to the exact source, but it is clear that these portions of the 51 data should be ignored when we analyze the data for flows. Luckily, the irregularities occur infrequently enough that results can be presented with confidence. Figure 5.4 Six separate waveforms corresponding to the current densities in kA / m A2 for each of the six ion collectors over the course of a single spheromak shot. Note how there are sudden drops and rises that we speculate correspond to arcing, which render some sections of the data intractable. Recall that there are six ion collectors, consisting of three pairs to determine the flow along three different axes as shown in Figure 5.5. Because these three axes are in the 8-z plane, we cannot make any measurements of radial flow, but the flows in the azimuthal and axial directions are actually overdetermined. Using a simple coordinate transformation, we can resolve all three flows into their axial and azimuthal directions. Also, the overdetermination in each direction provides a measure of "error" that reflects how closely the individual measurements of flow agree with one another. 52 z Figure 5.5 Each black dot in the figure corresponds to an ion collector in the Mach probe. Each opposing pair of ion collectors provides a measure of flow along an axis a, b, or c. The probe is oriented in the machine to be aligned in the 8-z plane as shown. Once the transformation is made, there is a clear depiction of the rotation of the plasma (azimuthal component) as well as the axial flow through the center of the chamber. An example of a typical result from the Mach probe over the course of a single spheromak shot is shown in Figure 5.6 below, along with a two-dimensional snapshot of flow right after the plasma has passed by the probe in Figure 5.7, shown with relative current densities. Figure 5.6 This graph is an example of the flow at the center of the chamber over the course of a single spheromak shot. As we would expect, the axial flows are much more dynamic than the azimuthal flows. 53 Figure 5.7 A snapshot of the flow 29 J..ls after plasma enters the chamber. Each of the six points surrounding the origin are set at a radius relative to their respective current densities. The arrow corresponds to the net flow. The error bars at the end of the arrow are a measure of error in each direction showing how closely the distinct measurements of flow from each pair of ion collectors agree. This graphing method was inspired by [3]. 5.3 Single Spheromak Results Both Mach probes were used to provide a radial profile of the flow at the midplane of the chamber with interesting results. Measurements were made every two centimeters from the center to the edge of the chamber, at a radius of 20 centimeters. Magnetic diagnostics show that, in the single spheromak case, one spheromak is shot from one side of the chamber and reaches a relatively stable state within the chamber for about 60 J..ls. When the spheromak is in this stabilized state, the magnetic fields appear as in Figure 2.4. The two times during the shot that turn out to be most notable occur right after the plasma passes by the probe and then 5-15 J..ls later. Radial profiles of the flows in the chamber at these times as measured by both probes are shown below. 54 Figure 5.8 Radial profiles of drift velocity according to both the small and large probes at 28 J..ls during a single spheromak shot, right after the plasma has started to pass by the probe. These drift velocities are the averages of results from 10 different shots at each radial position. 55 Figure 5.9 Radial profiles of drift velocity according to both the small and large probes at 40 J..ls during a single spheromak shot, after the plasma has stabilized in the chamber. These drift velocities are the averages of results from 10 different shots at each radial position. 56 There are some notable similarities and differences between the flows according to these two probes. At both times during the experiment and for both probes, the azimuthal velocity is very low, though generally in the +8 direction. Early in the shot, both Mach probes agree that there is a steady flow of approximately 70 kmI s for r > 10 cm in the axial direction. However, while the larger Mach probe records flows of the same magnitude at r < 10 cm, the smaller Mach probe detects flow of SO kmI s or less. The results at t=40 J.ls are even less consistent with each other. At r > 10 cm, both probes agree that there is a reversal of flow at the edges such that, by this time, the flows are 40 I km s in the opposite direction. This reversal at the outer edge appears to be the result of a generally linear shear of flow in the radial direction according to the large Mach probe, which shows the inner radii maintaining an axial flow of SO km Is even 20 J.ls after the spheromak enters the chamber. The smaller Mach probe tells a different story. Its results suggest that the axial flows at r < 12 cm are actually very low by this time and even the flows near the center of the spheromak are reversed. 5.4 Counter-Helicity Merging Results Recall that magnetic reconnection can be induced within the chamber by fIring two spheromaks with opposing magnetic fIelds toward each other simultaneously. As discussed in the theory section, radial and azimuthal flows are expected as a sign of magnetic reconnection. Flows were measured at radii of 10 cm and 14 cm at the midplane by both the small probe and the large probe during these counter-helicity merging shots. The data presented are averaged over SO shots total for the small Mach probe and 100 shots total for the large Mach probe. Looking at the flows from the small Mach probe, shown in Figure 5.10, the axial flows at both of the radii at which measurements were made stay relatively close to zero, maxing out at around 15 km Is in the E-7W direction. In general, the axial flow stays slightly negative throughout the period shown. Concerning the azimuthal flows, the flow rises to about 20 kmI s in the positive azimuthal direction at both radii between 40 and SO J.ls. Between SO and 65 J.ls, the azimuthal flow at the outer radius drops to nearly zero while the azimuthal flow at the inner radius remains at 20-25 km/ s• 57 Figure 5.10 Drift velocity results from the small Mach probe during counter-helicity merging, measured at radii of 10 em and 14 em, averaged over SO shots at each radius. 58 Figure 5.11 Drift velocity results from the large Mach probe during counter-helicity merging, measured at radii of 10 em (bottom waveform) and 14 em (top waveform), averaged over 100 shots at each radius. 59 Turning to the results from the large Mach probe during counter-helicity merging in Figure 5.11, again, the axial flows are close to zero. The azimuthal flows are also very close to zero, except for a period at 35 ± 5 s, over which time the azimuthal flow reaches a maximum of approximately 10 km / in the +8 direction at r = 10 cm. At r = 14 cm, the azimuthal flow varies between 0 and 5 km / during this time period. 60 s s Chapter 6 Interpretation The fIrst observation that stands out from the results for each of the Mach probes is that, in some crucial respects, they do not agree. Though some general conclusions can still be drawn from the results attained here, this disagreement still requires some discussion. Recall, from Mach probe theory, that it is appropriate to analyze Mach probe data differently depending on whether the probe is in the magnetized or unmagnetized realm. The two probes used here are in two different realms, so there are several possibilities that could explain this discrepancy. First, recall that it was determined that the calibration constant for a Mach probe in the unmagnetized realm is highly dependent on the temperatures of ions and electrons in the plasma. This would suggest that the calibration constant for the smaller probe (unmagnetized) is changing over time and possibly even as the Mach probe is moved radially. For example, it seems that there is more disagreement between the probes at smaller radii. Perhaps this is because of differences in temperature that are only reflected in the results of the smaller probe. The extent of these temperature changes and the effect they would have on an unmagnetized analysis of a probe in an intermediate temperature range are unclear. Does this mean that we should trust the results from the magnetized probe more? Not necessarily, despite the fact that results shown here present many reasons to question the dependency of the smaller probe. The calibration constant, according to one-dimensional collisionless kinetic theory, should be a constant at k=1.7, independent of temperature and magnetic fIeld strength. However, this calibration constant was established using a one-dimensional analysis which assumed a steady direction of the magnetic fIeld in the direction of flow and perpendicular to the probe stalk. This is a highly idealized situation, and doesn't at all resemble the orientation of the 61 magnetic fields in the plasma chamber, especially during highly energetic events such as magnetic reconnection. This is not to say that the results from either of these probes is worthless, for the calibration of each probe clearly passed the inversion test and they clearly demonstrate measurements of flow, when disregarding electrical anomalies. 6.1 Single Spheromak Flow After a single spheromak is released into the chamber, it passes the midplane of the chamber with a radially constant axial flow of 70 km / s. Over the next 10-15 J.ls, a radial shear in the flow develops until a reversal of flow is exhibited at the edges of the chamber. The flows at the edge of the chamber during this period of time reach approximately 40 km / s in the opposite direction from the original flows. At this point, the velocities remain relatively constant, decaying until the energy has dissipated at about 100 J.ls. Throughout the shot, the azimuthal flow is practically zero, showing a slight tendency toward the positive azimuthal direction. These results are consistent with the frozen-in flux condition postulated by MHD for a plasma that has a Lundquist number of 1000. The lack of radial shear in azimuthal flow reflects the same lack of shear in the azimuthal magnetic fields of a spheromak at the midplane, as shown in Figure 2.4. The radial shear in axial flow, which probably reflects plasma crashing into the opposite end of the chamber and bouncing back around the edges of the chamber, follows the poloidal fields in the spheromak. There is little apparent flow that acts perpendicularly to magnetic fields in the spheromak. 6.2 Flow During Counter-Helicity Merging It is not immediately clear exactly what to expect from flows in coounter-helicity merging. One would assume that the axial flow at the midplane should remain at approximately zero, assuming symmetry but allowing for some fluctuations. The data presented above roughly agrees with this condition. In the azimuthal direction, we expect to see flows due to bi-directional jets from magnetic reconnection. The geometry of the problem is as pictured in Figure 6.1, which depicts two spheromaks coming together, focusing only in the r-z plane at r > O. 62 to roidal fields po loidal fields reoo nn ectlon plane Into page reoo nn ection plane outofPage8 ® z Figure 6.1 A schematic depicting the reconnection geometry of one side of two spheromaks in the r-z plane. The fields of the bottom spheromak are tilted out of the page and to the left, while those on the top spheromak are pointing into the page and to the right. In this figure, the spheromaks contain opposing fields. The top spheromak is left-handed (the poloidal fields move counterclockwise around a toroidal field going into the page in the -8 direction) and the bottom spheromak is right-handed (the poloidal fields move counterclockwise around a toroidal field coming out of the page). Since the magnetic field lines are actually a superposition of these two fields, the magnetic field lines about to reconnect from the bottom spheromak are pointing to the left and out of the page, while the magnetic field lines about to reconnect from the top spheromak are pointing to the right and into the page. This suggests that, using the model of two-dimensional Sweet-Parker reconnection, the left part of the reconnection plane is tilted out of the page and the right part of the reconnection plane is tilted into the page. Therefore, the bi-directional jets from reconnection are not solely in the radial direction, as they would be if the poloidal fields were the only ones present, but in the azimuthal direction as well. At radii inside the reconnection region, some positive azimuthal flow should be recorded (+8 is directed out of the page) , and at radii outside the reconnection region, some negative azimuthal flow should be recorded. In experiments performed at SSX last year, a bidirectional radial flow like that expected during reconnection was detected using ion Doppler spectroscopy [13]. This radial flow occurred at 40 ± 10 j..ls, in agreement with other diagnostics that this is the time period during which reconnection occurs. Data from the smaller Mach probe is difficult to parse. At both radii, positive 63 azimuthal flows occur on average, and they both seem to reach maxima of 2S km / " higher than we would expect the azimuthal flows to be, at around SS J..ls, well after the expected average time period for magnetic reconnection. The data during the period between 30 and 40 J..ls is broken up, though it appears that there are positive azimuthal flows at a radius of 10 cm during this period, as we would expect from the model explicated above. The larger Mach probe provides much nicer data. For the measurements at r = 10 cm, a peak flow of 10 km / s occurs at 36 J..ls. The peak drops to approximately no flow by t = 43 J..ls. The measurement at r = 14 cm seems to predict two separate bursts of lesser magnitude. Each burst reaches a maximum flow of 7 km / s in the -8 direction, and the bursts occur at 33 J..ls and 36 J..ls. There is no flow at this radius after t = 39 J..ls. These azimuthal flows are in agreement with the expected bi-directional outflows due to magnetic reconnection. 64 Appendix A Glossary Force-free - A plasma configuration is described as force-free if it is not necessary to use exterior energy sources in order to maintain the plasma. One of the most notable characteristics of such a plasma is that the cross product between bulk currents and the magnetic field is always zero Gyroradius - The gyro radius of ions in a plasma is defmed as the radius of ions in a plasma, assuming that they are traveling at their average thermal velocity and given the strength of the magnetic field in which they are immersed. More specifically, it is given by (3.7). Magnetohydrodynamics (MHD) - This is a model popularly used to describe plasma that treats the plasma as an electrically conducting fluid. This model focuses on analyzing the bulk flow behavior of the plasma rather than the motion of individual particles. Some versions of this theory include idealMHD, in which the fluid is assumed to be perfectly conducting, and resistive MHD, in which the fluid is assumed to have a locally invariant resistance. Poloidal- A component in the coordinate system used to describe toroids. This is the direction which goes through the hole in the middle of the toroid and around the outside. It is always perpendicular to the toroidal component. Reversed field pinch (RFP) - A formation of plasma similar to the tokamak except for some notable characteristics of the magnetic field. The toroidal field reverses direction as one moves outward from the center of the toroid, and there is a non-zero poloidal field. This formation is 65 advantageous because it takes less energy to sustain than the tokamak, but it is vulnerable to less predictable turbulence and non-linear effects. Spheromak - A force-free toroidal formation of plasma and the focus of research at SSx. This formation is characterized by the presence of both toroidal and poloidal fields, as well as its stability without the imposition of exterior energy sources. Stellarator - A formation of plasma, similar to the tokamak, in which the plasma is contained in a nearly toroidal shape and driven by an exterior current. To avoid the buildup of magnetic energy that is inherent toward the center of tokamaks, stellarators are twisted toroids, such that at one time a given section of plasma is immersed in a higher magnetic field and at another it is in a lower magnetic field. Tokamak - A formation of plasma in which the plasma is driven around a ring containing toroidal magnetic fields. These toroidal fields are maintained by strong electrical currents surrounding the ring. Due to the discovery of higher confmement modes, this remains the most promising formation for the prospect of fusion energy. Toroidal- A component in the coordinate system used to describe toroids. If one were to make concentric rings about the center of a toroid, the toroidal direction is defmed as the direction along the path of these rings. It is always perpendicular to the poloidal component. Z-pinch - A formation of plasm in which a straight, pinched line of flowing plasma is maintained between two electrodes. The pinch occurs due to magnetic pressure contained in a magnetic field surrounding the plasma. 66 Appendix B Argument for Flow in MHD Equilibria The following is an argument for the necessity of flow in any fonn of steady state magnetohydrodynamic equilibria. In section 2.6, where spheromak theory is introduced, it is protested that a spheromak in a steady state is not necessarily void of bulk flows. The following derivation is intended as a supplement to that argument. This reductio was referred to but not fully fleshed out in [32]. We begin with the resistive Ohm's law: E+vXB=1]J \ (B.l) and assume there is absolutely no flow and a steady state. Thus, the above equation and Faraday's law (V X E dB dt = -) reduce to - E=1]l (B.2) and (B.3) Now we invoke the spheromak force-free condition, discussed in section 2.6, and recall the following formulas: 1xB=Q VX B= AB = f.l) . 67 (B.4) (B.S) This is where the inconsistency occurs. The steady state Faraday's law (B.3) would imply that the curl of the electric field is zero; however, if we take the curl of Ohm's law with no flow (B.2), we get - 1]AB VxE= VX(1]J)= V x - (B.6) 110 which is clearly nonzero, since 1]A is a constant and the curl of the magnetic field is nonzero by 110 (B.S). This argument has a couple key implications. The first is that, given MHD in a force-free plasma formation, there is either necessarily some flow or the plasma is not in a steady state (that is, its magnetic field is constantly changing). The latter is not witnessed in the laboratory in the case of stable force-free plasmas, so we must settle on the former - that MHD dictates a constant flow in order to satisfy Maxwell's equations and the steady-state condition. Another strong possibility is that MHD is just too simplified, so we are using the wrong Ohm's law. An Ohm's law containing more terms has been postulated with some success, and it seems that MHD is now developing into a more complex theory, accounting for more factors that are integral to electromagnetic energy storage and dissipation in plasmas [33]. 68 References [1] K.H. Burrell et al., Phys. Fluids B 2, 1405 (1990) [2] H. Yamada et al., Phys. Plasmas 9, 4605 (2002) [3] S.P. Gerhardt et al., Rev. of Sci. lnst. 75, 4621 (2004) [4] U. Shumlak et al., Phys. Plasmas 10, 1683 (2003) [5] ]. Ghosh et al., Phys. Plasmas 13, 022503 (2006) [6] R.F. Ellis et al., Phys. Plasmas 12, 055704 (2005) [7] M. Cecconello et al., Plasma Phys. Control. Fusion 48, 1311 (2006) [8] S.c. Hsu et al., Phys. Plasmas 8, 1916 (2001) [9] Y. Ono et al., Phys. Rev. Lett. 76, 3328 (1996) [10] A. Falk, Dynamics ofField-Reversed-Configuration in SSX, Senior honors thesis, Swarthmore College (2003) [11] F.F. Chen, Introduction to Plasma P0;sics and Controlled Fusion Vol 1: Plasma P0;sics, Second Edition, (plenum Press, N ew York, 1984) [12] S.R. Seshadri, Fundamentals of Plasma P0;sics, (American Elvesier Company, New York, 1973) [13] ]. Fung, High Resolution Flow and Ion Temperature Measurements with Ion Doppler Spectroscopy at SSX, Senior honors thesis, Swarthmore College (2006) [14] D. Biskamp, Magnetic Reconnection in Plasmas, (Cambridge University Press, Cambridge, 2000) [15] P.A. Sweet, Electromagnetic Phenomena in Cosmical P0;sics (Cambridge University Press, New York, 1958) [16] E.N. Parker,]. Geophys. Res. 62, 509 (1957) [17] E. Priest and T. Forbes, Magnetic Reconnection: MHD Theory andApplications, (Cambridge University Press, Cambridge, 2000) 69 [18] M. Landreman, The Three-Dimensional Structure ofMagnetic Reconnection in SSX, Senior honors thesis, Swarthmore College (2003) [19] P.M. Bellan, Spheromaks, (London: Imperial College Press, 2000) [20] lB. Taylor, Rev. Mod. Phys. 58, 741 (1986) [21] CG.R. Geddes et al., Phys. Plasmas 5, 1027 (1998) [22] CG.R. Geddes, Equilibrium Studies on SSX, Senior honors thesis, Swarthmore College (1997) [23] LH. Hutchinson, Phys. Rev. A 37,4358 (1988) [24] K.S. Chung and LH. Hutchinson, Phys. Rev. A 38,4721 (1988) [25] M. Hudis and L.M. Lidsky, l Appl. Phys. 41, 5011 (1970) [26] LH. Hutchinson, Phys. Plasmas 9, 1832 (2002) [27] LH. Hutchinson, Plasma Phys. Controlled Fusion 44, 1953 (2002) [28] V. Chaplin, High Time Resolution Measurements ofElectron Temperature in the SSX Plasma, Senior honors thesis, Swarthmore College (2007) [29] T. Gray, Densiry Studies on SSX, Senior honors thesis, Swarthmore College (2001) [30] T.W. Kornack, Magnetic Reconnection Studies on SSX, Senior honors thesis, Swarthmore College (1998) [31] C MacLatchy et al., Rev. Sci. Inst. 63, 3923 (1992) [32] M.R. Brown, l Plasma Physics 57, 203 (1996) [33] CD. Cothran et al., Geophys. Res. Lett. 32, L03105 (2005) 70 Acknowledgements I would especially like to thank my advisor and the founder of the SSX lab, Dr. Michael Brown, for his eagerness to share his expertise in this field and his willingness to let me playa big role in his project. Also, I would be remiss if I failed to recognize his boundless patience with my working habits. Though David Cohen was not a constant presence in the SSX laboratory, his eagerness to probe scientific issues was inspiring, and I appreciated having him as a colleague and an advisor. Chris Cothran was also very important to the success of this project, as a mentor who was willing to help me get my feet wet when I was new to the field, even when he was busy with a multitude of his own projects. Finally, I would like to thank all of the fellow students that have worked with me on this project, whose unfailing kindness and diligence have helped make my experience at the Swarthmore Spheromak Experiment both a fulfilling and a productive one. They are Brie Coeliner, Marc Chang, Vernon Chaplin, Jerome Fung, and Victoria Swisher. 71