1: %\documentclass{aastex}
2: %% preprint produces a one-column, single-spaced document:
3: \documentclass[12pt, epsfig, preprint2]{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{H$\alpha$ Filament in SN1006} \shortauthors{Raymond et al.}
11:
12:
13: \begin{document}
14:
15: %% one line. However, you may use \\ to force a line break if
16: %% you desire.
17: \title{The pre-shock gas of SN1006 from HST/ACS observations}
18: \author{J.C. Raymond\altaffilmark{1}, K.E. Korreck\altaffilmark{1}, Q.C. Sedlacek\altaffilmark{1},
19: W. P. Blair \altaffilmark{2}, P. Ghavamian\altaffilmark{2}, R. Sankrit\altaffilmark{3}}
20:
21: \altaffiltext{1}{Harvard-Smithsonian Center for
22: Astrophysics, 60 Garden Street, Cambridge, MA 02138}
23: \altaffiltext{2}{The Johns Hopkins University, Baltimore, MD}
24: \altaffiltext{3}{Space Sciences Laboratory, University of California, Berkeley, CA}
25:
26: \begin{abstract}
27:
28: We derive the pre-shock density and scale length along the line
29: of sight for the collisionless shock from a deep HST image that
30: resolves the H$\alpha$ filament in SN1006 and updated model
31: calculations. The very deep ACS high-resolution image of the
32: Balmer line filament in the northwest (NW) quadrant shows that
33: 0.25 $\le~n_0~\le$ 0.4 $\rm cm^{-3}$ and that the scale along
34: the line of sight is about $2 \times 10^{18}~\rm cm$, while
35: bright features within the filament correspond to ripples with
36: radii of curvature less than 1/10 that size. The derived
37: densities are within the broad range of earlier density
38: estimates, and they agree well with the ionization time scale
39: derived from the Chandra X-ray spectrum of a region just behind
40: the optical filament. This provides a test for widely used
41: models of the X-ray emission from SNR shocks. The scale and
42: amplitude of the ripples are consistent with expectations for
43: a shock propagating though interstellar gas with $\sim$ 20\%
44: density fluctuations on parsec scales as expected from studies
45: of interstellar turbulence. One bulge in the filament
46: corresponds to a knot of ejecta overtaking the blast wave,
47: however. The interaction results from the rapid deceleration of
48: the blast wave as it encounters an interstellar cloud.
49:
50:
51: \end{abstract}
52:
53: \keywords{ISM:individual(SN1006)--supernova remnants--shock waves--optical:ISM}
54:
55: \section{Introduction}
56:
57:
58: SN1006 (G327.6+14.6) is one of the best SNRs for studying the physics of collisionless
59: astrophysical shocks, in particular the acceleration of non-thermal particles.
60: It is a nearby Type $\textrm{I}$a supernova remnant at a distance of
61: 2.1 kpc (the distance we assume throughout) with a diameter of $\sim$ 18 pc \citep{wink03}
62: and a shock speed in the 2500 - 2900 $\rm km~s^{-1}$ range \citep{gha02, hm}.
63: The remnant has a high Galactic latitude and modest foreground
64: reddening, E(B-V)=0.11 $\pm$ 0.02 \citep{sch80}.
65:
66: SN1006 has been observed at radio \citep{rey93, moffett}, optical
67: \citep{gha02,kwc87,smi91}, ultraviolet \citep{ray95, korreck} and X-ray
68: \citep{koyama, wink03,lon03,bam03,dye04} wavelengths. Pure Balmer
69: line filaments were found in the optical by \citet{vdb}. In the radio and X-ray,
70: the remnant has a limb-brightened shell structure with cylindrical
71: symmetry around a SE to NW axis probably aligned with the ambient
72: galactic magnetic field \citep{rey96,jp89}. The NE shock front of
73: SN1006 shows strong non-thermal X-ray emission \citep{koyama, dye04}
74: while the NW shock shows very little non-thermal emission
75: at radio or X-ray wavelengths. On the other hand, the pre-shock
76: density is several times higher in the NW than the NE \citep{korreck}.
77: Knots of X-ray emission from shocked SN ejecta
78: are scattered through the interior of the remnant \citep{lon03, vink03}.
79:
80: A Balmer line filament defines the blast wave in the northwest quadrant of
81: SN1006, and the H$\alpha$ profile provides diagnostics for the
82: shock speed and ion-electron thermal equilibration.
83: The H$\alpha$ emission from a Balmer-dominated shock has a two component profile \citep{cr78}. The
84: broad component is due to charge exchange between neutrals and protons, which
85: produces a population of neutrals at nearly the post-shock proton temperature.
86: The narrow component is produced
87: when cold ambient neutrals pass through the shock and emit line
88: radiation before charge transfer or ionization occurs. The ratio
89: of the broad to narrow flux is sensitive to the electron and ion temperatures.
90: The FWHM of H$\alpha$ broad component is 2290 $\pm$ 80 km s$^{-1}$, and
91: models imply a shock speed of v$_{\rm{shock}}= 2890
92: \pm 100$ km s$^{\rm{-1}}$ for a shock with little
93: electron-ion equilibration \citep{gha02}. However, new models incorporating
94: some additional physics obtain a lower shock speed of 2509$\pm$111 $\rm km~s^{-1}$
95: \citep{hm}.
96:
97: The pre-shock density, $n_0$, is an important parameter for understanding
98: the evolution of SN1006, as well as for interpreting the X-ray spectra
99: in terms of ionization time scale and determining the relative contributions
100: of shocked ISM and SN ejecta to the X-ray emission. The density could
101: also be important for attempts to understand the high ratio of
102: non-thermal to thermal X-ray emission in this SNR. Estimates of
103: $n_0$ cover a wide range, from 0.05-0.1 $\rm cm^{-3}$ based on the
104: global X-ray emission \citep{hss86} to 1 $\rm cm^{-3}$ based on
105: interpreting the scale over which the X-rays brighten as the length
106: scale for ionization of the shocked plasma \citep{win97}. A related
107: parameter is the length scale along the line of sight. The ripples
108: in the SNR blast wave could in principle result from density
109: inhomogeneities in the ambient medium or from knots of SN ejecta
110: overtaking the blast wave. ISM density fluctuations have been inferred for
111: a section of the non-radiative shock in the Cygnus Loop \citep{ray03},
112: while an ejecta knot is clearly the cause of one bulge in the
113: H$\alpha$ filament of SN1006 \citep{lon03, vink03}. The issue is
114: important for estimating the amplitude of interstellar turbulence
115: on sub-parsec scales and for the interpretation of the distance
116: between the blast wave and the reverse shock in terms of particle
117: acceleration \citep{warren}.
118:
119: In this paper we use an H$\alpha$ image obtained with the ACS imager on the
120: Hubble Space Telescope to determine the pre-shock density and the length scale
121: along the line of sight. This is possible because the ACS images resolve the
122: thickness of the narrow zone behind the shock where hydrogen atoms are excited
123: and ionized, and that thickness scales inversely as the pre-shock density. We
124: have computed new models of the H$\alpha$ emissivity as a function of distance
125: behind the shock taking into account recent results by \citet{hm}. We
126: discuss the observations in the next section, then compare with models to derive the
127: shock parameters. Section 4 provides a limit on the brightness of any
128: shock precursor, compares the densities with other estimates and
129: discusses the implications for other analyses of SN1006 observations.
130: Section 5 summarizes the conclusions.
131:
132:
133: \section{Observations}
134:
135:
136: The Hubble Space Telescope's ACS Wide Field Camera (WFC) imaged a full
137: field of 202 x 202 arcsec$^2$ at coordinates $\alpha_{2000}$
138: =$15^{\rm{h}}$ $2^{\rm{m}}$ $19.02^{\rm{s}}$, $\delta_{2000}$ =-41$^o$
139: 44\arcmin\ 48.4\arcsec\ on 9 orbits from 15-17 February 2006. Exposures
140: with durations of 2,746.0 s, 2,848.0 s, and 2,828.0 s were obtained for
141: each subset of 3 orbits, for a combined 25,266 second exposure. The exposures
142: were taken with the F658N H-alpha filter, which has a flat response at wavelengths
143: above about 6558\AA , drops to half the peak transmission at about 6548 \AA \/ and
144: 20\% of the peak transmission at 6540 \AA . This means that it passes the narrow component,
145: all the red wing of the broad component, and about half the blue wing of
146: the broad component, or about 90\% of the H$\alpha$ emission.
147:
148: The image was centered on the position where Ghavamian et al. (2002) obtained
149: a low dispersion spectrum and Sollerman et al. (2003) obtained a high dispersion
150: spectrum, so we are able to use the shock speed, neutral fraction
151: and electron-ion equilibration derived from those observations.
152: The position was also observed in the ultraviolet with the
153: Hopkins Ultraviolet Telescope \citep{ray95, lam96} and
154: with FUSE \citep{korreck}, so that we can use the information
155: derived in those papers about ion-ion thermal equilibration to
156: constrain model parameters. The
157: H$\alpha$ filament also defines the outer boundary of position NW-1
158: in the X-ray spectral analysis of Long et al. (2003), who
159: found the spectrum to be consistent with thermal emission from
160: shocked interstellar gas.
161:
162: The raw data were combined and reduced with the standard ACS
163: calibration pipeline (CALACS), involving bias/dark-current
164: subtraction, flat-fielding, image combination, and cosmic-ray
165: rejection (Sirianni et. al. 2005), then 'drizzled' (Fruchter
166: and Hook 2002) onto a 0.03 arcsec pixel scale with the task
167: 'multidrizzle' to correct for
168: geometric distortion and improve the sampling of the point
169: spread function \citep{koekemoer}.
170:
171: Figure~\ref{xrayhalpha} shows the HST image overlaid on the Chandra X-ray image.
172: The morphology of the filament is clearly that of a rippled
173: sheet seen edge-on, with the bright rims corresponding to tangencies
174: to the line of sight \citep{hester}. For the present purposes we are interested
175: in the simplest tangencies, since those are amenable to modeling.
176: The bulge near the SW corner of the image is morphologically similar to a larger,
177: brighter region farther to the SW where a clump of ejecta is overtaking the shock
178: \citep{lon03, vink03}, but it shows only a slight X-ray enhancement.
179: The shock morphology in the bulge is more complex, so this paper
180: concentrates on the smoother regions of the filament.
181:
182: IDL was used to extract and plot the curve of the shock
183: front spanning the drizzled image and to find the approximate direction
184: perpendicular to the shock. At each of 40 positions we extracted
185: the spatial profiles across the shock for a range of angles near
186: the initial estimate and selected the the profile showing the
187: narrowest H$\alpha$ peak as the one closest to the shock normal direction. Many
188: of the profiles suffer from low signal to noise or from complexity
189: due to several tangencies to the line of sight. We have selected
190: 8 profiles with bright, simple H$\alpha$ peaks for further
191: analysis. In particular, we model sections of the trailing
192: edge of the filament in regions corresponding to sections F, G and H
193: of the \citet{wink03} proper motion analysis. Figure~\ref{halpha}
194: shows the boxes used to extract spatial profiles, starting with
195: profile 8 in the upper left and ending with profile 29 in the lower
196: right. The profiles were extracted using boxes 4.5\arcsec wide and
197: 9\arcsec long. Their positions and position angles are shown in
198: Table 1. The profiles extracted
199: were similar to those used by \citep{wink03} for
200: measuring the proper motion of the filament; here, however, the goal was
201: to determine the width, not the position, of the filament.
202:
203:
204: \section{Analysis and Results}
205:
206: To interpret the images we compute models of the H$\alpha$
207: brightness behind a curved shock, convolve the model intensity distribution
208: with the ACS point spread function, and compare the models to the observations.
209: In general, the pre-shock density controls the brightness drop off behind
210: the peak, because the thickness of the emission region is inversely
211: proportional to the density. The radius of curvature of the shock determines the fall off
212: ahead of the peak for the concave outward filaments that we model here.
213: The absolute intensity scales approximately as $n_0^2 f_{neut} R^{1/2}$, where
214: $n_0$ is the pre-shock density, $f_{neut}$ is the pre-shock neutral fraction,
215: and R is the radius of curvature of the shock.
216:
217: \subsection{Model Calculations}
218:
219: We start with model calculations similar to those of \citet{lam96}. Figure \ref{model}
220: shows the H$\alpha$ emissivity as a function of distance behind the shock for
221: a 2900 $\rm km~s^{-1}$ shock model with pre-shock density $n_0~=~0.25~\rm cm^{-3}$,
222: a neutral fraction of 0.1, and a ratio $T_e / T_p~= 0.05$ at the shock front. The
223: model follows neutral hydrogen as it passes through the shock and undergoes
224: collisional excitation and ionization by protons and electrons as well as charge
225: transfer with post-shock protons. The atomic rates are described by \citet{lam96},
226: and Coulomb collisions slowly transfer energy from the ions to the electrons. The
227: radiative transfer involved in the conversion of Ly$\beta$ photons to H$\alpha$
228: photons, important for the narrow component, is described in \citet{lam96}. A fraction
229: of the H$\alpha$ arises from converted Ly$\beta$ photons, and those H$\alpha$
230: photons are produced over a scale of about 1 Ly$\beta$ mean free path. That scale
231: is about 0.1\arcsec, which is small enough compared to the observed filament widths
232: that we ignore it.
233:
234: Evidence for a shock precursor has
235: been reported for a number of non-radiative shocks from low ionization emission
236: lines \citep{hes94, sol03}, from the velocity widths of their narrow components \citep{smi94, hes94},
237: and from the spatial distribution of the narrow component emission
238: \citep{lee06}. The narrow component in SN1006 shows no broadening beyond that
239: expected for the ambient ISM \citep{sol03}, so we do not include any precursor
240: emission in the models.
241:
242: Recently, \citet{hm} investigated the consequences of the sharp decline of the
243: the charge transfer cross section at speeds above about 2000 $\rm km~s^{-1}$ for the
244: Balmer line profiles. They estimated a shock speed of 2509$\pm$111 $\rm km~s^{-1}$
245: from the Balmer line profile presented by \citet{gha02}, as opposed to the value
246: 2890$\pm$100 $\rm km~s^{-1}$ found by Ghavamian et al. Both values apply to the
247: case of little ion-electron thermal equilibration in the shock. Strong equilibration
248: is ruled out by the X-ray spectrum of \citet{lon03} and by the combination of
249: broad to narrow intensity ratio given by \citet{gha02} and the Ly$\beta$ radiative
250: transfer calculations of \citet{lam96}. The smaller shock speed of \citet{hm} would decrease the
251: distance scale of the H$\alpha$ emission behind the shock. However, it would also
252: imply a smaller distance for SN1006 based the combination of the proper motion and the
253: and shock speed \citep{wink03}, so the angular width of the emission region would be unchanged
254: to first order.
255:
256: There is another important implication of the work of \citet{hm} for the present
257: study. The rapid drop in charge transfer cross section with increasing
258: velocity means that neutrals are more likely to undergo charge transfer with protons
259: moving away from the shock than with protons moving toward the shock. While the
260: velocity dependence of the cross section was included in earlier models, the anisotropy
261: of the resulting H I velocity distribution was not. Thus the earlier models implicitly
262: assumed that the broad component neutrals move away from the shock at $V_s$/4, the
263: same speed as the post-shock protons. By integrating the product of velocity
264: times charge transfer cross section, $\sigma_{cx} v$ \citep{redbook},
265: over Maxwellian distributions at the post shock temperatures, we
266: find that after 1 charge transfer event the average neutral is moving away from the shock
267: at 1500 $\rm km~s^{-1}$ behind a 3000 $\rm km~s^{-1}$ shock, rather than the 750 $\rm km~s^{-1}$
268: of the ionized gas. For a 2500 $\rm km~s^{-1}$ shock, the neutrals move at 1100 $\rm km~s^{-1}$
269: instead of 625 $\rm km~s^{-1}$. Thus for the interesting range of shock speeds, the
270: neutrals move away from the shock twice as fast as is assumed in the model shown in
271: Figure \ref{model}. The relative velocity is much smaller for the second charge transfer
272: event, so after two charge transfers the neutrals have a speed closer to that of the downstream
273: plasma. Kevin Heng (2007, private communication) has provided the average downstream
274: speed of the broad component neutrals computed by the \citet{hm} model code. For the
275: 2500 to 3000 $\rm km~s^{-1}$ velocity range it is 1.32 times the plasma speed. Therefore,
276: for comparison to observations we stretch the spatial scale of the broad component emission
277: shown in Figure 3 by a factor of 1.32. Particle conservation implies that the neutral
278: density is decreased by the same factor, so the broad component emissivity at each point
279: is reduced by a factor of 1.32. Finally,
280: for comparison with the observations we multiply the broad component
281: emission by 0.75 to account for the drop of sensitivity in the blue wing resulting from
282: the transmission of the F658N filter.
283:
284: To model the geometry of the filament, we assume a shock surface that is concave outward
285: with radius of curvature R. The H$\alpha$ emissivity from the planar model extends
286: behind the shock at each point. Numerical integration then gives the brightness as
287: a function of position relative to the tangent point of the shock. For comparison with
288: the observations, we convolve the model emissivity with the ACS point spread function.
289: In order to properly account for the pixel sampling in cuts made at about 45$^\circ$
290: to the rows and columns of the detector, we measured the Gaussian widths of stars near
291: the filament (2.31 pixel FWHM), then placed a dense series of circular Gaussians along
292: a 45$^\circ$ line, extracted the profile perpendicular to that line, and measured a
293: slightly broader 2.43 pixel width. This Gaussian is convolved with the models for
294: comparison with the observed spatial profiles.
295:
296: Figure~\ref{schem} is a schematic diagram of the geometry we imagine in a plane that
297: includes the line of sight (dashed line) and a radial vector from the center of the
298: SNR. The light line shows a large scale ripple in the shock front, and we have superposed
299: a smaller scale ripple to obtain the shape of the heavier curve that is shaded toward
300: the inside of SN1006. The shading indicates the H$\alpha$ emissivity, which fades
301: gradually from the shock front towards the inside of the SNR. The trace in the lower
302: right indicates the H$\alpha$ brightness obtained for a cut across the tangent point
303: of the shock by integrating the H$\alpha$ emissivity along different lines of sight.
304:
305: \subsection{Comparison of Models and Observations}
306:
307:
308: Figure~\ref{obspred10} compares a grid of models with different $n_0$ and R
309: to the spatial profile at position 10. The sharp spikes ahead of the H$\alpha$
310: peak are faint stars in the extraction region. The models have been scaled to match the
311: peak H$\alpha$ brightness of the filament by simply adjusting the neutral fraction.
312: Neutral fractions above 1 are obviously unphysical, so models with low $n_0$ and
313: small R are ruled out. \citet{gha02} estimated a pre-shock neutral fraction of
314: 0.1 from the ratios of He I and He II lines to H$\alpha$, so we take the permitted range
315: of neutral fractions to be $0.05 \le f_{neut} \le 0.2$. Figure~\ref{obspred10} shows that models with $n_0$ below
316: about 0.30 $\rm cm^{-3}$ fall off too slowly behind the shock, while the models with
317: $n_0$ above about 0.35 $\rm cm^{-3}$ give too sharp a peak.
318: The R=$5 \times 10^{16}$ cm model comes closest to
319: matching the observed profile. Models with
320: smaller R predict H$\alpha$ emission fainter than observed. We conclude
321: that $0.3~\le~n_0~\le~0.35$ and $10^{16.5}~\le~R ~<~10^{16.8}$ cm.
322:
323: Figure~\ref{obspred28} shows the analogous plots for position 28. The peaks in this
324: section of the filament are both brighter and broader, with a fairly constant H$\alpha$
325: intensity ahead of the brightness peak. None of the models match exactly, probably
326: because the rippled sheet does not follow the assumed shape of an arc of a circle.
327: From Figure~\ref{obspred28} we conclude that $n_0$ must be greater than 0.25 $\rm cm^{-3}$
328: to avoid a long tail toward the inside of the remnant, but that densities above
329: 0.4 produce too sharp a peak. R must be less than $2 \times 10^{17}$ cm, because
330: larger radii of curvature predict a shoulder on the outer side of the spatial profile
331: that exceeds the observations, while R less than $0.7 \times 10^{17}$ cm requires unacceptably
332: high values of $f_{neut}$ to match the brightness. We conclude that the acceptable
333: ranges are $0.25 ~\le~n_0~\le~0.35$ and $0.7 \times 10^{17}~\le~R~\le~1.5 \times 10^{17}$ cm.
334: We note, however, that the agreement between the model spatial profiles and the
335: observations is not as good as at position 10. Based on Figure~\ref{halpha} , it seems possible that
336: the bright filament at position 28 contains more than one tangency to the line of sight,
337: broadening the spatial profile and increasing the total brightness. Thus the upper
338: limit on $n_0$ and the lower limit on R are less secure than in the case of position 10.
339:
340: Table 1 presents the ranges of $n_0$ and R derived from other positions along the filament.
341: We did not attempt to model other positions where the emission peak is faint or where
342: complex morphology indicates a more complex structure than can be approximated by a simple
343: curved sheet. The profiles of positions 8 through 11 are qualitatively similar to that
344: of position 10, and those of positions 26 through 29 are like that of position 28.
345:
346: \section{Discussion}
347:
348: \subsection{Pre-shock density}
349:
350: The densities near 0.3 $\rm cm^{-3}$ that we derive fall in the middle of the range of
351: previous values. \citet{hss86} estimated 0.05-0.1 $\rm cm^{-3}$ from the global X-ray
352: spectrum. Since the pre-shock density in the NE, and quite likely the rest of the remnant,
353: is about 2.5 \citep{lon03} to 4 \citep{korreck} times smaller than in the region we observed, the
354: \citet{hss86} estimate is consistent with our results, even though most of the X-ray emission
355: was subsequently shown to be non-thermal in nature. \citet{win97} obtained a pre-shock
356: density near 1.0 from the spatial profile of the X-ray emission, but \citet{lon03}
357: found $n_0~\simeq~0.25$ from the value of $n_e t$ obtained from the Chandra spectrum
358: of their region NW-1, which lies immediately behind the region of the Balmer line filament
359: we observed. The Chandra spectrum showed solar abundances, which is consistent with
360: the interpretation of shocked interstellar gas, and the temperature of 0.7 keV is
361: consistent with very inefficient thermal equilibration between ions and electrons.
362: Thus the present results agree well with the results of shock wave models of the
363: X-ray spectra and confirm the parameters derived. However, as \citet{lon03} point
364: out, the shock models did not provide an acceptable $\chi$-squared fit to the data,
365: so some aspect of the physics remains to be understood. In particular, with $n_0$
366: as an independently measured quantity rather than a free parameter, reanalysis of
367: the Chandra spectrum might be able to place better limits on the non-thermal
368: emission in the NW part of SN1006.
369:
370: Another density estimate
371: for the NW region of SN1006 was obtained by \citet{lam96}, who found that the
372: relative intensities of the UV lines could be explained if only about half the
373: O VI emission fell within the aperture of the HUT telescope. This would require
374: an ionization length for the O VI of about 10$^\prime$$^\prime$, or $n_0~\sim~0.04~\rm cm^{-3}$.
375: This interpretation is not consistent with the values of $n_0$ derived for the
376: same region from the HST image, so we conclude that there is a problem either
377: with the models of \citet{lam96}, or perhaps more likely with the reddening
378: correction in the far UV.
379:
380: A density estimate that is independent of SN1006 itself comes from H I 21 cm
381: observations by \citet{dub02}. An H I feature with a column density of
382: $7 \times 10^{20}~\rm cm^{-2}$ at a velocity of -6 $\rm km~s^{-1}$ lies
383: just outside the NW rim of SN1006. While the velocity does not correspond
384: to the Galactic rotation value at the distance to SN1006
385: ($-25~\le~V_{LSR}~\le~-16 ~\rm km~s^{-1}$), the SNR is far enough from the
386: plane that the velocity could easily differ. \citet{dub02} estimate a
387: neutral hydrogen density of 0.5 $\rm cm^{-3}$, which with the low neutral fraction
388: from \citet{gha02} would imply a total density of 5 $\rm cm^{-3}$, well above
389: the values we derive. We conclude that either SN1006 is not interacting with
390: the cloud identified by Dubner et al., or that the shock has only reached the
391: low density outskirts of the cloud.
392:
393: \subsection{Length scales}
394:
395: The length scale given by the radius of curvature is considerably
396: shorter than the length of the filament itself. The more or less
397: straight portion of the H$\alpha$ filament extends for perhaps 7$^\prime$,
398: or about 4 pc, roughly 100 times typical radius of curvature derived for the ripples.
399: The actual line of sight scale of the filament is several
400: times that of the bright rim at the trailing edge of the
401: filament. The region ahead of the trailing bright rim at
402: position 28 has a nearly constant surface brightness of
403: $2~-~3 \times 10^{-5}~\rm photons~cm^{-2}~s^{-1}~\arcsec^{-2}$.
404: For $n_0$ = 0.3, $f_{neut}$ = 0.1 and $V_s$ = 2900 $\rm km~s^{-1}$,
405: this implies an angle of 6 to 10 degrees between the shock and
406: the line of sight. The roughly constant intensity region is
407: about $2.8 \times 10^{17}$ cm in radial exent, so the range of angles
408: implies a depth along the line of sight between 1.5 and $2.3 \times 10^{18}$ cm,
409: or about a tenth the scale of the filament in the plane of the sky.
410: The $10^{18}$ cm scale corresponds to the smoother curve in the schematic
411: diagram in Figure~\ref{schem}, while the $10^{17}$ cm scale derived from
412: the H$\alpha$ brightness profiles corresponds to the smaller scale ripple
413: superposed to obtain the shock geometry shown in the schematic.
414:
415: Even the larger scale inferred for the direction along the line of sight
416: is several times smaller than the length of the filament in the plane of
417: the sky. Very faint H$\alpha$ emission,
418: which is not apparent in Figure~\ref{halpha}, is seen in the very
419: deep H$\alpha$ image of \citet{wink03} extending out ahead of the
420: bright filament (their Figure 5). This is probably a shock in lower density
421: gas that may be farther from tangency with the line of sight, and it probably
422: extends for a distance comparable to the length of the filament.
423:
424: Figure 2 shows that ripples on the scale of $10^{18}$ cm can be seen in
425: the direction along the filament, for instance between the two boxes where
426: radial profiles were extracted. Ripples on the $10^{17}$ cm scale are
427: not apparent, perhaps because a $10^{17}$ cm ripple with 10\% amplitude
428: would be only a few resolution elements radially in Figure 2. However, such
429: small scale rippling undoubtedly contributes to the widths of the H$\alpha$
430: brightness peaks and prevents them from being as sharp as the model peaks,
431: as is particularly apparent in the position 28 profile. It is also quite
432: likely that the ripples are not isotropic. If the magnetic field lies
433: near the plane of the sky in the NE-SW direction, then density structures
434: would be elongated in that direction and smaller scales would appear along
435: the line of sight. There is some evidence that the field does lie in this
436: direction based on the cap-like morphology of the non-thermal X-ray emission
437: \citep{wil96}, but there is also a suggestion that the field lies in
438: the SE-NW direction based on the relative temperatures of protons and oxygen
439: ions \citep{korreck}.
440:
441: If we attribute the $\sim~2 \times 10^{18}$ cm scale length of the ripples to
442: density fluctuations in the interstellar gas, we can estimate the
443: amplitude of the density fluctuation from the amplitude of the ripple.
444: The amplitude of the ripple is about one tenth the wavelength,
445: and that should be about equal to $\delta$V/V. For constant ram pressure,
446: $n V^2$, this requires density fluctuations of about 20\%. As was found
447: for similar ripples in the blast wave of the Cygnus Loop, this agrees reasonably well with
448: the expectations from the spectrum of interstellar turbulence \citep{minter, ray03}.
449: However, the smaller scale ripples revealed by the $10^{17}$ cm radius of
450: curvature also seem to have amplitudes of order 1/10 the wavelength, and
451: a Kolmogorov spectrum of density fluctuations from the turbulent cascade
452: would lead one to expect a smaller amplitude \citep{minter}. \citet{beresnyak}
453: show that the spectrum of density fluctuations may be considerably flatter than
454: the Kolmogorov spectrum.
455:
456: One feature in the filament is probably not due to density fluctuations in the
457: ISM, however. The bulge near the western edge of the filament in Figure~\ref{halpha}
458: is very similar to one farther west along the filament that coincides
459: closely with a bright knot of X-ray emission with enhanced elemental
460: abundances \citep{lon03, vink03}. The X-ray knot (position NW-2 of \citet{lon03})
461: is clearly a knot of ejecta overtaking the blastwave. The bulge at the SW
462: corner of the ACS image is probably a similar structure with lower X-ray contrast
463: due to its smaller size and perhaps differences in density, ionization time scale
464: and abundances. According to hydrodynamic simulations of
465: Type Ia SNRs, ejecta knots form at the Rayleigh-Taylor unstable contact discontinuity,
466: but they reach at most 87\% of the blast wave radius \citep{wang01}. Thus
467: the ejecta knots should be over a parsec, or about 2$^\prime$ behind the
468: Balmer filament. \citet{warren} have pointed out that if a substantial fraction
469: of the energy dissipated by a shock goes into accelerating cosmic rays, the
470: ejecta knots can come much closer to the outer shock. In the case of the
471: bulge in the southwestern corner of the HST image and the X-ray knot farther to the SW,
472: the lack of synchrotron emission
473: at radio and X-ray wavelengths indicates that little energy goes into cosmic rays.
474: Instead, since the blast wave encounted the dense gas in the NW sector of SN1006 only about
475: 180 years ago \citep{lon03}, the ejecta knots have undoubtedly overtaken the blast wave
476: because the blast wave decelerated when it encountered denser gas.
477: This has an important implication for analyses
478: of the fraction of shock energy that goes into cosmic rays. \citet{warren} show
479: that ejecta knots in Tycho's SNR come much closer to the blast wave than predicted
480: by \citet{wang01}, and they interpret this in terms of energy that goes into
481: accelerating particles to high energies. That argument
482: can be applied to most of Tycho's SNR, but
483: regions where the SNR shell is flattened or where the Balmer line emission is especially
484: bright should be avoided in the analysis, because those are regions where the shock has
485: probably been decelerated by higher density gas.
486:
487: \subsection{Shock Precursor}
488:
489: Shock wave precursors have been inferred from anomalously high line widths
490: of the narrow component H$\alpha$ emission in a number of SNRs and attributed
491: to heating in the precursor predicted by diffusive shock acceleration models
492: or to broad component hydrogen atoms that overtake the shock and heat the
493: upstream gas \citep{smi94, hes94}. The small width of the narrow
494: component in SN1006 means that there is no evidence for such a precursor,
495: but it is nevertheless worthwhile to place on limit on the emission from
496: such a precursor. One expects a more or less exponential falloff of brightness
497: ahead of the shock peak on a scale given by $\kappa / V_s$, where $\kappa$
498: is the comic ray diffusion coefficient, or by $(n_0 \sigma)^{-1}$ where
499: $\sigma$ is the charge transfer cross section. The former would be about
500: 1\arcsec\/ for $\kappa~\sim~10^{25}~\rm cm^2 s^{-1}$.
501: For a speed near 3000 $\rm km~s^{-1}$ and the densities in Table 1, the
502: latter would be about 0.6\arcsec. The brightness could be a significant
503: fraction of the narrow component brightness, or up to half the total
504: brightness of the filament.
505:
506: The concave outward ripples we have
507: analyzed so far are not appropriate for searching for a precursor, because
508: the emission from the curved part of the shock front could easily resemble
509: emission from a precursor. Instead the bright rims about 10\arcsec\/ ahead
510: are convex outward ripples, so that any emission ahead of the peak would be
511: due to a precursor. Spatial profiles do show shoulders of order 1\arcsec
512: wide ahead of the leading bright rims in several places. However, there is
513: additional faint emission out ahead of the main filament along the entire
514: region imaged (\citet{wink03}, Figure 5). That suggests that the leading
515: rim might be {\it S}-shaped along the line of sight rather than a convex
516: outward arc, so the $H\alpha$ could be similar to that in the models described
517: above (Figures 4 and 5). In particular, the sections of the filament that show
518: an H$\alpha$ shoulder ahead of the peak emission seem to the the regions where
519: the diffuse emission ahead of the filament is especially bright.
520: Therefore, we can only place an upper limit on the precursor just ahead
521: of the shock of about 1/3 the peak brightness. Stricter limits of about 1/10 the
522: peak brightness can be placed in some sections. Given the lack of evidence
523: for heating in a precursor in SN1006, this is not surprising. It does
524: suggest, however, that morphology alone may not be enough to establish the
525: existence of a precursor without some additional information such as [N II]
526: or [S II] line emission \citep{hes94}, spatial separation of narrow
527: and broad emission \citep{lee06}, superthermal [N II] line widths \citep{sol03},
528: or narrow component line widths of order 40 $\rm km~s^{-1}$.
529:
530: \subsection{Bulk velocity contribution to line widths}
531:
532: The widths of the H$\alpha$ and UV line profiles have been analyzed under the
533: assumption that the broadening due to bulk motions in parts of the filament
534: that are not quite tangent to the line of sight is small. The range
535: of angles between the line of sight and the shock surface derived
536: above for the region between the leading
537: and trailing rims implies doppler shifts up to 1/6 the post-shock speed,
538: or $V_s$/8. Making the extreme assumption that both red- and blue-
539: shifts are present, and ignoring the contribution of the bright rims
540: that are tangent to the LOS, the contribution of bulk motions as large
541: as 370 $\rm km~s^{-1}$ could be present, or a full width of 750 $\rm km~s^{-1}$.
542: With measured line widths of 2290 $\rm km~s^{-1}$ for H$\alpha$ \citep{gha02}
543: and 2100 to 2600 $\rm km~s^{-1}$ for the UV lines \citep{ray95, korreck},
544: bulk motions contribute at most 6\% to the measured line widths when added
545: in quadrature to the thermal widths.
546:
547: \section{Summary}
548:
549: An HST image of the Balmer line filament in SN1006 resolves the region where neutral
550: hydrogen is excited and ionized. The thickness of the ionization zone implies a
551: total density of 0.25 to 0.40$\rm cm^{-3}$, roughly in the middle of the order of magnitude
552: range of earlier density estimates. It agrees well with the density inferred by
553: \citet{lon03} from the ionization time scale of the X-ray emission just behind the
554: Balmer line filament, providing confirmation for the shock wave models used to interpret
555: the X-ray spectra. Ripples in the filament on parsec scales are consistent with
556: those expected for a shock propagating through a medium with $\sim$ 20\% density
557: fluctuations on a parsec scale, in accordance with measurements of turbulence
558: in the ISM. The ripples on $10^{17}$ cm scales have higher amplitude than expected for
559: a Kolmogorov spectrum of density fluctuations, however. A bulge in the H$\alpha$
560: filament corresponding to a knot of ejecta
561: shows that ejecta knots can overtake the outer SNR shock when the blast wave
562: decelerates upon encountering a dense cloud.
563:
564: The density estimates in this paper are based upon model calculations for the
565: emission from non-radiative
566: shocks. The models took recent improvements by \citet{hm} into account in an
567: approximate way, but a more complete model would be valuable. Models of the
568: 3D morphology of the filament that go beyond the simple shape assumed could improve
569: the accuracy of the derived parameters, especially when spatially resolved spectra become
570: available to pin down the additional parameters involved.
571:
572: \acknowledgments
573:
574:
575: This work is based on observations made with the NASA HST, and we are grateful
576: to Max Mutchler of STScI for help in using multidrizzle on the images. We thank
577: Dick McCray and the referee, Kevin Heng, for very helpful
578: comments, and Kevin Heng for computing the average velocity of the fast
579: neutrals relative to the shock. This work is supported
580: by HST Grant GO-10577.01-A to the Smithsonian Observatory.
581: This work made use of the NASA Astrophysics Data System (ADS).
582:
583: {\it Facilities:} \facility{HST}
584:
585: \begin{thebibliography}{}
586: \bibitem[Bamba et al.(2003)]{bam03} Bamba, A., Yamazaki, R., Ueno, M. \& Koyama, K.
587: 2003, \apj, 589, 827
588: \bibitem[Barnett (1990)]{redbook} Barnett, C.F. 1990, Atomic Data for Fusion. Volume 1:
589: Collisions of H, $H_2$, He and Li Atoms and Ions with Atoms and Molecules, Oak Ridge Report
590: ORNL-6086
591: \bibitem[Beresnyak \& Lazarian(2006)]{beresnyak} Beresnyak, A., \& Lazarian, A. 2006, astro-ph/0610766
592: %\bibitem[Cardelli, Clayton, Mathis(1989)]{ccm89}Cardelli, J. A.,
593: %Clayton, G. C., Mathis, J. S., 1989, \apj, 345, 245
594: %\bibitem[Chevalier, Kirshner, \& Raymond(1980)]{kcr} Chevalier,
595: %R. A., Kirshner, R. P., Raymond, J. C., 1980, \apj, 235, 186
596: \bibitem[Chevalier \& Raymond(1978)]{cr78}Chevalier, R. A.,
597: Raymond, J. C., 1978, \apj, 225, L27
598: \bibitem[Dubner et al.(2002)]{dub02} Dubner, G.M., Giacani, E.B., Goss, W.M., Green, A.J. \&
599: Nyman, L.-\AA , \aa, 387, 1047
600: \bibitem[Dyer et al.(2004)]{dye04} Dyer, K. K., Reynolds, S. P. \&
601: Borkowski, K. J., 2004, \apj, 600, 752
602: %\bibitem[Ellison \& Reynolds(1991)]{ell92} Ellison, D.C., Reynolds, S. P., 1991, \apj,
603: %382, 242
604: \bibitem[Fruchter \& Hook(2002)]{fruchter} Fruchter, A.S., \& Hook, R.N. 2002, \pasp, 114, 144
605: \bibitem[Ghavamian et al.(2002)]{gha02}Ghavamian, P., Winkler, P. F., Raymond, J. C. \& Long, K. S. 2002, \apj, 572, 999
606: \bibitem[Hamilton, Sarazin \& Szymkowiak(1986)]{hss86} Hamilton, A.J.S., Sarazin, C.L. \& Szymkowiak, A.E.
607: 1986, \apj, 300, 698
608: \bibitem[Heng \& McCray(2006)]{hm} Heng, K., \& McCray, R. 2006, \apj, in press
609: \bibitem[Hester(1987)]{hester} Hester, J.J. 1987, \apj, 314, 187
610: \bibitem[Hester, Raymond \& Blair(1994)]{hes94}Hester, J. J., Raymond, J. C. \& Blair,
611: W. P. 1994, \apj, 420, 721
612: %\bibitem[Henry \& Murthy(1993)]{hm93} Henry, R. C., Murthy,
613: %J., 1993, \apj, 418, L17
614: \bibitem[Jones \& Pye(1989)]{jp89} Jones, L. R. \& Pye, J. P. 1989, \mnras, 238, 567
615: \bibitem[Kirshner, Winkler \& Chevalier(1987)]{kwc87}Kirshner, R. P.,
616: Winkler, P. F. \& Chevalier, R.A. 1987, \apj, 315, L135
617: \bibitem[Koekemoer et al. (2002)]{koekemoer} Koekemoer A. M., Fruchter A. S., Hook R. N., Hack W., 2002, in
618: HST Calibration Workshop, S. Arribas, A.M. Koekemoer \& B. Whitmore, eds, (STScI, Baltimore), p. 325
619: \bibitem[Korreck et al. (2004)]{korreck} Korreck, K.E., Raymond, J.C., Zurbuchen, T.H. \&
620: Ghavamian, P. 2004, ApJ, 615, 280
621: \bibitem[Koyama et al.(1995)]{koyama} Koyama, K., Petre, R., Gotthelf, E.V., Hwang, U., Matsura,
622: M., Ozaki, K. \& Holt, S.S. 1995, Nature, 378, 255
623: \bibitem[Laming et al.(1996)]{lam96}Laming, J. M., Raymond, J.
624: C., McLaughlin, B. M. \& Blair, W. P. 1996, \apj, 472, 267
625: \bibitem[Lee et al. (2006)]{lee06} Lee, J.J., Koo, B.-C., Raymond, J.C., Ghavamian, P., Pyo, T.-S.,
626: Tajitsu, A. \& Hayashi, M. 2006, submitted to \apj
627: %\bibitem[Lim \& Raga (1996)]{lim96}Lim, A. J., Raga, A. C., 1996, \mnras, 280, 103
628: \bibitem[Long et al.(2003)]{lon03} Long, K. S., Reynolds, S. P.,
629: Raymond, J. C., et.al. 2003, \apj, 586, 1162
630: \bibitem[Minter \& Spangler(1997)]{minter} Minter, A.H. \& Spangler, S.R. 1997, ApJ, 485, 182
631: \bibitem[Moffett, Goss \& Reynolds(1993)]{moffett} Moffett, D.A., Goss, W.M. \& Reynolds, S.P. 1993, \aj,
632: 106, 1566
633: %\bibitem[Pye et al.(1981)]{pye81}Pye, J. P., Pounds, K. A., Rolf,
634: %D. P., Smith, A., Willingale, R., Seward, F. D., 1981, \mnras, 194,
635: %569
636: %\bibitem[Raymond et al. (1983)]{ray83} Raymond, J.C., XXXX
637: \bibitem[Raymond (2003)]{ray03} Raymond, J.C. 2003, Rev. Mex. A\&A, 15, 258
638: \bibitem[Raymond, Blair \& Long(1995)]{ray95}Raymond, J. C., Blair, W. P. \& Long,
639: K. S., 1995 \apj, 454, L31
640: %\bibitem[Reynolds(2004)]{rey04}Reynolds, S. P., 2004, Adv. Space Res., 33, 461
641: \bibitem[Reynolds(1996)]{rey96}Reynolds, S. P., 1996 \apj, 459, L13
642: %\bibitem[Reynolds \& Chevalier(1981)]{rey81}Reynolds, S. P., Chevalier, R. A., 1981,
643: %\apj, 245, 912
644: %\bibitem[Reynolds \& Gilmore(1986)]{rey86}Reynolds, S. P., Gilmore, D. M.,
645: %1986, \aj, 92, 1138
646: \bibitem[Reynolds \& Gilmore(1993)]{rey93}Reynolds, S. P. \& Gilmore, D. M.
647: 1986, \aj, 106, 272
648: \bibitem[Schweizer \& Middleditch(1980)]{sch80}Schweizer, F. \&
649: Middleditch, J. 1980, \apj, 241, 1039
650: \bibitem[Sirianni et al.(2005)]{sirianni} Sirianni, M., et al. 2005, \pasp, 117, 1049
651: \bibitem[Smith et al.(1991)]{smi91}Smith, R. C., Kirshner, R. P.,
652: Blair, W. P. \& Winkler, P. F. 1991, \apj, 375, 652
653: \bibitem[Smith et al.(1994)]{smi94}Smith, R. C., Raymond, J. C. \& Laming, J. M. 1994,
654: \apj, 420, 286
655: \bibitem[Sollerman et al. (2003)]{sol03} Sollerman, J., Ghavamian, P., Lundqvist, P. \&
656: Smith, R.C. 2003, A\&A, 407, 249
657: %\bibitem[Spitzer (1956)]{spi56}Spitzer Jr., L., 1956, 'Physics of Fully
658: %Ionized Gases', Interscience Publishers, Inc.
659: %\bibitem[Tennekes \& Lumley (1972)]{ten57}Tennekes, H., Lumley, J.L., 1972, 'A First
660: %Course in Turbulence', The MIT Press
661: %\bibitem[Tanimori(1998)]{tan98}Tanimori, T., 1998, \iaucirc, 188, 121
662: %\bibitem[Vasyliunas \& Siscoe (1976)]{val76}Vasyliunas, V. M., Siscoe, G. L., 1976,
663: %\jgr, 81, 1247
664: \bibitem[van den Bergh (1976)]{vdb} van den Bergh, S. 1976, \apjl, 208, L17
665: \bibitem[Vink et al.(2003)]{vink03}Vink, J., Laming, J. M., Gu, M. F.,
666: Rasmussen, A. \& Kaastra, J. S. 2003, \apj, 587, L31
667: \bibitem[Wang \& Chevalier(2001)]{wang01} Wang, C.-Y. \& Chevalier, R.A. 2001, \apj, 549, 1119
668: \bibitem[Warren et al.(2005)]{warren}Warren, J.S., et al. 2005, \apj, 634, 376
669: \bibitem[Willingale et al.(1996)]{wil96}Willingale, R., West, R. G.,
670: Pye, J. P., Stewart, G. C., 1996, \mnras, 278, 749
671: \bibitem[Winkler \& Long(1997)]{win97} Winkler, P.~F., \&
672: Long, K.~S.\ 1997, \apj, 491, 829
673: \bibitem[Winkler, Gupta \& Long(2003)]{wink03}Winkler, P. F., Gupta, G.,
674: Long, K. S. 2003, \apj, 585, 324
675: \end{thebibliography}
676:
677: \clearpage
678:
679: \input{tb1.tex}
680:
681: \clearpage
682:
683: %% No more than seven \figcaption commands are allowed per page,
684: %% so if you have more than seven captions, insert a \clearpage
685: %% after every seventh one.
686:
687: %% There must be a \figcaption command for each legend. Key the text of the
688: %% legend and the optional \label in curly braces. If you wish, you may
689: %% include the name of the corresponding figure file in square brackets.
690: %% The label is for identification purposes only. It will not insert the
691: %% figures themselves into the document.
692: %% If you want to include your art in the paper, use \plotone.
693: %% Refer to the on-line documentation for details.
694:
695: \begin{figure}
696: \plotone{f1.ps}
697: \end{figure}
698:
699: \begin{figure}
700: \plotone{f2.ps}
701: \end{figure}
702:
703: \begin{figure}
704: \plotone{f3.eps}
705: \end{figure}
706:
707: \begin{figure}
708: \plotone{f4.ps}
709: \end{figure}
710:
711: \begin{figure}
712: \plotone{f5.ps}
713: \end{figure}
714:
715: \begin{figure}
716: \plotone{f6.ps}
717: \end{figure}
718:
719: \clearpage
720:
721: \figcaption[f1.ps]{A three-color figure showing HST/ACS F658N (in red) and two energy bands
722: of Chandra data (0.4 - 0.7 keV in green and 0.8 - 1.5 keV in blue). The field of view
723: shown is 279\arcsec\ $\times$ 331\arcsec, with north up and east to the left. The X-ray
724: data have been binned by 4 pixels and smoothed with a 3-pixel Gaussian filter to remove
725: pixelation. Note the extended region of X-rays coincident with the bulge in the H$\alpha$
726: image. The drop-off in X-ray at upper left is due to a CCD-chip boundary and is not real.
727: \label{xrayhalpha}}
728:
729:
730: \figcaption[f2.ps]{The full field of view ACS F658N (H$\alpha$) image
731: of the NW Balmer filament in SN 1006. Two 10\arcsec\ $\times$
732: 20\arcsec\ boxes indicate the filament tangencies discussed in detail
733: in this paper.
734: \label{halpha}}
735:
736: \figcaption[f3.eps]{H$\alpha$ emissivity as a function of position behind the shock for
737: a non-radiative shock with $n_0$ = 0.25 $\rm cm^{-3}$, $f_{neut}$ = 0.1 and
738: $T_e /T_p$ = 0.05 at the shock. This model was computed under the assumption that charge
739: transfer is isotropic. To account for the effects described by \citet{hm} we stretched
740: out the broad component emission by a factor of 1.5 when comparing with the observations.
741: The scale assumes that 1\arcsec = $3.1\times 10^{16}$ cm for a distance of 2.1 kpc.
742: \label{model}}
743:
744: \figcaption[f4.ps]{Schematic diagram of the geometry of the emitting filament along the
745: line of sight (LOS). The light line shows a large scale ripple in the shock front,
746: and the darker curve is the result of adding a smaller scale ripple to the light line.
747: The fading of the dark curve towards the inside of SN1006 indicates the falloff
748: of emissivity with distance behind the shock (Figure~\ref{model}). The trace in
749: the lower right shows the H$\alpha$ brightness as a function of position for lines
750: of sight passing near the tangency to the LOS. It is obtained by integrating
751: the H$\alpha$ emissivity along each line of sight.
752: \label{schem}}
753:
754: \figcaption[f5.ps]{Comparison of observed and computed H$\alpha$ spatial profiles for position 10.
755: The shock wave is concave outward, with the assumed radius of curvature shown in each panel.
756: The inside of the remnant is to the right.
757: The dashed curves correspond to the pre-shock densities listed, with the lowest density giving
758: the highest curve on the inside and the lowest on the outside. All the models have been scaled
759: to the observed peak intensity by using the neutral fraction shown. Neutral fractions above
760: 1 are obviously unphysical, implying that that combination of pre-shock density and radius
761: of curvature cannot account for the observed brightness.
762: \label{obspred10}}
763:
764: \figcaption[f6.ps]{Same as Figure 4 for position 28.
765: \label{obspred28}}
766:
767:
768:
769:
770:
771:
772: \end{document}
773:
774: