Evolution of the Plasma Universe:  I. Double Radio Galaxies, Quasars, and Extmgalactic Jets






Abstract—Mosaic piahma ph yzicz and our concept of the univ•••• '•      fields,   nonlinearity,   nonhomogeneity,    and   explicit  time


n a state of rapid revision. This change started witb in-i’ifu

ments of plasmas in Earth’s ionosphere, cemetery atmospheres, and planetary magnetospheres, the translation of hnowledge front laboru- tozye*peflmenCto sstrophysc*Uphenomenm, dbco*eHeofbMicdnnd

dependence in both the laboratory and magnetosphere has instilled confidence that the electromagnetic particle-in- cell approach can be used to study plasmas not accessible


illamentary plasma structures in the Galaxy and double radio sources,         IO  th  8fftl  measurement.

and the particle simulation of plasmas not accessible t '•-•      ••••-           Because  of  these  recent  developments,   our  concept  of


surement. Because of these, Birkeland (field-aligned) currents, double

layers, and magnetic-field-aligned electric fields are now hnown to be far more important to tbe evolution of spsce plesma, Including the ac- celeration of charged particles to high energies, than previously thought. This paper and its sequel Investigate the observational evi- dence for a plasma universe threaded by Birkeland currents or fila- ments. This model of the universe was insplred by the advent of three- dimensional fully electromagnetic  particle simulations and their  appli-

cation  to  the  study  of  latioratory  4  pinches.  This study  resulted  in to-

tally unexpected phenomena in the deta post-processed from the sim- ulation particle, field, and history dumps. In partlcular, when the simulation parameters were scaled to galactic dimensions, the inter- action between pinched filaments led to synchrotron radiation whose emission properties were found to share the following characteristics with double radio galaxies and quasars: power magnitude, lsophotel morphology, spectra, brightness along source, polarization, and Jets. The evolution of these plnched synchrotron emltting plasmas to ellip- tical, peculiar, and spiral galaxies by continuing the simulation run is addressed in a sequel paper.




cosmic plasma has changed considerably. Essential dif- ferences between the old and new paradigm include the following.

  • Electric double layers, which did not attract much attention a decade ago, are now known to accelerate charged particles to kilovolt energies in the terrestrial magnetosphere [43—[7]. They may also exist elsewhere and accelerate particles to even higher
  • The necessity of a global electric current description to describe the transfer of energy in magnetized cosmic plasmas. This leads to the necessity of drawing the cir- cuits in which the current flows {8], [9]. (This approach also emphasizes the discrete particle description in simu- lating cosmic phenomena, where particle currents may be explicitly )
  • In the magnetospheres, plasmas exist in both active

and passive states; this is probably  true for all   cosmic




PLASMA physics is at present in a state of which  is so rapid that it is appropriate  to

4)    Cosmic plasmas  often are not homogeneous,  but

rather exhibit filamentaiy strictures which are likely to be


speak of a change in paradigm [1]-[3]. This change started a decade ago and is precipitated by the following.

  1. fit situ measurements of the properties of plasma in the magnetospheres, including the solar magnetosphere. Active space experiments are beginning to make signifi- cant
  2. Laboratory investigations of plasmas with dimen-

sions ranging from miciometers to centimeters have pro- vided an increased understanding of bow to transfer knowledge from one plasma region to another and from one size to another.

  1. Recent discoveries including the observation of hel- ical and filamentaiy structures in the galactic center and extragalactic radio
  2. Multidimensional particle-in-cell simulations. The solution to problems with complex geometry, intense self-


Manuscript received January 2, 11186; revised August 2, 1986.

The author is with Los Alamos National Labomtory, Los Alamos, Nhi


IEEE Log Number 8610981.

associated with currents parallel to the magnetic field.

  • In the magnetospheres there are thin stable current layers which separate regions of different magnetization, density, and tempemture. Similar phenomena must also exist in more distant
  • In the case where a current flows in a partially ion- ized plasma, a chemical separation may take place. Due to this and other effects, space plasmas have a general tendency to be separated into regions of different chemical


The necessity of employing plasma physics to account for the observed electromagnetic radiation fmm cosmic sources has been pursued by a number of authors. In par- ticular, Sturrock [10] and Sturrock and Barnes [11] pro- posed a magnetized plasma radiogalaxy model in which the tearing modes in current-conducting sheets play an important role in the radiation and morphologies ob- served. Alfv6n postulates the existence of two neighbor- ing double layers of radio lobs dimensions in a helio- spheric pinched-current model involving the central el- liptical galaxy [12]. The importance of pinched  plasma


IXl93-3813/86/120t1-0639501.0£I O 1986 IEEE


M0                                        IEEE  TRAi"JSACTIONS  ON  PLAS5'iA  SCIENCE,  VOL.  PS-14,  NO.  6,  DECEMBER 1986



currents, beam instabilities, and hlamentation to radio- galaxy processes has been pursued further by Peratt and Green [13], Browne [14], and also by Lerner [15], who points out the close similarities between measured radia- tion spectra from cosmic .sources and that of the dense plasma focus and other pulsed power laboratory devices. Bostick’s theory of radio sources is based on his dense plasma focus investigations [16].

This paper investigates the filamentaiy electric-current aspect of cosmic plasma. Section II describes the basic model: interactions among galactic-dimensioned field- aligned current filaments. Section III describes the anal- ysis of the model with three-dimensional electromagnetic particle-in-cell simulations. The Biot—Savart force law for filaments  is discussed  in  Section  IV,  while synchrotron

have a general astrophysical interest far beyond that of understanding the space environment of our own Earth.”


  1. Birkeland Currents in Laboratory Plasma

In the labomtoiy, hlamentaiy structure is a common morphology exhibited by energetic plasmas. X-ray pin- hole photographs, optical streak and framing camera pho- tographs, and laser holograms often show a filamentary magnetic “rope-like” structure from plasmas produced in multiterawatt pulse power generators or in dense plasma focus machines. High-resolution etchings of electron beams onto witness plates show nearly identical vortex prohles ranging from a distension of a few micrometers in the dense plasma focus, to a few centimeters in cathode


radiation  from pinched

Sections VI and

is given  in Section  V .

radio sources, quasars,

electron beams [23]—[26]. This size variation of four or- ders of magnitude is extended to nearly nine orders  of


and magnetically  confined sheet electron beams  (jets).

The author’s conclusions are given in Section VIII. The evolution of cosmic plasma beyond the time fee tigated in this manuscript is presented in a sequel (denoted Paper II).

The only assumption made in the analysis in this pa- per—if it should be called an assumption—is that the basic properties of plasmas are the same everywhere, fmm sub-

millimeter dimensions to the Hubble distance (102a cm)

[l]—[3], [8], [9].



An electromotive force P = Je X B - de giving rise to electrical currents in conducting media is produced wher- ever a relative perpendicular motion of plasma and mag- netic-field lines exist [8], [17]—[19]. An example of this is the (nightside) sunward-directed magnetospheric plasma that cuts the earth’s dipole field lines in the equa- torial plane, thereby pmducing a potential supply that drives currents within the auioral circuit. The tendency for charged particles to follow magnetic lines of force and therefore produce field-aligned currents has resulted in the widespread use of the term “Birkeland currents” in space plasma physics [20], [21). Their discovery in the earth’s magnetosphere in 1974 has resulted in a drastic change in our understanding of aumra dynamics, now attributed to the filamentation of Birkeland charged-particle sheets fol- lowing the earth’s dipole magnetic-fteld lines into vortex current bundles. In anticipation of the importance of Birkeland currents in astrophysical settings, Falthammar states [22]:

“A reason why Birkeland currents are particularly interesting is that, in the plasma forced to carry them, they cause a number of plasma physical processes to occur (waves, instabilities, fine structure formation). These in turn lead to consequences such as acceleration of charged panicles, both positive and negative, and element separation (such as preferential ejection of ox- ygen ions). Both of these classes of phenomena should

magnitude when aumral vortex recordings are directly compared to the laboratory data [27]. With regard to ac- tual current magnitudes, fine-detail resolution of current filaments shows indistinguishable vortex patterns over nearly 12 orders of magnitude while coarser resolution shows that the phenomena probably tmnscend at least 14 orders of magnitude, fmm microampere to multimega- ampere electron beams.


  1. Birkeland Currents in Cosmic Plasma

As far as we know, most cosmic low-density plasmas also depict a filamentaiy stricture. For example, filamen- taiy structures are found in the following cosmic plasmas, all of which are observed to be associated with or are likely to be associated with electric currents.

  • In the aumra, filaments parallel to the magnetic field are very often observed. These can sometimes have di- mensions down to about 100
  • Inverted P events and the iii-situ measurements of strong electric fields in the magnetosphere (10 10‘ A, 10' in) demonstrate the existence of filamentary struc- tures.

3)  In the  ionosphere  of Venus,   ‘flux  ropes,” whose

filamentary diameters are typically 20 km, are observed.

  • In the sun, prominences (10'' .4), spicules, coronal streamers, polar plumes, , show filamentaiy structure whose dimensions are of the order 10 —10' m.
  • Cometaiy tails often have a pronounced filamentaiy

stricture (28].

  • In the interstellar medium and in interstellar clouds there is an abundance of hlamentaiy structures, e.g., the Veil nebula, the Lagoon nebula, the Orion nebula .(Fig. 1), the Crab nebula,
  • The center of the Galaxy, where twisting plasma fil- aments, apparently held together by a magnetic field possessing both azimuthal and poloidal components, ex- tend for nearly 500 light years (5 x 10" m) [29].
  • Within the  radio bright  lobes of double  radio  gal-

axies, where filamental lengths may exceed 20 kpc (6 X

in'°  › Iool.


PE9ATT:  EVOLUTION  OF'  Tj-fE  PLASMA  (jyyjy  j¿  g; j


Fig. 1. Detail of the Orion nebula ( $ky and Teles••R•, Nov. 1979). This paper investigates what an observer at f2 would see when looking into filamental interactions at points P in a plasma universe of similar struc-




  1. The Formation of Double Layers Within Birkeland Currents

Recent literature in the area of magnetospheric physics reflects considerable interest in magnetic-field-aligned electric fields. Such electric fields can have important consequences in cosmic plasma [31], [32], including the “unfreezing” of magnetic fields, the acceleration of elec- trons to very high energies, and the filamentation of the plasma itself.

In magnetized nonhomogeneous astrophysical plasma, a number of mechanisms are present that can generate field-aligned electric fields. These include anomolous re- sistivity caused by wave-particle interactions, collision- less therinoelectric effect due to energy-dependent wave— particle interactions, magnetic mirror effects involving trapped particles and magnetic-field gradients, and elec- tric double layers due to localized charge separation. While all of the above mechanisms have been studied in the laboratory and simulated by computer, it is the last mechanism that has been found to be remarkably prolific in producing appreciable potential drops in neutral plasma. Moreover, Birkeland currents and double layers appear to be associated phenomena, and both laboratory experi- ments (33] and computer simulations (34] have shown the formation of a series of double layers along current-car- rying plasma columns or filaments. Fig. 2 illustmtes the geometry at hand. When double layers (or a series of dou- ble layers) form in adjacent Birkeland current filaments, field-aligned electric fields are generated, which then serve to accelerate electrons within these regions.





F'ig. 2. Double layers in adjacent Birkeland cunenc filaments.




Fig. 3. Basic geometry under considcmtion; two pamllel Birkeland cur-




  1. Interacting Birkeland Currents Model

It is the purpose of this paper to extend the study of cosmic plasma to the case of galactic-dimensioned (50 kpc in width) Birkeland filaments by means of three-dim.en- sional, fully electromagnetic, and relativistic particle-in- cell simulations. Fig. 1 is a contest-enhanced photograph of the Orion nebula but serves the purpose of representing the morphology to be expected by an observer situated within a much larger filamentaiy metagalactic structure.

The simulation m6del consists of modeling a magnetic- field-aligned neutral plasma tilament (column) in the pres- ence of a field-aligned electric field. (Strictly speaking, because of the parallel electric field, the portion of the filament simulated is a double layer (35J .) To study the evolution of intemcting filaments, a second filament (nearly identical to the first) is placed adjacent as depicted in Fig. 3. (As many as six filaments have been investi- gated by simulation while up to 12 filaments have been studied experimentally. However, because of the r ' force between filaments, it would appear that a majority of cosmic plasma phenomena are the result of two, or at most


M2                                   IEC::E  TRANSACTIONS   ON  PLASMA  SCIENCE.  VOL.   PS-14,  NO.  6,  DECEMBER   1986



three, interactions among the closest filaments.) The re- mainder of this paper is concerned with what the signa- tures of existence would be to an observer situated within a nonhomogeneous plasma universe consisting of galac- tic-sized  Birkeland currents.



Specification of plasma density, temperature, magnetic field strength, acceleration field, and filamental width set the initial conditions for simulation. The parameters that delineate the physical characteristics of a field-aligned current-carrying filament of plasma are the electmn drift velocity

4  '“t/c                                         

the plasma thermal velocity


Because of the EMP-induced current f„ a galactic fil- ament can be expected to retain its columnar filamental form provided the Bennett-pinch condition is  satisfied,

i.e. , that


where ?If is the electron density per unit length [41].

In addition to confining plasma in filaments radially, the axial current flow produces another important effect; a long-range interactive force on other galactic filaments [2], [13], (38]. The Biot—Savart electromagnetic force be- tween filaments is


for all space, where y x II is the Lorentz force. If  the




and the thermal/magnetic  pressure  ratio

n kT  + n;kT;       [(‹»$ dr)(hp/A)]' 4(1 + T;lT ) B l 2                     (cdtlk )’ (( o› zlu›p ) ]


current path greatly exceeds the filament widths, the at- tractive force between two similarly oriented filaments is approximately  given by


where the subscripts  1 and 2 denote columns  I  and  2,



where n, = n; is the neutral plasma density, his the plasma temperature, k is Boltzmann’s constant, » is the perme- ability of free space,and the subscripts c and i denote elec- tron and ion particle species, respectively. The parameter dt 'is the simulation time step, A is the cell size, and c is the speed of light.

We choose a plasma temperature typical for cosmic Birkeland hlaments, a few kiloelectronvolts [31), by set- ting the initial dimensionless  simulation parameters  to

‹»p dt ——  0.25 (electron plasma  frequency), hp/A  = 0.25

respectively, and $‹2 is their separation. Because of the axial magnetic field By, the particles spiral as they drift or accelerate and thereby produce an azimuthal component in the generalized current I = zf, + 8f . The magnetic moment  associated  with  the  azimuthal  current  is  in  =

zB R’ ——          2      If the magnetic moments in adjacent lil-

aments are aligned, a short-range repulsive force is gen- erated  between them:




(the Debye length in cells), and cdtlb —-  1.0 (the speed    Hence, the electromagnetic  forces between filaments  are


of light). A field-aligned Birkeland filament is established          *

as It j'  (long-range  attractive) and It      (short-


by means of the parameter ‹»,q/‹»P ——   1 .5 where y  0 =    range repulsive).

eB! !and B.g —-  B(t ——  0) is the axial magnetic field.       During long-range  attraction,  the motion of either fila-


For this choice of parameters, Ash  - 0.0625 and, for T

= T;, ¢, = 0.0069. Current flow within the filament is initiated  by  setting E.  B.-     0.01c, so that  0  fi f1    :   1. The simulations are carried out with the three-dimen- sional  electromagnetic  particle  codes SPLASH and TRIS-

ment in the intemction region may be approximately de- scribed by the equation Mdr ldt’ -- ‹•*'*/4r(n — r), the solution of which is

u  = drldt  -—  *t  nL/2r3 )l/2  (ln a/a  —  r) '2  + u(0)     (8)


TAN [36]. The former code consists of 32 768 cells and where L is the length of the filamental region involved in 250 000 particles while the latter code has 4 194 304 cells, Biot-Savart attraction, if is the total mass, 2a is the dis- 2 million photons, and 5 million particles. For SPLASH, tance of separation between filaments, and v(0) is the rel- the radius of each filament is 3A and the center-to-center   ative velocity at time zero. For the case ii(0) = 0, and if

separation is llA, while for TRISTAN the radius and sep-  thC filaments are sufficiently sepamted so that the loga-

aration are  12A and 44A, respectively.                                         FlthmlC COiTection  is of order unity,  the  velocity  is   ap-

SPLASH has been benchmarked against the in = 1 hel-    proximately given by

ical instability in magnetized z pinches [37], interacting                        plasmoids  [38],  microampere  to  submegaampere cylin-


drical charged particle beams [26], thin-sheet beam prop- agation experiments [39], auroral magnetic storms (39], and barium plasma experiments in the magnetosphere [40].

{v - f,(2L/c'3f)'    - Std/2  2L/ f)ccs-           (9)

In dimensionless gaussian simulation units, (9) is




where  fi  =  N (up      3  (           +   1)rr) k  and  N  are the total simulation mass and number of interacting simula- tion filaments, respectively. The parameter 8 = dlv is the distance between filaments in cell widths, r, = r lk is the radius of a filament in cell widths, and

°é —— Bl(mcu›p ie) —— u› lump                        

is the magnetic field in dt time steps (for -—  B )  .


  2. The Application of the Synchrotron Mechanism to Cosmic Plasma

One of the most important processes that limit the ener- gies attainable in particle accelerators is the radiative loss by electrons accelerated by the magnetic field of a beta- tron or synchrotron. This mechanism was first brought to the attention of astronomers by Alfvén and Herlofson [42]; a remarkable suggestion at a time when plasma, magnetic fields, and laboratory physics were thought to have little, if anything, to do with a cosmos filled with “island” uni- verses (galaxies). While the Alfvén—Herlofson proposal was cast in terms familiar to astronomy, it is clear that the suggestion pertained to radiative emission of relativ- istic electrons in the trapping field of the interstellar mag- netic field, or in an intergalactic magnetic field.

Synchrotron radiation is characterized by a generation of frequencies appreciably higher than the cyclotmn fre- quency of the electrons, a continuous spectra (for a pop- ulation of electrons) whose intensity decreases with fre- quency beyond a critical frequency (near-intensity maxima), increasing beam directivity with increasing  y

(        (          2)  1/2)  and strongly polarized  electromag-

netic wave vectors. Many excellent reviews of synchro- tron radiation in laboratory and astrophysical sources are to be found in the literature [43]—[50].

The recognition that this mechanism of radiation is im- portant in astronomical sources has been one of the most fruitful developments in astrophysics. For example, it has made possible the inference that high-energy particles ex- ist in many types of astmnomical objects; it has given additional evidence for the existence of extensive mag- netic fields; and it has indicated that enormous amounts of energy may indeed be generated, stored, and released in cosmic plasma.


  1. Experimental Studies of Synchrotron-Emitting Bennett Pinches

Charged particle beams held together or pinched by their self-magnetic fields have been of general interest since their earliest investigation by Bennett [41]. The macroscopic picture of such a beam is that’ of a self-con- sistent magnetic confinement or compression against the expansion due to thermal pressure (4). On the micm- scopic scale, the individual particle orbits include radial oscillations due to the Lorentz force superimposed on the drift in the direction of mean flow. These are the betatmn oscillations. Since they imply particle acceleration, there


is electromagnetic radiation associated with them. Be- cause the force is a e x B force, the radiation from the relativistic  electrons is synchrotron radiation.

Manifestations of the pinch effect appear for a labora- tory observer as a rapidly occurring phenomenon. A burst of radiation from high-current discharges (with current densities of the order 10'' A/cm'), such  as  low-induct- ance vacuum sparks, plasma focus devices, and exploded wires is found over a broad spectral range: from the mi- crowave region to the hard X-ray region [51]—[54J . Re- corded data show that the radiation bursts are correlated with dips in the current waveform, i.e. , interruptions in the current flow. Analysis of the directional patterns of  the millimeter radiations shows that the microwave radia- tion is synchrotron radiation of electrons in the magnetic field of the proper current. The hard X-ray quanta are at- tributed to synchmtron radiation from the electrons at the transitions between Landau levels in this same current- induced magnetic field [55].

The total synchrotron power radiated incoherently from a Maxwellian distribution of election velocities, over all frequencies, is given by [563


,      ergs/cm'.       (12)


Similar expressions applicable for the determination of the average power radiated per unit length for relativistic Bennett pinches have been given by Meierovich [55) and Newberger [57], [58].

An enhancement of radiated power, given by (12), is achieved when the sum of the r x B radial forces seen by the ielativistic electrons (5) is increased, ask is the case when the azimuthal magnetic fields of neighboring pinches are present. While no theoretical treatment of synchro- tron enhancement from beam interactions is known, this phenomenon has been examined in some detail, both ex- perimentally and with simulations.

Fig. 4(a) illustrates the radiation yield versus the num- ber of columnar Bennett-pinched plasma filaments. The experimental data (solid dots) pertain to 30-mm-long 15- a-diameter metallic wires strung between the cathode and anode of a terawatt pulse-power generator [25]. In all cases, the delivered pulse is approximately 50 us long, while the radiation burst duration is approximately 5 ns long. As illustrated, an X-ray energy enhancement of 6 is obtained when two hlaments interact and enhancement factors — 12-30 are recorded when up to 12 filaments symmetrically ariayed about a common center interact. Laser shadowgiaphy diagnostics have shown the follow- ing sequence of events leading to broad-band electromag- netic radiation in pulsed power, Before the current teaches a value sufficient to fully establish the Bennett pinch (4) in’the conducting filaments, plasma flows off the filaments towards the geometrical center along a path defined by the magnetic separatrix (see Section YH). The subsequent collision (interaction) of the inflowing plasma is photo- gmphically recorded in the emission of X rays by means











































Fig. S.  Simulation synchrotron mdiation characteristics for two intemcting




























Fig. 4. (a) Radiated energy versus number of filaments. Solid dois—N- shell X-ray energy for I MV, 1.3 MA delivered lo single wire and wife array loads. Hollow dois—splash simulation. (b) X-my pinhole photo- graph (seen side-on) of plasma tripped in the magnetic sump between wires {25]. Wife filaments in absorption of X rays ftom the cenlml emit- ting region are observable in the photograph. While lhe wife length ex- cCcds the array diameter for the laboratory case, a double-layer-produced electric field within a galaxy may be of iht: order of or smaller than the distance between  intemcling Birkeland currents.



of pinhole cameras using X-ray film plates. The photo- graphs show the topology of the emitting plasma config- uration during the 5-ns burst, as-well-as the absorption profiles of the filamental pinches in between the camera and the emitting plasma, and the radiation “lit” surfaces of the filamental pinches behind the emission region. Fig.

4(b) shows that the morphology of the radiating plasma, at radiation burst, is both helical and filamentary. This stricture and the radiation enhancement is recorded only when the number of wires is greater than 1, i.e., the wires are interacting electromagnetically. The helicity is ex- tremely well defined in some X-ray photographs and dem- onstrates the presence of an axial component to the mag- netic field, even if none is externally imposed initially. (The second paper in [23] shows in a contact print on CR39 etchable plastic track detector a vortex pattern when no axial Sd, field was used in the plasma focus. This shows how the same morphology can be delineated using a to- tally different technique.) Late time-streak and framing camem diagnostics show the Biot—Savait attraction, ro- tation, and coalescence of the dense filaments themselves [25].

  1. Simulation Studies  of Synchrotron-Emitting Bennett


Fig. 5 is a plot of the simulation electric and magnetic energies lost as synchrotron radiation in arbitrary energy units (AEU) versus time in inverse units of plasma fre- quency ‹» ' (37].' Whenever the attractive force between simulation columns causes their sepamtion to be reduced to a distance such that the repulsive force (7) starts to become compamble to the attractive force (6), a burst in the radiation occurs (Fig. 5). For the parameters used in these simulations, this distance is of the order of several


'The author thanks a rcfcmc for pointing out th8t the arbitrary energy unit is not quite arbitrary, hut cotzcsponds to the totat    initial particle thcr-




pinch radii. As shown in Fig. 5, the radiation from the kiloelectronvolt particles is polarized in the transverse plane and the synchrotron enhancement (burst) is detected

in the z and y electric radiation energies (figs, HER,) and the z magnetic radiation energy (lFqp). The burst lasts until the induced axial magnetostatic energy lFg„ due to

the azimuthal current Off, is depleted (because the coun- terparallel azimuthal current force (7) brakes the azimu- thal electron flow in both filaments). For some simula- tional parameters, fig, can build up and discharge again in the form of additional bursts of synchrotron radiation. The long-time slowly varying increase in radiation in lFgg and lFgg is due to the buildup of electrostatic energy from charge separation in the particle number and size con- strained simulation model.


  1. DouacE RxuIO SOURCES

The existence of double radio galaxies presents a major challenge to cosmological theories. The discovery of dis- crete radio sources dates back to the pioneering survey of Reber (1944), who found two areas of enhanced intensity in Cygnus and Cassiopeia [59]—(62]. Cygnus A, the brightest radio source in the constellation Cygnus, has proved to be the “prototype” of double radio galaxies, and models of double radio galaxies are usually based on the characteristics of this source. Many excellent reviews on the properties of double radio sources observed from the 1960’s on are available in the literature [63]—(66], as are a number of models of sources. However, regardless of whatever ingredients are postulated as necessary in models used to “explain” their existence, what is ob- served from any radio source is synchrotron radiation,


Additionally, in d) it is also necessary to explain the ra- diation “hot spots” in classical radio doubles; their dia- metric displacement off an imaginary axis running from giant radio lobe to giant radio lobe through an elliptically shaped center; the relationship between “bridges” and so-called jets connecting the elliptical center with the lobes; the one-sidedness of jets in quasars and powerful sources; multiknotted features in hot spots, bridges, and jets; the apparent superluminosity of some sources; and the pmblem of a nearly uniformly fading of some jets, thousands of light years in extent, in a few decades.


  1. Scaling Simulations ie Galactic Dimensions

The scaling of plasma physics on cosmical and labo- ratory scales generally involves estimates of the diffusion in plasma, inertia forces acting on the currents, the Cor- iolis force, the gravitational force, the centrifugal force, and theJ x B electmmagnetic force [67], [68]. The sim- ulations reported in this paper are scaled to Cygnus A using the latter force law via (9).

Cygnus A consists primarily of two radio lobes of  di-

mansion  --  w  —-   1021      (35 kpc)  sepamted  on either

side of an elliptical galaxy  by  a distance a      1.22  X  10" in [43]—[46], [63]—[66], [69]—[72]. Typically, estimates of the relative velocities between cosmic plasmas mnge from hundreds of kilometers per second to 1000—8000 km/ s (for interactions between components of peculiar gal- axies [73]). Random velocities characteristic of the veloc- ity dispersion of galaxies in highly condensed clusters are of the order 2000 km/s [74), [753. Assuming a total plasma mass for the radio lobes of Cy  nus A of the    order


which requires only the presence of relativistic   electrons

of that observed in galaxies, M --

10     kg (while noting


in a magnetic field 2

  1. Requirements of the Model

Any plausible theory on double radio galaxies should be expected to explain the following:

  1. the origin and source of energy of double radio gal- axies;
  2. the total magnitude of the radio flux observed;
  3. the measured flux density as a function of fre- quency;
  4. the observed isophotal morphologies;
  5. the spatially varying power law within a source;
  6. the polarization properties of the incoherent syn- chrotron radiation measured; and
  7. the lifetime and evolution of a


'In addition to the model described in this paper, there am at lcest two other related plasma models of double mdio galaxies. Alfvtn proposes a circuit on the galactic scale where the cumnt is driven by the dynamo ac-

that all mass estimates are model dependent), and setting the velocity of attraction between filaments to 1000 km/ s, yields I, - 2.15 x 10"  A and B  --  I lw =--  2.5 X 10 ' T (2.5 x 10“ G). (The quantities I, and 5 are physically nonsensitive to the actual mass distribution, depending only on the square root of the mass per unit length; a slower velocity of 1000 km/s is chosen since this velocity increases with increasing current as the source develops in time.)

To convert simulation results to dimensional form, it is sufficient to fix the value of one physical quantity, e.g., 21 . Since we are scaling to the strong radio source Cygnus A, the value of 5 is applied at time step 90, the peak synchrotron burst energy. At this time, Bg has grown comparable in strength to 5 so that JI¿ = 0.0034. Ve- locity counts show T 2.8 keV at time T -- 90. 8ub- stituting Qp, Tz , and B into 3) yields a mean plasma den-

sity  n   =  1.79 x  103            (1.7   x   10  ' cm  '). Since

= 1.5, the axial magnetic field strength is S,q =


tion of a rotating galaxy. Ineidc the galaxy, the cumnt flows in the plane

of symmetry; outside, it follows back to the plane of symmetry. Alfv6n

2.0 x

10 ' T (2.0 x 10    G). These parameter values


suggests that double layers may occur in the axial parts of the ciictlit on either side of the centml galaxy, leading to double mdio lobe emission regions. Bostick proposes a model where electfomagnetically attmcting plasmas join to make a bar which is the armatute of a homopolar generator. Two plasma foci at the center of the armatuie shoot jeu of plusma perpen- dicular to the galactic plane, each of Which terminates in a tadio lobe.

characterize Cygnus A and are in close agreement with many previously published estimates using independent means (Table I). Additionally, from the simulation pa- rameter E z ! • -- 0.01c, the acceleration field within a filament is £,q = 63.4 mV/in.


                                                                                    IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. PS-14, NO. 6, DECEMBER  1986








Simulation derived value


Estimate (model and


Glactic  current,  f., A

2. 15 x 10"



Galactic magnetic field,

B. G

2.5 x l0—’ (Sp)

2.0  x  i0-’ (B z )


3. I6 x 10—*

Perlcy et ul. (3979)

Mills and Sturtock (1970)


3.00 x 70-*

1.20 x I0—°

t0—°— l0—'

de Young and Axfo•! t 1967) Hargrave and Ryle (1974) Perola (1981)

Thermal plasma


I — TO

Miley (l980J


tern perature, T, , keV

PI a sma  density,  n, ,  cm   ’     l .79 x  I0— °




Density of synchrotron         6.9 x  I0-’ omitting eleccrons,

4 x 10—*—2 x 10— °

io—•—io -'

0.6 x  T0-'


I.5 x l0—*

Gisler and Miley  (1979)

Pcrley ci uf. (1979) Miley (1980)

Gisler and Miley (1979)

Shklovsky ( 1960) Ginzburg and

Syrovatskii  (1963)



f'„.„  ,  ergs  s   '

6.2  x 10" (burst average)

8.1 x 10” (classical

1.6 x I0•’

4.4 x l0•’

Motfet (1975)

Shklovsky  (1960)




8.00 x 10‘


Sturrork and Barnes ( 1972)


de Young and Axford (1967)


S x 10*

Ryle and Windram (1968)

Total source energy,

6.30 x IO"


Sturrock and Barnes ( I9?2)

er 8'


de Young and AxFord (1967)

Average energy per



Perola (198 I )

electron, 7, , NfeV



Sturrock ( 1969)



To scale the simulation  spatial and  temporal dimen-    should be traversed  in d(16/ 1836)  x  2l92df’ = 205di'. sions,  A  =  4hp  =  2.97  x 1 4      and dt —-  (4ui )  ' =     The  measured  simulation  traversal  time  is somewhat

1.04  x   10   4  s,  to  Cygnus  A  re  uires  a  size/time  multi-        longer,   -3Wdt’,  because  of  the  presence  of  repulsive

plication  factor of ‹x  = 5.6 X             For this scaling, con-    forces  in the  full  simulation  model  (7) that  are not   in-

servation of the speed of light  dictates that                         eluded in (10).


cdt _ cdl’ _

A   " A’   ” '

where  A’  = ixA   = I .6d  X   10"    m  and  dt’   ——   odf   ——    5.87 X 10' ' s are the galactic equivalent cell and time step, respectively. The values of n, T, B, and A remain the same regardless of whether the simulations are scaled to A and dt or to A' and  dt'.

One immediate consequence of the rescaling is   that,

while the dimensionless simulation parameters remain un- touched, the resolution is reduced, i.e. ,

‹»p dt ——    › dt’ -— 0.25                            (14)

  1. Formation of Hot Spots in interacting Filaments

When the attraction between adjacent Bennett-pinched

Birkeland filaments reduces their mutual separation to a few  filamental  radii, the r‘4    pulsion  (7) produces  a re-

distribution of the synchrotron-emitting relativistic elec- tron currents within the filaments. The result of this re- pulsion is an “edge-brightening” (an increased current density in the form of rings at the outer edges of the fil- aments) as well as a diametrical displacement of current density, caused by the tendency of Birkeland currents to twist about each other at late time (Paper II). This process is depicted in Fig. 6, that shows contours of the magnetic


where ‹», = 4. 17  X  10   3 rad/s is the highest cpicyclic       field Allergy at the midway cross section of the two fila-

frequency resolvable (T ——  0.478 Myr).                             ITlCRts.  Additionally, this figure gmphically  illustrates the Checking the parameter values, the velocity between    tendency towards further filamentation as the current tubes the hydrogenic plasma filaments at T = 90 is (10),   v =          are flattened into sheet beams. The cross-sectional regions

4.5d  x  10  3 cells per time step, or 4.56  x  10  ' a' ldt’     of dense synchrotron-emitting electron currents are called

1260 km/s, in satisfactory agreement with the assumed          ‘hot spots” in analogy to their double radio galaxy coun-

value v =  1000 km/s. To traverse a distance of  10A’ =     terparts (see Section VI-D).

55 kpc between filaments requires 2192 time steps. An


exorbitant amount of computer time would be required to study the complete bulk-force interaction, that involves the twisting of filaments, in a three-dimensional EM code. To economically resolve the slow-time Biot—Savart forces, a time scale compression is made by reducing the mass  of thé  ions  in,  =  16m,.  For this  reduction, 10A’

  1. Radiated Power and Isoyhotal Morphologies of Strong Sources

An estimate of the total power emitted as synchrotron radiation follows directly fmm the results of Section V. During the radiation “burst era” at T = 90, the total en- ergy radiated in the form of electric and magnetic  field




Fig. 6. Contours of magnetic energy B' about two adjacent filaments at simulation cross section. P = 9—20 in 1 DC steps. The contours at the

l00Atl0ft  Of  th0  tWD  fIlAlT16f1tS  C0fTeS{IOf1d  IO  eftefgy  InaxiiTia  WhilC the

centml ellipse is an energy minimum. “Hot spots” in azimuthal field energy feeding synchrotron mdialion ate beginning to become noticeable in the later time framea.



energy is iPqd = 2.1 AEU while the total simulation mag-

netostatic energy  is lip,  = 350 AEU. Since, at T —-   9D,

= 2.0  x   10"   T,  Bz -—   2.5   x   10"   T, and P           63

 is  the  plasma  volume;  fig,  =  (2  )  'B's =  2.5  x 10 3 J , or 1 AEU = 7.1 x 10" J. Thé peak radiation burst

lasts -20di’ in the compressed time frame (Fig. 5), cor-

responding to an actual time of 20(6 x 10" s)V(1836/16)

= 1.28 x 10"  s. The total power emitted in synchrotron

radiation  is L  =  2. 1  X  7.1  x  1050 Ju l.28  x  1ol4 $

1.16 x 10" W, which is to be compared with the radio luminosity of Cygnus A of 1.6--4.4 x 1037               (Table I).





Fig. 7. (a) Isophotal contours of synchrotron mdiation at 150 MHz from Cygnus A. (b) Magnetic energy isophoies of filament cross sections near synchrotron burst I = 86.


  1. Induction Fields

In accordance with Faraday’s Law of Induction

V  x £ = — d&/6i                              (16)

the converging magnetic fields produce an axially di- rected induction field E„ whose strength, based upon a number of frames (time steps) of the cross-sectional view of the simulation magnetic field, can be approximated by

E   --  -B b’ldt’                        (16)

or ) £,        6.9  V/m,  using  the  parameters  of  Section


The fully electromagnetic simulation allows the calcu- lation of the induction field from (15) for the more com- plex magnetic fields B = B(r, 0), and these are shown in Fig. 8. Fig. 8(a) shows the magnetic energy densities ' while Fig. 8(b) shows the corresponding electric induc-

densities fi'. The induction field grows

- 0.7 ac T ——   �4 to      2         1.3 at 7

  1. In dimensionless simulation units, iE is


The synchrotron power derived above is also available

from (12). At T = 90, the velocity distribution of the

E —-         E



Birkeland electrons is nearly Maxwellian with a mean en- ergy of 26.49 keV, or J3, = 0.228.  From (12),   2. 17 X 10 "B'n,Q' —— 2.06 x 10 " W/cm'. The total power or luminosity is then L - › P = 2.06 x  10" W,

where hg = c/‹» is the electromagnetic “skin-depth” pa- mmeter and IP, = 51I keV is the rest mass energy of the electron. In MKS units, the induction electric field cor- responding to iE = 1 is 3.96 Vfm, so that the peak fields


i.e., about a factor of 6 léss than 1.16 x

10"  W resulting

r =  104 and 300 are 3.31 V/m and 4.52 V/m,


from the filament intemction (cf. Fig. 5).

Fig. 7 compares the induced magnetic energy isophotes of the synchrotron-emitting simulation currents to the synchrotron isophotes measured for Cygnus A at a fre- quency of 150 MHz [72].


* Roral D•••!•Rment of Double Radio Sources

During the time that the elcctromotivc force driving the Birkeland currents exists, both axial currents and concom-


648                                                                                              IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. PS-14. NO. 6, DECEMBER 1986


                                 E2                                                                     ROTATIOML  SYNNE7RY  SEQUENCE






  1. 3C €7







Fig. 9. Rotational symmetry sequence of double radio galaxies (after Miley









Fig. 8. Eleciric- (induction) field and magnetic-field energies . ' ' and ‹ ’ ' associated with two interacting Birkeland filaments. T -- I W-3W in 50 DC steps. Run TO6.



itant azimuthal magnetic fields increase, producing the distribution of magnetic-field energy shown above (Fig. 8(a)). As previously seen in Fig. 6 (T -• 1—20), the longer time sequence in Fig. 8 (T --• 100—300) also shows the tendency toward a rotational displacement in source mor- phology, from the strong source at T = 90 (peak burst) to the weaker source at T = 300. A suggested sequence in mtational symmetry as a property of double radio sources has, in fact, been noted and is reproduced in Fig. 9.

The approximate age of a double radio galaxy can be determined from its isophotal contour profile at a given wavelength; the isophotes of strong classical double (early) radio galaxies are largely determined by the mag- netic energy  density  of the  interacting  Birkeland fila-







Fig. 10. (a) Isophotes of 3C434. 1. Fornax A. and 3C 192. (b) Magnetic

and electric field energy (composiiesl simulation analogs.


ments, while (later time) double radio galaxies project more complicated patterns resulting from the induction acceleration of plasma confined between the magnetic isobars. Fig. 10 illustrates this situation by comparing the synchrotron isophotes to the induction electric-field en- ergy patterns for the sources 3C434. 1, Fornax A (NGC 1315), and the weaker later time source 3C192 [76].





32                                                             0
















Fig. 12. Isometric and planar views of self-consistent magnetic fields at T

= 255 showing elliptical sump and one-sided isobaric channel between eaeigy hot spots. Peak field (squamd) = 1.5 units, 0. I units/contour. Simulation rim TO6.


11) is      ' = 0.2,  or 2i,  =  0.592  x   10"   T.  The field induced pressure defining the boundary of the sump at this

'(29o)   'B') -  1.4  X   10  ' ' Pa  (1.4  x   10  10






Fig. 11. Development of a double radio galaxy. laometric view of mag- netic energy contours. Time increases from top to bottom.

At later time, the converging magnetic-field lines con- tinue to compress intergalactic plasma into two narrow channels formed on either side of the elliptical sump. Fig. 12 shows the isobaric contours and isometric veiw at T ——




255. At this time, a channel exists on the

of Fig.  12(b), subtended  by  a 3.8  x 10


(0.98 x



ELECTnon   Benns  (JeTS)

  1. The C.expression of Intergalactic Plasma by

Converging Magnetic Mirrors

As the axial currents and concomitant azimuthal mag- netic fields about each filament increase, a magnetic iso- baric configuration, shown as a function of time in Fig. 11, is produced. The effect of the “colliding” magnetic fields is to pmduce an isobaric sump with elliptical cross section at the separatrix between parallel currents [77]. At T -- 9D, the field strength squared in the vicinity of the sump (approximately midway up on the last frame in Fig.

10 'T) isobar that also encloses the sump. The length of the channel is approximately 9A’ = 1.5 x ID' in (49.8 kpc) while the width varies from about 0. 5A’ to 2.0A’ (2.8—11 kpc).

The condensation of plasma from the cosmic plasma medium involves two mechanisms: the pinching of plasma within the current-conducting filaments, and the capture and compression of plasma between the filaments. The rate of condensation thus depends on whether the plasma is internal or external to a filament.

Within a filament the convection velocity of plasma ra- dially inward is o = £ x B/S', so that at time 7’ - 100,


650                                                                                              IEEE  TRANSACTIONS  ON  PLASMA  SCIENCE.  VOL.  PS-14,  NO.  6,  DECEMBER     1986



u„„   - E  IB    - 60 mV/in  + 2  x   10"   T  = 3  x  10‘ ml s (3000 km/s), i.e. , the convection velocity is three times faster than the Biot—Savart attraction between filaments (the inflow velocity is 30 000 km/s where B - 2 x 10" T). Conversely, from Fig. 11, the velocity of a magnetic isobar towards the sump is, approximately, 0.032 cells per time step, or v,                          - 0.032A'/di’ —-  8.99 x 10‘ mls (8990 km/s), i.e. , the approximate velocity of isobaric compression in the region of the sump is nine times faster than the Biot-Savart attraction. It is noteworthy that the incoming plasma closely resembles the closing of two giant cymbals (e.g., Fig. 8(a), frame 3) as is often the case for peculiar galaxies such as NGC 5128 and Cyg A, located b::tween giant radio lobes, that possess “dust lanes” at their midsection [78). The convection velocity decreases with time, having its maximum value i›,pqp at the onset of plasma convection. The compression velocity increases (as f increases with particle acceleration in a constant E, field) until pressure equilibrium is reached in the sump.

At T - 100, the elliptical sump (defined by the bound-

ary of the 1.4 x 10 ''-Pa isobar) extends some 50 kpc and can balance the thermokinetic pressure of a 104

6-keV plasma. At T —- 255, the spatial extent of this iso- bar has been reduced to -20 kpc. At this time the mag- netic-field gradient at the sump is bBlb’ —— 0.0554, or 4.4 x  10 "  G/m (I .32 pG/kpc), so that nearly all of the  in-

tergalactic plasma originally present in a volume P - 4/3 r(2. 5A 1 )3      x 1062 md has been compressed into

the elliptical sump. For the 2—2.5 x 10" T contours, the

pressure is 2.5 X 10 Pa, allowing the confinement of a 10 -    3  (10"-cm")  2-eV  plasma.   (The  concentrated

plasma at the center of the sump has riot yet been modeled because of demands on spatial resolution in a simulation. Lerner has studied this state in some detail and points out the similarity to pinch regions in the dense plasma focus [l5].)

The phenomena associated with the capture and compression of intergalactic plasma between neighboring Birkeland currents in synchrotron emission can best be seen from the observational evidence itself (Figs. 13 and

14) {79]-[81). Fig. 13(a) illustmtes double sources for which no centml “object” of any kind is present, while Fig. 13(b)—(c) illustrates sources in which large-red-shift QSO’s have formed at or near the geometric center be- tween filaments. Fig. 14 shows the isophotes of double sources for which the central object is identified as gal- actic in nature; usually having a large red shift, and being elliptical or peculiar in morphology. This variety of source represents a later stage in temporal evolution and is ac- companied by somewhat richer isophotal patterns because of the action of the inductive field on the confined plasma.

  1. Temporal and SR•tial Characteristics of the

Induction Acceleration  field

A temporal-spatial history of the electric and magnetic fields within the simulation region is possible from field probes and field-slice plots. Fig. 15 shows the temporal

history of the axial induction electric field E, measured at the ordinate position 16A’ and along the “major axis” between the geometric centers of the Birkeland currents at abscissa points 10.SA’, l2A', l4A’,  l6A’,  l8A',  20A’, and 21A', respectively (probes 1—7, Fig. 12(b)). Fig. 16 illustrates the spatial variation of the E, field component along the abscissa at the ordinate slice position 16A'. The probe and slice data show the following.

  1. Variability in Synchrotron Flux: The temporal probe data (Fig. 15) show  a time variation doldt'  - 75/di'

(14.6 jiV/m per year) with a periodicity -5dt’ (i.e. , an expanded time - 106 years). The variability is apprecia-

ble even over a 30-year period (Section VII-C), i.e., 438

mV/in,  or  - 101 6 V  over a parsec-dimensioned  accelera-

tion length.

  1. One-Sidedness in  Synchrotron   Radiation     (82),

)83): Fig. 16 illustrates both the gross channel morphol- ogy and slice plots of the induction fields at selected times of Fig. IS. As shown, an isobaric channel has started to form on the right-hand side (RHS) of the sump, or core, at time T - 193. The channel continues to form and nar- row until T - 243, when a change in the converging mag- netic-field line morphologies ceases to support a channel on the RHS, but rapidly forms a new channel of the left- hand side (LHS). The plasma electrons confined in these channels are responsible for the synchrotron radiation ob- served. However, synchrotron radiation from the channel plasma is possible only when the polarity of the induction field in the channel is correct. (For example, the electric field must be directed in the —z direction (inward from the plane of the figure) for an outward observation of syn- chrotron radiation from electrons accelerating in the +z direction.) One-sidedness (field reversal or a large field ratio on either side of the core) was measured by probes 5 and 6 (located within the right-hand channel ) for T - 190-290 (Fig. 15; the actual field spatial profiles at se- lected times are shown in Fig. 16). Thus at time T —— 255 (Fig. 12(b)), the field component in the left-hand channel is directed along +z, while the strong-field component in the mole well-defined right-hand channel is directed along

— z. Thus this source projects a channel or “jet-like” ra- diation pattern in the +z direction that is connected to a strong core (I° u E ). This is also the direction in which the synchrotmn radiation from the electron Birkeland cur- rents making up the outer radio lobes is beamed.

  1. Electric and Magnetic Field Configuration: During the synchtotmn burst era 25 :s T < 140 the plasma being collected within the core and channels is tenuous. The induction field is positive on both sides of the core but, at times, the left and right components can differ appreciably in magnitude by a ratio as high as 5 : l . This era also cor- responds to a weak core field and magnetic fields that are largely transverse to the major axis. A transition from transverse to parallel magnetic fields is observed at about T -- As shown in Fig. 16, the 5 channels become increasingly “collimated” for Z’ N 200 and are accom- panied by an increasing core field when 250 x T « 360 (Fig. 15,  probe 4).



















  1. (2




























Fig. 13. Fifteen QSO’s with and without an optimally identifiable object between mdio lobes. (a) 3C86B, 3C427. I, 3C86A, 3C69. (b) 3C47, 3C194, 3C204, 3C215, 3C249. 1, 3C263, 3C288.1. (c) 3C323. 1, 3C336,

4C28. 3C 18. All measurements taken at 5 GHz.


652                                                                                               IREE TRAi'JSACTIOHS ON PLAShlA SCiEiicn, voL. rs-i‹, NO. 6, DECEMBER 1986





























Fig. 14. Twenty-one QSO’S hBving a galactic object situated midway be- tween radio lobes. (a) 3C61, 3C277.3, 3C295, 3C388, 3C411, 3C434, 3Cl90. (b) 3C154, 3C184, 3C17l, 3C223, 3C234, 3C298. 2, 3C284. (c) 3C332, 3C381, 3C357, 3C430, 3C456, 3C79.










probe 2


—         probe 5






             200                 4tXI          550

Time Step

Fig. 15. Electric field Ez versus time at probe positions I-7 (Fig. 12). The ordinate is given in simulation units. Calibmiion: I unitfdiv (probe 5, 2 units/div). Zero field, first line, piobee I , 2, and 4; second line, probes 3, 5, 6, 7. Run  TD6.



The total intensity along a source observed at radio and microwave wavelengths must be determined by the square of the induction field within the source. Fig. 17 depicts the total intensity distribution of the radio galaxy 3C326 on a crosscut along its major axis at 49 cm. The intensity profile recorded corresponds appmximately to that mea- sured in the simulation at T -206 in Fig. 16 (cf. [84]). Knowledge of the field strengths and plasma density al- lows a quantitative analysis of the brightness of plasma within the channels, and can be directly compared to data [85], [86]. These comparisons are not made in this paper.

  1. Polarization: Plots of the angular distribution of ra- diation from mildly relativistic electrons, 1—200 keV, are given by Oster [87]. For frequencies above the  second

Fig. 16. Spatial slice plots of Ez [spncirai brightness = E z) along source major axis at selected time T -- 193, 198, 206, 210. 218, 227, 237, 243,

and  254.  Ordinate.  2  simulation  units/div.  Abscissa,  0-35  A’.  insets

itlustratc the magnetic isobaric ptofile at time  step.



electron cyclotron harmonic, the radiation pattern of each filament is largely polarized in the z—y plane (cf. Figs. 5 and 18).

  1. Superluminosity: The rapid spatial variation of the induction field, as well as the changing of the field polar- ity along the channel, causes an apparent “superluminos- ity” effect as the field sweeps the channel confined Superluminosities are often associated with radio sources (88] and apparent “superluminosities” as high as seven times the speed of light have been measured in the simulation. (The decrease in the induction field strength because of a induction in the rate of change of the mag- netic field along the channel does not require a high de- gree of alignment of the source with the observer. More- over, while 7c is the highest phase velocity seen in the simulations, there is no reason why phase velocities can- not be appreciably higher than this value.) The electro-


654                                                                                               IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. PS-14, NO. 6, DECEMBER  1986









I arcmkxite


1 000



6 00














Fig. l7. (a) Westerbork 21-cm map of the mdio galaxy 3C 326 on which are superimposed vectors indicating the orientation of the magnetic field projected perpendicular to the line of sight. (b) A crosscut along the major axis of 3C 326 showing the 49-cm spectml index distribution of the outer lobes (after Willis [84]).




magnetic wave effect for illumination as a source of “su- perluminosity” was first suggested by Kellerman [89]. Isophotal contour maps and major-axis profile data mea- sured from quasars can be directly compared to the sim- ulation data in Fig. 16 [90]. The ignition of previously confined plasma is also demonstrated by nonradial ap- pearances of the isobaric profiles taken on by the plasma [91]. These profiles are not seen in side-on observations of charged particle beams, but are observable when a sheet beam is viewed end-on.


  1. Simulating the Induction Accelerated Sheet Beam

For the purpose of resolving the plasma magnetically confined within the 5 x 50-kpc. channel, a simulation with a temporal/spatial resolution increased approximately two and one-half times is used. The channel length and width are 24A“ and SA“, respectively, where A“ = 6.25 x 10"

m. The time step is dr“  = 2  x  10"  s. To insure that  all




Fig. 18. (a) Electric energy and magnetic elliptical sump at T = 237 in

simulation rim TO6. The arrows denote the magnetic-field polarization


the  particles  simulated  are   ma

10  3 Pa thermokinetic  isobar

netically   confined, the

chosen  to  define the

vectors in this synchrotron source. (b) Isophotal contours of the source 2355 + 490. The arrows indicate the orientation of the magnetic field projected perpendicular to the line of sight.

































REB      (d)









B2  g       ie›



















channel dimensions. (Because of the steepness of the iso- bars, this dimension is practically the same as that sub- tended by the 10    isobar.) Based on this pressure, suit-

able choices for the electron density and temperature are 2  X  103    3 and 300 eV, respectively.  A spatially  aver-

aged induction field Ed - 600 mV/in is used. The dimen- sionless simulation units are ‹» di —— 0.25, hp/A = 0.1, and w,  /‹»   =  5.0, and the scaling factor is ‹x”  =  2. 1    x

1015  The number of electrons per cubic cell is  10.

Because of the £, • ‹ diocotron instability, sheet beams filament into current bundles whenever a threshold determined by the length of propagation or current carried by the beam is exceeded [26]. The quantity £, is the spa-

tially dependent electric field across the sheet width in a nonneutral beam (or in an overall neutral beam when the 11 x 4B force produces charge separation). This phenom- enon was first discussed by Alfv6n in connection with charge bundles and folds in the context of auroras [92]. The optical synchrotron radiation, isophotes, and vortex filaments of the jet (sheet beam) in M87 are depicted in Fig. 19(a)—(c). Fig. 19(d) shows the filamentation within the simulation sheet beam. Also shown are the magnetic- field contours %' and the B-field-polarization vectors. As shown in Fig. 19, the LHS of the beam is tethered to the denser core, but the RHS is free to fold (within the con- straints determined by the strength of the external mag- netic pressure). The configuration shown is typical of sources such as M87 and 3C273 [93]—[102]. The polar- ization of the self-consistent magnetic held is also  shown




Fig. 20. (a) The jet in 1919 + 479. (b) Laboratory sheet beam at early time (after Peratt [391)


in Fig. 19(e) and is in agreement with measured polariza-

tion data [103]—[107].

Fig. 20 illustrates the filamentation within the extra- galactic source 1919 + 479 at 6 cm, shown together.with its laboratory analog. The bends in sheet beams, either extiagalactic or laboratory, delineate where the dyi:amics of the beam change fmm that of a (linear) rigid mtor to a (nonlinear) vortexing beam. Typically, this change occurs


































































Fig. 21. Ten examples of very large arfay observatiotls of mdio quasars having jets [115]. The general morphology ie that of a fi1ament°d stnic- ture in emission that is connected to a core located midway between two sttong mdio emitting plasmas. The last three pictures are designated as “faint” jets.



when the beam has rotated some 10°-30° about the major axis, depending again on the “collimation”  pressure of

the external magnetic isobars. It is emphasized that the  The temi “jet” was first used in astrophysics by Curtis sheet beam is pmduced between the converging isobars in 1918 to describe the elongated optical feature pmtrud- and is inductively accelerated out of (or into) the plane of ing from the core of the elliptical galaxy M87  [1083,  the figure.                                                                                      [1093. Later Baade and Minkowski suggested that such a







Year of Observation

Fig. 22. Independent estimates of the magnitude of the jet of M87 in the B wavcband plotted according to date of observation (after Warren-Smith ct tit. [116]).




jet might actually correspond to matter ejection after some active phase of the core [110]. Jets have been mapped in about 200 radio sources. Jets are found not only in double radio galaxies and quasars, but also in central compact radio sources located in the nucleus of associated (optical) galaxies (Paper II) [111)-[113]. They are measured from the electromagnetic (synchrotron) radiation they emit, from centimeter wavelengths to X rays. However, in spite of improved resolution and statistics of observations, de- finitive direct evidence that ordered streaming motions are present in jets or radiosources is still missing. Moreover, uncertainty exists in identifying structures as either “jets” or  “bridges”  [114].

Fig. 21 shows VLA observations of radio quasars hav- ing jets (seven out of 26 quasars in a 966-MHz Jodrell Bank survey) and three quasars with faint jets [115]. The general morphology is that of a filamented stricture in emission (see Section V-B) that is connected to a core (see Sections VII-A and -B) located midway between two strong radio plasma sources. These data are consistent with Fig. 16, which indicates that the source geometry can consist of a core and channel (that may be filamented and bent) or a core with a side node. The topology mapped is dependent on the polarity of the induction field and the direction of observation.

Lastly, as indicated in Section VII-B1, the induction field across the channel is variable, causing the precon-



its filamented channel versus observed time [1163. Com- parison of the integrated magnitude of the jet with pre- vious independent measurements over the period 1934— 1980 suggests that the jet is variable and has been fading more or less uniformly by about 0.8 mag per decade be- tween 1964 and 1980. The data imply that over the period 1952-1980, the total jet intensity fell by at least 2.5 mag. Comparisons of isophotes taken in 1964 and 1979 show no obvious differences in overall shape, apart from effects of variation and noise. This indicates that the fading has affected the whole channel more or less uniformly since 1964, i.e., the knots (i.e., filaments or vortices) have not been seen to move. However, between 1934 and 195d, knots d and Zt became significantly brighter than C.

At a distance of 11.4 Mpc, the channel length of M87 is 30 arcsec in the plane of tlie sky, or 5400 light years across. For this reason, and because side-on photographs of charged-particle beams display difierent morphologies, the explanation of observable jet fading based on side- ejected matter models is untenable. However, the mor- phology, field configuration, polarization, and variability of M87 are in agreement with laboratory and simulation analogs of outwardly  propagating  sheet jets.





In discussing the Biot-Savart forces between adjacent circuit currents a problem arises: Ampere’s problem, i.e. , the necessity of accounting for the complete electrical cir- cuits in which the currents flow. This problem applies to cosmic as well as laboratory plasmas. Unlike the previous sections, where the specification of a single physical quantity was sufficient to scale all parameters in the sim- ulation study of double radio galaxies, the discussion of the entire circuit is far more speculative since this is not modeled. (Particle codes have, in fact, advanced to a level where an external circuit consisting of lumped circuit ele- ments can be included in the simulation.) Nevertheless, based upon the properties of plasma in the laboratory, a reasonable guess about the properties of the cosmic circuit is possible.

In its simplest form, an electrical circuit consists of a voltage source, a load, and the transmission line between the source and load. For the case studied here, the trans- mission line and the load are the field-aligned current that is likely to form (a series of) double layers. Like most field-aligned currents ir cosmical plasma, the currents flowing from the source to the double layers are likely to be pinched. This means that they consist of hlaments [16], [117). The magnetic field is of the order B -- y fl 2or, where r is the radius of the filament or bunch of filaments. In circuit theory the magnetic energy of the circuit is de- scribed as being due to an inductance L, such that


fined plasma to brighten or fade. Fig. 22 shows the mag- nitude of the optical synchrotron radiation from M87 and

B’  dV




658                                                                                              IEEE TRA?'ISACTIONS  ON PLASMA  SCIENCE,  VOL.  PS-14.  NO. 6, DECEMBER 1986



where Iis the current. The integral should be taken over the  whole  region where  Iproduces  a magnetic field.

Since the circuit conducts filamentary currents, the cir- cuit consists of a resistance, an inductance, and electro- static double layers formed along the filaments, all fed by an EMF. The circuit equation is

measurement, i.e., in plasmas having the dimensions of galaxies or systems of galaxies. The necessity for a three- dimensional electromagnetic approach derives from the fact that the evolution of magnetized plasmas involves complex geometries, intense self-fields, nonlinearities, and explicit time dependence. Moreover, synchrotron ra- diation and double layers are discrete particle phenomena


L dI + r


' sourre                            DL


and cannot be studied using magnetofluid models of plasma. The importance of applying electromagnetism and


where E <D is the sum of the potentials over the double layers, and ?t and L are the resistance and inductance of the circuit.  respectively.

If ?I is negligible  (as is often  the case in cosmic   situa-

tions)  the  current  grows  as  long  as <source  >          DL• If

DL ' source , the current is constant. For the simulated pair of Birkeland  filaments  (Section VI-D),  jLf 2  2.1

x  10"  J and I = 2. 15 x  10"  A; so that L ——   1.08 X 10"

  1. The rate of increase of the current is Shit —— 2.15 X lol9 A/90 di’ ——   3.98  x   5   A/s.  For  these  values, ml di ——  4.2 x  1o20 v.

In analogy to the rotating magnetized plasma dynamos that drive cosmic Birkeland currents of lesser dimension, a magnetized plasma with a size of the order of the largest objects  observed  in  the  universe,   10—50  Mpc, moving

through 10 9 T field lines at 1000 km/s, will produce an

EMF of the order of 10'° V. Using Cygnus-A-diinen- sioned current cross sections, 35 kpc, and the knowledge

that the width/length ratio of laboratory filaments is gen- erally in a range 10 3  10 5  a typical filament length as-

sociated with the dynamo source may be of the order of

350 Mpc.

If suddenly *source becomes zero, the current will con- tinue to flow but it will decrease with the time constant T

= Lf/A T =  10 4 s.



Because of in-situ measurements of plasmas in Earth’s ionosphere, cometary plasmas, and the planetary magnet- ospheres and recent discoveries of helical and filamentary plasma structures in the Galaxy and in extragalactic ob- jects, our understanding of cosmic plasma has changed considerably during the last decade. In particular, Birke- land (field-aligned) currents, double layers, and field- aligned electric fields are now known to play a far more important mle in the evolution of plasma in space, in- cluding the acceleration of charged particles to high ener- gies. Because the properties of the plasmas in space are found to differ little ftom those in the laboratory, the em- pirical knowledge gained from earth-bound experiments has suddenly found application in situations orders of magnitude greater in dimension. Kirchhoft's laws for cur- rents in circuits appear equally valid regardless of whether the plasma has its dimensions measured in centimeters, kilometers, parsers, kiloparsecs, or megaparsecs.

With the advent of three-dimensional electromagnetic particle-in-cell simulations, investigations of Birkeland currents, double layers, and field-aligned electric fields have become possible in plasmas not accessible to in-situ

plasma physics to the problem of radio galaxy, galaxy, and star formation derives from the fact that the universe is largely matter in its plasma state, i.e. , a plasma uni- verse. The motion of this plasma in local regions can lead to pinches and ultimately condense states of matter. Where double layers form in the pinches, strong electric fields can accelerate the charged particles to high energies. Sim- ulations of the interactions between plasma pinched into filaments show:

  • a burst of synchrotron radiation of luminosity - 1037 W lasting 107   10' years as the interaction begins;
  • isophotal topologies of double radio galaxies and quasars, including juxtaposed “hot spots” in the ra- dio lobes (cross sections of the interacting Birkeland currents);
  • the formation of “dust-lane” peculiar and elliptical galaxies at the geometric center of quasars and radio galaxies (due to plasma trapped and compressed within the elliptical magnetic separatrix);
  • a spatially varying power law along the major axis of the simulated double radio galaxies in agreement with observations;
  • alternating beams of betatron pumped synchrotron- emitting electrons on either side of the elliptical cen- These have the morphologies (i.e., “knots” or vortices) and polarization properties of jets but are aligned in an acceleration direction toward the ob- server; and
  • a “superluminosity” and fading of jets as the beta- tron induced acceleration field sweeps over and ig- nites previously confined

The simulation time frame of this investigation lasted some 10'—109 years. The lifetime and evolution of qua-

sars and double radio sources, the so-called end problem of double radio galaxies, is addressed in a sequel  paper

(Paper II) by continuing the simulation run - 1-5 X 109

years farther in time. Above all, a translation of knowl- edge between laboratory, space, and cosmic plasma is in- dicated from the results outlined above.



[1]  H. Alfv£n,  Cosmic Plasma.     Dordrecht,  Holland:  Reidel, 1981.

(2)  H. Alfv£n,  “Pamdigm tmnsition in cosmic plasma physics,”  Pliys-

tea Scripts, vol. T2/ 1, pp. l0—19, 1982.

[3J H. Alfv£n, “Paradigm transition in cosmic plasma physics,’ ’ Geo-

pliys. Ales.  Lrti., vol.  10, pp. 487--488, 1983.

[4] L.P. Block, “Double layers in the labomtory and above thC AU- miv,’ ’ in Physics of Auroral Arc Formation (Geophysic8l  M0f10- gmph Series, vol. 25). Washington, DC: Amer. Geophys. Ufiioil, 1981, pp. 218—225.






[5] P. Carlqvist, “On the formation of double layers in plasma,” Cosmic Electrodyn. , vol. 3, pp. 377-388, 1972.

[6] L.P. Block, “A double layer teview,’’ Astrophys.  Space Sci. , vol.

55, pp. 59—83, 1978.

[7] C.-G. Falthammar, “Non-resistive electric drops in cosmical plas- mas,” in Particle Acceleration Mechanisms in Astrophysics, J. Anns, C. McKee, and C. Max, Eds. New York: Amer. Inst. of Physics,  t 979, pp. 27-4t.

  1. Alfvén, “Electric currents in cosmic plasma,” Itev. Geophys. Space Phys. , vol. 15, pp. 271-284, 1977.

[9] H. Alfv6n, “Model of the plasma universe,” IEEE Trans.  Plasma

Sci. , vol. PS-14, no. 6, pp. 629-638, Dec. 1986.

 P.A. Sturrock, “A model of quasi-stellar radio sources,” Nature,

vol. 211, pp. 697—700, 1966.

 P.A. Sturrock and C. Barnes, “Activity in galaxies and quasars,”

Astrophys. J . , vol. 176, pp. 3 I-45, ’1972.

[t21 H. Alfvén, “Double radio sources and the new approach to cosmic plasma physics,” Astrophys. Space Sci. , vol. 54, pp. 279-292, 1978.

[13] A.L. Peratt and J.C. Green, “On the evolution of interacting mag- netized galactic plasmas,” dstropfiys.  Spoce dci. , vol. 91, pp. 19—

33, 1983.

(14) P.F. Browne, “Magnetic vortex tubes, jets and nonthermal sources,” Astron. Astrophys. , vol. 144, pp. 298-314, 1985.

 E.J. Lemer, ’‘Magnetic self-compression in laboratory plasmas, quasars and indie galaxies, Part I,’’ £sscrs etid Particle Ileums, vol. 4, part 2, pp. 193—222, 1986.

[t6] W . H . Bostick The Coercingy and Morphology of the Electron, Photon, and Other Elementary Onta , chap. 3, to be published.

W.H. Bosiick, “The electromagnetic rant action o£ the plasma fo- cus as a paradigm for the genemtion of cosmic mys and the gigantic jets in active galaxies,” in Proc. Tenth Int. Cosmic day Conf., OF Sessions (La Jolla, CA), vol. 3, 1985, p.  18.

[l7] S.-I. Akasofu, “The magneiospheric cumnts: An introduction to m,agnetospheric cumnts,” in Magnetospheric Currents, T.A. Po- temra, Ed. (Geophysical Monograph No. 28). Washington, DC: Amer. Geophys. Union, 1984, pp. 29-48.

 W.J. Hiekkila, “Magnetospheric topology of fields and currents,” in Magnetospheric Currents, T.A. Poiemra, Ed. (Geophysical Monograph No. 28).   Washington, DC: Amer. Geophys. Union,

l 984,  pp. 208-222.

[t9] H.  Alfvdn,  “Space  plasma—The  fourth  state of  matter,’’ Johns

Hopkins APL Tech. Digest, vol. 6, pp. llXi- 102, 1985.

(20] A. Dessler, “Evolution of arguments regarding existence of field aligned cumnts,” in Af‹igneiosplieric Currents, T.A. Potemra, Ed. (Geophysical  Monogmph  No. 28).     Amer.  Geophys.  Union, !984.

 T.W. Hill, “Rotationally-induced Birkeland cumnt systems.” in Magneiospheric Currents, T.A. Potemra, Ed. (Geophysical Mon- ogmph No. 28). Washington, DC: Amer. Geophys. Union. 1984, pp. 340—349.

 C.-G.  Falthammar,  “Magnetosphere-ionospherc  coupling,”  in

Proc. ESA Workshop on Future Missions in Solar, Heliospheric, find 5p‹ice Plasmo Physics (Garmisch-Partenkiichen, Germany). Apr. 30-May 3, 1985. pp. 107- 133.

[23J V. Nardi, W.H. Bostick, J. Feugeas, and W. Prior, “Internal stnic- tute of electron beam filaments,” Phys. Ifev. d, vol. 27, pp. 22 I 1- 2217, 1980.

  1. Nardi, in Proc. Second Int. Conf. on Encrgy Storage. Compres-

sion, nn# Switching (Hanicn, Italy, 1978). New York: Plenum, 1982.

[24] B. Lchnert, “An instability of laminas flow of mercury caused by an external magnetic field,” f'roc. ?toy. Soc. d, vol. 233, pp. 299-


[25] A.L. Pcatt and P. Kocrt, “Pulsed electromagnetic acceleration of exploded wire plasmas,” /. Jpp/. Phys., •o. 54, pp. 6292-630t , 1983.



in Cygnus A,” Astroph ys.  J . , vol. 285, pp.  L35-L38,   l98'4.

(311 C.-G. Falthammar, “Generation mechanisms for magnetic-field- aligned electric fields in lhe magnetosphere,”  J.  Geomegn.  Gen-

electr. , vol. 30, pp. 419-434, 1978.

[323 C.-G. Falthammar,  ’‘Magnetic-field-aligned  electric fields,” ESA

  1. , vol. 7, pp. 385-401, 1983.

[33] C. Chan and N. Heishkovitz, “Tmnsition from single to multiple double layers,” Phys. Fluids, vol. 25, pp. 2135-2137, 1982.

[34] A.L. Pemtt and M.E. Jones, “Particle-in-cell simulations of heavy ion plasma double layers,” presented at IEEE Int. Conn Plasma Sci. , Saskatoon, Canada, 1986.

[35] H. Alfvén, “Double layers and circuits in asirophysics,” IEEE Trans. Plasma Sci. , vol. PS- 14, no. 6, Dec. 1986.

  1. Buneman, C.W. Barnes. J.C. Green, and D.E. Nielsen, ‘‘Prin- ciples and capabilities of 3-D, E-M particle simulations,’’ J. Comp. Phys. , vol. 38. pp. I—44, 1980.

[37] D. Nielsen, J. Green, and O. Buneman, “Dynamic evolution of a z-pinch,’’ Phvs. Itev. Lrtf. , vol. 42. pp. 1274- 1277, 1979.

(38] A.L. Perati, J. Green, and D. Nielsen, “Evolution of colliding plasmas,” I•Iiys. ?tev. £rtt. , vol. 44, pp. 1767-1770, 1980.

[393 A.L. Perati, C.M. Snell, and H . F. Webster, “Spiral structure in magnetically confined electron sheet beams,” in Conf. Record IEEE fnt. Cont. Pfiismn Sci. (Pittsburgh, PA), IEEE Cat. 85CH2199-8, 1985, pp. 46—47.

[40] W. Peter, “Time-dependent interaction between space plasmas and transverse magnetic fields.” US:£ Trans. Plasma $ei. , voi . PS- 14, no. 6, Dec. 1986.

[4t] W.H. Bennett, “Magnetically self-focussing streams,” Phys. Rev. ,

'vol. 45, pp. 890-897, 1934.

  1. Alfv6n and N. Herlofson, “Cosmic radiation and radio stars,”

Phys. Rev., vol. 78, p. 616,  1950.

[43] A.T. Moffet, “Strong nonthermal radio emission from galaxies,” ’in Galaxies and the Universe, Sandage, Sandage. and Kristian. Eds. 1975. ch. 7, pp. 2 I I -281.

 K.C. Westfold, ‘’The polarization of synchrotmn radiation,” Its- trophys. J., vol. 150, pp. 241-258, 1959.

 Y.L.  Ginzbur$  and  S.  1.  Syrovatskii.  '‘Cosmic  magnetobrems- strahlung (synchrotron radiation),’’ Aiimtzif Itet'iew s nf Astronomy end ztstropfiysics, L. Goldberg. A.J. Deulsch. and D. Layzer. Eds. Palo Alto, CA: Annual Reviews, 1965, pp. 297-350.

[46] V.L. Ginzburg and S. 1. Syiovatskii. 27ie Origin ‹i/ Cosmic ftnys.

New York: Gordon and Breach, 1969.

[47] R.C. Bless, ”The theory of synchrotron radiation,” in Nebulae and Interstellar fatter, B.M. Middlehurst and L.H. A fler, Eds. Chi- cago, IL: University of Chicago Press, 1968, ch. 12.

  1. van Stcenbergen, “Synchrotron radiation sources,” 1£££ Trans. Null. Sci., vnI . NS-26, pp. 3785-3790, 1979.

[49] W.K.H. Panofsky and M. Phillips. Cf‹irsicof £fectricit' end Mag- netism.  Reading, MA: Addison-Wesley. 1962, ch. 20.

[50] G.R. Burbidge, “On synchrotron radiation from Messier 87.’’ is-

trophys. J . , •oI . 124, pp. 416-429, 1956.

[5 I] T.N. Lie and R.C. Elton, "X radiation from optical and inner- shell transition in a highly ionized dense plasma.” Phvs. Rev. A, vol. 3, pp. 865-871. 1971.

[52] T.I. Welch and E.J. Clothiaux , ‘‘X-ray structure of a pinched plasma in a vacuum spark,” J. Appl. Phys., ver. 45, pp. 3825- 3827, 1974.

  1. Fukai and E.J. Clolhiaux, “Mechanism for the hard-X-ray emis- sion in vacuum spark discharges,” Ph)7s. he'd. Letl. , vol. 34, pp. 863-866. 1975.
  2. 1. Baum, A. Brascnahl, and R.S. White, “X-rey and electron specim from the double inverse pinch device,” Phys. Fluids. vol. 16. pp. 226-230, 1973.

 B.E. Meierovich, “Electromagnetic collapse, problems of stabil- ity, emission of radiscion and evolution of a dense piach," Phys. Jtrportx. vol. 104, pp. 259-347, t 984.


[26] q. L. Pemtt and C. bI. Snell, “Ivlictowavc generation  from fila-       [561 G. Bckcfi. Radiation Processes in Plasmas.     New York: Wiley,


mentation and vortex formation  within  magnetically  confined elec- t ops begms,” Pfiyi. ftcv. Lrtt. , vol. 54, pp. 1167-1170, 1985.

[27] H.F. Webster and T.J. Hallinan, “Instabilities in charge sheets and cumnt sheets and their possible occumnce in the aurora,” III- die Sci. , vol. 8, pp. 475-482. 1973.

[28] D.A. Mendis, “On the hydromagnetic model of comets,” 77ie Afooti

and the Planets, vol. 19, pp. 361 -369, 1978.

1966, ch. 6.

 B.S. Newberger, ”Synchrotron radiation from self-pinched beams,” Los Alamos Nat. Lab. , Los Alamos, NM, Memorandum X-8(8- fl3)5, 1983.

[58]  B.S. Newberger,  M.I. Buchwald,  R.R. Karl,  D.C. Moir, and  T.

  1. Starke, "Synchrotron  radiation  from Bennett beams,"  6u//. Jmcr.

Phys. Soc. , vol. 29, p.  1435, 1984.


[29]  F.  Yusef-Zudeh,  M. Morris,  and D. Chance,  “Large,  highly or-           (59] G. Reber, ‘‘Cosmic static,’’ Proc. IRE, vol. 30, pp. 367-378, 1940.

gani¿ed radjo strictures near the galactic cenlre,” Nature, vol. 310,              {60]  G. Reber, “Cosmic static,” Astrophys. J. , vol. 100, pp.  279-287,

  1. 557-561, 1984. 1944.

[30] R.A. Perley, J.W. Dreher, and I. J. Cowan, “The jet and filaments             [61) G. Reber and 3. C. Gteenstein, “Radio frequency investigations of


660                                                                                               IEEE  TRANS.ACTIONS   ON  PLASh4A  SCIENCE.  VOL.  PS-14,  NO.  6.  DECEMBER  1986



astronomical interest,’’ Observatory, Feb.  1947.

[62] G. Reber, “Cosmic static,” Proc. IRE, vol. 36, pp. l2l5—1218, 1948.

(633 1. S. Shklovsky,  “Extragalactic  radio waves,’ ’ in  Cosmic Radio

iF‹ives.   Cambridge, MA: Harvard Univ. Press, 1960, ch. VI.

 A.G. Pachoczyk, Radio Galaxies.   New York: Pergamon,  1977.

[65] G. Miley, “The structure of extended extragalactic radio sources,” Annu. Kiev. Astron. Astrophys. , vol. 18, pp. 165-215, 1980.

[66] G.C. Perola, “Radio galaxies: Observations and theories of their extended components,’ ’ Fundamentals Cosmic Phys. , vol. 7, pp. 59- 130, 1981.

(67] W.H. Bostick, ''Possible hydromagnetic simulations of cosmical phenomena in the laboratory,’’ Itev. fled. Phys. , vol. 30, pp. 1090- 1094, 1958.

[683 B. Lehneri, “Plasma physics on cosmical and labomtoiy scale,”

?Vuovo Cimcnfo, vol.  13 (supplement), pp. 59— 107, 1959.

[69] E.A. Parker, “Precise measurements of the flux densities of mdio sources Cas A and Cyg A at metre wavelengths,” hon. Notts. Hey. Astron. Soc. , •oI . 138, pp. 404—422, 1968.

[70] S. Mitton and M. Ryle, “High resolution observation of Cygnus A at 2. 7GHz and 5GHz,” 3fozi. Netic. Roy. Astron. Soc., vol. 146,  pp.  15—27, 1969.

[71] P.J. Hargrave and M. Ryle, “Observations of Cygnus A w'ith the 5-km radio telescope,” Eton. Netic. Roy. Astron. Soc. , vol. 166, pp. 305-327, 1974.

[72] A.J. Winter ci ‹if. , “The structure of Cygnus A at 150 MHz,” Eton.

Netic. Roy. Astron. Soc. , vol. 192, pp. 931—944, 1980.

[733 A. Toomre, “Mergers and some consequences,’’ in fire £vofuiiozi of Galaxies and Stellar Populations’, B.M. Tinsley and R.B. Lar- son, Eds. New Haven, CT: Yale Univ. Observatory, 1977, pp.  401 -427.

[74] G.R. Burbidge,  “Possible sources of mdio emission in clusters   OF

galaxies,” Astrophys.  J. , vol.  128, pp.  1 —8, 1958.

[75] V.L. Afanasiev, 1. D. Karachentsev, V.P. Arkhipova, V.A. Postal, and V.G. Metlov, “Radial velocities of some interacting galax- ies,” Herron. Astrophys. , voi . 91, pp. 302-304, 1980.

[76] M.J. Cameron,  “The structure of bright galaxies at 408   MHz,’’

Mon. Notts. Roy. Astron. Soc. , vol. 152, pp. 439-460, 1971.

[77] A.L. Peratt and J.C. Green, “Magnetic sump formation and plasma confinement between z-pinched plasmas,” in Proc. 1981 IEEE Int. Coii/. f’fnsma Sci. (Saute Fe, NM), IEEE Conf. Record 81CH 1640- 2,  198 I, p. b4.

[78] R.J. Dufour ei nf. , “Picture processing analysis of the optical struc- ture of NGC5128 (Centaurus A),” dnrz›zi. J . , vol. 84, pp. 284-301, 1979.

[79] G.G. Polley and S.N. Henbest, “Observations of 48 extragalactic radio sources with the Cambridge 5-km telescope at 5 Ghz,” Eton. Notts. ftoy. Astr. Soc. , vol. 169, pp. 477-526, 1974.

[80] J.M. Riley and G. £i. Polley, “Observations of 31 extmgalactic sources with the Cambridge 5-km telescope at 5 Gliz,” 3femo. A‹›y. Osiron. Soc. , vol. 80, pp. 105—137, 1975.

[81] R.T. Schilizzi and W.B. McAdam, “Observations of extended sources at 408 MHz-I. The indie structures,” 3femo. ltoy. Astron. Soc., vol. 79, pp.  1 -73, 1975.

[89]  K.I. Kellerman,  “Radio  galaxies  and  quasars,”  Mini.  Ne'w  York

Acrid.  Sci.,  vol.  336, pp.  I—II, 1980.

[903  S.C. Unwin  ct ml.,  ‘‘Superluminal  motion  in the quasar 3C345,’’

Astroph ys.  J. ,  vol. 271,  pp.  536-550, 1985.

[91] R.L. Moore, J.A. Biretta, A.C.S. Readhead, and L. Baath, “Su- perluminal accelemtion of the new component in 3C345,” in PL8f and CO<R**• Radio Sources, R. Fanti, K. Kellermann, and G. Setli, Eds.   Dordrecht, Holland i Reidel, 1984, pp. 109— 110.

[92]  H. Alfvtn,  Cosmic Electrodynamics.     New  York:  Oxford,  1950, p.


[93] B.D. Turland, “Observations of M87 at 5 GHz with the 5-km tel- escope,” Eton. fofic. Floy. Astr. Soc., voi . 170, pp. 281—294, 1975.

[94] J.R. Forster, J.  Dteher,  M.C.H. Wright,  and W.J. Welch,  “Virgo  A at  1.3 centimeters,” dstropfiys.  J . . •oI .  221, pp.  L3-L6,  1978.

(953 G.D. Schmidt, B.M. Peterson, and E.A. Beaver, “Imaging po- larimetty of the jets of M87 and 3C273,” Asirophys. J . , vol.  220, pp. L31-L36, 1978.

[96] H. Andcrnach, I.R. Baker, A. von Kap-herr, and R. Wielebinski, “The mdio continuum of M87,” Astron. Astrophys. , vol. 74, pp. 93-99, 1979.

(97] F.N. Owen, P.E. Hardee, and R.C. Bignell, “VLA observiiions of chc M87 jet at 6 and 2 cencimctcrs," Astrophys. 1., xo. 239, pp. L11—US, 1980.

[98] J.-L.  Nieto  and  G.  Lelitvre,  “High-resolution  observations   of

M87,” Astron. Astrophys. , vol. i 09, pp. 95-100, 1982.

  1. Lelitvre, J.-L. Nieto, D. Horville, L. Renard, and B. Scrvan, “Optical structures of the M87 and 3C273 jets,’’ dsrr‹iii. Astro- phys. , voi . 138, pp. 49-56, 1984.
  2. Flatters and R.G. Conway, “The indie jet of the quasar 3C273,”

Nature, col. 314, pp. 424--426,   1955.

 R.G. Conway and D. Stannard, “Radio brightness distribution of 3C273,” Nature, vol. 255, pp. 310—312,  1975.

 R.A. Perley, A.G. Willis. and J.S. Scott, “The structure of  the mdio jets in 3C449,” Nature,  vol.  281, pp.  437-442,  1979.

[103] W.  Baade,  “Polarization  in the jet  of Messier  87.’’ Astrophys.  J . ,

vol.  123, pp. 550-551,  1956.

 G.K. Miley and H. van der Laan, “The polarization distribution of eleven 3C mdio galaxies at 1415 MHz,” Astron.  Astnophys. ,  vol. 28, pp. 359-372, 1973.

 J.W.  Dreher,  “Polarization  maps  of  Cygnus  A  at  23 gigahertz,”

Astroph ys. J . , vol. 230, pp. 687—696, 1979.

 S.F. Burch,  “Multifrequency  studies of seven 3CR mdio  sources-

  1. Results and interpretations,” Afpn. Notts. Roy. ystron. Soc. , vol. 186, pp. 519-553, 1979.

[107] R.G. Slrom and A.G.  Willis,  “Multifrequency  observations  of very large mdio galaxies, II. 3C236,” ztsiron. Astrophys. , vol. 85, pp. 36-54,  1980.

[1083  A.  Ferrari,  “Jets in galaxies,”  ESA J . ,  vol.  SP-207, pp. 259—273,


 A.H. Bridle and R.A. Perley, “Extmgaliictic indie jets,” dztmi.

Rev.  Astron.  Astrophys. ,  vol.  22, pp.  319-358, 1984.

  1. Baade and R. Minkowski, “Identification of the mdio sources  in Cassiopeia, Cygnus A, and Puppis A,’’ Astrophys. J . , vol. 119,  pp.  206—231, 1954.


[82] A.H. Bridle, “Sidedness, field configuration, and collimation of          11113  R.I. Potash  and J. F.  C. Whardle,  “4C 32.69:  A  quasar with  a

extrtgalactic radio jets,” Astron. J . , vol. 89, pp. 979-986, 1984.                            mdio jet,” Astrophys. J. , vol. 239, pp. 42--49, 1980.

[83] L. Rudnick and B.K. Edgar, “Alternating-side ejection  in extra-          (I 12]  R. Linfield,  “VLBI observations  of jets  in double mdio  galaxies,”

galactic radio sources,’’ Astrophys. J . , vol. 279, pp. 74-85, 1984.                          Astrophys. J . , vol. 244, pp. 436—446, 1981.

 A.G. Willis,  “The large  scale  stmciuie  of extra-galactic mdio           11131  B.H. Wills,  “A Seyfert galaxy joins the jet set,” Stature, vol. 313,


sources ( >arcsecond),”  Physics  Scripts,  vol.  17, pp. 243-255,


[85] J.M. Riley and N. J; B.A. Branson, “New observations of 3C382, 3C452, and 3C465 tit 2.7 and 5 GHz,” Unit. Notts. Itoy. Astron. Soc. , vol. 164, pp. 271-287, 1973.

{86] J.A. Hiigbom, “A study of the mdio galaxies 3C111, 192, 219,

  1. 741, 1985.

[1t4) J. 0. Bums and W.A. Christiansen, “Radio jets and bridges in luminous double sources: 3C388 and 0816 + 326,” Univ. of New Mexico, Albuquerque,  Rep., 1980.

 F.N. Owen and J.J. Puschell, “VLA observations of Jodtell Bank

mdio quasars,” Astron. J . , vol. 89, pp. 932-957, 1984.


223, 315,  and 452,’’ dsiron.  dszropliys.  Suppl. , vol.  36, pp.  173-      [I 163  R.F. Wamn-Smith, D.J. King, and S.M. Scarrott, “Polarimctiy


192, 1978.

[87] L. Oster, “Effects of collisions on the cyclotron mdiation from re-

lativistic particles,” Phys. Rev.. vol. 119, pp.  1444-1454, 1960.

[883 1. W.A. Browne, “MERLIN observations of superluminal indie sources,” Nature, vol. 299, pp. 788-793, 1982.

and photomctiy of M87: Is the jet fading?," Lion. Notts. Roy. As- tron. Soc., vol. 210, pp. 415--424, 1984.

 B.E. MClCfOVlch, ''Macmscopic stability of charged part.icle beams,” I i££ Trans. Plasma Sci. , vol. PS-l4, no. 14, pp. 683- 689, Dec. 1986.




IEEE  TRANSACTIONS   ON  PLASMA   SCIENCE,   VOL.   PS-14,   NO.   6,   DECEMBER   1986                                                                                                  763

Evolution of the Plasma Universe: II. The Fomation of Systems of Galaxies






Abstr‹zcf—The model of the plasma universe, inspired by totally un- expected phenomena observed with the advent and application of fully three-dimensional electromagnetic particle-in-cell simulations to fila- mentary plasmas, consists of studying the interaction between field- aligned current-conducting, galactic-dimensioned plasma sheets or fil- aments (Birheland currents). In a precedi•B papec, the evolution of the interaction spanned some 10’—10’ years, where simulational ana- logs of synchrotron-emitting double radio galaxies and quasars were discovered. This paper reports the evolution through the next 10’-5 x 10’ years. In particular, reconfiguration and compression of tenuous cosmic plasma due to the self-consistent magnetic fields from currents conducted through the filaments leads to the formation of elliptical, peculiar,  and  barred  and  normal  spiral  galaxies.   The   importance   of' the electromagnetic pinch in producing condense states and initiating gravitational collapse of  dusty  galactic  plasma to  stellisimals,  then  stars, is discussed. Simulation diita are directly compared to galaxy mor- phology types, synchrotron flux, Hi distributions, and fine detail struc-

ture in rotiitional velocity curves. These comparisons suggest that knowledge obtained from laboratory, simulation, and magnetospheric plasmas offers not ° 7 to enhance our understanding of the universe, but also to provide feedback informiition to laboratory plasma exper- iments from the unprecedented source of plasms data provided by the plasma universe.










and the formation of stars within the dusty galactic plas- mas [2]—[4] .

Although the gravitational force is weaker than the electromagnetic force by 39 orders of magnitude, gravi- tation is one of the dominant forces in astrophysics when electromagnetic forces neutralize each other, as is the case when large bodies form [5]. Indicative of the analogy of forces for the motion of electrons and ions in the electro— magnetic field and the motion of large bodies in the grav- itational field is the ease with which a plasma model may be changed to a gravitational model. This transformation requires only a change of sign in the (electrostatic) poten- tial calculation such that like particles attract instead of repel, followed by setting the charge-to-mass ratio equal to the square root of the gravitational constant (a gravi- tational model cannot be simply changed to an electro- magnetic model as the full set of Maxwell’s equations are required in the latter).

The basic model of the plasma universe, inspired by the observation that most cosmic plasmas are filamentary in structure [6], consists of study '•8 the interaction between



of cosmic  plasma  from  a filamentary

current-conducting         galactic-dimensioned        plasma       fila-


to the development of double radio sources and quasars was investigated in the first part to this sequel paper  (Paper I)  [l]. The time  frame of  this  study, based

upon scaling simulation parameters to galactic dimen- sions, spanned  some  108  10’ years.  In this paper (Paper

II), the evolution for the next 1—5 x 10’ years under the influence of electromagnetic forces acting on the plasma  is investigated.

The importance of electromagnetic forces in galactic and stellar evolution derives from the fact that the uni- verse is largely matter in its plasma state. The observed stars are composed of plasmas, as are the interstellar and interplanetaiy media and the outer atmospheres of planets. The neutral Hi regions in galaxies are also plasma al- though the degree of ionization is probably only 10". Both the intra- and intergalactic media then consist of plasma, leading to the coinage of the term “plasma uni- verse.’ Electromagnetic forces can then be expected to play a crucial role in the development of the plasma uni- verse including both the formation of systems of galaxies


Manuscript received June 6, 1986; revised 3uly l7, 1986.

The author is with the Los Alamos National Laboratory, Los Alamos, NM 87545.

IEEE Log Number 8610914.

ments aligned along magnetic  field lines (Birkeland   cur-

rents, [ l]). Although the entire filamental circuit is ex- pected to be hundreds of megaparsecs in length and is probably no less complex than the magnetospheric current distribution near Earth, only a fraction of the circuit is simulated. The fraction of the length corresponds to a lo- cal region that is capable of interacting with an adjacent local region in a neighboring filament. Strictly speakin 8 the neighboring regions in adjacent fi laments are double layers since the model consists of a parallel electric field  in each of the pinched plasma filaments [7]. Because of the axial E field, electrons within the pinch are acceler- ated along the filament producing strong currents and also radiating away energy in the form of sy nchrotron radia- tion at radio, optical, and X-ray   wavelengths.

It is the purpose of this paper to continue the investi- gation of the dynamics of the denser interacting plasmas pinched within the filaments by means of the electromag-

netic and 8 vitational force laws. That this is possible is due largely to the advent of the particle simulation of  dy-

namic systems in three dimensions on large computers, allowing the computation of up to many millions of charge and mass particles according to their respective force laws. This approach to the study of cosmic plasma is labeled “gravito-electrodynamic”    [8).


0093-3813/86/ 1200-0763$0 I .00 fi 1986 IEEE





                                                                                     IEEE   TRANSACTION  S   ON   PLASMA   SCIENCE,   POL.   PS-14,    NO.    6,   DECEMBER     1986



PROPERTiES O             CHhOTRONq-EM    TINc GALACTiC Rsnio Souxc  s

     Gcomeev and Dimension                                                                                                                                 Red-Shift. z          two extended radio lobrs separated                                                                                                                                0.01- 1.8

icns of Kgc to icns of Mgc; oftenfimcs          0.3-3.8

c*tcndcd fialoes








                          clewntaiy  spirals ct a few




two radio componcms: -l0Kpc coincident with spital disl:, Up:












(Andronicda is blue-





Log P( 1.4 GHz) - W Hz-'


Fig. I .   The radio luminosity  turret iti• " 8• •xiex u nd quasars at lhe   pres-

en1 ctixnitilt›;;ical eptich (adapted from Fant i and Perola } 10]).


The gross radio properties of galaxies are reviewed in Section II. Section III describes a transistion through the following sequence of cosmic objects: double radio gal- axy to radioquasar; radioquasar to radioquiet quasi—stel- far objects (QSO’s) (9); radioquiet QSO’s to peculiar and Seyfert spiral galaxies; and peculiar and Seyfert galaxies to nomial and barred galaxies. The various classifications of elliptical and spiral galaxies are discussed  in Sections IV and V, respectively. The importance of electromag— netic effects in describing both the bulk— and fine-detail structure in the velocity curves of spiral galaxies is also reported in Section V. Multiple interacting galaxies are studied  in Section VI. The chemical composition  and  the

distribution of neutral hydrogen in 8•laxies is discussed in Section VII. Section VIII covers the Alfvén—Carlqvisl model for star formation in pinched plasma hlaments while  Section  IX  reports  the  extension  of three-dimen-

sional electromagnetic particle simulation techniques to include gravitational forces with the formation of stars.


an order of magnitude greater than the power of our Gal- axy) the power comes principally  from spiral galaxies.

In the region 10" —10" W/Hz, an overlap of spiral and elliptical galaxies occurs. The “ellipticals” are in fact a hybrid class, containing bona fide ellipticals along with N galaxies (a bright nucleus surrounded by a faint   nebulos-

ity) to that of cD galaxies  (giant ellipticals with very   ex-

tended radio  ‘halos”). Noteworthy  in Fig. 1 is a “break’ in synchrotron power from ellipticals at 1024 5    /Hz  and

another at 10'' ' W/Hz for spirals.

The size of the radio-emitting regions in galaxies spans a  very  wide  range.  At  powers  larger  than  about  1  23

W/Hz at I .4 GHz the radio emission is generally domi- nated by an extended component, whose size goes from tens of kiloparsecs  (e.g. , Cyg A) to tens of  megaparsecs

(e.g. , 3C236). Often a very compact central radio com- ponent is present, whose power ranges from 10 2  up to 1025  W/Hz,  and  which  may  be  seen  to  vary  with  time.

Extended and central radio components are typically found also in quasars.

At  powers  less than about  1023     /Hz the  size of  the

radio region in elliptical galaxies is generally measured in kiloparsecs and often reduces to a compact central com- ponent.  In spiral  galaxies,  the  next  stage of a suggested


The radio power L of galaxies, integrated  from 10 MHz  epochological  sequence  in  Fig.  1, the situation  is differ- to 100 GHz, ranges from about 10" to about 10” W, and ent. Apart from  the  radical  change  in  morphology  be- relative to their optical luminosity from less than 10 ' to tween elliptical  and  spiral, the spiral galaxies  have  not  about  l  [10], [l l ]. The distribution  in power is described   only  a compact  nuclear  component  (of  radio  dimension by means of the radio luminosity function (RLF), which between 0. l and l kpc) bul also a component of size - 10 represents the number of radio-emitting  galaxies per unit     kpc coincident with the spiral    disk.

volume  as a  function  of  the  monochromatic  power at a        Overlapping the powerful  radio ellipticals  having Sey- certain  frequency.  Fig.  l  illustrates  the RLF at  1.4 GHz                fert-like  nuclear  spectra  and  the  spiral  galaxies are the at the present  cosmological  epoch,  the “local” RLF.             Seyfert  spirals  themselves,  comprising  1 percent  of all The  RLF  suggests  a  continuity  in the morphological  spiral  types.  In contrast  to the ellipticals,  spiral galaxies

types  of  radio-emitting  galaxies.  Above  10 6 W/Hz,  the    rarely  have compact  nuclear  sources and are rarely asso-

main contribution comes from quasars and classical dou- ciated with extended radio lobes. Table I delineates the ble radio galaxies. In the region l0"- 10" W/Hz, the el- properties of galactic radio sources with decreasing red- liptical galaxies dominate, while below 10" W /Hz (about shift.





PERATT:   EV OLUT ioN   or   Ton  PLASMA   U Nl V ERSE. II














10"    —






10"   —



















sI   *










° °e

e     ee





! e                  10"

namics  of  the  plasma  within  the  bulk  forms).  These   are

shown in Figs. 3—7.

Figs. 4, 5, and 6 pertain to the cross-sectional views of two plasma filaments (the “extended  components’ ’)  of width - 35 kpc and separation 80 kpc.  (The axial ex-  tent is determined either by lhe length of the “micro- pinch” within the filament (in comparison to the analogy of laboratory filaments) or to the width of the double layer formed in the Birkeland current; these are typically com- parable to the filamental width). Fig. 5 (w,./ui, — 3) has  an axial magnetic  field strength  twice that  of  Fig. 4 (w,./  wq = 1.5). Fig. 6 pertains to w,.let, = 3.0, and an initial thermal plasma of 32 KeV. Fig. 7 compares selected  sim-

ulation  frames of a single simulation  run to three  galaxy


t0"  —                               Sg






10"  —                           s         S


0.01          0. 1            1              10           ltD         J (Q)



Fig. 2. Plot of the niono‹:hroniat ie radio power at 1 . 4 GHz versus linear diameter for classes of extragalactic radio sources. The sy mbol e denotes extended radio s‹›urees assticiated with el liptieal galaxies while q and s denote quasars and  spiral  galax ies.  respectively  (adapted  from  Eckers




Fig. 2 shows the positions of the radio sources, both extended and compact on  a  linear-size  radio-luminosity plot [12]. As seen, the bulk of the classical double radio galaxies possessing an elliptical galaxy have a spatial  ex- tent  between  a few  tens of  k iloparsecs  to many  hundreds

of kiloparsecs, with radio luminosities of L - l0”—10"

W . Some transitional  radio galaxies,  8—80 kpc, L  - 1034

W, are also present. The radioquasars appear in two  dis-  tinct populations; extended sources with dimensions of several  kiloparsecs  to  several  hundreds  of  kiloparsecs  (L

- 10"—10"  W), and compact  sources  - 2—8 pc  (L -

1  37   10"  W).  Most of the spiral galaxies are found  to  be

clustered according to a size-luminosity of - 10—80 kpc,

- 10' - l0'2.5 w .

Finally, unlike the other properties, the radio spectra of  the spiral galaxies are similar to those of the radio gal- axies. The remainder of this paper addresses the evolu- tionary sequence suggested by the data  above.



The transition of two interacting Birkeland currents of galactic dimensions [1] into the morphology of a spiral galaxy is best seen in time-lapse photography (16-mm films) from the long-time simulation runs. The reason for this is the rapidity with which the spiral is formed once  the distance between interacting plasmas has closed to the order of a filament radii. Nevertheless, single frame pho- tographs from the simulations can perhaps convey the gross  morphological  change  (but do not  convey  the dy-

morphological types.

The transition proceeds as follows. For the case of a field-aligned electric field 6JJ It, oriented along the + z di- rection (out of the plane of the figure) in the columns, the electrons in both columns spiral downward in counter- clockwise rotation. Likewise, the ions spiral upward in clockwise rotation.  The  current  density  is J  =  ii,.q„ e„  + n, q, u, where the subscripts e and i denote electrons and ions, respectively. The quantity q is the charge on the particle, u, = ii, and u,. >> u, are the average  (drift) velocities. For the initial separations depicted in the first few frames of Figs. 4-6, the Biot—Sava rt force is predom- inantly attractive [ 1, sec. lV] and the  relative  velocity  of the two  plasma  filaments  are approx imately  1000 km/s  (T

=  100, [I, sec. VI-B]).

The relative velocity of the plasmas increases directly with increasing current and reaches u velocity of several thousand kilometers per  second  near  minimum  separation (T -- 400, Fig. 5). The simulation  current  ranges  from  about 2 x i 0" to 4  x  10"  A  iR  Fig.  S. Collision  does not occur because the repulsive force  of  the  counterpar- allel azimuthal currents becomes  equal  to,  then  exceeds, the attractive force at separations of  the  order  of  the  plasma radii. At this time the translational momentum is converted into angular momentum  because of the i  =  in x II torque between filaments (left-hand or clockwise twist) where in is the magnetic moment of a filament. Concomitant  with  the  attraction,  repulsion,  and  rotation, is a reconfiguration of the current/plasma cross-sectional profiles in the filaments.  In  particular,  the  initially  circu- lar cross sections are deformed into oval shapes that  then take on a ‘jelly-bean-like’ profile prior to forming em- bryonic spiral arms. During this process, the elliptically shaped quasar formed midway between the  two  synchro- tron radiating plasmas is enclosed by the plasmas them- selves, as they spiral inward. During this time, the quasar dimensions narrow  and  the  enclosed  sump  density  in- creases because  of the  increasing  isobaric  pressure. The

- 50-kpc channel (T -- 255) is reduced to a few kilopar- secs length [ l, fig. 12]. The trailing spiral arms then lengthen with time. Bostick [ 13], [14] was the h rst to rec- ord the formation of spiral structures in the laboratory from interacting plasmoids and to note the striking simi- larity to their galactic analogs.
































(b)                                  (a)

Fig. 7.  (a) Selected frames from galaxy simulation run TDD. w, /=„  =   3.0,

T z -- T,z  —-  2 keV. Acceleration field, 0.002 cells per time step squared.

(b) Optical photographs of the galax ies NGC 3187,  M95, dnd   M64.






0        100      200       3tX1       400       500



3) a burst of synchrotron radiation from each filament with a total luminosity L  - 1037 W and duration  T

                            = 70—110 [1, fig. 5].

The double radio source with the elliptical plasma at the center and the QSO with the very active nuclear compo- nent differ only in the strength of the induction field mid- way between  filament cross sections.

  1. Radioquasar to Ra‹iioquiet QSO

This phase of the evolution is marked by 1) a decrease in radio lobe (extended component) luminosity when T > 110; and 2) an increase in central component activity for  T - 250—350. Thus the power in the extended compo- nents fades while the increased induction field and the in-


Fig. 8. E electric field waveform at center of simulation region for run TD6. Abscissa is given in simulation units        (Paper I).



creased compression of plasma in the elliptical sump in- tensifies the synchrotron radiation from the compact component. The compact component can appear as an isolated synchrotron source during this period (e.g. , Fig. 2, linear  size  -0.01 kpc).

  1. Radioquiet QSO to Seyfert Spiral

The development of the plasma and field morphologies of a radioquiet QSO to a Seyfert spiral actually encom- passes  a  transitional  period  involving  the  formation of





                                                                                     IEEE  TRANSACTION S ON  PLASM A  SCIENCE,  VOL.  PS-14,  NO.  6,  DECEMBER 1986


components of radio galaxies and radioquasars. Fig. 10 is an example of this geometry. Like S0 galaxies (galaxies with little or no evidence of star-forming activity in the disk) E galaxies are found most frequently in regimes characterized by high galaxy density, i.e. , areas most sus- ceptible to interactions.
















Fig. 9. lsodensitometer tracing of the elliptical galaxy M87, made from a 60-min exposure on 111a-I emulsion with the I .2-in Palomar Schmidt telescope. The inner circle is the diameter given in the Shapely-Annes catalogue, while the outer ellipse spans as much as 70 arcmin. The hor- izontal extent of the image frame is 500 arcmin. Note that the inner isophotes have vertical major axes, but the outer isophotes show notice- able clockwise twisting (cf. [ I, Fig. 8]).



several peculiar plasma geometries  (Fig. 4,  450— 550). (If the ratio of the nucleus luminosity to the lumi- nosity of a galaxy is 0. 1 to unity  or more,  the galaxy  is of the Seyfert type.) During this period the incoming plasma of the extended components has not yet coalesced with the quasar center. The synchrotron radio activity in the extended components has decreased markedly (Fig. 8) and the quasar core ( a 1 kpc) consists of highly activated,

strongly radiating, dense ( - 1010     ') plasma because of

the continued magnetic sump compression ( - 10 Pa). The density decreases with isobaric gradient, being ten- uous at the periphery of the sump. Time-lapse photogra- phy shows that the inward velocities of the isobars are not quite linear in time but pulsate as they   compress.

The extended components, quieted for T > 110, reap- pear as the peculiar or spiral plasma morphologies that form about the compact nucleus.

  1. Peculiar and Seyfert Galaxies to Spiral Galaxies

Beyond T - 600, coalescence of the outer plasma com- ponents on the excited compact center begins. Addition- ally, the continued electromagnetic compression on plasma confined in the sump can be expected to start the gravitational collapse of this material. This manifests two effects: the disappearance of the quasar plasma emission and the appearance of stars (Section VIII).



Elliptical (E) galaxies, as distinct from peculiars, irreg- ulars, and spirals, are characterized by a very smooth tex- ture, a bright nucleus, and a tenuous outer envelope of large extent (sensitive photographic plates show that the envelope may be 20 times the diameter of the nucleus, Fig. 9) [15]. As mentioned in Section II, Ellipticals are most  often  found  midway  between  the  extended radio

  1. Irregular Galaxies and  “Dust Lane ” E Galaxies

The elliptical sump formed midway between two ex- tended plasma components is the result of the coming to- gether of the magnetic field lines of two (or more) adja- cent filaments. At early times, the topology of these field lines is that of two “clashing  cymbals”  (cf.  [1, figs.  6 and 8], that is also the shape taken on by intergaJactic plasma between the extended ciirrent-conducting compo- nents. Examples include not only irregularly shaped gal- axies (Fig. 11) but also E and some 50 galaxies with “dust lanes” [12]. The dust lanes are usually aligned perpen- dicular to the major axis between extended components (Fig. 12), as they must be for plasma pushed in from either filament.

  1. Flattened E and 50 Galaxies

Elliptical galaxies are classified in a sequence from E0 to E7, according to the degree of apparent flattening. E0 systems appear c;ircular, and E7 are the most oblong known. All flatter galaxies seem to be spirals and, even among the E7’s, many may be type S0, systems that re- semble spirals but lack distinctive arms. In fact, E7 gal- axies often show slight spiral arms or perturbances. The morphologies described above are the shape that Birke- land currents take when they are closely spaced (e.g., tens

of kiloparsecs) or carry weaker galactic currents (I - 1017

  1. A) (Fig. 13). As shown, when closely adjacent Birkeland currents form out of panicle flows along magnetic field lines, a merging and torquing of these plasmas produces forms that have a distinct gap at midsection. These then merge together in rotation to produce an elliptical shape that may exhibit very weak spiral-arm



Hubble originally believed that elliptical galaxies evolve into spiral galaxies [16] and this seems to be borne out by the simulations. However, the simulations show that another class of galaxy, the peculiars [17], [18], bridge the formation of ellipticals to spirals.

Spiral galaxies are the most abundant type known; of all classified galaxies 78 percent are spiral (75 percent normal spiral and 25 percent barred spiral), 18 percent are elliptical, and 4 percent are irregular (Table II).

  1. Normal and Barred Spirals

Whether a normal spiral (S) galaxy or a barred spiral (SB) galaxy forms out of the plasma interaction depends primarily on the profile or cross section of the current- carrying filaments, its density distribution, and strength of the azimuthal magnetic fields. Bars form when the in- teracting plasma regions are sharply divided in  plasma

















































































phtitt›gru ph  front the Hubble atlas (after R.D. Eckers  [ 12]).




NGC 51Z8











N6C J947
























0 151-497, NGC 1947. N €iC .5266, IC 4:170. and NGC’ 1316.



density,    while   normal    spirals   tend   to   form   when   the   in-      ri . 13. Single-frame stills ot‘ plasma in Galaxy simulation f102. ui,

tergalaclic   plasniii  supporting  the  current-conducting   fil-

aments is more homogeneous overall.


  1. Rotation Ctiurarteristi‹ s of Spiral Galaxies

Fig. 14 shows the radial velocity versus distance from the galaxy center typical of spiral galaxies [ 19]—[23]. These data show I )  a nearly linear solid-body rotation for

the galaxy  center (the first few aicminutes from    center);

2) a nearly radially independent velocity profile in the spiral arms; and 3) distinct structure  in the spiral arms  that appears on the so-called  flat p‹›rtion  tif the  velocity








(TIME T —— 0—2000)


classification             observed            time in cifssi fixation stale, simulated





20a 57a

(T=50 to 450)

(T=450 to 550)




(T=550 to 20tXI)














' NGC 5t28

























linear dimension


Fig. 14. Spiral galaxy rotational velocity characteristics. The bottom right- hand-side  curve  is  taken  from  simulation  run  DD4  during  time   steps  1745- 1746.



curve (beyond the first few arcminutes or, equivalently,     instability is produced. The effect of this instability shows the first few kiloparsecs).                                                        up in both the cross-sectional views of the  spiral arms The simulated velocity curve in Fig. 14 corresponds to         (Figs. 4-7, late time) and in the velocity profile (Fig. 14, the spiral galaxy in Fig. 6 at time T —— 1750 (this curve is                simulation curve). Because the plasma in the spiral arms the differential rotation measured between T —— 1749 and                          is very nearly neutral (the B x CB charge separation pre- 1750). The simulation data illustrate that 1) the plasma                                                                                                                vents completely  local neutrality), the diocotron instabil- core rotates very nearly as a solid body, and 2) the spirals                ity is moderated somewhat. The azimuthal (sideways) ve- arms grow in length as they trail out along the magnetic                locity of this instability  is given by  P = P’ cot(J), where

isobars. Concomitant with the lengthening of the arms is      P’ is the slowly varying radial velocity component and

a thinning of the arms. Because of this, and the axial cur-     = f (n„  B, v ), where *e iS the effective electron collision rent conducted through the thin plasma arms, a diocotron  frequency in the arm. Good examples of this are found in











772                                                                                                IEEE  TRANSACTIONS  ON  PLASMA  SCIENCE,  VOL.  PS-14,  NO.  6,  DECEMBER  1986


Fig. 15. Optical photographs of the galaxy NGC 3646. Nole the well-de-

fined diocotron  instability  structure in the spiral’s arm.




Fig. 17. The Centaurus chain of galaxies. Note thai the ring of NGC 4650A (top) is transverse to the chain or filament axis.
















Fig. 16. (a) Single frame from galaxy simulation SAR at I = 2300. (b) The interacting pair NGC 450-UCC 807. (c) The imeracting galaxy pair IC 2163-NGC 2207.

the Sc-type galaxies M101, NGC 253, and NGC 2998 (Fig. 14). Fig. 15, NGC 3646, is an example of a very large diocotron instability in the spiral arms.

  1. Evolutionary Sequence for Spiral Galaxies

It is often the case in the literature that Sa and Sb spirals and Sc and Sd spirals are referred to as “early Hubble types” and “late Hubble types,” respectively. However, Hubble emphasized that his classification sequence was not meant to be an evolutionary sequence [16]. The sim- ulations show first the formation of elliptical galaxies. Then, as the synchrotron-radiating Birkeland current-con- ducting outer plasma components move inward on the el- liptical core, peculiar galaxies form and in sequence the spiral types Sd, Sc, Sb, and Sa (or their barred equivalents SBd, SBc, SBb, SBa). Stars form first in the densely com- pressed elliptical core (Population II stars) and then in the pinched plasma that make up the spiral arms (Population I stars). For Sd and Sc galaxies, the axial Birkeland cur-

rents are just reaching the Alfvén-Carlqvist threshold 0.1 N 10 20 A/m' (Section VIII) and star formation is irreg- ular. For Sb and Sa galaxies the current is a     20 A/m'

and star formation follows closely the morphology of the











plasma in the spiral arms that are usually fragmented be- cause of the diocotron instability. The well-known Baade description that stars in spiral arms appear “like beads on  a string” is also an equally apropos description for the simulated 8alaxies. The older elliptical galaxies, com- pressed in size to make up the core of spiral galaxies, are forced to rotate as a rigid body by the in X 2f torques exerted by the incoming outer components while the spi- ral arms trail behind.  The vortex  motion  of the beads  of

               stars in the arms provides the characteristic cot(J) motion on the rotational  velocity  curves of spiral galaxies.





Birkeland currents often occur in sheets and where these have dimensions of a hundred kiloparsecs or more, fila- mentation of the plasma into a number of galaxies can  take place. Because of the  r  '  force there is a tendency  for filaments to pair up, eventually leading to neighboring spiral galaxies. The classification of multiple interacting galaxies can be spiral—spiral or spiral—peculiar but do not include E or 50 galaxies. Moreover, the spirals tend to be Sd, Sc, or Sb, or the barred equivalents SBd, SBc, or SBb. With this pairing is a decrease in cluster density for the multiply interacting spiral galaxies. Fig. 16(a) shows ad- jacent spiral galaxies formed from a single sheet Birke- land current.  The connection of a spiral configuration  to   a spherical or oval configuration by a thin filament is a commonly observed morphology, both in simulations and in laboratory experiments [24]. A galactic example of this behavior is the Markarian 205— NCG4319 pair [25]. Fig. 16(b) and (c) shows the interacting spiral pairs IC2163- NGC2207 and UGC807— NGC450 [26)-[29]. Because the

Birkeland current is part of a closed-circuit element, 8al- axies occur periodically along the gigaparsec—subgigapar- sec filament where double layers form and where inter- actions with neighboring filaments occur. The Centaurus chain of galaxies (Fig. 17) may be an example of galaxies forming along filaments.




Cosmic plasma generally is composed of the elements in varying abundances. As such. plasma chemistry plays an important role in the formation of galaxies [30]. The elements in cosmic plasma are detected by absorption and emission lines in the electromagnetic spectrum of cosmic


Neutral hydrogen is detected from galaxies via the Van

de Hulst radio-emission line at h = 21. 11 cm (/ = 1420.4 MHz), which arises from the transition between the hy- perfine-structure sublevels of the 8•ound state of a hydro- gen atom [31]. The importance of the study of interstellar hydrogen in this line consists of the fact that this is the sole procedure for direct observation of neutral hydrogen in galaxies.























(a)                                                   (b)

Fig. 18. Hi distributions superimposed on optical photographs of galaxies. (a) NGC 4736 (36] . NGC 5033 [37). and NGC 4 l51 [35). (b) NGC  3198

[37] and M83 (34] .



  1. Marklund E X B Convection and Neutralization of Plasma in Galaxies

Under the influence of the E 5 B force, both the elec- trons and ions drift with the  velocity



so that the plasma as a whole moves radially inward. This mechanism provides a very efficient convection process for the accumulation of matter from plasma [32]. The ma- terial should form as a filamentary structure about the twisted magnetic flux tubes, the lines of which are com- monly referred to as “magnetic ropes” because of their qualitative pattern [5]. Magnetic ropes should tend to co- incide with material filaments that have a hi8her density than the surroundings (this is also the case for the fila- ments in the current sheath of the plasma focus). The cosmic magnetic flux tubes are not directly observable themselves, but the associated filaments of condensed matter can be observed by the radiation they emit and ab- sorb.


774                                                                                               IEEE  TRANSACTIONS  ON  PLASMA  SCIENCE,  VOL.  PS-14,  NO.  6,  DECEMBER  1986


























Fig. 20. (a) Galaxy M5l. Top—Infrared photograph, 0.71-0. 88-pm, 1.2- in Palomar Schmidt telescope. Bottom—Ultraviolet, 175-275-rim, 0.3-in rocket-borne telescope. (b) Simulation run TO6, time step 600. Top—Plasma distribution. Bottom—Electric (induction) field energy dis-









Fig. 19. (a) HI distribution superimposed on an optical photograph of NGC 4 151. (b) Simulation electromagnetic energy density superimposed on simulated plasma galaxy.



Marklund found a stationary state when the inward con- vection of ions and electrons toward the axis of a filament was matched by recombination and outward diffusion of the neutralized plasma. The equilibrium density of the ionized component normally has a maximum at the axis. However, because of the following mechanism, hollow cylinders, or modifications of hollow cylinders of matter, will form about the flux  tubes.

Because of the radiated loss of energy, the filaments cool and a temperature gradient is associated with the plasma. As the radial transport depends on the ionization potential of the element, elements with the lowest ion- ization potentials are brought closest to axis. The most abundant elements of cosmical plasma can be divided into groups of roughly equal ionization potentials as follows: He(24  eV);  H,O,N(13  eV);  C,S(11  eV);  and Fe,Si,Mg(8

eV). These elements can be expected to form hollow cyl- inders whose radii increase with ionization potential. He- lium will make up the most widely distributed outer layer; hydrogen,  oxygen,  and nitrogen should  form the middle

layers; and iron, silicon, and magnesium will make up the inner layers. lnterlap between the layers can be expected and, for the case of galaxies, the metal-to-hydrogen ratio should be maximum near center and decrease outwardly. Both the convection process and the luminosity increase with the field Ed.

For the case of a fully ionized hydrogenic plasma, such as that of the simulation model, the ions drift inwards un- til they reach a radius where the temperature is well below the ionization potential and the rate of recombination of the hydrogen plasma is considerable. Because of this “ion pump” action, hydrogenic plasma will be evacuated from the surroundings and neutral hydrogen will be most heavily deposited in regions of strong magnetic flux [33].

  1. Distribution of Neutral Hydrogen in Galaxies

High-resolution observations of neutral hydrogen in ir- regular and spiral galaxies usually reveal very extended Hi distributions. Contour maps of the HI typically show a relative lack of Hi in the cores of spiral galaxies but high Hi content in the surrounding region, usually in the shape of a “horseshoe” [34]—[40]. This region is not uniform but may have two or more peaks in neutral hydrogen con- tent. Fig. 18 shows several examples of Hi distributions in spiral galaxies.

A direct comparison of the simulation predictions is possible by overlaying the simulation galaxy intense 21- field distribution (i.e., regions of strong Marklund con- vection) with the galaxy and also overlaying the HI dis-




tribution of a galaxy with its optical image. Fig. 19 shows the Seyfert NGC 4151 and its simulation analog, while Fig. 20 shows the Seyfert M51 with its simulation analog. The simulation allows the two peaks in neutral hydro- gen to be traced back to their origin. Both are found to be the remnants of the originally extended components, i.e., the original Birkeland filaments. The hydrogen deficient center is the remnant of the elliptical galaxy   or quasar



  1. Pinch Compression of Dark Interstellar Clouds

The importance of the pinch effect in interstellar plasma clouds located in  galactic-dimensioned  current-conduct- ing plasma is illustrated  by  the following  two examples [ 1

  • Consider first   an  interstellar   cloud  of   100 solar



  1. The Motion of Solid Bodies Condensed in Plasma

The motion of a solid particle in a plasma obeys the equation


d/ "

where q is the electric charge of the particle, —Ali is due to viscosity, and/is the sum of all other forces, including the  radiation pressure.

Depending on the size of the particle, we have three typical cases.

  1. Very Small Particles: The term q(E + v X B) dom- inates over mg and the particle is part of a dusty plasma. Under cosmic conditions this is true if the size of the par- ticle is < 10 nm [30]. In the case of large electric charges the limiting size may rise to 100 nm.
  2. Groins: If the size of the particle is so large that the electromagnetic term is negligible, we have an interme- diate case dominated by viscosity and gravity. The par-


masses, M   ——   2  x  1032 k

, occupying a volume of  the

ticles in this regime are referred to as grains. Their equa-


linear  dimension  I    ——


  1. The temperature of the

tion of motion is


cloud is T ——   10—10' K. This  represents  a cloud of   ap-

proximately the same mass as the Orion nebula. The num- ber of atoms present in the  cloud  is  -1 59 implying  a mean density  of n  = 10' m"   (l 2  cm    ) and giving N ——

1   42  atoms   -             1      Putting  the  latter  figure  and  the tem-

perature  above into the Bennett  relation  [l, eq.  (4)], we

find  that  an  electric  current  of  I   -5  X   1013—2   x   1014  A

has to flow throughout the cloud in order to produce a considerable compressional effect.

2) Consider next a cloud of one solar mass only, Mf, =

2 x  1030  g  having a tern  erature of T ——  10—1 2 K and

Under conditions in interstellar clouds this may be valid for particles of the order of 10 pm.

  1. Large Solid Bodies: For “particles” of the size of kilometers or more, the inertia and gravitational terms dominate. Electromagnetic forces are negligible, and vis- cous forces can be considered as perturbations which may change the orbit slowly. Depending on the properties of the cosmic cloud, viscous forces become important for meter or centimeter sizes. The equation of motion is then


an extent of I,  =

  1. Hence, u  =  1 6  10’     3                                    dv


(1-10" cm") and N  -—   1040   104    atoms  -         1   The cur-

rent needed  to compress  the cloud  is now  found to be I

dt ——  mg —  gv.



l 2_ 5  x  1013 A .

The current density for both cases outlined above is ap- proximately 0. I —1.0 x 10 20 A/ 2 For the simulated spiral galaxy, I        - 10 kpc, 1      20  A, so that J  -0. I  X

10"' A/   2 and  the threshold  for star  formation  is  met.

(Star formation in the elliptical sump is expected to start earlier because of the compressive forces on the dense plasma  contained  there.)

In the analytical examples above, the interstellar clouds were assumed to be contracted by a Bennett pinch, imply- ing a pure toroidal magnetic field and a pure axial current. According to Alfvén and Carlqvist [2] all magnetic con- figurations ranging from this kind of pinch to the force- free states would also give rise to contractive forces. For the plasma to pinch in these cases the total current must  be larger than that given by the Bennett relation. In order that the total current I, conducted by the galaxy reach the threshold current density for star formation, it is required that the double layer electric field E, exist at least to times comparable with the time of spiral formation, i.e. , T- 1000. (The time constant of the gigapaisec length  cir-



The transition of plasma into stars involves the forma- tion of dusty plasma, the sedimentation of the dust, the formation of stellesimals, and then the collapse into a stel- lar state. While the above process appears amenable to particle simulation, a crude approximation of proceeding directly from charged particles (actually a cloud of charged particles [1, sec. III]) to mass particles is inves- tigated.

The transition of charge particles to mass particles in- volves the force constant, that is, the ratio of the coulomb electrostatic force between two charges q sepamted a dis- tance r



to the gravitational force between two masses in separated a distance r


cuit is of order  10 4 s [1, sec.  VIII].)

Fg (r) ——   — G 2   r’.






776                                                                                               IEEE  TRANSACTION S  ON  PLASMA  SCIENCE,  YOL.  PS-14,  NO.  6,  DECEMBER  1986


In   the  particle   algorithm   this   change   is  elected   by     ponents  is  caused  by  a  reconfiguration  of  the magnetic

  1. a) changing all particles  to a single species; b)  limiting  isobars  so that  first plasmoids,  then  flattened  “jets”  ap-  the axial extent of the simulation to be of the order of less  pear on either side of the elliptical sump. (These geome-  than the extent or the radial dimension, i.e. , about the size  tries are formed as the isobaric profiles act on the confined  of the expected double layer dimension; c) setting the ax-  )  The  flattened  magnetic  field  minima  on  either ial velocities  to zero; and d)  setting the charge-to-mass   side of the sump produce  a plasma  trap so that  quasars    ratio equal to the negative of the square root of the grav- tend to form along this line. Spiral arms form as the outer itational constant (times 4reb). This last change produces plasmas from the extended radio components move in- attractive mass particles via the transformation  +c(r) —— ward  and rotate around,  and then  coalesce  with the el-   mq(r) ‘in the force equation F ——  —4 ‹p , where liptical center.





p p(r) ——        /4re  r                                  (7)



ve(r) —— — Gm2 / r                                  (8)


The difficulties encountered in explaining the dynamics of elliptical and spiral galaxies in the absence of magnetic fields and plasma physics are well known [15], [48]. In a


are the electrostatic and gravitation potentials, respec- tively.

Preliminary investigations of the electromagnetic-grav- itation potential transformation in three-dimensional codes have been reported elsewhere [41]. The actual process is probably somewhat more complicated than the simplified procedure described above and may involve two interac- tions: the gravitational force of the plasma arms with stars formed in the arms, and then the gravitational interactions among the stars (themselves gravitationally bound plas- mas) in the arms. The earlier formed stars in the elliptical sump plasma must also be included in the simulation. It has been suggested that stars may disperse under gravi- tational self-forces as the arms rotate, leading to the for- mation of Sa- and SBa-type galaxies from earlier plasma Sc and Sb and SBc and SBb galaxies  [42].


  1. DisC USSION


Between the years 1967 and 1969, excellent  reviews and theories on quasars had already appeared that stressed the importance of magnetic fields and plasma theory in order to explain not only the synchrotron radiation ob- served from quasi-stellar objects but also the morpholo- gies of the extended radio components and the bright cores [43], [44]. These ideas were reexamined by Ginzburg and Ozemoy in a criticism of dense gravitational  objects  as the source of energy  in radio galaxies  [45].

The association of an elliptical class of peculiar galax- ies with pairs of radio sources has been advocated by Arp [25]-[28]. Arp noted that radio sources with similar flux densities tend to form pairs separated 2°—6° on the sky and stated there was a tendency for a certain class of pe- culiar galaxy to fall approximately on the line joining the pair. These peculiar galaxies are often elliptical galaxies that show evidence for structures that may have been ejected from (or are falling into) them  [46],  [47].

The simulations show that, depending on the time of evolution, both processes occur. At early time, the compression of intergalactic plasma in the sump produces the peculiar elliptical-like morphologies observed (Sec- tion IV-A  and Fig.  9). At yet later time,  the  ‘ejection”  of plasma  on a line connecting  two extended  radio com-

very perceptive paper, Toomre [49] questions the di— lemma that photographic data of galactic mergers present to gravitational A/-body simulations. The complexity of this problem is further compounded by the existence of peculiar galaxies, the morphologies and topologies of which do not inspire confidence in theories postulating linear waves or shocks as the molding forces [50]—[52]. Observationally, these difficulties are perhaps best sum- marized in The Atlos of Peculiar  Galaxies [18].

The Atlas as it has been realized in the following pages illustrates again that galaxies cannot be char- acterized as just assemblages of stars, radiation, and gravitation. The following Atlas pictures emphasize the importance of dust in some; they particularly im- ply a much more important role for the gas in gen- eral and point to the existence of either new forces or forces which previously hate been little consid- ered. For example, the twisted distorted shapes and curious linkages pictured here attest to the fact that there are viscosity-like forces present that in some cases are dominant. Probably these forces are due to magnetic effects. Vorontsov-Velyaminov has stressed in the past the probable magnetic nature of these effects. Magnetic forces are very difficult to study, but may be very important in our Universe. The recent radio-astronomy discoveries of violent events in galaxies reveal sources of energetic charged particles. These charged particles interact with magnetic fields and offer the hope of mapping, measuring, and understanding cosmic magnetic fields. Exploration of the connection between the plasmas observed with the radio telescopes and the optical evidences of plasma effects pictured in the present Atlas ‘is now open to us.

This paper and its predecessor (Paper I) have addressed the evolution of the plasma universe in order to under- stand both electromagnetic spectra over ten octaves, as well as the visual morphologies observed at optical wave- lengths.

It is emphasized that the simulations described in Pa- pers I and II do not pertain to wandering plasma  clouds  or  galaxies  whose  happenchance   encounters  might  be
















thought to produce tails or spiral structure. In a universe of plasma, Birkeland current sheets (i.e. , the flow of charged particles along magnetic field lines) can occur wherever a circuit and potential source exist. The source can simply be the transverse motion of plasma of dimen- sion 10—50 Mpc through a 10"-T field at - 1000 km/s. The circuit is thought to be composed of plasma whose filamental length, in analogy to laboratory filaments, may be hundreds of megaparsecs. As in the laboratory, Birke- land current sheet filaments, and lumps within the fila- ments interact with their neighbors to produce the phe- nomena outlined in this paper.

Two plasma columns, as modeled here, represent the two closest filaments in a Birkeland sheet (Fig. 16 is a simulation where the initial plasma configuration was a Birkeland sheet). When the plasma profile, density, tem- perature, strength, and orientation of any external fields have been specified, no additional assumptions are pos- sible in a simulation; the configuration will evolve through cascades of nonlinear states according to the electromag- netic (and when present, gravitational) forces on the plasma. In this regard, the data presented here were ob- tained by postprocessing the simulation particle, field, probe, and history dumps. The totally unexpected (to the author) phenomena replicated in the simulations appear to be far more universal than the interacting laboratory z pinches for which they were intended. When scaled to cosmic dimensions the simulations show:

  • a burst of synchrotron radiation of luminosity 3

W lasting 10 —10' years as the interaction began;

  • isophotal topologies of double radio galaxies and quasars, including juxtapositioned “hot spots”  in the radio lobes (cross sections of the interacting Birkeland currents);
  • the formation of “dust lane” peculiar and elliptical galaxies at the geometric center of quasars and radio galaxies (due to plasma trapped and compressed within the elliptical magnetic separatrix);
  • a spatially varying power law along the major axis of the simulated double radio galaxies in agreement with observations;

alternatin 8 beams of betatron-pumped synchrotron- emitting electrons on either side of the elliptical cen- ter (these have the morphologies (i.e. , “knots” or vortices) and polarization properties of jets);  and

a “superluminosity” and fading of jets as the beta- tron-induced acceleration field sweeps over and ig- nites previously confined plasma.

The simulation time frame of this investigation lasted some  1  8  10’ years.  The lifetime and evolution  of qua-

sars and double radio sources, the so-called end problem of double radio galaxies, was addressed in this paper (Pa- per II) by continuing the simulation run - 1-5 X 10’ years farther in time. This extension of the simulation   showed:

  1. the transition of double radio galaxies to radioqua- sars to radioquiet QSO’s to peculiar and Seyfert gal- axies, finally ending in spiral galaxies;



  • the formation of irregular and dust lane galaxies, as well as more flattened E and S0 galaxies within the magnetic separatrix;
  • barred and normal spiral galaxies resulting from the inflow of plasma from the outer Birkeland currents onto the the elliptical galactic center; the character- istic rotational velocities of spiral galaxies including the fine-detail vortex cotangent structure on the “flat” portions of the spiral—arm velocity compo— nents;
  • replications of the morphologies of multiple inter- acting galaxies;

“horseshoe” like regions of  nearly  neutral  Hi  gas in spiral galaxies resulting from the convection and neutralization of plasma into regions of strong ga— lactic magnetic  fields; and

6)  toroidal  and  poloidal  components  of  the   galactic

magnetic field with field strengths reaching 2 X  10 G at the galactic center (fields as high as 10  2 G can

occur in concentrated regions). These results were reported prior to their observation in the Galaxy [41].

Finally, the pinch effect from the currents carried in the galactic plasmas illustrates the importance of the electro- magnetic field in initiating the first stages of dusty plasma collapse into stellisimals, then into stars  [53J .

The investigation of complex filamentation processes in laboratory plasmas by the particle-in-cell approach has thus led to unexpected analogies in cosmic plasmas. As computer powers (i.e. , speed and memory) increase, the ability to resolve and understand the evolution of the plasma universe in greater detail can be expected. Perhaps as important, as the universe represents an unprecedented plasma data bank, insights on laboratory data can also be expected.



A.L. Peratt, “Evolution of the plasma universe: 1. Double radio gal- axies, quasars, and extragalactic jets,” IEEE Trans. Plasma Sci. , sob PS- 14, no. 6, pp. 639—660, Dec. 1986.

[2] H. Alfvén and P. Carlqvist, “Interstellar clouds and the formation of stars,’ ’ Astroph ys.  Space Sci. ,  vol.  55, pp. 484-509, 1978.

[3] R. Beck, “Magnetic helds and spiral structure,” in Internal Kine- mnfics and Dynamics of Galaxies, E. Athanassoula, Ed. Amer. As- tron. Union, 1983, pp. 159- 160.

[4]  R.  Beck,  ‘ ‘Interstellar  magnetic  fields,’ ’  IEEE  Trans.  Plasma  Sci. .

vol. PS-14, no. 6, pp. 740—747, Dec.   1986.

[5]  H. Alfvén,  Cosmic Plasma.    Dordrecht,  Holland:  Reidel,  198 I .

[6] H. Alfvén, “ Paradigm transition in cosmic plasma physics,’’ Phys- ica. Scripta, vol. T2/ I , pp.  10- 19,  1982.

  1. Alfvén, “Double layers and circuits in astrophysics,” IEEE Trans.

f•insmn Sci. , vol.  PS- 14, no. 6, pp.  779-793,  Dec. 1986.

(8] H. Alfvén and D.A. Mend is, “Plasma effects in the formation, evo- lution and present configuration of the Satumian ring system,” Adv. Space Res. ,  vol. 3, pp. 95- 104,   1983.

{9] G. Burbidge and M. Burbidge, Quasi-Stellar Objecfs.  San  Fran- cisco, CA: Freeman, 1967, pp. 85—94.

[l0)  R.  Fanti  and  G.C.  Perola,   in  Radio  Astronomy   and  Cosmology.

D.L. Jauncy, Ed. (lAU Symp. 74).  Dordrecht,  Holland:  Reidel,  1977, p. 17 I .

G.C. Perola, “Radio galaxies: Observations and theories of their extended components,” Fundamentals of Cosmic Phys. , vol. 7, pp. 59- 130, 1981 .

{ 12]  R.D. Ekers,  “Radio observations  of the nuclei of galaxies,”  in  The




778                                                                                               IEEE TRANSACTIONS  ON PLASMA  SCIENCE,  VOL.  PS-14,  NO.  6, DECEMBER   1986



Formation and Dynamics of Galaxies, J.R. Shakeshaft, Ed. (IAU Symp.  58).    1974, pp. 257-277.

[133 W.H. Bostick, '‘Simulation of astrophysical processes in the labo- ratory,”  Nature,  vol.  173, p. 214,  1957.

[ 14] W.H. Bostick, “Experimental study of plasmoids,’’ in Electromag- netic Phenomena in Cosmical Physics, B. Lehnert, Ed. (Int. Astron. Union Symp. 6, Stockholm, Sweden, 1956). Cambridge, England: Cambridge,  1958, pp. 87—98.

 K.M. Strom and S.E. Strom, ‘‘Galactic evolution: A survey  of  re- cent progress,’’ Science, vol. 2 16, pp. 57 l—580, 1982.

[ 16] E. Hubble,  77ie Realm of the Nebulae.    New York:  Dover,  1958.

[17) B.A. Vorontsov-Velyaminov, Atlas and Catalogue of Interacting Golnxies, vol. 1. Moscow, USSR: Stemberg Inst. , Moscow  State Univ. ,  Moscow, 1959.

[18]  H.  Arp,  Atlas  of  Peculiar  Galaxies  (Astrophys. J.  Supplement series

123- 132), vol. 14, 1966, pp. 1—57.

[19] V: C. Rubin, W.K. Ford, Jr. , and N. Thonnard, “Extended rotation curves of high-luminosity spiral galaxies. IV. Systematic dynamical properties, Sa-Sc,” Astroph ys.  J. , vol. 225, pp. Ll07-L 111, 1978.

[203 G. delaucouleurs, “Survey ing velocity fields in galaxies,”  Sky Telesc. , vol. 62, pp. 406—410, 1981.

  1. Marcelin, J. Boulesteix, and  G.  Courtés,  “The  velocity  field  of the ionized gas in the barred galaxy NGC 925,” ztsiron. Astropfiys. , vol.  108, pp. 124- 140, 1982.

[22] M. Marcelin, J. Boulesteix, G. Courses, and B. Milliard,  ‘ ‘NGC 5128—A galaxy with a recently formed disk,” Nature, vol. 297, pp. 38—42, 1982.

[23]  R.  Louise  and  M.  Marcelin,  ‘‘Mass model  of spiral  galaxies disk,”

Astrophys.  Space Sci. , vol. 91, pp. 93—98,  1983.

[24] H.F. Webster and T.S. Hallinan, “instabilities in charge sheets and current sheets and their possible occurrence in the aurora,” Radio Sci. , vol. 8, pp. 475—482, 1973.

[25] J.W. Sulentic, “Confirmation of the luminous connection between NGC 4319 and Markarian 205,’ ’ Astrophys. J. Lett. , vol. 265, pp. L49-L53,  1983.

[26) H. Arp. “Three new cases of galaxies with large discrepant red- shifts,” dsir‹›pIiys.  J . , vol. 239, pp. 469—474,  1980.

  1. Arp, ‘‘Characteristics of companion galaxies,” Astrophys. J. , vol. 256, pp. 54—74, 1982.

[28] H. Arp, “Further examples of companion galaxies with discordant redshifts and their spectral peculiarities,” Astroph ys. J. , voY 263, pp. 54—72, 1982.

[29] H. Arp, “Further observations and analysis of quasars near compan- ion galaxies,”  Astroph ys.  J. ,  vol.  271, pp. 479-506,  1983.

  1. Alfvén and G. Arrhenius, £vofii/ion o/ the Solar System. Wash- ington, DC:  NASA,  1976,  SP-345.

[31] S.A. Kaplan and S.B. Pikelner, The interstellar Medium. Cam- bridge, MA: Harvard Univ. Press, 1970.

[32] G.T. Marklund, “ Plasma convection in force-free magnetic fields as a mechanism for chemical separation in cosmical plasmas,’’ Nature, vol. 277, pp.  370-37 I, 1979.

[33] 1. F. Mirabel and R. Morras, “Evidence for high-velocity inflow of neutral hydrogen toward the galaxy,” Astroph ys. J. , vol. 279, pp. 86—92, 1984.

[34] D.H. Rogstad, I. A. Lockhart,  and M.C.H. Wright,   “Aperture-


synthesis observations of the Hi in the galaxy  M85,’ ’ Astrophys.  J. ,

vol.  193, pp. 309—3 19, 1974.

[35] A. Bosnia, R.D. Ekers, and J. Lequeux, “A 21 -cm study of the Scyfert galaxy NGC 4151,’’ Astron. Astrophys. , vol. 57, pp. 97—104, 1977.

(36] A. Bosnia, L M. Van de Hulst, and W.T. Sullivan III, “A neutml hydrogen study of the spiral galaxy NGC 4736,” Astron. Astrophys. , vol.  57, pp.  373-381, 1977.

[373 A. Bosnia, “21-cm line studies of spiml galaxies. I. Observations of the  galaxies  NGC 5033, 3198,  5055, 2841, and 7331,” Detroit.   J. ,

vol. 86, pp. 1791 —1824, 1981 .

(38] A. Bosnia, “2I-cm line studies of spiml galaxies. II. The distribution and kinematics of neutml hydmgen in spiral galaxies of various mor- phological type,”  Astron.  J. ,  vol. 86, pp.  1825- 1846,  1981.

(39] E. Hummel and A. Bosnia, “Radio continuum observations of the spiral galaxies NGC 2841, NGC 5055, and NGC 7331,” Astron.  J . , vol. 87, pp. 242—251, 1982.

(40] H. Van Woerden, W. Van Dricl,  and  U.J. Schwarz,  “Distribution and motions of atomic hydrogen in lenticular galaxies,” in fnternnf dinemiitics rind Dyn‹imics of Galaxies, E. Athanassoula, Ed. Amer. Astron. Union, 1983, pp. 99—104.

[41]  A.L. Pemtt,  “Simulating  spiral galaxies,”  51:y  Tefesc. ,  vol.  68, pp.

118-122, 1984.

[423 E.J. Lemer, private communication, 1986.

E.J. Lemer, ‘‘Magnetic self-compression in laboratory plasmas, quasars and radio galaxies,” fzisers rind P‹irticfe Seems, vol. 4, part 2, pp.  193-222, 1986.

[433 H.J. Smith, “Observed characteristics of quasars,” in Quasars and high Energy Astronomy, K.N. Douglas et ‹if. , Eds. New  York:  Gordon and Breach, 1969, pp. 167- 180.

 P.A. Sturrock, “Explosions in galaxies and quasars,” in f'f‹isitin Instabilities in Astrophysics, D.G. Wentzcl and D.A. Tidman, Eds. New York: Gordon and Breach, 1969, pp. 297—328.

[45] V.L. Ginzburg and L.M. Ozcrney, “On the nature of quasars and active galactic nuclei,” Astroph ys. Space Sci. , vol. 50, pp. 23-41, 1977.

[46]  H.  Arp,  “Peculiar  galaxies  and  radio  sources,”  Astroph ys.  J. , vol.

148, pp. 321 -365, 1967.

[47] H. Arp, “Ejection from galaxies and galaxy formation,”  in Problems of Physics find Evolution off the Universa, L.V. Mirzoyan, Ed. Er- evan, Armenia: Armenian Acad. of Sci. , 1978, pp. 65—96.

[483 A.  Toomre,  “Theories of spiral  structure,”  mini.  Itev.  dsiron.  As-

trophys. , vol. IS, pp. 437—478, 1977.

[49] A. Toomre, ‘ ‘Mergers and some consequences,” in 77ie Evolution of Gaf‹iries iin# 5teffnr Populations, B.M. Tinsley and R.B.  Larson, Eds.   New Haven, CT: Yale Univ. Observatory, 1977, pp. 401 —426.

[50] H. Arp,  ‘ ‘The persistent  problem of spiral galaxies,” IEEE Trans.

Plasma Sci. ,  vol.  PS-14, no. 6, pp. 748—762,  Dec. 1986.

 J.H. Piddington, “The density-wave theory of galactic spirals,’ ’ is- trophys. J. , vol. 179, pp. 755-770, 1973.

[52] J.H. Piddington, “The role of magnetic fields in extragalactic as- tronomy,”  Astrophys.  Space dci. ,  voY  80, pp. 457-471,  1981 .

(533 B.E. Meierovich, ‘‘Electromagnetic collapse, problems of stability, emission of radiation and evolution of a dense pinch,’ ’ Phys. Reports, vol. 104, pp. 259—347, 1984.