1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[]{emulateapj}
3:
4:
5:
6: %%%%%%%%%%%%%%%%%%
7: % Macros
8: %%%%%%%%%%%%%%%%%%
9: \newcommand {\omegape} {\omega_\mathrm{pe}}
10: %%%%%%%%%%%%%%%%%%
11:
12:
13:
14: \shorttitle{COLLISIONLESS SHOCKS IN UNMAGNETIZED PLASMAS}
15: \shortauthors{KATO AND TAKABE}
16:
17: \begin{document}
18:
19:
20:
21:
22:
23: \title{Non-relativistic Collisionless Shocks in Unmagnetized Electron-Ion Plasmas}
24:
25: \author{Tsunehiko N. Kato and Hideaki Takabe}
26: \affil{
27: Institute of Laser Engineering, Osaka University,
28: 2-6 Yamada-oka, Suita, Osaka 565-0871, Japan
29: }
30: \email{kato-t@ile.osaka-u.ac.jp}
31:
32:
33:
34:
35:
36: \begin{abstract}
37: We show that the Weibel-mediated collisionless shocks
38: are driven at non-relativistic propagation speed ($0.1c < V < 0.45c$)
39: in unmagnetized electron-ion plasmas
40: by performing two-dimensional particle-in-cell simulations.
41: It is shown that
42: the profiles of the number density and the mean velocity
43: in the vicinity of the shock transition region,
44: which are normalized by the respective upstream values,
45: are almost independent of the upstream bulk velocity,
46: i.e., the shock velocity.
47: In particular,
48: the width of the shock transition region
49: is $\sim 100$ ion inertial length
50: independent of the shock velocity.
51: For these shocks
52: the energy density of the magnetic field generated by the Weibel-type instability
53: within the shock transition region
54: reaches typically 1-2\% of the upstream bulk kinetic energy density.
55: This mechanism probably explains the robust formation of collisionless shocks,
56: for example, driven by young supernova remnants,
57: with no assumption of external magnetic field in the universe.
58: \end{abstract}
59:
60:
61:
62:
63: \keywords{shock waves --- plasmas --- instabilities --- magnetic fields}
64:
65:
66:
67:
68: %%%%%%%%%%%%%%%%%%
69: \section{Introduction}
70: %%%%%%%%%%%%%%%%%%
71: Collisionless shocks are important phenomena in the universe;
72: they heat and compress interstellar or intergalactic plasmas
73: and often accelerate charged particles or cosmic rays.
74: The dissipation mechanism of the collisionless shocks is
75: highly complex and generally involved with
76: instabilities in the collisionless plasma.
77:
78:
79: The Weibel instability is driven in anisotropic or
80: counter-streaming collisionless plasmas \citep{Weibel59, Fried59}
81: and generates strong magnetic fields.
82: Collisionless shocks
83: inevitably generate an anisotropy
84: within their transition region,
85: where the upstream and downstream plasmas are mixed up,
86: therefore, the Weibel instability can develop
87: and generate strong magnetic fields there
88: if the background magnetic field is weak.
89: \citet{Moiseev63} considered such a process in non-relativistic
90: collisionless shocks and
91: \citet{Medvedev99} discussed the magnetic field generation
92: by the Weibel instability in relativistic collisionless shocks
93: of gamma-ray bursts.
94:
95:
96: In this context,
97: the Weibel instability in relativistic counter-streaming plasmas
98: has been investigated extensively by means of particle-in-cell simulations
99: \citep[e.g.,][]{Kazimura98, Gruzinov01, Haruki03, Silva03, Nishikawa03, Frederiksen04,
100: Spitkovsky05}.
101: Recent research with simulations showed that
102: the ``Weibel-mediated'' collisionless shocks can form
103: at relativistic propagation speed
104: in electron-positron plasmas \citep{Kato07,Chang08}
105: and also in electron-ion plasmas \citep{Spitkovsky08}
106: without background magnetic field;
107: in these shocks the dissipation mechanism is
108: effectively provided by the magnetic fields
109: generated by the Weibel-type instability
110: in the shock transition region.
111: In the universe,
112: these shocks can be driven
113: associated with relativistic phenomena, such as
114: gamma-ray bursts, jets from active galactic nuclei,
115: or pulsar winds.
116:
117:
118: There is also a possibility that
119: these Weibel-mediated shocks are driven
120: even in non-relativistic phenomena,
121: e.g.,
122: shocks of supernova remnants
123: whose propagation speed is typically $\sim 1,000-10,000$ km s$^{-1}$.
124: However,
125: in this case,
126: the Weibel instability can be inefficient
127: because the linear growth rate of the instability
128: is proportional to the flow velocity $V$.
129: Therefore,
130: it is not clear yet whether this kind of shocks can exist
131: in non-relativistic regime.
132: In this paper,
133: however,
134: we show that the Weibel-mediated collisionless shocks
135: propagating at non-relativistic speed
136: can form in unmagnetized electron-ion
137: plasmas by performing numerical simulations.
138: This also suggests a possibility to proof the formation of
139: these shocks in laboratory plasmas
140: with large-scale laser facilities.
141:
142:
143:
144:
145: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
146: \section{Method}
147: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
148: In order to investigate the collisionless shocks
149: in electron-ion plasmas without background magnetic fields,
150: we performed numerical simulations.
151: The simulation code is a relativistic and electromagnetic particle-in-cell code
152: with two spatial and three velocity dimensions (2D3V)
153: which was developed based on a standard method described by \cite{Birdsall}.
154: The basic equations are the Maxwell's equations
155: and the (relativistic) equation of motion of particles.
156: The simulation plane is the $x-y$ plane.
157: The $z$ axis is perpendicular to the plane.
158: Since we consider collisionless shocks in
159: unmagnetized plasmas,
160: the electromagnetic fields are initially set to zero
161: over the entire simulation box.
162: The boundary condition of the electromagnetic field
163: is periodic to each direction.
164:
165:
166: In the simulations
167: a collisionless shock is driven
168: by means of the ``injection method.''
169: There are two rigid walls at the left-hand side (smaller $x$) and right-hand side (larger $x$)
170: of the simulation box
171: and these walls reflect particles specularly.
172: Initially, both electrons and ions are uniformly loaded
173: in the region between the two walls
174: at the bulk flow velocity of $V$ in the $+x$-direction.
175: At the early stage of the simulations,
176: the particles located near the right wall
177: were reflected by the wall
178: and then interacted with the incoming particles,
179: i.e., the upstream plasma,
180: via some instabilities.
181: Such instabilities induce strong electric or magnetic fields that
182: provide dissipation mechanism of the shock and,
183: eventually, a collisionless shock is formed.
184: Note that the frame of the simulation
185: corresponds to the downstream rest frame.
186:
187:
188:
189:
190:
191:
192:
193: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
194: \section{Results}
195: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
196: We have carried out a series of simulations
197: with upstream bulk velocities of $V = 0.45c$, $0.2c$, and $0.1c$,
198: where $c$ is the speed of light.
199: The thermal velocity of electrons is typically taken as
200: one-tenth of the bulk velocity
201: and that of ions is determined so that the temperatures of electrons and ions are same.
202: Because of the computational power,
203: we used a reduced ion mass of $m_i = 20 m_e$.
204: Here,
205: we take $\omegape^{-1}$ as the unit of time
206: and the electron skin depth $\lambda_e \equiv c \omegape^{-1}$
207: as the unit of length,
208: where $\omegape \equiv (4\pi n_{e0} e^2 / m_e)^{1/2}$
209: is the electron plasma frequency defined for the electron
210: number density of the upstream plasma, $n_{e0}$.
211: We also define the ion inertial length, $\lambda_i \equiv (m_i/m_e)^{1/2} \lambda_e$.
212: The units of electric and magnetic fields are both taken as
213: $E_* = B_* \equiv c (4\pi n_{e0} m_e)^{1/2}$.
214:
215:
216:
217: %%%
218: \subsection{Results for $V=0.45c$}
219: %%%
220: The simulation for $V = 0.45c$ was performed
221: on a grid of $N_x \times N_y = 4096 \times 512$
222: with $\sim 27$ particles per cell per species.
223: The size of the simulation box is $L_x \times L_y = 2240 \lambda_e \times 280 \lambda_e$.
224: (We choose the grid size so that the system is stable against the cold beam nonphysical
225: instability, c.f. \citet{Birdsall}.)
226: The number density of ions
227: obtained from the simulation at $\omegape t = 2100$
228: is shown in Fig.~\ref{fig:npV045}.
229: The color shows the number density normalized by the upstream value,
230: namely, $n_{e0}$.
231: The upstream plasma flows from the left to the right
232: and goes through the transition region which has a filamentary structure
233: and then reaches the almost uniform downstream state ($x / \lambda_e > 1900$).
234: %%%
235: \begin{figure}[h]
236: \plotone{f1.eps}
237: \caption{
238: Number density of ions for the upstream bulk velocity of $V = 0.45c$
239: at $\omegape t = 2100$.
240: The left-hand side and right-hand side are the upstream and downstream of
241: the shock, respectively.
242: Filamentary structure is visible in the shock transition region,
243: $1300 < x/\lambda_e < 1900$.
244: }
245: \label{fig:npV045}
246: \end{figure}
247: %%%
248:
249:
250: Figure \ref{fig:npV045dev} shows the time evolution of the number density of ions.
251: The number density shown in color is averaged over the $y$-direction.
252: The transition region, or ``shock front'',
253: which is visible as a steep increase in the number density,
254: propagates upstream at an almost constant speed
255: ($\sim -0.18c$ measured in the downstream frame)
256: after $\omegape t \sim 500$.
257: Note that
258: the particles that are reflected at the right wall
259: at the early stage of the simulation,
260: which are remains of the initial condition,
261: eventually fade away at later times.
262: %%%
263: \begin{figure}[h]
264: \plotone{f2.eps}
265: \caption{
266: Time evolution of the number density of ions for $V = 0.45c$.
267: The horizontal and vertical axes are the $x$-coordinate and the time,
268: respectively.
269: The color shows the number density averaged over the $y$-direction.
270: The shock transition region, or ``shock front'', is visible as a steep increase in the number density.
271: }
272: \label{fig:npV045dev}
273: \end{figure}
274: %%%
275:
276:
277: Figure \ref{fig:profiles_V045} shows profiles of quantities
278: averaged over the $y$-direction at $\omegape t = 2100$:
279: (a) the ion number density, (b) the mean velocity in the $x$-direction,
280: and (c) the energy densities of electric and magnetic fields.
281: Both number density and mean velocity rapidly
282: change to reach the downstream values through the transition region
283: which extends with the width of $W \sim 500 \lambda_e$.
284: It is remarkable that a strong magnetic field is generated
285: within the shock transition region.
286: The energy density of the magnetic field
287: reaches approximately $1\%$ of the upstream bulk kinetic energy density
288: (measured in the downstream rest frame),
289: $U_\mathrm{KE} = n_{e0} (m_i + m_e) V^2 / 2$.
290: This strong magnetic field deflects and isotropize the particles coming from upstream;
291: this provides an effective dissipation mechanism
292: for this collisionless shock.
293: %%%
294: \begin{figure}[h]
295: \plotone{f3.eps}
296: \caption{
297: Profiles of (a) the ion number density,
298: (b) the mean velocity in the $x$-direction
299: both normalized by the respective upstream values,
300: and (c) the energy densities of
301: the magnetic (solid curve) and electric (dashed curve) fields
302: both normalized by the particle bulk kinetic energy density
303: of the upstream plasma.
304: }
305: \label{fig:profiles_V045}
306: \end{figure}
307: %%%
308:
309:
310: Figure \ref{fig:JxBz_V045} shows the current density
311: in the $x$-direction, $J_x$, and the $z$-component of the magnetic
312: field, $B_z$, at $\omegape t = 2100$.
313: It is evident that a number of current filaments exist
314: within the transition region.
315: These filaments consist of ions which are surrounded by the diffuse electrons
316: similar to the case of the relativistic counter-streaming plasmas \citep{Frederiksen04}.
317: The filamentary structure seen in the number density in Fig.~\ref{fig:npV045}
318: reflects the existence of them.
319: As seen in the figure,
320: these filaments generate the strong magnetic field around themselves
321: which is also observed in Fig.~\ref{fig:profiles_V045} (c).
322: This structure would be formed through the Weibel-type instability
323: like the relativistic shocks in unmagnetized plasmas \citep{Kato07,Spitkovsky08}.
324: %%%
325: \begin{figure}[h]
326: \plotone{f4.eps}
327: \caption{
328: (a) The current density $J_x$
329: and
330: (b) the magnetic field $B_z$
331: in the vicinity of the shock transition region
332: at $\omegape t = 2100$.
333: }
334: \label{fig:JxBz_V045}
335: \end{figure}
336: %%%
337:
338:
339:
340:
341:
342: \subsection{Dependence on upstream bulk velocity and ion mass}
343:
344: We also performed simulations with the bulk velocities of $0.2c$ and $0.1c$.
345: In both cases we confirmed that
346: the structure of the shock is qualitatively identical
347: to that of $V=0.45c$ case.
348: The ion number density and the magnetic field
349: obtained for $V = 0.1c$ are shown in Fig.~\ref{fig:npBzV01}.
350: %%%
351: \begin{figure}[h]
352: \plotone{f5.eps}
353: \caption{
354: (a) The ion number density and (b) the magnetic field $B_z$
355: obtained from the simulation for $V = 0.1c$.
356: }
357: \label{fig:npBzV01}
358: \end{figure}
359: %%%
360:
361:
362: It should be noted that these Weibel-mediated shocks
363: have similarities in the profiles of some quantities
364: under changing the upstream bulk velocity $V$.
365: Figure \ref{fig:npVprofile} shows
366: the profiles as Fig.~\ref{fig:profiles_V045} at different upstream bulk velocities:
367: $V = 0.45c$, $0.2c$ and $0.1c$.
368: Here, the horizontal axes are the $x$-coordinate normalized by
369: the ion inertial length including the relativistic effect,
370: $\lambda_i^* \equiv \Gamma^{1/2} \lambda_i$,
371: where $\Gamma \equiv [1 - (V/c)^2]^{-1/2}$ is the Lorentz factor
372: of the upstream bulk velocity.
373: It is remarkable that
374: the profiles in (a) and (b) are almost identical independent of $V$.
375: In particular,
376: the widths of the shock transition region are $W \sim 100 \lambda_i^*$
377: in \textit{all} cases.
378: The energy densities of magnetic fields
379: within the transition region
380: also reach typically 1-2\% of the upstream bulk kinetic energy density at maximum
381: independent of $V$.
382: %%%
383: \begin{figure}[h]
384: \plotone{f6.eps}
385: \caption{
386: Profiles of
387: (a) the ion number densities,
388: (b) the mean velocities in the $x$-direction $V_x$,
389: and (c) the energy densities of magnetic fields $U_B$
390: for $V = 0.45c$ (red solid curves), $0.2c$ (blue short-dashed curves),
391: and $0.1c$ (green long-dashed curves).
392: The horizontal axes are the $x$-coordinate normalized by the ion inertial length
393: (including the relativistic effect), $\lambda_i^*$.
394: The normalizations of the profiles are same as in Fig.~\ref{fig:profiles_V045}.
395: }
396: \label{fig:npVprofile}
397: \end{figure}
398: %%%
399:
400:
401: In order to see the effect of the mass ratio,
402: we performed simulations for $m_i / m_e = 50$ and $100$
403: with $V = 0.45c$.
404: The normalized ion number densities for the different three mass ratios
405: are shown in Fig.~\ref{fig:M_profile} as in Fig.~\ref{fig:npVprofile} (a).
406: The width of the transition region
407: is approximately $W \sim 100 \lambda_i^*$ in all cases
408: and this suggests the width may be the same order even for the real mass ratio.
409: However, there is a slight difference in the profiles and
410: this will be investigated in detail in the forthcoming paper.
411: %%%
412: \begin{figure}[h]
413: \plotone{f7.eps}
414: \caption{
415: Profiles of the ion number densities
416: for different ion-to-electron mass ratios:
417: $m_i/m_e = 20$ (red solid curve), $50$ (blue short-dashed curve),
418: and $100$ (green long-dashed curve).
419: The upstream bulk velocities are $V = 0.45c$ in all cases.
420: The horizontal and vertical axes are same as in Fig.~\ref{fig:npVprofile} (a).
421: }
422: \label{fig:M_profile}
423: \end{figure}
424: %%%
425:
426:
427:
428:
429:
430:
431: %%%%%
432: \section{Discussion}
433: %%%%%
434: The similarities in the normalized profiles shown in Fig.~\ref{fig:npVprofile}
435: can be explained as follows.
436: The structure of the transition region is essentially
437: determined by the coalescence history of the ion current filaments.
438: Here, we simply approximate the coalescence of two filaments as that of
439: two straight currents.
440: In this approximation,
441: the current in each filament is proportional to the bulk velocity $V$,
442: and a dimensional analysis of the equation of motion of the filaments
443: shows the time scale of the coalescence of two filaments
444: due to the Lorentz force, $\tau$,
445: is proportional to $V^{-1}$.
446: Thus, if we consider that the width in which the scale of the filamentary structure
447: becomes twice is given by $\Delta W \sim V \tau$, it is independent of $V$.
448: On the other hand,
449: the downstream end of the transition region is determined by
450: the saturation of the magnetic field of the ion current filaments;
451: i.e., the typical gyro-radius of ions becomes
452: comparable with the filament size, $r_{g,i} \sim r_f$,
453: like the case of electron-positron current filaments \citep{Kato05}.
454: This gives the condition $r_f \sim \lambda_i$,
455: which is independent of $V$.
456: Therefore,
457: if we consider the initial scale of the ion current filaments
458: is on the order of the electron skin depth $\lambda_e$,
459: the total width of the transition region is also independent of $V$.
460:
461:
462: Though the results described in the previous section
463: were obtained for unmagnetized plasmas,
464: these Weibel-mediated shocks can also be driven in magnetized
465: plasmas if the magnetic field is sufficiently weak.
466: One of the criteria for neglecting the upstream background magnetic field, $B_0$,
467: may be given by $r_{g,i} / W \gg 1$,
468: where $r_{g,i}$ is the typical gyro-radius of ions defined for $B_0$
469: and $W$ is the typical width of the shock transition region.
470: If we define the magnetization parameter $\sigma$ as
471: $\sigma \equiv [B_0^2/8\pi] / [n_{e0} (m_i + m_e) V^2 /2]$
472: and use the result obtained in the previous section $W \sim 100 \lambda_i^*$,
473: the criterion can be rewritten $\sigma \ll 10^{-4}$.
474: For shocks in supernova remnants (SNRs) of $\sim 1000$ years old,
475: typical parameters are
476: $n_{e0} \sim 1$ cm$^{-3}$, $V \sim 2000$ km s$^{-1}$, and $B_0 \sim 3\mu$G.
477: In this case,
478: the magnetization parameter becomes
479: $\sigma \sim 10^{-4} - 10^{-5}$
480: and therefore
481: there is a possibility that
482: the Weibel-mediated shocks are driven in SNRs.
483: For younger SNRs with higher shock speeds of $\ge 10,000$ km s$^{-1}$
484: \citep[e.g.,][]{Reynolds08},
485: these shocks would be formed more easily.
486:
487:
488: If the Weibel-mediated shocks are driven even
489: at the velocity of $\sim 1000$ km s$^{-1}$,
490: they can be investigated in laboratories
491: because laser experiments are capable of generating
492: collisionless plasma flow at this velocity \citep{Nishimura81, Ripin90}.
493: For this plasma flow with $n_{e0} \sim 10^{20}$ cm$^{-3}$
494: (even at this density,
495: the plasma can be collisionless during the shock formation),
496: the width of the shock transition region is estimated to be $W \sim 2$ mm.
497: This is a reasonable value as the plasma size produced with the present-day
498: large-scale laser facilities.
499: To demonstrate the shock formation in laboratories
500: is a challenging attempt
501: but it would be particularly worthwhile
502: as one of the key subjects for laboratory astrophysics
503: \citep[e.g.,][]{Takabe04}.
504:
505:
506:
507:
508:
509:
510:
511:
512: \acknowledgments
513: We would like to thank Y.~Sakawa and Y.~Kuramitsu for their useful discussions
514: and A.~Kageyama for her helpful comments.
515: Numerical computations were carried out at Cybermedia Center, Osaka University.
516:
517:
518:
519:
520: %%%%%%%%%%%%
521: % References
522: %%%%%%%%%%%%
523: \begin{thebibliography}{}
524: %
525: \bibitem[Birdsall \& Langdon (1991)]{Birdsall}
526: Birdsall,~C.~K., \& Langdon,~A.~B. 1991,
527: Plasma Physics via Computer Simulation (IOP Publishing: Bristol).
528: %
529: \bibitem[Chang et al.~(2008)]{Chang08}
530: Chang,~P., Spitkovsky,~A., \& Arons,~J. 2008,
531: \apj, 674, 378
532: %
533: \bibitem[Frederiksen et al.~(2004)]{Frederiksen04}
534: Frederiksen,~J.~T., Hededal,~C.~B., Haugb{\o}lle,~T., \& Nordlund, \AA. 2004,
535: \apj, 608, L13
536: %
537: \bibitem[Fried (1959)]{Fried59}
538: Fried,~B.~D. 1959,
539: Phys. Fluids, 2, 337
540: %
541: \bibitem[Gruzinov (2001)]{Gruzinov01}
542: Gruzinov,~A. 2001,
543: \apj, submitted (astro-ph/0111321)
544: %
545: \bibitem[Haruki \& Sakai (2003)]{Haruki03}
546: Haruki,~T., \& Sakai,~J. 2003,
547: Phys. Plasmas, 10, 392
548: %
549: \bibitem[Kato (2005)]{Kato05}
550: Kato,~T.~N. 2005,
551: Phys. Plasmas, 12, 080705
552: %
553: \bibitem[Kato (2007)]{Kato07}
554: Kato,~T.~N. 2007,
555: \apj, 668, 974
556: %
557: \bibitem[Kazimura et al.~(1998)]{Kazimura98}
558: Kazimura,~Y., Sakai,~J.~I., Neubert,~T., \& Bulanov,~S.~V. 1998,
559: \apjl, 498, L183
560: %
561: \bibitem[Medvedev \& Loeb (1999)]{Medvedev99}
562: Medvedev,~M.~V., \& Loeb,~A. 1999,
563: \apj, 526, 697
564: %
565: \bibitem[Moiseev \& Sagdeev (1963)]{Moiseev63}
566: Moiseev,~S.~S., \& Sagdeev,~R.~Z. 1963,
567: J. Nucl. Energy C, 5, 43
568: %
569: \bibitem[Nishikawa et al. (2003)]{Nishikawa03}
570: Nishikawa,~K.-I, Hardee,~P., Richardson,~G., Preece,~R.,
571: Sol,~H., \& Fishman,~G.~J. 2003,
572: \apj, 595, 555
573: %
574: \bibitem[Nishimura et al.~(1981)]{Nishimura81}
575: Nishimura,~H., et al. 1981,
576: \pra, 23, 2011
577: %
578: \bibitem[Reynolds et al.~(2008)]{Reynolds08}
579: Reynolds,~S.~P., Borkowski,~K.~J., Green,~D.~A., Hwang,~U.,
580: Harrus,~I., Petre,~R. 2008,
581: \apjl, in press
582: %
583: \bibitem[Ripin et al.~(1990)]{Ripin90}
584: Ripin,~B.~H., et al. 1990,
585: Laser Part. Beams, 8, 183
586: %
587: \bibitem[Silva et al.~(2003)]{Silva03}
588: Silva,~L.~O., Fonseca,~R.~A., Tonge,~J.~W., Dawson,~J.~M.,
589: Mori,~W.~B., \& Medvedev,~M.~V. 2003
590: \apjl, 596, L121
591: %
592: \bibitem[Spitkovsky (2005)]{Spitkovsky05}
593: Spitkovsky,~A. 2005,
594: in AIP Conf. Proc. 801, Astrophysical Sources of High
595: Energy Particles and Radiation, ed. T.~Bulik, B.~Rudak, \& G.~Madejski
596: (New York: AIP), 345
597: %
598: \bibitem[Spitkovsky (2008)]{Spitkovsky08}
599: Spitkovsky,~A. 2008,
600: \apjl, 673, L39
601: %
602: \bibitem[Takabe (2004)]{Takabe04}
603: Takabe,~H. 2004,
604: Nucl. Fusion, 44, S149
605: %
606: \bibitem[Weibel (1959)]{Weibel59}
607: Weibel,~E.~S. 1959,
608: \prl, 2, 83
609: %
610: \end{thebibliography}
611:
612:
613:
614:
615:
616:
617:
618: \end{document}
619:
620: