0809.3165/p.tex
1: \documentclass[oldversion]{aa}
2: \pdfoutput=1
3: 
4: \usepackage[intlimits]{amsmath}
5: \usepackage{txfonts}
6: \usepackage[american]{babel}
7: \usepackage{graphicx}
8: \usepackage{upgreek}
9: \usepackage{xspace}
10: 
11: 
12: % BibTeX ...
13: \usepackage{natbib}
14: \citestyle{aa}
15: \bibliographystyle{aa}
16: \def\jcop{J.~Comput.~Phys.}      % Journal of Computational Physics
17: 
18: 
19: \newcommand{\Halpha}{H$\upalpha$\xspace}
20: \newcommand{\Alfven}{Alfv\'{e}n\xspace}
21: \newcommand{\Alfvenic}{Alfv\'{e}nic\xspace}
22: \newcommand{\Rjetn}{\ensuremath{R_{\jet,\nozzle}}}
23: \newcommand{\taug}{\ensuremath{\tau_\mathrm{g}}}
24: \newcommand{\tauc}{\ensuremath{\tau_\mathrm{c}}}
25: \newcommand{\tauK}{\ensuremath{\tau_K}}
26: \newcommand{\nozzle}{\ensuremath{\mathrm{n}}}
27: \newcommand{\jet}{\ensuremath{\mathrm{jet}}}
28: \newcommand{\ee}{\ensuremath{\mathrm{e}}}
29: \newcommand{\simm}{\mathord{\sim}}
30: \newcommand{\proptoo}{\mathord{\propto}}
31: \newcommand{\Nabla}{\vec{\nabla}}
32: \newcommand{\unit}[1]{\ensuremath{\,\mathrm{#1}}}
33: \newcommand{\abs}[1]{\ensuremath{\left|#1\right|}}
34: \newcommand{\pmag}{\ensuremath{p_\text{mag}}}
35: \newcommand{\pgas}{\ensuremath{p_\text{gas}}}
36: \newcommand{\vA}{\ensuremath{v_\mathrm{A}}}
37: \newcommand{\vAn}{\ensuremath{v_\mathrm{A,\nozzle}}}
38: \newcommand{\vAr}{\ensuremath{v_{\mathrm{A}r}}}
39: \newcommand{\vArn}{\ensuremath{v_{\mathrm{A}r\mathrm{,n}}}}
40: \newcommand{\vAphi}{\ensuremath{v_{\mathrm{A}\varphi}}}
41: \newcommand{\vAphio}{\ensuremath{v_{\mathrm{A}\varphi0}}}
42: \newcommand{\vexp}{\ensuremath{v_\mathrm{e}}}
43: \newcommand{\cs}{\ensuremath{c_\mathrm{s}}}
44: \newcommand{\csn}{\ensuremath{c_\mathrm{s,n}}}
45: \newcommand{\cso}{\ensuremath{c_{\mathrm{s}0}}}
46: \newcommand{\er}{\ensuremath{\hat{\vec{e}}_r}}
47: \newcommand{\ephi}{\ensuremath{\hat{\vec{e}}_\phi}}
48: \newcommand{\evarphi}{\ensuremath{\hat{\vec{e}}_\varphi}}
49: \newcommand{\etheta}{\ensuremath{\hat{\vec{e}}_\theta}}
50: \newcommand{\evartheta}{\ensuremath{\hat{\vec{e}}_\vartheta}}
51: \newcommand{\ez}{\ensuremath{\hat{\vec{e}}_z}}
52: \newcommand{\de}{\ensuremath{\mathrm{d}}}
53: \newcommand{\const}{\ensuremath{\text{const}}}
54: \newcommand{\degree}{\ensuremath{^\circ}}
55: \newcommand{\efrate}{\ensuremath{\mathcal{E}}}
56: 
57: 
58: 
59: 
60: \begin{document}
61: 
62: \title{Kink instabilities in jets from rotating magnetic fields}
63: \author{R. Moll \and H. C. Spruit \and M. Obergaulinger}
64: \offprints{\protect\raggedright R. Moll,\\\email{rmo@mpa-garching.mpg.de}}
65: \institute{Max-Planck-Institut f\"{u}r Astrophysik, Karl-Schwarzschild-Str. 1, 85741 Garching, Germany}
66: \date{Accepted on September 15, 2008}
67: \abstract{We have performed 2.5D and 3D simulations of conical jets driven by
68: the rotation of an ordered, large-scale magnetic field in a stratified
69: atmosphere.  The simulations cover about three orders of magnitude in distance
70: to capture the centrifugal acceleration as well as the evolution past the
71: \Alfven surface.  We find that the jets develop kink instabilities, the
72: characteristics of which depend on the velocity profile imposed at the base of
73: the flow. The instabilities are especially pronounced with a rigid rotation
74: profile, which induces a shearless magnetic field. The jet's expansion appears
75: to be limiting the  growth of \Alfven mode instabilities.
76: }
77: \keywords{Magnetohydrodynamics (MHD) -- Instabilities -- ISM: jets and outflows
78: -- ISM: Herbig-Haro objects -- quasars: general -- Gamma rays: bursts}
79: \maketitle
80: 
81: 
82: 
83: 
84: 
85: 
86: \section{Introduction}
87: 
88: 
89: Strong magnetic fields on large scales may play an essential, active role in
90: the formation and evolution of jet-like outflows.  The general idea is that a
91: poloidal magnetic field, embedded in a plasma and anchored e.g. in an accretion
92: disk or a black hole, is forced into rotation at the anchor point, a toroidal
93: field develops and the plasma is accelerated by what can be interpreted as a
94: centrifugal force in a corotating frame
95: \citep{1982Blandford,1996Spruit,2000Koenigl}.  However, this magnetocentrifugal
96: acceleration is only effective up until the \Alfven surface, defined as the
97: surface where the flow velocity equals the \Alfven velocity.  Beyond this
98: point, the magnetic field will be strongly wound-up. Such a field configuration
99: is potentially unstable with respect to certain MHD instabilities.
100: 
101: MHD jets are susceptible to a variety of instabilities.  Kelvin-Helmholtz (KH)
102: instabilities are fed by the relative kinetic energy between the jet and the
103: ambient medium.  They can distort the jet surface only (ordinary modes) or the
104: whole beam \citep[e.g.][]{1991Birkinshaw}, provoking shocks, mixing with
105: ambient material and possibly a disruption of the jet
106: \citep{1995Bodo,1998Bodo}.  The presence of strong magnetic fields, poloidal or
107: toroidal, is expected to hamper the growth of KH instabilities
108: \citep{1992Appl,1999Keppens}.
109: 
110: The free energy associated with the toroidal magnetic field is responsible for
111: another class of instabilities, which is traditionally known as current-driven
112: (CD) and of notorious importance in controlled fusion devices \citep[for an
113: introduction, see, e.g.][]{1987Freidberg,1978Bateman}.  The relevance for
114: magnetized astrophysical jets has been pointed out by
115: \citet{1993Eichler,1997Spruit,1998Begelman} and others.  Among CD
116: instabilities, $m=1$ kink instabilities are the most effective.  An ideal kink
117: mode is characterized by helical displacements of the cylindrical cross
118: sections of a plasma column.  It is expected to grow on a dynamical time scale
119: with respect to an \Alfven wave crossing the unstable column. The
120: susceptibility is strongly dependent on the magnetic pitch, a measure for the
121: degree of wind-up.  Kink instabilities might destroy the ordered, symmetric
122: state of a jet, leading to its disruption or, through magnetic reconnection,
123: the associated dissipation of magnetic fields and steepening of the magnetic
124: pressure gradient, to its acceleration \citep{2002Drenkhahn,2006Giannios}.
125: 
126: Different kinds of instability can mix and interact. For example,
127: \citet{2002Baty} show how CD instabilities can stabilize KH vortices at the jet
128: surface. For this work, we used conditions under which CD kink instabilities
129: are expected to dominate (low plasma-$\beta$, small magnetic pitch).
130: 
131: For a self-consistent study of kink instabilities, numerical simulations need
132: to be carried out in 3D.  \citet{1996Lucek} did so using a simple model in
133: which a toroidal magnetic field configuration was allowed to expand into a
134: uniform atmosphere. This generated a jet which was subject to kink
135: instabilities. \citet{2004Nakamura} performed 3D simulations of MHD jets in
136: variously stratified atmospheres, finding that they can develop kink-like
137: distortions in the trans-\Alfvenic region.  Laboratory experiments of MHD jets
138: have been performed by \citet{2005Hsu}, confirming that the magnetic pitch
139: plays a crucial role for the formation of kink instabilities.
140: 
141: 
142: \subsection{Effects of jet expansion}
143: 
144: Jets from protostars, and especially AGN and microquasars, expand in width $d$
145: by orders of magnitude after passing through their \Alfven radius. In an
146: expanding flow there is no clean separation between time dependence due to
147: instability and that due to the expansion itself, making the question of
148: stability less well defined. Analytical studies thus tend to focus on
149: instabilities in a cylindrical geometry, with constant diameter (such as in the
150: ``magnetic tower'' picture of \citealt{2003Lynden}). Expansion has strong
151: consequences on the behavior of instabilities, however, compared with jets
152: modeled as cylinders of constant width.
153: 
154: First, there is the tendency for the toroidal (azimuthal, around the jet axis)
155: component of the magnetic field to dominate in an expanding jet. From the
156: induction equation, the poloidal and toroidal  components of the field vary as
157: $B_\mathrm{p}\sim 1/d^2$ and $B_\varphi\sim 1/d$ respectively (for constant
158: jet velocity). Expansion thus causes a continual increase of the ratio
159: $B_\varphi/B_\mathrm{p}$. Even when dissipation were to decrease the toroidal
160: field at some point, the ratio increases again on further expansion. Free
161: energy available in the toroidal field thus remains the dominant form of
162: magnetic energy, and one may expect the question of stability and dissipation
163: to remain relevant on all length scales. It also follows that the nonlinear
164: development of instabilities in an expending jet is expected to be very
165: different from the constant-diameter case.
166: 
167: Secondly, expansion has a strong effect even on the conditions for occurrence
168: of instability. It has a stabilizing effect, since magnetic instabilities
169: become ineffective when their signal speed (the \Alfven speed) drops below the
170: lateral expansion speed of the jet. This is discussed further below.
171: 
172: 
173: \subsection{Rationale of the calculations}
174: 
175: The aim of the calculations reported here is to study how kink instability
176: operates under these conditions of expansion of the jet over several orders of
177: magnitude in width.
178: 
179: The degree of instability to be expected in a jet driven by a rotating magnetic
180: field is intimately tied to the way it is collimated.  If, instead of being
181: cylindrical, the jet has a non-vanishing opening angle $\vartheta$, the
182: expected incidence of instability depends on the details of the
183: dependence of $\vartheta$ on distance $r$.  An opening angle increasing
184: with distance reduces instability, while for an asymptotically vanishing
185: opening angle instability must always set in at some distance, if it was not
186: present already from the start (see discussion in Sect.~\ref{sec:growthexp}).
187: 
188: The setup in the simulations presented here produces jets in the intermediate
189: case of an (approximately)  constant opening angle: a ``conical'' outflow. It
190: turns out that in this case the presence of instabilities and their amplitude
191: depends on secondary conditions such as the rotation profile imposed at the
192: base of the flow, hence it is a good test case for the incidence of
193: instabilities.
194: 
195: Since the observed jets travel over such large distances, even marginal forms
196: of instability can become effective. An important goal of the present
197: calculations is therefore to cover a large range in distance, about 3 orders of
198: magnitude. This is achieved by the use of a grid adapted to the approximately
199: conical shape of the jet.
200: 
201: 
202: 
203: 
204: \subsection{Expected instability growth in expanding jets}
205: \label{sec:growthexp}
206: 
207: In the following, we estimate how the sideways expansion affects the growth of
208: instabilities in broadening jets. In spherical coordinates
209: $(r,\vartheta,\varphi)$, we assume that the jet radius (distance to the jet's
210: central axis) is given by
211: \begin{equation}
212:     R = R' \left( \frac{r}{r'} \right)^\alpha \quad \text{with} \quad R' = r' \sin\vartheta'
213: \label{}
214: \end{equation}
215: where the prime stands for evaluation at a reference distance $r'$, for which
216: we take a distance somewhat beyond the \Alfven radius. The flow has then
217: approximately reached its asymptotic speed $v_r \approx \const$, and the
218: magnetic field has become predominantly azimuthal. In the absence of magnetic
219: dissipation due to instabilities, the field strength then varies as $B_\varphi
220: \propto R^{-1}$ (magnetic flux conservation) and the density as $\rho \propto
221: R^{-2}$ (mass conservation). Since the growth rate $\varGamma_\mathrm{g}$ is
222: expected to scale with the \Alfven crossing rate $\vAphi/R$, we introduce a
223: dimensionless instability rate $\varkappa$ of order unity:
224: \begin{equation}
225:     \varGamma_\mathrm{g} = \varkappa \frac{\vAphi}{R}
226:         = \varkappa \frac{\vAphi}{R'} \left( \frac{r}{r'} \right)^{-\alpha}
227: \label{}
228: \end{equation}
229: where $\vAphi=B_\varphi/\sqrt{4\pi \rho}$ is the azimuthal \Alfven velocity.
230: The expansion rate is estimated by
231: \begin{equation}
232:     \varGamma_\mathrm{e} = \frac{\de \ln R}{\de t}
233:                       = \frac{1}{R} \frac{\de r}{\de t} \frac{\de R}{\de r} = \frac{\alpha v_r}{r} .
234: \label{}
235: \end{equation}
236: We find
237: \begin{equation}
238:     \frac{\varGamma_\mathrm{g}}{\varGamma_\mathrm{e}} =
239:             \frac{\varkappa}{\upsilon \alpha \sin \vartheta'} \left( \frac{r}{r'} \right)^{1-\alpha}
240: \label{}
241: \end{equation}
242: with $\upsilon \coloneqq v_r/\vAphi \approx \const$ according to the ballistic
243: approximations mentioned above.  Consequently, the instability growth rate
244: dominates at some distance $r$ for a collimating jet ($\alpha<1$).
245: Decollimation ($\alpha>1$), on the other hand, thwarts the growth of
246: instabilities. A conical jet ($\alpha = 1$) constitutes a limiting case where
247: all depends on the combination of parameters $\varkappa/(\upsilon\, \sin
248: \vartheta')$, which is of order unity.  A numerical simulation is necessary to
249: find out whether the instability or expansion prevails.
250: 
251: The paper is organized as follows. In Sect.~\ref{sec:model}, we introduce the
252: magnetocentrifugal jet model and account for the assumptions made in our
253: simulations.  A detailed description of the numerical setup, the coordinate
254: system and the scale-free units employed in the analysis is given in
255: Sect.~\ref{sec:methods}.  In Sect.~\ref{sec:simulatedcases} we give the
256: parameters of the simulated cases and in Sect.~\ref{sec:results} we present the
257: results.  There, we start by making predictions on the characteristics of
258: instabilities by examining the relevant properties of our simulated jets. We
259: proceed with an analysis of the instabilities that actually appeared and
260: complete with looking for effects on the jets' dynamics. We finish with a
261: discussion and conclusions in Sect.~\ref{sec:discussion}.
262: 
263: 
264: 
265: \section{The model}
266: \label{sec:model}
267: 
268: 
269: The model is  construed to apply to jets produced by ordered magnetic fields
270: anchored in an accretion disk. This has become the default interpretation for
271: the jets observed in AGN, microquasars and protostellar objects, though it must
272: be kept in mind that observational evidence of the key ingredient in this
273: model, the ordered field \citep{1976Bisnovatyi,1982Blandford}, is still somewhat
274: indirect. 
275: 
276: More uncertain is the shape of this field. The strength of the field anchored
277: in the disk is likely to scale in some way with the orbital kinetic energy (or
278: gas pressure) in the disk, hence will decline with distance $R$ from the
279: rotation axis. In the absence of more detailed information, we consider a
280: simple form for a field of this kind, one in which the vertical (normal to the
281: disk) component at the surface $B_z$ varies as $B_z(R) \propto
282: [1+(R/z_0)^2]^{-\nu}$.  Neglecting gas pressure and fluid motions, the field
283: above the disk would be a potential field, its shape defined uniquely by $B_z$.
284: For $\nu=3/2$ it is the field of a monopole with the source at a depth $z_0$
285: below the center of the disk.
286: 
287: The initial state of the model is a gas distribution in hydrostatic equilibrium
288: in a field of this monopolar shape. Rotation is applied at the lower boundary,
289: in a region $R<R_0$ (see Sect.~\ref{sec:iandbcs} for details). This generates
290: an outflow with an approximately constant opening angle on the order $R_0/z_0$
291: (a ``conical'' outflow). The surrounding volume remains approximately in static
292: equilibrium, and serves to collimate the outflow to the desired opening angle.
293: 
294: The magnetic field responds to the rotation by winding up. That is, a toroidal
295: magnetic field $B_\varphi$ is produced and gives, together with the poloidal
296: field $B_r$, rise to helical field lines.  The magnetic pressure gradient
297: (minus the tension force) associated with $B_\varphi$ gives rise to a poloidal
298: acceleration of the plasma which is of centrifugal nature in a corotating
299: frame.  Beyond the \Alfven radius, the acceleration ceases to be effective,
300: while the field becomes predominantly azimuthal.  The further development
301: depends on how strongly the jet is affected by instabilities in this highly
302: wound-up field.  Possibly, they endanger the jet's integrity and/or facilitate
303: magnetic reconnection events. Magnetic field dissipation can entail further
304: acceleration of the jet \citep{2002Drenkhahn}.  As discussed above
305: (Sect.~\ref{sec:growthexp}), the ``conical outflow'' produced in our monopolar
306: background field is of special interest as it marks the boundary between cases
307: expected to be strongly respectively weakly unstable. 
308: 
309: A self-consistent investigation of the problem requires a full 3D treatment,
310: because kink instabilities are non-axisymmetric.  In addition to every 3D
311: simulation we also performed an axisymmetric (2.5D) simulation with the same
312: boundary and initial conditions.  This way, we could detect whether the jet
313: evolves differently due to the instabilities. 
314: 
315: The basic parameters of the model are the magnitude of the rotation velocity,
316: its profile $\varOmega(R)$, the relative strength of the magnetic field as
317: measured by plasma-$\beta$ of the initial state, and the jet's opening angle.
318: The parameter values are chosen such that the \Alfven radius of the resulting
319: outflow is located within the computational volume, so that the centrifugal
320: acceleration process is covered in the simulation, but close to the lower
321: boundary, so that the subsequent evolution can be followed over as large a
322: distance as numerically feasible. Increasing the imposed rotation rate moves
323: the \Alfven radius toward the lower boundary.  Due to numerical limitations,
324: however, $v_\varphi$ could not be increased indefinitely in the simulations,
325: and a compromise was necessary.  In the results reported below the region
326: inside the \Alfven radius occupies about 10--20\% of the box length.
327: 
328: 
329: 
330: \section{Methods}
331: \label{sec:methods}
332: 
333: 
334: We employ a spherical grid for our jet simulations.  This enables us to follow
335: jets with opening angles over a much longer distance than is possible with a
336: Cartesian grid, because the jet need not be overresolved at large heights in
337: order to properly resolve its base.  The obvious choice of letting the jet
338: propagate along the polar axis is numerically problematic if non-axisymmetric
339: flows are involved, because the grid is singular there. We therefore let the
340: jet flow in equatorial direction. The computational volume covers a range
341: $\Delta\theta=\Delta\phi$ in the polar and azimuthal angles, adjusted to
342: the opening angle of the flow.
343: 
344: 
345: \subsection{MHD equations and numerical solver}
346: \label{sec:equations}
347: 
348: We numerically solved the ideal adiabatic MHD equations, including a
349: temperature-dependent temperature-control term $K=K(T(t))$. Explicitly, the
350: equations are:
351: \begin{gather}
352:     \frac{\partial \rho}{\partial t} + \Nabla \cdot (\rho \vec{v}) = 0 , \label{eq:continuity_equation}\\
353:     \frac{\partial \vec{v}}{\partial t} + \vec{v} \cdot \Nabla \vec{v} = 
354:         - \frac{1}{\rho} \Nabla p + \frac{1}{4 \pi \rho} (\Nabla \times \vec{B} ) \times \vec{B}
355:         - \Nabla \Phi,    \label{eq:momentum_equation}\\
356:     \frac{\partial e}{\partial t} + \Nabla \cdot \left[ \left( e+p+\frac{B^2}{8\pi} \right) \vec{v}
357:         - \frac{1}{4 \pi} \vec{B}(\vec{B} \cdot \vec{v}) \right]
358:         = - \rho \vec{v} \cdot \Nabla \Phi + K,   \label{eq:energy_equation} \\
359:     \frac{\partial \vec{B}}{\partial t} = \Nabla \times (\vec{v} \times \vec{B}), \label{eq:induction_equation}
360: \end{gather}
361: where
362: \begin{equation}
363:     e = \frac{1}{2} \rho v^2 + \frac{B^2}{8\pi} + \frac{p}{\gamma - 1}
364: \label{eq:total_energy}
365: \end{equation}
366: is the total energy density, $\gamma = 5/3$ is the adiabatic index, $p$ is the
367: gas pressure, $\Phi$ is the gravitational potential (external, no self-gravity)
368: and the other symbols have their usual meanings. A notorious problem with
369: low-$\beta$ MHD simulations in fully conservative form, as in the code used
370: here, is the amplification of discretization errors that occurs because the gas
371: pressure is only a small contribution to the  total energy
372: (cf.~\ref{eq:total_energy}), which is dominated by the magnetic energy. As in
373: the case of highly supersonic flows, these errors manifest themselves in the
374: form of ``negative pressures'' at occasional grid points. This problem does not
375: occur when an equation for the thermal energy equation is used instead of the
376: total energy. We compute, in parallel, an alternative update of the gas
377: pressure from the thermal energy equation, in the form
378: \begin{equation}
379:     \frac{\partial p}{\partial t} + \Nabla \cdot ( p \vec{v} )
380:             = - (\gamma-1) p \Nabla \cdot \vec{v} + K.
381: \label{eq:intenergy_equation}
382: \end{equation}
383: Where negative pressures appear, they are replaced by this value.
384: 
385: Another device that turns out very useful to avoid negative pressures is the
386: temperature-control term $K$ in Eq.~\eqref{eq:energy_equation}. For this we use
387: a scheme loosely modeled after Newtonian cooling, or an optically thin
388: radiative loss process. After every full time step, we add/subtract thermal
389: energy according to \begin{equation} \frac{\Delta p(t)}{p(t=0)} = - \frac{T(t)
390: - T(t=0)}{T(t=0)} \frac{\Delta t}{\tauK} , \label{eq:cooling_term}
391: \end{equation} where $T$ is the temperature and $\tauK$ is a time scale chosen
392: so as to keep the temperature within about a factor 30 of the initial
393: atmospheric value.
394: 
395: With the MHD Poynting vector
396: \begin{equation}
397:     \vec{S} = - \frac{1}{4\pi} ( \vec{v} \times \vec{B} ) \times \vec{B},
398: \label{}
399: \end{equation}
400: Eq.~\eqref{eq:energy_equation} can also be written as
401: \begin{equation}
402:     \label{eq:energy_equation_alt}
403:     \frac{\partial ( e + \rho \Phi) }{\partial t}
404:        + \Nabla \cdot \left[ 
405:             \left( \frac{1}{2} \rho v^2 + \frac{\gamma}{\gamma-1}p + \rho \Phi \right) \vec{v}
406:                            + \vec{S} \right] = K ,
407: \end{equation}
408: describing the change of total energy including gravitational potential energy.
409: We will employ this form later when we look at the energy flow rates.
410: 
411: We used a newly developed Eulerian MHD code (Obergaulinger et al., in
412: preparation) to solve
413: Eqs.~(\ref{eq:continuity_equation}--\ref{eq:induction_equation}).  It is based
414: on a flux-conservative finite-volume formulation of the MHD equations and the
415: constraint transport scheme to maintain a divergence-free magnetic field
416: \citep{1998Evans}.  Using high-resolution shock capturing methods
417: \citep[e.g.,][]{1992LeVeque}, it employs various optional high-order
418: reconstruction algorithms and approximate Riemann solvers based on the
419: multi-stage method \citep{2006Toro}.  The simulations presented here were
420: performed with a fifth order monotonicity-preserving reconstruction scheme
421: \citep{1997Suresh}, together with the HLL Riemann solver \citep{1983Harten}
422: and third order Runge-Kutta time stepping.
423: 
424: 
425: \subsection{Grid coordinates}
426: 
427: 
428: \begin{figure}[t]
429: \begin{center}
430: \includegraphics[width=\linewidth]{figures/coords.pdf}
431: \end{center}
432: \caption{Computational domain and coordinate nomenclature (schematic drawing).
433: The jet propagates along the $y$-axis ($\theta=\phi=\pi/2$), near which the
434: grid is quasi-Cartesian in the normal plane. To describe the results of the
435: simulations, we use the alternate spherical coordinate system
436: $(r,\vartheta,\varphi)$ shown in the upper right inset, which takes the
437: $y$-axis as the polar axis.}
438: \label{fig:coords}
439: \end{figure}
440: 
441: In the 3D simulations, our computational domain was centered around the
442: $y$-axis in a ``standard'' spherical coordinate system $(r,\theta,\phi)$ with
443: $\theta$ being the polar angle from the $z$-axis and $\phi$ being the azimuthal
444: angle about the $z$-axis in the $x$-$y$-plane.  See Fig.~\ref{fig:coords} for
445: an illustration.  Positioning the domain in equatorial (rather than polar)
446: direction yields a grid which is free of singularities and quasi-regular in
447: transverse jet direction: Near the $y$-axis, the distance between two
448: $\phi=\const$ curves is $\Delta x \approx r\,\Delta\phi$ if we neglect terms of
449: order $(\theta-\pi/2)^2$ and of order $(\phi-\pi/2)^3$. The distance between
450: two $\theta=\const$ curves is then $\Delta z \approx r\,\Delta\theta$.  With
451: uniform spacings  $\Delta\theta$ and $\Delta\phi$, we thus obtain a grid
452: whose area elements are approximately those of an equidistant Cartesian grid in
453: a plane normal  to the $y$-axis.
454: 
455: For data analysis and plotting we use the ``alternative'' spherical coordinate
456: system $(r,\vartheta,\varphi)$ with $\vartheta$ being the polar angle from the
457: $y$-axis and $\varphi$ being the azimuthal angle about the $y$-axis in the
458: $z$-$x$-plane. It is adapted to the propagation direction of the jet and as
459: such better suited to describe its physics.  $R = r\sin\vartheta$ denotes the
460: distance to the $y$-axis.  Generally, we refer to the $r$-direction with
461: ``poloidal'' or ``radial'', the $\varphi$-direction with ``toroidal'' or
462: ``azimuthal'' and the $y$-axis as ``polar axis'' or ``central axis''.
463: 
464: In the 2.5D simulations, the jet propagates along the $z$-axis and the
465: $\phi$-direction is taken to be symmetric. However, to avoid confusion, we use
466: the same nomenclature as in the 3D simulations throughout this paper.  That is,
467: we visualize the jet as propagating in $y$-direction and employ the
468: $(r,\vartheta,\varphi)$ system for describing its physics.
469: 
470: For a proper resolution at all radii, it turned out to be necessary to employ
471: logarithmic grid spacing \citep{1996Park} in $r$-direction.  The $r$-left
472: interface of grid cell $i$ is situated at
473: \begin{equation}
474:     r_\mathrm{l}^i = r_\mathrm{l}^0 \left( \frac{r_\mathrm{r}^{n-1}}{r_\mathrm{l}^0} \right)^{i/n}
475: \end{equation}
476: where $n$ is the total number of cells in the domain which is bounded by
477: $r=r_\mathrm{l}^0$ and $r=r_\mathrm{r}^{n-1}=r_\mathrm{l}^n$. The cell center
478: of grid cell $i$ is situated at $r_\mathrm{c} =
479: (r_\mathrm{l}^i+r_\mathrm{r}^i)/2$.
480: 
481: 
482: \subsection{Initial and boundary conditions}
483: \label{sec:iandbcs}
484: 
485: The initial magnetic field is
486: \begin{equation}
487:     \vec{B}(r) = \frac{g}{r^2} \er,
488: \label{}
489: \end{equation}
490: corresponding to a magnetic monopole of charge $g$ located at the coordinate
491: origin.  The associated vector potential
492: \begin{equation}
493:     \vec{A} = \frac{g}{r} \tan \left( \frac{\theta}{2} \right) \ephi
494: \label{}
495: \end{equation}
496: was employed in the numerical setup to ensure solenoidality of the discretized
497: magnetic field.  Satisfying $\Nabla \times \vec{B}=0$, the initial magnetic
498: field is force-free.
499: 
500: We impose the static gravitational field of a point mass $M$, also located at
501: the origin. The stratification of gas pressure in this potential is chosen such
502: that the plasma-$\beta \coloneqq p / \pmag$ in the initital state is constant
503: throughout the computational domain. Hence, since $B \propto r^{-2}$, $p
504: \propto r^{-4}$.  The density  in the initial state is determined from
505: hydrostatic  equilibrium: $\rho GM = - r^2 \de p / \de r \propto r^{-3}$ where
506: $G$ is the gravitational constant. The
507: temperature, sound speed and \Alfven velocity vary as $T \propto r^{-1}$, $\cs
508: \propto r^{-1/2}$ and $\vA \propto r^{-1/2}$ in this stratification.
509: 
510: The lower boundary of the computational volume, where the jet's ``nozzle''
511: resides, is at a distance $r_\nozzle$ from the origin. The conditions at this
512: surface are related to the gravitational potential by 
513: \begin{equation}
514:     \Phi(r) = -\frac{GM}{r} \quad \text{with} \quad
515:         M = \frac{4 p_\nozzle r_\nozzle}{G \rho_\nozzle} ,
516: \label{}
517: \end{equation}
518: where the subscript $\nozzle$ (for ``nozzle'') denotes the values at $r_\nozzle$.
519: 
520: At the sides ($\theta$ and $\phi$) and top (upper $r$) of the domain, we use
521: open boundaries which allow for an almost force-free outflow of material: $p$,
522: $\rho$, all components of $\vec{v}$ and the transverse components of $\vec{B}$
523: are mirrored across the boundary interface to the opposing ``ghost cells'', the
524: normal component of $\vec{B}$ is determined by the solenoidality condition.
525: The open boundaries work well and cause only minimal artefacts in the form of
526: reflections.  At the bottom (lower $r$) of the domain, where the jet emanates,
527: $p$ and $\rho$ are kept fixed at their initial values in all ghost cells.  The
528: magnetic field is determined by extrapolation from the interior of the domain
529: using the same scheme as for open boundaries.  The velocity is prescribed to be
530: zero except for the ghost cells below the nozzle area ($R \le \Rjetn$ at
531: $r=r_\nozzle$). There, an azimuthal velocity field $\vec{v}=v_\varphi \evarphi$
532: is maintained, with either a Keplerian velocity profile
533: \begin{equation}
534:     v_\varphi = \begin{cases} v_{\varphi,\nozzle}^\text{max} \sqrt{\frac{0.2\Rjetn}{R}}
535:         & \text{for } 0.2 \Rjetn \le R \le \Rjetn \\
536:         0
537:         & \text{elsewhere}
538:         \end{cases}
539: \label{eq:Keplerian}
540: \end{equation}
541: or a rigid rotation profile
542: \begin{equation}
543:     v_\varphi = \begin{cases} v_{\varphi,\nozzle}^\text{max} \frac{R}{\Rjetn}
544:         & \text{for } R \le \Rjetn \\
545:         0 & \text{elsewhere} .
546:         \end{cases}
547: \label{eq:rigid}
548: \end{equation}
549: Note that we use the term ``Keplerian'' to indicate only that $v_\varphi
550: \propto R^{-1/2}$. The central mass $M$ only serves to balance our chosen
551: stratification and is not to be understood as the center of an accretion disk.
552: 
553: 
554: 
555: 
556: 
557: \subsection{Units}
558: 
559: 
560: \begin{table}
561: \caption{Normalization units}
562: \centering
563: \begin{tabular}{ccc}
564: \hline\hline
565: Quantity & Symbol(s) & Unit \\
566: \hline
567:     length          & $x$,$y$,$z$,$r$,$R$,$h$   & $l_0$ \\
568:     gas pressure    & $p$                   & $p_0$ \\
569:     density         & $\rho$                & $\rho_0$ \\
570:     velocity        & $v$                   & $\cso = \sqrt{ \gamma p_0 / \rho_0 }$ \\
571:     time            & $t$,$\tau$            & $t_0 = l_0 / \cso$ \\
572:     energy density  & $e$                   & $p_0$ \\
573:     energy flow rate& $\efrate$             & $p_0 l_0^3 / t_0$ \\
574:     force density   & $F$                   & $p_0 / l_0$ \\
575:     magnetic flux density       & $B$       & $B_0 = \sqrt{8 \pi p_0} $ \\
576:     current density & $j$                   & $j_0 = B_0 c / (4\pi l_0)$ \\
577: \hline
578: \end{tabular}
579: \label{tab:units}
580: \end{table}
581: 
582: The setup described above is unambiguously determined by 6 parameters,
583: $B_\nozzle$,  $p_\nozzle$, $\rho_\nozzle$, $R_{\jet,\nozzle}$, $r_\nozzle$ and
584: $v_{\varphi,\nozzle}^\text{max}$, but they are not all independent. As units of
585: length, pressure and density we use $l_0 \equiv 2 R_{\jet,\nozzle}$, $p_0
586: \equiv p_\nozzle$ and $\rho_0 \equiv \rho_\nozzle$. The physical quantities
587: expressed in these units are  listed in Table~\ref{tab:units}. Since these
588: units are arbitrary, the number of independent parameters defining the problem
589: reduces to $6-3=3$.  These are a plasma-$\beta$ value (which determines
590: $B_\nozzle$), an opening angle $\vartheta_\mathrm{o} \coloneqq \arcsin ( l_0 /
591: 2 r_\nozzle)$, and a Mach number for the rotation: either
592: $v_{\varphi,\nozzle}^\text{max} / \csn$ or $v_{\varphi,\nozzle}^\text{max} /
593: \vAn$. 
594: 
595: The sound speed, \Alfven velocity and escape velocity at $r=r_\nozzle$,
596: expressed in the normalization units, are $\csn = \cso$,
597: \begin{align}
598:     \vAn &= \sqrt{\frac{2}{\gamma}} \frac{B_\nozzle}{B_0} \cso
599:                         \approx 1.1 \frac{B_\nozzle}{B_0} \cso \\
600:  \text{and} \quad   v_{\text{esc},\nozzle} &= \sqrt{\frac{8}{\gamma}} \cso \approx 2.2 \cso .
601: \end{align}
602: 
603: For the sake of clarity, we usually omit the normalization unit.  For example,
604: $v=5$ would denote a velocity of $5\cso$, which is sonic Mach $5$ at the jet
605: nozzle.
606: 
607: 
608: \section{Cases studied}
609: \label{sec:simulatedcases}
610: 
611: In the following we present the results of two 3D simulations for two different
612: rotation profiles imposed at the nozzle, the Keplerian and rigid rotation
613: profiles given by  (\ref{eq:Keplerian}, case K3) and (\ref{eq:rigid}, case R3).
614: These are compared with two 2.5D simulations with the same initial and boundary
615: conditions (cases K2 and R2).
616: 
617: In all cases, the initial state has a constant plasma-$\beta$ of $1/9$
618: ($B_\nozzle=3$), the maximum rotation velocity at the boundary is
619: $v^\text{max}_{\varphi,\nozzle} = 0.33 \csn = 0.1 \vArn$ and the initial (half)
620: opening angle is $\vartheta_\mathrm{o} = 5.7\degree$ ($r_\nozzle = 5$).  This
621: choice of parameters yields a jet with a magnetic pitch low enough to be
622: unstable to kinks. At the same time, it avoids numerical problems found to
623: arise with higher (supersonic) rotation velocities as a boundary condition.
624: The ``grid noise'' in the 3D simulations (the grid is not axisymmetric in the
625: rotation direction $\varphi$) turned out to be sufficient to excite
626: instabilities, so we did not need to apply a perturbation by hand. For the
627: temperature-control term, we used $\tauK=2$.
628: 
629: In the 3D simulations, we used 384 logarithmically spaced grid cells in radial
630: direction and 96 uniformly spaced grid cells in each of the two angular
631: directions.  The physical extent of the simulated domain was $500$ in the
632: radial direction and $33.8\degree$ in each angular direction.  The ratio
633: between the maximum and minimum $r$ is $505/5 = 101$. The 2.5D simulations were
634: performed with the same resolution in the radial direction and $64$ grid cells
635: in the evolved angular direction which had an extent of $16.9\degree$. The 3D
636: simulations each ran for about one week (wall clock time) on 64 processors with
637: MPI parallelization.
638: 
639: 
640: 
641: 
642: 
643: \section{Results}
644: \label{sec:results}
645: 
646: \begin{figure}[t]
647: \begin{center}
648: \includegraphics[width=\linewidth]{figures/thetab_KR23.pdf}
649: \end{center}
650: \caption{Jet border, defined as the polar angle within which a given percentage
651: of energy flows (see Eq.~\ref{eq:efrate} for the definition of
652: $\efrate_\text{tot}$), as a function of distance.  In all cases, most of the
653: energy flow is contained within the
654: $\vartheta\approx\vartheta_\mathrm{o}=5.7\degree$ surface, where
655: $\vartheta_\mathrm{o}$ is the initial opening angle.  A minor but increasing
656: amount of energy flows outside this angle. The energy density in the jet
657: (especially the toroidal field) causes it to decollimate somewhat compared with
658: the conical configuration in which it is embedded.}
659: \label{fig:border}
660: \end{figure}
661: 
662: \begin{figure}[t]
663: \begin{center}
664: \includegraphics[width=\linewidth]{figures/flines_K3.jpg}
665: \end{center}
666: \caption{Selected magnetic field lines in the 3D simulation with a Keplerian
667: velocity profile. The right-hand plot shows the entire domain ($r=5\ldots505$),
668: the left-hand plot only the lower part up to $r \approx 200$.  The color coding
669: represents the strength of the azimuthal magnetic field. The magnetic field
670: lines, which were purely radial to begin with, wind many times around the
671: central axis, rendering it susceptible to kink instabilities.}
672: \label{fig:flines}
673: \end{figure}
674: 
675: \begin{figure}[t]
676: \begin{center}
677: \includegraphics[width=.8\linewidth]{figures/rhot_K3.jpg}
678: \end{center}
679: \caption{Density and temperature in a meridional slice in simulation K3
680: (Keplerian rotation profile, 3D).  The density is encoded in intensity, with
681: bright colors representing regions that are overdense with respect to the
682: environment.  The (square root of the) temperature is encoded in the hue, with
683: blue meaning cold and red meaning hot with respect to the environment.  The
684: hoop stress associated with $B_\varphi$ squeezes the plasma towards the central
685: axis and creates an underdense cavity around the central part of the jet. At
686: the boundary of this cavity the environment exerts the stress that confines the
687: jet. The observed opening angle for a jet like this would be smaller than the
688: width of the cavity.}
689: \label{fig:rhot}
690: \end{figure}
691: 
692: 
693: The jets are initially accelerated mainly by gas pressure. This holds up to the
694: sonic surface, which lies about halfway to the \Alfven surface.  Then, the
695: Lorentz force becomes the dominant driving force.  The \Alfven radii are at $r
696: \approx 30\ldots140$, depending on the simulation and the direction
697: $\vartheta$: near the axis (small $\vartheta$), the poloidal field $B_r$ is
698: amplified and the \Alfven radii are at larger distances than in the outer
699: regions (large $\vartheta$), where $B_r$ is attenuated.  The opening angle of
700: the jets increase somewhat with distance, but to a first approximation the flow
701: can still be treated as conical, see Fig.~\ref{fig:border}.  The jet front
702: crosses the upper boundary ($r=505$) at $t\approx260$ in the 2.5D simulations
703: and at $t\approx330$ in the 3D simulations. The latter are subject to more
704: numerical dissipation of kinetic energy, because the grid there is not
705: symmetric in azimuthal direction. This reduces the injected Poynting flux, see
706: Fig.~\ref{fig:eflow}. Apart from that, 2.5D and 3D simulations give, for our
707: purposes, comparable results.  Fig.~\ref{fig:flines} shows how the magnetic
708: field is wound up inside the jet in one of the simulations.  A typical density
709: and temperature distribution is shown in Fig.~\ref{fig:rhot}.
710: 
711: 
712: 
713: 
714: \subsection{Expected instabilities}
715: \label{sec:liability}
716: 
717: \begin{figure}[t]
718: \begin{tabular}{@{}l@{}r}
719: \includegraphics[width=.5\linewidth]{figures/h_K2.pdf} &
720: \includegraphics[width=.5\linewidth]{figures/h_R2.pdf}
721: \end{tabular}
722: \caption{Magnetic pitch as a function of distance in the 2.5D simulations. The
723: magnetic pitch is also the smallest possible wavelength of an instability. The
724: dependence of $h$ on the polar angle $\vartheta$ is stronger in the Keplerian
725: case (left plot), suggesting higher stability.}
726: \label{fig:pitch}
727: \end{figure}
728: 
729: \begin{figure}[t]
730: \begin{tabular}{@{}l@{}r}
731: \includegraphics[width=.5\linewidth]{figures/vAphi_vR_K2.pdf} &
732: \includegraphics[width=.5\linewidth]{figures/vAphi_vR_R2.pdf}
733: \end{tabular}
734: \caption{The expected onset of instability depends on the ratio of the
735: azimuthal \Alfven speed and the expansion velocity. This ratio is shown here
736: for the 2.5D simulations. The horizontal dotted lines are for two estimates of
737: the condition under which growth is possible (see text). Below the respective
738: line, expansion prevails and an instability cannot grow.}
739: \label{fig:growthcondition}
740: \end{figure}
741: 
742: 
743: \begin{figure}[t]
744: \begin{tabular}{@{}l@{}r}
745: \includegraphics[width=.5\linewidth]{figures/tauc_K2.pdf} &
746: \includegraphics[width=.5\linewidth]{figures/tauc_R2.pdf}
747: \end{tabular}
748: \caption{\Alfven crossing times with $\iota = \pi$ in the 2.5D simulations.  In
749: both cases, $\tauc$ and with it the expected instability growth time increases
750: with radius and distance from the central axis. The curves for $r \approx 100$
751: represent an underestimate, because the jet still accelerates at this
752: distance.}
753: \label{fig:acrosstime}
754: \end{figure}
755: 
756: 
757: The observed instabilities can be compared with expectations from linear
758: stability theory. To do this, we extract from the axisymmetric simulations the
759: quantities  that enter the stability conditions, and then compare the result
760: with the evidence of non-axisymmetric instability in the corresponding 3D
761: simulation. The available stability conditions apply only to steady or static
762: configurations and have been derived either in the context of controlled fusion
763: or cylindrical jets \citep{2000Appl,2000Lery}, hence the comparison can only be
764: indicative.
765: 
766: According to the Kruskal-Shafranov criterion, the longitudinal wavelength of an
767: instability must be at least as high as the magnetic pitch on the unstable
768: surface, defined to be the distance covered during one revolution of a helical
769: field line about the central axis.  Besides being the result of linear
770: stability analyses in the context of controlled fusion, it can be derived
771: heuristically from geometric arguments \citep{1958Johnson}.  Therefore, it
772: should give a convenient scale also in cases for which it was not originally
773: intended, like the expanding jets studied here.  Deviations can be expected
774: e.g. from the effect of one-ended line-tying, which in some cases has been
775: found to lead to increased instability as opposed to a configuration without a
776: free end \citep{2006Furno,2006Lapenta,2008Sun}.
777: 
778: For a conical jet, the magnetic pitch is
779: \begin{equation}
780:     h = 2\pi R \abs{\frac{B_r}{B_\varphi}}
781: \label{eq:pitch}
782: \end{equation}
783: on the $\vartheta=\const$ surface. See Appendix~\ref{app:conicalpitch} for a
784: derivation of this expression. In the simulations, $h$ decreases with $r$ and
785: settles to a constant value above the \Alfven radius, see Fig.~\ref{fig:pitch}
786: (compare also Fig.~\ref{fig:flines}). The variation of the pitch with
787: $\vartheta$ depends on the kind of rotation imposed at the lower boundary. In
788: the Keplerian case, the dependence is strong, with the asymptotic pitch being
789: approximately $10$ near the axis and $40$ at the limb of the jet.  In the rigid
790: rotation case, $h \approx 25$ in all directions within the jet.
791: 
792: 
793: 
794: The \Alfven crossing time in a conical, unaccelerated jet, defined as the time
795: it takes an azimuthal \Alfven wave to orbit the central axis, is given by
796: \begin{equation} \tauc = \frac{\iota R}{\vAphi - \iota v_R}
797: \label{eq:acrosstime} \end{equation} where $\iota=2\pi$ for a full revolution,
798: $\vAphi=\const$ is the azimuthal \Alfven speed and $v_R = v_r \sin \vartheta$
799: is the expansion velocity.  In the simulations, $\vAphi \approx \const$ above
800: the \Alfven radius.  This is as expected theoretically from conservation of
801: mass and magnetic flux in a conically expanding, steady axisymmetric jet.
802: $\tauc$ is finite and physically meaningful only if the condition
803: \begin{equation}
804:     \frac{\vAphi}{v_R} > \iota
805: \label{eq:growthcondition}
806: \end{equation}
807: is satisfied. If it is not, the expansion takes place too fast for an \Alfven
808: wave to cross the jet and an \Alfven mode instability cannot grow.  While the
809: critical value of $\iota$ is arguable, we note that causal contact across the
810: jet by Alfv\'en waves is only possible if $\iota \ge \pi$.  The situation in
811: our simulations is illustrated in Figs.~\ref{fig:growthcondition}
812: and~\ref{fig:acrosstime}. Instabilities can grow only slowly on magnetic
813: surfaces with large $\vartheta$.  Depending on the $\iota$ needed for efficient
814: growth, they may even be stalled due to the jet's expansion.  In any case,
815: instabilities grow most rapidly if they start at small $r$.  In regions where
816: the jet is accelerating ($\de v_r / \de r > 0$) or decollimating ($\de
817: \vartheta / \de r > 0$ along a field line), the effective \Alfven crossing time
818: is underestimated by Eq.~\eqref{eq:acrosstime}. The jet is then stabler than
819: condition~\eqref{eq:growthcondition} suggests.
820: 
821: 
822: \subsection{Instabilities found in the simulation}
823: 
824: \begin{figure}[t]
825: \includegraphics[width=\linewidth]{figures/jr_Br0_KR3.jpg}
826: \caption{Radial component of the current density ($\Nabla \times \vec{B}$) in
827: the 3D simulations just before the jets reach the upper boundary. The rod in
828: the middle of the jets is their central axis ($y$-axis).  Helical distortions,
829: characteristical for kink instabilities, can be seen in the backward current
830: (blue) in both cases.  The amplitudes and wavelengths are significantly larger
831: in the simulation with a rigid rotation profile (right-hand image).}
832: \label{fig:jrvolren}
833: \end{figure}
834: 
835: \begin{figure}[t]
836: \includegraphics[width=\linewidth]{figures/jrbary_R3.png}
837: \caption{Position of the barycenter of the backward current (blue material in
838: Fig.~\ref{fig:jrvolren}) in simulation R3. The left-hand map shows the
839: amplitude of the instabilities and the right-hand one its phase. The blank
840: region is where the jet has not been yet, its border marks the jet front.
841: Observers moving with the flow follow a time-position curve which is
842: (approximately) parallel to the front, i.e.  the instabilities are at rest with
843: respect to a comoving frame.}
844: \label{fig:jrmbarycenter}
845: \end{figure}
846: 
847: \begin{figure}[t]
848: \includegraphics[clip=true,width=\linewidth]{figures/comthj_1_R3.pdf} \\
849: \includegraphics[width=\linewidth]{figures/comthj_130_R3.pdf}
850: \caption{Amplitude of the perturbations (crosses, left-hand axis) as seen by an
851: observer (dotted line, right-hand axis) located just behind the jet front
852: (upper panel) and located 130 length units behind the jet front (lower panel)
853: in simulation R3.  The exponential fits (solid line, left-hand axis) yield the
854: instability growth time.}
855: \label{fig:comovingthj}
856: \end{figure}
857: 
858: 
859: We observe non-axisymmetric, kink-like distortions in the magnetic field and
860: other quantities in both 3D simulations. They emerge near the \Alfven radius,
861: propagate with the flow and grow in amplitude along the way.  It is convenient
862: to look at the current density $\vec{j}=\frac{c}{4\pi}\Nabla \times \vec{B}$
863: for a quantitative analysis.  The radial component $j_r$ is related to
864: $B_\varphi$ and is as such characteristic for the distortions in the magnetic
865: field.  In the unperturbed case, it is concentrated about the central axis and
866: along the outer boundary of the cavity illustrated in Fig.~\ref{fig:rhot}, with
867: respectively opposite orientation.  In our simulations, $B_\varphi$ is directed
868: in negative $\varphi$-direction and the axial current, accordingly, in negative
869: $r$-direction. We denote this backward current with $j_r^-$, so that $j_r =
870: j_r^+ + j_r^-$.
871: 
872: In the rigid rotation case (R3), the distortions attained large amplitudes of
873: several degrees. Looking at $B_\varphi$ in the $r=\const$ plane, we find that
874: the whole jet is affected by the kink.  The number of visible radial nodes is
875: 2--4, corresponding to wavelengths on the order of $150$, i.e. several times
876: larger than the magnetic pitch.  Owing to the distortions in the magnetic
877: field, the axial current was perturbed as shown in Fig.~\ref{fig:jrvolren} on
878: the right-hand image.  $j_r^-$ helically twines around the central axis in
879: reminiscence of ``ideal'' kink instabilities with an azimuthal mode number
880: $m=1$.
881: 
882: To analyze the unstable displacements, we determine the barycenter of the
883: backward current $j_r^-$ in the $r=\const$ plane, denoting its location with
884: $(\vartheta_j,\varphi_j)$. The result, from which one can directly read off
885: amplitudes and wavelengths, is shown in Fig.~\ref{fig:jrmbarycenter}.  The
886: slope of the points of constant phase $\varphi_j$ in the $r$-$t$ diagram
887: corresponds with the flow velocity $v_r$.  That is, the instabilities are at
888: rest with respect to a comoving frame.  We estimate the growth time in such a
889: frame by introducing an observer moving with flow and measure $\vartheta_j$ in
890: doing so.  We find strictly increasing, exponential growth if the observer is
891: located just behind the jet front, see upper panel in
892: Fig.~\ref{fig:comovingthj}. The exponential growth time $\taug$  is generally
893: on the order of the \Alfven crossing times shown in Fig.~\ref{fig:acrosstime}.
894: For observers which are farther behind the jet front, the amplitude does not
895: follow a simple exponential increase, see lower panel in
896: Fig.~\ref{fig:comovingthj} for an example. Rather, it saturates and even
897: declines in some cases. The reason for this is not clear.  We cannot rule out
898: the possibility that there is stabilizing feedback from the upper boundary.
899: Considering that the flow is super-\Alfvenic there, this seems unlikely
900: though.
901: 
902: In the Keplerian rotation case (K3), the jet also exhibits kink-like
903: distortions, see left-hand image in Fig.~\ref{fig:jrvolren}. However, the
904: perturbation amplitudes are much smaller, with $\vartheta_j$ attaining peak
905: values of about $1.4\degree$ directly behind the jet front and only about
906: $0.5\degree$ farther behind.  Unlike in the rigid rotation case, only inner
907: regions of the jet are affected by the kinks, the jet border is relatively
908: unharmed.  The wavelengths are on the order of 25--50, i.e. there are more
909: radial nodes than in the rigid rotation case.  Even for an observer traveling
910: just behind the jet front, the amplitude is not strictly increasing, but
911: saturates and tapers off after an initial rise with a growth time of $\taug
912: \approx 50$, i.e. also on the order of the relevant \Alfven crossing times in
913: Fig.~\ref{fig:acrosstime}.
914: 
915: 
916: 
917: \subsection{Impact on dynamics and energetics}
918: 
919: 
920: \begin{figure}[t]
921: \includegraphics[width=\linewidth]{figures/Eflow_R23.pdf}
922: \caption{Energy flow rates through $r=\const$ in 3D (thick lines) and in the
923: 2.5D (thin lines) simulations with a rigid rotation profile. There is no clear
924: evidence of additional conversion of Poynting flux to kinetic energy.  The
925: energy flow through the lateral boundaries is virtually zero at all times. The
926: situation is similar in the Keplerian case (K3 and K2).}
927: \label{fig:eflow}
928: \end{figure}
929: 
930: \begin{figure}[t]
931: \includegraphics[clip=true,width=\linewidth]{figures/vr_K23.pdf} \\
932: \includegraphics[width=\linewidth]{figures/vr_R23.pdf}
933: \caption{Radial velocities in the 3D (thick lines) and in the 2.5D (thin lines)
934: simulations.  The jets do not accelerate above the \Alfven radius, despite
935: instabilities. The value of $v_r$ near the central axis in simulation R3,
936: though rising, does not give conclusive evidence of acceleration, because the
937: jet's axis is strongly distorted by the instabilities.}
938: \label{fig:vr}
939: \end{figure}
940: 
941: 
942: Instabilities release magnetic energy by transforming it into kinetic energy, but
943: for dissipation in the sense of magnetic reconnection sufficiently small length
944: scales have to develop. In ideal MHD simulations like ours, such dissipation is
945: present in the form of numerical diffusion due to the effect of interpolations
946: across adjacent grid cells. The effect cannot be modeled by Ohmic resistivity
947: but can be quantified through secondary effects like changes in the energy
948: fluxes.
949: 
950: We did not find conclusive evidence of magnetic field dissipation provoked by
951: the instabilities in the 3D simulations.  For example, the flow of nonradial
952: magnetic field
953: \begin{equation}
954:     \int_{r=\const} \sqrt{B_\vartheta^2+B_\varphi^2}\,  v_r \,\de A 
955: \end{equation}
956: shows asymptotically the expected $\simm r$ behavior (as $B_\varphi \sim
957: r^{-1}$ and $A \sim r^2$) with minor fluctuations but no decreasing trend when
958: compared to the 2.5D simulations.
959: 
960: It is helpful to look at the energy fluxes.  From
961: Eq.~\eqref{eq:energy_equation_alt}, the energy flow rate in poloidal direction
962: is
963: \begin{equation}
964:     \efrate_\text{tot}(t,r) = \int_{r=\const} \left[ \left( 
965:             \frac{1}{2} \rho v^2 + \frac{\gamma}{\gamma-1}p + \rho \Phi \right) v_r
966:             + S_r \right] \de A .
967: \label{eq:efrate}
968: \end{equation}
969: Decomposing the integral from left to right, we identify the flow rates of
970: kinetic energy $\efrate_\text{kin}$, thermal enthalpy $\efrate_\text{thrm}$,
971: gravitational potential energy $\efrate_\text{grav}$ and magnetic enthalpy
972: $\efrate_\text{mag}$.  We plotted these in Fig.~\ref{fig:eflow} for the
973: simulations R3 and R2.  The total energy flow $\efrate_\text{tot}$ rises for
974: small $r$ due to the temperature-control term $K$ in
975: Eq.~\eqref{eq:energy_equation}.  The conversion of Poynting flux to kinetic
976: energy flux in the 3D simulation looks qualitatively the same as in the 2.5D
977: comparison simulation. In particular, there is no evidence of an additional
978: conversion of $\efrate_\text{mag}$ to $\efrate_\text{kin}$ due to magnetic
979: field dissipation. This agrees with the fact that there is no additional
980: acceleration of the jets, see Fig.~\ref{fig:vr}.
981: 
982: 
983: \section{Discussion and conclusions}
984: \label{sec:discussion}
985: 
986: We have simulated magnetocentrifugally driven, conical jets over a range in
987: distance of 1000 times the initial jet radius, in both 3D and axisymmetric
988: 2.5D.  The calculations extend to a factor of about 5--10 beyond the \Alfven
989: surface.  The 3D jets developed non-axisymmetric instabilities of the kink
990: kind.
991: 
992: The violence of the instabilities depends on the rotation profile applied at
993: the base.  With a rigid rotation ($\proptoo R$) profile, the perturbations grow
994: to much larger amplitudes than with a Keplerian ($\proptoo R^{-1/2}$) profile.
995: We suspect that the reason for the differing behavior lies in the magnetic
996: shear, defined as the variation of the magnetic pitch with distance to the
997: axis. In the rigid rotation case, there is virtually no shear as opposed to the
998: Keplerian case, for which the pitch increases with distance from the axis,
999: see Fig.~\ref{fig:pitch}.  A shear-free configuration is expected to be
1000: unstable to non-resonant modes, whereas a configuration with increasing pitch
1001: is expected to be unstable to modes with a resonant surface inside the jet
1002: \citep{2000Appl,2000Lery}.  This fits well with what we observe in the
1003: simulations, viz. that the kink is confined inside the jet in the Keplerian
1004: case. Heuristically speaking, the differing behavior could be attributed to the
1005: fact that the outer (high $\vartheta$) layers of the jet, which are more stable
1006: (higher magnetic pitch), damp internally arising modes in the Keplerian case.
1007: 
1008: In both cases, the longitudinal wavelength of the instabilities is $\simm5$
1009: times larger than the value of the magnetic pitch near the axis. The relation
1010: is qualitatively consistent with the findings of \citet{2000Appl} for a
1011: cylindrical jet. The growth time of the instabilities is on the order of the
1012: \Alfven crossing time.  The exact relation is difficult to determine, because
1013: the crossing time as well as the location of the resonant surface can only be
1014: estimated.  As the azimuthal magnetic field strength and with it the azimuthal
1015: \Alfven speed decrease past the \Alfven surface, opposing parts of the jets
1016: become causally disconnected from each other. Thus, the jet expands too fast
1017: for \Alfven mode instabilities to grow. The effect is amplified if the jet is
1018: diverging.  Recollimation, on the other hand, should boost the growth of
1019: instabilities.
1020: 
1021: As found in other studies, the conversion of magnetic enthalpy (Poynting flux)
1022: to kinetic energy is fairly efficient, on the order 70\%. Dissipation of
1023: magnetic fields by internal instabilities is expected to contribute additional
1024: acceleration of the flow \citep{2002Drenkhahn}.  The calculations do not show a
1025: clear signature of this process.  It seems that either the observed region is
1026: too small, and/or the numerical dissipation of magnetic fields is too weak.
1027: Also, from a macroscopic point of view, the instabilities were not violent
1028: enough to bring together fields with an antiparallel component, as is necessary
1029: for magnetic reconnection to occur.  Moreover, most of the magnetic enthalpy
1030: was already converted in the magnetocentrifugal acceleration process.
1031: Therefore, even if there was magnetic dissipation, the effect would not be
1032: dramatic. Nevertheless, we found that the magnetic field gets significantly
1033: distorted by the instabilities.  This should facilitate magnetic field
1034: dissipation further downstream but it may be necessary to extend the
1035: calculations to larger distances to see the effect.
1036: 
1037: It is tempting to compare the instability-related structures in the simulations
1038: with structures in observed jets. The 3D jet structure in Fig.~\ref{fig:rhot},
1039: for example, is reminiscent of the semi-regular patterns seen in \Halpha images
1040: taken of outflows from young stellar objects (YSO) like HH 34
1041: \citep{2002Reipurth}.  There are, however, other possible interpretations of
1042: the observed structure. The wiggles in YSO jets could also be the result of a
1043: precessing or orbitally moving source \citep[and references
1044: therein]{2002Masciadri}. The symmetric nature of structures often seen in jet
1045: and counterjet \citep[e.g. HH 212][]{2001Wiseman} suggests a modulation of the
1046: outflow speed or mass flux originating at the source of the outflow rather than
1047: an instability developing further away. The irregularities caused by the
1048: instabilities studied here are possibly more important for internal magnetic
1049: energy release inside the jet than for major observable structures like the
1050: knots and wiggles in YSO jets, though they are likely to contribute to these at
1051: some level as well.
1052: 
1053: 
1054: \begin{acknowledgements}
1055: We thank an anonymous referee for constructive comments. We are also grateful
1056: to Dimitrios Giannios for stimulating discussions.
1057: \end{acknowledgements}
1058: 
1059: \bibliography{ref}
1060: 
1061: 
1062: \appendix
1063: 
1064: \section{Magnetic pitch for a conical jet}
1065: 
1066: \label{app:conicalpitch}
1067: \begin{figure}[t]
1068: \begin{center}
1069: \includegraphics{figures/conicalpitch.pdf}
1070: \end{center}
1071: \caption{Magnetic field line with pitch $h$ on a conical surface.}
1072: \label{fig:conicalpitch}
1073: \end{figure}
1074: 
1075: The radial progress of the field line depicted in Fig.~\ref{fig:conicalpitch}
1076: is determined by
1077: \begin{equation}
1078:     \frac{\de r}{R \de \varphi} = \abs{\frac{B_r}{B_\varphi}} \eqqcolon a(r)
1079: \end{equation}
1080: where $R=r \sin\vartheta$ is the distance to the polar axis and $a(r)$ is
1081: the unsigned slope. Assuming that $B_r \propto r^{-2}$ and $B_\varphi \propto
1082: r^{-1}$ due to magnetic flux conservation, we can write
1083: \begin{equation}
1084:     a(r) = a_0 \frac{r_0}{r} .
1085: \end{equation}
1086: By integrating the resulting expression we obtain the radial distance covered
1087: after one revolution:
1088: \begin{equation}
1089:     \int_{r_0}^{r_0+h} \de r = \int_0^{2\pi} a_0 R_0 \, \de\varphi \quad \Rightarrow \quad h = 2\pi a_0 R_0 .
1090: \end{equation}
1091: Alternatively, we could also define a local magnetic pitch $\tilde{h}$ by
1092: taking $a = a_0 = \const$ for the slope. The result is
1093: \begin{equation}
1094:     \frac{\tilde{h}}{r_0} = \exp\left( 2\pi a_0 \sin\vartheta \right) - 1 .
1095: \end{equation}
1096: The difference between $h$ and $\tilde{h}$ turned out to be insignificant in
1097: our analysis.  This is understandable since $\tilde{h} \rightarrow h$ for small
1098: $a_0 \sin\vartheta$.
1099: 
1100: 
1101: 
1102: 
1103: 
1104: \end{document}
1105: 
1106: