0804.3808/ms.tex
1: %\documentclass{aastex}
2: %% preprint produces a one-column, single-spaced document:
3: %\documentclass[12pt, epsfig, preprint]{aastex}
4: %% preprint2 produces a double-column, single-spaced document:
5: \documentclass[preprint2]{aastex}
6: %\usepackage{graphicx}
7: 
8: %\newcommand{\vdag}{(v)^\dagger}
9: \newcommand{\myemail}{jraymond@cfa.harvard.edu}
10: \shorttitle{Non-Maxwellian Proton Distributions in Shocks} \shortauthors{Raymond \& Laming}
11: 
12: 
13: \begin{document}
14: 
15: %% one line. However, you may use \\ to force a line break if
16: %% you desire.
17: \title{Non-Maxwellian Proton Velocity Distributions in Nonradiative Shocks}
18: \author{J.C. Raymond}
19: 
20: \affil{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138}
21: 
22: \author{Philip A. Isenberg}
23: 
24: \affil{Department of Physics and Institute for the Study of Earth, Oceans and Space, 
25: University of New Hampshire, Durham}
26: 
27: \author{J.M Laming}
28: 
29: \affil{Naval Research Laboratory, Code 7674L, Washington DC 20375-5321}
30: 
31: 
32: \begin{abstract}
33: The Balmer line profiles of nonradiative supernova remnant shocks 
34: provide the means to measure the post-shock proton velocity distribution.
35: While most analyses assume a Maxwellian velocity distribution,
36: this is unlikely to be correct.  In particular, neutral atoms that pass
37: through the shock and become ionized downstream form a nonthermal
38: distribution similar to that of pickup ions in the solar wind.
39: We predict the H$\alpha$ line profiles from the combination of
40: pickup protons and the ordinary shocked protons, and
41: we consider the extent to which this distribution could affect the
42: shock parameters derived from H$\alpha$ profiles.  The 
43: Maxwellian assumption could lead to an underestimate of shock
44: speed by up to about 15\%.   The isotropization of the pickup ion
45: population generates wave energy, and we find that for the most favorable
46: parameters this energy could significantly heat the thermal particles. 
47: Sufficiently accurate 
48: profiles could constrain the strength and direction of the magnetic 
49: field in the shocked plasma, and we discuss the distortions from
50: a Gaussian profile to be expected in Tycho's supernova remnant.
51: 
52: \end{abstract}
53: 
54: \keywords{supernova remnants--shock waves--ISM: lines and bands--line: profiles--turbulence}
55: 
56: \section{Introduction}
57: 
58: Fast interstellar shock waves that encounter partially neutral gas 
59: are observable as filaments of pure Balmer line emission
60: if they are young compared to their radiative cooling times 
61: (Chevalier \& Raymond 1978; Raymond 1991; Ghavamian et al. 2001).
62: The Balmer lines are
63: produced in the thin layer just behind the shock where hydrogen atoms are
64: excited and ionized, and this layer is thin enough that Coulomb collisions
65: cannot bring different particle species into thermal
66: equilibrium.  Hence the Balmer lines can be used to probe the physical
67: processes in collisionless shocks.
68:  
69: The Balmer lines have two component line profiles.  The broad component arises from neutral H atoms 
70: created by charge transfer with post-shock protons, and its velocity 
71: width is comparable to the downstream proton thermal velocity.  The narrow component
72: comes from neutrals that have passed through the shock, but that have 
73: not been ionized by charge transfer.  Therefore, its
74: velocity width corresponds to the temperature of the pre-shock gas.  The
75: intensity ratio of the broad and narrow components depends on electron
76: and proton temperatures, $T_e$ and $T_p$, so that it can serve as a diagnostic 
77: for $T_e / T_p$ immediately behind the shock.  This is an important quantity for interpreting
78: X-ray spectra of SNRs and for understanding collisionless shocks.  In a few cases it has been
79: possible to measure the widths of UV lines of other elements, and therefore 
80: the kinetic temperatures, $T_i$, of other ions
81: \citep{ray95, laming96, ray03, korreck07}.  The overall result is that
82: the plasma behind relatively slow shocks ($\sim$ 300 $\rm km~s^{-1}$ )
83: is close to thermal equilibrium, while shocks faster than about
84: 1000 $\rm km~s^{-1}$ are far from equilibrium, with $T_e / T_p$ $<$ 0.1
85: and $T_i / T_p \sim m_i / m_p$
86: \citep{rakowski, ghavamian07}.  Other important applications of Balmer line diagnostics
87: for collisionless shocks are estimates of shock speed, which can be combined
88: with proper motions to find SNR distances (Winkler, Gupta \& Long 2003), and inferences
89: of cosmic ray diffusion coefficients from the properties of shock precursors
90: \cite{smi94, hes94, lee07}.  
91: 
92: All of the current models used to interpret the Balmer line profiles assume that 
93: the post-shock proton velocity distribution
94: is Maxwellian (Chevalier et al. 1980; Lim \& Raga 1996; Laming et al. 1996),
95: though there 
96: is no solid justification for that assumption.  Coulomb
97: collisions are not able to bring the protons to a Maxwellian rapidly enough, and
98: it is not clear what sort of distribution would be produced by plasma turbulence.
99: Heng \& McCray (2007) have recently drawn attention
100: to the importance of sequential charge transfer events in determining the
101: profile of the broad component at shock speeds above about 2000 $\rm km~s^{-1}$,
102: where the charge transfer cross section changes rapidly with energy, and this
103: affects some of the diagnostics.  Heng et al. (2007) have extended the 
104: model effort to a fuller treatment of the hydrodynamics than is usually
105: employed, but they keep the assumption that the proton distribution is 
106: Maxwellian.
107: 
108: Ion velocity distributions directly measured in the solar wind are essentially
109: never Maxwellian in the vicinity of  shocks.  Ion distributions typically have
110: power-law tails or strong anisotropies, with beamed components upstream and
111: highly perpendicular enhancements downstream (Schopke et al. 1983; 1990; Gosling and Robson
112: 1985; Thomsen 1985; Kucharek et al. 2004).
113: The Balmer line profiles of non-radiative
114: shocks provide a unique opportunity to search for non-Maxwellian velocity
115: distributions in astrophysical plasmas.  
116: 
117: In this paper we will keep the assumption that protons that pass through
118: the shock have a Maxwellian distribution at the temperature given by the 
119: Rankine-Hugoniot jump conditions (Draine and McKee 1993),
120: but we will add the manifestly non-Maxwellian distribution of
121: protons that pass through the shock as neutrals and become ionized.
122: We will consider the potentially observable
123: effects on the Balmer line profiles including line widths and centroid shifts
124: and how they might affect shock parameters derived from H$\alpha$ profiles.
125: We will also discuss the implications of magnetic field strength and direction and of plasma turbulence
126: on the profiles and the possibility that observed profiles could constrain the field
127: parameters.  We will briefly consider the implications of non-Maxwellian distributions for
128: the heating of electrons and minor ions.
129: % Excludes cosmic ray seed particle discussion.
130: 
131: \section{Pickup Ions}
132: 
133: Neutral particles that are ionized in the post-shock flow are very much like the pickup ions
134: (PUI) measured by spacecraft in the solar wind \cite{moebius85, gloeckler93, isenberg95}.  When neutral atoms slowly flow into the interplanetary medium,
135: they can be ionized by photons from the Sun, by charge transfer with solar
136: wind ions, or by collisions with electrons.  At that point, the new ions are streaming
137: with respect to the solar wind plasma at the solar wind speed, $V_{SW}$, which is much
138: larger than the local Alfv\'{e}n speed, $V_A$.  These ions are immediately
139: swept up by the magnetic field in the solar wind.  Their velocity component perpendicular to the
140: local magnetic field becomes a gyrovelocity around the field, which, in combination with
141: the instantaneous parallel component, initially forms a monoenergetic ring-beam in velocity
142: space.  This ring-beam is unstable, and the particles rapidly scatter toward isotropy by
143: interacting with ambient or self-generated waves, resulting in a velocity-space shell
144: \cite{sagdeev86, leeip, isenberg, bogdan}. 
145: 
146: In the solar wind, pickup protons are distinguished by their unusual velocity distributions,
147: but heavier pickup ions can also be recognized by their single
148: ionization states, such as $\rm He^+$ or $\rm O^+$, which stand out among solar wind
149: ions that are much more highly ionized.  The pickup ions add significant mass and momentum to the 
150: solar wind in the outer heliosphere, and the waves they generate play an important
151: role in heating the solar wind beyond 5 AU \cite{csmith01, isenberg03, isenberg}.  
152: The composition and charge state of these pickup ions indicate that they form the source 
153: particles for the observed anomalous component of cosmic rays (Garcia-Munoz et al. 1973, 1975; 
154: Fisk et al. 1974; Cummings \& Stone, 2007).  These particles must therefore be preferentially 
155: accelerated to several tens of MeV/nucleon at the solar wind termination shock or in the 
156: heliosheath beyond.  However, recent in situ observations during and after the Voyager encounters 
157: with the termination shock (Stone et al. 2005; Stone 2007) have shown that the energization process 
158: is still not well understood.  
159: 
160: \section{Consequences for SNR shocks}
161: 
162: Consider a planar shock in which the downstream magnetic field makes an angle $\theta$
163: with the shock normal.  Since the field component perpendicular to the flow is
164: compressed by the shock, $\theta$ is typically 60$^\circ$ to 85$^\circ$, though
165: of course pure parallel and pure perpendicular shocks maintain their field
166: directions.  For a strong shock with a compression ratio of 4, a neutral 
167: passing through the shock moves at $\frac{3}{4} V_S $ relative
168: to the post-shock protons.  Thus
169: when it becomes a pickup ion it acquires a gyro velocity 
170: 
171: \begin{equation}
172: V_{\bot} = \frac{3}{4} V_S ~sin (\theta)
173: \end{equation}
174: 
175: \noindent
176: and a velocity along the field direction of
177: 
178: \begin{equation}
179: V_{\|} = \frac{3}{4} V_S ~cos (\theta) 
180: \end{equation}
181: 
182: \noindent
183: relative to the post-shock plasma.  These monoenergetic particles  form an
184: unstable ring distribution in velocity space.  They can emit plasma waves
185: and interact with these waves to scatter into a more isotropic distribution.  
186: Generally, the dominant isotropization process is pitch-angle scattering through
187: the cyclotron resonant interaction with parallel-propagating  ion-cyclotron and
188: fast-mode waves (Wu \& Davidson 1972; Winske et al. 1985; Lee \& Ip 1987; see
189: also Zank 1999; Szeg\"{o} et al. 2000).
190: 
191: 	The ring-beam distribution may also be subject to other plasma instabilities,
192: depending on the relative density and downstream conditions.  In principle, a 
193: downstream magnetic field nearly parallel to the flow can result in bump-on-tail 
194: (Gary 1978) or firehose-like instabilities (Winske et al. 1985; Sagdeev et al. 1986).
195: The saturation of the Landau bump-on-tail instability leaves a highly anisotropic beam 
196: which still scatters in pitch angle through the cyclotron resonance.  The firehose 
197: instability could disrupt the beam, but requires both a high density of pickup ions 
198: relative to the background plasma and an ionization time-scale much shorter than the 
199: time-scale for cyclotron resonant pitch-angle scattering.  If the downstream magnetic 
200: field is nearly perpendicular to the flow, a mirror-mode instability can be excited 
201: (Winske \& Quest 1988; McKean et al. 1995), but this instability saturates at a much 
202: lower level than the resonant ion-cyclotron instability (Yoon 1992), and so may be 
203: neglected.  In this paper, we will take the ring-beam of newly-ionized protons to 
204: quickly stabilize through cyclotron-resonant pitch-angle scattering.  In particular, 
205: we will assume the rapid formation of a bispherical distribution.
206: 
207: 
208: \subsection{Bispherical Distribution}
209: 
210: Under most conditions, a given energetic proton is cyclotron-resonant with two
211: parallel-propagating electromagnetic modes.  If the proton parallel speed is much
212: faster than the Alfv\'{e}n speed, $V_A$, both of these waves will be Alfv\'{e}n
213: waves -- one propagating along the field in the same direction as
214: the proton and the other in the opposite direction.  Resonant scattering away
215: from the ring-beam will result in the amplification of one of these modes and the
216: damping of the other.  Which mode is unstable depends on the position of the ring-beam
217: in velocity space, as determined by the angle of the local magnetic field to the plasma flow
218: direction.  The resonant interaction with either wave yields a diffusion which conserves the
219: proton energy in the frame of the wave phase speed, scattering the particles along
220: a sphere in velocity space centered on one of the points $v_{\|} = \pm V_A$, as
221: shown in Figure 1.  A useful
222: analytical result is obtained in the case where the damped
223: mode can be neglected and the scattering at each point in velocity space is only due to 
224: interactions with the unstable mode.  In this case, a steady ring-beam will be scattered 
225: to a bispherical distribution: a uniformly populated shell formed by the two spherical
226: caps which meet at the position of the original ring-beam (Galeev \& Sagdeev 1988; Williams
227: \& Zank 1994).
228: 
229: Many of the basic properties of this distribution may be obtained geometrically.  If the 
230: ring-beam of the newly ionized protons is located at ($V_{\|},~V_{\bot}$) as given by
231: equations (1) - (2), the radii of the two spherical caps are $v_{\pm}^2 = V_{\bot}^2 + (V_{\|} \pm V_A)^2$.
232: The area of each cap in velocity space is $a_\pm = 2 \pi v_\pm (v_\pm \mp V_{\|} - V_A)$.
233: Since the particles are distributed uniformly over these areas, the net
234: streaming speed of the bispherical shell is
235: 
236: 
237: \begin{equation}
238: v_{bulk} = \frac{1}{a_T} [ V_A (a_+-a_-) + \pi V_{\bot}^2 (v_- - v_+) ]
239: \end{equation}
240: 
241: \noindent
242: where the total shell area $a_T = a_+ + a_-$.  Clearly, the case of flow perpendicular
243: to the magnetic field, $V_{\|} = 0$, gives $v_+ = v_-$ and $v_{bulk} = 0$.  Similarly,
244: for parallel flow faster than the Alfv\'{e}n speed, the distribution reduces to a
245: single sphere of radius $V_{\|} - V_A$, and the bulk speed is slowed to $v_{bulk} = V_A$.
246: In general, the streaming speed of the bispherical distribution is bounded by $\pm V_A$.
247: Figure 2 shows this streaming speed as a function of magnetic field angle $\theta$ for
248: several values of the downstream field strength, taking a shock speed of 2000 $\rm km~s^{-1}$
249: and an upstream proton density of 1 $\rm cm^{-3}$.
250: 
251: These simple properties may be modified for realistic conditions.  For instance, 
252: dispersion of the resonant waves will systematically shift their phase speed, and so 
253: distort the shape of the shell away from a sphere (Isenberg \& Lee 1996).  This 
254: distortion can be significant if the speed difference between the neutrals and the 
255: downstream plasma is comparable to $V_A$.
256: In addition, an efficient turbulent cascade could maintain the stable wave mode intensity
257: despite the damping by the pickup protons.  In this case, the multiple wave-particle
258: interactions with both stable and unstable waves can yield a much different distribution,
259: and even result in particle acceleration through the second-order Fermi mechanism
260: (Isenberg et al. 2003; Isenberg 2005).  However, the bispherical expressions provide a
261: reasonable first approximation to the pickup proton distribution expected downstream of a
262: strong supernova shock.  In this initial study, we will retain the bispherical assumptions,
263: and address these simplifications in the discussion section.
264: 
265: 
266: \subsection{Total Proton Distribution}
267: 
268: At any point in the downstream plasma the velocity distribution is
269: the sum of the distributions of the shocked protons and the protons
270: formed by ionization or charge transfer in the downstream gas.  If the
271: preshock neutral fraction is small, the distribution is dominated
272: by the shocked protons, and it will be difficult to detect the effects
273: of the pickup protons.  These effects will be much easier to see in the shocks of
274: Tycho's SNR, where the neutral fraction is around 0.85 \cite{ghavamian00}
275: than in SN1006, where it is around 0.1 \cite{ghavamian02}.  Figure 3
276: shows a simple model of a shock propagating at 2000 $\rm km~s^{-1}$ into
277: a medium with $n_H = n_p = 0.5 \rm cm^{-3}$, roughly similar to the values
278: expected for Tycho's SNR. The proton density just
279: behind the shock is the density of thermal protons, so the increase downstream
280: represents the addition of pickup protons.  The neutrals immediately behind
281: the shock make up the slow or narrow component.  Their density drops as
282: charge transfer converts them to pickup protons and relaces them with fast or broad 
283: component neutrals.  Eventually, collisional ionization removes all neutrals,
284: leaving a fully ionized plasma far downstream.  The rate coefficients for charge transfer
285: and ionization by electrons and protons were adopted from Laming et al. (1996).
286: Note that this plot assumes that the pickup protons move with the same bulk speed as the
287: background plasma.  This will be strictly true only for a perpendicular shock,
288: since the scattered shell of pickup ions will generally retain some streaming
289: motion with respect to the thermal plasma if the field has a component along
290: the flow.
291: 
292: Figure 4 shows the thermal proton and pickup ion distributions for one choice of 
293: the parameters.  For the modest Alfv\'{e}n speeds expected behind
294: SNR shocks, the PUI distribution is not far from spherical.  Thus
295: the total velocity distribution shows a peak with a sharp
296: cutoff plus high velocity wings from the thermal distribution.
297: 
298: The broad components of the Balmer line profiles will reflect the proton distributions, though
299: they are weighted by the charge transfer cross section.  Figure 5
300: shows the proton velocity distribution profiles in the direction parallel
301: to the shock front obtained by adding the background plasma distribution
302: to the PUI distribution.  We have chosen this
303: direction because strong limb brightening is required to make the H$\alpha$
304: emission from a non-radiative shock bright enough that a high
305: S/N profile can be obtained.  The projection is obtained
306: by multiplying the velocity along the magnetic field direction by sin$\theta$.
307: If the observer is not in the plane containing the pre-shock and post-shock
308: magnetic field, the centroid shift will be reduced.
309: In this paper we do not compute Balmer line profiles, since they depend on
310: specific shock parameters.  Such calculations will be needed for the
311: interpretation of observations, but for shocks below roughly 2000 $\rm km~s^{-1}$
312: the variation of charge transfer cross section with velocity is weak
313: enough that the H$\alpha$ profile should closely resemble the proton
314: velocity distribution (see Heng and McCray 2006).  It should be kept in
315: mind, however, that the broad component of H$\alpha$ is emitted from a
316: region of varying pickup ion fraction, with values near zero near the shock
317: and approaching the pre-shock neutral fraction far downstream.  Roughly speaking,
318: the H$\alpha$ profile will correspond to a pickup ion fraction of about
319: half the pre-shock neutral fraction. 
320: 
321: It is apparent from Figure 5 that the departure from a Maxwellian
322: ought to be detectable with sufficiently high S/N data.  The difficulty
323: is that the narrow component, whose intensity is generally dominant,
324: obscures the center of the broad component profile.
325: The usual procedure of fitting the sum of two Gaussians to the total profile
326: provides enough degrees of freedom to absorb modest departures from the
327: assumed Gaussians, especially if the far wings of the profile and the
328: background level are poorly defined.  For very fast shocks, the dropoff
329: in charge transfer cross section with velocity may suppress the high
330: velocity tail in any case.
331: 
332: As an estimate of the error that could be made by assuming a Maxwellian
333: proton distribution and using the resulting
334: broad line width to derive a shock speed, we fit the profiles in Figure
335: 5 with single Gaussians and compared those widths to the widths of a pure Gaussian
336: at the temperature expected from shock speed.
337: We find that the Gaussian widths estimated from the Figure 5 distributions
338: are as much as 14\% narrower than those predicted
339: for a pure thermal distribution of protons, so the shock speed
340: could be underestimated by 14\%.  This is the extreme case, however,
341: and underestimates about half that large would be typical.  These underestimates
342: would be partly countered if the pickup process provides additional heating to the 
343: plasma.
344: 
345: 
346: \subsection{Plasma Heating}
347: 
348: Another possible consequence of the pickup process is plasma heating by the waves
349: generated in the isotropization of the initial ring-beam of newly 
350: ionized protons.  The energy lost by the protons in scattering from the
351: ring-beam to the final nearly isotropic shell is transferred to the
352: resonant waves.  These waves in turn may heat the plasma, either directly or through
353: a turbulent cascade to dissipative modes.  In the simple bispherical picture of
354: section 3.1, the energy available to the waves is given by
355: 
356: \begin{equation}
357: E_w = E_o - E_{BD+} - E_{BD-}
358: \end{equation}
359: 
360: \noindent
361: where $E_o = mn ( V_{\bot}^2 + V_{\|}^2)/2$ is the energy in the initial ring-beam
362: and the energy in the bispherical distribution is 
363: 
364: \begin{equation}
365: E_{BD\pm} = \frac{nm\pi v_\pm^2}{a_T} [\frac{V_{\|}}{v_\pm}
366: (V_{\|} V_A \pm V_A^2 \mp v_\pm^2) + (v_\pm^2 - V_A)^2]
367: \end{equation}
368: 
369: Figure 6 shows the ratio of the total bispherical energy, $E_{BD} = E_{BD+}+E_{BD-}$
370: to the initial energy for various combinations of the Alfv\'{e}n speed and the
371: downstream magnetic field angle.  The wave energy
372: in (4) is essentially a maximum estimate, since the bispherical distribution
373: has a lower energy than the distributions obtained by including dispersive
374: effects or the replenishment of the stable wave modes (Isenberg \& Lee 1996;
375: Isenberg 2005).
376: 
377: The form of the plasma heating which results from the pickup proton generated
378: waves is an active area of research in the solar wind.   A phenomenological
379: model which assumes that these waves feed a turbulent cascade which dissipates by 
380: heating the thermal
381: ions has been shown to reproduce the observed proton temperatures in 
382: the outer heliosphere reasonably
383: well (Smith et al.2001, 2006; Isenberg et al. 2003; Isenberg 2005).  This 
384: heating can be important when
385: the upstream neutral fraction is large, and it may therefore affect the estimates of the
386: shock speed from the observed Balmer line width.
387: 
388: \subsection{Electron Heating}
389: 
390: Electron heating is observed to be very inefficient in fast shocks, so
391: the observed electron temperatures could provide a strong constraint 
392: on the wave energy even if only a modest fraction of the wave energy
393: is transferred to the electrons.  Observations of young SNRs show that $T_e / T_i$ is
394: less than 0.1 in shocks faster than about 1500 km/s (Rakowski 2005).
395: Ghavamian et al. (2006) propose that cosmic rays diffusing ahead of a fast
396: shock produce lower hybrid waves which then heat the electrons to a temperature of about 0.3 keV,
397: and this can reproduce the observed variation of $T_e / T_i$ with shock speed.
398: 
399: Alternatively, if the pickup proton ring distribution generates lower hybrid waves,
400: they could heat electrons.  The lower hybrid heating is inefficient unless the Alfv\'{e}n speed
401: is large (Omelchenko et al. 1989; Cairns \& Zank 2002), but the Alfv\'{e}n speed downstream
402: of SNR shocks is very poorly known.  In the absence of information about $V_A$, one cannot
403: make quantitative predictions.  In Tycho's SNR, which has a large neutral fraction in
404: the pre-shock gas, the observed low electron temperature precludes efficient transfer of energy
405: to the electrons if $V_A > 0.1 V_S$. 
406: 
407: 
408: \subsection{Downstream Heating of Heavy Ions}
409: 
410: 
411: Other elements with ionization potentials at least as large as that of hydrogen
412: will be partially neutral in the pre-shock gas.  They will also be ionized and picked
413: up except that they will be more likely to undergo electron or proton 
414: collisional ionization rather than charge transfer, so the process will
415: occur over a thicker region behind the shock.  Thus O, N and especially
416: Ne and He should initially form ring distributions and be picked up by the plasma.  As with the protons,
417: the initial width of the ring varies as sin$\theta$ and the initial speed
418: along the field as cos$\theta$.
419: 
420: Heavy ions present in the upstream plasma can also have peculiar 
421: downstream distributions due to their passage
422: through the shock. 
423: They are decelerated by the electric potential jump associated with the
424: shock, and because of their large mass to charge ratios they are decelerated
425: less than the protons.  Fuselier \& Schmidt (1997) find that the initial
426: ring velocity in this case is 
427: 
428: \begin{equation}
429: V_{\bot} = V_s (((m/q-1)+1/16)/(m/q))^{1/2} 
430: \end{equation}
431: 
432: \noindent
433: for a strong perpendicular shock.  Thus we expect that heavy ions passing through
434: the shock will have values of $V_{\bot}$ between $V_S$ and $3 V_S / 4$.
435: A few observations exist to test this expectation.
436: The line widths of C IV and He II lines in the Hopkins Ultraviolet Telescope
437: spectrum of SN1006 (Raymond et al. 1995) are the same as the width of H$\alpha$
438: within substantial uncertainties, and the O VI line observed with FUSE is
439: consistent with the same width (Korreck et al. 2004).  Ghavamian et al. (2002)
440: find that the pre-shock neutral fraction of H is about 0.1, while that of He is at least 0.7.
441: Since C has a lower ionization potential than H, and O has the same ionization
442: potential has H, these elements also have small pre-shock neutral fractions.  Thus
443: C and O should have larger values of $V_{\bot}$ than H, while He should be primarily
444: a pickup ion distribution.  Unfortunately, the 10\% to 30\% uncertainties in the
445: line widths preclude a definitive  comparison, but with somewhat higher quality
446: profiles for the UV lines one could begin to constrain the magnetic field direction.
447:  
448: \subsection{Cosmic Ray Modified Shocks}
449: 
450: Except for some consideration of magnetic field amplification, the discussion above 
451: assumes a simple magnetohydrodynamic shock.  However, both observations and theory (e.g., 
452: Drury \& V\"{o}lk 1981; Malkov et al. 2000; Warren et al. 2005)
453: indicate that a substantial fraction of the energy dissipated in the shock, as much
454: as 80\%, can be converted to cosmic rays.  This results in a ``modified shock" structure
455: with several interesting features; 1) a particle velocity distribution such as a Maxwellian with
456: a power law tail, 2) a smooth transition rather than a sharp shock jump,
457: 3) a compression ratio higher than the hydrodynamic factor of 4, and, 4) a lower
458: proton temperature for a given shock speed, since less energy is available to
459: heat the gas. 
460: 
461: The Balmer line profiles are not expected to reveal the non-thermal tails predicted 
462: for strong diffusive shock acceleration of cosmic rays, since only a very small fraction
463: of the particles ($\sim 10^{-3}$) are accelerated. Also, the charge transfer cross
464: section declines rapidly at speeds above about 2000 $\rm km~s^{-1}$ (e.g., Schultz et al. 2008), 
465: so the faster protons are less likely to produce broad component neutrals.  Therefore,
466: direct detection of the power law tail will be very difficult.
467:    
468: The smooth transition could change the profile in a manner incompatible with observations, in that
469: the gradually increasing temperature would give a composite H$\alpha$ profile which
470: is the sum of profiles formed at all the temperatures in the shock transition.  It
471: would probably not resemble the easily separable broad and narrow component profiles 
472: observed.  This difficulty would be avoided if the smooth transition occurs on a length 
473: scale smaller than the length scale for charge transfer, since few broad component
474: neutrals would form in the intermediate temperature region.  The length scale for a modified
475: shock is $\kappa / V_s$, where $\kappa$ is the cosmic ray diffusion coefficient.  The
476: charge transfer length scale is $V_s / (n_p q_{CT})$, where $q_{CT}$ is the charge transfer
477: rate coefficient.  Since $q_{CT} \sim 3 \times 10^{-7}~\rm cm^3 s^{-1}$ in the downstream
478: plasma, $\kappa$ should be smaller than about $10^{23}~\rm cm^2 s^{-1}$.  Values of
479: $\kappa$ of this order are required to accelerate cosmic rays to high energies within
480: an SNR lifetime, but they are comparable to the Bohm limit, and therefore at the
481: low end of the range of plausible values.
482: 
483: A high compression ratio, say 7 rather than the usual 4, would mean that the narrow 
484: component neutrals  move at $6 V_S / 7$ relative to the postshock gas, rather than 
485: $3 V_S / 4$, so the PUI component will have larger initial parallel and perpendicular 
486: velocities by 14\%. On the other hand, if a large fraction of the shock energy goes 
487: into nonthermal particles, the thermal speed of the protons will be reduced by a
488: factor $(1+P_C / P_G)^{-1/2}$, where $P_C$ and $P_G$ are the cosmic ray and gas pressures.
489: If $P_C$ is comparable to $P_G$, the thermal part of the line width will be seriously
490: affected and the shock speed will be underestimated if $P_C$ is assumed to be small.  
491: Most of the Balmer line filaments studied to date show very weak radio 
492: emission (e.g., the NW filament in SN1006 and the northern filament in the Cygnus Loop;
493: Ghavamian et al. 2001, 2002), so $P_C / P_G$ is probably small. 
494:   
495: \section{Application to Tycho's SNR}
496: 
497: Tycho's supernova remnant presents a good opportunity to search for the
498: effects described above because of its relatively high pre-shock neutral
499: fraction \cite{ghavamian00}, the excellent high and low resolution
500: spectra of knot g \cite{smi91, ghavamian01, lee07},
501: and the extensive X-ray and radio studies of both the thermal and
502: non-thermal shocks \cite{dickeljones, dickel, hwang, warren, bamba}.   
503: The preshock density is approximately
504: 1 $\rm cm^{-3}$, the pre-shock neutral fraction is approximately 0.85 and the shock
505: speed is approximately 2000 $\rm km~s^{-1}$ \citep{ghavamian00, ghavamian01}.
506: The magnetic field is likely to be amplified in strong SNR shocks 
507: (e.g., Lucek \& Bell 2000; Vink \& Laming 2003),
508: but its strength is not accurately known.  Non-thermal synchrotron
509: emission from nearby parts of the blast wave of Tycho's SNR, implies that the magnetic field
510: is on the order of 100 $\mu$G $\equiv 1 B_{100}$ \cite{warren}, yielding an Alfv\'{e}n speed
511: of approximately 100 $\rm km~s^{-1}$.  The field direction is not known
512: with certainty, though Dickel et al. (1991) show that the polarization indicates
513: a predominantly radial field on scales of a few arcseconds behind the shock.
514: The thermal pressure implied by the Rankine-Hugoniot jump conditions with
515: the shock speed and the pre-shock density above yields a plasma $\beta$ of 12/$B_{100}^2$.
516: 
517: There is a significant shift between the centroids of the broad and narrow
518: components of the H$\alpha$ profiles in Tycho. 
519: Smith et al. (1991) and Ghavamian et al. (2001) found shifts between
520: the broad and narrow components of H$\alpha$ of 240$\pm$60 and
521: 132$\pm$35 $\rm km~s^{-1}$, respectively, for two slit positions in knot g.
522: Smith et al. interpreted the shift as an indication that the
523: shock normal does not lie in the plane of the sky, so that the shift
524: represents a small component of the post-shock plasma speed.  This
525: interpretation is consistent with the observation of Lee et al. (2004), who
526: showed that the centroid of the narrow component is shifted with
527: respect to the centroid of the ambient gas in that region (though there
528: is some uncertainty about the size of this shift;  Lee et al. 2007).  However,
529: the shock normal cannot be very far from the plane of the sky, since very strong
530: limb brightening is required to account for the observed brightness of knot g.
531: 
532: Alternatively, it is possible that the shift between broad and narrow component centroids
533: is related to the projection of $v_{bulk}$ onto the line of sight.  
534: The shift is limited to approximately $V_A$, so a shift of the magnitude
535: measured would require that the projection of the magnetic field direction
536: onto the direction parallel the line of sight be fairly large, and therefore $\theta$
537: must be near 90$^\circ$.  Within the limits of the data now available, we
538: cannot tell whether the shift between broad and narrow centroids is
539: essentially a geometrical effect, as proposed by Smith et al. (1991) or a 
540: result of the pickup ion bulk speed discussed above.
541: 
542: A second puzzle relates to the nature of turbulence downstream from the shock.
543: If the 100 $\mu$G field is generated by turbulent amplification
544: in the shock front, it will be fairly disordered.  The 
545: non-resonant mechanism proposed by Bell (2004) predicts that the
546: scale of the turbulence is smaller than the gyroradius of cosmic ray protons
547: (Zirakashvili et al. 2008), and generally yields a perpendicular shock.
548: Compression by the shock would also make the mean field direction more perpendicular
549: to the shock normal.  Giacalone \& Jokipii (2007) and  Zirakashvili \& Ptuskin (2008) 
550: study the effects of density inhomogeneities on magnetic field generated downstream of 
551: the shock. Both works find significant magnetic amplification, and Zirakashvili \& Ptuskin
552: (2008) remark that the magnetic field component parallel to the shock normal is
553: more enhanced.  The interaction between the
554: the pickup ions and the field also tends to bend the field toward the shock
555: normal, and the observed field in Tycho's SNR is nearly radial at
556: the edge of the SNR \cite{dickel}.
557: A turbulent field would suggest that the pickup process occurs over
558: a large range of $\theta$, smearing out the profile as in Figure 5d.
559: Detection of a non-Gaussian profile in H$\alpha$
560: would provide some idea of the nature of the turbulence.  This will
561: require very good data and careful assessment of the instrument profile
562: and the background level, however, and existing data do not provide useful
563: constraints.
564: 
565: 
566: \section{Discussion}
567: 
568: \subsection{Caveats}
569: 
570: There are several qualifications to the analysis presented here. 
571: One is the use of PUI analysis based on Alfv\'{e}n
572: waves, which is appropriate for a cold plasma.  As mentioned above,
573: $\beta$ is around 12 for Tycho's SNR, and that will be typical for
574: the strong shocks seen as Balmer line filaments.  Thus, other wave modes may
575: be important.  It is unknown whether they will tend to 
576: change the directions, rather than the energies, of the protons
577: in the way that Alfv\'{e}n waves do.  It is also possible that they
578: will provide better coupling to the electrons.
579: 
580: Another question is whether the amplified B field behind the shock is
581: strongly turbulent on small scales.  If so, PUI would be generated
582: over a broad range of angles \cite{isenbergsw9, nemeth}, tending to wash out any
583: line-shift signature in
584: the H$\alpha$ profile.  The polarization measurements of Dickel et al
585: (1991) indicate that the field is reasonably well ordered on the scale
586: of their resolution, but it could be highly random on the 0.1$^\prime$$^\prime$
587: scale over which the H$\alpha$ is produced.
588: 
589: Finally, there is the question of momentum conservation when a significant
590: fraction of the downstream plasma has been picked up and streams along an oblique
591: magnetic field.  In this case, the thermal plasma would presumably act to 
592: cancel the transverse momentum, resulting in a rotation of the field
593: toward the shock normal.  We plan to quantitatively investigate this 
594: interaction in the near future.  The resolution may lie in the density gradient
595: of the pickup ions, but further calculations are needed.
596: 
597: \subsection{Implications for Balmer line filament analysis}
598: 
599: If the pickup ions provide a significant contribution to the H$\alpha$
600: profile, values of $V_s$ derived from Gaussian fits are somewhat in error.
601: This error would propagate into distances derived by combining shock speeds
602: derived from the Balmer line profiles with proper motions (e.g., Winkler,
603: Gupta \& Long 2003).  The modifications are probably not severe, and in cases
604: such as SN1006, where the pre-shock neutral fraction is only 10\% and the
605: contribution of pickup ions to the Balmer line profiles is only 5\%, they
606: would be completely negligible.  In cases where the pre-shock neutral fraction
607: is of order 50\%, as much as 25\% of the broad component emission could 
608: arise from atoms produced by charge transfer from pickup ions.  In such cases
609: $V_s$ would probably be underestimated.  On the other hand, waves emitted by the
610: pickup ions as they isotropize could heat the protons and lead to a compensating effect.
611: 
612: %The pickup ions might be efficiently injected into the diffusive shock acceleration
613: %process, providing seed particles just as heliospheric pickup ions are seed particles
614: %for the anomalous cosmic rays.  However, this would have implications for cosmic
615: %ray abundances that seem at odds with observations.   The waves produced as the
616: %pickup ions isotropize might heat the background plasma
617: 
618: If non-Maxwellian distributions can be observed by way of distortions of the H$\alpha$
619: profiles of non-radiative shocks, they could contain unique information about
620: the strength and direction of the magnetic field and the level of turbulence in
621: the region where the H$\alpha$ emission arises.  The most promising SNR where
622: non-Maxwellian distributions might be found is probably Tycho, thanks to its
623: large neutral fraction and relatively bright H$\alpha$ emission.
624: 
625: \bigskip
626: The authors thank Marty Lee for important suggestions.
627: We would also like to acknowledge very useful discussions at the Lorentz Center workshop
628: ''From Massive Stars to Supernova Remnants" and HST Guest Observer Grant GO-10577 
629: and FUSE Guest Observer grant NNG05GD94G to the Smithsonian Astrophysical Observatory.  
630: This work was supported in part by NSF Grant ATM0635863 and NASA Grant NNX07AH75G.
631: 
632: 
633: \begin{thebibliography}{}
634: 
635: %\bibitem[Chevalier, Kirshner, \& Raymond(1980)]{kcr} Chevalier,
636: %R. A., Kirshner, R. P., Raymond, J. C., 1980, \apj, 235, 186
637: 
638: \bibitem[Bamba et al. 2005]{bamba} Bamba, A., Yamazaki, R., Yoshida, T., 
639: Terasawa, T. \& Koyama, K.  2005, ApJ, 621, 793
640: 
641: \bibitem[Bell 2004]{bell} Bell, A.R. 2004, MNRAS, 353, 550
642: 
643: \bibitem[Bogdan et al. 1991]{bogdan} Bogdan, T.J., Lee, M.A \& Schneider, P.1991, JGR, 96, 161
644: 
645: \bibitem[Cairns \& Zank 2002]{cairns} Cairns, I.H., \& Zank, G.P. 2002, GRL, 29, 10.1029
646: 
647: %\bibitem[Chalov 2005]{chalov} Chalov, S.V. 2005, AdSpR, 35, 2106
648: 
649: \bibitem[Chevalier \& Raymond 1978]{cr78} Chevalier, R. A., Raymond, J. C., 1978, \apj, 225, L27
650:  
651: \bibitem[Chevalier et al. 1980]{ckr} Chevalier,
652: R. A., Kirshner, R. P., Raymond, J. C. 1980, \apj, 235, 186
653: 
654: \bibitem[Cummings \& Stone 2007]{cummings} Cummings, A.C., \& Stone, E.C. 2007, Space Sci. Rev., 130, 389
655: 
656: \bibitem[Dickel \& Jones 1985]{dickeljones} Dickel, J.M., \& Jones, E.M. 1985, ApJ, 288, 707
657: 
658: \bibitem[Dickel et al. 1991]{dickel} Dickel, J.R., van Bruegel, W.J.M., \& Strom, R.G. 1991, AJ, 101, 2151
659: 
660: \bibitem[Draine \& McKee 1993]{drainemckee} Draine, B.D., \& McKee, C.F. 1993, ARA\&A, 31, 373
661: 
662: \bibitem[Drury \& V\"{o}lk 1981]{drury} Drury, L.O'C. \& V\"{o}lk,H.J. 1981, ApJ, 248, 344
663: 
664: \bibitem[Fisk et al. 1974]{fisk} Fisk, L.A., Kozlovsky, B. \& Ramaty, R. 1974, ApJL, 190, L35
665: 
666: \bibitem[Fuselier \& Schmidt 1997]{fuselier} Fuselier \& Schmidt 1997, JGR, 102, 11273
667: 
668: \bibitem[Galeev \& Sagdeev 1988]{galeev} Galeev, A.A., \& Sagdeev, R.Z. 1988, Astr. Sp. Sci., 144, 427
669: 
670: \bibitem[Garcia-Munoz et al. 1973]{garcia1} Garcia-Munoz, M., Mason, G.M. \& Simpson, J.A. 1973, ApJL, 182, L81
671: 
672: \bibitem[Garcia-Munoz et al. 1975]{garcia2} Garcia-Munoz, M., Mason, G.M. \& Simpson, J.A. 1975, ApJ, 202, 265
673: 
674: \bibitem[Gary 1978]{gary} Gary, S.P. 1978, JGR, 83, 2504
675: 
676: \bibitem[Ghavamian et al. 2000]{ghavamian00} Ghavamian, P., Raymond, J. Hartigan, P. Blair, W.P. 
677: 2000, ApJ, 535, 266 
678: 
679: \bibitem[Ghavamian et al. 2001]{ghavamian01} Ghavamian, P., Raymond, J., Smith, R.C., \& 
680: Hartigan, P. 2001, ApJ, 547, 995
681: 
682: \bibitem[Ghavamian et al. 2002]{ghavamian02} Ghavamian, P., Winkler, P.F., Raymond, J.C. \& Long, K.S. 2002,
683: ApJ, 572, 888
684: 
685: \bibitem[Ghavamian et al. 2007]{ghavamian07} Ghavamian, P., Laming, J.M., \& Rakowski, C.E. 2007, ApJL, 654, L69
686: 
687: \bibitem[Giacalone \& Jokipii 2007]{gj} Giacalone, J. \& Jokipii, J.R. 2007, ApJL, 663, L41
688: 
689: \bibitem[Gloeckler et al. 1993]{gloeckler93} Gloeckler, G., Geiss, J., Balsiger, H., Fisk, L.A., Galvin, A.B., Ipavich, F.M., Ogilvie, K., von Steiger, R., and Wilken, B. 1993, Science, 261, 70
690: 
691: \bibitem[Gosling \& Robson 1985]{gosling} Gosling, J.T. \& Robson, A.E. 1985, in {\it Collisionless
692: Shocks in the Heliosphere: Reviews of Current Research}, ed. B.T. Tsurutani \& R.G. Stone 
693: (Washington, D.C.; AGU), 141
694: 
695: \bibitem[Heng \& McCray 2007]{hm} Heng, K., \& McCray, R. 2007, \apj, 654, 923
696: 
697: \bibitem[Heng et al. 2007]{hetal} Heng, K., van Adelsberg, M., McCray, R. \& Raymond, J.C. 2007, ApJ, 668, 275
698: 
699: % \bibitem[Hester 1987]{hester} Hester, J.J. 1987, \apj, 314, 187
700: 
701: \bibitem[Hester et al. 1994]{hes94} Hester, J. J., Raymond, J. C. \& Blair,
702: W. P. 1994, \apj, 420, 721
703: 
704: \bibitem[Hwang et al. 2002]{hwang} Hwang, U., Decourchelle, A., Holt, S.S. \& Petre, R. 2002, 
705: ApJ, 581, 1001
706: 
707: \bibitem[Isenberg 1995]{isenberg95} Isenberg, P.A. 1995, Rev. Geophys. Supp., U.S. Nat. Rep. IUGG, 623
708: 
709: \bibitem[Isenberg 1998]{isenbergsw9} Isenberg, P.A. 1999, {\it Proceedings of the Ninth
710: International Solar Wind Conference}, S.R. Habbal, R. Esser, J.V. Hollweg and P.A.
711: Isenberg, eds., (AIP Conference Proceedings 471), p. 815 
712: 
713: \bibitem[Isenberg et al. 2003]{isenberg03} Isenberg, P.A., Smith, C.W. \& Mattheus, W.H. 2003, ApJ, 592, 564
714: 
715: \bibitem[Isenberg 2005]{isenberg} Isenberg, P.A. 2005, ApJ, 623, 502
716: 
717: \bibitem[Isenberg \& Lee 1996]{isenberglee} Isenberg, P.A., \& Lee, M.A. 1996, JGR, 101, 11055
718: 
719: \bibitem[Korreck et al. 2004]{korreck} Korreck, K.E., Raymond, J.C., Zurbuchen, T.H.  \&
720: Ghavamian, P. 2004, ApJ, 615, 280
721: 
722: \bibitem[Korreck et al. 2007]{korreck07} Korreck, K.E., Zurbuchen, T.H., Lepri, S.T. \& Raines, J.M. 2007, 
723: ApJ, 659, 773
724: 
725: \bibitem[Kucharek et al. 2004]{kucharek} Kucharek, H.M., M\"{o}bius, E., Scholer, M., Mouikis, C., Kistler, L.,
726: Horbury, T.S., Balogh, A.R., R\'{e}me, H. \& Bosqued, J. 2004, Ann. Geophys. 22, 2301
727: 
728: \bibitem[Laming et al. 1996]{laming96} Laming, J. M., Raymond, J.
729: C., McLaughlin, B. M. \& Blair, W. P.  1996, \apj, 472, 267
730: 
731: \bibitem[Lee et al. 2004]{lee04} Lee, J.J., Koo, B.-C. \& Tatematsu, K. 2004, ApJL, 605, 113
732: 
733: \bibitem[Lee et al. 2007]{lee07} Lee, J.J., Koo, B.-C., Raymond, J.C., Ghavamian, P., 
734: Pyo, T.-S., Tajitsu, A. \& Hayashi, M. 2007, \apjl, 659, L133
735: 
736: \bibitem[Lee \& Ip 1987]{leeip} Lee, M.A. \& Ip, W.-H. 1987, JGR, 92, 11041
737: 
738: \bibitem[Lim \& Raga 1996]{limraga} Lim, A. J., Raga, A. C., 1996, \mnras, 280, 103
739: 
740: \bibitem[Lucek \& Bell 2000]{lucek} Lucek, S.G. \& Bell, A.R. 2000, MNRAS, 314, 65
741: 
742: \bibitem[Malkov et al. 2000]{Malkov} malkov, M.,A., Diamond, P.H. \& V\"{o}lk, H.J. 2000,
743: ApJL, 533, 171
744: 
745: \bibitem[McKean et al. 1995]{mckean} McKean, M.E., Omidi, N, \& Krauss-Varban, D. 1995, JGR, 100, 3427
746: 
747: \bibitem[Moebius et al. 1985]{moebius85} Moebius, E., et al. 1985, Nature, 318, 426
748: 
749: \bibitem[Meyeret et al. 1997]{meyer} Meyer, J.-P., Drury, L. O'C. \& Ellison, D.C. 1997, 
750: ApJ, 487, 182
751: 
752: \bibitem[N\'{e}meth et al. 2000]{nemeth} N\'{e}meth, Z., Erd\"{o}s, G. \& Balogh, A. 2000
753: GRL, 27, 2793
754: 
755: \bibitem[Omelchenko et al. 1989]{omelchenko} Omelchenko, Y.A., Sagdeev, R.A., Shapiro, V.D.
756: et al. 1989, Sov. J. Plasma Phys. 15, 427
757: 
758: \bibitem[Pittard et al. 2003]{pittard} Pittard, J.M., Hartquist, T.W. \& Ashmore, I. 2003,
759: A\&A, 408, 813
760: 
761: \bibitem[Rakowski 2005]{rakowski} Rakowski, C.E. 2005, Adv. Sp. Res., 35, 1017
762: 
763: \bibitem[Raymond 1991]{ray91} Raymond, J.C. 1991, PASP, 103, 781
764: 
765: \bibitem[Raymond et al. 1995]{ray95} Raymond, J. C., Blair, W. P. \& Long,
766: K. S., 1995 \apj, 454, L31
767: 
768: \bibitem[Raymond et al. 2003]{ray03} Raymond, J.C., Ghavamian, P., Sankrit, R.,
769: Blair, W.P. \& Curiel, S. 2003, ApJ, 584, 770
770: 
771: \bibitem[Sagdeev et al. 1986]{sagdeev86} Sagdeev, R.Z., Shapiro, V.D., Shevchenko, V.I.,
772: \& Szego, K. 1986, GRL, 13, 85
773: 
774: \bibitem[Schultz et al. 2008]{schultz} Schultz, D.R., Krstic, P.S., Lee, T.G. \&
775: Raymond, J.C. 2008, ApJ, in press
776: 
777: \bibitem[Smith et al. 2006]{csmith} Smith, C.W., Isenberg, P.A., Matthaeus, W.H. \& Richardson, J.D.
778: 2006, ApJ, 638, 508
779: 
780: \bibitem[Smith et al. 2001]{csmith01} Smith, C.W., Mattheus, W.H., Zank, G.P., Ness, N.F., Oughton, S. \& Richardson, J.D. 2001, JGR, 106, 8253
781: 
782: \bibitem[Smith et al. 1991]{smi91}Smith, R. C., Kirshner, R. P.,
783: Blair, W. P. \& Winkler, P. F. 1991, \apj, 375, 652
784: 
785: \bibitem[Smith et al. 1994]{smi94} Smith, R. C., Raymond, J. C. \& Laming, J. M. 1994,
786: \apj, 420, 286
787: 
788: \bibitem[Stone et al. 2005]{stone} Stone, E.C., Cummings, A.C., McDonald, F.B., Heikkila, B.C., Lal, N.
789: \& Webber, W.R. 2005, Science, 309, 2017
790: 
791: \bibitem[Stone et al. 2007]{stone07} Stone, E.C. 2007, Eos Trans. AGU, 88(52), Fall Meet. Suppl., Abstract SH11A-01
792: 
793: \bibitem[Szeg\"{o} et al. 2000]{szego} Szeg\"{o}, K., et al. 2000, Sp. Sci. Rev., 94, 429
794: 
795: \bibitem[Thomsen 1985]{thomsen} Thomsen, M.F. 1985, in {\it Collisionless Shocks in the Heliosphere:
796: Reviews of Current Research}, ed. B.T. Tsurutani \& R.G. Stone (Washington, D.C.: AGU), 253
797: 
798: \bibitem[Vink \& Laming 2003]{vink} Vink, J. \& Laming, J.M. 2003, ApJ, 584, 758
799: 
800: \bibitem[Warren et al. 2005]{warren} Warren, J.S. et al. 2005, ApJ, 643, 376
801: 
802: \bibitem[Williams and Zank 1994]{williams} Williams, L.L., \& Zank, G.P. 1994, JGR, 99, 19229
803:  
804: \bibitem[Winkler, Gupta and Long 2003]{winklong03} Winkler, P.F., Gupta, G. \& Long, K.S. 2003, ApJ, 585, 324
805:  
806: \bibitem[Winske et al. 1985]{winske85} Winske, D., Wu, C.S., Li, Y.Y., Mou, Z.Z. \& Guo, S.Y.
807: 1985, JGR, 90, 2713
808: 
809: \bibitem[Winske \& Quest 1988]{winske88} Winske, D. \& Quest, K.B. 1988, JGR, 77, 5399
810: 
811: \bibitem[Yoon 1992]{yoon} Yoon, P.H. 1992, Phys. Fluids B, 4, 3627
812: 
813: \bibitem[Zank 1999]{zank} Zank, G.P. 1999, Sp. Sci. Rev., 89, 413
814: 
815: \bibitem[Zirakashvili et al. 2008]{zirakashvili} Zirakashvili, V.N., Ptuskin, V.S.
816: \& V\"{o}lk, H.J.  2008, astro-ph/0801.4486
817: 
818: \bibitem[Zirakashvili \& Ptuskin 2008]{zirakashvili2} Zirakashvili, V.N. \& Ptuskin, V.S.
819: 2008, astro-ph/0801.4488
820: 
821: \end{thebibliography}
822: 
823: \clearpage
824: 
825: %\begin{figure}
826: %\plotone{stability.eps}
827: %\end{figure}
828:  
829: %\figcaption{Regions in the $\theta$, $V_A$ plane that are stable and unstable
830: %against ion-sound production by the double-peaked velocity distribution of the
831: %pickup protons.
832: %\label{stability}}
833: 
834: \begin{figure}
835: \plotone{f1.eps}
836: \end{figure}
837: \figcaption{Bispherical distribution in velocity space where the shock speed equals 1.0.
838: The Alfv\'{e}n speed for this example was 0.2 $V_S$ and $\theta$ was 60$^\circ$.
839: \label{schem}}
840: 
841: %\begin{figure}
842: %\plotone{bisph_alf20.25.2.eps}
843: %\end{figure}
844: %\figcaption{Proton velocity distribution for an angle $\theta$=20$^\circ$
845: %between the field an the shock front, and Alfv\'{e}n speed $V_A$=100 $\rm km~s^{-1}$
846: %and a pickup ion density of 0.25 the total density.
847: %\label{bisphere1 }}
848: 
849: \begin{figure}
850: \plotone{f2.eps}
851: \end{figure}
852: \figcaption{ Bulk velocity of a bispherical distribution along the field
853: direction for a 2000 $\rm km~s^{-1}$ shock with a pre-shock density of 
854: 1 $\rm cm^{-3}$ and a post-shock density of
855: 4 $\rm cm^{-3}$ for a range of post-shock magnetic field strengths and
856: angles between the shock normal and the field. The corresponding Alfv\'{e}n
857: speeds are 880, 710, 530, 350, 180 and 35 $\rm km~s^{-1}$.
858: \label{vbulk}}
859: 
860: \begin{figure}
861: \plotone{f3.eps}
862: \end{figure}
863: \figcaption{Variation of proton and neutral hydrogen densities behind a 2000 $\rm km~s^{-1}$
864: shock.  This model does not include the effects of weighting with the charge transfer
865: cross section or of drift on the pickup ions along the magnetic field, both of which tend
866: to increase the velocity of the pickup ions relative to the shock front and reduce their
867: density.
868: \label{quilha}}
869: 
870: \begin{figure}
871: \plotone{f4.eps}
872: \end{figure}
873: \figcaption{Proton velocity distribution for an angle $\theta$=70$^\circ$
874: between the field and the shock normal, an Alfv\'{e}n speed $V_A$=100 $\rm km~s^{-1}$,
875: and a pickup ion density of 0.25 the total density. The lower dashed line is
876: the bispherical distribution, the upper dashed line is the thermal proton
877: distribution, and the solid line is the total.
878: \label{bisphere1 }}
879: 
880: \begin{figure}
881: \plotone{f5.eps}
882: \end{figure}
883: \figcaption{ Velocity distributions for various combinations of parameters.  a) Ratios of pickup ions
884: to thermal protons ranging from 0.05 (outermost curve)to 0.65, b) Angles between the field and the shock
885: normal of 45$^\circ$ (outermost curve) and 75$^\circ$, c) Alfv\'{e}n speeds of
886: 0.05 $V_S$ (outermost curve) and 0.20 $V_S$, and d) the velocity distribution for a distribution of
887: angles assuming isotropic turbulence upstream and compression of $B_\bot$ by a factor of 4.
888: \label{params }}
889: 
890: \begin{figure}
891: \plotone{f6.eps}
892: \end{figure}
893: \figcaption{ Energy of the bispherical distribution as a fraction of the initial
894: energy of the neutral atoms.  The energy is computed in the rest frame of the post-shock
895: gas.  The curves correspond to values of $\theta$ of 85$^\circ$, 75$^\circ$, 65$^\circ$, 55$^\circ$,
896: 45$^\circ$, 35$^\circ$, 25$^\circ$,  15$^\circ$ and 5$^\circ$ from top to bottom.  
897: \label{energy }}
898: 
899: \end{document}
900: