1: %%
2: %% February 8, 2006
3: %%
4: %% First notes on the Battery Letter
5: %%
6:
7:
8: \documentclass[12pt,preprint]{aastex}
9: %% manuscript produces a one-column, double-spaced document:
10:
11:
12: %\documentclass[11pt,manuscript]{aastex}
13: %% preprint2 produces a double-column, single-spaced document:
14:
15:
16:
17:
18: \newcommand{\vdag}{(v)^\dagger}
19:
20:
21: \newcommand{\myemail}{icontop@academyofathens.gr}
22:
23:
24: \shorttitle{Battery Letter}
25:
26:
27: \shortauthors{Contopoulos, Kazanas \& Christodoulou}
28:
29:
30:
31: \begin{document}
32:
33:
34: \title{The Cosmic Battery Revisited}
35:
36:
37: \author{Ioannis Contopoulos,\altaffilmark{1}
38: Demosthenes Kazanas,\altaffilmark{2} and
39: Dimitris M. Christodoulou\altaffilmark {3}}
40:
41:
42: \altaffiltext{1}{Research Center for Astronomy, Academy of Athens,
43: GR-11527 Athens, Greece. \\
44: Email: icontop@academyofathens.gr}
45: \altaffiltext{2}{NASA/GSFC, Code 665, Greenbelt, MD 20771, USA. \\
46: Email: kazanas@milkyway.gsfc.nasa.gov}
47: \altaffiltext{3}{Math Methods, 54 Middlesex Turnpike, Bedford, MA 01730,
48: USA. \\
49: Email: dimitris@mathmethods.com}
50:
51:
52: \begin{abstract}
53: We reinvestigate the generation and accumulation of magnetic
54: flux in optically thin accretion flows
55: around active gravitating objects. The source of the
56: magnetic field is the azimuthal electric current
57: associated with the Poynting-Robertson drag on the electrons
58: of the accreting plasma.
59: This current generates magnetic field loops which open up because of the
60: differential rotation of the flow. We show through simple numerical
61: simulations that what regulates the generation and
62: accumulation of magnetic flux near the center
63: is the value of the plasma conductivity. Although the conductivity is
64: usually considered to be effectively infinite for the fully ionized
65: plasmas expected near the inner edge of accretion disks,
66: the turbulence of those plasmas may actually render them much
67: less conducting due to the presence of anomalous resistivity.
68: We have discovered that if the resistivity is sufficiently high
69: throughout the turbulent disk while it is suppressed
70: interior to its inner edge, an interesting steady-state process is
71: established: accretion carries and accumulates magnetic flux
72: {\em of one polarity} inside the inner edge of the disk, whereas
73: magnetic diffusion releases magnetic flux
74: {\em of the opposite polarity} to large distances.
75: In this scenario, magnetic flux of one polarity grows and accumulates at a steady rate in the region
76: inside the inner edge and up to the point of equipartition when it becomes dynamically important.
77: We argue that this inward growth and outward expulsion of oppositely-directed
78: magnetic fields that we propose may account for the $\sim 30$~min cyclic
79: variability observed in the galactic microquasar GRS1915+105.
80:
81:
82: \end{abstract}
83:
84:
85: \keywords{accretion, accretion disks---MHD---plasmas---stars: magnetic fields}
86:
87:
88: \section{Introduction}
89:
90:
91: The issue of the orgin of cosmic magnetic fields, despite much progress in the
92: field, remains an open issue in astrophysics. The problem is basically of a
93: topological nature, because the magnetic field, being a solenoidal
94: vector ${\bf B}=\nabla\times {\bf A}$ (${\bf A}$ is the magnetic vector
95: potential), is necessarily absent in a homogeneous and isotropic universe.
96: For the same reason, the magnetic field cannot be generated by the potential
97: fluid motions that result
98: from the growth of scalar inhomogeneities in such a universe, as current models
99: contend. As such, the evolution of the magnetic field
100: is akin to that of vorticity ${\bf \omega = \nabla
101: \times {\bf v}}$ (${\bf v}$ is the flow velocity),
102: a fact that has been noticed by at least one (to our knowledge) paper
103: in the literature (see Eqs.~[4] and [7] in Kulsrud et al. [1997]).
104: It is also of interest to note that the source of both ${\bf\omega}$ and ${\bf B}$
105: is proportional to $\nabla p \times \nabla \rho/\rho^2$ ($p$ is the pressure and
106: $\rho$ is the density),
107: a quantity which is nonzero only in the presence of nonbarotropic fluids (Kulsrud
108: et al. [1997] consider as such curved shocks and photoheating). This source term,
109: while generally small, is nonetheless of great importance because in its
110: absence the initial values of {\bf $ \omega$}, {\bf B} = 0 are preserved by the evolution
111: equations. It is generally referred to as the {\em Biermann battery} and leads to
112: magnetic fields that are usually quite weak initially ($\sim 10^{-20}$ G)
113: but thought to be subsequently amplified to the observed mean galactic values ($\sim
114: 10^{-6}$ G) by dynamo processes.
115:
116:
117: Some years ago, we proposed a source-term alternative to the term of the Biermann battery
118: (Contopoulos \& Kazanas [1998], hereafter CK). More specifically, we
119: posited that the Poynting-Robertson radiation force acting predominantly on the
120: electrons of an accretion flow around an active astrophysical source may generate
121: toroidal electric currents sufficiently large to support poloidal magnetic
122: fields that in certain situations could approach equipartition values. The
123: Poynting-Robertson source term is proportional to $\nabla\times(\ell \, {\bf\omega})$,
124: where
125: $\ell= L \, \sigma_{\rm T}/4 \pi r m_ec^3$ is the usual accreting source compactness
126: ($L$ is the source luminosity, $\sigma_T$ is the Thompson cross-section,
127: $m_e$ is the electron mass, $c$ is the
128: speed of light, and $r$ is the distance from the center);
129: as such, besides its much larger magnitude compared to the term of the Biermann battery,
130: this term is additionally important because it provides a hitherto unexplored,
131: direct coupling between the magnetic field and the vorticity equations of motion.
132:
133:
134: In our earlier work (CK) we concentrated our attention on the effects of the
135: Poynting--Robertson source term on the evolution of the magnetic field
136: within the volume of the disk alone, by solving the corresponding (induction)
137: equation in that region. Our study showed that, in a highly conducting plasma,
138: currents and fields grow only for a restricted time of the order of the accretion
139: time, and thus they quickly saturate to values well below equipartition (but
140: still much higher than those induced by the Biermann battery). However, for
141: plasma conductivity below a certain critical value (and a free inner boundary
142: condition that allows the accumulation of magnetic flux interior to that point),
143: we found a different regime in which the magnetic field accumulated
144: around the gravitating object grows at a constant rate; we then showed that in several
145: astrophysically interesting systems, magnetic fields may reach even equipartition
146: values within astrophysically relevant time scales.
147:
148:
149: This `unrestricted' field growth has its origin on the fact that the
150: %works only when one takes into account the fact that
151: magnetic field loops, generated by the toroidal Poynting-Robertson current
152: inside the disk, open up to infinity once they reach the surface of the disk
153: because of its differential rotation (and the ideal MHD conditions outside the
154: disk). This opening up, coupled with
155: sufficiently low conductivity, allows the separation of
156: {\em inward advected} field of one polarity from the {\em outward diffusing}
157: field of the opposite polarity. This effect of magnetic loop opening above and below
158: the disk is central in our mechanism and it has been clearly described in our
159: earlier paper.
160: %While our work went for the most part unnoticed, it was nonetheless
161: Our work was criticized by Bisnovatyi-Kogan, Lovelace \& Belinski (2002), who,
162: arguing along the lines of earlier work by Bisnovatyi-Kogan \& Blinnikov (1977),
163: claimed that the magnetic field grows only over an accretion time scale or
164: so and saturates at values well below equipartition. In support of their view
165: they provided an exact solution of the induction equation with the
166: Poynting-Robertson term included that indeed exhibited the claimed saturation
167: in the ideal MHD limit, while in an Appendix to their paper they argued
168: that their results would not change significantly for finite plasma conductivity.
169:
170: %As such, their solution agreed with ours in that particular limit, as discussed
171: %above; however, we should stress that the field growth found in CK depends
172: %crucially on the presence of finite conductivity and violation of the ideal
173: %MHD conditions.
174:
175:
176: In our opinion, the reason for the difference in the conclusions of the two
177: treatments lies in the consideration of the inner boundary condition to the
178: solution of essentially the same equation. In addition, the one-dimensional
179: character of the model used, although it showed that unrestricted field
180: growth can take place, did not help the reader to visualize the actual
181: geometry of the magnetic field on the poloidal plane. We therefore decided
182: to return to this problem and perform more sophisticated two-dimensional numerical
183: simulations with a variety of boundary conditions which, we hope, will
184: convince the reader that the mechanism described in our original paper
185: can produce astrophysically interesting results. In \S~2 we describe the
186: setup of our simulations and in \S~3 we present the results of our calculations.
187: Finally, in \S~4 our results are discussed and some conclusions are drawn.
188:
189:
190:
191:
192: \section{Simulation Setup}
193:
194:
195: Our simulations improve on those of CK by extending the treatment
196: to two spatial dimensions while incorporating the essential elements
197: of the cosmic battery scenario. Our treatment is still linear in that
198: the dynamics of the accretion disk are prescribed and we solve for the
199: magnetic field evolution; we expect that this simplified treatment
200: will serve as a guide for our upcoming detailed, resistive, numerical MHD
201: simulations of the combined magnetic field and vorticity equations
202: that is in preparation.
203:
204:
205: Our comptuational domain consists of two regions: a viscous, resistive,
206: geometrically thick accretion disk of specified flow velocity field and an
207: overlying force-free magnetosphere where ideal MHD conditions apply.
208: Inside the disk, the magnetic
209: field generated through the Poynting-Robertson azimuthal drag
210: on plasma electrons is dynamically insignificant (at least initially),
211: hence magnetic field lines follow the flow as dictated by the
212: time-dependent induction equation
213: %
214: \begin{equation}
215: \frac{\partial {\bf A}}{\partial t}=
216: \frac{L\sigma_Tv_\phi}{4\pi r^2ce}\hat{\bf \phi}
217: +{\bf v}\times{\bf B}
218: -\eta \nabla\times{\bf B}
219: \label{A}
220: \end{equation}
221: %
222: (see also CK and Bisnovatyi-Kogan, Lovelace \& Belinski~2002).
223: Here, ${\bf A}$ is the magnetic vector potential,
224: ${\bf B}=\nabla\times {\bf A}$ is the magnetic field,
225: $e$ is the electron charge,
226: and $\eta$ is the magnetic diffusivity.
227: The Poynting-Robertson radiation force on electrons is
228: equal to $-L\sigma_Tv_\phi/(4\pi r^2 c)\hat{\phi}$.
229: Henceforth, we will work in a spherical system of coordinates
230: ($r$, $\theta$, $\phi$). We assume the disk to be Keplerian
231: ($v_{\phi} \propto v_K$) and thick (height $h \simeq r$), i.e.
232: very similar to an ADAF
233: (Narayan \& Yi 1994)\footnote{The picture one may
234: have in mind is that of small scale turbulent
235: magnetic fields building-up to equipartition values due to the
236: magneto-rotational instability. This effectively
237: thermalizes the kinetic energy of the azimuthal
238: motion on time scales not much longer than the local
239: free fall time, thus leading to a disk of
240: height $h \simeq r$, which is precisely the ADAF geometry we
241: consider.}. The remaining
242: unknown quantity relevant to our calculation is the radial
243: velocity $v_r$ which (for $h \simeq r$) is given by
244: %
245: \begin{equation}
246: v_r = -\alpha v_K \ ,
247: \end{equation}
248: %
249: where $\alpha$ is the usual accretion disk parameter.
250:
251:
252: In an $\alpha-$disk of the type considered here,
253: it is natural to assume that the flow turbulence that is responsible
254: for the viscosity required for accretion is also responsible for
255: other dissipative processes, namely the anomalous plasma resistivity
256: or equivalently anomalous magnetic diffusivity (see e.g. Heitsch
257: \& Zweibel 2003).
258: This can be expressed through the introduction of
259: the magnetic Prandtl number ${\cal P}_m$ in the expression for the
260: anomalous magnetic diffusivity
261: %
262: \begin{equation}
263: \eta \sim {\cal P}_m r|v_r|
264: \label{eta}
265: \end{equation}
266: %
267: \cite{RRS96}. In what follows, ${\cal P}_m$ is taken to be
268: a free parameter. Once ${\bf v}$, $L$ and $\eta$ are specified,
269: eq.~(\ref{A}) can be directly integrated to yield the magnetic
270: field as a function of time and position in the interior of the
271: accretion flow. Note that the disk's differential
272: rotation will continually wind up the field in the azimuthal direction.
273:
274:
275:
276: The Poynting-Robertson source of poloidal electric current
277: extends up to the surface of the disk, and hence
278: magnetic field loops emerge into a magnetosphere
279: above and below the disk
280: (note that the Poynting-Robertson effect is of course at work
281: everywhere, only its effectiveness drops quickly with
282: distance from the central luminosity source).
283: The physical conditions
284: in the magnetosphere are assumed to be very different in that
285: the dominant dynamical factor is the magnetic field and not
286: the plasma inertia, as is the case in the disk
287: interior\footnote{We assume
288: the simplest possible configuration, one with minimal
289: magnetospheric plasma loading. We also neglect the corona
290: region above and below the disk which is believed
291: to be the origin of disk wind outflows. The study of the
292: plasma dynamics in that region is outside
293: the scope of our present investigation, and we do refer the
294: interested reader to the relevant literature (e.g. Krasnopolsky,
295: Li \& Blandford 2003).}.
296: Under the magnetospheric ideal MHD conditions,
297: electric currents develop that quickly establish
298: a force-free magnetic field equilibrium.
299: In particular, the azimuthal winding of the field described above
300: will generate electric currents that will flow along the
301: magnetic loops and will tend to open them up. It has been
302: shown in the solar--photosphere literature
303: \cite{AJJ84} that there exists a critical amount
304: of winding in the azimuthal direction, beyond which
305: each loop will quickly break into two disconnected open parts
306: with their footpoints on the rotating disk.
307:
308:
309: In reality, the disk and the magnetosphere
310: evolve at the same time, however the evolution speed in the
311: magnetosphere (the Alfven speed $v_A\sim c$) is much higher than
312: the evolution speeds within the disk ($v_K, v_r << v_A$).
313: We decided not to follow the full evolution of these loops
314: in the magnetosphere and consider instead a sequence of
315: force-free MHD equilibria above and below
316: the disk, characterized by
317: %
318: \begin{equation}
319: (\nabla\times{\bf B})\times {\bf B}=0\ .
320: \label{FF}
321: \end{equation}
322: %
323: We would like to acknowledge here that force-free is
324: an approximation of the physical conditions in the
325: magnetosphere above the disk corona.
326: Under conditions of axisymmetry,
327: eq.~(\ref{FF}) can be rewritten in spherical
328: coordinates as
329: %
330: \begin{equation}
331: \frac{\partial^2\Psi}{\partial r^2}
332: -\frac{1}{r^2\tan\theta}\frac{\partial\Psi}{\partial \theta}
333: +\frac{1}{r^2}\frac{\partial^2\Psi}{\partial \theta^2}=
334: -I\frac{{\rm d}I}{{\rm d}\Psi}\ ,
335: \label{FF1}
336: \end{equation}
337: %
338: We have introduced here the magnetic flux function
339: $\Psi \equiv r\sin\theta A_\phi$ and the poloidal electric current
340: distribution $I=I(\Psi)$. The three field components are then
341: given by
342: %
343: \begin{eqnarray}
344: B_r & = & \frac{1}{r^2 \sin\theta}
345: \frac{\partial\Psi}{\partial\theta} \ , \\
346: B_\theta & = & -\frac{1}{r \sin\theta}
347: \frac{\partial\Psi}{\partial r} \ , \\
348: B_\phi & = & \frac{I}{r \sin\theta}\ .
349: \end{eqnarray}
350: %
351: Eq.~(\ref{FF1}) is an elliptic
352: equation with well-defined boundary conditions
353: (the field distribution $\Psi(r,\theta)|_{\rm surface}$
354: on the surface of the disk) that allows us to determine
355: the magnetic field configuration in the disk's atmosphere.
356:
357:
358: As we argued above, the magnetic field loops get wound up
359: by the disk's differential rotation
360: up to a critical amount of winding, where all loops
361: break into two disconnected open parts. Under
362: the ideal force-free MHD conditions that we have
363: assumed for the disk magnetosphere, this critical
364: configuration contains a certain distribution $I(\Psi)$ of
365: poloidal electric current along the magnetic field.
366: It is easy to check that $I(\Psi)$ is a monotonically
367: growing function of $\Psi$, since in that case the
368: Lorentz force $(\nabla\times {\bf B})\times B_\phi\hat{\phi}$
369: acts to open up the magnetic loop.
370: The exact functional form depends on the detailed
371: disk-magnetosphere interaction. We decided to choose the
372: following simple form
373: %
374: \begin{equation}
375: I(\Psi)=-\lambda \Psi(2-\Psi/\Psi_{\rm max})\ ,
376: \label{I}
377: \end{equation}
378: %
379: known to us from a different problem of
380: magnetospheric field line opening, the problem
381: of the axisymmetric pulsar magnetosphere (Contopoulos, Kazanas
382: \& Fendt 1999).
383: The constant of proportionality $\lambda$ is determined by
384: the requirement that all magnetospheric field lines
385: open up to infinity because of the disk differential rotation.
386: $\Psi_{\rm max}$ is the maximum value of the magnetic
387: flux function on the surface of the disk.\footnote{
388: The magnetic field is generated in the disk interior,
389: and the total amount of magnetic flux crossing the surface
390: of the disk is zero, hence $\Psi(r,\theta)|_{\rm surface}$ has a maximum value
391: $\Psi_{\rm max}$ at a certain distance,
392: and decreases again to zero farther out.}
393: Our choice simplifies considerably the numerical calculation
394: but it is obviously not unique. As we shall see, such a distribution
395: of poloidal electric current does indeed lead to a largely open
396: magnetospheric configuration.
397:
398: Inside the disk, the azimuthal field winding is limited by field
399: diffusion as described by the poloidal component of eq.~1. One
400: may easily check that, in the region outside $r_{\rm in}$ where
401: $\eta$ is of the same order as $rv_r$, the azimuthal field is
402: limited to a value of the order of $v_K/v_r$ times its poloidal
403: value. Outside the disk, the azimuthal field winding is limited by
404: the opening-up of the coronal loops to a value of the same order as its
405: poloidal value (e.g. Aly~1984). We thus expect a small discontinuity
406: in the azimuthal field in a transition region across the surface of the
407: disk.
408:
409:
410: The problem is now fully specified and the numerical simulations
411: proceed as follows:
412: \begin{enumerate}
413: \item At each timestep, we evolve the magnetic field in the
414: interior of the disk through eq.~(\ref{A}).
415: \item We obtain the
416: distribution of $\Psi(r,\theta)|_{\rm surface}$.
417: \item This distribution is used as a boundary condition for the calculation
418: of a force-free ideal MHD magnetospheric equilibrium
419: via eq.~(\ref{FF1}).
420: %, with the poloidal magnetospheric
421: %electric current distribution as given in Eq.~\ref{I}.
422: \item The above sequence of steps is repeated for each subsequent timestep.
423: \end{enumerate}
424:
425:
426: Before we proceed with the results of our numerical integrations, we would like
427: to note that the magnetic-field configurations in the disk
428: and magnetosphere are closely coupled. It is obvious that the field
429: generated in the disk is the source of the magnetospheric
430: field. What is not obvious though it that, in the
431: presence of magnetic field diffusion, the magnetospheric
432: field also acts back on the disk's magnetic field:
433: the opening up of the magnetospheric
434: field generates field bending and
435: field tension on the surface of the disk, and as a result,
436: the field diffuses into the interior of the disk, in the
437: direction that will release the magnetic tension.
438: This back--reaction of the magnetosphere onto the disk's magnetic
439: field has not been taken into account by
440: Bisnovatyi-Kogan, Lovelace \& Belinski~(2002)
441: in their criticism of our original paper.
442: This effect is a central element in our cosmic battery mechanism.
443:
444:
445: \section{Numerical Integration}
446:
447:
448: We will work in renormalized variables where distances, times,
449: and magnetic fields are rescaled, respectively, to the inner radius $r_{\rm in}$,
450: the dynamical time $t_{\rm in}\equiv
451: r_{\rm in}/v_{K,{\rm in}}$ at the inner radius,
452: and the magnetic field $B_{P-R}\equiv t_{\rm in}\times
453: L\sigma_T v_{K,{\rm in}}/(4\pi cer_{\rm in}^3)$ that is generated
454: at the inner radius within a dynamical time, viz.,
455: %
456: \begin{equation}
457: r/r_{\rm in}\rightarrow r\ , \ \
458: t/t_{\rm in}\rightarrow t\ , \ \
459: B/B_{P-R}\rightarrow B\ .
460: \end{equation}
461:
462:
463: As was discussed above, our simulation setup allows us to study
464: any accretion disk model in conjunction with
465: any magnetic diffusivity distribution in the disk.
466: The main elements of the cosmic battery
467: mechanism may, however, be demonstrated clearly
468: by using the following simple flow pattern:
469: %
470: \[
471: v_\phi = \left\{ \begin{array}{ll}
472: r^3 & \mbox{$r<1$, \ $60^{\circ}<\theta<120^{\circ}$} \\
473: r^{-1/2} & \mbox{$r\geq1$, \ $60^{\circ}<\theta<120^{\circ}$} \\
474: 0 & \mbox{otherwise}
475: \end{array}
476: \right. \]
477: \[
478: v_r = -0.1 v_\phi
479: \]
480: \[
481: v_\theta = 0.
482: \]
483: We emphasize once again that the flow pattern
484: dictated by the specific accretion disk model
485: is independent of the Poynting-Robertson generated
486: magnetic field, as long as the latter remains below
487: equipartition values. This allows us to decouple the
488: magnetic field evolution from the flow kinematics.
489:
490:
491: The anomalous magnetic diffusivity $\eta$ is calculated
492: from eq.~(\ref{eta}) at all distances. We assume that
493: interior to $r = 1$ the accretion disk terminates and
494: the velocity profiles are different from those for $r>1$.
495: The behavior given above corresponds qualitatively to
496: settling on a stellar surface, although one could also produce
497: the appropriate behavior for accretion onto a black hole.
498: As the flow rotation decreases to match that of
499: the underlying object, the shear decreases, and
500: the flow becomes laminar (in the case of a black hole,
501: the flow again becomes laminar interior three Schwarzschild
502: radii because matter is in free fall there). Laminar
503: flow implies the decay of the MHD turbulence that led
504: to the anomalous disk magnetic diffusivity, and
505: therefore we recover perfect field-plasma
506: coupling, i.e. $\eta\approx 0$ inside $r<1$.
507: This turns out to be important in obtaining the increase in
508: magnetic flux found in CK.
509:
510:
511:
512:
513: The poloidal magnetic field evolution in the disk is determined by
514: the $\phi$--component of eq.~(\ref{A}) written in dimensionless
515: form as
516: \begin{equation}
517: \frac{\partial \Psi}{\partial t} = \frac{v_\phi}{r}
518: +rv_r B_\theta -{\cal P}_m rv_r \left[
519: \frac{\partial}{\partial r}(rB_\theta)-
520: \frac{\partial B_r}{\partial\theta}\right]\ ,
521: \end{equation}
522: %
523: %\begin{eqnarray}
524: %\frac{\partial A_r}{\partial t}
525: %& = & -v_\phi B_\theta-{\cal P}_m \frac{v_r}{\sin\theta}
526: %\frac{\partial}{\partial\theta}(\sin\theta B_\phi) \\
527: %\frac{\partial A_\theta}{\partial t}
528: %& = & -v_r B_\phi + v_\phi B_r
529: %+{\cal P}_m v_r
530: %\frac{\partial}{\partial r}(r B_\phi) \\
531: %\frac{\partial A_\phi}{\partial t} = \frac{v_\phi}{r^2}
532: %+v_r B_\theta -{\cal P}_m v_r \left[
533: %\frac{\partial}{\partial r}(rB_\theta)-
534: %\frac{\partial B_r}{\partial\theta}\right]\ .
535: %\end{eqnarray}
536: %
537: which was integrated for 100 dynamical times. In Fig.~\ref{fig1},
538: we plot the resulting poloidal magnetic field configurations
539: for various values of the magnetic Prandtl number ${\cal P}_m$ assumed constant
540: throughout the disk.
541: The proportionality constant in the expression for
542: the magnetospheric poloidal electric current (eq.~[\ref{I}]) is set equal to
543: $\lambda=2.5$ because
544: we found that, for smaller values, a large fraction of closed
545: magnetic field loops remains in the magnetosphere.
546: For small values of the magnetic Prandtl number, the magnetic
547: field near the inner edge of the disk
548: grows for about one accretion time ($\sim \alpha^{-1}$ in dimensionless
549: time units), and reaches an asymptotic
550: value of the order of $\alpha^{-1}$ (Fig.~\ref{fig2}).
551: The total magnetic flux accumulated around the
552: center is also of the same order.
553: The field O-point in the disk is
554: displaced from its original position around $r=1$ to near
555: the inner radial boundary of our computational domain.
556: In the high diffusivity limit,
557: magnetic-field diffusion wins over accretion,
558: the field O-point moves outside $r=1$, and the magnetic field
559: generated by the Poynting-Robertson source eventually saturates
560: in this case as well.
561: %The . In all
562: %cases, a steady-state accretion configuration is established
563: %with zero field growth.
564:
565:
566: These results are not in contradiction with those of CK because
567: CK assumed a free inner boundary for the flow that can
568: only absorb the incident $B_z-$flux. To simulate this situation,
569: we obtained a second set of solutions in which we set ${\cal P}_m=0$
570: inside $r=1$, a reasonable condition, as was discussed above.
571: Our results are shown in Figs.~\ref{fig3} and \ref{fig4}.
572: For low values of the Prandtl number, ${\cal P}_m\ll 1$,
573: the field evolution is similar to the former case, namely the
574: field saturates within a few accretion times. However,
575: beyond a critical value of ${\cal P}_m\sim 1$, we obtain
576: a very different behavior, with new magnetic field loops centered
577: around $r=1$ continuously being generated. These loops open up because of
578: the disk's differential rotation and accretion
579: carries inward magnetic flux of one polarity, whereas diffusion acts to remove
580: to large distances flux of the opposite polarity. The high
581: conductivity at $r \le 1$ can then hold any magnetic flux
582: that enters this region. It is interesting that
583: the growth of $B_z$ inside $r=1$ may also
584: contribute to the suppression of MHD turbulence
585: in that region, as seen in recent numerical simulations
586: (Camenzind, personal communication).
587: We thus recover the effect discovered by CK,
588: namely that for values of the magnetic diffusivity higher
589: than a critical value ${\cal P}_m\sim 1$,
590: the accumulated magnetic field grows without limit.
591:
592:
593: Finally, in order to demonstrate that the opening up
594: of magnetic field lines is of central importance in this mechanism,
595: we obtained one more set of solutions in which we set $\lambda = 0$,
596: i.e. we ignored the magnetospheric field twisting because of the
597: disk's differential rotation. In that case (Figs.~\ref{fig5} and
598: \ref{fig6}), magnetic-field lines in the magnetosphere
599: remain closed and the magnetic field again saturates to a finite value
600: well below equipartition.
601:
602:
603:
604:
605: \section{Discussion and Conclusions}
606:
607:
608: We presented above a two-dimensional analysis of the cosmic battery mechanism first
609: proposed in CK, paying
610: particular attention to the criticism of Bisnovatyi-Kogan, Lovelace \& Belinski~(2002);
611: while still simplified, our
612: present treatment represents a significant improvement over our earlier work, mainly
613: because of its two-dimensional character which is essential in capturing the structure
614: and evolution of the magnetic field on the poloidal plane.
615:
616:
617: The two-dimensional models confirm our previous results. When we use appropriate
618: values of the plasma conductivity
619: and the proper magnetospheric boundary conditions, these models produce
620: configurations in which the magnetic flux
621: interior to the accretion disk's inner edge $r_{\rm in}$ increases linearly with time.
622: At the same time, the present
623: results complement our previous investigation and delineate conditions under
624: which such field growth is possible. We have thus found that, for steady magnetic-field
625: growth near the central object of an accretion disk, we need large diffusivity in the disk
626: and a region
627: near the central object in which the diffusivity practically drops to zero.
628: The zero-diffusivity condition in the central region proves to be necessary because then
629: the magnetic flux that crosses into radii $r < r_{\rm in}$ cannot diffuse back out
630: and becomes effectively trapped near the central object
631: (a finite diffusivity in the central region would eventually allow for the flux to leak out to
632: $r>r_{\rm in}$, saturating thus the field growth, as was argued in the Appendix of
633: Bisnovatyi-Kogan, Lovelace \& Belinski~[2002]).
634: We believe that the zero-diffusivity condition is
635: reasonable in the central region where the accretion disk does not extend
636: and the associated dissipative processes are thus not operative.
637: %Then, our present computations clarify the reasons for disagreement between
638: %the previous two treatments which has now been traced to (a) the inner
639: %edge boundary condition; and (b) the opening up of the magnetospheric field
640: %loops due to the disk's differential rotation.
641:
642: In our present simplified picture, we have ignored the transition region
643: between the conducting interior
644: (where $\eta$ is effectively zero and the field is greatly
645: amplified from a field a fraction of a Gauss to a field six or more orders of
646: magnitude higher) and the diffusive exterior. In reality, any small amount
647: of diffusivity present will
648: `smooth out' the abrupt field transition. Magnetic flux will continue
649: to accumulate as long as the thickness of the transition region remains
650: smaller than $r_{\rm in}$. Were the thickness of the transition region to
651: become of the same order as $r_{\rm in}$, the mechanism would then saturate,
652: and the field would diffuse outward accross $r_{\rm in}$ at the same rate
653: that it is brought in. If we take into account the Spitzer collisional diffusivity
654: estimated as $\eta_{\rm Spitzer}\sim 10^3 T_6^{1.5}\mbox{cm}^2\mbox{s}^{-1}$, where
655: $T_n$ is the plasma temperature in units of $10^n$~K (e.g.~Zombeck~1992),
656: the field growth would saturate when the ratio $B(r<r_{\rm in})/B(r>r_{\rm in})$
657: reaches a value of the order of $v_r r_{\rm in}/\eta_{\rm Spitzer}> 10^{12}$.
658: We conclude that the Spitzer resistivity does not place any practical limit to the
659: growth of the field interior to $r_{\rm in}$.
660:
661:
662:
663: Finally, we conclude that this battery
664: mechanism could be important in accounting for the observed magnetic fields in
665: several astrophysical sites as outlined in CK. Around a black hole in particular,
666: our mechanism generates a magnetic field of the right large--scale dipolar
667: topology required for the Blandford-Znajek mechanism of electromagnetic energy
668: extraction from a spinning black hole to work (Blandford \& Znajek~[1977];
669: Blandford~[2002]).
670: In this respect we would like to point the reader's attention to the
671: intriguing possibility that these considerations may in fact apply to
672: understanding the low frequency cyclic variability observed in the
673: galactic microquasar GRS 1915+105 (e.g. Pooley \& Fender [1998]; Ueda {\em et al.}
674: [2002]). A distinguishing characteristic of this source
675: are its X-ray/IR/radio flares that repeat over time scales of $\sim 20-40$ minutes and
676: lead to outflows with $\Gamma \simeq 2-3$. It is interesting to note
677: that if magnetic fields of order $B_z \simeq 0.1$ G generated by the mechanism
678: proposed herein in one accretion time of $\sim$0.01 sec
679: (in a model with $M=10M_\odot$, $L=0.1 L_{\rm Eddington}$, and $\alpha=0.1$)
680: were to increase linearly
681: with time as envisioned in this work, then after a period of $\sim$1000 seconds
682: they would achieve values which could easily account for the observed flares and outflows.
683:
684:
685:
686:
687:
688: %There is observational evidence that such a magnetic field topology
689: %also exists in the center of our Galaxy (e.g. Yusef-Zadeh~1989; Morris~1990;
690: %Morris, Uchida \& Do~2006).
691: %It is worth pointing out that while we have focused our attention to the
692: %value of the magnetic flux interior to $r_{\rm in}$, an equal amount of
693: %magnetic flux of the opposite polarity escapes to infinity. While the
694: %associated magnetic field is small, the flux is nonetheless significant and is
695: %increasing linearly with time, just as that concentrated interior to
696: %the disk's inner edge does; this flux will permeate the interstellar space with
697: %consequences that are beyond the scope of the present work. However,
698: %assuming that the black hole at the galactic center was active near
699: %its Eddington level ($\sim 10^{44}$ erg/s) for $\sim 10^8$ years, it
700: %would release a flux of order $\sim 10^{36}$ G cm$^2$, a value not very different
701: %from that of our Galaxy. Clearly, these estimates are approximate and have to
702: %be viewed with caution but they are indicative of the potential significance
703: %of this process.
704:
705:
706: DK would like to acknowledge support by INTEGRAL and HST GO grants.
707:
708:
709: \begin{thebibliography}{}
710: \bibitem[Aly 1984]{AJJ84}
711: Aly, J. J. 1984, ApJ, 283, 394
712: \bibitem[Bisnovatyi-Kogan, Lovelace \& Belinski 2002]{B-KLB02}
713: Bisnovatyi-Kogan, G. S., Lovelace, R. V. E. \& Belinski, V. A. 2002,
714: ApJ, 580, 380
715: \bibitem[Blandford 2002]{B02}
716: Blandford, R. D. 2002, in Proceedins of the MPA/ESO, 381
717: \bibitem[Blandford \& Znajek 1977]{BZ77}
718: Blandford, R. D. \& Znajek, R. L. 1977, MNRAS, 179, 433
719: \bibitem[Blinnikov \& Bisnovatyi-Kogan 1977]{BB-K77}
720: Blinnikov, S. I. \& Bisnovatyi-Kogan, G. S. 1977, A\& A, 59, 111
721: \bibitem[Contopoulos \& Kazanas 1998]{CK}
722: Contopoulos, I. \& Kazanas, D. 1998, ApJ, 508, 859 (CK)
723: \bibitem[Contopoulos, Kazanas \& Fendt (1999)]{CCKF99}
724: Contopoulos, I., Kazanas, D. \& Fendt, C. 1999, ApJ, 511, 351
725: %\bibitem[Han \& Wielebinski 2002]{HW02}
726: %Han, J.-L. \& Wielebinski, R. 2002,
727: %Chinese Journal of Astronomy \& Astrophysics, Vol. 2, p. 293-324,
728: %arXiv:astro-ph/0209090
729: \bibitem[Heitch \& Zweibel 2003]{HZ03}
730: Heitch, F. \& Zweibel E. G. 2003, ApJ, 583, 229
731: \bibitem[Krasnopolsky, Li \& Blandford 2003]{KLB03}
732: Krasnopolsky, R., Li, Z-Y \& Blandford, R. D. 2003, ApJ, 595, 631
733: \bibitem[Kulsrud {\em et al.} 1997]{KCOR97}
734: Kulsrud, R. M., Cen, R., Ostriker, J. P. \& Ryu, D. 1997, ApJ, 480, 481
735: %\bibitem[Morris 1990]{M90}
736: %Morris, M. 1990, in IAU Symp. 140, Galactic and Intergalactic Magnetic
737: %Fields, ed. R. Beck, P. P. Kronberg \& R. Wielebinski
738: %(Dordrecht: Kluwer), 361
739: \bibitem[Narayan \& Yi 1994]{NY94}
740: Narayan, R. \& Yi, I. 1994, ApJ, 428L, 13
741: \bibitem[Pooley, Fender 1998]{PF98}
742: Pooley, G. G. \& Fender, R. P. 1998, in IAU Col. 164, Radio Emission from
743: Galactic and Extraglactic Compact Sources, eds. J. A. Zensus,
744: G. B. Taylor \& J. M. Wrobel, ASP Conference Series, 144
745: \bibitem[Reyes-Ruiz \& Stepinski 1996]{RRS96}
746: Reyes-Ruiz, M. \& Stepinski, T. F., 1996, ApJ, 459, 653
747: \bibitem[Ueda et al 2002]{U02}
748: Ueda, Y. {\em et al.} 2002, ApJ, 571, 918
749: \bibitem[Zombeck 1992]{Z92}
750: Zombeck, M. V. 1992, Handbook of Space Astronomy and Astrophysics
751: (2nd ed.; Cambridge: Cambridge Univ. Press)
752: %\bibitem[Yusef-Zadeh 1989]{YZ89}
753: %Yusef-Zadeh, F. 1989, in IAU Symp. 136, The Center of the Galaxy,
754: %ed. M. Morris (Dordrecht: Kluwer), 243
755: \end{thebibliography}
756:
757:
758:
759:
760:
761:
762:
763:
764:
765: \clearpage
766:
767:
768:
769:
770:
771: \begin{figure}
772: %\epsscale{.80}
773: %\includegraphics[angle=270,scale=.80]{f1.eps}
774: %\centerline{\includegraphics[angle=270,scale=.80]{f1.eps}}
775: \includegraphics[angle=270,scale=.80]{f1.eps}
776: \caption{We plot the poloidal magnetic field structure after
777: 100 dynamical times for various values of the Prandtl number ${\cal P}_m$.
778: $\alpha=0.1$.
779: The disk extends from $60^{\circ}<\theta<120^{\circ}$, and
780: is surrounded by a force-free atmosphere. Magnetic field lines
781: that enter the atmosphere are wound up by the disk differential
782: rotation and open up. A poloidal electric current distribution
783: as defined by Eq.~\ref{I} is established in the magnetosphere.}
784: \label{fig1}
785: \end{figure}
786:
787: \clearpage
788:
789: \begin{figure}
790: %\epsscale{.80}
791: %\includegraphics[angle=270,scale=.70]{time1.ps}
792: %\centerline{\includegraphics[angle=270,scale=.70]{f2.eps}}
793: \includegraphics[angle=270,scale=.80]{f2.eps}
794: \caption{We plot the evolution of the magnetic flux $\Psi$
795: contained within $r=1$ on the disk midplane for various
796: values of the Prandtl number (${\cal P}_m=0.001$, 0.01, 0.1, 1
797: for the solid, dashed, short-dashed and dotted line respectively).
798: In all cases, the flux accumulation saturates.}
799: \label{fig2}
800: \end{figure}
801: \begin{figure}
802: %\epsscale{.80}
803: %\includegraphics[angle=270,scale=.80]{multi2.ps}
804: %\centerline{\includegraphics[angle=270,scale=.80]{f3.eps}}
805: \includegraphics[angle=270,scale=.80]{f3.eps}
806: \caption{Same as Fig.~\ref{fig1} only now we set $\eta=0$ inside
807: $r=1$. We see here the effect first discovered in CK,
808: namely that for values of the magnetic diffusivity higher
809: than a critical value the accumulated magnetic field grows
810: without limit. }
811: \label{fig3}
812: \end{figure}
813: \begin{figure}
814: %\epsscale{.80}
815: %\includegraphics[angle=270,scale=.70]{time2.ps}
816: %\centerline{\includegraphics[angle=270,scale=.70]{f4.eps}}
817: \includegraphics[angle=270,scale=.80]{f4.eps}
818: \caption{Same as Fig.~\ref{fig2} for
819: values of the Prandtl number ${\cal P}_m=0.001$, 0.01, 0.1, 1
820: for the solid, dashed, short-dashed, and dotted
821: line respectively. For low values of the Prandtl number,
822: the mechanism saturates
823: as before. Above a critical value of the Prandtl number
824: ${\cal P}_m\approx 1$, however, we observe unlimited steady field growth.}
825: \label{fig4}
826: \end{figure}
827: \begin{figure}
828: %\epsscale{.80}
829: %\includegraphics[angle=270,scale=.80]{multi3.ps}
830: %\centerline{\includegraphics[angle=270,scale=.80]{f5.eps}}
831: \includegraphics[angle=270,scale=.80]{f5.eps}
832: \caption{Same as Fig.~\ref{fig3} only now we set $I(\Psi)=0$
833: in the magnetosphere, and the magnetic field loops do not
834: open up. The effect shown in Fig.~\ref{fig3}
835: disappears.}
836: \label{fig5}
837: \end{figure}
838: \begin{figure}
839: %\epsscale{.80}
840: %\includegraphics[angle=270,scale=.70]{time3.ps}
841: %\centerline{\includegraphics[angle=270,scale=.70]{f6.eps}}
842: \includegraphics[angle=270,scale=.80]{f6.eps}
843: \caption{Same as Fig.~\ref{fig4} for
844: values of the Prandtl number ${\cal P}_m=0.001$, 0.01, 0.1, 1
845: for the solid, dashed, short-dashed, and dotted
846: line respectively. $I(\Psi)=0$. The magnetic field again saturates.}
847: \label{fig6}
848: \end{figure}
849:
850:
851:
852:
853: %\appendix
854:
855:
856: %\section{Appendix}
857:
858:
859: \end{document}
860:
861:
862:
863:
864:
865: Note that we do not have a detailed model for the region
866: inside the inner edge of the accretion disk. In particular,
867: we do not know the mechanism by which the flow stops rotating
868: and eventually comes to a halt at the center.
869: Our simple model for the anomalous magnetic
870: diffusivity is not expected to work there, and therefore, we
871: are free to think about other possibilities. As we will see,
872: in order to recover the result of our first paper, we will
873: need to modify our choice for $\eta$ inside $r=1$, namely
874: choose $\eta=0$ as is expected if one considers
875: that the turbulent viscosity mechanism that generats
876: the accretion flow is suppressed inside $r=1$.