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

2: 

3: \shorttitle{SPECTROPOLARIMETRY OF THE Ba~{$\!$\sc ii} D LINES}

4: \shortauthors{BELLUZZI, TRUJILLO BUENO \& LANDI DEGL'INNOCENTI}

5: 

6: 

7: \begin{document}

8: 

9: \title{THE MAGNETIC SENSITIVITY OF THE Ba~{$\!$\sc ii} D$_1$ AND D$_2$\\ 

10: LINES OF THE FRAUNHOFER SPECTRUM}

11: 

12: \author{L{\sc{uca}} B{\sc{elluzzi}}\altaffilmark{1,2}, 

13: J{\sc{avier}} T{\sc{rujillo}} B{\sc{ueno}}\altaffilmark{1,3} \\

14: {\sc{and}} \\ 

15: E{\sc{gidio}} L{\sc{andi}} D{\sc{egl}}'I{\sc{nnocenti}}\altaffilmark{2}}

16: 

17: \altaffiltext{1}{Instituto de Astrof\'\i{}sica de Canarias, V\'\i{}a L\'actea 

18: s/n, E-38205 La Laguna, Tenerife, Spain}

19: \altaffiltext{2}{Universit\`a degli Studi di Firenze, Dipartimento di 

20: Astronomia e Scienza dello Spazio, \\

21: Largo Enrico Fermi 2, I-50125 Firenze, Italy}

22: \altaffiltext{3}{Consejo Superior de Investigaciones Cient\'\i ficas, Spain}

23: 

24: 

25: \begin{abstract}

26: The physical interpretation of the spectral line polarization produced by the

27: joint action of the Hanle and Zeeman effects offers a unique opportunity to 

28: obtain empirical information about hidden aspects of solar and stellar 

29: magnetism. To this end, it is important to achieve a complete understanding 

30: of the sensitivity of the emergent spectral line polarization to the presence

31: of a magnetic field. Here we present a detailed 

32: theoretical investigation on the role of resonance scattering and 

33: magnetic fields on the polarization signals of the Ba~{$\!$\sc ii} D$_1$ 

34: and D$_2$ lines of the Fraunhofer spectrum, respectively at 4934~{\AA} 

35: and 4554~{\AA}.

36: We adopt a three-level model of Ba~{$\!$\sc ii}, and we take into account the 

37: hyperfine structure that is shown by the $^{135}$Ba 

38: and $^{137}$Ba isotopes. Despite of their relatively small abundance (18\%), 

39: the contribution coming from these two isotopes is indeed fundamental for the 

40: interpretation of the polarization signals observed in these lines.

41: We consider an optically thin slab model, through which we can investigate in a rigorous way the essential physical mechanisms involved (resonance 

42: polarization, Zeeman, Paschen-Back and Hanle effects), avoiding complications 

43: due to radiative transfer effects. We assume the slab to be illuminated from 

44: below by the photospheric solar continuum radiation field, and we investigate 

45: the radiation scattered at 90$^\circ$, both in the absence and in 

46: the presence of magnetic fields, deterministic and microturbulent.

47: We show in particular the existence of a differential magnetic

48: sensitivity of the three-peak $Q/I$ profile that is observed in the D$_{2}$ 

49: line in quiet regions close to the solar limb, which is of great interest for 

50: magnetic field diagnostics.

51: \end{abstract}

52: 

53: 

54: \keywords{atomic processes --- line: profiles --- polarization --- scattering 

55: --- Sun: magnetic fields}

56: 

57: 

58: \section{INTRODUCTION} 

59: \label{sect:introduction}

60: Probably, the most interesting aspect of spectropolarimetry is that

61: it allows us to diagnose magnetic fields in astrophysics. To this

62: end, it is crucial to achieve a complete physical understanding of

63: the magnetic sensitivity of the emergent spectral line radiation

64: given the fact that it can occur through a variety of rather

65: unfamiliar physical mechanisms, not only via the Zeeman effect (e.g.,

66: Landi Degl'Innocenti \& Landolfi 2004 (hereafter LL04); see also the 

67: reviews by Stenflo 2003 and Trujillo Bueno 2003). In this respect, the 

68: main aim of this paper is to help decipher the physical mechanisms that

69: control the magnetic sensitivity of the polarization of the D-lines

70: of Ba~{$\!$\sc ii}, with particular interest in developing a powerful

71: diagnostic tool for mapping the magnetic field of the lower solar

72: chromosphere.\\

73: \indent In the atmospheres of the Sun and of other stars there is a

74: fundamental mechanism producing polarization in spectral lines, which

75: has nothing to do with the familiar Zeeman effect. There, where light

76: escapes through the stellar ``surface'', the atomic system is

77: illuminated anisotropically. The radiative transitions

78: produce population imbalances and quantum coherences between pairs of

79: magnetic sublevels, even among those pertaining to different levels.

80: The mere presence of this so-called atomic level polarization

81: produces spectral line polarization, without the need of a magnetic field.

82: This is usually referred to as resonance line polarization.

83: The important point is that a magnetic field

84: can modify the atomic polarization of the upper and/or lower levels

85: of the spectral line under consideration and the ensuing polarization

86: of the emergent spectral line radiation. Interestingly, the possible

87: presence of crossings and repulsions among magnetic sublevels of 

88: fine-structure and/or hyperfine-structure multiplets can enhance

89: dramatically the magnetic sensitivity of the emergent spectral line

90: polarization. A remarkable example is the enhancement of the line-core 

91: scattering polarization of the D$_2$ line of Na~{$\!$\sc i} by a

92: vertical magnetic field, which is due to interferences between particular

93: hyperfine structure (HFS) magnetic sublevels of the $^2P_{3/2}$ upper level

94: \citep{jtb02}. It is of interest to note that this theoretical prediction

95: was observationally confirmed by \citet{Ste02} via filter

96: polarimetry of the solar atmosphere.\\

97: \indent In contrast with the case of sodium, which has one single isotope with 

98: nuclear spin I=3/2, barium has five (even) isotopes with I=0 (with an overall

99: abundance of $82.18\%$) and two (odd) isotopes with I=3/2 (with an

100: abundance of $17.82\%$).

101: Moreover, the HFS splitting of the odd isotopes of barium 

102: is about a factor five larger than for the case of sodium. 

103: Obviously, the emergent fractional linear polarization

104: (i.e., the $Q/I$ profile, where $I$ and $Q$ are two of the Stokes

105: parameters) has contributions from all the barium isotopes.

106: In fact, as pointed out by \citet{Ste97b}, the $Q/I$ pattern of the

107: Ba~{$\!$\sc ii} D$_2$ line that \citet{Ste97a} observed in very

108: quiet regions close to the solar limb shows a three-peak structure,

109: with a prominent central $Q/I$ peak due to the even isotopes (which

110: are devoid of HFS) and two less significant peaks in the red and blue

111: wings caused by the contributions from the odd isotopes (which have

112: HFS). Therefore, we think that for the D$_2$ line of Ba~{$\!$\sc ii} we

113: should also have enhancement of scattering polarization by a vertical

114: field, but only around such wing wavelengths because the required

115: interferences occur only between the magnetic sublevels of the 

116: $^2P_{3/2}$ upper level of the barium isotopes endowed of HFS.

117: Actually, the scientific motivation that led us to undertake the

118: theoretical investigation presented here was to develop a novel

119: plasma diagnostic tool based on the idea that such isotopes of barium

120: must have a different behavior in the presence of a magnetic field, 

121: with respect to those devoid of HFS.\\

122: \indent While the physical origin of the observed $Q/I$ profile of the 

123: Ba~{$\!$\sc ii} D$_2$ line seems to be clear, nobody has yet been able to

124: model the $Q/I$ profiles

125: of the Ba~{$\!$\sc ii} D$_1$ line that \citet{Ste98} observed

126: in two different regions close to the solar limb. Interestingly, while

127: the $Q/I$ profile shown in Fig.~1 of \citet{Ste98} might

128: perhaps be the result of the interaction of the D$_1$ line with the

129: continuum (see the bottom panel to the right-hand-side), the

130: symmetric $Q/I$ profile shown in the panel just above the previous

131: one is very similar to the enigmatic $Q/I$ profile of the Na~{$\!$\sc i} 

132: D$_1$ line \citep[see also Fig.~3 in][]{Ste00}. Although the main

133: scientific target of our paper is to understand the magnetic

134: sensitivity of the above-mentioned Ba~{$\!$\sc ii} D$_2$ line, we present

135: also some results of our $Q/I$ calculations for the D$_1$ line with

136: the aim of helping to clarify its physical origin.\\

137: \indent The outline of this paper is the following. The formulation of the

138: problem is presented in Section~\ref{sect:formulation}, where we establish 

139: our modeling assumptions, we briefly discuss the relevant equations, 

140: and we describe the atomic model, showing the behaviour of the magnetic 

141: sublevels of the odd barium isotopes in the presence of an increasing 

142: magnetic field. The magnetic sensitivity of the atomic

143: polarization of the lower and upper levels of the Ba~{$\!$\sc ii} D-lines

144: is discussed in Section~\ref{sect:polarization}, pointing out the 

145: similarities with the behavior of the Na~{$\!$\sc i} levels. 

146: Section~\ref{sect:emergent} focuses on the emergent spectral line

147: polarization in the absence and in the presence of a 

148: magnetic field, with emphasis on the D$_2$ line of Ba~{$\!$\sc ii}, but showing 

149: also some interesting results for the D$_1$ line. Finally, 

150: Section~\ref{sect:conclusions} summarizes

151: our main conclusions with an outlook to future research. The three

152: appendices give detailed information that may help the reader to

153: understand better the complexity of the problem we are investigating.

154: 

155: 

156: \section{FORMULATION OF THE PROBLEM} 

157: \label{sect:formulation}

158: We consider an optically thin slab of Ba~{$\!$\sc ii} ions illuminated 

159: from below by the photospheric solar continuum radiation field

160: (see Fig.~\ref{fig:geometry}). 

161: The slab is assumed to be located 1000~km above the $\tau_{5000}\!=\!1$ 

162: photospheric level, the approximate height at which, according to 

163: semi-empirical models of the solar atmosphere, the line-core optical depth of 

164: the D$_2$ line is unity along a line-of-sight specified by 

165: $\mu\!\equiv\!{\rm cos}{\theta}\!=\!0.1$, with ${\theta}$ the 

166: heliocentric angle.\\

167: \indent In the following subsections we summarize the basic

168: equations derived within the framework of the quantum theory of spectral

169: line polarization that will be used

170: in this work (\S~\ref{sect:equations}), we describe the atomic model

171: that we have adopted to describe the Ba~{$\!$\sc ii} ion (\S~\ref{sect:atom}),

172: and finally we show how the main properties of the continuum photospheric

173: radiation field incident on the slab have been calculated

174: (\S~\ref{sect:radiation}).

175: 

176: 

177: \subsection{{\it The Basic Equations}}

178: \label{sect:equations}

179: We describe the excitation state of the Ba~{$\!$\sc ii} levels using 

180: the density matrix formalism, a robust theoretical framework very suitable to 

181: treat the atomic polarization (population imbalances and quantum interferences,

182: or coherences, among the magnetic sublevels) that anisotropic pumping processes 

183: induce in an atomic system. 

184: Referring to one of the isotopes of barium (see \S~\ref{sect:atom}) that 

185: shows HFS, we indicate with $I$ its nuclear

186: spin quantum number. In the absence of magnetic fields, using Dirac's 

187: notation, the energy eigenvectors can be written in the form 

188: $|\alpha \, J \, I \, F \, f >$,

189: where $\alpha$ represents a set of inner quantum numbers

190: (specifying the electronic configuration and, if the atomic system is described 

191: in the L-S coupling scheme, the total electronic orbital and spin angular 

192: momenta), $J$ is the total electronic angular momentum quantum number, 

193: $F$ is the total angular momentum quantum number (electronic plus nuclear:

194: {\boldmath $F$=$J$+$I$}), and $f$ is its projection along the 

195: quantization axis.

196: %%

197: \begin{figure}[!t]

198: \begin{center} 

199: \includegraphics[width=0.5\textwidth]{f1.eps}

200: \caption{\footnotesize{Geometry of the problem under investigation.}}

201: \label{fig:geometry}

202: \end{center}

203: \end{figure}

204: %%

205: In principle, to have a suitable description of atomic polarization, one has 

206: to take into account all the coherences of the form

207: %

208: \begin{equation}

209: <\alpha\,J\,I\,F\,f\,|\,\hat{\rho}\,|\,\alpha\,J^{\prime}\,I\,

210: F^{\prime}\,f^{\prime}> \;\;,

211: \end{equation}

212: %

213: where $\hat{\rho}$ is the density matrix operator.

214: The approximation that is used in this paper 

215: consists in restricting the description of the atomic system to the 

216: $J$-diagonal density matrix elements

217: %

218: \begin{equation}

219: <\alpha\,J\,I\,F\,f\,|\,\hat{\rho}\,|\,\alpha\,J\,I\,F^{\prime}\,f^{\prime}> \;\;,

220: \end{equation}

221: %

222: or, in other words, in neglecting coherences between different $J$-levels.

223: The resulting model-atom is referred to as the ``multi-level atom with 

224: hyperfine structure'' according to LL04.

225: The approximation appears to be fully justified for the investigation of the 

226: Ba~{$\!$\sc ii} D-lines given the large frequency separation between the levels 

227: $6p\;^{2}P_{1/2}$ and $6p\;^{2}P_{3/2}$ (see \S~\ref{sect:atom}), and given the 

228: relatively low abundance of barium in the solar 

229: atmosphere\footnote{Note that this 

230: approximation is not justified for the D-lines of Na~{$\!$\sc i} and for the 

231: H and K lines of Ca~{$\!$\sc ii}, given the larger abundance of these ions.}.\\

232: \indent In the presence of a magnetic field, according to the general approach 

233: of the Paschen-Back effect theory, the energy eigenvalues and eigenvectors have 

234: to be found by diagonalization of the total

235: Hamiltonian (unperturbed atomic Hamiltonian plus magnetic Hamiltonian) on

236: each $J$-level subspace.

237: Taking the quantization axis along the magnetic field direction, it can be 

238: demonstrated that

239: the total Hamiltonian commutes with the projection along the quantization

240: axis of the total angular momentum operator ($f$ is a good quantum

241: number), while, in general, it does not commute with the total angular

242: momentum operator. The eigenvectors of the total Hamiltonian can be expressed 

243: in the form (e.g. LL04)

244: %

245: \begin{equation}

246: |\alpha\,J\,I\,i\,f>=\sum_{F}C_{F}^{\,i}\,(\alpha\,J\,I,\,f\,)\,|\alpha\,

247: J\,I\,F\,f> \;\; ,

248: \label{eq:eigenvectors}

249: \end{equation}

250: %

251: where the index $i$ labels the energy eigenstates belonging to the

252: subspace corresponding to assigned values of the quantum numbers $\alpha$, $J$, 

253: $I$ and $f$, and where the coefficients $C_{F}^{\,i}$ can be chosen to be real.

254: In the energy eigenvectors representation the atomic system

255: will therefore be described by means of the matrix elements

256: %

257: \begin{equation}

258: <\alpha\,J\,I\,i\,f\,|\,\hat{\rho}\,|\alpha\,J\,I\,i^{\prime}\,f^{\prime}>

259: \equiv\rho_{\alpha\,J\,I}(i\,f,\,i^{\prime}\,f^{\prime}) \;\;.

260: \label{eq:rho}

261: \end{equation}

262: %

263: If the magnetic field is so weak that the magnetic energy is much smaller than 

264: the energy intervals between the HFS $F$-levels, we are in the so-called 

265: {\it Zeeman effect regime} (of HFS), where the energy eigenvectors are still 

266: of the form $|\alpha\, J\, I\, F\, f> $ ($C_{F}^{\,i} \simeq \delta_{Fi}$), 

267: and the splitting between the HFS magnetic sublevels is 

268: linear with the magnetic field strength.

269: For stronger magnetic fields it is necessary to apply the Paschen-Back effect 

270: theory, and one enters the so-called {\it incomplete Paschen-Back effect 

271: regime}.

272: In this regime the energy eigenvectors have the general form of

273: equation~(\ref{eq:eigenvectors}) (the magnetic field produces a $F$-mixing of 

274: the various HFS levels originating from a particular $J$-level) and, 

275: as we will show in detail for the case of Ba~{$\!$\sc ii} in 

276: \S~{\ref{sect:atom}},

277: the splitting among the various HFS magnetic sublevels is no longer linear with 

278: the magnetic field. Several crossings 

279: among HFS magnetic sublevels with different $f$ quantum number take place in 

280: this regime, as well as a repulsion among the magnetic sublevels with the same 

281: $f$ quantum number. 

282: This behaviour of the magnetic sublevels has important consequences on the 

283: atomic polarization, as pointed out by \citet{Bom80} and described in detail 

284: in LL04, and produce interesting effects, sometimes referred to as 

285: {\it level crossing effect} and {\it anti-level-crossing effect},

286: on the polarization signals produced by resonance scattering 

287: \citep[e.g.,][]{jtb02}.

288: In \S~\ref{sect:D2-vertical-field} we will show their effect on the

289: linear polarization of the Ba~{$\!$\sc ii} D$_{2}$ line.

290: If the magnetic field strength is further increased the 

291: so-called {\it complete Paschen-Back effect regime} is reached. 

292: In this regime the energy 

293: eigenvectors are of the form $|\alpha\, J\, I\, M_{J}\, M_{I}>$, and the 

294: splitting among the HFS magnetic sublevels is again linear with the magnetic 

295: field strength: the atom behaves in this regime as if it were devoid of HFS.

296: Going from the Zeeman effect regime to the complete Paschen-Back effect regime,

297: the magnetic field produces therefore an energy eigenvectors basis 

298: transformation.\\

299: \indent In the following we will work in the spherical statistical tensor

300: representation.

301: The conversion of the density matrix elements of equation~({\ref{eq:rho}})

302: into this representation is given by the relation (cf.~LL04)

303: %

304: \begin{eqnarray}

305: \rho_{\alpha\,J\,I}(i\,f,\,i^{\prime}\,f^{\prime}\,) = \sum_{FF^{\prime}}

306: C_{F}^{\,i}(\alpha\,J\,I,\,f)\,C_{F^{\prime}}^{\,i^{\prime}}

307: (\alpha\,J\,I,\,f^{\prime})\,

308: \rho_{\alpha\,J\,I}(F\,f,\,F^{\prime}\,f^{\prime}\,)=

309: \;\;\;\;\;\;\;\;\;\;\;\;\; & & \nonumber \\

310: =\sum_{FF^{\prime}}C_{F}^{\,i}(\alpha\,J\,I,\,f)\,

311: C_{F^{\prime}}^{\,i^{\prime}}(\alpha\,J\,I,\,f^{\prime})

312: \sum_{KQ}(-1)^{F-f}\sqrt{2K+1}

313: \Bigg( \begin{array}{ccc}

314: \!F & \!\!F^{\prime} & \!\!K \\

315: \!f & \!\!-f^{\prime} & \!\!-Q

316: \end{array} \!\!\Bigg)

317: \,^{\alpha\,J\,I}\rho^{K}_{Q}(F,F^{\prime}) \;\; . & &

318: \end{eqnarray}

319: %

320: The Statistical Equilibrium Equations (SEEs) and the radiative transfer 

321: coefficients for a multi-level atom with HFS, in the spherical statistical

322: tensor representation, written taking the quantization axis directed along

323: the magnetic field, can be found in \S~7.9 of LL04.  

324: Here we write only the expression for the emission coefficient in the 

325: transition between the upper level ($\alpha_u, J_u$) and the lower level 

326: ($\alpha_\ell, J_\ell$)

327: %

328: \begin{eqnarray}

329:  & & \varepsilon_{j}(\nu,\mathbf{\Omega})=\frac{2h\nu^3}{c^2}

330:  \frac{h\nu}{4\pi}\mathcal{N} (2J_{u}+1)

331:  B(\alpha_{u}J_{u}I \to \alpha_{\ell} J_{\ell}I) \nonumber \\

332:  & & \times \sum_{KQK_{u}Q_{u}} \sqrt{3(2K+1)(2K_{u}+1)}

333:  \sum_{i_{u} F_{u} F_{u}^{\prime} F_{u}^{\prime\prime} i_{\ell}F_{\ell}

334:  F_{\ell}^{\prime}} \,\, \sum_{f_{u}f_{u}^{\prime}f_{\ell}qq^{\prime}}

335:  (-1)^{1+F_{u}^{\prime}-f_{u}+q^{\prime}} \nonumber \\

336:  & & \times C_{F_{\ell}}^{\,i_{\ell}}(\alpha_{\ell}J_{\ell}I,f_{\ell})\,

337:  C_{F_{\ell}^{\,\prime}}^{\,i_{\ell}}(\alpha_{\ell}J_{\ell}I,f_{\ell})\,

338:  C_{F_{u}}^{\,i_{u}}(\alpha_{u}J_{u}I,f_{u})\,

339:  C_{F_{u}^{\,\prime\prime}}^{\,i_{u}}(\alpha_{u}J_{u}I,f_{u}) \nonumber \\

340:  & & \times \sqrt{(2F_{\ell}+1)(2F_{\ell}^{\prime}+1)(2F_{u}+1)

341:  (2F_{u}^{\prime}+1)} \nonumber \\

342:  & & \times \Bigg( \begin{array}{ccc}

343:  F_{u} & F_{\ell} & 1 \\

344:  -f_{u} & f_{\ell} & -q

345:  \end{array} \Bigg)

346:  \Bigg( \begin{array}{ccc}

347:  F_{u}^{\,\prime} & F_{\ell}^{\,\prime} & 1 \\

348:  -f_{u}^{\,\prime} & f_{\ell} & -q^{\,\prime}

349:  \end{array} \Bigg)

350:  \Bigg( \begin{array}{ccc}

351:  1 & 1 & K \\

352:  q & -q^{\,\prime} & -Q

353:  \end{array} \Bigg) \nonumber \\

354:  & & \times \Bigg( \begin{array}{ccc}

355:  F_{u}^{\,\prime} & F_{u}^{\,\prime\prime} & K_{u} \\

356:  f_{u}^{\,\prime} & -f_{u} & -Q_{u}

357:  \end{array} \Bigg)

358:  \Bigg\{ \begin{array}{ccc}

359:  J_{u} & J_{\ell} & 1 \\

360:  F_{\ell} & F_{u} & I

361:  \end{array}\Bigg\}

362:  \Bigg\{\begin{array}{ccc}

363:  J_{u} & J_{\ell} & 1 \\

364:  F_{\ell}^{\,\prime} & F_{u}^{\,\prime} & I

365:  \end{array} \Bigg\} \nonumber \\

366:  & & \times {\rm{Re}} \Big[{\mathcal{T}}_{Q}^{K}(j,\mathbf{\Omega})\,

367:  ^{\alpha_{u}J_{u}I}\!\rho_{Q_{u}}^{K_{u}}(F_{u}^{\prime},F_{u}^{\prime\prime})

368:  \, \Phi(\nu_{\alpha_u J_u I i_u f_u, \alpha_{\ell} J_{\ell} I i_{\ell}

369:  f_{\ell}}-\nu) \Big] \; ,

370: \label{eq:epsilon}

371: \end{eqnarray}

372: %

373: where $j\!=\!0,1,2,3$, standing respectively for the Stokes parameters 

374: $I,Q,U$ and $V$, $\mathcal{N}$ is the number density of atoms, 

375: $B(\alpha_{u}J_{u}I \to \alpha_{\ell} J_{\ell}I)$ is the Einstein coefficient 

376: for stimulated emission, ${\mathcal{T}}_{Q}^{K}(j,\mathbf{\Omega})$ is a 

377: geometrical tensor (cf.~LL04), and $\Phi$ the profile of the line.\\

378: \indent It is important to note that the previous equations

379: are valid under the {\it{flat-spectrum approximation}}. 

380: For a multi-level atom with

381: HFS this approximation requires that the incident radiation

382: field should be flat (i.e.~independent of frequency) across a spectral interval

383: $\Delta\nu$ larger than the frequency intervals among the HFS levels (possibly 

384: split by the magnetic field), and larger than the inverse 

385: lifetimes of the same levels. 

386: This is a good approximation for the D$_1$ and D$_2$ lines of Ba~{$\!$\sc ii} 

387: if we restrict to magnetic fields smaller or of the order of 1~kG.

388: 

389: 

390: \subsection{{\it The Atomic Model}} 

391: \label{sect:atom}

392: %

393: \begin{table}[t!]

394: \label{tab:isotopes}

395: \begin{center}

396: {\footnotesize

397: \begin{tabular}{ccccccccc}

398: \tableline

399: \noalign{\smallskip}

400: Isotope & Abund. & $I$ & \multicolumn{2}{c}{Isotope Shifts (MHz)$^{1}$} & \multicolumn{4}{c}{HFS Constants (MHz)$^{2}$} \\

401: \noalign{\smallskip}

402: \cline{4-5}

403: \cline{6-9}

404: \noalign{\smallskip}

405:  & (\%)$^{\rm{a}}$ & & D$_1$ & D$_2$ & $^2S_{1/2}$ & $^2P_{1/2}$ & 

406: \multicolumn{2}{c}{$^2P_{3/2}$} \\

407: \noalign{\smallskip}

408:  & & & & & $\mathcal{A}$ & $\mathcal{A}$ & $\mathcal{A}$ & $\mathcal{B}$ \\

409: \tableline

410: \noalign{\smallskip}

411: $^{130}$Ba & 0.106 & 0 & 355.3$^{\rm{b}}$ & 372.3$^{\rm{b}}$ & & & \\

412: $^{132}$Ba & 0.101 & 0 & 278.9$^{\rm{b}}$ & 294.9$^{\rm{b}}$ & & & \\

413: $^{134}$Ba & 2.417 & 0 & 222.6$^{\rm{c}}$ & 234.6$^{\rm{c}}$ & & & \\

414: $^{135}$Ba & 6.592 & 3/2 & 348.6$^{\rm{b}}$ & 360.7$^{\rm{b}}$ & 3591.67$^{\rm{d}}$ & 664.6$^{\rm{e}}$ & 113.0$^{\rm{e}}$ & 59.0$^{\rm{e}}$ \\

415: $^{136}$Ba & 7.854 & 0 & 179.4$^{\rm{b}}$ & 186.9$^{\rm{b}}$ & & & \\

416: $^{137}$Ba & 11.232 & 3/2 & 271.1$^{\rm{b}}$ & 279.0$^{\rm{b}}$ & 4018.87$^{\rm{d}}$ & 743.7$^{\rm{e}}$ & 127.2$^{\rm{e}}$ & 92.5$^{\rm{e}}$ \\

417: $^{138}$Ba & 71.698 & 0 & \multicolumn{2}{c}{reference isot.} & & & \\

418: \noalign{\smallskip}

419: \tableline

420: \end{tabular}

421: }

422: \end{center}

423: \footnotesize{$^{1}$A positive I.S. means that the line is shifted to higher

424: frequencies with respect to the reference isotope.}\\

425: \footnotesize{$^{2}$The HFS constant $\mathcal{B}$ is defined according

426: to the convention of the American literature.}\\

427: \footnotesize{$^{\rm{a}}$NIST on-line database; $^{\rm{b}}$\citet{Wen84};

428: $^{\rm{c}}$\citet{Wen88}; $^{\rm{d}}$\citet{Bec81}; $^{\rm{e}}$\citet{Vil93}.}

429: \caption{\footnotesize{Isotopes considered in this work.}}

430: \end{table}

431: %

432: We adopt a three-level model of Ba~{$\!$\sc ii} consisting in the ground level

433: ($6s\;^{2}S_{1/2}$), the upper level of the D$_{1}$ line ($6p\;^{2}P_{1/2}$) 

434: and the upper level of the D$_{2}$ line ($6p\;^{2}P_{3/2}$).

435: There are seven stable isotopes of barium, whose mass numbers and relative 

436: abundances are listed in Table~1. 

437: In this work we take into account the contributions coming from all 

438: the seven isotopes.\\

439: \indent The mass and volume differences between the nuclei of the various 

440: isotopes involve small but appreciable differences on the energies of the fine 

441: structure levels of different isotopes (isotopic effect).

442: We use the values of the isotopic shifts in the D$_{1}$ and D$_{2}$ 

443: lines listed in Table~1 to 

444: correct the energies of the $^{2}P_{1/2}$ and $^{2}P_{3/2}$ levels of the 

445: variuos isotopes. For the reference isotope

446: (138) we use the energy values given by \citet{Moore}.\\

447: %%

448: \begin{figure}[!t]

449: \begin{center} 

450: \includegraphics[width=0.9\textwidth]{f2.eps}

451: \caption{{\footnotesize Energies, corrected for the isotopic effect, of the 

452: fine-structure and HFS levels of the isotope $^{137}$Ba~{\sc ii}. 

453: The right panels show the energies of 

454: the HFS magnetic sublevels as functions of the magnetic field strength 

455: (in each panel the zero of the energy scale is chosen at the energy of the 

456: corresponding fine structure $J$-level).

457: \label{fig:paschen}}}

458: \end{center}

459: \end{figure}

460: %%

461: \indent Isotopes with even mass number have nuclear spin $I\!=\!0$, while 

462: those with odd mass number (135 and 137) have nuclear spin $I\!=\!3/2$. 

463: The odd isotopes show therefore HFS due to nuclear spin.

464: Introducing the total angular momentum, characterized by the quantum number

465: $F$, we observe (see Fig.~\ref{fig:paschen}) that the levels

466: $^{2}S_{1/2}$ and $^{2}P_{1/2}$ split into two HFS levels ($F\!=\!1,2$), 

467: while the level $^{2}P_{3/2}$ splits into four HFS 

468: levels ($F\!=\!0,1,2,3$).

469: It is possible to demonstrate that the HFS Hamiltonian can be 

470: expressed as an infinite series of electric and magnetic multipoles 

471: \citep[e.g.][]{Kop58}.

472: To calculate the energies of the various HFS levels we 

473: consider the first two terms of the series (magnetic dipole and electric 

474: quadrupole terms), and we use the HFS constants listed in 

475: Table~1.

476: This investigation will clearly show the importance of the HFS

477: effects for a correct modeling of the 

478: polarization produced by scattering processes in a stellar atmosphere.\\

479: \indent The Land\'e factors have been calculated theoretically, assuming L-S 

480: coupling for the Ba~{$\!$\sc ii} ion. The values obtained differ by less than 

481: 1.3\% from the experimental ones \citep{Moore}, reported in 

482: Figure~\ref{fig:paschen}. 

483: This can be considered as a proof 

484: that the L-S coupling is quite a good approximation for the

485: Ba~{$\!$\sc ii} ion.\\

486: \indent The energies of the HFS magnetic sublevels of the isotope 

487: 137, as functions of the magnetic field strength, are shown in 

488: Figure~\ref{fig:paschen}.

489: We can see that in the range between 0 and 1000~G the splitting 

490: of the various magnetic sublevels originating from the ground level 

491: $^{2}S_{1/2}$ is linear with the magnetic field strength (Zeeman effect 

492: regime). A similar behaviour is shown by the magnetic sublevels originating 

493: from the level $^{2}P_{1/2}$ for magnetic fields smaller than about 600~G. 

494: For stronger magnetic fields the linearity of the splitting appears to be 

495: slowly lost, which indicates that the incomplete Paschen-Back effect regime is 

496: reached.

497: The splitting observed among the magnetic sublevels originating from the level 

498: $^{2}P_{3/2}$ shows instead that a complete transition from the Zeeman effect 

499: regime to the complete Paschen-Back effect regime takes place for magnetic 

500: fields ranging from 0 to 1000~G.

501: As described in \S~\ref{sect:equations}, 

502: in the intermediate incomplete Paschen-Back effect regime

503: several level crossings among HFS magnetic sublevels can be observed, 

504: as well as a repulsion among the sublevels with the same 

505: $f$ quantum number. 

506: 

507: 

508: \subsection{{\it The Incident Radiation Field}}

509: \label{sect:radiation}

510: As already stated in \S~\ref{sect:formulation}, we consider an optically 

511: thin plane-parallel slab, composed of Ba~{$\!$\sc ii} ions, located at 

512: approximately 1000~km above the $\tau_{5000}\!=\!1$ photospheric level, 

513: and we assume that the slab is illuminated from below 

514: (hence, anisotropically) by the photospheric continuum radiation. 

515: Under these hypothesis the atomic polarization can be calculated solving 

516: directly the SEEs for the given continuum radiation field coming from 

517: the photosphere.\\

518: %

519: \begin{table}[t!]

520: \label{tab:transitions}

521: \smallskip

522: \begin{center}

523: {\footnotesize

524: \begin{tabular}{ccccc}

525: \tableline

526: \noalign{\smallskip}

527: Line & $\lambda$~(\AA) & $A\,(s^{-1})$ & $\bar{n}_{\nu}$ & $w_{\nu}$ \\

528: \noalign{\smallskip}

529: \tableline

530: \noalign{\smallskip}

531:  D$_1$ & 4934.09 & 0.955$\times 10^8$ & 0.323$\times 10^{-2}$ & 0.159 \\

532:  D$_2$ & 4554.03 & 1.17$\times 10^8$ & 0.225$\times 10^{-2}$ & 0.176 \\

533: \noalign{\smallskip}

534: \tableline

535: \end{tabular}

536: \caption{\footnotesize{Wavelength (in air), and Einstein coefficient of the 

537: transitions considered; mean number of photons and anisotropy factor of the 

538: photospheric continuum at the wavelength of the transitions, 1000~km above 

539: the $\tau_{5000}=1$ level.}}

540: }

541: \end{center}

542: \end{table}

543: %

544: \indent Let us take a reference system with the {\it{z}} axis directed along the

545: local vertical, and let us describe the continuum radiation field incident on 

546: the slab by means of the tensor 

547: %

548: \begin{equation}

549: J^{K}_{Q}(\nu)=\int\frac{{\rm d}\Omega}{4\pi}\sum_{i=0}^{3}

550: {\mathcal{T}}_{Q}^{K}(i,\mathbf{\Omega}) S_i(\nu,\mathbf{\Omega}) \:\: ,

551: \end{equation}

552: %

553: where $S_{i}=I,Q,U,V$.

554: Assuming that the incident radiation field is unpolarized and has cylindrical

555: symmetry around the local vertical, it is easy to verify that the only non

556: vanishing components are

557: %

558: \begin{equation}

559: J^{0}_{0}(\nu)\! \! =\! \! \oint\frac{{\rm{d}}\Omega}{4\pi}I(\nu,\mu) 

560: \;\;\;\;\; {\rm{and}} \;\;\;\;\; 

561: J^{2}_{0}(\nu)\! \! =\! \! \oint\frac{{\rm{d}}\Omega}{4\pi}\Big(\frac{1}

562: {2\sqrt{2}} (3\mu^2-1)I(\nu,\mu) \Big) \;\; .  

563: \label{eq:JKQ}

564: \end{equation}

565: %

566: Note that $J^{0}_{0}$ is just the mean intensity of the incident radiation

567: (averaged over all directions), while $J^{2}_{0}$ gives a measure of 

568: the anisotropy of the radiation field\footnote{In particular $J_{0}^{2}$ 

569: quantifies the unbalance between vertical and horizontal illuminations.}.

570: Instead of $J^{0}_{0}$ and $J^{2}_{0}$, we can use the quantities 

571: $\bar{n}(\nu)$, the mean number of photons, and $w_{\nu}$, the so called 

572: anisotropy factor. The new quantities are related to the previous radiation 

573: field tensor components through the relations

574: %

575: \begin{equation}

576: \bar{n}(\nu)\! =\! \frac{c^{2}}{2h\nu^{3}}J^{0}_{0}(\nu) 

577: \;\;\;\;\; {\rm and} \;\;\;\;\;

578: w_{\nu}\! =\! \sqrt{2}\,\frac{J^{2}_{0}(\nu)}{J^{0}_{0}(\nu)} \;\; .

579: \end{equation}

580: %

581: To calculate the values of $\bar{n}(\nu)$ and $w_{\nu}$ of the photospheric 

582: continuum at the height of 1000~km above the visible solar ``surface'', at the 

583: frequencies of the D$_{1}$ and D$_{2}$ lines of Ba~{$\!$\sc ii}, we follow 

584: \S~12.3 of LL04.

585: The values of the specific intensity of the radiation coming from the solar 

586: disk center and of the limb-darkening coefficients are taken from 

587: \citet{Allen}. The values obtained for $\bar{n}_{\nu}$ and $w_{\nu}$ are listed 

588: in Table~2.

589: At this point the SEEs can be solved numerically.

590: Their expression becomes simpler if we rewrite them in the reference system

591: with the quantization axis directed along the local vertical direction, as

592: in this case only two components of the radiation field tensor are non zero

593: ($J_{0}^{0}$ and $J_{0}^{2}$).

594: It can be demonstrated that all the radiative rates are invariant under a

595: rotation of the reference system so that only 

596: the magnetic kernel has to be modified with respect to the expression 

597: given in equation~(7.66) of LL04.

598: 

599: 

600: \section{THE POLARIZATION OF THE ATOMIC LEVELS} 

601: \label{sect:polarization}

602: We solve numerically the SEEs 

603: (which implies, for each isotope with HFS of our 

604: model atom, the solution of a linear system of 384 equations in the unknowns 

605: $\rho^K_Q(F,F^{\,\prime}\,)$) for magnetic field strengths between 0 and 

606: 1000~G, and for various inclinations of the magnetic field with respect to the 

607: local vertical.

608: We recall that the $\rho^0_0(F,F\,)$ elements quantify the populations of the 

609: various $F$-levels, the $\rho^2_Q$ elements ({\it{alignment}} components) 

610: contribute to the linear polarization of the scattered radiation, while 

611: the $\rho^{1}_{Q}$ elements ({\it{orientation}} components) contribute to the 

612: circular polarization of the scattered radiation.

613: As shown in LL04, an anisotropic, unpolarized, 

614: flat spectrum radiation field generally induces only 

615: alignment in the atomic system, while orientation can be

616: originated by the so called alignment-to-orientation conversion mechanism.

617: Note that all the levels of the isotopes 135 and 137, because of the HFS, 

618: can carry alignment while, for all the other isotopes, 

619: only the level $^2P_{3/2}$, the upper level of the D$_{2}$ line, can carry it.\\

620: \indent Let us consider isotope 137.  In complete analogy with 

621: the case of Na~{$\!$\sc i}, having a single isotope with $I\!=\!3/2$ 

622: \citep[see Trujillo Bueno et al.~2002 and][and the discussion therein]{Cas02},

623: only the level $^2P_{3/2}$ is polarized directly 

624: via the anisotropic illumination. The ground level becomes polarized because of 

625: a transfer of polarization via spontaneous emission in the D$_{2}$ line, while 

626: the level $^2P_{1/2}$ becomes polarized via radiative absorption

627: ({\it{repopulation pumping}}) in the D$_{1}$ line. This explains the fact that

628: the upper and lower levels of the D$_{1}$ line are equally sensitive to the

629: magnetic field strength, independently of its inclination 

630: (see Fig.~\ref{fig:sigma20}).\\

631: %%

632: \begin{figure}[!t] 

633: \begin{center}

634: \includegraphics[angle=270,width=0.9\textwidth]{f3.eps}

635: \caption{{\footnotesize Fractional atomic alignment, 

636: $\sigma^{2}_{0}=\rho^{2}_{0}(F,F) / \rho^{0}_{0}(F,F)$, of the various 

637: HFS levels of the isotope 137, calculated in the local 

638: vertical reference system, as functions of the magnetic field vector, for 

639: various inclination angles. Top row $\theta_{B}\!=\!0^{\circ}$, middle rows 

640: $\theta_{B}\!=\!30^{\circ}$ and $\theta_{B}\!=\!60^{\circ}$, bottom row 

641: $\theta_{B}\!=\!90^{\circ}$. The angle $\theta_B$ is defined in 

642: Figure~\ref{fig:geometry}.}}

643: \label{fig:sigma20}

644: \end{center}

645: \end{figure}

646: %%

647: \indent It is well known that a magnetic field is able to modify the atomic 

648: polarization, and therefore the polarization of the scattered radiation

649: (Hanle effect\footnote{\label{note:Hanle}In this work by Hanle 

650: effect we mean any modification of the atomic polarization which is due to 

651: the action of a magnetic field. Note, however, that in the literature it is 

652: often meant by Hanle effect only the relaxation of coherences (defined in the 

653: magnetic field reference system) having $Q\ne0$.

654: Within this second meaning, it is often stated that there is no Hanle effect 

655: in the presence of a vertical magnetic field. Note that according to our 

656: definition this statement is true only if we are dealing with an 

657: isolated level (i.e.~if we neglect the quantum interferences among the magnetic 

658: sublevels originating from different hyperfine (or fine) structure levels).

659: We prefer to adopt the former more general definition because, as we are 

660: dealing with a quite complex atomic system with HFS, and as we are 

661: investigating the role of magnetic fields with  

662: complex configurations (random-azimuth and microturbulent), it becomes quite 

663: difficult to understand which effects due to the magnetic field can be 

664: considered as `Hanle effect' according to the latter definition.}). 

665: The behaviour of the various spherical statistical tensors, 

666: written in the local vertical reference system, as functions of the magnetic 

667: field inclination and strength, is qualitatively equal to the case of 

668: Na~{$\!$\sc i}, investigated by \citet{jtb02}.

669: We point out that the first decrease of the atomic polarization of 

670: the D$_1$ levels, which takes place only for inclined fields 

671: (see Fig.~\ref{fig:sigma20}), is due to the so called lower-level Hanle 

672: effect\footnote{This definition has been formulated within the latter 

673: definition of the Hanle effect given in the footnote~\ref{note:Hanle}.}.

674: As can be demonstrated (see LL04 for details), the spherical 

675: statistical tensors $\rho^K_Q(F,F)$ (that describe population imbalances and 

676: quantum interferences among magnetic sublevels originating from the same 

677: $F$-level, and that mainly affect the polarization of the scattered radiation 

678: at the line center) are significantly modified by a magnetic field when the 

679: Zeeman splitting is of the same order of magnitude as the inverse lifetime of 

680: the level. That is, as a rough estimation, when the magnetic field ranges 

681: between (see LL04) 

682: %

683: \begin{equation}

684: \label{eq:formulita}

685: 0.1\,B_{\rm c} \le B \le 10 \, B_{\rm c} \;\; ,

686: \end{equation}

687: %

688: with

689: %

690: \begin{equation}

691: B_{\rm c}\approx\frac{1.137\times10^{-7}}{t_{\rm life}\,g_{\rm L}} \;\; ,

692: \end{equation}

693: %

694: where $t_{\rm life}$ (in seconds) is the radiative lifetime of the lower 

695: or upper level of the line transition under consideration, $g_{\rm L}$ 

696: is its Land\'e factor and $B_{\rm c}$ is the critical magnetic field intensity

697: in G\footnote{The previous expression of $B_{\rm c}$ is exact only for an 

698: isolated level (see footnote\ref{note:Hanle}).}. 

699: Since the relevant atomic level here

700: is the ground level, it is important to note that its radiative lifetime is 

701: $t_{\rm life} \approx 1/(B_{\ell u}\, J_{0}^{0})$. 

702: As seen in Figure~\ref{fig:sigma20} this 

703: first decrease takes place for magnetic fields of the order of $10^{-1}$G, 

704: consistently with our simplifying assumption that the pumping radiation field 

705: is the continuum radiation tabulated by \citet{Allen}. If, on the contrary, one 

706: takes into account the line profile (resulting in a smaller value of 

707: $J_{0}^{0}$) the decrease will take place for smaller magnetic fields 

708: \citep[cf.~the Na~{$\!$\sc i} results of][]{jtb02}. 

709: Note that this decrease of the atomic polarization in the ground level 

710: has a feedback even on the D$_2$ upper level.

711: Concerning the second sudden decrease of the atomic polarization of the 

712: D$_1$ levels, for $B$ larger than 100~G, we recall that this is due to the

713: inhibition of the repopulation pumping mechanism discussed by \citet{jtb02} 

714: and by \citet{Cas02}, which sets in when the electronic and nuclear angular 

715: momenta, $J$ and $I$, of the $^{2}P_{3/2}$ level are decoupled, 

716: (i.e.~when this level, the only one that can carry alignment even in the 

717: absence of HFS, enters the complete Paschen-Back effect regime). 

718: Note that for the case

719: of sodium this sudden decrease occurs for $B$ larger than 10~G simply because 

720: the complete Paschen-Back effect regime is reached for weaker magnetic fields 

721: in sodium than in barium.\\ 

722: \indent As expected, because of the symmetry of the problem,

723: for a vertical magnetic field only the components

724: with $Q\!=\!0$ are non zero. For different orientations of the magnetic field 

725: we have in general contributions coming from all the density matrix elements;

726: in particular it is possible to demonstrate that the

727: components with $Q\!=\!0$ are independent of the magnetic field azimuth, 

728: the components with $Q\!=\!1$ change sign under an azimuth rotation of 

729: $180^{\circ}$, the components with $Q\!=\!2$ change sign under an azimuth 

730: rotation of $90^{\circ}$, and so on.\\

731: \indent The expressions of the emission coefficients (eq.~\ref{eq:epsilon}), 

732: as well as the expressions of all the other radiative transfer coefficients 

733: given in LL04 hold in the magnetic reference system.

734: Therefore we have to transform the spherical statistical tensors, obtained

735: solving the SEEs written in the local vertical 

736: reference system, into the magnetic field reference system.

737: Indicating with $[\rho^{K}_{Q}(F,F^{\,\prime}\,)]_{B}$ the spherical statistical

738: tensor components in the magnetic field reference system, and with

739: $[\rho^{K}_{Q}(F,F^{\,\prime}\,)]_{V}$ the spherical statistical tensor 

740: components in the local vertical reference system, we have

741: %

742: \begin{equation}

743: [\rho^{K}_{Q}(F,F^{\,\prime}\,)]_{B}=

744: \sum_{Q^{\prime}}\,[\rho^{K}_{Q^{\prime}}(F,F^{\,\prime}\,)]_{V}\,

745: D^{K}_{Q^{\prime}Q}(R)^{\ast} \;\; ,

746: \end{equation}

747: %

748: where $D^{K}_{Q^{\prime}Q}(R)$ is the rotation matrix calculated for the

749: rotation $R$ which carries the vertical reference system into the magnetic

750: reference system (referring to Fig.~{\ref{fig:geometry}}, we have 

751: $R\!=\!(\chi_B,\theta_B,0)$), and where the apex 

752: ``$\,\ast\,$'' indicates the complex conjugate.

753: We observe that after the rotation, because of the symmetry of the problem,

754: the spherical statistical tensors in the magnetic reference

755: system do not depend on the azimuth but only on the inclination of the

756: magnetic field with respect to the local vertical reference system.

757: 

758: \section{THE POLARIZATION OF THE EMERGENT SPECTRAL LINE RADIATION} 

759: \label{sect:emergent}

760: For the case of a tangential observation in a plane-parallel atmosphere, it 

761: can be shown that, under the approximation of a weakly polarizing

762: atmosphere ($\varepsilon_{I}\gg\varepsilon_{Q},\varepsilon_{U},

763: \varepsilon_{V}$; $\eta_{I}\gg\eta_{Q},\eta_{U},\eta_{V}$),

764: the emergent fractional polarization is given by 

765: \citep[e.g.,][]{jtb03}

766: %

767: \begin{equation}

768: \label{eq:dichroism}

769: \frac{X(\nu,\mathbf{\Omega})}{I(\nu,\mathbf{\Omega})} \approx

770: \frac{\varepsilon_{X}(\nu,\mathbf{\Omega})}

771: {\varepsilon_{I}(\nu,\mathbf{\Omega})}-

772: \frac{\eta_{X}(\nu,\mathbf{\Omega})}

773: {\eta_{I}(\nu,\mathbf{\Omega})} \;\;\;\;\;\; {\rm{with}}\; X=Q,U,V \;\; .

774: \end{equation}

775: %

776: We point out that the first term of equation~(\ref{eq:dichroism}) is the 

777: contribution to 

778: the emergent fractional polarization due to selective emission processes, while 

779: the second one is caused by dichroism (selective absorption of polarization 

780: components).

781: As shown in Figure~\ref{fig:sigma20}, for the D$_{2}$ line the contribution due 

782: to dichroism is much smaller than that due to selective emission. 

783: For this reason, from now on we will describe the polarization properties of 

784: the radiation emergent from the slab using the relation

785: %

786: \begin{equation}

787: \frac{X(\nu,\mathbf{\Omega})}{I(\nu,\mathbf{\Omega})}=

788: \frac{\varepsilon_{X}(\nu,\mathbf{\Omega})}

789: {\varepsilon_{I}(\nu,\mathbf{\Omega})} \;\;  .

790: \label{eq:fract-polar1}

791: \end{equation}

792: %

793: We recall that the polarization properties of the

794: emergent radiation will always be described assuming the reference direction 

795: for positive $Q$ parallel to the slab.\\

796: \indent It is important to remember that the expressions for the emission and 

797: absorption coefficients given in LL04 take into account only the line processes.

798: For this reason, in order to be able to reproduce qualitatively the 

799: observed profiles, we need to add the contribution coming from the continuum.

800: Assuming that the continuum is not polarized and constant across

801: the line, we have

802: %

803: \begin{equation}

804: \frac{X(\nu,\mathbf{\Omega})}{I(\nu,\mathbf{\Omega})}=

805: \frac{\varepsilon_{X}^{\, l}(\nu,\mathbf{\Omega})}

806: {\varepsilon_{I}^{\, l}(\nu,\mathbf{\Omega})+\varepsilon^{\,\rm c}_{I}} \;\;  ,

807: \label{eq:fract-polar2}

808: \end{equation}

809: %

810: where $\varepsilon^{\,\rm c}_{I}$ is the continuum contribution to the 

811: intensity of the emergent radiation, and where the apex ``{ \it l} '' is to 

812: recall that the corresponding quantity refers only to the line processes.

813: In Appendix~\ref{app:dichroism} we show and briefly discuss some results

814: obtained by applying equation~(\ref{eq:dichroism}).\\

815: \indent The Ba~{$\!$ \sc ii} D-lines that we are investigating are strong lines:

816: according to theoretical models of the solar atmosphere the wings of these 

817: lines originate in the photosphere, while the line-cores in the high 

818: photosphere-low chromosphere.

819: The optically thin slab model illuminated by the photospheric continuum 

820: that we are considering in this paper is therefore just a zero-order 

821: approximation. 

822: Nevertheless, it allows us to take into account 

823: in a very rigorous way the atomic physics involved in the problem, and to 

824: understand its essential role on the magnetic sensitivity of the polarization 

825: profiles of these lines, avoiding complications coming from radiation transfer 

826: effects. This is the first step of our investigation, in a forthcoming paper 

827: we will propose more realistic models, where radiation transfer effects

828: will be taken into account.\\

829: \indent Once the SEEs in the vertical reference system 

830: have been solved numerically, and 

831: the spherical statistical tensors have been rotated to the magnetic

832: field reference system, we can calculate the 

833: emission coefficients for $90^{\circ}$ scattering 

834: ($\theta \!= \! 90^{\circ}$ and $\chi \!= \! 0^{\circ}$ in 

835: Fig.~\ref{fig:geometry}) 

836: by means of equation~(\ref{eq:epsilon}), and the polarization of the 

837: scattered radiation through equation~(\ref{eq:fract-polar2}). In the 

838: following subsections we present our results for the Ba~{$\!$\sc ii} D$_2$ and 

839: D$_1$ lines, for various magnetic field configurations.

840: 

841: 

842: \subsection{{\it D$_2$ Line -- No Magnetic Field Case:\\

843: Origin of the Three Peaks Structure and Choice

844: of Parameters Values}}

845: \label{sect:D2-zero-field}

846: As a first step, we observe the laboratory positions of the various 

847: HFS components of the D$_{2}$ line. The three isotopes without HFS contribute 

848: to the D$_2$ line with just one component each, those which have HFS 

849: contribute to this line with six components each, which may overlap.

850: As seen in Figure~\ref{fig:hfscomp} (right panel), it is possible to divide the 

851: various HFS components into three groups. The central group, at about 

852: 4554.03{\AA}, is composed by the five components (three visible in the figure) 

853: due to the five isotopes without HFS (note that the main contribution comes 

854: from the isotope 138 because of its high abundance, while isotopes 130 and 132 

855: bring a negligible contribution, not visible in the figure). 

856: The other components, due to the isotopes with HFS, fall at different 

857: wavelengths but can be gathered into two groups

858: because of the large splitting of the ground level into the two $F\!=\!1$ and 

859: $F\!=\!2$ HFS levels (see Fig.~\ref{fig:paschen}). In particular, the group at 

860: about 4553.995{\AA} is composed by the HFS components, of both isotopes 135 

861: and 137, associated to the transitions towards the lower $F\!=1\!$ HFS level, 

862: while the group at about 4554.045{\AA} is composed by the HFS components

863: associated to the transitions to the lower $F\!=\!2$ HFS level.

864: %%

865: \begin{figure}[!t] 

866: \begin{center}

867: \plottwo{f4a.eps}{f4b.eps}

868: \caption{{\footnotesize Laboratory positions of the various 

869: HFS components of the D$_{1}$ (left panel) and D$_{2}$ (right panel) lines 

870: resulting from the seven isotopes considered. The horizontal axis gives the 

871: wavelength (in \AA) measured from 4934.075~{\AA} for the D$_{1}$ line,

872: and from 4554.029~{\AA} for the D$_{2}$ line. 

873: These are the wavelengths (in air) of the isotope 138

874: D$_1$ and D$_2$ lines, respectively.}}

875: \label{fig:hfscomp}

876: \end{center}

877: \end{figure}

878: %%

879: As we will see in more detail below, and as already

880: pointed out by \citet{Ste97b}, the origin of the three peaks structure of 

881: the D$_{2}$ line lies in this splitting of the various components into these 

882: three groups. Similar considerations

883: could be done about the position and relative strength of the various

884: components of the D$_{1}$ line, with the difference that now the upper 

885: level just has two HFS levels (instead of four).\\

886: \indent Let us begin our analysis of the D$_{2}$ profile by considering only 

887: the isotope 138 (without HFS), and let us assume a Doppler width of 30~m{\AA}. 

888: For this isotope, as shown in the left panels of 

889: Figure~\ref{fig:D2-zero-field}, 

890: the ratio $\varepsilon_{Q}^{\, l}(\lambda)/\varepsilon_{I}^{\, l}(\lambda)$ 

891: (often referred to as fractional polarization)

892: is constant and different from zero\footnote{See Appendix~\ref{app:two-level} 

893: for an analytical proof of this result.}. 

894: Adding the contribution of the continuum, the same ratio remains unchanged in 

895: the line-core (where $\varepsilon_{I}^{\, l}\gg\varepsilon_{I}^{\, \rm c}$), 

896: while it goes to zero at the wavelengths corresponding to the wings of the 

897: intensity profile.

898: The same considerations can be done for all the other isotopes without HFS.

899: %%

900: \begin{figure}[!t] 

901: \begin{center}

902: \includegraphics[angle=270, width=0.95\textwidth]{f5.eps}

903: \caption{{\footnotesize Theoretical profiles of the emission coefficients 

904: $\varepsilon_{I}^{\,l}(\lambda)$ and $\varepsilon_{Q}^{\,l}(\lambda)$,

905: and of the ratio 

906: $\varepsilon_{Q}^{\,l}(\lambda)/(\varepsilon_{I}^{\,l}(\lambda)+

907: \varepsilon_{I}^{\, \rm c})$, of the Ba~{\sc ii} D$_2$ line, for the isotope 

908: 138, for the isotope 137 and for all the seven isotopes together, in the 

909: absence of a magnetic field. The emission profiles are 

910: normalized to the maximum value of the intensity emission profile, 

911: calculated taking into account the contribution of all the seven isotopes, 

912: $(\varepsilon_{I}^{l})_{\rm{max}}$.  

913: The vertical dashed lines show the positions of the three groups 

914: of transitions (see text). 

915: The last row shows the 

916: $\varepsilon_{Q}^{\,l}(\lambda)/(\varepsilon_{I}^{\,l}(\lambda)+

917: \varepsilon_{I}^{\, \rm c})$ profiles, without continuum (solid), 

918: with a continuum $\varepsilon_{I}^{\, \rm c}/(\varepsilon_{I}^{l})_{\rm{max}}$ 

919: of $10^{-5}$ (dot), of  

920: $3\times 10^{-5}$ (long dash), and of

921: $9\times 10^{-5}$ (dash-dot).

922: \label{fig:D2-zero-field}}}

923: \end{center}

924: \end{figure}

925: %%

926: Let us consider now the isotope 137 (with HFS). For this

927: isotope, assuming the same Doppler width, the profiles of the line emission 

928: coefficients $\varepsilon_{I}^{\, l}$ and 

929: $\varepsilon_{Q}^{\, l}$ show two peaks at the wavelength positions

930: of the two groups of HFS transitions (see Fig.~\ref{fig:D2-zero-field}).

931: The ratio $\varepsilon_{Q}^{\, l}(\lambda)/\varepsilon_{I}^{\, l}(\lambda)$ is

932: no longer constant but decreases showing two broad minima at the wavelength 

933: positions corresponding to the wings of the $\varepsilon_{I}^{\, l}$ and 

934: $\varepsilon_{Q}^{\, l}$ profiles, and assumes the same value as the isotopes

935: without HFS moving away from the line-core\footnote{See 

936: Appendix~\ref{app:two-level} for an analytical proof of this result.}. 

937: This profile clearly shows the depolarizing effect of the HFS.

938: The role of the continuum is the same as observed for the isotopes without

939: HFS. The same arguments hold for the isotope 135.\\

940: \indent This investigation on the isotopes 138 and 137 shows that the

941: central peak of the observed $Q/I$ profile is due to the isotopes without 

942: HFS, while the two secondary peaks are due to the isotopes with HFS. 

943: The position and amplitude of these secondary peaks appear to be strongly 

944: dependent on the background continuum emissivity $\varepsilon_{I}^{\,\rm c}$,

945: a physical quantity that, given the exploratory character of this paper, we will

946: just parametrize in order to reproduce at best the observed profile.\\

947: \indent We consider now all the isotopes together and we adjust the

948: Doppler width, the anisotropy factor and the continuum

949: intensity in order to obtain the best fit to the $Q/I$ profile observed by

950: \citet{Ste97a}, still assuming that no magnetic field is present.

951: Changing the Doppler width we can modify the separation between the two 

952: minima of the profile

953: $\varepsilon_Q^{\, l}(\lambda)/\varepsilon_I^{\, l}(\lambda)$. 

954: To obtain the same separation as the observed profile we need a value of 

955: about 30~m{\rm{\AA}}. This value seems

956: to be very reasonable as it can be obtained assuming a temperature of

957: about 6000~K and a microturbulent velocity of about 1.8~km/s, values which are

958: in good agreement with those given by semi-empirical chromospheric models 

959: at the height of about 1000~km.\\

960: \indent Modifying the anisotropy factor we simply scale the ratio

961: $\varepsilon_Q^{\, l}(\lambda)/\varepsilon_I^{\, l}(\lambda)$.

962: Radiative transfer effects are disregarded in our model and, since the

963: Ba~{$\!$\sc ii} D$_{2}$ line is a strong line, we can expect that the 

964: calculated value for the anisotropy factor taking into account only the 

965: photospheric continuum (see \S~\ref{sect:radiation}),

966: is probably overestimated.

967: %%

968: \begin{figure}[!t]

969: \begin{center}

970: \includegraphics[width=0.70\textwidth]{f6.eps}

971: \caption{{\footnotesize {\it Panel} a: observed $Q/I$ profile of the 

972: Ba~{\sc ii} D$_{2}$ line \citep{Ste97a}.

973: {\it Panel} b: theoretical $Q/I$ profile

974: obtained in the absence of a magnetic field, choosing a Doppler width

975: ($\Delta \lambda_D$) of 30~m{\rm{\AA}}, an anisotropy factor ($w$) of 0.037,

976: and a continuum $\varepsilon_{I}^{\,\rm c}/(\varepsilon_{I}^{l})_{\rm{max}}$ of

977: $9 \times 10^{-5}$.

978: {\it Panel} c: theoretical $Q/I$ profile obtained in the presence of a

979: microturbulent magnetic field of 5~G, choosing the following values of the free

980: parameters: $\Delta \lambda_D$=30~m{\rm{\AA}}, $w$=0.052,

981: $\varepsilon_{I}^{\,\rm c}/(\varepsilon_{I}^{l})_{\rm{max}}$=$1.77 \times

982: 10^{-4}$.

983: {\it Panel} d: theoretical $Q/I$ profile obtained in the presence of a

984: vertical magnetic field of 40~G, choosing the following values of the free

985: parameters: $\Delta \lambda_D$=30~m{\rm{\AA}}, $w$=0.037,

986: $\varepsilon_{I}^{\,\rm c}/(\varepsilon_{I}^{l})_{\rm{max}}$=

987: $7.5\times 10^{-5}$.}}

988: \label{fig:bestfit}

989: \end{center}

990: \end{figure}

991: %%

992: We find indeed that the anisotropy factor has to be decreased to the value of 

993: 0.037 (approximately $1/5$ of the value 0.176 mentioned in 

994: \S~\ref{sect:radiation}) to obtain the observed value of the ratio 

995: $Q/I$ at the wavelength position of the central peak.\\

996: \indent In this paper we are not taking into account the continuum processes 

997: in a rigorous way. As mentioned above, we describe the effect of the 

998: continuum simply through the parameter $\varepsilon^{\,\rm c}_{I}$. 

999: As shown in Figure~\ref{fig:D2-zero-field}, the continuum modifies the 

1000: wings of the profile. Taking as a reference

1001: the ``red'' secondary peak of the observed profile, we find that the best fit

1002: is obtained assuming a value for 

1003: $\varepsilon_{I}^{\,\rm c}/(\varepsilon_{I}^{l})_{\rm{max}}$ of

1004: $9\times 10^{-5}$.

1005: With these values of the parameters, applying equation~(\ref{eq:fract-polar2}), 

1006: we get the profile shown in the panel b of Figure~\ref{fig:bestfit} which 

1007: reproduces quite well the 

1008: observed profile, and which is very similar to a theoretical profile already

1009: obtained by \citet{Ste97b}.\\

1010: \indent The main aim of this work is to investigate the magnetic 

1011: sensitivity of the 

1012: linear and circular polarization of the D-lines of Ba~{$\!$\sc ii}, and not

1013: merely to reproduce as better as possible the observed profiles of these lines.

1014: For this reason, we prefer to perform our investigation by sticking 

1015: to the values of the parameters found for the simpler, unmagnetized case. 

1016: Obviously there is 

1017: no reason to think that the best agreement with the observed profile has to be 

1018: found in the absence of a magnetic field. It is likely enough that the best 

1019: agreement could be obtained in the presence of a deterministic or

1020: microturbulent magnetic field, using different values of the parameters. 

1021: For example, in panels c and d of Figure~\ref{fig:bestfit} we show the best

1022: theoretical profiles that we have obtained in the presence of a microturbulent

1023: magnetic field of 5~G and a vertical magnetic field of 40~G, choosing different

1024: values of the free parameters (see caption to Fig.~{\ref{fig:bestfit}}).

1025: 

1026: 

1027: \subsection{{\it D$_2$ Line -- The Influence of a Magnetic Field on the 

1028: Emergent Polarization}}

1029: \label{sect:magnetic}

1030: Depending on its strength, and on its direction with respect to the local 

1031: vertical and to the direction of the scattered radiation, a magnetic field 

1032: will differently modify the linear and circular polarization of the line 

1033: through the Zeeman and the Hanle effects\footnote{Hereafter, regardless of the 

1034: particular regime (Zeeman effect regime, 

1035: incomplete or complete Paschen-Back effect regime), any polarization signal 

1036: that originates from the splitting among the magnetic sublevels will be 

1037: referred to as Zeeman effect.}. 

1038: It is well known that the Zeeman effect produces 

1039: in general elliptical polarization, which degenerates into linear polarization 

1040: if the magnetic field lies on the plane perpendicular to the line-of-sight 

1041: (LOS), and into circular polarization if the magnetic field lies along the 

1042: LOS. The Zeeman effect dominates the polarization of the scattered radiation 

1043: if the splitting among the magnetic sublevels is of the same order of magnitude 

1044: or larger than the Doppler width of the line. This criterion gives a critical 

1045: value of the magnetic field strength for the Ba~{$\!$\sc ii} D$_2$ line of 

1046: about 3000~G. However, if the magnetic field 

1047: is not too weak, and if there are no other mechanisms that dominate the 

1048: polarization, it is possible to identify Zeeman effect signatures on the 

1049: fractional polarization profiles even for intensities much smaller than the 

1050: critical value. As we will see below in 

1051: Figures~\ref{fig:D2-vertical-field} and \ref{fig:D2-vertical-Zeeman}, for 

1052: the line under investigation magnetic fields of about 50~G

1053: are enough in order the transverse Zeeman effect to produce 

1054: appreciable modifications of the linear polarization signal.\\ 

1055: \indent On the other hand, as described in \S~\ref{sect:polarization}, 

1056: a magnetic field is able to modify the atomic polarization, and therefore the 

1057: polarization of the scattered radiation.

1058: Depending on the configuration of the magnetic field, and on the geometry of 

1059: the scattering event, different signatures of the Hanle effect can be

1060: produced on the polarization profiles. 

1061: As the observed scattering polarization in the Ba~{$\!$\sc ii} 

1062: D$_2$ line is dominated by the atomic polarization of the upper level,

1063: recalling that for the $^2P_{3/2}$ level

1064: $t_{\rm life} \approx 1/A_{u\ell} \approx 10^{-8}$~s and $g_{\rm L}=1.33$,

1065: applying equation~(\ref{eq:formulita}), we find that the line

1066: is expected to be sensitive to the Hanle effect for magnetic field 

1067: strengths ranging approximately between 1~G and 100~G (values which are 

1068: smaller than the ones needed for the transverse Zeeman effect to be 

1069: appreciable). 

1070: 

1071: 

1072: \subsection{{\it D$_2$ Line -- Vertical Magnetic Field}}

1073: \label{sect:D2-vertical-field}

1074: In this section we consider the effect of a vertical magnetic field on the 

1075: theoretical $Q/I$ profile\footnote{Hereafter by `theoretical $Q/I$ profile' 

1076: we will always refer to the profile obtained by applying 

1077: equation~(\ref{eq:fract-polar2}).} of Figure~\ref{fig:bestfit} (panel b). 

1078: The results are shown in Figure~\ref{fig:D2-vertical-field}.

1079: The first interesting feature is the enhancement of the linear polarization 

1080: at the wavelength positions of the two dips between the line-core peak and 

1081: the two secondary peaks, for magnetic fields relatively 

1082: weak (less than 50~G), for which the influence of the transverse 

1083: Zeeman effect is negligible. 

1084: In order to understand which physical mechanism is at the origin of this and

1085: other features shown by these $Q/I$ profiles, in the various ranges of magnetic 

1086: field intensity, we try to distinguish which polarization properties of the 

1087: emergent radiation are due to the atomic polarization effects and which 

1088: ones are due to the Zeeman effect. 

1089: %%

1090: \begin{figure}[!t]

1091: \plottwo{f7a.eps}{f7b.eps}

1092: \caption{{\footnotesize Theoretical $Q/I$ profile of the Ba~{\sc ii} D$_2$ line 

1093: in the presence of a vertical magnetic field ($\theta_B \!=\! 0^{\circ}$). 

1094: In the left panel the magnetic field varies between 0 and 100~G while

1095: in the right panel it varies between 0 and 1000~G. Note the difference in the 

1096: scale of the two figures.}}

1097: \label{fig:D2-vertical-field}

1098: \end{figure}

1099: %%

1100: To this aim, we can obtain 

1101: interesting information by plotting the profiles obtained through 

1102: equation~(\ref{eq:fract-polar2}) according to two different strategies: 

1103: \begin{itemize}

1104: \item[{\sc a)}] taking the nominal values for the 

1105: spherical statistical tensors, but setting $B\!=\!0$ when calculating the energy

1106: eigenvalues and eigenvectors,

1107: \item[{\sc b)}] setting equal to zero all the 

1108: spherical statistical tensors, except $\rho^{0}_{0}$, but taking properly into 

1109: account the influence of $B$ on the energy eigenvalues and eigenvectors. 

1110: \end{itemize}

1111: %%

1112: \begin{figure}[!t]

1113: \includegraphics[width=0.95\textwidth]{f8.eps}

1114: \caption{{\footnotesize {\it Left column}: theoretical $Q/I$ profiles of the 

1115: Ba~{\sc ii} D$_{2}$ line obtained in the presence of a vertical magnetic field. 

1116: {\it Middle column}: theoretical $Q/I$ profiles obtained neglecting the Zeeman 

1117: effect (only atomic polarization effects). 

1118: {\it Right column}: theoretical $Q/I$ profiles obtained neglecting atomic 

1119: polarization effects (only Zeeman effect). The zero of the wavelength scales 

1120: is taken at 4554~\AA.}}

1121: \label{fig:D2-vertical-Zeeman}

1122: \end{figure}

1123: %%

1124: In the former case (Fig.~\ref{fig:D2-vertical-Zeeman}, middle column) we are 

1125: taking into account only the effects due to atomic polarization, neglecting 

1126: the Zeeman effect, while in the latter case 

1127: (Fig.~\ref{fig:D2-vertical-Zeeman}, right 

1128: column) we are taking into account only this second effect, within the 

1129: framework of the Paschen-Back effect theory.\\

1130: \indent From Figure~{\ref{fig:D2-vertical-Zeeman}}, it is clear that the 

1131: above-mentioned enhancement of the linear polarization at the wavelength 

1132: positions of the dips, in the presence of a weak vertical magnetic field, 

1133: is not due to the Zeeman effect, but to the Hanle effect acting only on 

1134: the isotopes with HFS (see \S~\ref{sect:D2-zero-field}). 

1135: We note in fact that the central peak, which is due to the isotopes without HFS 

1136: is not sensitive, in this range, to the magnetic field.

1137: Actually this particular behaviour can be explained in terms of two different 

1138: mechanisms: the anti-level-crossing effect (briefly introduced in 

1139: \S~\ref{sect:equations}), and the change of coupling scheme 

1140: of the atomic system\footnote{According to our definition, these effects should 

1141: be better considered as particular cases of the Hanle effect.}.

1142: We note first that if the magnetic field lies along the symmetry 

1143: axis of the radiation field, only the statistical tensors $\rho_0^0$ and 

1144: $\rho_0^2$ are different from zero (see also \S~\ref{sect:radiation}). 

1145: As the incomplete Paschen-Back effect regime is reached, the HFS magnetic 

1146: sublevels with the same $f$ quantum number separate from each other

1147: (see \S~\ref{sect:equations}), as a consequence the terms

1148: $\rho_0^2(F,F^{\prime})$ (which 

1149: quantifies the corresponding quantum interferences) decrease, and 

1150: this causes an increase of the polarization of the scattered radiation 

1151: (anti-level-crossing-effect, see LL04 for details).

1152: On the other hand, as already stated in \S~\ref{sect:equations}, going 

1153: from the Zeeman effect regime to the complete Paschen-Back effect regime, 

1154: the magnetic field produces an energy eigenvectors basis transformation. 

1155: This transformation implies a changing of the coupling scheme of the atomic 

1156: system, which affects the polarization state of the atomic 

1157: system\footnote{The complex 

1158: mechanism of inhibition of atomic polarization transfer discussed in 

1159: \citet{Cas02} is a particular consequence of this coupling scheme 

1160: transformation.}.

1161: Both these mechanisms begin to play an appreciable role as the upper level of 

1162: the D$_2$ enters the incomplete Paschen-Back effect regime.

1163: The order of magnitude of the magnetic field strength needed to reach 

1164: this regime can be estimated from the relation

1165: $0.1 \le \Delta\lambda_{B}/\Delta\lambda_{{\rm{hfs}}} \approx 1$, where

1166: $\Delta\lambda_{B}$ is the splitting induced by the magnetic field, and

1167: $\Delta\lambda_{{\rm{hfs}}}$

1168: is the wavelength separation between the HFS $F$-levels.

1169: Applying this relation to the Ba~{$\!$\sc ii} D$_2$ line, we find that this 

1170: effect is expected to take place for magnetic fields larger than 10~G, as it is

1171: observed in Figure~\ref{fig:D2-vertical-Zeeman}.

1172: The possibility of an enhancement of the scattering polarization in the

1173: presence of a vertical magnetic field, through this kind of mechanisms,

1174: was already pointed out by \citet{jtb02} for the case of the

1175: Na~{$\!$\sc i} D$_2$ line.\\

1176: \indent Increasing the magnetic field strength, the transverse 

1177: Zeeman effect eventually 

1178: becomes appreciable and, besides the previous effect, we see an increase of 

1179: the polarization at the wavelength position of the two peaks on the wings of 

1180: the profile (see Fig.~\ref{fig:D2-vertical-field} and 

1181: Fig.~\ref{fig:D2-vertical-Zeeman}).

1182: Going to magnetic fields of about 200~G or stronger we enter the transverse 

1183: Zeeman effect regime, and the linear polarization profile takes the typical 

1184: symmetrical shape\footnote{Note that here we are plotting the ratio

1185: $\varepsilon_{Q}(\lambda)/\varepsilon_{I}(\lambda)$ and not just

1186: $\varepsilon_{Q}(\lambda)$.} (see Fig.~\ref{fig:D2-vertical-field} 

1187: and Fig.~\ref{fig:D2-vertical-Zeeman}). 

1188: 

1189: 

1190: \subsection{{\it D$_2$ Line -- Horizontal Magnetic Field, Perpendicular to 

1191: the Line of Sight}}

1192: \label{sect:D2-horizontal-field}

1193: %%

1194: \begin{figure}[!t]

1195: \plottwo{f9a.eps}{f9b.eps}

1196: \caption{{\footnotesize Theoretical $Q/I$ profiles of the Ba~{\sc ii} D$_2$ line

1197: in the presence of a horizontal magnetic field, perpendicular to the line of 

1198: sight ($\theta_B \!=\! 90^{\circ}, \, 

1199: \chi_B \!=\! \pm 90^{\circ}$)}}

1200: \label{fig:D2-horizontal-field}

1201: \end{figure}

1202: %%

1203: In the presence of a weak horizontal magnetic field, perpendicular

1204: to the line of sight, as the field increases we observe

1205: a decrease of the linear polarization at the wavelength position of the

1206: central peak, due to the Hanle effect.

1207: However, consistently with the fact that the Hanle effect has to vanish in

1208: the far wings of the line, the two peaks on the wings remain almost unaffected,

1209: as can be seen in the left panel of Figure~\ref{fig:D2-horizontal-field}.

1210: Similarly to what happens in the presence of a vertical magnetic field,

1211: going to intensities of about 50~G or stronger we enter the

1212: transverse Zeeman effect regime, and the $Q/I$ profile takes the well known

1213: shape shown in the right panel of Figure~\ref{fig:D2-horizontal-field}.

1214: 

1215: 

1216: \subsection{{\it D$_2$ Line -- Horizontal Magnetic Field, Directed Along the 

1217: Line of Sight}}

1218: \label{sect:D2-longitudinal-field}

1219: %%

1220: \begin{figure}[!t]

1221: \plottwo{f10a.eps}{f10b.eps}

1222: \caption{{\footnotesize Theoretical $Q/I$ (left panel) and $U/I$

1223: (right panel) profiles of the Ba~{\sc ii} D$_2$ line in 

1224: the presence of a longitudinal magnetic field ($\theta_B \!=\! 90^{\circ}, \, 

1225: \chi_B \!=\! 0^{\circ}$). For $\chi_B \! = \! 180^{\circ}$ the $U/I$ profile 

1226: would be the same except for an overall sign switch.}}

1227: \label{fig:D2-longitudinal-field-QU}

1228: \end{figure}

1229: %%

1230: In the presence of a longitudinal magnetic field of increasing

1231: strength, there is again a decrease of the linear polarization at the

1232: wavelength position of the central peak, due to the Hanle effect, while the

1233: two peaks on the wings are not affected.

1234: In this geometry the Zeeman effect does not modify the linear

1235: polarization and, going to stronger magnetic fields, we enter a regime of

1236: saturation, as shown in the left panel of 

1237: Figure~\ref{fig:D2-longitudinal-field-QU}.

1238: Because of the Hanle effect, we have in this case a rotation of the plane of

1239: linear polarization. This implies the presence of the non zero $U/I$ signal

1240: shown in the right panel of Figure~\ref{fig:D2-longitudinal-field-QU}.

1241: Finally we have a typical antisymmetric $V/I$ signal due to the longitudinal

1242: Zeeman effect, as shown in Figure~\ref{fig:D2-longitudinal-field-V}.

1243: For weak magnetic fields (of the order of about 100~G or

1244: weaker) the signal increases linearly with the magnetic field strength.

1245: Going to stronger fields the linearity is slowly

1246: lost, and the profile starts to saturate.

1247: %%

1248: \begin{figure}[!t]

1249: \plottwo{f11a.eps}{f11b.eps}

1250: \caption{{\footnotesize Theoretical $V/I$ profiles of the Ba~{\sc ii} D$_2$ line

1251: in the presence of a longitudinal magnetic field.}}

1252: \label{fig:D2-longitudinal-field-V}

1253: \end{figure}

1254: %%

1255: 

1256: 

1257: \subsection{{\it D$_2$ Line -- Random-Azimuth Magnetic Field}}

1258: \label{sect:D2-random-azimuth}

1259: %%

1260: \begin{figure}[!t]

1261: \begin{center}

1262: \includegraphics[width=0.9\textwidth]{f12.eps}

1263: \caption{{\footnotesize Theoretical $Q/I$ profiles of the Ba~{\sc ii} D$_2$ line

1264: in the presence of a random-azimuth magnetic field for three different 

1265: inclinations: 30$^\circ$ (top), 60$^\circ$ (middle), and 90$^\circ$ (bottom).}}

1266: \label{fig:D2-random-azimuth}

1267: \end{center}

1268: \end{figure}

1269: %%

1270: In this section we present the results obtained for the fractional polarization 

1271: in the presence of magnetic fields with a given inclination and a random 

1272: azimuth (i.e.~the results obtained in the presence of a magnetic field of 

1273: given strength and inclination, averaged over the azimuth). 

1274: Figure~\ref{fig:D2-random-azimuth} shows the theoretical profiles 

1275: obtained in the presence of random-azimuth magnetic fields with inclinations of 

1276: 30$^\circ$, 60$^\circ$ and 90$^\circ$.

1277: In the 30$^\circ$ case we observe that increasing the magnetic field 

1278: strength the linear polarization decreases at the wavelength position of 

1279: the central peak and of the two dips close to it, because of the Hanle effect. 

1280: For magnetic fields 

1281: of about 100~G it is possible to observe the first signatures of the Zeeman 

1282: effect, which dominates the linear polarization as we further increase its 

1283: intensity. Note that as the main component of the magnetic field is vertical, 

1284: the Zeeman effect produces the typical three lobes profiles with the same 

1285: signs as in the case of a deterministic vertical magnetic field.

1286: Similar considerations hold for a random-azimuth magnetic field with an 

1287: inclination of 60$^\circ$ and 90$^\circ$. As the main component of the 

1288: magnetic field is now horizontal, as far as 

1289: the Zeeman effect starts to dominate the polarization (which happens for 

1290: fields of about 200~G or stronger in the 60$^\circ$ case, and for fields 

1291: of about 50~G or stronger in the 90$^\circ$ case), we obtain the well known 

1292: Zeeman effect profiles 

1293: with the same signs as in the case of a deterministic horizontal magnetic 

1294: field, perpendicular to the LOS.

1295: Stronger magnetic fields are needed for the Zeeman effect to be 

1296: appreciable in the presence of a random-azimuth magnetic field with an 

1297: inclination of 60$^\circ$ because in this case the vertical and the 

1298: horizontal components of the magnetic field (which produce Zeeman effect 

1299: profiles with opposite signs) are comparable.

1300: 

1301: 

1302: \subsection{{\it D$_2$ Line -- Microturbulent Magnetic Field}}

1303: \label{sect:D2-microturbulent}

1304: Averaging the emission coefficient over all the possible orientations of the 

1305: magnetic field\footnote{See Appendix~\ref{app:average} for the details  

1306: implied in performing this average.}, 

1307: we can investigate the polarization properties of the line in the presence 

1308: of a unimodal microturbulent magnetic field.

1309: %%

1310: \begin{figure}[!t]

1311: \begin{center}

1312: \includegraphics[scale=0.32]{f13.eps}

1313: \caption{{\footnotesize Theoretical $Q/I$ profiles of the Ba~{\sc ii} D$_2$ line

1314: in the presence of a microturbulent magnetic field.}}

1315: \label{fig:D2-microturbulent}

1316: \end{center}

1317: \end{figure}

1318: %%

1319: As shown in Figure~\ref{fig:D2-microturbulent}, we observe that the linear 

1320: polarization at the wavelength position of the central peak decreases,

1321: while the two peaks on the wings remain constant as the magnetic

1322: field strength is increased.

1323: This behaviour can be easily understood: in the presence of a

1324: microturbulent magnetic field there is no observational polarization signal

1325: due to the Zeeman effect, while the Hanle

1326: effect produces a decrease of the linear polarization only in the line-core.

1327: For $B \! > \! 100$~G, we enter a saturation regime and, contrary to the case 

1328: of a longitudinal field, we still have a non-zero signal in the line-core.

1329: 

1330: 

1331: \subsection{{\it The Polarization of the D$_{1}$ Line}}

1332: \label{sect:lineD1}

1333: %%

1334: \begin{figure}[!t]

1335: \begin{center}

1336: \includegraphics[angle=270, width=0.95\textwidth]{f14.eps}

1337: \caption{{\footnotesize Theoretical profiles of the emission coefficients

1338: $\varepsilon_{I}^{\,l}(\lambda)$ and $\varepsilon_{Q}^{\,l}(\lambda)$,

1339: and of the ratio

1340: $\varepsilon_{Q}^{\,l}(\lambda)/(\varepsilon_{I}^{\,l}(\lambda)+

1341: \varepsilon_{I}^{\, \rm c})$, of the Ba~{\sc ii} D$_1$ line, for the isotope 

1342: 138, for the isotope 137 and for all the seven isotopes together, in the 

1343: absence of a magnetic field. The emission profiles are

1344: normalized to the maximum value of the intensity emission profile,

1345: calculated taking into account the contribution of all the seven isotopes,

1346: $(\varepsilon_{I}^{l})_{\rm{max}}$.

1347: The vertical dashed lines show the positions of the three groups

1348: of transitions (see \S~\ref{sect:D2-zero-field}).

1349: The last row shows the

1350: $\varepsilon_{Q}^{\,l}(\lambda)/(\varepsilon_{I}^{\,l}(\lambda)+

1351: \varepsilon_{I}^{\, \rm c})$ profiles without continuum (solid), and

1352: with a continuum $\varepsilon_{I}^{\, \rm c}/(\varepsilon_{I}^{l})_{\rm{max}}$ 

1353: of $9\times 10^{-5}$ (dot).}}

1354: \label{fig:D1-zero-field}

1355: \end{center}

1356: \end{figure}

1357: To understand the physical origin of the polarization signal observed 

1358: in the D$_{1}$ line and, in particular, the role of HFS, we start 

1359: our investigation considering separately the

1360: isotopes 138 (without HFS) and 137 (with HFS), before taking into 

1361: account the contribution coming from all the seven isotopes together.

1362: We use the values of the average number of photons and of the anisotropy factor 

1363: given in \S~\ref{sect:radiation}, and a Doppler width of 30~m{\AA}. 

1364: In Figure~\ref{fig:D1-zero-field} we can see that the emission 

1365: coefficient $\varepsilon_{Q}^{\, l}$ of the isotopes without HFS 

1366: is constant in wavelength and equal to zero, while 

1367: it shows an anti-symmetrical profile, that goes rapidly to zero moving away 

1368: from the line-core, in the isotopes with HFS. For these isotopes the 

1369: frequency integrated Stokes $Q$ emission coefficient is equal to 

1370: zero\footnote{All these properties can be derived analytically, and are 

1371: briefly discussed in Appendix~\ref{app:two-level}.}. 

1372: The profile of the ratio $\varepsilon_{Q}^{\, l}/\varepsilon_{I}^{\,l}$ of 

1373: the isotopes with HFS, contrary to the case of the D$_{2}$ line, goes slowly 

1374: to zero moving away from the line-core.

1375: For this reason the effect of the continuum is less important in this line, as 

1376: it just `pushes' more rapidly to zero the wings of this profile without

1377: modifying significantly its shape.

1378: In the following description of our investigation on the magnetic 

1379: sensitivity of this line we have neglected the contribution to the 

1380: intensity coming from the continuum.\\

1381: %%

1382: \begin{figure}[!t]

1383: \begin{center}

1384: \includegraphics[width=0.95\textwidth]{f15.eps}

1385: \caption{{\footnotesize Theoretical profiles for the ratio 

1386: $\varepsilon_{Q}^{\, l}/\varepsilon_{I}^{\, l}$ of the Ba~{\sc ii} D$_1$ line, 

1387: in the presence of a vertical, horizontal perpendicular to the line-of-sight, 

1388: longitudinal and microturbulent magnetic field of various intensities.}}

1389: \label{fig:D1-magnetic-fields}

1390: \end{center}

1391: \end{figure}

1392: %%

1393: \indent As seen in Figure~\ref{fig:D1-zero-field}, within 

1394: the framework of our modeling approach for weak magnetic fields 

1395: ($B\! < 50\!$ G) it is not possible to obtain the symmetric 

1396: $Q(\lambda)/I(\lambda)$ profile observed by \citet{Ste00}. The 

1397: theoretical profile that we have obtained has no evident symmetries, and its 

1398: main peak does not coincide in wavelength with the central peak of the 

1399: observed profile. Changing the Doppler width and the anisotropy factor we can 

1400: modify the width and the amplitude of the peaks, adding a continuum 

1401: contribution to the $Q$ Stokes parameter we can shift the profile along the 

1402: polarization scale in order to get values of $Q/I$ in the wings 

1403: close to the observed ones, but we cannot really modify the shape of the profile.

1404: In Figure~\ref{fig:D1-magnetic-fields} we show the theoretical 

1405: $\varepsilon_{Q}^{\, l}/\varepsilon_{I}^{\, l}$ profiles in the presence of a 

1406: vertical, horizontal perpendicular to the line-of-sight, longitudinal and 

1407: microturbulent magnetic field.

1408: In complete analogy with the D$_{2}$ line, the profiles are modified by the 

1409: combined action of the Hanle and Zeeman effects.

1410: As already pointed out by \citet{jtb02} and by \citet{Cas02}

1411: for the Na~{$\!$\sc i} D$_1$ line case, it appears that, at this level of 

1412: approximation, the only way to get a symmetric profile, centered at the 

1413: wavelength position of the central peak of the observed profile 

1414: \citep[see][]{Ste00}, 

1415: is to be in the presence of a magnetic field strong enough to enter the 

1416: transverse Zeeman effect. 

1417: However, as expected, the transverse Zeeman effect produces in the 

1418: $\varepsilon_{Q}^{\, l}/\varepsilon_{I}^{\, l}$ profile two wing lobes that 

1419: are more significant than the central one, which is not the case of the 

1420: observed profile. 

1421: 

1422: \section{CONCLUSIONS}

1423: \label{sect:conclusions}

1424: 

1425: The most interesting general conclusion of

1426: our theoretical investigation on the magnetic sensitivity 

1427: of the D-lines of Ba~{$\!$\sc ii} is that

1428: the observation and modeling of the Hanle and Zeeman effects in these resonance lines provide a novel

1429: diagnostic tool for mapping the magnetic fields of the

1430: upper photosphere and lower chromosphere.\\

1431: \indent In particular, the Ba~{$\!$\sc ii} D$_2$ line at 4554~{\AA} is

1432: particularly interesting because the emergent linear polarization has

1433: contributions from different isotopes, contributions that are easily 

1434: resolved and have a different behavior in the presence of a magnetic field. 

1435: As a result, there is a

1436: differential magnetic sensitivity of the emergent linear polarization at 

1437: line center (where the signal is produced by the even isotopes without 

1438: HFS) with respect to the line wings (where the signals are produced by

1439: the odd isotopes with HFS). 

1440: For instance, for the case of a vertical magnetic field

1441: with a strength between 10~G and 100~G, approximately, only the isotopes

1442: with HFS are sensitive to the Hanle effect, which produce an enhancement of

1443: the scattering polarization at the two $Q/I$ wing wavelengths. For the case of

1444: a horizontal field between about 1 and 100~G

1445: the most conspicuous observable effect is the line core depolarization

1446: produced by the Hanle effect of the barium isotopes devoid of HFS.

1447: In both cases, the transverse Zeeman effect begins to play an increasingly

1448: dominating role for field intensities larger than 100~G, approximately.

1449: Useful information on the magnetic sensitivity of the $Q/I$ profile of the

1450: calculated emergent radiation in the D$_2$ line

1451: can be seen in Figure~{\ref{fig:D2-random-azimuth}, which corresponds to the 

1452: case of a random azimuth magnetic field with a fixed inclination.

1453: Of particular interest is the case of an unimodal microturbulent and isotropic 

1454: magnetic field (see Figure~{\ref{fig:D2-microturbulent}}), for which there is 

1455: no contribution from the Zeeman effect and Stokes $U$ and $V$ are zero.\\

1456: \indent Concerning the enigmatic Ba~{$\!$\sc ii} D$_1$ line it is important to

1457: note that in the absence of magnetic fields only the $18\%$ isotopes with

1458: HFS are capable of producing linear polarization through bound-bound

1459: transitions. As with the sodium D$_1$ line, this is possible thanks to the

1460: fact that in the absence of depolarizing mechanisms

1461: only the upper and lower levels of the D$_1$ line 

1462: transition in the odd isotopes are significantly

1463: polarized. Interestingly, a $Q/I$ profile with a conspicuous blue-shifted 

1464: peak is obtained if only the selective emission of

1465: polarization components that results from the upper-level polarization are 

1466: taken into account (see the last panel of the third row of 

1467: Fig.~\ref{fig:D1-zero-field})\footnote{Note, however, that in the absence of a 

1468: magnetic field the theoretical $Q/I^{max}$ profile of the Ba~{$\!$\sc ii} 

1469: D$_1$ line (see the last panel in the second row of 

1470: Fig.~{\ref{fig:D1-zero-field}}) has an antisymmetrical shape, as already found 

1471: for the Na~{$\!$\sc i} D$_1$ line \citep[see Fig.~2 in][]{jtb02}, when only 

1472: selective emission processes are considered.}. 

1473: Under such circumstances one could argue that a detailed radiative

1474: transfer solution for the Ba~{$\!$\sc ii} D$_1$ line including the Doppler

1475: shifts caused by the convective motions and waves that are present in the

1476: solar atmospheric plasma could perhaps produce a symmetric $Q/I$ profile

1477: for the Ba~{$\!$\sc ii} D$_1$ line, as observed by Stenflo et al. (2000).

1478: However, as shown in this paper, for the Ba~{$\!$\sc ii} D$_1$ line we should

1479: expect also a significant contribution from ``zero-field" dichroism --that

1480: is, from the selective absorption of polarization components that results

1481: from the lower-level polarization. In fact, when both selective emission

1482: and absorption processes are taken into account through the approximation

1483: of equation~(\ref{eq:dichroism}), we then obtain the nearly 

1484: antisymmetric $Q/I$ profile of Figure~\ref{fig:D1-dichroism}.

1485: Note that there is no possibility of destroying the lower level polarization

1486: without simultaneously destroying the atomic polarization of the upper level

1487: of the D$_1$ line \citep[see Fig.~1 of][and our Fig.~\ref{fig:sigma20} 

1488: for barium levels]{jtb02}. As far as dichroism is neglected, within the 

1489: framework of our present modeling assumptions one might then be tempted to conclude that the only possibility of obtaining a symmetric $Q/I$ peak for the 

1490: Ba~{\sc ii} D$_1$ line is via the transverse

1491: Zeeman effect, even though, through this kind of mechanism, a profile with wing 

1492: lobes more significant then the central one is obtained, and magnetic fields quite intense are needed. 

1493: Our approach neglects, however, the radiative transfer

1494: effects that we certainly have in the real solar atmosphere, with its

1495: vertical stratification, horizontal inhomogeneities and the Doppler shifts

1496: caused by the above-mentioned upflows, downflows and waves.

1497: Therefore, detailed radiative transfer simulations using realistic solar

1498: atmospheric models are 

1499: urgently needed in order to be able to conclude

1500: whether a symmetric $Q/I$ profile for the Ba~{$\!$\sc ii} D$_1$ line with a significant line-center peak may be

1501: obtained within the framework of the density matrix theory we have applied

1502: in this paper, either because of the influence of the atomic level

1503: polarization of the $18\%$ of the barium isotopes endowed of HFS (which would

1504: require the presence of atmospheric regions with very weak fields), or due

1505: to the transverse Zeeman effect of all the barium isotopes (which would

1506: require the presence of a sufficiently strong magnetic field, and the 

1507: absence of significant saturation effects at the line-center wavelength).\\

1508: \indent Finally, we would like to finish this paper by emphasizing the

1509: importance of pursuing high-spatial resolution polarimetric observations

1510: of the Ba~{$\!$\sc ii} D$_2$ line (e.g., via Fabry-Perot polarimetry) in

1511: order to help decipher the spatial and temporal fluctuations of the magnetic

1512: field vector, in both active and quiet regions of the solar atmosphere.

1513: 

1514: 

1515: \acknowledgments

1516: This research has been partially funded by the European Commission through 

1517: the Solar Magnetism Network, and by the Spanish Ministerio de Educaci\'on y 

1518: Ciencia through project AYA2004-05792.

1519: 

1520: \appendix

1521: 

1522: \section{RESULTS OBTAINED TAKING INTO ACCOUNT DICHROISM EFFECTS}

1523: \label{app:dichroism}

1524: %%

1525: \begin{figure}[!t]

1526: \begin{center}

1527: \includegraphics[angle=270,width=0.85\textwidth]{f16.eps}

1528: \caption{{\footnotesize {\it Left column:} theoretical profiles for the 

1529: ratios (from the top) $\varepsilon_{Q}^{\, l}/\varepsilon_{I}^{\, l}$, 

1530: $\eta_{Q}^{\, l}/\eta_{I}^{\, l}$ and for their difference, in the presence 

1531: of a vertical magnetic field of various intensities. {\it Right column:} line 

1532: profiles taking into account the continuum contribution to the intensity 

1533: emission and absorption 

1534: ($\varepsilon_{I}^{\, \rm c}/(\varepsilon_{I}^{l})_{\rm{max}}=9 \times 10^{-5}$,

1535: $\eta_{I}^{\, \rm c}$ calculated through eq.~[\ref{eq:eta-cont}]). 

1536: Calculations refer to the Ba~{\sc ii} D$_{2}$ line.}}

1537: \label{fig:D2-dichroismA}

1538: \end{center}

1539: \end{figure}

1540: %%

1541: We show here the results obtained taking into account the absorption effects in 

1542: the D$_{2}$ line. Including the contribution of the continuum, 

1543: equation~(\ref{eq:dichroism}) takes the form

1544: \begin{equation}

1545: \frac{X(\nu,\mathbf{\Omega})}{I(\nu,\mathbf{\Omega})}\approx

1546: \frac{\varepsilon_{X}^{\, l}(\nu,\mathbf{\Omega})}

1547: {\varepsilon_{I}^{\, l}(\nu,\mathbf{\Omega})+

1548: \varepsilon^{\,\rm c}_{I}}-

1549: \frac{\eta_{X}^{\, l}(\nu,\mathbf{\Omega})}

1550: {\eta_{I}^{\, l}(\nu,\mathbf{\Omega})+

1551: \eta^{\,\rm c}_{I}} \;\; ,

1552: \label{eq:dichroism-cont}

1553: \end{equation}

1554: where $\eta_{I}^{\rm c}$ is the continuum contribution to the

1555: total intensity absorption.

1556: This quantity can be calculated from the continuum intensity emission 

1557: coefficient through the relation

1558: \begin{equation}

1559: \eta^{\,\rm c}_{I}=\frac{\varepsilon^{\,\rm c}_{I}}{B_{\rm{P}}

1560: (\nu_{0},T=5800K)} \;\; ,

1561: \label{eq:eta-cont}

1562: \end{equation}

1563: where $B_{\rm{P}}(\nu_{0},T=5800K)$ is the Planck function calculated at the

1564: central wavelength ($\nu_0$) of the line we are investigating and at the

1565: effective solar temperature.

1566: Figures~\ref{fig:D2-dichroismA} and \ref{fig:D2-dichroismB} show the results 

1567: for the D$_{2}$ line. We consider first the results obtained without 

1568: continuum.

1569: %%

1570: \begin{figure}[!t]

1571: \begin{center}

1572: \includegraphics[angle=270,width=0.85\textwidth]{f17.eps}

1573: \caption{{\footnotesize Same as Figure~\ref{fig:D2-dichroismA} for larger 

1574: values of the magnetic field.}} 

1575: \label{fig:D2-dichroismB}

1576: \end{center}

1577: \end{figure}

1578: %%

1579: In the absence of magnetic fields, since the atomic polarization in the lower 

1580: level is much smaller than in the upper level of D$_{2}$ (see 

1581: Fig.~\ref{fig:sigma20}), the absorption effects are completely negligible, as 

1582: already observed in \S~\ref{sect:emergent}. 

1583: Introducing a vertical magnetic field the profile of 

1584: the ratio $\eta_{Q}^{\, l}/\eta_{I}^{\, l}$ is modified by the 

1585: transverse Zeeman effect and, as the magnetic field is increased, it assumes a 

1586: shape which is very similar to the one observed for

1587: $\varepsilon_{Q}^{\, l}/\varepsilon_{I}^{\, l}$ 

1588: (see Fig.~\ref{fig:D2-dichroismB}). For this reason, the profile 

1589: obtained through equation~(\ref{eq:dichroism}) does not show any detail 

1590: due to the transverse Zeeman effect 

1591: since the contribution coming from emission and absorption cancel out. 

1592: The situation is somewhat different in the presence of the continuum. 

1593: In this case, as shown in Figure~\ref{fig:D2-dichroismB}, for magnetic fields 

1594: stronger than about 100~G the contribution coming from the absorption term 

1595: becomes more important. 

1596: At this point it is important to remember that the continuum has been 

1597: considered 

1598: as a parameter in our investigation and, as stressed in 

1599: \S~\ref{sect:D2-zero-field}, we have to be careful when dealing with spectral 

1600: details that find their origin in this physical aspect of the problem.

1601: Similar considerations can be done for the other magnetic field geometries 

1602: considered in this work.\\

1603: %%

1604: \begin{figure}[t]

1605: \begin{center}

1606: \includegraphics[angle=270,width=0.4\textwidth]{f18.eps}

1607: \caption{{\footnotesize Theoretical $Q/I$ profile of the Ba~{\sc ii} D$_{1}$ 

1608: line calculated according to equation~\ref{eq:dichroism}, in the absence of 

1609: magnetic fields and without any contribution of the continuum.

1610: For the Doppler width, the anisotropy factor, and the average number of photons 

1611: have been used the same values of \S~\ref{sect:lineD1}.}}

1612: \label{fig:D1-dichroism}

1613: \end{center}

1614: \end{figure}

1615: %%

1616: \indent Figure~\ref{fig:D1-dichroism} shows the ratio $(\varepsilon_{Q}^{\, l}/

1617: \varepsilon_{I}^{\, l})-(\eta_{Q}^{\, l}/\eta_{I}^{\, l})$ for the  D$_{1}$ 

1618: line in the absence of magnetic fields. As expected, since the atomic 

1619: polarization is quite similar in the upper and lower levels of the D$_{1}$, 

1620: dichroism is much more important in this line, and the profiles obtained 

1621: through equations~(\ref{eq:fract-polar1}) and (\ref{eq:dichroism}) are quite

1622: different from each other. However, even taking into account dichroism, we 

1623: are not able to reproduce the observed profile.

1624:  

1625: 

1626: \section{ANALYTICAL RESULTS FOR A TWO-LEVEL ATOM APPLIED TO THE Ba~{$\!$\sc ii} 

1627: D LINES}

1628: \label{app:two-level}

1629: In this work we have described the Ba~{$\!$\sc ii} ion through a three 

1630: level model atom that allowed us to study both the D$_{1}$ and D$_{2}$ lines; 

1631: we have taken into account the ground level polarization, the stimulated 

1632: emission effects and, finally, the effects of a magnetic field.

1633: At this level a numerical approach of the problem is absolutely necessary.

1634: However, as described in detail in \S~10 of LL04, by introducing some 

1635: simplifying approximations it is possible to obtain analytical expressions 

1636: for the atomic density-matrix elements and, through these, for the radiation 

1637: transfer coefficients of the atomic system. 

1638: These analytical expressions are very useful in order to understand 

1639: the physics of the phenomenon under investigation, which could remain quite 

1640: hidden within a numerical approach.

1641: The basic approximation is to consider a two-level atom. 

1642: The only difference between our three-level model atom and a two-level atom 

1643: lies in the fact that in the SEEs of our model the ground level feels the 

1644: effect of both transitions towards the two upper levels considered. 

1645: However, the numerical solution of the SEEs showed that the upper 

1646: level of the D$_{2}$ line is much more polarized than the ground level,

1647: so that, as far as the D$_{2}$ line is considered, the polarization of the 

1648: ground level, as a first approximation, can be neglected and our atomic 

1649: system can be treated as a two-level atom. 

1650: We can therefore apply the equations of \S~10 of 

1651: LL04 to calculate the upper level density-matrix elements or the 

1652: emission coefficients of the various Ba~{$\!$\sc ii} isotopes, described as 

1653: two-level atoms with HFS, and two-level atoms without HFS, with unpolarized 

1654: lower level.

1655: Neglecting the stimulated emission effects (which is a good approximation 

1656: whenever the incident radiation field, as in our case, is weak), and in 

1657: the absence of the magnetic field, the emission coefficient of a two-level 

1658: atom without HFS is (see eq.[10.16] of LL04)

1659: %

1660: \begin{eqnarray}

1661: \varepsilon_{i}(\nu,\mathbf{\Omega}) &\! =\! & 

1662: \frac{h\nu}{4\pi}\mathcal{N}_{\ell}

1663: B(\alpha_{\ell}J_{\ell}\rightarrow \alpha_{u}J_{u})\phi(\nu_{0}-\nu) \nonumber\\

1664: & & \times \sum_{KQ}W_{K}(J_{\ell},J_{u})(-1)^{Q}

1665: \mathcal{T}_{Q}^{K}(i,\mathbf{\Omega})J_{-Q}^{K}(\nu_{0}) \;\; ,

1666: \label{eq:eps-two-level}

1667: \end{eqnarray}

1668: %

1669: where $\mathcal{N}_{\ell}$ is the number density of atoms in the ground 

1670: level\footnote{Note that for a two level atom 

1671: $\mathcal{N}=\mathcal{N}_{\ell}+\mathcal{N}_{u}$ while, for our model atom,

1672: $\mathcal{N}=\mathcal{N}_{\ell}+\mathcal{N}_{u}^{D_{2}}+

1673: \mathcal{N}_{u}^{D_{1}}$. As far as lower level polarization is neglected, 

1674: this is the only difference between a two-level atom and our model atom.}, 

1675: and where 

1676: %

1677: \begin{equation}

1678: W_{K}(J_{\ell},J_{u})=3(2J_{u}+1)

1679: \Bigg\{ \begin{array}{ccc}

1680: 1 & 1 & K \\

1681: J_{u} & J_{u} & J_{\ell}

1682: \end{array} \Bigg\}^{2} \;\; .

1683: \end{equation}

1684: %

1685: For $90^{\circ}$ scattering of a radiation field with cylindrical 

1686: symmetry around the direction of propagation (as in our case), from 

1687: equation~(\ref{eq:eps-two-level}) we obtain

1688: %

1689: \begin{equation}

1690: p_{Q}\equiv \frac{\varepsilon_{Q}}{\varepsilon_{I}}=

1691: \frac{3W_{2}(1/2,3/2)}{4/w-W_{2}(1/2,3/2)} \;\; .

1692: \label{eq:fract-pol-two-level}

1693: \end{equation}

1694: %

1695: This expression shows that for the isotopes without HFS the fractional 

1696: polarization $p_{Q}$ does not depend on frequency, as found in 

1697: \S~\ref{sect:D2-zero-field} (Fig.~\ref{fig:D2-zero-field}).

1698: Substituting the numerical values of the various quantities ($w=0.037$ and 

1699: $W_{2}(1/2,3/2)=0.5$) we obtain the value of 1.4\%, as found with numerical 

1700: calculations.

1701: The expression of the emission coefficient of a two-level atom with HFS 

1702: (eq.~[10.166] of LL04) is much 

1703: more complicated and will not be written here. Anyway, as shown in

1704: \S~10.22 of LL04, at frequencies very distant from the `center 

1705: of gravity', the multiplet behaves in resonance scattering as a simple 

1706: transition between two levels without HFS. The fractional polarization, 

1707: therefore, at these frequencies is still described by 

1708: equation~(\ref{eq:fract-pol-two-level}).

1709: This result justifies the fact that at large distances from the line center 

1710: the fractional polarization of the isotopes with HFS reaches the same value 

1711: as the isotopes without HFS. 

1712: It is important to stress that this asymptotic behaviour of the 

1713: isotopes with HFS is strongly dependent on the interferences between the 

1714: various HFS magnetic sublevels of the D$_{2}$ upper level.

1715: Because of the small frequency distance between the various components of the 

1716: HFS multiplet with respect to their natural width, the analytical expression 

1717: of the emission coefficients of the isotopes with HFS cannot be simplified in 

1718: the neighbourhood of the various transitions. Nevertheless, we can 

1719: qualitatively 

1720: justify the decrease of the fractional polarization at the wavelength positions 

1721: of the various components of the HFS multiplet by considering the frequency 

1722: integrated emission coefficients of the isotopes with HFS

1723: %

1724: \begin{eqnarray}

1725: \tilde{\varepsilon}_{i}(\mathbf{\Omega}) & \! = \! & \int_{\Delta \nu} 

1726: \varepsilon_i(\nu,\mathbf{\Omega}) {\rm d}\nu \nonumber \\

1727: & = \! & \frac{h\nu}{4\pi}\,\mathcal{N}_{\ell}\,

1728: B(\alpha_{\ell}J_{\ell}\rightarrow \alpha_{u}J_{u})

1729: \sum_{KQ}\,[W_{K}(\alpha_{\ell}J_{\ell}I,\alpha_{u}J_{u})]_{\rm hfs}\,(-1)^{Q}\,

1730: \mathcal{T}_{Q}^{K}(i,\mathbf{\Omega})J_{-Q}^{K}(\nu_{0}) \;\; ,

1731: \end{eqnarray}

1732: %

1733: where the interval $\Delta \nu$ is sufficiently broad to fully cover all the 

1734: Zeeman components of the line, and where

1735: %

1736: \begin{equation}

1737: [W_{K}(\alpha_{\ell}J_{\ell}I,\alpha_{u}J_{u})]_{\rm hfs}=W_{K}(J_{\ell},J_{u})

1738: [D_{K}(\alpha_{u}J_{u}I)]_{\rm hfs} \;\; .

1739: \end{equation}

1740: %

1741: The quantity $[D_{K}(\alpha_{u}J_{u}I)]_{\rm hfs}$ is the depolarizing factor 

1742: due to HFS and it is given by

1743: %

1744: \begin{eqnarray}

1745: [D_{K}(\alpha JI)]_{\rm hfs} & = & \frac{1}{(2I+1)}\sum_{FF^{\prime}}(2F+1)

1746: (2F^{\prime}+1) 

1747: \Bigg\{ \begin{array}{ccc}

1748: J & J & K \\

1749: F & F^{\prime} & I

1750: \end{array} \Bigg\}^{2} \nonumber \\

1751: & & \times \frac{1}{1+2\pi {\rm i} \nu_{\alpha J I F^{\prime},\alpha J IF}/

1752: A(\alpha J \rightarrow \alpha_{\ell} J_{\ell})} \;\; .

1753: \label{eq:hfs-depolar-fact}

1754: \end{eqnarray}

1755: %

1756: For the isotope 137 the depolarizing factor 

1757: $[D_{2}(J\!\!=\!\!3/2,I\!\!=\!\!3/2)]_{\rm hfs}$ is equal to 0.27,

1758: and the frequency integrated fractional polarization, that in complete analogy 

1759: with equation~(\ref{eq:fract-pol-two-level}) has the form

1760: %

1761: \begin{equation}

1762: \tilde{p}_{Q}\equiv\frac{\tilde{\varepsilon}_{Q}}{\tilde{\varepsilon}_{I}}=

1763: \frac{3W_{2}(J_{\ell},J_{u})[D_{K}(\alpha_{u}J_{u}I)]_{hfs}}{4/w-

1764: W_{2}(J_{\ell},J_{u})[D_{K}(\alpha_{u}J_{u}I)]_{hfs}} \;\; ,

1765: \label{eq:hfs-freq-int-fract-pol}

1766: \end{equation}

1767: %

1768: is equal to 0.0038. Comparing this value with 0.014, the value 

1769: of $p_{Q}$ previously found for the isotopes without HFS (note that for these 

1770: isotopes $p_{Q}$ is equal to $\tilde{p}_{Q}$), we can clearly see 

1771: the depolarizing effect of the HFS.\\

1772: %%

1773: \begin{figure}[!t]

1774: \begin{center}

1775: \includegraphics[angle=270,width=0.4\textwidth]{f19.eps}

1776: \caption{{\footnotesize Frequency integrated fractional polarization for the

1777: isotope 137 in a 90$^{\circ}$ scattering event in the presence of a vertical 

1778: magnetic field, as function of the magnetic field strength. Calculations 

1779: refer to the Ba~{\sc ii} D$_2$ line.}}

1780: \label{fig:wkkpqb}

1781: \end{center}

1782: \end{figure}

1783: %%

1784: \indent For the D$_{1}$ line the quantity $W_{2}(1/2,1/2)$ is zero and, 

1785: taking into account equations~(\ref{eq:eps-two-level}) and 

1786: (\ref{eq:hfs-freq-int-fract-pol}), 

1787: it is easy to see that the Stokes $Q$ emission coefficient of the isotopes 

1788: without HFS has to be constant and equal to zero, while the frequency integrated

1789: Stokes $Q$ emission coefficient of the isotopes with HFS has to be zero, as 

1790: found with numerical calculations in \S~\ref{sect:lineD1}.

1791: However, as previously said, for the case of the D$_{1}$ line the approximation 

1792: of unpolarized lower level is not good anymore. Some analytical results that 

1793: it is possible to obtain taking into account the polarization of the lower 

1794: level are derived in LL04.\\

1795: \indent In the presence of a vertical magnetic field 

1796: equation~(\ref{eq:hfs-freq-int-fract-pol}) generalizes into

1797: %

1798: \begin{equation}

1799: \tilde{p}_{Q}\equiv\frac{\tilde{\varepsilon}_{Q}}{\tilde{\varepsilon}_{I}}=

1800: \frac{3[W_{220}(\alpha_{\ell}J_{\ell}I\alpha_{u}J_{u};B)]_{\rm hfs}}{4/w-

1801: [W_{220}(\alpha_{\ell}J_{\ell}I\alpha_{u}J_{u};B)]_{\rm hfs}} \;\; ,

1802: \label{eq:magnetic-hfs-freq-int-fract-pol}

1803: \end{equation}

1804: %

1805: where the general expression of the quantity 

1806: $[W_{KK^{\prime}Q}(\alpha_{\ell}J_{\ell}I\alpha_{u}J_{u};B)]_{\rm hfs}$ 

1807: is given in LL04 (eq.~[10.167]).

1808: Figure~\ref{fig:wkkpqb} shows that there is an increase of $\tilde{p}_{Q}$ for 

1809: vertical magnetic fields ranging between 0 and 500~G 

1810: (see Fig.~\ref{fig:D2-vertical-field}).

1811: 

1812: 

1813: \section{AVERAGE OF THE EMISSION COEFFICIENTS OVER THE MAGNETIC FIELD 

1814: DIRECTIONS}

1815: \label{app:average}

1816: The only quantity that depends on the magnetic field orientation in the 

1817: general expression of the emission coefficients (\S~\ref{sect:equations}, 

1818: eq.~[\ref{eq:epsilon}]) is the product 

1819: \begin{equation}

1820: \Big[{\mathcal{T}}_{Q}^{K}(j,\mathbf{\Omega})\Big]_{B}

1821: \Big[{}^{\alpha_{u}J_{u}I}\!\rho_{Q_{u}}^{K_{u}}(F_{u}^{\prime},F_{u}^

1822: {\prime\prime})\Big]_{B} \; \; ,

1823: \label{eq:magnetic-ref}

1824: \end{equation}

1825: where the label $B$ means that the quantity is calculated in the 

1826: magnetic field reference system.

1827: We already observed (\S~\ref{sect:polarization}) that the spherical statistical

1828: tensors calculated in the magnetic reference system, because of the symmetry

1829: of the problem, do not depend on the azimuth of the magnetic field, $\chi_{B}$,

1830: but only on its inclination with respect to the local vertical, $\theta_{B}$.

1831: For this reason all the dependence on $\chi_{B}$ is included into the 

1832: geometrical tensor $\big[{\mathcal{T}}_{Q}^{K}(j,\mathbf{\Omega})\big]_{B}$.

1833: The expression of the tensor $\mathcal{T}_{Q}^{K}$ in terms of rotation 

1834: matrices is given by (eq.~[5.159] of LL04)

1835: \begin{equation}

1836: \Big[{\mathcal{T}}_{Q}^{K}(j,\mathbf{\Omega})\Big]_{B}=

1837: \sum_{P}t^{K}_{P}(j)\mathcal{D}^{K}_{PQ}(R_0) \;\; ,

1838: \label{eq:tauKQ}

1839: \end{equation}

1840: where $\mathcal{D}$ is the rotation matrix, $R_0$ is the rotation bringing 

1841: the reference system $(\mathbf{e}_{\it{a}}(\mathbf{\Omega}),\mathbf{e}_{\it{b}}

1842: (\mathbf{\Omega}),\mathbf{\Omega})$, with 

1843: $\mathbf{e}_{\it{a}}(\mathbf{\Omega})$ the 

1844: reference direction, into the reference system with the $z$ axis 

1845: directed along the magnetic field, and where $t^{K}_{P}(j)$ is a scalar 

1846: quantity that does not depend on the particular geometry of the problem 

1847: (cf.~eq.~[5.160] of LL04).

1848: The relation between the geometrical tensor 

1849: $\mathcal{T}_{Q}^{K}$ calculated in the magnetic field reference system, and 

1850: the same quantity calculated in the local vertical reference system 

1851: (with the $z$ axis directed along the local vertical) is

1852: \begin{equation}

1853: \Big[{\mathcal{T}}_{Q}^{K}(j,\mathbf{\Omega})\Big]_{B}=

1854: \sum_{A}\Big[{\mathcal{T}}_{A}^{K}(j,\mathbf{\Omega})\Big]_{V}

1855: \mathcal{D}^{K}_{AQ}(R_1) \;\; ,

1856: \label{eq:tauKQ-rotat}

1857: \end{equation}

1858: where the label $V$ means that the corresponding quantity is calculated into

1859: the local vertical reference system, and

1860: where $R_1$ is the rotation bringing the local vertical reference system into 

1861: the magnetic field reference system. In the geometry of our problem the 

1862: rotation $R_1$ is defined by the Euler angles $(\chi_B,\theta_B,0)$.

1863: Obviously, because of its definition, 

1864: $\big[{\mathcal{T}}_{A}^{K}(j,\mathbf{\Omega})\big]_{V}$ does not 

1865: depend on the magnetic field orientation: in 

1866: analogy with equation~(\ref{eq:tauKQ}) we can write

1867: \begin{equation}

1868: \Big[{\mathcal{T}}_{A}^{K}(j,\mathbf{\Omega})\Big]_{V}=

1869: \sum_{P}t^{K}_{P}(j)\mathcal{D}^{K}_{PA}(R_2) \;\; ,

1870: \end{equation}

1871: where $R_2$ is the rotation that brings the reference system 

1872: $(\mathbf{e}_{\it{a}}(\mathbf{\Omega}),\mathbf{e}_{\it{b}}(\mathbf{\Omega}),

1873: \mathbf{\Omega})$ into the local vertical reference system.

1874: Substituting equation~(\ref{eq:tauKQ-rotat}) into the expression 

1875: (\ref{eq:magnetic-ref}) we obtain

1876: \begin{equation}

1877: \Big[{\mathcal{T}}_{Q}^{K}(j,\mathbf{\Omega})\Big]_{B}

1878: \Big[{}^{\alpha_{u}J_{u}I}\!\rho_{Q_{u}}^{K_{u}}(F_{u}^{\prime},F_{u}^

1879: {\prime\prime})\Big]_{B}=\sum_{A}\Big[{\mathcal{T}}_{A}^{K}

1880: (j,\mathbf{\Omega})\Big]_{V}\mathcal{D}^{K}_{AQ}(R_1)

1881: \Big[{}^{\alpha_{u}J_{u}I}\!\rho_{Q_{u}}^{K_{u}}(F_{u}^{\prime},F_{u}^

1882: {\prime\prime})\Big]_{B} \;\; ,

1883: \end{equation}

1884: where we see that all the dependence on $\chi_B$ is included in the rotation

1885: matrix $\mathcal{D}^{K}_{AQ}(R_1)$.

1886: Recalling the expression of the rotation matrices in terms of the reduced 

1887: rotation matrices (cf. eq.~[2.68] of LL04), and recalling the Euler 

1888: angles corresponding to the rotation $R_1$, we have

1889: \begin{equation}

1890: \mathcal{D}^{K}_{AQ}(R_1)={\rm{e}}^{-{\rm{i}}A\chi_{B}}d^{K}_{AQ}(\theta_{B})

1891: \;\; ,

1892: \end{equation}

1893: where $d^{K}_{AQ}$ is the reduced rotation matrix, which

1894: depends only on the second Euler angle of the rotation, in our 

1895: case the inclination of the magnetic field.

1896: Averaging on $\chi_{B}$ reduces therefore to calculate the integral

1897: \begin{equation}

1898: \frac{1}{2\pi}\int_{0}^{2\pi}{\rm{e}}^{-{\rm{i}}A\chi_{B}}{\rm{d}}\chi_{B} \;\;.

1899: \end{equation}

1900: It is easy to see that the integral is different from zero only if $A=0$, and 

1901: that, in this case, it is equal to 1.

1902: Averaging the emission coefficients over $\chi_{B}$, the magnetic field 

1903: azimuth, is therefore equivalent to substitute expression 

1904: (\ref{eq:magnetic-ref}) with

1905: \begin{equation}

1906: \Big[{\mathcal{T}}_{0}^{K}(j,\mathbf{\Omega})\Big]_{V}

1907: d^{K}_{0Q}(\theta_B)

1908: \Big[{}^{\alpha_{u}J_{u}I}\!\rho_{Q_{u}}^{K_{u}}(F_{u}^{\prime},F_{u}^

1909: {\prime\prime})\Big]_{B} \;\; .

1910: \end{equation}

1911: Now all the quantities depend only on the magnetic field inclination, and we 

1912: can complete numerically (for example by a Gaussian quadrature)

1913: the average over the magnetic field orientation.

1914: 

1915: 

1916: \begin{thebibliography}{}

1917: \bibitem[Becker et al.(1981)]{Bec81}

1918: Becker, W., Blatt, R., \& Werth, G. 1981, J. Physique Coll., 42, 

1919: C8-339

1920: \bibitem[Bommier(1980)]{Bom80}

1921: Bommier, V. 1980, \aap, 87, 109

1922: \bibitem[Casini et al.(2002)]{Cas02}

1923: Casini, R., Landi Degl'Innocenti, E., Landolfi, M., \& Trujillo Bueno, J. 

1924: 2002, \apj, 573, 864

1925: \bibitem[Landi Degl'Innocenti \& Landolfi(2004)]{Libro}

1926: Landi Degl'Innocenti, E., \& Landolfi, M. 2004, Polarization in Spectral

1927: Lines (Dordrecht: Kluwer)

1928: \bibitem[Kopfermann(1958)]{Kop58}

1929: Kopfermann, H. 1958, Nuclear Moments (New York: Academic Press)

1930: \bibitem[Moore(1958)]{Moore}

1931: Moore, C. E. 1958, Atomic Energy Levels: as derived from the analyses of 

1932: optical spectra, Vol.III (Washington National Bureau of Standards)

1933: \bibitem[Pierce(2000)]{Allen}

1934: Pierce, K. 2000, in Allen's Astrophysical Quantities, ed. A. N. Cox

1935: (4th ed.; New York: Springer), 355 

1936: \bibitem[Stenflo(1997)]{Ste97b}

1937: Stenflo, J. O. 1997, \aap, 324, 344

1938: \bibitem[Stenflo \& Keller(1997)]{Ste97a}

1939: Stenflo, J. O., \& Keller, C. U. 1997, \aap, 321, 927

1940: \bibitem[Stenflo et al.(1998)]{Ste98}

1941: Stenflo, J. O., Keller, C. U., \& Gandorfer, A. 1998, \aap, 329, 319

1942: \bibitem[Stenflo et al.(2000)]{Ste00}

1943: Stenflo, J. O., Keller, C. U., \& Gandorfer, A. 2000, \aap, 335, 789

1944: \bibitem[Stenflo et al.(2002)]{Ste02}

1945: Stenflo, J. O., Gandorfer, A., Holzreuter, R., Gisler, D., Keller, C. U.,

1946: \& Bianda, M. 2002, \aap, 389, 314

1947: \bibitem[Stenflo(2003)]{Ste03}

1948: Stenflo, J.O. 2003, in ASP Conf. Ser. 307, Solar Polarization 3, 

1949: ed. J. Trujillo Bueno, \& J. S\'anchez Almeida (San Francisco: ASP), 385

1950: \bibitem[Trujillo Bueno(2003)]{jtb03}

1951: Trujillo Bueno, J. 2003, in ASP Conf. Ser. 307, Solar Polarization 3,

1952: ed. J. Trujillo Bueno, \& J. S\'anchez Almeida (San Francisco: ASP), 407

1953: \bibitem[Trujillo Bueno et al.(2002)]{jtb02}

1954: Trujillo Bueno, J., Casini, R., Landolfi, M., \& Landi Degl'Innocenti, E. 

1955: 2002, \apj, 566, L53

1956: \bibitem[Villemoes et al.(1993)]{Vil93}

1957: Villemoes, P., Arnesen, A., Heijkenskj\"old, F., \& W\"annstr\"om, A. 1993,

1958: J. Phys. B, 26, 4289

1959: \bibitem[Wendt et al.(1984)]{Wen84}

1960: Wendt, K., Ahmad, S. A., Buchinger, F., Mueller, A. C., Neugart, R., \&

1961: Otten, E. W. 1984, Z. Phys. A, 318, 125

1962: \bibitem[Wendt et al.(1988)]{Wen88}

1963: Wendt, K., Ahmad, S. A., Ekstr\"om, C., Klempt, W., Neugart, R., \& Otten, 

1964: E. W. 1988, Z. Phys. A, 329, 407

1965: \end{thebibliography}

1966: 

1967: 

1968: \end{document}

1969: