1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2: %% Filename:      ms.tex

3: %% Version:       

4: %% Description:   

5: %% Author:        Sebastien Bongard-Canet <bongard@lyopc32.in2p3.fr>

6: %% Created at:    Tue Nov 13 16:24:13 2001

7: %% Modified at:   Mon May  5 13:22:55 2008

8: %% Modified by:   Eddie Baron <baron@ou.edu>

9: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

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

11: %\documentclass{emulateapj}

12: %\usepackage[T1]{fontenc} 

13: %\usepackage[english]{babel}    % langue 

14: %\RequirePackage{amsfonts}

15: %\RequirePackage{amssymb}

16: 

17: % Polices and general def  ========================================

18: 

19: % Depending on the fonts used, try the 2 commands

20: %\newcommand{\scname}[1]{{\normalfont\scshape #1}}

21: \newcommand{\scname}[1]{\textsc{#1}}

22: 

23: % Eddie stuff, remember to ask him what this does

24: \def\ifundefined#1{\expandafter\ifx\csname#1\endcsname\relax}

25: \ifundefined{ensuremath}\def\ensuremath#1{\relax\ifmmode{#1}}

26: \else${#1}$\fi\else\relax\fi

27: 

28: %\def\doublespace{\setlength{\baselineskip}{23pt}}

29: %\def\singlespace{\setlength{\baselineskip}{14pt}}

30: 

31: %\newcommand{\articlesize}{

32: %\doublespace

33: %\setlength{\textwidth}{6.5in}

34: %\setlength{\textheight}{8.4in}

35: %\voffset=-0.75in

36: %\hoffset=-0.55in}

37: \newcommand{\checkit}{\fbox{{\tiny$\surd$}}}

38: %\newcommand{\rem}[1]{\textbf{[\textsl{#1}]}}

39: %\newcommand{\rem}[1]{\textsl{[\textbf{#1}]}}

40: 

41: % Acronymes & abbreviations ========================================

42: 

43: % Instituts 

44: \newcommand{\cral}{\scname{cral}}

45: \newcommand{\ipnl}{\scname{IPNL}}

46: \newcommand{\SF}{\scname{Supernov\ae{} Factory}}

47: \newcommand{\SCP}{\scname{SCP}}

48: \newcommand{\Highz}{\scname{High-z}}

49: 

50: % Instruments

51: 

52: \newcommand{\snifs}{\scname{SNIFS}} 

53: \newcommand{\snls}{\scname{SNLS}} 

54: \newcommand{\snap}{\scname{SNAP}} 

55: 

56: % Codes

57: \newcommand{\phoenix}{\texttt{PHOENIX}} 

58: \newcommand{\phx}{\texttt{PHOENIX}}

59: \newcommand{\synow}{\scname{SYNOW}} 

60: 

61: 

62: %Cosmology

63: \newcommand{\OM}{\ensuremath{\Omega_{M}}}      % 

64: \newcommand{\OK}{\ensuremath{\Omega_{k}}}      % 

65: \newcommand{\OL}{\ensuremath{\Omega_{\Lambda}}}      % 

66: \newcommand{\dmz}{\ensuremath{\Delta_{m_{15}}}}

67: 

68: % Units

69: 

70: \newcommand{\electron}{e$^{-}$}      % electron

71: \newcommand{\dC}{$^{\circ}$C} % degre celsius

72: 

73: \newcommand{\kms}{\ensuremath{\textrm{km}~\textrm{s}^{-1}}}

74: \newcommand{\arcsecsq}{\ensuremath{\textrm{arcsec}^{2}}}

75: \newcommand{\surfbright}{\ensuremath{\textrm{mag}/\textrm{arcsec}^{2}}} 

76: \newcommand{\Reff}{\ensuremath{R_{e}}}

77: 

78: \newcommand{\Msun}{\ensuremath{M_{\odot}}}

79: \newcommand{\Lsun}{\ensuremath{L_{\odot}}}

80: 

81: \newcommand{\gcm}{g~cm$^{-3}$}

82: \newcommand{\kmps}{\ensuremath{\mbox{km}~\mbox{s}^{-1}}}

83: \newcommand{\is}{\mbox{s}$^{-1}$}

84: \newcommand{\iyr}{\mbox{yr}$^{-1}$}

85: \newcommand{\msol}{\ensuremath{{\mbox{M}_\odot}}}

86: \newcommand{\foe}{\ensuremath{10^{51}}}

87: 

88: % Astronomical terminology

89: \newcommand{\supernovae}{supernov\ae} 

90: \newcommand{\Supernovae}{Supernov\ae} 

91: \newcommand{\SNIa}{{SN~Ia}} 

92: \newcommand{\SNeIa}{{SNe~Ia}} 

93: \newcommand{\WD}{\textbf{WD}} 

94: 

95: 

96: % Spectroscopy

97: %\newcommand{\SiII}{Si~\scname{ii}}

98: %\newcommand{\SiIII}{Si~\scname{iii}}

99: %\newcommand{\FeII}{Fe~\scname{ii}}

100: %\newcommand{\FeIII}{Fe~\scname{iii}}

101: %\newcommand{\CaII}{Ca~\scname{ii}}

102: %\newcommand{\CoII}{Co~\scname{ii}}

103: %\newcommand{\NiII}{Co~\scname{ii}}

104: %\newcommand{\TiII}{Ti~\scname{ii}}

105: %\newcommand{\SII}{S~\scname{ii}}

106: %\newcommand{\CII}{C~\scname{ii}}

107: %\newcommand{\OII}{O~\scname{ii}}

108: %\newcommand{\HeII}{He~\scname{ii}}

109: \newcommand{\SiII}{\ion{Si}{2}}

110: \newcommand{\SiIII}{\ion{Si}{3}}

111: \newcommand{\FeII}{\ion{Fe}{2}}

112: \newcommand{\FeIII}{\ion{Fe}{3}}

113: \newcommand{\CaII}{\ion{Ca}{2}}

114: \newcommand{\CoII}{\ion{Co}{2}}

115: \newcommand{\NiII}{\ion{Ni}{2}}

116: \newcommand{\TiII}{\ion{Ti}{2}}

117: \newcommand{\SII}{\ion{S}{2}}

118: \newcommand{\CII}{\ion{C}{2}}

119: \newcommand{\OII}{\ion{O}{2}}

120: \newcommand{\HeII}{\ion{He}{2}}

121: 

122: 

123: \newcommand{\SiIIred}{$\lambda 6355$}

124: \newcommand{\SiIIblue}{$\lambda 5972$}

125: \newcommand{\TiIIRSi}{$\lambda 6013$}

126: 

127: \newcommand{\RSi}{$\Re_{Si}$}

128: \newcommand{\RCa}{$\Re_{Ca}$}

129: \newcommand{\RCaS}{\ensuremath{\Re_{CaS}}}

130: \newcommand{\RSiS}{\ensuremath{\Re_{SiS}}}

131: \newcommand{\RSiSS}{\ensuremath{\Re_{SiSS}}}

132: 

133: \newcommand{\vSiII}{\ensuremath{v_{\textrm{\tiny\SiII}}}}

134: \newcommand{\vSiIIdot}{\ensuremath{\dot{v}_{\textrm{\tiny\SiII}}}}

135: 

136: \newcommand{\gamray}{$\gamma$-ray}

137: \newcommand{\halpha}{H$\alpha$}

138: \newcommand{\hbeta}{H$\beta$}

139: % Nuclear physics

140: \newcommand{\nni}{\nuc{56}\mbox{Ni}}

141: \newcommand{\xni}{\ensuremath{\mbox{X}_{\mbox{Ni}}}}

142: 

143: % Radiative transfer

144: \newcommand{\nomw}{W7}

145: \newcommand{\lte}{LTE}

146: \newcommand{\nlte}{NLTE}

147: 

148: \newcommand{\I}{\ensuremath{I(\vecr, \vecn, \nu, t)}}

149: \newcommand{\Inu}{\ensuremath{I_{\nu}}}

150: \newcommand{\Jnu}{\ensuremath{J_{\nu}}} 

151: \newcommand{\Snu}{\ensuremath{S_{\nu}}} 

152: 

153: \def\ang{\mbox{\AA}}

154: \def\Tmod{\ensuremath{T_{\mbox{model}}}}

155: \def\Teff{\ensuremath{T_{\mbox{model}}}}

156: \def\teff{\ensuremath{T_{\mbox{model}}}}

157: \def\tstd{\ensuremath{\tau_{\mbox{std}}}}

158: \def\Rzero{\ensuremath{R_0}}

159: %Vectors

160: \newcommand{\vecr}{\mathbf{r}}

161: \newcommand{\vecv}{\mathbf{v}}

162: \newcommand{\vecn}{\mathbf{n}}

163: \newcommand{\vecdS}{\mathbf{dS}}

164: 

165: %Maths

166: \def\pder#1#2{{\partial #1 \over \partial #2}}

167: \newcommand{\fst}{\ensuremath{1^{\textrm{st}}}}

168: \newcommand{\snd}{\ensuremath{2^{\textrm{nd}}}} 

169: \newcommand{\trd}{\ensuremath{3^{\textrm{rd}}}} 

170: \def\xth#1{{#1\ensuremath{^{\textrm{th}}}}}

171: \ifundefined{nuc}\def\nuc#1#2{\relax\ifmmode{}^{#1}{\protect\textrm{#2}}

172: \else${}^{#1}$#2\fi}\else\relax\fi 

173: 

174: \def\alog#1{\times 10^{#1}}

175: 

176: \def\la{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$<$}}}}

177: \def\ga{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$>$}}}}

178: \newcommand{\be}{\begin{eqnarray}}

179: \newcommand{\ee}{\end{eqnarray}}

180: 

181: 

182: %colors

183: \newcommand{\red}[1]{{\color{red} #1}}

184: \newcommand{\green}[1]{{\color{green} #1}}

185: \newcommand{\blue}[1]{{\color{blue} #1}}

186: \newcommand{\orange}[1]{{\color{orange} #1}}

187: 

188: \newcommand{\rem}[1]{\textsl{[\textbf{#1}]}}

189: 

190: %Graphics

191: \graphicspath{{./}{redone/}}

192: %\graphicspath{{figOk/}{newfigs/}}

193: %\graphicspath{{apjfigs/}}

194: %\DeclareGraphicsExtensions{{.eps},{.ps}}

195: %\DeclareGraphicsExtensions{{.eps}}

196: 

197: %\received{} 

198: %\accepted{} 

199: %\journalid{}{} 

200: %\articleid{}{} 

201: \shortauthors{Bongard, S. et~al.}

202: \shorttitle{Spectral Formation in SNe~Ia}

203: 

204: 

205: % Document ============================================================

206: \bibliographystyle{apj}

207: 

208: \begin{document}

209: 

210: \title{Multi-layered Spectral Formation in \SNeIa\ Around Maximum Light} 

211: 

212: 

213: \author{ 

214:   Sebastien Bongard\altaffilmark{1,2,4},

215:   E.~Baron\altaffilmark{2,3},

216:   G.~Smadja\altaffilmark{4},  David

217: Branch\altaffilmark{2},  and  Peter

218: H.~Hauschildt\altaffilmark{5}

219:  }

220: 

221: \altaffiltext{1}{Physics Division, Lawrence Berkeley

222:   National Laboratory, MS 50A-5014, 1 Cyclotron Rd, Berkeley, CA

223:   94720-8139 USA; email: {sbongard@lbl.gov}}

224: 

225: \altaffiltext{2}{Homer L.~Dodge Department of Physics and Astronomy, University of

226: Oklahoma, 440 West Brooks, Rm.~100, Norman, OK 73019, USA; email:

227: {baron@ou.edu, {branch@nhn.ou.edu}}}

228: 

229: 

230: \altaffiltext{3}{Computational Research Division, Lawrence Berkeley

231:   National Laboratory, MS 50F-1650, 1 Cyclotron Rd, Berkeley, CA

232:   94720-8139 USA}

233: 

234: 

235: \altaffiltext{4}{Institut de Physique Nucl\'eaire Lyon, B\^atiment Paul Dirac

236: Universit\'e Claude Bernard Lyon-1

237: Domaine scientifique de la Doua

238: 4, rue Enrico Fermi

239: 69622 Villeurbanne cedex, France; email: {smadja@in2p3.fr}}

240: 

241: \altaffiltext{5}{Hamburger Sternwarte, Gojenbergsweg 112,

242: 21029 Hamburg, Germany; email: {yeti@hs.uni-hamburg.de}}

243: 

244: \begin{abstract}

245:   We use the radiative transfer code \phx\ to study the line formation

246:   of the wavelength region 5000--7000~\ang.  This is the region where

247:   the SNe~Ia defining Si~II feature occurs. This region is important

248:   since the ratio of the two nearby silicon lines has been shown to

249:   correlate with the absolute blue magnitude. We use a grid of LTE

250:   synthetic spectral models to investigate the formation of line

251:   features in the spectra of SNe~Ia. By isolating the main

252:   contributors to the spectral formation we show that the ions that

253:   drive the spectral ratio are \FeIII, \FeII, \SiII\ and \SII. While

254:   the first two strongly dominate the flux transfer, the latter two

255:   form in the same physical region inside of the supernova. 

256:   We also show that the

257:   na\"ive blackbody that one would derive from a fit to the observed

258:   spectrum is far different than the true underlying continuum.

259: \end{abstract}

260: 

261: \keywords{cosmology --- stars: atmospheres --- supernovae} 

262: 

263: %\tableofcontents

264: 

265: \section{Introduction}

266: 

267: Type Ia \supernovae\ have been used as ``standardizable candles'' for

268: more than 10 years, thanks to the correlation between their light

269: curve shape and their absolute blue magnitude at maximum light. This

270: correlation is well matched by a spectroscopic sequence

271: \citep{nugseq95} defined by the ratio, \RSi, of the depth

272: of two absorption features usually identified as \SiII\  \SiIIblue\ and

273: \SiIIred\ lines. Using the radiative transfer code \phoenix\

274: \citep[][and references therein]{hbjcam99,bhpar298,hbapara97,phhnovetal97,phhnovfe296},

275: this sequence has been matched 

276: to a temperature sequence that could physically be related to the

277: amount of \nni\ produced in the context of Chandrasekhar-mass white

278: dwarf explosions.   

279: 

280: \citet{hathighv00} showed that \RSi\ and the velocity derived from the

281: minimum of the $\approx 6100$~\AA\ feature correlated poorly,

282: suggesting that a one-parameter description of \SNeIa\ was

283: insufficient.  This velocity derived from the P-Cygni feature

284: associated with the \SiII\ \SiIIred\ line, and its time evolution

285: break \SNeIa\ into subclasses depending upon more than one parameter

286: \citep{benetti05}.  \citet{branchcomp206} and \citet{hach06} also

287: showed that different spectroscopic indicators exist which can be used

288: to distinguish variations even within the ``normal'' \SNeIa.  In the

289: context of Chandrasekhar-mass explosions, these variations cannot be

290: interpreted only in terms of differences in the amount of \nni\

291: produced.  The velocity scatter could be due to properties of the

292: progenitor system that impact on the kinetic energy released during

293: the explosion. For example, the ratio ($^{54}$Fe+$^{58}$Ni)/$^{56}$Ni

294: has been proposed as a second physical parameter which could account

295: for this dispersion \citep{mazzpod06}.

296: 

297: This paper is an attempt to unravel the spectral formation in the

298:  $5000-7000$~\AA region (henceforth the \RSi\ wavelength region),

299:  as a way to address the  

300: physical reasons for their correlation with luminosity, and especially

301: the link between this correlation and the temperature sequence of

302: \SNeIa.

303: A previous attempt using  the radiative transfer code \synow\

304: \citep{fisher00}, was performed by \citet{garn99by04} who explained  

305: the counter-intuitive correlation of \RSi\ with luminosity by the

306: temperature sensitivity of a \TiII\ line. 

307: On the other hand

308: synthetic spectral investigations  \citep{stehle02bo05,branchcomp206}

309: failed to find any important \TiII\ line contributions in this part of

310: the spectrum.      

311: 

312: 

313: We use the radiative transfer code \phx\ differently than it has

314: usually been used previously. Instead of fitting a supernova by varying the time

315: and bolometric luminosity, we use a grid of simulated models to probe

316: the line formation process. Our input model is the parametrized

317: deflagration model W7 \citep{nomw7}, homologously expanded to

318: calculate the abundances at any given time.  The knowledge of the

319: abundance structure of the explosion model, and the self-consistent

320: physical structure provided by \phoenix\ are used to describe more

321: precisely how the spectrum is formed in \SNeIa.  \citet{bongard06a}

322: introduced the line ratio \RSiS, which is defined as the ratio of two

323: integrals of the flux over a chosen wavelength range, where each integral

324: is centered around the two lines used in the original definition of

325: \RSi\ (see \S\ref{rsisdef}). It is designed to be useful for spectra

326: which have lower signal 

327: to noise than well-observed nearby SNe~Ia.

328: 

329: This understanding of the formation of the spectrum allows us to

330: explain why the line ratio \RSiS\ is a good probe of the temperature

331: sequence, and reinforces the link between this sequence and the

332: absolute blue magnitude of \SNeIa.  These results also allow us to

333: rule out a \TiII\ contribution to the correlation of \RSi\ with

334: luminosity.  We emphasize that even though the synthetic spectra

335: presented here \emph{do not} do a good job of reproducing observed

336: SNe~Ia spectra, we consider W7 to be the fiducial SNe~Ia model.

337: \citet{bbbh06} showed that reasonably good synthetic spectra can be

338: produced with W7 near maximum light. We make differential comparisons

339: based on that model, keeping in mind that \citet{bbbh06} emphasized

340: that W7 did not reproduce the wavelength region that we examine in

341: detail here.

342: 

343: 

344: % a word on time evolution 

345: 

346: We will address time

347: evolution only qualitatively. 

348: The goal of this paper is to understand the temperature sequence of

349: SNe~Ia spectra around maximum light, thus we concentrated on day 20 

350: after explosion, since maximum light in $B$ in SNe~Ia has been shown  

351: to occur 18--25 days after explosion.

352: 

353: 

354: \section{Analysis tool: \phx\ simulations}

355: \subsection{The LTE grid}

356:  

357: Using the multi-purpose radiative transfer code \phx\ we converged LTE

358: calculations for the \nomw\ explosion model at 10, 15, 20, and 25 days

359: after explosion, for a range of bolometric luminosities spanning the

360: ``normal'' \supernovae\ blue magnitudes.

361: 

362: 

363: \subsection{Single Ion spectra\label{subsec:sing_ion}}

364: 

365: Each model converged with \phx\ not only provides a spectrum, but also

366: the knowledge of the physical structure and line optical depth in the

367: supernova. This gives us the ability to compute ``single ion 

368: spectra'' as follows: Using the converged output of the \phx\

369: simulation, we artificially  turn off all but the

370: continuum and one single ion line opacities. The

371: ``continuum opacities'' denote all the bound-free and free-free

372: opacities as well as electron and Rayleigh

373: scattering.  We then recalculate the solution of

374: the scattering problem in order to get the ``single ion

375: spectrum'', with the level populations and free electron number kept

376: fixed.  

377: 

378: 

379: The relative strength of the absorption and emission features in this

380: spectrum give an indication of the contribution of each line to the

381: complete spectrum. Since the solution of the scattering

382: problem is recalculated, the features are \emph{not} expected to look

383: exactly the same as in the complete spectrum.  In any case, due to the

384: global nature of the transfer equation, the

385: full spectrum in \emph{not} the linear combination of all the single

386: ion spectra.

387: 

388: In these single ion spectra, the flux transfer from the blue

389: to the red due to scattering will be smaller, due to  the opacity

390: decrease caused by the suppression of most of the ions. They will

391: thus appear bluer than the complete spectrum. The back scattering will

392: also be decreased for the same reason, resulting in ``single ion

393: spectra'' brighter than the full synthetic spectrum. 

394: 

395: The ratio of scattering to absorption is larger in the single

396: ion spectra compared to the full synthetic spectrum because the

397: full scattering problem is solved with fixed level populations and

398: fewer atomic lines. The same effect enhances stimulated emission in the

399: single ion spectra, resulting in strong net emission in the

400: P-Cygni profiles.

401: 

402: Since the single ion spectra are always brighter than the

403: full synthetic spectrum we must normalize the luminosity in order to

404: compare the two spectra. We do this by setting the single ion

405: bolometric luminosity to that of the full synthetic spectrum. This

406: procedure is

407: trivial since the synthetic spectra include fluxes at all wavelengths.

408: 

409: The special case we name ``continuum-only spectrum'' is that where all

410: line opacities are set to zero, and therefore the spectrum is due to

411: pure continuum processes.  In a moving atmosphere a

412: photoionization edge can produce a feature that looks like a P-Cygni

413: line \citep{b94i2}. We can turn on and off different

414: species individually, thus we can, for example, have a ``two-ion''

415: spectrum that consists of, for example, only Si~II and S~II.

416: 

417: 

418: \section{Using single ion spectra: ruling out a \TiII\

419:   contribution to \RSi}

420: 

421: \subsection{The \RSi\ puzzle }

422: 

423: Because of their similar redshift with respect to \SiII\ \SiIIred\ and

424: \SiII\ \SiIIblue\ lines it is tempting to consider the $6100$~\AA\

425: and  $5800$~\AA\ troughs to be part of their P-Cygni  profiles. These two

426: lines share the 4P level, the \SiIIred\ line is the 

427: 4S-4P transition and the \SiIIblue\ is the 4P-5S transition. Their optical

428: depths can thus be estimated in the Sobolev approximation using \citep{atlas99}:

429: \begin{equation}

430:   \label{eq:tauSi}

431:   \tau=(\frac{\pi e^{2}}{m c})f\, \lambda \, t \, n_{l}\, [1-\frac{g_{l}

432:       n_{u}}{g_{u} n_{l}}],

433: \end{equation}

434: where $n_{l}$ and $n_{u}$ are the number densities

435: of the lower and upper level associated with the transition, $f$ is its

436: oscillator strength, $t$ is the time since explosion, $g_{l}$ and

437: $g_{u}$ are the statistical weights of the levels,  and $\lambda$ is the

438: wavelength of the transition. Assuming thermal equilibrium, the 

439: ratio of the \SiII\ \SiIIred\ line optical depth ($\tau_{red}$) over

440: the \SiII\ \SiIIblue\ line optical depth ($\tau_{blue}$)  becomes: 

441: \begin{equation}

442:   \label{eq:SiTauRedBlue}

443:   \frac{\tau_{red}}{\tau_{blue}} \propto

444:   \frac{\frac{1}{3}~e^{\Delta_{E_{red}}/(k_{B}T)}-1}{1-3~e^{-\Delta_{E_{blue}}/(k_B

445:       T)}}

446: \end{equation}

447: where $\Delta E$ is the energy difference between the upper and

448: lower levels of the transition.

449: This ratio decreases monotonically with respect to a

450: temperature increase, as also shown in Fig.~\ref{fig:RSiseqTex},

451: where \synow\  was used to compute \SiII\ spectra for excitation

452: temperatures ranging from 5000~K to 40000~K.  

453:  

454: 

455: 

456: 

457: Since P-Cygni troughs become deeper with increasing optical depth, a  

458: higher \SiII\ excitation temperature should increase \RSi. Under the

459: reasonable assumption that  higher luminosity \SNeIa\ have

460: higher \SiII\ excitation temperatures, \RSi\ should increase and    

461: not decrease with luminosity. Observations show the opposite

462: behavior. 

463: 

464: \clearpage

465: \begin{figure}[ht]

466: \centering

467: \includegraphics[width=0.70\textwidth, clip]{f1}

468: \caption{\SiII\ in \synow\ for increasing excitation 

469:   temperatures.

470:   The trough of the blue feature due to \SiII\

471:   \SiIIblue\ increases relative to the trough of the

472:   \SiII\ \SiIIred\ feature 

473:   as temperature increases, as predicted by

474:   Eq.~\ref{eq:SiTauRedBlue}. The \SiII\ reference line is \SiII\

475:   \SiIIred\, and only the excitation temperature is varied, which

476:   explains the $\sim 6100$~\AA\ trough stability. (Note that the

477:   unrealistic excitation temperature of 40000K leads to a very strong

478:   feature around 5400~\AA\ that is not observed in SNe~Ia spectra.)

479: } 

480: \label{fig:RSiseqTex}

481: \end{figure}

482: \clearpage

483: 

484: 

485: Fig.~\ref{fig:RSiTempComp} displays the temperature structure from

486: our grid of calculations 20 days after explosion with increasing

487: bolometric luminosities. We restricted the plot to the

488: $9000-16000$~\kms\ region, where silicon is found in W7. This shows the

489: physical temperature increases with bolometric luminosity as one

490: would expect. In LTE, a temperature increase results in a \SiII\

491: excitation temperature 

492: increase with luminosity which would lead to an \RSi\ increase with

493: luminosity. Therefore, unless one assumes a temperature 

494: structure inversion in the $9000-16000$~\kms\ region due to NLTE

495: effects, \SiII\ lines alone are unable to explain the trend of the

496: \RSi\ correlation with \SNeIa\ blue magnitude. 

497: 

498: \clearpage

499: \begin{figure}[ht]

500: \begin{center}

501: %\includegraphics[width=0.70\textwidth, clip]{fig2_redone}

502: \includegraphics[width=0.70\textwidth, clip]{f2}

503: \caption{Temperature 

504: for day 20 models in the $8000$-$16000$~\kms\ region

505: where \SiII\ ions are found, as a function of the model parameter

506: $M_{bol}$. The horizontal blue line marks the 7000 K limit at which

507: \TiII\ lines have been suggested to dominate over \SiII\ \SiIIblue\ 

508: \citep{garn99by04}.}

509: \label{fig:RSiTempComp}

510: \end{center} 

511: \end{figure}

512: \clearpage

513: 

514: \subsection{\TiII\ impact on the ``\RSi\ wavelength region''}

515: 

516: \citet{garn99by04} proposed that \RSi\ variation with luminosity could be

517: accounted for by a blend of  \TiII\ lines. We focus on the

518: behavior of the \TiIIRSi\ line. \citet{garn99by04} suggested that the

519: the \RSi puzzle, described above could be solved by the

520: temperature 

521: evolution of this line, which increases much faster than the \SiII\

522: \SiIIblue\ line decreases below 7000~K. 

523: 

524: In order to probe this assumption we display in

525: Fig.~\ref{fig:phxTiIISiII} the day 20 two-ion spectra of \SiII\ and

526: \TiII\ alone as well as the full spectrum for our lower 

527: luminosity model. 

528: We also display the continuum-only spectrum as a reference. 

529: For \SiII, the 6100~\AA\ trough and the concomitant emission peak

530: are, as expected, dominated by \SiII\ lines, but the  \SiII\ \SiIIblue\

531: P-Cygni profile lacks the full spectrum blue edge, hinting at a

532: missing contribution to the full spectrum line profile.  

533: This other contribution can not be \TiII\ as no \TiII\ lines appear in

534: the \RSi\ wavelength region, even though the \TiII\ \TiIIRSi\ line has

535: been checked to form between 12000 and 16000~\kms, where the

536: temperature is below 7000~K as indicated by the black line in

537: Fig.~\ref{fig:RSiTempComp}.   

538: 

539: One could argue that this line does not appear in our two-ion

540: spectrum because of a too low abundance of titanium in \nomw. We rule

541: out the low abundance in W7 as an explanation for this effect since

542: the \TiII\ single ion spectrum shows a strong feature to the blue of

543: 5000~\AA, which does a reasonably good job of reproducing the observed trough

544: seen in fast decliner SNe~Ia such as SN~1991bg.

545: 

546: 

547: \clearpage

548: \begin{figure}[ht]

549:   \centering

550: %  \includegraphics[width = 0.7\textwidth, clip]{fig3_redone}

551:   \includegraphics[width = 0.7\textwidth, clip]{f3}

552:   \caption{The single ion day 20 spectrum of \SiII\ and \TiII\ alone as

553:     well as the full spectrum for our lower luminosity model. We also

554:     display the continuum-only spectrum as a

555:     reference.}

556:   \label{fig:phxTiIISiII}

557: \end{figure}

558: \clearpage

559: 

560: Fig.~\ref{fig:RSiTempComp} shows the \phoenix\ temperature structure

561: between 10000 and 16000~\kms, where the \SiII\ and \TiII\ lines of the

562: \RSi\ wavelength region form in all our synthetic spectra models.

563: Using \synow\ we were able to fit the \RSi\ wavelength region with

564: \SiII\ and \TiII\ lines alone, but required a \TiII\ excitation temperature

565: of 40000~K. This temperature is way too high compared to the  detailed

566: \phx\ results, but note that the temperature in \synow\ is dependent

567: on the chosen reference line. However, even with \synow, one

568: expects that \TiII\ line strength increases with \emph{decreasing}

569: temperature. What is happening in this case, is that in order to

570: obtain a noticeable contribution of \TiII\ (with respect to \SiII), we

571: must go to very high excitation temperatures combined with extremely

572: large optical depths in \TiII: the 

573: temperature dependence is not independent of the \TiII\ optical depth,

574: which in \synow\ is a free parameter.

575: 

576: This very high excitation temperature does not necessarily indicate that the

577: assumptions of \synow\ have broken down, since the \phoenix\ and

578: \synow\ \TiII\ spectra agree quite well for similar temperatures. 

579: In Fig.~\ref{fig:SiTiSynowPhx} we display the comparison between the

580: \synow\ results and the \phx\ results in the temperature range

581: appropriate to the physical results from \phx.  The 

582: \SiII\ and the \TiII\ line strengths agree.  We also agree with the point of

583: \citet{garn99by04} that the temperature dependence of the 

584: strength of \TiII\ is such that the line-strength increases

585: dramatically below 7000~K; however, the  relative strength of the line

586: with respect to \SiII\ does not increase fast enough for the

587: \TiII\ line to become important at reasonable physical

588: temperatures. This is due to the fact that the initial line

589: strength of \TiII\ is so small.

590: \clearpage

591: \begin{figure}[ht]

592:   \includegraphics[width = 0.47\textwidth, clip]{f4a}

593:   \includegraphics[width=0.47\textwidth]{f4b}

594:   \caption{\synow\ and \phoenix\ single ion spectra 

595:     with \synow\ excitation temperature

596:     matching \phoenix\ temperature structure. Left panel: \SiII, right

597:   panel \TiII.}

598:   \label{fig:SiTiSynowPhx}

599: \end{figure}

600: \clearpage

601: 

602: 

603: 

604: \section{Multi-layered spectrum formation}

605: 

606: \subsection{The deepest layer}

607: 

608: 

609: 

610: The multi-layer spectrum formation picture alters the way that one

611: thinks about how spectrum formation occurs. In the photospheric

612: picture, one thinks about features forming in a reversing layer, above

613: a true continuum. In the multi-layer picture, features can form

614: throughout the supernova atmosphere and can in principle imprint a

615: shape into the ``continuum'' that is altered higher up in the

616: atmosphere. The most important difference between the multi-layer

617: picture and the photospheric picture is that features from multiple

618: ionization stages can strongly affect the overall spectrum.

619: 

620: \clearpage

621: \begin{figure}[ht]

622: \centering

623:   \includegraphics[width = 0.45\textwidth,clip]{f5a}\\

624:   \includegraphics[width = 0.45\textwidth,clip]{f5b}

625:   \caption{The na\"ive blackbody fit would peak where the full

626:     synthetic spectrum does, that is around 5000~\ang, far to the red

627:     of the physical blackbody. 

628:     Top Panel: Full \phx\ day 20 spectrum in black,

629:     continuum-only spectrum forming below 2000~\kms\ in dotted blue,

630:     $13685K$ blackbody in plain blue. Bottom Panel: Continuum-only day 20

631:     spectrum in dotted blue, and $13685K$ blackbody in plain

632:     blue.}

633:   \label{fig:phxOnlyNoone}

634: \end{figure}

635: \clearpage

636: 

637: In Fig.~\ref{fig:phxOnlyNoone} (top panel) we display the deepest layer of the 

638: spectral formation of our lower bolometric luminosity model: the

639: continuum-only spectrum.

640: It forms below 3000~\kms\ for all our range

641: of bolometric luminosities, where the continuum optical depth becomes

642: greater than one,  which is at a much

643: lower velocity than where the photosphere is usually considered to be

644: in the  photospheric model. 

645: 

646: 

647: 

648: Figure~\ref{fig:phxOnlyNoone} (bottom panel) clearly shows that the na\"ive

649: blackbody fit (that is the blackbody which peaks at the same

650: wavelength as the smoothed full synthetic spectrum) to the full

651: synthetic spectrum would give a blackbody temperature that is much

652: cooler than the true underlying continuum. While the  

653: continuum-only spectrum is brighter and bluer (as described in

654: \S\ref{subsec:sing_ion}) Fig.~\ref{fig:phxOnlyNoone} (bottom

655: panel) shows that 

656: the blackbody obtained from the physical temperature 13685~K is also

657: significantly bluer than the na\"ive blackbody. This is a very

658: important point since it shows that the na\"ive blackbody has nothing

659: to do with the real physical conditions.

660: 

661: It is crucial to keep this point in mind when analyzing SNe~Ia

662: features, especially when using techniques like principle component

663: analysis \citep[PCA,][]{james_pca06}.

664: It is tempting to subtract a ``continuum'' case by

665: case in order to maximize the contribution of the lines. Since    

666: the ``pseudo-continuum'' is responsible for a large part of the 

667: spectral structure, this procedure yields the opposite result: the

668: subtracted  ``continuum'' adds a fake relationship between features. 

669: The PCA eigenvectors will thus no longer account only for real SNe~Ia

670: diversity, but also for the arbitrary continuum.

671: 

672: The blackbody of Fig.~\ref{fig:phxOnlyNoone} (bottom panel)

673: corresponds to the temperature where $\tau_{e} \sim 1$. Even at depth

674: of 2000~\kms\ ($\tau \sim 3 $ at 2000~\kms), the supernova

675: spectrum is \emph{not} a blackbody spectrum. The spectrum of the full

676: \nomw\ model is obtained from the steep narrow blue continuum-only spectrum

677: by line interactions that transfer flux toward the red by Doppler

678: shift or fluorescence. 

679: 

680: 

681: 

682: \subsection{Iron lines}

683: 

684: In Fig.~\ref{fig:FeIIFeIIIonly} we display the \FeII, \FeIII\ and

685: \FeII +\FeIII\ two-ion spectra of our faintest

686: model at day 20. The \RSi\ wavelength region ($5000-7000$~\AA) gives us some

687: insight into the spectral formation process because of the smaller

688: number of strong lines than  in the bluer spectral

689: regions. The \FeIII\ and  

690: the \FeII\ peaks at 5900~\AA\ and 6100~\AA\ are blends of weak

691: lines formed at velocities of $\sim$ 5000 and

692: $\sim$ 9000~\kms, respectively. It follows from the optical depths

693: shown in 

694: Fig.~\ref{fig:tauFeII} that the 5128~\AA\ \FeIII\ peak and the

695: 5170~\AA\ \FeII\ peak arise from strong lines formed at

696: velocities  of $\sim$ 7000 and $\sim$ 15000~\kms, respectively.   

697: 

698: Fig.~\ref{fig:FeIIFeIIIonly} shows how \FeIII\

699: and \FeII\ lines blend together. The strong \FeIII\ 5128~\AA\ feature is

700: shielded by the strong \FeII\ 5170~\AA\ one, and the two blends above

701: 5900~\AA\ merge together. 

702: 

703: \clearpage

704: \begin{figure}[ht]

705:   \centering

706:   \includegraphics[width=0.48\textwidth]{f6a}\\

707:   \includegraphics[width=0.48\textwidth]{f6b}

708:   \caption{\FeII\ and \FeIII\ single ion spectra for our faintest

709:     simulated spectra, 20 days after explosion. The top panel is

710:     \FeIII\ in red and \FeII\ in pink. The bottom panel is \FeII\ +

711:     \FeIII\ two-ion spectra in red. In both cases the black

712:     spectrum is the full simulated spectrum and the blue spectrum is

713:     the continuum-only spectrum. }

714:   \label{fig:FeIIFeIIIonly}

715: \end{figure}

716: 

717: 

718: 

719: In the photospheric model the spectrum is considered to be a blackbody

720: continuum emitted by the photosphere on top of which strong, 

721: well-separated lines of ions present in the line

722: forming 

723: region add structure mainly by scattering light.

724: We define the ``pseudo-continuum'' as the spectrum formed by

725: weak line blends which produces more structure than a blackbody

726: spectrum. The photosphere is replaced in this picture by

727: a ``pseudo-photosphere'' forming at depth depending on

728: wavelength. This concept is widely used in the radiative transfer community.

729: 

730: This ``pseudo-photosphere'' is clearly seen in the results of

731: \citet{branchcomp105} who were able to model the spectrum of the

732: normal SN~Ia 1994D using \synow\ at 115 days past maximum

733: light, at least in the blue. Specifically, the results of

734: \citet{branchcomp105} show that even at late times (thus in the

735: deepest layers) the spectrum is dominated by permitted lines that can

736: be treated in resonant scattering. 

737: 

738: Similarly, the line forming region of the pseudo-photospheric model

739: does not have a fixed depth. We define it as the region where the

740: strong lines that create the final structure of the spectrum form.

741: In this picture, the strong  \FeIII\ 5128~\AA\ feature does not belong

742: to the line forming region because it forms in the same physical

743: region as the weak \FeII\ lines of the pseudo-photosphere with which

744: it blends, as can be seen in Fig. \ref{fig:tauFeII}. It adds

745: structure to the pseudo-continuum and is shielded by the strong

746: \FeII\ 5170~\AA\ feature forming further out. 

747: The existence of strong lines that form deep and of weaker lines that

748: imprint their features further out in the atmosphere leads to a

749: convergence between the pseudo-photospheric region and the line

750: forming region, making it almost impossible to separate them into two

751: clearly distinct physical regions.

752: 

753: 

754: The \FeII+\FeIII\ spectrum we display in Fig.~\ref{fig:FeIIFeIIIonly}

755: can be considered to be the pseudo-continuum spectrum in the

756: $5000-7000$~\AA\ region. It is formed by a large peak due to  

757: \FeIII\ $\lambda$5128 and neighboring strong \FeII\ lines,

758: followed by a flat region forming between 5000 

759: and 9000~\kms\ depending on wavelength. This  shows that

760: by construction \RSi\ couples spectral regions that form at very

761: different depths. 

762: 

763: In the ``pseudo-photospheric'' model, the flux transfer from the blue

764: toward the red is dominated by \FeIII\ and \FeII\ weak line

765: blends. Of course, weak cobalt and nickel lines also contribute,

766: but the study of their effects will be postponed to a later paper in

767: which the more complicated spectral formation in the $\lambda <

768: 5000$~\AA\ region will be discussed. The colors of \SNeIa\ are therefore

769: dominated by weak lines of the heavy elements whose blending shapes

770: the pseudo-photosphere spectrum.   

771: 

772: 

773: 

774: \section{Iron lines and colors}

775: 

776: 

777: As the bolometric luminosity of our models is increased, the

778: temperature also rises, ionizing \FeII\ to \FeIII. The \FeII\

779: number abundance depletion decreases the 5170~\AA\ feature strength,

780: leaving \SII\ the dominant ion of the blue edge of the \RSi\

781: feature at $\sim$ 5455~\AA. 

782: This effect can be seen in

783: Figs.~\ref{fig:justfe_evol}--\ref{fig:allfour} where the \FeIII,

784: \FeII, \SiII, and \SII\ 

785: spectrum changes with bolometric luminosity are

786: displayed. The \FeII\ 5170~\AA\ feature strength decreases as \FeII\ is

787: ionized into \FeIII, unshielding the strong \FeIII\ 5128~\AA\ 

788: feature. Simultaneously, the flux transfer towards the red 

789: decreases because \FeII\ is much more efficient at transferring flux

790: toward the red than \FeIII\ as can be seen in

791: Fig.~\ref{fig:FeIIFeIIIonly}. This efficient flux transfer from

792: the blue to the red is due to numerous blue and UV \FeII\ lines that

793: amount to a large opacity because of line blanketing.

794: 

795: 

796: 

797: \clearpage

798: \begin{figure}[ht]

799:   \centering

800: %\includegraphics[width=0.8\textwidth]{BmVphx}\\

801: \includegraphics[width=0.8\textwidth]{f7}\\

802: \caption{$B-V$ color as a function of 

803:   $M_{bol}$ for the different epochs in our grid. 

804: }

805: \label{fig:bmvevolution}

806: \end{figure}

807: 

808: 

809: \begin{figure}[ht]

810:   \centering

811: \includegraphics[width=0.45\textwidth]{f8a}\\

812:   \includegraphics[width=0.45\textwidth]{f8b}

813: \caption{\FeII\ (top panel) and  \FeIII\ (bottom panel) day 20 single ion spectra as a function of

814:   $M_{bol}$. Each spectrum is rescaled to the bolometric flux of the

815:   corresponding full synthetic spectrum. Since \FeIII\ forms deeper it is

816:   less sensitive to a change in the bolometric luminosity. \FeII\ lines are

817:   more efficient at transferring flux from blue to red. 

818: }

819: \label{fig:justfe_evol}

820: \end{figure}

821: 

822: \begin{figure}[ht]

823:   \centering

824:   \includegraphics[width=0.45\textwidth]{f9a}\\

825:   \includegraphics[width=0.45\textwidth]{f9b}

826:   \caption{\FeII\ + \FeIII\  (top panel) and \SiII\ + \SII\

827:     (bottom panel) 

828:     two-ion day 20 spectra as a function of $M_{bol}$. The

829:     flux transfer from the blue to the red is dominated by the iron

830:     (and other iron peak elements in the full spectrum). All spectra

831:     are rescaled to have the same flux as the corresponding full

832:     synthetic spectra.} 

833: \label{fig:RSiSFeSi}

834: \end{figure}

835: 

836: 

837: \begin{figure}[ht]

838:   \centering

839:   \includegraphics[width=0.45\textwidth]{f10}

840: \caption{\FeIII+\FeII+\SiII+\SII\ four-ion day 20 spectra as a

841:   function of $M_{bol}$. 

842: }

843: \label{fig:allfour}

844: \end{figure}

845: \clearpage

846: 

847: \begin{figure}[ht]

848:  \centering

849:  \includegraphics[width=0.48\textwidth]{f11}

850:  \caption{Optical depth of the most important lines of the \RSi\

851:    wavelength region at day 20. Weak lines always form deeper than the strong

852:    ones for \FeII\ and \FeIII. \FeIII\ $\lambda$5128 and \FeII\

853:    $\lambda$ 5170 are displayed as typical examples of strong \FeIII\

854:    and \FeII\ lines in the \RSi\ wavelength region.}

855:  \label{fig:tauFeII}

856: \end{figure}

857: \clearpage

858: 

859: Figure \ref{fig:FeIIFeIIIonly} also shows that in the 5000-8000~\AA\

860: wavelength region, \FeIII\ features remain very stable over the range

861: of bolometric luminosities explored, while \FeII\ features vary much

862: more. This is not surprising, as \FeIII\ lines form much deeper, in a

863: region that is hotter and denser, where physical conditions vary less than in

864: the outer regions.

865: 

866: We see in Figure \ref{fig:RSiSFeSi} (top panel) that the \FeIII+\FeII\

867: two-ion 

868: spectrum undergoes a strong  increase of flux below 5500~\AA,

869: while it remains very stable above, when the bolometric luminosity

870: increases. This translates into a strong variation in color of the

871: full synthetic spectrum. 

872: On the other hand, the \SiII+\SII\ two-ion spectra

873: displayed in Fig.~\ref{fig:RSiSFeSi} (bottom panel) remain very consistent in 

874: shape, while their flux follows the continuum-only spectrum flux increase.  

875: 

876: The flux variation displayed in Fig.~\ref{fig:allfour} is thus

877: dominated by the ``pseudo-continuum'' of \FeIII\ and \FeII\ lines,

878: especially the weak line blends, on top of which the strong \SiII\

879: and \SII\ lines add structure. The flux transferred by the well

880: separated lines formed in the line forming region is small compared to

881: the flux transferred at depth by iron lines, but is by  no means negligible

882: when using SNe~Ia for precision cosmology. It is therefore crucial to

883: have a good description of the full spectrum in order to perform

884: K-corrections. 

885: 

886: 

887: Figure \ref{fig:full_spect_tempevolution} shows very close

888: agreement between the full synthetic spectrum and the

889: \FeIII+\FeII+\SiII+\SII\ four-ion spectrum above 5000~\AA,

890: proving that in the W7 model, the other metallic lines have a secondary

891: importance in transferring flux compared to \FeIII\ and \FeII\ lines.

892: 

893: 

894: Faint \SNeIa\ are expected to be redder because of a larger

895: proportion of \FeII\ (as compared to \FeIII) and not because the

896: underlying na\"ive blackbody is colder. That is, accepting that the

897: primary variation for the peak luminosity is due to nickel mass

898: \citep{nugseq95}, dimmer supernovae are redder because the underlying

899: ionization stage is primarily \FeII. Figure~\ref{fig:bmvevolution}

900: clearly displays this brighter-bluer evolution for each epoch we

901: simulated. 

902: 

903: While the variation with epoch is beyond the scope of this

904: work, the variation with luminosity in part mimics the variation with

905: epoch to some extent. Moreover, the spectra at 10 days after explosion

906: are much bluer than the later ones, because the supernova is much

907: denser and hotter at this epoch, leading to a higher degree of

908: ionisation of the iron core.

909: 

910: This \FeIII-\FeII\ effect on flux transfer agrees with the 

911: analysis of \citet{kasen06b}, which explains the  second red bump in

912: \SNeIa\ light curves by the change in opacity due to the recombination

913: of \FeIII\ into \FeII: when the temperature decrease in the \SNeIa\

914: envelope is enough for \FeIII\ to recombine into \FeII\ more flux is

915: transferred from the blue into the red, causing the second red bump.   

916: 

917: \subsection{Silicon and Sulfur lines}

918: 

919: Fig.~\ref{fig:SiIIonly} (top panel) displays \FeII +\FeIII\ two-ion spectrum

920: and \SiII\ single ion spectrum. \SiII\ lines form in the line

921: forming region, contributing little to the global flux transfer but 

922: adding lots of structure to the spectrum. 

923: As previously mentioned, the $\sim 6100$~\AA\ trough as well as the

924: \SiII\ \SiIIred\ emission peak are dominated by a strong \SiII\ line. The

925: \SiII\ \SiIIblue\ line is also seen to dominate the second ``P-Cygni''

926: trough, but \SiII\ alone fails to account for the blue edge at

927: $\sim 5600$~\AA. We already ruled out \TiII, and the \FeII\ 5170~\AA\

928: peak is too blue to account for this edge.  

929: Fig.~\ref{fig:SiIIonly} (bottom panel) displays the

930: \SII\ single ion spectrum in addition to the previous ones,

931: showing that \SII\ lines can account for the bluer edge of the \RSi\

932: wavelength region. 

933: \clearpage

934: \begin{figure}[ht]

935:   \centering

936: %   \includegraphics[width=0.48\textwidth]{fig12a_redone}\\

937: %   \includegraphics[width=0.48\textwidth]{fig12b_redone}

938:   \includegraphics[width=0.48\textwidth]{f12a}\\

939:   \includegraphics[width=0.48\textwidth]{f12b}

940:   \caption{Single ion spectra for our faintest luminosity 20 days

941:     after explosion. The top panel shows spectra without \SII, the

942:     bottom panel includes \SII. They show that \SiII\ lines alone 

943:     can not account for the \RSi\ wavelength 

944:   region shape. \FeII\ and \FeIII\ contribute with their weak lines to

945: the ``pseudo-continuum''. The blue edge ($<$5600~\AA) of the \RSi\

946: region is dominated by \SII\ lines. }

947:   \label{fig:SiIIonly}

948: \end{figure}

949: \clearpage

950: 

951: The \SiII+\SII\ spectrum of Fig.~\ref{fig:SIIFeIIFeIII} (top panel) 

952: shows that the relative strengths of the peaks and depths due to

953: \SiII\ and \SII\ in the \RSi\ wavelength region are consistent with 

954: the spectrum of the full \nomw\ model. In Fig.~\ref{fig:SIIFeIIFeIII}

955: (bottom panel), 

956: where all the spectra are rescaled to have the same total

957: flux, if the \RSi\ line features are dominated by

958: \SiII\ and \SII, it is the \FeIII\ and \FeII\ lines that provide the

959: flux background.  

960: The \FeIII+\FeII+\SiII+\SII\ spectrum shows that between 5000~\AA\ 

961: and 8000~\AA\ the spectrum of the full \nomw\ model is well reproduced by 

962: these 4 ions alone.  This is an important result for the

963: understanding of the theoretical explanation of the correlation

964: between the line ratios and the absolute $B$ magnitude. Since we have

965: shown that the \RSi\ ratio, that is the flux in this wavelength range,

966: depends only on the iron,  silicon, and sulfur; to correctly model the

967: line ratio, one must only have a model that correctly accounts for

968: these ions. Clearly we are not there yet, but this result

969: significantly reduces the parameter space that must be studied to

970: obtain physical understanding. 

971: 

972: However, we must be cautious due to the fact that our models do not

973: reproduce the observed spectra. For example, W7 does not reproduce the

974: shape of the Si~II 6100\ang\ feature \citep{bbbh06} and thus we could

975: be missing at least one important species in our calculations. Turning

976: this around shows that obtaining the correct variation of \RSiS\ with

977: luminosity could be an important filter for obtaining physically

978: correct models. 

979: 

980: Similarly, the strong iron features around 4500~\AA\ and 5170~\AA\ are

981: unrealistically large compared to observed spectra. These strong

982: features are a finer probe of a model quality than the overall

983: shape of the spectrum: the thick blend of lines of the iron core is

984: much less sensitive to an abundance variation than the strong lines,

985: forming where the iron blanket has become optically thin. 

986: 

987: \clearpage

988: \begin{figure}[ht]

989:   \centering

990:   \includegraphics[width=0.48\textwidth]{f13a}\\

991:   \includegraphics[width=0.48\textwidth]{f13b}

992:   \caption{\SiII+\SII\ two-ion day 20 spectra (top panel, not rescaled,

993:     bottom panel, rescaled)

994:     } 

995:   \label{fig:SIIFeIIFeIII}

996: \end{figure}

997: \clearpage

998: 

999: 

1000: \section{\RSiS: observational support of the ``multi-layered spectral

1001:   formation''} 

1002: 

1003: \subsection{\RSiS\ predicted trend: a hotter \phx\ spectrum\label{rsisdef}}

1004: 

1005: We defined a line ratio \RSiS\ \citep{bongard06a} as 

1006: \begin{equation}

1007:   \mbox{\RSiSS}=\frac{\int_{5500}^{5700}F_{\lambda}d\lambda}{\int_{6450}^{6200}F_{\lambda}d\lambda}.    

1008:   \label{RSiSdef}

1009: \end{equation}

1010: We showed that it was better correlated to luminosity than \RSi, and

1011: deferred the motivation to the present paper. Here is

1012: where we address this question.

1013: 

1014: As shown in the previous sections, the \RSi\ wavelength region is

1015: dominated by \FeIII, \FeII, \SiII, and \SII\ lines. Moreover, for

1016: luminosities corresponding to ``normal'' \SNeIa, the bluer edge

1017: of this wavelength region is dominated by \SII, since the envelope is

1018: still too hot to allow the strong \FeII\ $5170$~\AA\ feature to dominate. As

1019: the luminosity decreases, the strength of the \FeII\ $5170$~\AA\ 

1020: feature increases, this behavior is illustrated in

1021: Fig.~\ref{fig:full_spect_tempevolution}. This effect will have 

1022: differing importance for the two different definitions of the line

1023: ratio, \RSi\ and \RSiS. \RSi\ will hardly be affected since the depth of

1024: the blue \SiII\ feature does not vary significantly; however, \RSiS\

1025: uses just this wavelength region and will thus be strongly impacted by

1026: the growth of the \FeII\ $\lambda5170$ line.

1027: 

1028: 

1029: The  ``multi-layered line formation''  picture shows that the depth of

1030: the  troughs are not the pertinent quantities to probe the temperature

1031: sequence, since the ``absorption'' features are formed by a complex

1032: blend of absorptions and emissions due to ions spanning a wide

1033: range of depth. The intermediate $\sim 5900$~\AA\ \RSi\  peak has

1034: the same problem, as it is a blend of  \FeIII, \FeII, and \SiII\ lines.   

1035: A direct corollary is that \RSi\ does not measure line strength

1036: evolution as is commonly believed. 

1037: 

1038: \clearpage

1039: \begin{figure}[ht]

1040: \centering

1041:   \includegraphics[width = 0.8\textwidth]{f14}

1042:   \caption{The full synthetic day 20 spectrum (solid line) is compared to the

1043:     day 20 four-ion spectra including \FeIII, \FeII, \SiII,

1044:     and \SII\ (dotted line).  The top panel shows $M_{bol}= -18.0$, the

1045:     bottom panel 

1046:     $M_{bol}= -19.0$. The pink vertical line indicates  \FeIII\ 

1047:     $5128$~\AA\ feature, while the green vertical region shows \FeII\

1048:     strong feature $5170-5291$~\AA\ that dominates at lower

1049:     luminosities. The blue vertical zones are zones where \RSiS\  is

1050:     defined.  

1051: }

1052:   \label{fig:full_spect_tempevolution}

1053: \end{figure}

1054: \clearpage

1055: 

1056: 

1057: 

1058: The motivation for the definition of \RSiS\ was to isolate contributions from

1059: \SiII\ and \SII\ lines forming in the same physical region between

1060: 10000~\kms\ and 15000~\kms. 

1061: 

1062: There are at least two options for the correlation of \RSiS\ with

1063: absolute blue magnitude:  the silicon and sulfur lines form in the

1064: same physical region and their strength is simply determined by the

1065: local conditions, or more likely \RSiS\ does not measure the region where

1066: silicon and sulfur lines form, but rather the conditions where the

1067: iron ``pseudo-continuum'' forms.

1068: 

1069: Keeping in mind that one should be careful when interpreting single

1070: ion spectra, Fig.~\ref{fig:RSiSFeSi} supports this latter explanation.

1071: Fig.~\ref{fig:RSiSFeSi} (top panel) shows that the \FeII\ and \FeIII\

1072: blend varies little in the 6100~\AA\ wavelength zone used to

1073: compute \RSiS. The consistency in the \SiII+\SII\ two-ion

1074: spectra shape displayed in Fig.~\ref{fig:RSiSFeSi} (bottom panel) suggests a

1075: small variation of the relative strength of the \SiII\ and \SII\

1076: features. We therefore favor the hypothesis that \RSiS\ traces the

1077: evolution in shape of the ``pseudo-continuum'' dominated by iron

1078: lines. As such, it should be better correlated with luminosity, since

1079: it probes the deeper layers of the supernova. On the other hand, it

1080: will be more sensitive to pollution by lines forming further out,

1081: especially strong \FeII\ lines. 

1082: 

1083: 

1084: \subsection{Synthetic vs observed spectra}

1085: 

1086: Up to now we have used the \nomw\ model as a reasonable physical explosion

1087: model of \SNeIa, but we did not try to

1088: match with observed spectra. We 

1089: used the knowledge of the abundance structures and the physical

1090: structure obtained by \phoenix\ to probe the line formation process,

1091: considering the \nomw\ model as a theoretical supernova of which we

1092: had complete knowledge. 

1093: 

1094: The \RSiS\ ratio correlates well with

1095: absolute blue magnitude \citep{bongard06a}. In Fig.~\ref{fig:RSiSphx}

1096: we display \RSi\ 

1097: and \RSiS\ for our day 20 and day 25  models as well as

1098: real \SNeIa\ data. The calculated \RSi\ appears to reproduce the

1099: observed data quite well for both days 20 and 25. It follows the

1100: ``normals'' quite well and may even extend to the regime of very

1101: low luminosity SNe~Ia such as SN 1991bg. At first glance, it would

1102: appear that the calculated \RSiS\ fails to match the observations;

1103: however, if one restricts one's attention to the ``normals'', 

1104: [Fig.~\ref{fig:RSiSphx} (bottom panel)], in fact the agreement between

1105: \RSiS\ (for day 25) and the data is better than that for \RSi. The

1106: change in slope we observe around $M_{B} = -19.1$ for day 25 comes

1107: from the strong \FeII\ 5170~\AA\ feature that dominates over \SII\ at

1108: lower luminosities. 

1109: 

1110: It should be noted that the W7 model is not expected to reproduce SNe~Ia

1111: light curves, and certainly cannot be expected to match the observed

1112: diversity from fast to slow decliners.

1113: Therefore, the time since explosion should be considered

1114: only as indicative, and related only with great care to real

1115: supernov\ae\ epochs.  

1116: 

1117: \clearpage

1118: \begin{figure}[ht]

1119: \centering

1120:  \includegraphics[width=0.7\textwidth]{f15a}\label{fig:RSiSphx_a}\\

1121: \includegraphics[width =0.7\textwidth]{f15b}\label{fig:RSiSphx_b}

1122:   \caption{\RSi\ vs $M_{B}$ (top panel) and \RSiS\ vs $M_{B}$ (bottom

1123:     panel) The real

1124:     data are displayed for reference, and are discussed thoroughly in

1125:     \citet{bongard06a}. The trends of real and synthetic spectra for 20 and

1126:     25 days after maximum are similar. The change in slope between day 20

1127:     and day 25 days after explosion is due to \FeII\ lines starting to

1128:     dominate the \RSi\ wavelength region.}

1129:   \label{fig:RSiSphx}

1130: \end{figure}

1131: \clearpage

1132: 

1133: 

1134: 

1135: 

1136: \section{Conclusion}

1137: 

1138: 

1139: We used a grid of LTE \phoenix\ synthetic spectra using the \nomw\

1140: model, to study the spectral formation in \SNeIa\ around the time of maximum

1141: light. Instead of focusing on fitting synthetic to observed spectra,

1142: we probed the detailed line formation in the synthetic spectrum to try

1143: to understand which ion was responsible for which feature, while

1144: taking detailed line blending into account.

1145: 

1146: We analyzed the synthetic spectra in detail to study the

1147: line formation in the \RSi\ wavelength region, showing that the

1148: \SiIIred\ peak and the 6100~\AA\ trough were dominated by \SiII\ lines

1149: forming over a blend of  \FeII/\FeIII\ lines. The intermediate

1150: 5800~\AA\ peak has been shown to be a complex blend of  \SiII\

1151: \SiIIred, \SiII\ \SiIIblue, \FeIII, and \FeII\ weak lines. The \TiII\

1152: contribution to the 5500~\AA\ trough has been ruled out. We 

1153: showed that the redder edge was dominated by the \SiIIblue\  \SiII\

1154: line whereas the bluer one was dominated by \SiII\ 5455~\AA\ line or

1155: \FeII\ 5170~\AA\ depending on luminosity. Based on our unraveling of

1156: the line formation in this wavelength region, we are able to

1157: illustrate our motivation for the definition of \RSiS,  

1158: isolating the \SiII\ and \SII\ contributions of the \RSi\ region.   

1159: 

1160: We described a multi-layer model where the 

1161: observed pseudo-continuum is formed throughout the entire supernova and

1162: thus strong features from multiple ionization stages can occur in the

1163: observed spectrum.

1164: We stress the importance of the numerous weak iron

1165: lines and show that their blends dominate the flux transfer. As a

1166: corollary, in the multi-layer model, the inner spectrum is not close to

1167: a Planck function but contains much more structure background. 

1168: We showed that the ionization stage of the iron core

1169: dominates the \SNIa\ colors, explaining the brighter-bluer relation at

1170: maximum light. 

1171: Even though a detailed study of time evolution is beyond

1172: the scope of this paper, we showed that our results where also

1173: qualitatively consistent in this regard, our earlier spectra being

1174: systematically much bluer than our later ones.

1175: 

1176: 

1177: \acknowledgements

1178:   We thank the anonymous referee for comments which significantly

1179:   improved the organization of this paper.

1180:   This work was supported in part by by NASA grants NAG5-3505 and

1181:   NAG5-12127,  NSF grants AST-0307323, AST-0506028, and

1182:   AST-0707704, and US DOE Grant DE-FG02-07ER41517.  S.~Bongard and

1183:   E.~Baron acknowledge support from the 

1184:   US Department of Energy Scientific Discovery through Advanced

1185:   Computing program under contract DE-FG02-06ER06-04.  

1186:   This research used resources of the National Energy Research

1187:   Scientific Computing Center (NERSC), which is supported by the

1188:   Office of Science of the U.S.  Department of Energy under Contract

1189:   No. DE-AC03-76SF00098; and the H\"ochstleistungs Rechenzentrum Nord

1190:   (HLRN).  We thank all these institutions for a generous allocation

1191:   of computer time.

1192: 

1193: 

1194: %\bibliography{apj-jour,refs,baron,sn1bc,sn1a,sn87a,snii,stars,rte,cosmology,gals,agn,crossrefs}

1195: 

1196: \begin{thebibliography}{22}

1197: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi

1198: 

1199: \bibitem[{Baron {et~al.}(2006)Baron, Bongard, Branch, \& Hauschildt}]{bbbh06}

1200: Baron, E., Bongard, S., Branch, D., \& Hauschildt, P. 2006, ApJ, 645, 480

1201: 

1202: \bibitem[{Baron {et~al.}(1999)Baron, Branch, Hauschildt, Filippenko, \&

1203:   Kirshner}]{b94i2}

1204: Baron, E., Branch, D., Hauschildt, P.~H., Filippenko, A.~V., \& Kirshner, R.~P.

1205:   1999, ApJ, 527, 739

1206: 

1207: \bibitem[{Baron \& Hauschildt(1998)}]{bhpar298}

1208: Baron, E. \& Hauschildt, P.~H. 1998, ApJ, 495, 370

1209: 

1210: \bibitem[{{Benetti} {et~al.}(2005){Benetti}, {Cappellaro}, {Mazzali},

1211:   {Turatto}, {Altavilla}, {Bufano}, {Elias-Rosa}, {Kotak}, {Pignata}, {Salvo},

1212:   \& {Stanishev}}]{benetti05}

1213: {Benetti}, S., {Cappellaro}, E., {Mazzali}, P.~A., {Turatto}, M., {Altavilla},

1214:   G., {Bufano}, F., {Elias-Rosa}, N., {Kotak}, R., {Pignata}, G., {Salvo}, M.,

1215:   \& {Stanishev}, V. 2005, ApJ, 623, 1011

1216: 

1217: \bibitem[{Bongard {et~al.}(2006)Bongard, Baron, Smadja, Branch, \&

1218:   Hauschildt}]{bongard06a}

1219: Bongard, S., Baron, E., Smadja, G., Branch, D., \& Hauschildt, P. 2006, ApJ,

1220:   647, 480

1221: 

1222: \bibitem[{{Branch} {et~al.}(2005){Branch}, {Baron}, {Hall}, {Melakayil}, \&

1223:   {Parrent}}]{branchcomp105}

1224: {Branch}, D., {Baron}, E., {Hall}, N., {Melakayil}, M., \& {Parrent}, J. 2005,

1225:   PASP, 117, 545

1226: 

1227: \bibitem[{{Branch} {et~al.}(2006){Branch}, {Dang}, {Hall}, {Ketchum},

1228:   {Melakayil}, {Parrent}, {Troxel}, {Casebeer}, {Jeffery}, \&

1229:   {Baron}}]{branchcomp206}

1230: {Branch}, D., {Dang}, L.~C., {Hall}, N., {Ketchum}, W., {Melakayil}, M.,

1231:   {Parrent}, J., {Troxel}, M.~A., {Casebeer}, D., {Jeffery}, D.~J., \& {Baron},

1232:   E. 2006, PASP, 118, 560

1233: 

1234: \bibitem[{Fisher(2000)}]{fisher00}

1235: Fisher, A. 2000, PhD thesis, Univ. of Oklahoma, unpublished

1236: 

1237: \bibitem[{Garnavich {et~al.}(2004)}]{garn99by04}

1238: Garnavich, P.~M. {et~al.} 2004, ApJ, 613, 1120

1239: 

1240: \bibitem[{{Hachinger} {et~al.}(2006){Hachinger}, {Mazzali}, \&

1241:   {Benetti}}]{hach06}

1242: {Hachinger}, S., {Mazzali}, P.~A., \& {Benetti}, S. 2006, MNRAS, 370, 299

1243: 

1244: \bibitem[{Hatano {et~al.}(1999)Hatano, Branch, Fisher, Deaton, \&

1245:   Baron}]{atlas99}

1246: Hatano, K., Branch, D., Fisher, A., Deaton, J., \& Baron, E. 1999, ApJS, 121,

1247:   233

1248: 

1249: \bibitem[{Hatano {et~al.}(2000)Hatano, Branch, Lentz, Baron, Filippenko, \&

1250:   Garnavich}]{hathighv00}

1251: Hatano, K., Branch, D., Lentz, E., Baron, E., Filippenko, A.~V., \& Garnavich,

1252:   P. 2000, ApJ, 543, L49

1253: 

1254: \bibitem[{Hauschildt \& Baron(1999)}]{hbjcam99}

1255: Hauschildt, P.~H. \& Baron, E. 1999, J. Comp. Applied Math., 109, 41

1256: 

1257: \bibitem[{Hauschildt {et~al.}(1997{\natexlab{a}})Hauschildt, Baron, \&

1258:   Allard}]{hbapara97}

1259: Hauschildt, P.~H., Baron, E., \& Allard, F. 1997{\natexlab{a}}, ApJ, 483, 390

1260: 

1261: \bibitem[{Hauschildt {et~al.}(1996)Hauschildt, Baron, Starrfield, \&

1262:   Allard}]{phhnovfe296}

1263: Hauschildt, P.~H., Baron, E., Starrfield, S., \& Allard, F. 1996, ApJ, 462, 386

1264: 

1265: \bibitem[{Hauschildt {et~al.}(1997{\natexlab{b}})Hauschildt, Schwarz, Baron,

1266:   Starrfield, Shore, \& Allard}]{phhnovetal97}

1267: Hauschildt, P.~H., Schwarz, G., Baron, E., Starrfield, S., Shore, S., \&

1268:   Allard, F. 1997{\natexlab{b}}, ApJ, 490, 803

1269: 

1270: \bibitem[{{James} {et~al.}(2006){James}, {Davis}, {Schmidt}, \&

1271:   {Kim}}]{james_pca06}

1272: {James}, J.~B., {Davis}, T.~M., {Schmidt}, B.~P., \& {Kim}, A.~G. 2006, \mnras,

1273:   370, 933

1274: 

1275: \bibitem[{{Kasen}(2006)}]{kasen06b}

1276: {Kasen}, D. 2006, ApJ, 649, 939

1277: 

1278: \bibitem[{{Mazzali} \& {Podsiadlowski}(2006)}]{mazzpod06}

1279: {Mazzali}, P.~A. \& {Podsiadlowski}, P. 2006, MNRAS, L32+

1280: 

1281: \bibitem[{Nomoto {et~al.}(1984)Nomoto, Thielemann, \& Yokoi}]{nomw7}

1282: Nomoto, K., Thielemann, F.-K., \& Yokoi, K. 1984, ApJ, 286, 644

1283: 

1284: \bibitem[{Nugent {et~al.}(1995)Nugent, Phillips, Baron, Branch, \&

1285:   Hauschildt}]{nugseq95}

1286: Nugent, P., Phillips, M., Baron, E., Branch, D., \& Hauschildt, P. 1995, ApJ,

1287:   455, L147

1288: 

1289: \bibitem[{Stehle {et~al.}(2005)Stehle, Mazzali, Benetti, \&

1290:   Hillebrandt}]{stehle02bo05}

1291: Stehle, M., Mazzali, P., Benetti, S., \& Hillebrandt, W. 2005, MNRAS, 360, 1231

1292: 

1293: \end{thebibliography}

1294: 

1295: 

1296: \end{document}

1297: 

1298: