1: %% AASTeX v5.x LaTeX 2e macros.

2: 

3: %% Produce a one-column, single-spaceddocument.

4: 

5: %\documentclass[apj]{emulateapj}

6: \documentclass[12pt,preprint]{aastex}

7: 

8: %% manuscript produces a one-column, double-spaced document:

9: 

10: %\documentclass[manuscript]{aastex}

11: 

12: %% preprint2 produces a double-column, single-spaced document:

13: 

14: %\documentclass[preprint2,longabstract]{aastex}

15: 

16: %Bibliographie suivant le style A&A

17: 

18: \usepackage{natbib}

19: \usepackage{color} % texte en couleur -> \textcolor{red}{texte}

20: \usepackage{ulem}  % texte barre -> \st{texte}; vague -> \uwave{texte}

21: \usepackage{psfrag}

22: 

23: \bibpunct{(}{)}{;}{a}{}{,}

24: 

25: \newcommand{\modif}{\bf }

26: 

27: \slugcomment{Accepted for publication in ApJ -- February 21, 2008}

28: 

29: \shorttitle{Azimuthal variations of the CR ion acceleration at the BW of SN 1006} 

30: 

31: \shortauthors{Cassam-Chena{\"\i} et al.}

32: 

33: %=====================================================================

34: 

35: \begin{document}

36: 

37: \title{Morphological evidence for azimuthal variations of the cosmic ray ion

38: acceleration at the blast wave of SN 1006}

39: 

40: \author{\mbox{Gamil Cassam-Chena{\"\i}\altaffilmark{1}, John P.

41: Hughes\altaffilmark{1}, Estela M. Reynoso\altaffilmark{2,3}, 

42: Carles Badenes\altaffilmark{4}, David Moffett\altaffilmark{5}}}

43: 

44: \altaffiltext{1}{Department of Physics and Astronomy, Rutgers

45: University, 136 Frelinghuysen Road, Piscataway, NJ 08854-8019;

46: \texttt{chenai@physics.rutgers.edu, jph@physics.rutgers.edu}} 

47: \altaffiltext{2}{Instituto de Astronom\'ia y F\'isica del

48: Espacio (IAFE), CC 67, Sucursal 28, 1428 Buenos Aires, Argentina}

49: \altaffiltext{3}{Departamento de F\'isica, Facultad de Ciencias Exactas y

50: Naturales, Universidad de Buenos Aires, Argentina}

51: \altaffiltext{4}{Department of Astrophysical Sciences, Princeton University,

52: Peyton Hall, Ivvy lane, Princeton, NJ 08544-1001}

53: \altaffiltext{5}{Department of Physics, Furman University, 

54: 3300 Poinsett Highway, Greenville, SC 29613} 

55: 

56: \begin{abstract}

57: Using radio, X-ray and optical observations, we present evidence for

58: morphological changes due to efficient cosmic ray ion

59: acceleration in the structure of the southeastern region of the

60: supernova remnant SN 1006. SN 1006 has an apparent bipolar morphology

61: in both the radio and high-energy X-ray synchrotron emission. In the

62: optical, the shock front is clearly traced by a filament of Balmer

63: emission in the southeast. This optical emission enables us to trace

64: the location of the blast wave (BW) even in places where the synchrotron

65: emission from relativistic electrons is either absent or too weak to

66: detect. The contact discontinuity (CD) is traced using images in the

67: low-energy X-rays (oxygen band) which we argue reveals the

68: distribution of shocked ejecta. We interpret the azimuthal variations

69: of the ratio of radii between the BW and CD plus the X-ray and radio

70: synchrotron emission at the BW using CR-modified hydrodynamic models.

71: We assumed different azimuthal profiles for the injection rate of

72: particles into the acceleration process, magnetic field and level of

73: turbulence.  We found that the observations are consistent with a

74: model in which these parameters are all azimuthally varying, being

75: largest in the brightest regions.

76: \end{abstract}

77: 

78: \keywords{ acceleration of particles --- cosmic rays --- shock waves

79: --- ISM: individual (SN 1006) --- supernova remnants }

80: 

81: 

82: %*******************************************

83: \section{Introduction\label{sect-intro}}

84: %*******************************************

85: 

86: Collisionless shocks in young supernova remnants (SNRs) are thought to

87: be responsible for the production and acceleration of the bulk of

88: Galactic cosmic rays (CRs)  at least up to the ``knee'' ($\sim 3$ PeV)

89: of the CR spectrum \citep[e.g.,][]{bev07}. The theoretical mechanism

90: is believed to be the first order Fermi acceleration, also known as

91: diffusive shock acceleration (DSA), where electrons, protons and other

92: ions scatter back and forth on magnetic fluctuations or Alfv\'en wave

93: turbulence through the velocity discontinuity associated with the

94: shock \citep[][and references therein]{joe91, mad01}.  The presence of

95: the turbulence is a key ingredient in the process.  The higher the

96: turbulence level, the higher the scattering rate and the faster the

97: acceleration proceeds, allowing particles to reach higher energies.

98: The fast acceleration implies also that the CR pressure becomes

99: significant very quickly so that the shock structure is highly

100: nonlinear \citep[e.g.,][]{elb00, ba07_NL}. The back-reaction of the

101: particles on the shock hydrodynamics increases the level of magnetic

102: turbulence and the injection rate in turn, i.e., the accelerated

103: particles create the scattering environment by themselves

104: \citep{be78a, blo78, vob03}. Moreover, the production of relativistic

105: particles (extracted from the thermal plasma) along with their escape

106: from the shock system increases the overall density compression ratio

107: \citep{bee99, bl02}.

108: 

109: This theoretical picture leads to several observational consequences.

110: The intensity of the synchrotron emission from shock-accelerated

111: electrons will be larger as the acceleration becomes more efficient

112: since both the number of relativistic particles and intensity of the

113: magnetic field must increase. In the region of very efficient

114: acceleration, the X-ray synchrotron emission from the highest energy

115: electrons will be concentrated in the form of thin sheets behind the

116: blast wave due to synchrotron cooling in the high turbulent magnetic

117: field amplified at the shock whereas the radio synchrotron emission

118: will be much broader in comparison \citep{elc05}. Finally, the

119: morphology of the interaction region between the blast wave (BW) and

120: the ejecta interface or contact discontinuity (CD) will become much

121: thinner geometrically as the acceleration becomes more efficient due

122: to the change in the compressibility of the plasma \citep{dee00,

123: eld04}. Because shocks are believed to put far more energy into ions

124: than electrons, this last point, if observed, would provide direct

125: evidence for the acceleration of CR ions at the BW.

126: 

127: The detection of $\gamma$-rays with a neutral pion-decay signature

128: resulting from the interaction of shock-accelerated protons with the

129: ambient matter would provide evidence for CR ion acceleration as well.

130: The best chance to see a clear pion-decay signal is when a SNR

131: interacts with a dense medium, the prototype case being RX

132: J1713.7--3946 \citep{cad04b}. Very high energy $\gamma$-rays have been

133: detected from this remnant \citep{aha07_RXJ1713} and also from a few

134: others \citep[e.g.,][]{aha07_RXJ0852,hol07_RCW86}. However, the

135: pion-decay signal is generally difficult to extract from these

136: observations because there are other processes efficiently producing

137: $\gamma$-rays but involving CR electrons, namely bremsstrahlung and

138: Inverse-Compton scattering off the 2.7 K cosmic microwave background

139: and Galactic photons. The nature of the underlying process leading to

140: $\gamma$-rays and therefore the nature of the emissive particles is

141: subject to intense debate \citep[e.g.,][]{bev06_RXJ1713, pom06,

142: uca07, pl08}. Besides, in a large number of SNRs where bright and

143: geometrically thin X-ray synchrotron-emitting rims are observed (e.g.,

144: Keper, Tycho, SN 1006), ambient densities are often very low leaving

145: open the question of CR ion production in these SNRs \citep[see

146: also][]{kaw08}.

147: 

148: The idea that the gap between the BW and CD allows quantifying the

149: efficiency of CR ion acceleration was proposed by \cite{de05} in the

150: case of Tycho, Kepler and Cas A. In Tycho, a remnant where thin X-ray

151: synchrotron-emitting rims are observed all around with little

152: variations in brightness, \citet{wah05} showed that the BW and CD are

153: so close to each other that they cannot be described by standard

154: hydrodynamical models (i.e., with no CR component), hence providing

155: evidence for efficient CR ion acceleration in this SNR.  In a later

156: study, \citet{cah07} showed that the X-ray and to some extent the

157: radio properties of the synchrotron emitting rim as well as the gap

158: between the CD and BW were consistent with efficient CR ion

159: acceleration modifying the hydrodynamical evolution of Tycho.

160: 

161: Ideally, one would like to find a SNR where both efficient and

162: inefficient particle acceleration take place so that differential

163: measurements can be made. This offers a great advantage for the

164: astrophysical interpretation since a number of uncertainties (e.g.,

165: distance and age) can be eliminated in the comparison.  Furthermore,

166: by directly observing a difference in the gap between the BW and CD in

167: regions of efficient and inefficient acceleration, one obtains a

168: direct, model-independent, confirmation that CRs do indeed modify the

169: hydrodynamical evolution of the BW. SN 1006 provides a valuable

170: laboratory in that regard since the gradual variations of the

171: intensity of the radio and X-ray synchrotron emission along the BW

172: point to underlying variation of CR ion acceleration.  The task

173: of measuring the ratio of radii between the BW and CD in

174: the regions of inefficient and efficient CR acceleration will be

175: challenging. Nonetheless, SN 1006 is one of the rare objects (perhaps

176: even the unique one) where this can be accomplished as we explain below.

177: 

178: In this paper, we focus our attention on the poorly studied

179: southeastern (SE) quadrant of SN 1006, i.e., along the shock front

180: lying between the northeastern and the southwestern

181: synchrotron-emitting caps. Our first goal is to measure the ratio of

182: radii between the BW and CD there. We can fairly easily follow the BW

183: in the bright synchrotron-emitting caps (presumably the regions of

184: efficient CR acceleration) and the BW can still be traced from the

185: H$\alpha$ emission in the region where both the radio and X-ray

186: synchrotron emissions are either too faint to be detected or absent

187: (presumably the regions of inefficient CR acceleration). As for the

188: CD, it can be traced from the thermal X-ray emission associated with

189: the shocked ejecta. Our second objective is to measure the azimuthal

190: variations of the synchrotron brightness in the radio and several

191: X-ray energy bands.  With these observational key results in hand, we

192: use CR-modified hydrodynamic models of SNR evolution to constrain the

193: azimuthal variations of the acceleration parameters, i.e., the

194: injection rate of particles in the acceleration process, strength of

195: the magnetic field, particle diffusion coefficient and maximum energy

196: of electrons and ions.

197: 

198: This paper is organized as followed. In \S \ref{sect-sn1006}, we

199: describe the basic characteristics of SN 1006. In \S \ref{sect-data},

200: we present the data used in our study. In \S \ref{sect-results}, we

201: present the key observational results in SN 1006. In \S

202: \ref{sect-models-vs_obs}, we try to relate hydrodynamical models with

203: those results in the context of CR-unmodified and -modified shocks. 

204: Finally, we discuss our results (\S \ref{sect-discussion}) and present

205: our conclusion (\S \ref{sect-conclusion}).

206: 

207: %*******************************************

208: \section{Basic characteristics of SN 1006\label{sect-sn1006}}

209: %*******************************************

210: 

211: SN 1006 was a thermonuclear supernova (SN) widely seen on Earth in the

212: year 1006 AD. More than a thousand year later, the remnant from this

213: explosion is a huge shell of $30\arcmin$ angular size.

214: 

215: The synchrotron emission is detected from the radio up to the X-rays

216: and dominates in two bright limbs -- the northeast (NE) and southwest

217: (SW) -- where several thin ($20\arcsec$ width or so) rims/arcs are

218: running at the periphery, sometimes crossing each other \citep{bay03,

219: lor03, rob04}. Those apparent ripples are most likely the result of

220: the projection of undulating sheets associated with the shock front.

221: The radio emission is well correlated with the nonthermal X-rays

222: \citep{reg86, lor03}. Both show an abrupt turn-on with coincident

223: edges in the radial directions, but the X-ray emission has more

224: pronounced narrow peaks. This points to a scenario in which the rims

225: are limited by the radiative losses\footnote{Note however that such a

226: model is incapable of reproducing the observed sharp turn-on of radio

227: synchrotron emission at SN 1006's outer edge, as in Tycho

228: \citep{cah07}.} in a highly turbulent and amplified magnetic field

229: \citep{bek02, bek03, elc05, ba06}.  Finally, the observed synchrotron

230: morphology in SN 1006 is best explained if the bright limbs are polar

231: caps with the ambient magnetic field parallel to the shock velocity

232: \citep{rob04}. In the caps, the maximum energy reached by the

233: accelerated particles, as well as their number, must be higher than

234: elsewhere \citep{rob04}.

235: 

236: In contrast to the nonthermal X-ray emission, the very faint thermal

237: X-ray emission seems to be distributed more or less uniformly

238: \citep{rob04}. This is best seen in the oxygen band (0.5-0.8 keV). It

239: is, however, difficult to separate the thermal emission from the

240: nonthermal emission in the caps. Moreover, it is not clear whether the

241: thermal X-rays should be attributed to the shocked ambient medium

242: \citep{yak07} or the shocked ejecta \citep{lor03}, but both the

243: over-solar abundances required to fit the X-ray spectra in the inner

244: northwest (NW) and NE parts of the SNR and the clumpiness on a

245: $30\arcsec$ to $1\arcmin$ scale of the low-energy X-ray emission favor

246: an ejecta origin. The SE region seems to be slightly

247: different than the rest of the SNR in terms of ejecta composition and

248: clumpiness too. There is, in particular, evidence for the presence of

249: cold \citep{wil05} and reverse-shock heated \citep{yak07} iron. In

250: terms of dynamics, there is evidence for ejecta extending

251: to/overtaking the BW in the NW \citep{lor03, vil03b_SN1006, rak07}.

252: 

253: In the optical, the remnant has a very different morphology.  Deep

254: H$\alpha$ imaging reveals very faint Balmer emission around almost the

255: entire periphery with a clear arc or shock front running from east to

256: south \citep{wil97b_SN1006, wig03}. Interestingly, no synchrotron

257: emission is detected in X-ray in this southeast (SE) region whereas

258: extremely faint (and highly polarized) synchrotron emission is

259: detected in the radio \citep{reg93}. Besides, none of the optical

260: emission is of synchrotron origin. There is also a clear Balmer line

261: filament in the NW of SN 1006 which has been observed in great detail

262: in the optical \citep{ghw02_SN1006, sog03, rak07}, ultraviolet

263: \citep{rab95, kor04, haf07} and X-ray bands \citep{lor03,

264: vil03b_SN1006, acb07}, providing diagnostics for the shock speed ($V_s

265: \sim 2400-3000\: \mathrm{km/s}$), ion-electron thermal equilibration

266: at the shock ($T_e/T_p \le 0.07$) and preshock ambient density ($0.25

267: \: \mathrm{cm}^{-3} \le n_0 \le 0.4 \: \mathrm{cm}^{-3}$). Optical

268: proper motions \citep{wig03} combined with the shock speed estimate

269: \citep{ghw02_SN1006, hem07} along this NW filament leads to a distance

270: to the SNR between $1.8$ and $2.3$ kpc. However, because the remnant

271: is interacting with a denser medium in the NW \citep{mog93, dug02},

272: the shock speed and the ambient density, in particular, might be

273: respectively higher and lower elsewhere \citep{acb07}. For instance,

274: the broadening of the oxygen \citep{vil03b_SN1006} and silicon

275: \citep{yak07} X-ray emission-lines suggest a shock speed $V_s \ge 4000

276: \: \mathrm{km/s}$ (although these measurements were carried out at

277: different locations in SN 1006). Such high velocities are compatible

278: with recent proper motion measurements based on radio data

279: \citep{moc04, rec08}. Moreover, the lack of observed TeV $\gamma$-ray

280: emission \citep{aha05_SN1006} sets an upper limit on the ambient

281: density of $n_0 \le 0.1 \: \mathrm{cm}^{-3}$ \citep{ksb05} and the

282: high latitude ($\sim 500 \: \mathrm{pc}$ at a distance of $2$ kpc) of

283: SN 1006 suggests that $n_0 \sim 0.03-0.04 \: \mathrm{cm}^{-3}$

284: \citep{fe01_ISM}.  

285: 

286: %*******************************************

287: \section{Data\label{sect-data}}

288: %*******************************************

289: 

290: \subsection{X-ray\label{subsect-data-xray}}

291: 

292: We used the most recent \textit{Chandra} data of SN 1006 (ObsId 3838

293: and ObsId 4385 up to 4394) which were obtained in April 2003 with the

294: ACIS-I imaging spectrometer in timed exposure and very faint data

295: modes. Eleven pointings were necessary to cover the entire extent of

296: the remnant. The final image is shown in Figure

297: \ref{fig_SN1006_3color_xray}. The X-ray analysis was done using CIAO

298: software (ver. 3.4). Standard data reduction methods were applied for

299: event filtering, flare rejection, gain correction. The final exposure

300: time amounts to $\sim 20$ ks per pointing.

301: 

302: \subsection{Radio\label{subsect-data-radio}}

303: 

304: Radio observations were performed in 2003 (nearly at the same

305: time as the X-ray observations) with both the Australia Telescope

306: Compact Array (ATCA) and the Very Large Array (VLA).

307: 

308: The ATCA consists of an array of six 22-m antennae that attain a

309: maximum baseline of 6 km in the East-West direction.  The ATCA

310: observations occurred on three separate occasions in antenna

311: configurations that optimize high spatial resolution: January 24 in

312: its 6-km B configuration for 12 hours, March 3 in its 6-km A

313: configuration for 12 hours, and June 12 in its 750-m C configuration

314: for 7 hours. Observations were made using two 128-MHz bands (divided

315: into 32 channels each) centered at 1384 and 1704 MHz.  The source PKS

316: 1934--638 was used for flux density and bandpass calibration, while PKS

317: 1458--391 was used as a phase calibrator.  The total integration time

318: on SN 1006 was over 1500 minutes.

319: 

320: The VLA consists of 27 25-m antennae arranged in a ``Y'' pattern.

321: Observations with the VLA were carried out during 4 hours on January

322: 25th in its hybrid CnD configuration (arranged to maximize spatial

323: resolution when observing southern declination sources).  Observations

324: were made using two 12.5-MHz channels centered at 1370 and 1665 MHz.

325: 3C286 and 1451--400 were observed for flux and phase calibration.  The

326: total VLA integration time on SN 1006 was 140 minutes.

327: 

328: Data from all of the observations were combined, then uniformly

329: weighted during imaging to minimize the effects of interferometric

330: sidelobes.  The final image, shown in Figure \ref{fig_SN1006_radio},

331: has a resolution of $\sim 6\arcsec \times 9\arcsec$ and an

332: off-source rms noise of $20 \: \mu\mathrm{Jy/beam}$ near the rims.

333: Since the angular size of SN 1006 is comparable to the primary beam of

334: both instruments (half a degree), corrections were applied to the

335: image to recover the lost flux. After primary beam corrections, the

336: total flux density was recovered to within $< 5\%$ of the expected value. 

337: With primary beam correction, noise increases in the faint SE portion.

338: For the study of the azimuthal variations of the radio emission at the shock, 

339: we use non beam corrected data (\S \ref{subsubsect-azimu-synch-emiss}).

340: 

341: \subsection{Optical\label{subsect-data-optical}}

342: 

343: We used the very deep H$\alpha$ image presented by \cite{wig03}.  This

344: image taken in June 1998, that is 5 years before the X-ray data, is shown

345: in Figure \ref{fig_SN1006_Halpha}.

346: To compare the optical and X-ray images, we had to correct for the

347: remnant's expansion. In the optical, proper motions of $0.280\pm0.008

348: \: \arcsec \: \mathrm{yr}^{-1}$ were measured (from April 1987 to June

349: 1998) along the NW rim where thin nonradiative Balmer-dominated

350: filaments are seen \citep{wig03}. In the radio, an overall expansion

351: rate of $0.44\pm0.13 \: \arcsec \: \mathrm{yr}^{-1}$ was measured

352: \citep[from May 1983 to July 1992,][]{mog93}, although higher values

353: have been measured using more recent observations \citep{moc04,

354: rec08}. This value does not include the NW rim where the optical

355: filaments are observed because there is simply little or no radio

356: emission there.  The observed expansion rate is clearly higher in the

357: radio than in the optical and is consistent with the picture in which

358: the BW encounters a localized and relatively dense medium in the NW. 

359: For simplicity, when using radial profiles, we use the value of

360: $0.40\arcsec \: \mathrm{yr}^{-1}$, so that the optical profiles are

361: shifted by $2\arcsec$ to be compared with the X-ray profiles. As we

362: show below, this correction is negligible compared to other

363: uncertainties when trying to locate the fluid discontinuities.

364: 

365: %*******************************************

366: \section{Key observational results\label{sect-results}}

367: %*******************************************

368: 

369: \subsection{Fluid discontinuities\label{sub-sect-radii-SE}}

370: 

371: In this section, we determine the location of the fluid

372: discontinuities -- i.e., the BW and CD -- in the SE quadrant of SN

373: 1006. Our goal is to measure the gap between the BW and CD

374: as a function of azimuth.

375: 

376: To trace the location of the BW, we use the H$\alpha$ image of SN 1006

377: which shows very faint filaments of Balmer emission around much of the

378: periphery of the remnant (Fig.~\ref{fig_SN1006_all}, \textit{top-right

379: panel}).  In the southeastern quadrant, the H$\alpha$ emission follows

380: a nearly circular arc.  We extracted radial profiles by azimuthally

381: summing over $4^{\circ}$ wide sectors starting from $158^{\circ}$

382: (from west) to $302^{\circ}$ (in the clockwise direction).  The center

383: of these profiles was determined to provide the best match to the

384: curvature of the H$\alpha$ rim.  The coordinates of the center are

385: $(\alpha_{\mathrm{J2000}}, \delta_{\mathrm{J2000}}) =

386: (15^{\mathrm{h}}02^{\mathrm{m}}56.8^{\mathrm{s}},-41^\circ 56\arcmin

387: 56.6\arcsec)$. This center is very close to the geometrical center of

388: the SNR, which, given its overall circularity, may not be too far from

389: the overall expansion center which has not yet been determined. In

390: practice we found it very difficult to determine the location of the

391: optical rim based purely on a numerical value (e.g., contour value or

392: enhancement factor above the local background level) because of the

393: presence of faint stellar emission, poorly subtracted stars, and faint

394: diffuse H$\alpha$ emission across the image.  So, we identified

395: locations where the H$\alpha$ emission increased slightly in the

396: radial profiles by eye and then further checked (again visually) the

397: corresponding radii on the optical image. We associated generous

398: uncertainties with the radii derived from this procedure: $12 \arcsec$

399: outward and $24 \arcsec$ inward (our process tended to overestimate

400: the radii, hence the asymmetric errors).

401: 

402: To trace the location of the CD, we use the X-ray image in the

403: low-energy band ($0.5-0.8$ keV) which contains most of the oxygen

404: lines (Fig. \ref{fig_SN1006_all}, \textit{bottom-left panel}). While

405: it is not clear whether the oxygen emission comes from the shocked

406: ambient medium or the shocked ejecta, we consider here an ejecta

407: origin. We discuss and justify this assumption below (\S

408: \ref{subsect-oxygen-ejecta-or-ISM}). Another issue comes from the fact

409: that the X-ray emission in the oxygen band may contain some nonthermal

410: contribution in the bright limbs. The three-color composite X-ray

411: image (Fig. \ref{fig_SN1006_3color_xray}) reveals that the oxygen emission

412: (red color) is in fact fairly different from the nonthermal X-rays

413: (white color), and appears notably clumpier even in the synchrotron

414: limbs (see the eastern limb for instance). This tells us that we can

415: still use the oxygen emission to trace the CD in the bright limbs, at

416: least in the azimuthal range we selected ($158^{\circ}-302^{\circ}$)

417: where the synchrotron emission is not completely overwhelming.  Radial

418: profiles were extracted from a flux image, summing over $1^{\circ}$

419: wide sectors, with the same center and azimuthal range as for the

420: optical. In the radial profiles, we selected the radii for which the

421: brightness becomes larger than $1.5 \times 10^{-5}$

422: ph/cm$^2$/s/arcmin$^2$.

423: 

424: Figure \ref{fig_radius_Ha_xray} summarizes the above-described

425: measurements and shows the azimuthal variations of the radii as

426: determined from the H$\alpha$ emission (\textit{black lines}) and

427: low-energy ($0.5-0.8$ keV, \textit{red lines}), mid-energy ($0.8-2.0$

428: keV, \textit{green lines}) and high-energy ($2.0-4.5$ keV,

429: \textit{blue lines}) X-ray emissions. The cross-hatched lines

430: indicate the location of the bright synchrotron limbs. They correspond

431: to the places where the contour of the outer high-energy X-ray

432: emission (\textit{blue lines}) can be determined. Between the two

433: synchrotron caps (angles between $200^\circ$ and $270^\circ$), there

434: is a clear gap between the BW and CD which is easily seen in Figure

435: \ref{fig_SN1006_all} (\textit{bottom-left} panel).  

436: In the bright limbs, however, the BW and CD are

437: apparently globally coincident.  Fingers of ejecta (indicated by the

438: \textit{stars}) are even sometimes visible clearly ahead of the

439: H$\alpha$ emission. In these places, the H$\alpha$ emission might not

440: be the best tracer of the BW location. The mid-energy X-ray emission

441: (\textit{green lines}) shows that these fingers are also present.  It

442: is not clear whether this emission traces the BW or, again, the ejecta

443: (mostly silicon).  The high-energy X-ray emission (\textit{blue

444: lines}) cannot be used, nor a X-ray spectral analysis, to answer this

445: point because of the low number of X-ray counts in this region.

446: 

447: Figure \ref{fig_Rs_o_Rc} shows the ratio of radii for the H$\alpha$

448: and low-energy X-ray emission, which to first approximation

449: corresponds to the ratio of radii between the BW and CD, $R_{\rm

450: BW}/R_{\rm CD}$.  With increasing azimuthal angles the ratio of radii

451: increases from values near unity (within the northeastern

452: synchrotron-emitting cap) to a maximum of

453: $R_{\mathrm{BW}}/R_{\mathrm{CD}} \simeq 1.10^{+0.02}_{-0.04}$ before

454: falling again to values near unity (in the southwestern

455: synchrotron-emitting cap). Over the entire azimuthal region where the

456: synchrotron emission is faint, the azimuthally averaged ratio of radii

457: is $R_{\rm BW}/R_{\rm CD} \simeq 1.04\pm0.03$. In the regions within

458: the synchrotron rims, $R_{\rm BW}/R_{\rm CD} \simeq 1.00$.

459: 

460: \subsection{Synchrotron emission at the blast wave\label{subsubsect-azimu-synch-emiss}}

461: 

462: In this section, we measure the azimuthal variations of the

463: synchrotron flux extracted at the BW and we investigate how these

464: variations compare when measured at different frequencies.

465: 

466: For that purpose, we use radio and X-ray data (Fig.

467: \ref{fig_SN1006_all}, \textit{left panels}).  We extracted the flux

468: in a $30\arcsec$-wide region behind the BW along the SE rim. To define

469: the BW location, we used the well-defined radii obtained from the

470: high-energy X-ray image for the northeastern ($\theta \leq 200^\circ$)

471: and southwestern ($\theta \geq 270^\circ$) regions while, in the

472: remaining SE quadrant ($200^\circ < \theta < 270^\circ$), we used the

473: radii determined from the H$\alpha$ image since no synchrotron

474: emission is detected there.

475: 

476: In Figure \ref{fig_flux_vs_angle_radio_xrays} (\textit{top panel}), we

477: show the azimuthal variations of the projected brightness in the radio

478: (1.5 GHz, \textit{black lines}), and several X-ray energy bands:

479: $0.5-0.8$ keV (\textit{red lines}), $0.8-2.0$ keV (\textit{green

480: lines}) and $2.0-4.5$ keV (\textit{blue lines}). The radio brightness

481: was multiplied by $2 \times 10^{-5}$ to make it appear on the same

482: plot with the X-ray data.  When moving from the northeastern or

483: southwestern synchrotron-emitting caps toward the faint SE region, we

484: see that the projected brightness gradually decreases in both radio

485: and X-ray bands. The radio, medium- (\textit{green lines}) and

486: high-energy (\textit{blue lines}) X-rays show mostly the variations of

487: the synchrotron emission, while the low-energy X-rays (\textit{red

488: lines}) show mostly the variations of the thermal (oxygen) emission,

489: but may contain some nonthermal contribution as well.

490: 

491: In Figure \ref{fig_flux_vs_angle_radio_xrays} (\textit{bottom panel}),

492: we show the same azimuthal variations but rescaled roughly to the same

493: level in the synchrotron-emitting caps. The logarithmic scale shows

494: that the brightness contrast between the bright caps and faint SE

495: region is of order $20$ to $50$ in the radio and X-rays. In fact, this

496: value might even be larger since the radio and high-energy X-ray

497: fluxes in the faint SE quadrant are compatible with no emission.  We

498: note that the azimuthal variations of the radio synchrotron emission

499: are much less pronounced than those in the X-rays. 

500: 

501: %*******************************************

502: \section{Relating hydrodynamical models to the observations\label{sect-models-vs_obs}}

503: %*******************************************

504: 

505: In the previous section, we have found evidence for azimuthal

506: variations of the separation between the BW and CD and synchrotron

507: brightness just behind the BW (\S \ref{sect-results}). Both the

508: separation and the brightness are correlated: the brighter the BW, the

509: smaller the separation. We measured in particular an unexpectedly small

510: separation between the BW and CD ($\sim 1.04$) in the SE quadrant of

511: SN 1006 where little to no synchrotron emission is detected.  These

512: measurements depend on the assumption that the oxygen emission comes

513: from the shocked ejecta.  After justifying this approach below (\S

514: \ref{subsect-oxygen-ejecta-or-ISM}), we first investigate the

515: different scenarios that could potentially lead to a small ratio

516: of radii between the BW and CD assuming no CR acceleration at the

517: shock (\S \ref{subsect-no-CR-accel}). Finally, we interpret the

518: observed azimuthal variations of the ratio of radii and synchrotron

519: brightness using CR-modified hydrodynamic models (\S

520: \ref{subsect-CR-accel}).

521: 

522: \subsection{Does the oxygen come from the ejecta?\label{subsect-oxygen-ejecta-or-ISM}}

523: 

524: A strong assumption made in the measurement of the ratio of radii

525: between the BW and CD in SN 1006 is that the thermal X-ray emission

526: (from the oxygen) is associated with the shocked ejecta.  

527: 

528: There is, however, still the possibility that the oxygen emission is

529: associated with the shocked ambient medium. \cite{yak07} have shown

530: that the integrated X-ray spectrum of the SE quadrant can be described

531: by a combination of three thermal plasmas in non-equilibrium

532: ionization and one power-law component.  One of the thermal

533: components, assumed to have solar abundances and therefore associated

534: with the shocked ambient medium, was able to produce most of the

535: observed low-energy X-rays (in particular the K$\alpha$ lines from O

536: \textsc{vii}, O \textsc{viii} and Ne \textsc{ix}).  The two other

537: thermal components, with non-solar abundances, were able to reproduce

538: most of the K$\alpha$ lines of Si, S and Fe, and were attributed to

539: the shocked ejecta.

540: 

541: There are several arguments from an examination of the high resolution

542: \textit{Chandra} images, however, that favor an ejecta origin for the

543: low-energy X-ray emission.  The three-color composite X-ray image

544: (Fig.~\ref{fig_SN1006_3color_xray}) shows that the spectral character

545: of the X-ray emission in the SE quadrant (where there is little to no

546: synchrotron emission) does not change appreciably as a function of

547: radius behind the BW, which would be expected if the \cite{yak07}

548: picture were correct. The lack of virtually any X-ray emission between

549: the BW and CD can be explained by the expected low density of the

550: ambient medium \citep[$\sim 0.03-0.04$ cm$^{-3}$][]{fe01_ISM} which

551: can reduce both the overall intensity and the level of line emission

552: (due to strong nonequilibrium ionization effects). If the soft X-ray

553: emission were from the shocked ambient medium its brightness should in

554: principle rise gradually behind the BW as a result of the projection

555: of a thick shell onto the line-of-sight \citep{wah05}, but the radial

556: profile of the low-energy X-ray emission (not shown) in fact turns on

557: rather quickly at the periphery.

558: 

559: Another argument in favor of an ejecta origin is the observed clumpiness

560: of the oxygen emission throughout the SE region. There is also the

561: protuberances seen right at the edge of SN 1006, which are suggestive

562: of Rayleigh-Taylor hydrodynamical instabilities that are expected at

563: the CD (see Fig. \ref{fig_SN1006_3color_xray}). Note that the outer edge of

564: the long, thin filament of X-ray emission in the NW -- which overlaps

565: the brightest H$\alpha$ emission and therefore is likely to be at least

566: partly due to shocked ambient medium -- is much smoother than the edge of the

567: remnant in the SE. Finally, high emission measures of oxygen in the

568: shocked ejecta at young dynamical SNR ages are compatible with most

569: thermonuclear explosion models. Essentially, all 1-D and 3-D

570: deflagration models, and all delayed detonation and pulsating delayed

571: detonation models have oxygen in the outer layers (but not prompt

572: detonations or sub-Chandrasekhar explosions) \citep{bab03}.

573: 

574: 

575: \subsection{The small gap in the faint SE region\label{subsect-no-CR-accel}}

576: 

577: We have presented evidence in \S \ref{sub-sect-radii-SE} for a small

578: ratio of radii $R_{\mathrm{BW}}/R_{\mathrm{CD}} \sim 1.04$ between the

579: BW and CD in the SE quadrant of SN 1006.  This is also the location

580: where the radio and X-ray nonthermal emission are either very weak or

581: undetected.  The most straightforward explanation for this lack of

582: synchrotron emission is an absence of efficient CR acceleration at the

583: BW here. In the following sections, we first investigate whether pure

584: 1-D hydrodynamical models for SNR evolution (i.e., without CR

585: acceleration) can reproduce such a small ratio of radii.  We consider

586: the role of ambient density, ejecta profile, and explosion energy on

587: the size of the gap between CD and BW.  Then we consider the effects

588: of 3-D projection on the hydrodynamically unstable CD.

589: 

590: \subsubsection{Predictions from standard hydrodynamical models\label{subsubsect-hydro-simu-no-CRs}}

591: 

592: To understand how a small gap can be produced in a region where there

593: is apparently no evidence for efficient particle acceleration, we used

594: standard one dimensional (1-D) spherically symmetric numerical

595: hydrodynamical simulations that follow the interaction of the ejecta

596: with the ambient medium. These simulations do not include any CR

597: component. 

598: 

599: Key inputs to these simulations are the initial density profiles of

600: the ejecta and ambient medium. In the following, we consider a uniform

601: density in the ambient medium and two different initial ejecta density

602: profiles: exponential and power-law. The exponential form has been

603: shown to be most representative of explosion models for thermonuclear

604: SNe \citep{dwc98}, but we also present results with a power-law

605: distribution to quantify the impact of the shape or compactness of the

606: ejecta profile on $R_{\mathrm{BW}}/R_{\mathrm{CD}}$.

607: 

608: In general, an exponential profile is expected for thermonuclear SNe

609: because the explosion is driven by the continuous release of energy

610: from the burning front as it propagates through the star, while the

611: power-law profile (more compact) is expected for core collapse SNe

612: because the explosion is driven by a central engine (core bounce), and

613: the shock loses energy as it propagates through the star

614: \citep{mam99}. Power law profiles with $n=7$ have been used to

615: represent 1-D deflagration models \citep[in particular, model W7

616: from][]{not84}, while exponential profiles are more adequate to

617: represent delayed detonation models \citep{dwc98}. When comparing

618: these two analytical profiles, it is worth noting that they transmit

619: momentum to the ISM in a different manner. The more compact power law

620: profile is a more efficient piston for ISM acceleration, and will lead

621: to smaller gaps than the less compact exponential profile

622: \citep{dwc98, bab03}.

623: 

624: The hydrodynamical simulations provide the radii of the BW and CD at

625: any time for a given ejecta profile, ambient medium density, and

626: explosion energy. In Figure \ref{fig_n0_vs_ratio}, we quantify the

627: impact of these three contributions on the ratio of radii

628: $R_{\mathrm{BW}}/R_{\mathrm{CD}}$. In particular, we plot the ratio

629: obtained for a wide range of ambient medium density ($0.001 \:

630: \mathrm{cm}^{-3} \leq n_0 \leq 30 \: \mathrm{cm}^{-3}$) using 1-D

631: exponential (EXP) and powerlaw (PL7) ejecta profiles with different

632: kinetic energies. The cross-hatched domains define the range of ratio

633: of radii\footnote{In Figure \ref{fig_n0_vs_ratio}, it is important to

634: keep in mind that the maximum value of the ratio found in Tycho comes

635: from a presumably efficient particle acceleration shock region while

636: in SN 1006, it comes from an inefficient particle acceleration

637: region.} observed in SN 1006 and Tycho

638: ($R_{\mathrm{BW}}/R_{\mathrm{CD}} \leq 1.10$) and range of typical

639: ambient densities ($0.01 \: \mathrm{cm}^{-3} \leq n_0 \leq 0.06 \:

640: \mathrm{cm}^{-3}$ in SN 1006; $0.1 \: \mathrm{cm}^{-3} \leq n_0 \leq

641: 0.6 \: \mathrm{cm}^{-3}$ in Tycho) consistent with constraints derived

642: from the observations \citep{acb07, cah07}.  It is clear that the 1-D

643: hydrodynamical simulations are unable to reproduce a ratio of radii as

644: small as the one observed in both SNRs.  We note however that such a

645: comparison between the observations and the models is not

646: straightforward since our measurements are likely to be affected by

647: projection and other effects as we detail next.

648: 

649: \subsubsection{Three dimensional projection effects\label{subsubsect-geom-proj}}

650: 

651: One of the limitations of our numerical simulations is the one

652: dimensionality.  Hydrodynamical simulations in 2-D or 3-D show that

653: hydrodynamical (Rayleigh--Taylor) instabilities project pieces of

654: ejecta ahead of the 1-D CD.  In these simulations, the outermost

655: pieces of ejecta reach half way of the gap between the 1-D CD and BW

656: \citep[e.g.,][]{chb92, wac01}. Therefore, because of such protrusions,

657: the line-of-sight projected CD radius, $\hat{R}_{\mathrm{CD}}$, will

658: appear larger than the true average CD radius, $R_{\mathrm{CD}}$.  On

659: the other hand, because the BW is not as highly structured as the CD

660: interface (it is not subject to Rayleigh--Taylor instabilities), it is

661: reasonable to assume that the true average BW radius,

662: $R_{\mathrm{BW}}$, is very close to the projected value,

663: $\hat{R}_{\mathrm{BW}}$.  It follows that the ratio of projected radii

664: between the BW and CD, $\hat{R}_{\mathrm{BW}} /

665: \hat{R}_{\mathrm{CD}}$, will be smaller than the true average ratio,

666: $R_{\mathrm{BW}} / R_{\mathrm{CD}}$. Quantifying the effects due to

667: projection is of considerable importance for interpreting the ratio of

668: radii between the BW and CD in the context of particle

669: acceleration\footnote{It is not even clear how the 1-D CR-modified

670: hydrodynamical results on the ratio of radii need to be adjusted given

671: multi-D Rayleigh--Taylor instability effects. Preliminary studies on

672: this have been done by \cite{ble01} but this work does not fully

673: quantify all relevant effects.}.

674: 

675: There are several attempts aimed at estimating the projection

676: correcting factor, $\xi$, where $\xi$ is defined as

677: $\hat{R}_{\mathrm{CD}} = (1 + \xi) \: R_{\mathrm{CD}}$. Based on the

678: projection with ejecta protrusions determined from a power-spectrum

679: analysis done at the observed CD in Tycho, \cite{wah05} estimated an

680: amount of bias of $\xi_{\mathrm{proj}} \simeq 6\%$. Based on the

681: projection of a shell of shocked ejecta with protrusions calculated in

682: a 2-D hydrodynamical simulations, \cite{dw00} and \cite{wac01} found a

683: slightly larger value of $\xi_{\mathrm{proj}} \simeq 10\%$. Taking

684: $\xi=10\%$, the predicted ratio of projected radii,

685: $\hat{R}_{\mathrm{BW}} / \hat{R}_{\mathrm{CD}} = R_{\mathrm{BW}} /

686: R_{\mathrm{CD}} \: / \: (1 + \xi)$, becomes $1.18 / (1+0.1) \simeq

687: 1.07$, where $1.18$ is the ratio of radii found at an age of 1000

688: years using 1-D power-law ejecta profiles assuming a kinetic energy of

689: the explosion of $10^{51}$ ergs and an ambient medium density of $0.03

690: \: \mathrm{cm}^{-3}$. Using exponential ejecta profiles, we find

691: $\hat{R}_{\mathrm{BW}} / \hat{R}_{\mathrm{CD}} =1.14$ because they

692: produce higher ratio of radii (i.e., $R_{\mathrm{BW}} /

693: R_{\mathrm{CD}}=1.25$). In either case (power-law or exponential

694: ejecta profiles), the average ratio (resp. 1.07 or 1.14) is still

695: larger than the average value of 1.04 measured in the SE quadrant of

696: SN 1006 (cf. \S \ref{sub-sect-radii-SE}). 

697: 

698: There are several possibilities to explain such discrepancy between

699: the models and the observations. For instance, the value of

700: $\xi_{\mathrm{proj}}$ could be larger in 3-D than in 2-D. Indeed,

701: models show that hydrodynamical instabilities can grow considerably

702: faster (by $\sim 30\%$) and penetrate further in 3-D than in 2-D

703: \citep{kaa00}. There can also be a certain amount of inhomogeneity in

704: the ejecta density distribution. Spectropolarimetric observations of

705: thermonuclear SNe do show that ejecta are clumpy on large scales

706: \citep{lel05}. Besides, \cite{wac01} have shown that high density

707: clump traveling with high speed in the diffuse SN ejecta can reach and

708: perturb the BW. This requires a high density contrast of a factor 100.

709: The density contrast is certainly not as high in SN 1006 because

710: otherwise we would see it in the low-energy X-ray and/or in the

711: optical image, but may be non-negligible. According to the previous

712: calculation, the ejecta clumping is required to contribute an

713: additional $\xi_{\mathrm{clump}} \sim 3-10\%$ factor (so that $\xi =

714: \xi_{\mathrm{proj}}+\xi_{\mathrm{clump}}$) in order to make the

715: observed and predicted ratios of radii agree. Finally, although we

716: could not come up with a reasonable solution, we cannot exclude the

717: possibility of a complex 3-D geometry in the SE quadrant of SN 1006

718: where the outermost extent of the shocked ejecta would not be directly

719: connected to the shock front emission in the hydrodynamical sense

720: (i.e., the CD and BW would corresponds to different parts of the

721: remnant seen in projection).

722: 

723: \subsection{Azimuthal variations of the CR ion acceleration\label{subsect-CR-accel}}

724: 

725: In this section, we try to explain both the observed azimuthal

726: variations of the ratio of radii between the BW and CD (\S

727: \ref{sub-sect-radii-SE}) and synchrotron flux at the BW (\S

728: \ref{subsubsect-azimu-synch-emiss}) in SN 1006 in the context of

729: CR-modified shock hydrodynamics where the efficiency of CR

730: acceleration is varying along the BW.

731: 

732: In the following, we use CR-modified hydrodynamic models of SNR

733: evolution which allow us for a given set of acceleration parameters to

734: calculate the ratio of radii, $R_{\mathrm{BW}}/R_{\mathrm{CD}}$, and

735: the synchrotron flux, $F_{\nu}$, at any frequency $\nu$. Relevant

736: parameters are the injection rate of particles into the DSA process,

737: the magnetic field strength and the particle diffusion coefficient (or

738: equivalently the turbulence level). Our goal will not consist in

739: finding the exact variations of these parameters along the shock by

740: fitting the data, but rather to provide a qualitative description of

741: the variations based on the theory to verify whether the model

742: predictions are consistent with the observations or not.

743: 

744: \subsubsection{Conceptual basis for modeling\label{subsect-basis-modelling}}

745: 

746: Here, we explicate the various assumptions concerning the spatial

747: distribution of the ambient density and magnetic field before running

748: our CR-modified hydrodynamic models. The knowledge of the remnant's

749: environment is important since it affects its evolution and also its

750: emission characteristics.

751: 

752: First, we assume that the ambient density is uniform. This is

753: suggested by the quasi-circularity of the H$\alpha$ filament in the SE

754: of SN 1006 and a reasonable assumption for remnants of thermonuclear

755: explosion \citep{bah07}. Second, we assume that the ambient magnetic

756: field direction is oriented along a preferred axis. In SN 1006, this

757: would correspond to southwest-northeast axis which is the direction

758: parallel to the Galactic plane \citep[following][]{rob04}. The

759: magnetic field lines are then parallel to the shock velocity in the

760: brightest part of the synchrotron caps in SN 1006 and perpendicular at

761: the equator. The angle between the ambient magnetic field lines and

762: the shock velocity will determine the number of particles injected in

763: the DSA process, i.e., the injection rate. The smaller the angle, the

764: larger the injection rate \citep{elb95, vob03}.

765: 

766: With this picture of an axisymmetry around the magnetic field

767: orientation axis in mind, the azimuthal conditions at the BW are

768: essentially spatially static, the evolution is self-similar and hence

769: temporal evolution of gas parcels can be followed with a 1-D

770: spherically symmetric code. We will run such a code with specific

771: initial conditions at each azimuthal angle along the BW (viewing the

772: SNR in a plane containing the revolution axis of the magnetic field).

773: We implicitly assume a radial flow approximation in our approach,

774: i.e., that each azimuthal zone is sufficiently isolated from the

775: others so that they evolve independently.

776: 

777: We run the self-similar models assuming a power-law density profile in

778: the ejecta with an index of $n=7$, an ejected mass and kinetic energy

779: of the ejecta of 1.4 $M_{\odot}$ and $10^{51}$ ergs, respectively,

780: which are standard values for thermonuclear SNe. We assume that the

781: SNR evolves into an interstellar medium whose  density is $n_0 = 0.03$

782: cm$^{-3}$ and pressure 2300 K cm$^{-3}$. For such a low density, the

783: use of the self-similarity is well justified as shown in Figure

784: \ref{fig_n0_vs_ratio}: the ratio of radii

785: $R_{\mathrm{BW}}/R_{\mathrm{CD}}$ obtained with a 1-D purely numerical

786: hydrodynamic simulation with the same initial conditions remains

787: indeed more or less constant even until the age of 1000 yrs

788: (\textit{solid yellow lines}). Other input parameters will be

789: specified later. First, we describe the CR

790: acceleration model used with the hydrodynamical model.

791: 

792: \subsubsection{CR acceleration model\label{CR-hydro-model}}

793: 

794: The models consist of a self-similar hydrodynamical calculation

795: coupled with a nonlinear diffusive shock acceleration model, so that

796: the back-reaction of the accelerated particles at the BW is taken into

797: account \citep{dee00}. For a given injection rate of protons,

798: $\eta_{\mathrm{inj}}$, far upstream magnetic field,  $B_{\mathrm{u}}$,

799: and diffusion coefficient, $D$, the shock acceleration model

800: determines the shock jump conditions and the particle spectrum from

801: thermal to relativistic energies \citep{bee99}. This allows us to

802: compute the ratio of radii and, by following the downstream evolution

803: of the particle spectrum associated with each fluid element, the

804: synchrotron emission within the remnant \citep{cad05}.

805: 

806: {The CR proton spectrum at the BW is a piecewise power-law

807: model with an exponential cutoff at high energies:

808: \begin{equation} \label{fp}

809: f_{\mathrm{p}}(E) = a \:  E^{-\Gamma(E)} \:

810:  \exp \left( -

811:  E / E_{\mathrm{p,max}} \right),

812: \end{equation}

813: where $a$ is the normalization, $\Gamma$ is the power-law index which

814: depends on the energy $E$, and $E_{\mathrm{p,max}}$ is the maximum

815: energy reached by the protons. Typically three distinct energy regimes

816: with different $\Gamma$ values are assumed \citep{bee99}.  The

817: normalization, $a$, is given by:

818: \begin{equation}\label{a}

819: a =  \: \frac{n_{\mathrm{inj}} \: q_{\mathrm{sub}}}{4 \: \pi \: p_{\mathrm{inj}}^3}

820: \end{equation}

821: where $n_{\mathrm{inj}} = \eta_{\mathrm{inj}} \: n_0 \

822: (r_{\mathrm{tot}}/r_{\mathrm{sub}} )$ is the number density of gas

823: particles injected in the acceleration process, with

824: $r_{\mathrm{tot}}$ and $r_{\mathrm{sub}}$ being the overall density

825: and subshock compression ratios, $q_{\mathrm{sub}} = 3 \;

826: r_{\mathrm{sub}} / (r_{\mathrm{sub}} - 1)$ and $p_{\mathrm{inj}}$ is

827: the injection momentum. Here, $p_{\mathrm{inj}} = \lambda \:

828: m_{\mathrm{p}} \: c_{s2}$ where $\lambda$ is a parameter (here fixed

829: to a value of 4) which encodes all the complex microphysics of the

830: shock and $c_{s2}$ is the sound speed in the shock-heated gas

831: immediately downstream from the subshock \citep[see \S 2.2

832: in][]{bee99}. We note that because there is a nonlinear reaction on

833: the system due to the injection, the parameter $\eta_{\mathrm{inj}}$

834: should be in fact related to $p_{\mathrm{inj}}$ \citep[see \S 5

835: in][for more details]{blg05}.

836: 

837: The CR electron spectrum at the BW is determined by assuming a certain

838: electron-to-proton density ratio at relativistic energies,

839: $K_{\mathrm{ep}}$, which is defined as the ratio between the electron

840: and proton distributions at a regime in energy where the protons are

841: already relativistic but the electrons have not yet cooled radiatively

842: \citep[e.g.,][]{elb00}. In the appropriate energy range, the CR

843: electron spectrum is then:

844: \begin{equation} \label{fe}

845: f_{\mathrm{e}}(E) = a \: K_{\mathrm{ep}} \: E^{-\Gamma(E)} \:

846:  \exp \left( - 

847:  E / E_{\mathrm{e,max}} \right),

848: \end{equation}

849: where $E_{\mathrm{e,max}}$ is the maximum energy reached by the

850: electrons, which could be eventually lower than that of protons

851: ($E_{p,\mathrm{max}}$) through synchrotron cooling of electrons

852: depending on the strength of the post-shock magnetic field.

853: $K_{\mathrm{ep}}$ is left as a free parameter. We will obtain

854: typically $K_{\mathrm{ep}} \sim 10^{-4}-10^{-2}$.

855: 

856: The maximum energies $E_{p,\mathrm{max}}$ and $E_{e,\mathrm{max}}$

857: contain information on the limits of the acceleration. They are set by

858: matching either the acceleration time to the shock age or to the

859: characteristic time for synchrotron losses, or by matching the

860: upstream diffusive length to some fraction, $\xi_s$, of the shock

861: radius, whichever gives the lowest value. We set $\xi_s=0.05$,

862: a value that allows to mimic the effect of an expanding spherical

863: shock  as compared to the plane-parallel shock approximation assumed

864: here \citep[see][]{be96, elb00}. A fundamental parameter in

865: determining the maximum energy of particles is the diffusion

866: coefficient, $D$, which contains information on the level of

867: turbulence and encodes the scattering law \citep{pam06}. For the sake

868: of simplicity, we assumed the Bohm regime for all particles (i.e.,

869: diffusion coefficient proportional to energy). We allow deviations

870: from the Bohm limit (which corresponds to a mean free path of the

871: charged particles equal to the Larmor radius, which is thought to be

872: the lowest possible value for isotropic turbulence) via the parameter

873: $k_0$ defined as the ratio between the diffusion coefficient, $D$, and

874: its Bohm value, $D_{\mathcal{B}}$. Hence $k_0 \equiv D /

875: D_{\mathcal{B}}$ is always $\geq 1$ and $k_0=1$ corresponds to the

876: highest level of turbulence. The smaller the diffusion coefficient (or

877: $k_0$), the higher the maximum energy (the maximum energy scales as

878: $1/\sqrt{k_0}$ in the radiative loss case and as $1/k_0$ in the age-

879: and escape-limited cases).

880: 

881: \subsubsection{Heuristic models}

882: 

883: In order to make predictions for the azimuthal variations of the ratio

884: of radii and synchrotron flux, we must make some assumptions about the

885: azimuthal variations of the input parameters used in our CR-modified

886: hydrodynamical models. Relevant input parameters are the injection

887: rate ($\eta_{\mathrm{inj}}$), the upstream magnetic field

888: ($B_{\mathrm{u}}$), the electron-to-proton ratio at relativistic

889: energies ($K_{\mathrm{ep}}$) and the level of turbulence or diffusion

890: coefficient relative to the Bohm limit ($k_0$).

891: 

892: To understand the influence of these four basic input parameters, we

893: begin first with a heuristic discussion to set the stage for later

894: detailed and more physical models (\S \ref{subsubsect-good-model}).

895: In Figure \ref{fig_heuristic_model_profiles}, we show three models

896: where we vary the two most important parameters, $\eta_{\mathrm{inj}}$

897: and $B_{\mathrm{u}}$, with azimuthal angle ($K_{\mathrm{ep}}$ is held

898: constant and $k_0$ is set to 1 at all angles). Model 1 (\textit{left

899: panels}) has $\eta_{\mathrm{inj}}$ varying from $10^{-5}$ to

900: $10^{-3}$, while $B_{\mathrm{u}}$ is held constant at the value $25 \;

901: \mu\mathrm{G}$. Model 2 (\textit{middle panels}) has constant

902: $\eta_{\mathrm{inj}} = 10^{-4}$ and $B_{\mathrm{u}}$ varying from $3

903: \; \mu\mathrm{G}$ to $25 \;\mu\mathrm{G}$. Model 3  (\textit{right

904: panels}) is a combination of models 1 and 2 and has

905: $\eta_{\mathrm{inj}}$ varying from $10^{-5}$ to $10^{-3}$ and

906: $B_{\mathrm{u}}$ varying from $3 \; \mu\mathrm{G}$ to $25

907: \;\mu\mathrm{G}$.

908: 

909: Note that in all these models, the functional form of the profiles

910: when $\eta_{\mathrm{inj}}$ and $B_{\mathrm{u}}$ vary is rather

911: arbitrary and just used for illustration purposes (for completeness we

912: note that the angular range shown corresponds to the data extraction

913: region in Fig.~\ref{fig_flux_vs_angle_radio_xrays}). Moreover, the

914: upstream magnetic field, $B_{\mathrm{u}}$ (\textit{second panels,

915: solid lines}), is allowed to vary with azimuthal angle because we

916: implicitly assume that it can be significantly amplified by the

917: CR-streaming instability. Our model does not include self-consistently

918: the magnetic field amplification believed to occur at SNR shocks, but

919: is provided with a simple compression.  Assuming that the magnetic

920: turbulence is isotropic ahead of the shock, the immediate post-shock

921: magnetic field, $B_{\mathrm{d}}$ (\textit{dashed lines}), will then be

922: larger than upstream by a factor $r_B \equiv B_{\mathrm{d}} /

923: B_{\mathrm{u}} = \sqrt{(1 + 2 \: r_{\mathrm{tot}}^{2})/3}$. This

924: relation is assumed at each azimuthal angle\footnote{Here, the

925: compression does not depend on the angle between the upstream magnetic

926: field and shock velocity as considered in \cite{re98} or \cite{orb07}.

927: Instead we assume that after the amplification process, the magnetic

928: field becomes entirely turbulent, so its originally ordered character

929: has been lost.}.

930: 

931: We show the predicted profiles of the ratio of radii between the BW

932: and CD (\textit{third panels}) and synchrotron brightness projected

933: along the line-of-sight in the radio and different X-ray energy bands

934: (\textit{bottom panels}). We explain later how those brightness

935: profiles were precisely constructed (\S \ref {subsubsect-good-model}).

936: Clearly, model 2 does not predict the expected correlation between the

937: ratio of radii and the synchrotron brightness and hence can be

938: rejected immediately\footnote{Looking at the BW to CD ratio in

939: model 2, it naively seems surprising to have a less modified shock

940: (i.e., higher BW to CD ratio) where the magnetic field is larger.

941: This effect is due to the presence of Alfv\'en heating in the

942: precursor \cite[see][]{bee99}, although there is debate about whether

943: this is the most efficient mechanism for turbulent heating in the

944: precursor \citep[see][]{amb06}.}. On the other hand, models 1 and 3

945: lead to the expected correlation: the smaller the BW/CD ratio, the

946: brighter the synchrotron emission. Comparing models 1 or 3 with model

947: 2 allows us to see that only an azimuthal variation of the injection

948: rate leads to the appropriate variation in the ratio of BW to CD radii

949: (\textit{third panels}). Comparing model 1 with model 3 shows the

950: impact of the magnetic field variations through synchrotron losses on

951: the azimuthal profiles of the X-ray synchrotron emission

952: (\textit{bottom panels}).

953: 

954: Although models 1 and 3 seem to describe the key observational

955: constraints fairly well, they suffer weaknesses as regards their

956: underlying astrophysical premises.  In model 1, it is difficult to

957: justify why the magnetic field should be enhanced in the region of

958: inefficient particle acceleration ($\eta_{\mathrm{inj}} \leq 6 \times

959: 10^{-5}$).  The model proposes a post-shock magnetic field there of

960: $\sim 80 \: \mu\mathrm{G}$, when a value of $10-15 \: \mu\mathrm{G}$ would

961: more likely reflect the value of the compressed ambient field.  In

962: model 3, it is difficult to understand why the turbulence level would

963: saturate (i.e., $k_0 = 1$) everywhere. Hence, we are led to develop

964: more elaborated models where $k_0$ and/or $K_{\mathrm{ep}}$ are

965: allowed to vary with azimuthal angle. A variation of either $k_0$ or

966: $K_{\mathrm{ep}}$ will not significantly affect the ratio of BW to CD

967: radii, but will only modify the profiles of the synchrotron emission.

968: In fact, a model with varying $K_{\mathrm{ep}}$ can be already

969: excluded because this quantity changes the radio and X-ray synchrotron

970: intensities in the same way while the observed radio and X-ray

971: profiles (Fig.~\ref{fig_flux_vs_angle_radio_xrays}) vary with azimuth

972: in different ways. Varying $k_0$ (as we show in \S \ref

973: {subsubsect-good-model}) does allow us to modify the relative

974: azimuthal profiles.

975: 

976: \subsubsection{Can CR-modified models provide $R_{\mathrm{BW}}/R_{\mathrm{CD}}=1.00$?\label{ratio-unity}}

977: 

978: Although CR-modified hydrodynamical simulations predict a ratio of

979: radii $R_{\mathrm{BW}}/R_{\mathrm{CD}}$ smaller than standard

980: hydrodynamical simulations, they still predict a lower limit on that

981: ratio that is strictly larger than unity. For instance, Figure

982: \ref{fig_heuristic_model_profiles} (\textit{third-right panel}) shows

983: that the modeled ratio of radii where the synchrotron emission is

984: strong is $\sim 1.13$. This value was obtained assuming an injection

985: rate of $10^{-3}$ and an ambient magnetic field of 25 $\mu$G. Keeping

986: all the same inputs, but reducing the magnetic field value to 3

987: $\mu$G, we obtain the lowest possible value of

988: $R_{\mathrm{BW}}/R_{\mathrm{CD}}=1.06$.

989: 

990: On the other hand, the observations of SN 1006 show that the ratio of

991: radii is of order 1.00 in the synchrotron-emitting caps, a value that

992: is not possible to obtain from our CR-modified model under the DSA

993: framework.  However, as already mentioned before in our discussions of

994: the small gap between the BW and CD in the region of inefficient

995: acceleration (\S \ref{subsect-no-CR-accel}), we need to consider

996: projection effects due to hydrodynamical instabilities at the CD which

997: can reduce the gap by $6-10\%$.  This does not fully account for the

998: difference with the model and, therefore, we may need to invoke

999: another effect, such as clumping of the ejecta, to fully explain the

1000: offset between the observation and models.  It is important to note

1001: that in both the regions of efficient and inefficient CR acceleration,

1002: we require essentially the same numerical factor to bring the modeled

1003: ratio of radii (for the appropriate model in each case) into agreement

1004: with the observed ratio. In the following, we introduce an ad-hoc

1005: rescaling of the ratio of radii by decreasing the modeled values by a

1006: constant factor ($\sim 13\%$) at all azimuthal angles. By doing this,

1007: we implicitly assume that this factor contains the total contribution

1008: of effects that are unrelated to the CR acceleration process. The

1009: numerical value used depends, of course, on the particular SN

1010: explosion model used. For example, the factor would need to be

1011: increased slightly in the case of an exponential ejecta profile (cf.,

1012: Fig.~\ref{fig_n0_vs_ratio}).

1013: 

1014: \subsubsection{Toward a good astrophysical model\label{subsubsect-good-model}}

1015: 

1016: In Figure \ref{fig_model_k0_var_profiles}, we present in more detail

1017: what  we believe is a reasonable astrophysical model. This model

1018: assumes that $\eta_{\mathrm{inj}}$ increases from a value of $5 \times

1019: 10^{-5}$ to a value of $10^{-3}$ while at the same time

1020: $B_{\mathrm{u}}$ increases from a value of $3 \;\mu\mathrm{G}$ to a

1021: value of $20 \; \mu\mathrm{G}$ and $k_0$ increases from an arbitrary

1022: value of 100 to 1 ($K_{\mathrm{ep}}$ is kept constant at all angles).

1023: The azimuthal variations of these parameters is consistent with the

1024: picture of DSA where, once the magnetic field fluctuations at the

1025: shock exceed the background ISM fluctuations, any low injection rate

1026: leads to growth of the turbulent magnetic field in the upstream

1027: region, which in turns leads to an increase of the injection rate

1028: \citep{vob03}. The acceleration efficiency, defined as the

1029: fraction of total energy flux crossing the shock that goes into

1030: relativistic particles \citep[see \S 3.3 in][]{bee99}, is $51\%$ in

1031: the case with $\eta_{\mathrm{inj}} = 10^{-3}$ and $7\%$ in the case

1032: with $\eta_{\mathrm{inj}} = 5 \times 10^{-5}$.

1033: 

1034: Note that the form of the azimuthal profiles of $\eta_{\mathrm{inj}}$

1035: and $B_{\mathrm{u}}$ is again arbitrary (here a squared-sine

1036: variation) and just used for illustration purpose.  Our goal is not to

1037: find the exact form of the profiles based on an accurate fit of the

1038: model predictions to the observations, but rather to describe

1039: qualitatively how the acceleration parameters vary along the BW in SN

1040: 1006. The extremum values for $\eta_{\mathrm{inj}}$ were chosen so

1041: that the predicted variations of the ratio of radii

1042: $R_{\mathrm{BW}}/R_{\mathrm{CD}}$ roughly matches the observations. A

1043: maximum (minimum) value of $\eta_{\mathrm{inj}}$ between $10^{-4}$ and

1044: $10^{-2}$  (resp. $10^{-5}$ and $5 \times 10^{-5}$) will not

1045: significantly affect our results concerning the profile of

1046: $R_{\mathrm{BW}}/R_{\mathrm{CD}}$, but differences will be seen in the

1047: absolute intensity in the synchrotron emission. The maximum value of

1048: $B_{\mathrm{u}}$ was adjusted so that it yields an immediate postshock

1049: value, $B_{\mathrm{d}} \simeq 80 \: \mu{\mathrm{G}}$ (\textit{panels

1050: e}), consistent with the value derived from the thickness of the X-ray

1051: synchrotron-emitting rims in SN 1006, under the assumption that this

1052: thickness is limited by the synchrotron losses of the highest energy

1053: electrons \citep[e.g.,][]{ba06}. Finally, we adjusted the minimum

1054: value of $B_{\mathrm{u}}$ in the faint region so that the predicted

1055: and observed radio synchrotron flux are roughly the same; here we show

1056: profiles with a value of $B_{\mathrm{u}} = 3 \: \mu{\mathrm{G}}$,

1057: i.e.,  the typical ISM value.

1058: 

1059: In Figure~\ref{fig_model_k0_var_profiles} (\textit{panel g}), we show

1060: the predicted azimuthal profile of ratio of radii between the BW and

1061: CD (\textit{dashed lines}). The ratio of radii was normalized

1062: (\textit{solid lines}) so that it roughly equals 1.00 in the region of

1063: efficient CR acceleration as in the observations. This normalization

1064: procedure (which we justified in \S \ref{ratio-unity}) works

1065: remarkably well. The variations of the ratio of radii reflects nothing

1066: but the variations of the overall density compression ratio,

1067: $r_{\mathrm{tot}}$ (\textit{panel d}). For low injection rates

1068: ($\eta_{\mathrm{inj}} \leq 6 \times 10^{-5}$), the compression ratio

1069: comes close to the value obtained in the test-particle case (i.e.,

1070: $r_{\mathrm{tot}} = 4$) and as the injection rate increases, the

1071: plasma becomes more compressible with $r_{\mathrm{tot}} \sim 6$.

1072: However, once the magnetic field has been considerably amplified

1073: ($B_{\mathrm{u}} \geq 10 \; \mu\mathrm{G}$), $r_{\mathrm{tot}}$ starts

1074: to slightly decrease due to the heating of the gas by the Alfv\'en

1075: waves in the precursor region \citep[see][]{bee99}.

1076: 

1077: In Figure~\ref{fig_model_k0_var_profiles} (\textit{panel h}), we show

1078: the azimuthal profiles of the synchrotron emission in the radio and

1079: several X-ray energy bands. To obtain these profiles, we first

1080: computed the radial profile of the synchrotron flux (calculated in a

1081: given energy band) at each azimuthal angle. Then, we projected the

1082: flux along the line-of-sight and finally extracted the projected flux

1083: in $30\arcsec$-wide regions behind the BW (assuming a distance of 2

1084: kpc) as we did in the observations of SN 1006 (\S

1085: \ref{subsubsect-azimu-synch-emiss}). In this procedure, we assumed a

1086: spherically symmetric distribution for the emissivity.  This is a

1087: reasonable assumption as long as we restrict the analysis close to the

1088: BW, considering that the bright limbs in SN 1006 are in fact polar

1089: caps, i.e., where the emissivity distribution is axisymmetric.

1090: Finally, we fixed the radio flux in the synchrotron caps to be the

1091: same as in the observations by adjusting the electron-to-proton

1092: density ratio at relativistic energies, $K_{\mathrm{ep}}$.  This is

1093: obtained for a value of $K_{\mathrm{ep}} = 7 \times 10^{-4}$.

1094: 

1095: Relaxing the assumption of the Bohm diffusion (i.e., $k_0$) at each

1096: azimuthal angle significantly modifies the azimuthal profiles of the

1097: synchrotron flux in the X-ray band  (while the radio profile is

1098: unchanged) compared to the model~3 of Figure

1099: \ref{fig_heuristic_model_profiles} in which $k_0$ was always equal to

1100: 1.  This is because an increase of $k_0$ decreases the maximum energy

1101: of electrons, $E_{e,\mathrm{max}}$. Since the synchrotron emission is

1102: very sensitive to the position of the high-energy cutoff in the

1103: electron distribution (i.e., $E_{e,\mathrm{max}}$), all profiles do

1104: not show anymore a plateau in the region of efficient CR acceleration

1105: but are now gradually decreasing (\textit{panel h}). Besides, the

1106: synchrotron brightness in the X-ray bands now decreases faster than in

1107: the radio (\textit{panel i}). This is roughly consistent with what is

1108: observed in SN 1006. We note that in this model, the maximum energy of

1109: both protons and electrons (\textit{panel f}) decrease rather quickly

1110: as we move toward regions of inefficient CR acceleration as opposed to

1111: the previous model~3 with $k_0=1$ where the maximum energy of

1112: electrons remained approximately constant (not shown). Overall, our

1113: results from this model are consistent with those of \cite{rob04}

1114: which were based on a different approach (i.e., measure of the

1115: azimuthal variations of the cutoff-frequency all around SN 1006) using

1116: \textit{XMM-Newton} observations.

1117: 

1118: 

1119: \subsubsection{Other azimuthal profiles for $\eta_{\mathrm{inj}}$, $B_{\mathrm{u}}$ and $k_0$}

1120: 

1121: We have found that a simple model in which the injection rate of

1122: particles in the acceleration process ($\eta_{\mathrm{inj}}$),

1123: amplified magnetic field ($B_{\mathrm{u}}$) and level of turbulence

1124: ($k_0$) were gradually increasing around the shock front was able to

1125: provide predictions in a good agreement with the observations for the

1126: azimuthal variations of the ratio of radii and radio and X-ray

1127: synchrotron fluxes. The input parameters and their azimuthal

1128: variations were chosen based on the physical picture where the

1129: injection of particles and amplification of the magnetic turbulence

1130: are coupled.

1131: 

1132: One can imagine however more complicated scenario and hence more

1133: complicated input profiles for the acceleration parameters

1134: ($\eta_{\mathrm{inj}}$, $B_{\mathrm{u}}$, $k_0$). For instance, the

1135: coupling between the injection and the turbulence which occurs at the

1136: beginning of the acceleration process may end at some point when the

1137: amplitude of the magnetic field fluctuations become so high that their

1138: further growth is prevented by strong dissipation processes

1139: \citep{vob03}. The injection, magnetic field and level of turbulence

1140: may in fact saturate (over some azimuthal range) as we reach regions of

1141: very efficient CR acceleration.  Another example of scenario can be

1142: imagined if the shock velocity is not uniformly distributed along the

1143: shock (as opposed to what we have assumed in our models) due to a

1144: asymmetrical explosion. Such a scenario could explain why the bright

1145: synchrotron-emitting caps in SN 1006 seem to have larger radii than

1146: the faint regions. If so, the injection could be larger in the region

1147: of higher velocity while the level of turbulence and magnetic field

1148: could potentially saturate there. In general, these new scenarios will

1149: not lead to a significant modification of the profile of the ratio of

1150: radii. However, we do expect some changes in the azimuthal profiles of

1151: the radio and X-ray synchrotron flux. They will generally be flatter

1152: in the regions of very efficient CR acceleration.

1153: 

1154: %*******************************************

1155: \section{Discussion\label{sect-discussion}}

1156: %*******************************************

1157: 

1158: \subsection{Polar cap morphology\label{sub-morphology}}

1159: 

1160: The model in which the injection rate of particles, magnetic field and

1161: level of turbulence are all varying as a function of azimuth was

1162: successful in reproducing the overall azimuthal variations of the

1163: radio and X-ray synchrotron flux observed along the BW of SN 1006

1164: (Fig.~\ref{fig_model_k0_var_profiles}). In fact, it is actually

1165: possible to build a 2-D projected map of the radio and X-ray

1166: synchrotron morphology comparable to the radio and X-ray images of SN

1167: 1006 (Figs. \ref{fig_SN1006_3color_xray} and \ref{fig_SN1006_radio}).

1168: This requires us to know the three-dimensional distribution of the

1169: emissivity. Here, we consider polar caps as suggested by \cite{rob04}.

1170: 

1171: To simplify the calculation, we consider that the radial profiles of

1172: the emissivity obtained from the different values of injection,

1173: magnetic field and turbulence level have an exponential form

1174: characterized by a maximum emissivity at the shock and width $\Delta

1175: R$ over which the emissivity decreases (see Appendix

1176: \ref{app-projection}). We also fixed the BW radius, 

1177: $R_{\mathrm{BW}}$, to be the same at each azimuthal angle where the

1178: injection and magnetic field vary, although it slightly depends on the

1179: CR acceleration efficiency. These are good approximations.

1180: 

1181: In the case of the radio synchrotron emission,  the width $\Delta R$

1182: does not depend on the azimuthal angle (from the pole to the equator),

1183: and we have typically $\Delta \tilde{R} \equiv \Delta R /

1184: R_{\mathrm{BW}} = 0.01$. The projected radio morphology is shown in

1185: Figure \ref{fig_projection} (\textit{top-right panel}). It has a

1186: bipolar limb-brightened morphology as is observed in SN 1006. 

1187: 

1188: In the case of the X-ray synchrotron emission, the width $\Delta R$ is

1189: expected first to increase (starting from the equator) as particles

1190: reach higher energies and the emission builds up, and then to slightly

1191: decrease when the effect of radiative losses starts to become

1192: important due to the larger post-shock magnetic field as we move

1193: toward the pole. Because the brightness is much lower at the equator

1194: than at the pole, a reasonable approximation is to consider the width

1195: constant with typically $\Delta \tilde{R} = 2 - 5 \times 10^{-3}$

1196: (\textit{bottom-left panel}) or even a slightly increasing width from

1197: the pole to the equator (\textit{bottom-right panel}). A slightly

1198: increasing width will not change the overall morphology. It still

1199: produces a thin rim whose projected width is decreasing starting from

1200: the pole to the equator. In Figure \ref{fig_projection}, we have

1201: assumed that the emissivity contrast at the BW between the pole and

1202: the equator is always a factor 100. In fact, because the contrast

1203: between the pole and equator is larger in the X-rays than in the

1204: radio, we will obtain a similar morphology but with the size of the

1205: caps slightly reduced in X-ray. In other words, the X-ray synchrotron

1206: emitting rims become geometrically thinner and of smallest azimuthal

1207: extent as we go to higher energy. All these effects would be however

1208: slightly attenuated if we had included the instrumental effects

1209: (particularly the PSF), but overall this is consistent with what is

1210: observed in SN 1006.

1211: 

1212: \subsection{CR acceleration in a partially ionized medium\label{sect-neutral-med}}

1213: 

1214: When trying to delineate the BW, we found that the X-ray

1215: synchrotron-emitting rims and the diffuse filament of H$\alpha$

1216: emission were coincident over some little azimuthal range, notably for

1217: instance at the edge of the eastern cap where the synchrotron emission

1218: starts to turn on (see Fig. \ref{fig_SN1006_3color_xray}). SN 1006 is

1219: one of the rare remnants where this characteristic is clearly

1220: observed, although this likely happens in the Tycho SNR too.

1221: 

1222: In general, H$\alpha$ emission from non-radiative shocks is believed

1223: to arise when the blast wave encounters partially ionized gas. This

1224: points to the presence of neutral atoms in the ambient medium around

1225: SN 1006 \citep{ghw02_SN1006}. Theoretical studies have shown that in

1226: such medium, the scattering Alfv\'en waves should be damped,

1227: henceforth quenching the acceleration of high energy particles

1228: \citep{drd96}. In fact, recent theoretical studies suggest that the

1229: magnetic field could be turbulently amplified even in such medium

1230: \citep{byt05}. The fact that we observe the H$\alpha$ and synchrotron

1231: X-rays to be coincident gives support to this statement. Other

1232: observations of H$\alpha$ emission coexisting with synchrotron X-rays

1233: would be very interesting in that regard.

1234: 

1235: \subsection{Comparison between SN 1006 and Tycho}

1236: 

1237: Like SN 1006, the Tycho SNR is believed to be the remnant of a

1238: thermonuclear explosion and, as such, is expected to develop an

1239: approximately exponential ejecta density profile and to evolve in a

1240: uniform ambient medium. Measurements of the gap between BW and CD have

1241: been done in the Tycho SNR using \textit{Chandra} observations

1242: \citep{wah05}. We comment on the Tycho results in the light of our

1243: results.

1244: 

1245: Because the thin and bright X-ray synchrotron-emitting rims

1246: concentrated at the BW are observed all around Tycho, in contrast to

1247: SN 1006, the BW's location can be easily determined. In addition,

1248: because the X-ray emission from the shocked ejecta is dominant in

1249: Tycho, in contrast to SN 1006, it is also fairly easy to determine the

1250: CD's location. The observed azimuthally-averaged

1251: ratio of radii between the BW and CD was found to be $\sim 1.04$ in

1252: Tycho \citep{wah05}. After correcting, approximately, for effects due

1253: to projection of the highly structured CD, the ratio of radii became

1254: $\sim 1.075$.  This is still a lot smaller than the (unprojected)

1255: ratio of $\sim 1.25$ derived using standard one-dimensional

1256: hydrodynamical simulations with no CR acceleration, at the current age

1257: of Tycho ($430$ yrs).  It was then concluded that efficient CR ion

1258: acceleration was occurring around nearly the entire BW of Tycho.

1259: 

1260: More recently, \citet{cah07} applied a CR-modified self-similar

1261: hydrodynamic model to the radio and X-ray observations of the Tycho

1262: SNR using the observed ratio of BW to CD radii as an important input

1263: constraint.  Using reasonable values for the hydrodynamical parameters

1264: (ambient medium density, SN explosion energy, ejected mass), DSA

1265: parameters (injection efficiency, magnetic field, diffusion

1266: coefficient, electron-to-proton ratio at relativistic energies), and

1267: the distance, a good description of the observational properties at

1268: the blast wave of Tycho was obtained. This detailed study provides a

1269: self-consistent/coherent picture of efficient CR ion acceleration.

1270: 

1271: Regarding SN 1006, we have some good news and some bad (compared to

1272: the case of Tycho, where CR-modified models seem to describe the data

1273: well).  The good news is that there is a significant variation in the

1274: ratio of BW to CD radii as a function of azimuthal angle that is

1275: correlated with the varying intensity of the X-ray and radio

1276: synchrotron emission.  Where the gap is large the synchrotron emission

1277: is faint (or not detected) and where the gap is small the emission is

1278: bright.  This is consistent with an azimuthal variation in the effects

1279: of CR-modification to the remnant dynamics.  The bad news is that in

1280: regions of presumably efficient acceleration (i.e., the bright rims),

1281: the ratio of BW to CD radii approaches small values, very close to 1.

1282: The puzzle is that our CR-modified hydrodynamical models are unable to

1283: produce ratio values less than 1.06, while the specific scenarios

1284: presented above produce ratio values closer to $1.12$. In order

1285: to make progress in our study we decided to introduce an ad-hoc

1286: rescaling of the ratio of radii by decreasing the modeled values at

1287: all angles by a constant factor of $\sim 12\%$.  This procedure

1288: works remarkably well and suggests to us that something else,

1289: unrelated to the CR acceleration process (such as 3-D projection

1290: effects and ejecta clumping), may be responsible for the smaller than

1291: expected gap between the BW and CD in SN 1006.

1292: 

1293: %*******************************************

1294: \section{Conclusion\label{sect-conclusion}}

1295: %*******************************************

1296: 

1297: Using a combination of radio, optical (H$\alpha$) and X-ray images, we

1298: have located the positions of the BW and CD radii along the

1299: southeastern sector of SN 1006. We found that with increasing

1300: azimuthal angles the ratio of radii between the BW and CD increases

1301: from values near unity (within the northeastern synchrotron-emitting

1302: cap) to a maximum of about 1.10 before falling again to values near

1303: unity (in the southwestern synchrotron-emitting cap). These variations

1304: reflect changes in the compressibility of the plasma attributed to

1305: variations in the efficiency of the BW at accelerating CR ions and

1306: give strong support to the overall picture that SNR shocks produce

1307: some fraction of Galactic CRs.  However, at the present time we do not

1308: have a detailed explanation for the apparent overall smallness of the

1309: measured ratios of radii in SN 1006: the minimum value predicted by

1310: our CR-modified self-similar dynamical models is 1.06.  In this study

1311: we simply rescaled the ratio of radii by a constant factor

1312: (independent of azimuthal angle) of $\sim 12\%$ with the expectation

1313: that some process, other than CR acceleration itself, was responsible

1314: for driving the edge of the ejecta closer to the BW all along the rim

1315: of SN 1006. The lack of a definitive astrophysical explanation for

1316: this discrepancy is a significant factor limiting our ability

1317: to understand the CR acceleration process in SN 1006. Further research

1318: into the effects of hydrodynamical instabilities at the CD and ejecta

1319: clumping (two possible explanations for the small BW/CD ratio) would

1320: incidentally provide new insights into the acceleration process.

1321: 

1322: In addition to the azimuthal variations of the ratio of radii between

1323: the BW and CD we also interpreted the variations of the synchrotron

1324: flux (at various frequencies) at the BW using CR-modified hydrodynamic

1325: models.  We assumed different azimuthal profiles for the injection

1326: rate of particles in the acceleration process, magnetic field and

1327: turbulence level.  We found the observations to be consistent with a

1328: model in which these quantities are all azimuthally varying, being the

1329: largest in the brightest regions. Overall this is consistent with the

1330: picture of diffusive shock acceleration. In terms of morphology, we

1331: found that our model was generally consistent with the observed

1332: properties of SN 1006, i.e., a bright and geometrically-thin

1333: synchrotron-emitting rim at the poles and very faint synchrotron

1334: emission at the equator and in the interior.  In the model, the X-ray

1335: synchrotron-emitting rims are geometrically thinner and of smallest

1336: azimuthal extent than the radio rims, which is in broad agreement with

1337: observations.  This is because the most energetic electrons

1338: accelerated at the blast wave lose energy efficiently in the amplified

1339: post-shock magnetic field.  Based on this picture, it would be worth

1340: trying to measure the azimuthal variation of the magnetic field

1341: strength (from the X-ray rim widths) and thereby gain further insight

1342: into the amplification process.

1343: 

1344: 

1345: \acknowledgments It's G.C.-C.'s pleasure to acknowledge

1346: \mbox{J.~Ballet}, \mbox{A.~Decourchelle} and \mbox{D.~C.\ Ellison} for

1347: previous enlightening discussions and work. G.C.-C.\ would like also

1348: to thank \mbox{F.~Winkler} for providing the precious H$\alpha$ image.

1349:  C.B.\ thanks John Blondin for making the VH-1 code available (this

1350: code was used to produce Figure \ref{fig_n0_vs_ratio}). E.M.R.\ is

1351: member of the Carrera del Investigador Cient\'ifico of CONICET,

1352: Argentina. E.M.R.\ was supported by grants PIP-CONICET 6433,

1353: ANPCyT-14018 and UBACYT A055 (Argentina). C.B.\ was supported by NASA

1354: through Chandra Postdoctoral Fellowship Award Number PF6-70046 issued

1355: by the Chandra X-ray Observatory Center, which is operated by the

1356: Smithsonian Astrophysical Observatory for and on behalf of NASA under

1357: contract NAS8-03060. Financial support was also provided by Chandra

1358: grants GO3-4066X and GO7-8071X to Rutgers, The State University

1359: of New Jersey.

1360: 

1361: 

1362: %===========================================================

1363: 

1364: \appendix

1365: 

1366: \section{Projection along the line-of-sight in the case of polar caps\label{app-projection}}

1367: 

1368: Let $r$ be the distance to the center $O$ of a sphere of radius $R_s$,

1369: $\phi$ the azimuthal angle in the $(O\vec{x},O\vec{z})$ plane with $0

1370: \leq \phi \leq 2 \pi$, and $\theta$ the latitude with $- \pi / 2 \leq

1371: \theta \leq \pi / 2$.  The projection of a

1372: spherical emissivity distribution $\mathcal{E}(r,\theta,\phi)$ onto

1373: the plane $(O\vec{x},O\vec{y})$ results in a two-dimensional

1374: brightness distribution:

1375: \begin{equation}\label{brightness}

1376: \mathcal{B}(x,y) = 2 \: \int_{0}^{\ell} \mathcal{E}(r,\theta,\phi) \: dz,

1377: \end{equation}

1378: where $\ell^2 = x_s^2 - x^2$, and $0 \leq x \leq x_s \equiv R_s \:

1379: \cos \theta$ and $0 \leq y \leq R_s$.  At any point of coordinate

1380: $(x,y,z)$, the emissivity is $\mathcal{E}(x,y,z) =

1381: \mathcal{E}(r,\theta,\phi)$ where $r^2 = x^2 + y^2 + z^2$ and $\theta

1382: = \arcsin \left( y / r \right)$ and $\phi = \arccos( x / \sqrt{x^2 +

1383: z^2} )$.  When the emissivity has a symmetry of revolution around the

1384: $O\vec{y}$ axis (e.g., polar caps), there is no dependency on $\phi$.

1385: 

1386: Let us consider a sphere where the emissivity $\mathcal{E}$ is

1387: radially decreasing from the maximum $\mathcal{E}_{\mathrm{max}}$ at

1388: the surface of the sphere with a characteristic width $\Delta R$. In

1389: the case of an exponential decrease and cylindrical symmetry, we have:

1390: \begin{equation}\label{emissivity}

1391:   \mathcal{E}(r,\theta,\phi) = \mathcal{E}_{\mathrm{max}}(\theta) \:

1392:   \exp \left( \frac{r - R_s}{\Delta R(\theta)} \right).

1393: \end{equation}

1394: In the case of polar caps, $\mathcal{E}_{\mathrm{max}}$ will

1395: be larger at the poles ($\theta=\pi / 2$ and $\theta = - \pi / 2$).

1396: 

1397: In Figure \ref{fig_projection}, we show the projected morphology

1398: obtained with the emissivity spatial distribution given by Eq.

1399: (\ref{emissivity}) in the range $0 \leq \theta \leq \pi / 2$ assuming

1400: different values and angular dependencies for the width $\Delta

1401: \tilde{R} \equiv \Delta R / R_s$. In those plots, a squared-sine

1402: variation was assumed for ${\log}_{10} \left(

1403: \mathcal{E_{\mathrm{max}}} \right)$, having the value $0.0$ at the

1404: pole and $-2.0$ at the equator. Decreasing the value at the equator

1405: produces similar rims but with a lower azimuthal extent (i.e.,

1406: smaller polar caps).

1407: 

1408: %===========================================================

1409: 

1410: \newpage

1411: 

1412: %++++++++++++++++++++++++++++++++++++++++++++++++++++

1413: 

1414: % Bibliographie a cet endroit ( style A&A )

1415: 

1416: \bibliographystyle{aa} % A&A style

1417: 

1418: \bibliography{ms} % ne surtout pas mettre le .bib !!!

1419: 

1420: %+++++++++++++++++++++++++++++++++++++++++++++++++++++

1421: 

1422: %############################################## 

1423: 

1424: %          FIGURES

1425: 

1426: %##############################################

1427: 

1428: \clearpage

1429: 

1430: \begin{figure*}[t] 

1431: \centering

1432: \includegraphics[width=14cm]{f1.eps}

1433: \caption{Three-color \textit{Chandra} image of SN 1006: 0.5-0.8 keV

1434: (\textit{red}), 0.8-2 keV (\textit{green}) and 2-4.5 keV

1435: (\textit{blue}). The regions dominated by synchrotron emission

1436: from high-energy relativistic electrons appears in white and are

1437: naturally associated with the BW. The regions dominated by the

1438: thermal emission from the shocked gas (mostly oxygen) appears in red

1439: and are most likely associated with the ejecta. Note that the oxygen

1440: is found slightly ahead of the BW in some places (e.g., east

1441: and south).  Point sources have been removed here so that only the

1442: diffuse emission from the remnant is visible. Images were background

1443: subtracted, corrected for vignetting and slightly smoothed.  The

1444: intensity scaling is square-root with a maximum fixed at $5.0 \times

1445: 10^{-4}$, $5.0 \times 10^{-4}$ and $2.0 \times 10^{-4}$

1446: ph/cm$^2$/s/arcmin$^2$ for the red, green and blue bands,

1447: respectively.} 

1448: \label{fig_SN1006_3color_xray} 

1449: \end{figure*}

1450: 

1451: \begin{figure*}[t] 

1452: \centering

1453: \includegraphics[width=14cm]{f2.eps}

1454: \caption{Radio image of SN 1006 at 1.5 GHz. The

1455: intensity scaling is square-root with a maximum fixed at $1.0$ mJy$/$beam.

1456: The bright elongated source in the western rim is a background radio galaxy.} 

1457: \label{fig_SN1006_radio} 

1458: \end{figure*}

1459: 

1460: \begin{figure*}[t] 

1461: \centering

1462: \includegraphics[width=14cm]{f3.eps}

1463: \caption{H$\alpha$ image of SN 1006.} 

1464: \label{fig_SN1006_Halpha} 

1465: \end{figure*}

1466: 

1467: \begin{figure*}[t] 

1468: \centering

1469: \includegraphics[width=8cm]{f4a.eps}

1470: \includegraphics[width=8cm]{f4b.eps}

1471: \includegraphics[width=8cm]{f4c.eps}

1472: \includegraphics[width=8cm]{f4d.eps}

1473: \caption{Images of SN 1006 in several energy bands. \textit{Top-left}:

1474: Radio image at 1.5 GHz. The maximum intensity was fixed at $0.6$

1475: mJy$/$beam. \textit{Top-right}: H$\alpha$ image with the contour

1476: errors associated with the shock front seen in the southeast

1477: (\textit{red lines}). \textit{Bottom-left}: \textit{Chandra} X-ray

1478: image in the oxygen band (0.5-0.8 keV). The maximum intensity was

1479: fixed at $5.0 \times 10^{-4}$ ph/cm$^2$/s/arcmin$^2$.

1480: \textit{Bottom-right}: \textit{Chandra} image in the mid-energy X-ray

1481: band (0.8-2.0 keV). The maximum intensity was fixed at $6.0 \times

1482: 10^{-4}$ ph/cm$^2$/s/arcmin$^2$. Both X-ray images were background

1483: subtracted, corrected for vignetting and slightly smoothed. In the

1484: radio and X-ray images, we show the contour derived from the H$\alpha$

1485: emission (\textit{red lines}). This contour lies within the contour

1486: errors shown in the H$\alpha$ image. In all the images, the intensity

1487: scaling is linear.}

1488: \label{fig_SN1006_all} 

1489: \end{figure*}

1490: 

1491: \begin{figure*}[t] 

1492: \centering

1493: \includegraphics[width=13cm]{f5.eps} 

1494: \caption{Azimuthal variation of the outer H$\alpha$ emission

1495: (\textit{black lines}) and X-ray emission as obtained in several X-ray

1496: energy bands: 0.5-0.8 keV (\textit{red lines}), 0.8-2 keV

1497: (\textit{green lines}) and 2-4.5 keV (\textit{blue lines}). Regions

1498: where the X-ray synchrotron emission is dominant are indicated by

1499: cross-hatched yellow lines.  In those regions, the fingers

1500: (\textit{stars}) indicate the presence of shocked ejecta found at or

1501: even slightly ahead of the BW. The X-ray contours correspond

1502: to places where the brightness becomes larger than $1.5 \times

1503: 10^{-5}$ ph/cm$^2$/s/arcmin$^2$ in the $0.5-0.8$ keV band, $0.6 \times

1504: 10^{-5}$ ph/cm$^2$/s/arcmin$^2$ in the $0.8-2$ keV band and $0.4

1505: \times 10^{-5}$ ph/cm$^2$/s/arcmin$^2$ in the $2-4.5$ keV band. Note

1506: that the radii measurements are quite sensitive to the contour values,

1507: but the range of variations is contained between the red and blue

1508: lines.}

1509: \label{fig_radius_Ha_xray} 

1510: \end{figure*}

1511: 

1512: \begin{figure*}[t] 

1513: \centering

1514: \includegraphics[width=13cm]{f6.eps} 

1515: \caption{Ratio of the outer H$\alpha$ radius (\textit{black lines} in

1516: Fig.  \ref{fig_radius_Ha_xray}) to the 0.5-0.8 keV X-ray radius

1517: (\textit{red lines} in Fig. \ref{fig_radius_Ha_xray}). Regions where

1518: the X-ray synchrotron emission is dominant are indicated by

1519: cross-hatched yellow lines. Error bars include uncertainties on the

1520: radius derived from the (expanded) H$\alpha$ image and that

1521: done in the low-energy X-rays choosing different

1522: contour values ($2.5 \pm 1.0 \times 10^{-5}$

1523: ph/cm$^2$/s/arcmin$^2$). The large error bars between

1524: $200^\circ-205^\circ$ are due to a low exposure there.}

1525: \label{fig_Rs_o_Rc} 

1526: \end{figure*}

1527: 

1528: \begin{figure*}[t] 

1529: \centering

1530: \includegraphics[width=13cm]{f7a.eps}

1531: \includegraphics[width=13cm]{f7b.eps}

1532: \caption{\textit{Top}: Azimuthal variations of the brightness at the

1533: BW of SN 1006 at several frequencies/energies: 1.5 GHz (\textit{black

1534: lines}), 0.5-0.8 keV (\textit{red lines}), 0.8-2 keV (\textit{green

1535: lines}) and 2-4.5 keV (\textit{blue lines}). The brightness in each

1536: azimuthal bin was calculated between the BW radius and the radius

1537: $30\arcsec$ behind.  For comparison, we show the level of the radio

1538: rms noise level and X-ray background located $1\arcmin$ outside the BW

1539: and extracted from a $30\arcsec$-wide box (\textit{dashed lines}).

1540: \textit{Bottom}: Same but with the curves rescaled roughly to the same

1541: level in the brightest regions and shown with a log scale.}

1542: \label{fig_flux_vs_angle_radio_xrays} 

1543: \end{figure*}

1544: 

1545: \begin{figure*}[t] 

1546: \centering 

1547: \includegraphics[width=13cm]{f8.eps} 

1548: \caption{Ratio of radii between the BW and CD,

1549: $R_{\mathrm{BW}}/R_{\mathrm{CD}}$, as obtained from 1-D numerical

1550: hydrodynamical simulations (assuming no CR acceleration at the BW) and

1551: calculated for a large range of ambient density, $n_0$, for a fixed

1552: SNR age (\textit{dashed lines}: 400 yrs; \textit{solid lines}: 1000

1553: yrs).  We show curves obtained with a 1-D exponential (EXP) ejecta

1554: profile with a kinetic energy of the explosion $E_{51} \equiv E /

1555: 10^{51}$ ergs $=1.0$ (\textit{red lines}), $E_{51} = 0.8$

1556: (\textit{blue lines}), $E_{51} = 1.4$ (\textit{green lines}) and a 1-D

1557: powerlaw (PL7) ejecta profile ($n=7$) with $E_{51} = 1.0$

1558: (\textit{yellow lines}). The cross-hatched domains correspond to the

1559: intersection of the range of ratio of radii observed in SN 1006 and

1560: Tycho ($\leq 1.10$, \textit{vertical solid lines}) and range of

1561: ambient medium density typically encountered in their vicinity

1562: (\textit{horizontal dotted lines}). The 1-D hydrodynamical simulations

1563: are clearly unable to predict a ratio of radii as small as the one

1564: observed in both SNRs. For completeness, we indicate by a

1565: filled circle the value of the ambient density at which the simulated

1566: SNR radius matches the one observed for each model. An angular BW

1567: radius of $14.6\arcmin$ (resp. $256\arcsec$) and a distance to the SNR

1568: of $2.0$ kpc ($2.8$ kpc) were assumed for SN 1006 (resp. Tycho),

1569: resulting in a physical BW radius of $\sim 8.5$ pc (resp. $\sim 3.5$

1570: pc). Larger values for the physical radius (due for instance to a

1571: larger distance to the remnant) would correspond to lower ambient

1572: densities.}

1573: \label{fig_n0_vs_ratio} 

1574: \end{figure*}

1575: 

1576: \begin{figure*}[t] 

1577: \centering 

1578: \includegraphics[width=13cm]{f9.eps} 

1579: \caption{Various models assuming different azimuthal profiles for the

1580: injection rate, $\eta_{\mathrm{inj}}$ (\textit{top}), and far

1581: upstream magnetic field, $B_{\mathrm{u}}$ (\textit{second, solid lines}). Each

1582: model predicts azimuthal variations of the ratio of radii between the

1583: BW and CD, $R_{\mathrm{BW}}/R_{\mathrm{CD}}$ (\textit{third}), and

1584: synchrotron brightness (projected along the line-of-sight) at different 

1585: frequencies (\textit{bottom}).}

1586: \label{fig_heuristic_model_profiles} 

1587: \end{figure*}

1588: 

1589: \begin{figure*}[t] 

1590: \centering \includegraphics[bb=0 60 720 720,clip,width=18cm]{f10.eps}

1591: \caption{\textit{Left column:} azimuthal profiles assumed for the

1592: injection efficiency, $\eta_{\mathrm{inj}}$ (\textit{panel a}), far

1593: upstream magnetic field, $B_{\mathrm{u}}$ (\textit{panel b}) and

1594: diffusion coefficient normalized to the Bohm-limit value, $k_0$

1595: (\textit{panel c}).  \textit{Middle column:} model profiles of the

1596: overall density ($r_{\mathrm{tot}}$) and magnetic field ($r_{B}$)

1597: compression ratios (\textit{panel d}), immediate post-shock value of

1598: the magnetic field, $B_{\mathrm{d}}$ (\textit{panel e}) and maximum

1599: energies (\textit{panel f}).  \textit{Right column:} predicted

1600: (\textit{thick} lines) and observed (\textit{thin} histograms)

1601: azimuthal profiles for the ratio of radii between the BW and CD,

1602: $R_{\mathrm{BW}}/R_{\mathrm{CD}}$ (\textit{panel g, solid lines}) and

1603: line-of-sight projected synchrotron brightness at several frequencies

1604: (\textit{panels h and i}).}

1605: \label{fig_model_k0_var_profiles} 

1606: \end{figure*}

1607: 

1608: \begin{figure*}[t] 

1609: \centering 

1610: \includegraphics[width=6cm]{f11a.eps} 

1611: \includegraphics[width=6cm]{f11b.eps} 

1612: \includegraphics[width=6cm]{f11c.eps} 

1613: \includegraphics[width=6cm]{f11d.eps} 

1614: \caption{Projected morphology predicted in the case of polar caps (see

1615: Appendix \ref{app-projection}) with $\Delta \tilde{R}=5 \times 10^{-2}$

1616: (\textit{top-left}), $\Delta \tilde{R} = 10^{-2}$ (\textit{top-right}),

1617: $\Delta \tilde{R} = 5 \times 10^{-3}$ (\textit{bottom-left}), $\Delta \tilde{R}$

1618: linearly varying from $2 \times 10^{-3}$ at the pole to $8 \times

1619: 10^{-3}$ at the equator (\textit{bottom-right}).  

1620: The radio and the high-energy X-ray

1621: morphology predicted by the model shown in Figure

1622: \ref{fig_model_k0_var_profiles} correspond roughly here to the

1623: top-right and bottom-right panels, respectively. In

1624: all panels, the intensity scaling is square-root.}

1625: \label{fig_projection} 

1626: \end{figure*}

1627: 

1628: %############################################## 

1629: 

1630: %          TABLES

1631: 

1632: %##############################################

1633: 

1634: 

1635: 

1636: \end{document}

1637: