astro-ph0408360/ms.tex
1: \documentclass[12pt, preprint]{aastex}
2: %\documentclass[preprint2]{aastex}
3: 
4: %\usepackage{txfonts}
5: \usepackage{epsfig}
6: %\usepackage{amsmath}   % AMS enviroments
7: 
8: \begin{document}
9: 
10: \title{Concerning the existence of a `turbulent' magnetic field
11: in the quiet Sun}
12: 
13: %\titlerunning{Turbulent magnetic fields in the quiet Sun}
14: \shorttitle{Turbulent magnetic fields in the quiet Sun}
15: 
16: \author{Rafael Manso Sainz\altaffilmark{1,2}, 
17: Egidio Landi Degl'Innocenti\altaffilmark{3} \& Javier Trujillo Bueno\altaffilmark{2,5}}
18: \altaffiltext{1}{High Altitude Observatory, National Center for Atmospheric Research\altaffilmark{4}, P. O. Box 3000,
19: Boulder CO 80307-3000}
20: \altaffiltext{2}{Instituto de Astrof\'{\i}sica de Canarias, E-38205, La Laguna,
21: Tenerife, Spain}
22: \altaffiltext{3}{Dipartimento di Astronomia e Scienza dello Spazio, Largo Enrico Fermi 2,
23: I-50125, Firenze, Italy.}
24: \altaffiltext{4}{The National Center for Atmospheric Research is sponsored by the National Science Foundation}
25: \altaffiltext{5}{Consejo Superior de Investigaciones Cient\'{\i}ficas (Spain)}
26: 
27: %\date{Received / Accepted}
28: 
29: \begin{abstract}
30: 
31: 	We report on the $a\,{}^5\!F$-$y\,{}^5\!F^\circ$ 
32: 	multiplet of Ti {\sc i} and its interest for the study of 
33: 	``turbulent'' magnetic fields in the quiet solar photosphere.
34: 	In particular, we argue that the sizable scattering polarization
35: 	signal    of the 4536 \AA\ line (whose lower and upper levels
36: 	have Land\'e factors equal to zero),
37: 	relative to the rest of the lines in the multiplet, gives
38: 	{\em direct} evidence for the existence of 
39: 	a ubiquitous, unresolved magnetic field.
40: 	We cannot determine precisely the strength
41: 	of the magnetic field,
42: 	but its very existence is evidenced 
43: 	by the differential Hanle effect technique that this 
44: 	Ti {\sc i} multiplet provides.
45: \end{abstract}
46: 
47: %
48: \keywords{Polarization --- Scattering --- Line: formation 
49: --- Sun: magnetic fields --- Sun: atmosphere}
50: 
51: Magnetic fields in the solar atmosphere
52: may be detected most clearly by the 
53: circular polarization pattern they induce in spectral lines. 
54: However, this Zeeman effect technique is not very suitable 
55: for investigating magnetic fields that
56: have complex unresolved geometries because the
57: contributions of opposite magnetic polarities within the spatio-temporal resolution element of the observation tend to cancel out.
58: Consequently, extremely weak magnetic fields or even strong fields with a rather convoluted topology may remain virtually undetectable.
59: 
60: This is unfortunate since, according to our general picture of
61: solar magnetism, most regions
62: of the solar atmosphere are expected to be filled by magnetic fields
63: whose geometries are complex and far from 
64: being spatially resolved.
65: In particular, `turbulent' magnetic fields driven by highly chaotic fluid motions are supposed to 
66: pervade the `quiet' regions of the solar photosphere, with mixed magnetic polarities even below the 
67: photon mean-free-path (e.g., Cattaneo 1999).
68: At higher chromospheric and coronal levels the medium rarefies exponentially,
69: the magnetic fields weaken and the picture becomes still more problematic.
70: 
71: The presence of relatively weak magnetic fields may be revealed by
72: the modification they produce in the linear polarization pattern
73: generated by scattering processes 
74: in spectral lines (Hanle effect).
75: Typically (but not always), the change
76: consists in a net depolarization and a rotation of the polarization plane.
77: If the azimuth of the
78: magnetic field has a random variation 
79: within the observed resolution element, rotations
80: of polarization planes cancel out, but the reduction of the
81: scattering polarization amplitude remains. Therefore, the
82: Hanle effect has the potential for detecting the presence of
83: tangled magnetic fields on subresolution scales 
84: in the solar atmosphere (Stenflo 1982).%; Stenflo et al. 1998).
85: 
86: It seems then natural to rely on this technique to probe the kind of 
87: `microturbulent' photospheric magnetic fields discussed above.
88: This has been tried out by some authors in the past, concluding 
89: that depolarization by a 'turbulent' field is required to be able 
90: to explain the available observations of scattering polarization 
91: on the sun (Stenflo 1982; Faurobert-Scholl et al.
92: 1995, 2001; Stenflo et al. 1998).
93: The main problem of these analyses is that they require a reference
94: zero-field polarization amplitude which, in turn, heavily relies
95: on theoretical calculations. 
96: Only very recently such radiative transfer calculations have started 
97: to be sufficiently realistic, taking into account the three-dimensional 
98: and dynamic nature of the solar atmosphere (Trujillo Bueno, Shchukina 
99: \& Asensio Ramos 2004).
100: In any case, the question may always be raised of whether all the other 
101: relevant physical mechanisms that can produce depolarization have been 
102: accounted for and whether the observed depolarization can then be safely 
103: ascribed to the presence of a magnetic field. 
104: In order to achieve model independence it is useful to consider various
105: spectral lines with different sensitivities to the Hanle effect 
106: (Stenflo et al. 1998).
107: Unfortunately, the particular combination of atomic lines used by
108: Stenflo et al. (1998) is far from optimum, because they have different
109: line formation properties and none of them is insensitive to magnetic fields.
110: %\footnote{Moreover, it has to be 
111: %remarked that the results obtained by Stenflo et al. (1998) 
112: %are based on the assumption that all lines have positive intrinsic
113: %polarization. Since this assumption is not physically justified,
114: %such results have to be taken with some caution.}.
115: Here we present an analysis of the polarization pattern of the 
116: multiplet 42 of Ti {\sc i}.
117: We argue that the available observations of scattering polarization
118: in this multiplet provide direct evidence that magnetic depolarization 
119: due to a `turbulent' field is really
120: at work in the quiet solar photosphere.
121: 
122: 
123: Recently, Manso Sainz \& Landi Degl'Innocenti (2003a,b; 
124: hereafter Papers {\sc i} and {\sc ii}, respectively) have investigated
125: the physical origin of the observed linearly-polarized, 
126: solar limb spectrum of Ti {\sc i}.
127: To this end, these authors used
128: a very realistic multilevel model for 
129: the Ti {\sc i} atom, which includes a very interesting
130: multiplet that may be of great relevance for the study of solar magnetism
131: via the Hanle effect.
132: This multiplet is formed by the transitions between the 
133: lowest term of the titanium's quintet system ($a\,{}^5\!F$) 
134: and the excited $y\,{}^5\!F^\circ$ term (see Fig.~1).
135: Interestingly, all the lines in the multiplet lie in the blue part
136: of the visible spectrum and their scattering polarization signals have, 
137: in fact, been recently observed in quiet regions near the solar
138: limb (Gandorfer 2002).
139: The important thing about this multiplet is that the transition
140: at 4536 \AA\ shows
141: a large polarization peak relative to the other lines of the multiplet
142: and that the Land\'e factor of both its lower and upper level 
143: ($a\,{}^5\!F_1$ and $y\,{}^5\!F^\circ_1$, respectively) is zero.
144: In other words, the line at 4536 \AA\ is,
145: unlike the others of the same multiplet, essentially insensitive
146: to magnetic fields.
147: We show here that the most natural explanation 
148: of the reported linear polarization observations invokes the existence
149: of a tangled magnetic field permeating the solar atmosphere 
150: within the resolution element.
151: 
152: Panel (a) of Fig.~2 shows the observed fractional linear polarization 
153: in the $a\,{}^5\!F$-$y\,{}^5\!F^\circ$ multiplet as reported by
154: Gandorfer (2002).
155: The transition at 4527.3~\AA\ will not be considered further in our 
156: analysis for its being blended with two other transitions of 
157: Ce {\sc ii} and Cr {\sc i}, two species seemingly quite `active' 
158: in scattering line polarization.
159: The first thing to note is that all these lines are strong
160: in the sense of Paper {\sc ii}, i.e., they have residual intensities
161: $(I-I_c)/I_c$ larger than 0.5, and all of them, with the notable exception of $\lambda$4536, have $Q/I$ signals below the continuum polarization level.
162: In the presence of a significantly polarized continuum 
163: the contribution of the strongest spectral lines 
164: to the linearly polarized spectrum cannot be simply added up
165: and the depolarization of the continuum by the presence of 
166: the spectral line cannot be neglected.
167: Nevertheless, we can always assume a similar effect for 
168: all the lines in the multiplet and compare the $Q/I$ signals
169: between them, in particular, with the 4536~\AA\  line.
170: 
171: A theoretical estimate for the $Q/I$ structure of the multiplet is
172: shown in panel (b) of Fig.~2.
173: It has been calculated for a 90$^\circ$ observation of a layer of
174: atomic titanium gas which is illuminated from below by a photospheric-like
175: radiation field. Details on the atomic model and radiation field
176: are given in Paper {\sc i}. Firstly,
177: we have calculated the atomic polarization
178: that optical pumping processes induce
179: in all the levels of the atomic model in the absence of any
180: depolarizing mechanism. Secondly, we have calculated
181: the $Q/I$ signal in each transition taking into
182: account the emissivity as well as dichroism effects 
183: (see Eq.~(14) of Paper {\sc i}).
184: As demonstrated in Paper {\sc i}, this method is able
185: to reproduce (at least qualitatively) the structure of the 
186: whole second solar spectrum of Ti {\sc i}.
187: 
188: In order to model the depolarization 
189: due to a weak microturbulent magnetic field we assume one and the same 
190: depolarizing rate for all the atomic levels, except for
191: the levels $a\,{}^5\!F_1$ and
192: $y\,{}^5\!F^\circ_1$ which have zero Land\'e factor (Sugar \& Corliss 
193: 1985)\footnote{See {\tt http://physics.nist.gov/}}. 
194: We point out that all the ${}^5\!F$ terms of titanium 
195: are well described by the LS-coupling. Therefore,
196: according to the well known formula for the Land\'e 
197: factor we can assume that all the ${}^5\!F_1$ and ${}^5\!F^\circ_1$ 
198: levels are completely insensitive to the magnetic field. 
199: Panels (c) and (d) show the $Q/I$ structure of the 
200: $a\,{}^5\!F$-$y\,{}^5\!F^\circ$ multiplet assuming depolarizing rates $D$
201: of $10^8$ s$^{-1}$ and $5{\times}10^8$ s$^{-1}$, respectively, for all
202: except the ${}^5\!F_1$ and ${}^5\!F^\circ_1$ levels (thick lines).
203: If we roughly estimate an `equivalent' magnetic field intensity
204: through the equation
205: $2\pi\nu_L g=8.79{\times}10^6\,B\,g\approx D$ 
206: ($\nu_L$ and $g$ being the Larmor frequency
207: and Land\'e factor, respectively), then, taking $g{\approx}1$,
208: a depolarizing rate $D=10^8$ s$^{-1}$ 
209: corresponds to a magnetic field of the order of 12 gauss, while 
210: $D=5{\times}10^8$ s$^{-1}$ implies 60 gauss.
211: For comparison, dotted lines show the $Q/I$ values when all 
212: the levels are depolarized without exception.
213: Only the transitions involving at least one level
214: with zero Land\'e factor
215: ($\lambda\lambda$4527, 4536 and 4544) change their behavior,
216: as expected.
217: 
218: It is of interest to note that this $D=5{\times}10^8$ s$^{-1}$ 
219: case produces a relatively good agreement with 
220: the $Q/I$ observations of Gandorfer (2002), 
221: and that Trujillo Bueno et al. (2004) concluded that
222: when one assumes a volume-filling and single-valued microturbulent
223: field then the best theoretical fit
224: to the observed scattering polarization in the Sr {\sc i} 4607 \AA\ line
225: is obtained for $B_{\rm microturbulent}{\approx}60$ gauss.
226: 
227: Since we are considering transitions belonging to the same multiplet,
228: we can assume that they `form' in the same atmospheric region and hence,
229: under similar environmental conditions.
230: If this formation region were
231: essentially devoid of magnetic fields
232: (say, a significant volume filled with magnetic fields below one gauss), then we would expect relative $Q/I$ signals qualitatively resembling
233: the solid line in Fig.~2b or the dotted lines in Figs.~2c, d.
234: In particular, the ${\Delta}J=0$ transitions 
235: $\lambda\lambda$4533, 4534 and 4535 would show $Q/I$ signals similar 
236: or even larger than the $\lambda$4536 transition.
237: However, the observed large polarization peak in the 4536 \AA\ line 
238: relative to the others, can only be explained if the depolarization
239: is of magnetic origin and therefore does not affect the ${}^5\!F_1$ levels.
240: 
241: Finally, we would like to point out that the main deficiency of our analysis
242: is our neglecting of the continuum polarization.
243: At such blue wavelengths the continuum polarization due to Rayleigh and/or Thomson
244: scattering is already significant, and lines as strong as the ones
245: in this 
246: multiplet show an important interaction with it 
247: that cannot be neglected (see discussion in Paper {\sc ii}).
248: For this reason we have presented our theoretical calculations 
249: in Fig.~2 without superimposing the continuum.
250: These considerations should be taken into account in further,
251: more quantitative analyses.
252: Yet, they do not change our conclusions.
253: 
254: 
255: %In summary, the argument of this letter is very simple:
256: Summarizing: 
257: among the several observed solar lines of the $a\,{}^5\!F$-$y\,{}^5\!F^\circ$ multiplet of titanium, only the line $\lambda$4536 shows a 
258: polarization peak, all others showing just a continuum depolarization.
259: Since the only apparent peculiarity of
260: the 4536\AA\ line lies in its lower and upper levels having zero 
261: Land\'e factor (i.e., its insensitivity to magnetic fields),
262: the observed polarization pattern may be considered 
263: as direct evidence for
264: the presence of a tangled magnetic field in the solar
265: atmosphere, which is depolarizing  the scattering polarization signals
266: of all the lines in the multiplet except, 
267: obviously, for $\lambda$4536.
268: 
269: 
270: The arguments above rely on a direct interpretation 
271: of observations and on a relative independence of theoretical modeling
272: (actually, this is where their strength lies).
273: However, since we are not solving the radiative transfer problem
274: we cannot account for the details of the polarization patterns
275: arising from the different illumination and depolarizing conditions
276: at the height of formation of the individual lines. 
277: This prevents, in general, our comparing polarization signals 
278: between spectral lines belonging to different multiplets
279: to obtain quantitative estimates of the magnetic field.
280: Notwithstanding, we can compare lines belonging 
281: to the same multiplet assuming that they form under 
282: similar thermodynamical, radiative and magnetic conditions.
283: We can then make qualitative interpretations of the most conspicuous 
284: polarization patterns.
285: That has been precisely the case here.
286: 
287: In this respect, it must be noticed the interesting multiplet 
288: $a{}^3\!H$-$x{}^3\!H$, whose line at 4742.8 \AA\ 
289: shows a very conspicuous polarization signal (actually one of the strongest
290: of the whole second solar spectrum of Ti {\sc i}), which is twice
291: as large as the signal observed in 
292: the $\lambda\lambda$4758.1 and 4759.3 lines of the multiplet.
293: This cannot be easily explained in terms of the simple modeling 
294: of Paper {\sc i} (all three would show large, but similar 
295: amplitudes), 
296: nor by some kind of 'differential' action of the magnetic
297: field, since all upper levels of the three transitions have similar
298: critical fields (the field at which the Zeeman splitting is of the order
299: of the natural width of the level): $\approx$ 11, 9 and 8 gauss, respectively.
300: This anomalous behavior is probably related to an interesting spectroscopic
301: feature of the lower level of $\lambda$4742.8.
302: Contrary to the other levels involved in the multiplet, this level 
303: (with total angular momentum $J=4$) is not 
304: well described as a purely LS-coupling level.
305: Actually, it is a 52\% -- 44\% mixing of the two configurations
306: $3d^3({}^2\!H)4s \, {}^3\!H$ and $3d^3({}^2\!G)4s\, {}^1\!G$, respectively
307: (see Sugar \& Corliss 1985).
308: (It is interesting to note that 
309: the lower level ($b{}^1\!G$) of the transition  
310: showing the largest observed $Q/I$ signal of
311: the titanium's second solar spectrum ($\lambda$5644.1) 
312: is a 48\% -- 48\% mixing
313: of the very same configurations $3d^3({}^2\!H)4s \, {}^3\!H$ and 
314: $3d^3({}^2\!G)4s\, {}^1\!G$, respectively.)
315: Due to this accidental coincidence the lower and upper levels of 
316: $\lambda$4742.8 are spectroscopically similar to an equivalent
317: LS-coupling level (Land\'e factor, total deexcitation probability,
318: and critical field), but the line itself has a smaller transition 
319: probability relative to $\lambda\lambda$4758.1 and 4759.3. 
320: Consequently,  the line  $\lambda$4742.8 probably forms 
321: at slightly different 
322: height in the photosphere than the other two, where the 
323: radiation field may be dissimilar.
324: 
325: 
326: We think that, all in all, the interpretation of the polarization
327: pattern in the  $a\,{}^5\!F$-$y\,{}^5\!F^\circ$ multiplet
328: of neutral titanium given in this letter is direct 
329: evidence for the existence of an unresolved 
330: magnetic field permeating large volumes of the 
331: quiet solar atmosphere. 
332: In any case, further study of the formation of the second solar spectrum
333: of Ti {\sc i} is necessary
334: in order to  arrive to more definite
335: conclusions.
336: By combining the Hanle-effect
337: line ratio technique proposed here for the relatively strong
338: 13 lines of this Ti {\sc i} multiplet with that reported by Trujillo Bueno (2003)
339: for the weak ${\rm C}_2$ lines of the Swan system,
340: we can now aim at further exploring
341: the full complexity of the small-scale magnetic activity
342: of the `quiet' Sun. To this end, we have already initiated
343: systematic observations of both atomic and molecular lines as well as 
344: more detailed theoretical investigations.
345: 
346: 
347: \acknowledgments
348: This research has been partly funded by the
349: Spanish Ministerio de Educaci\'on y Ciencia
350: through project AYA2001-1649 and by the European
351: Solar Magnetism Network. We thank 
352: Roberto Casini for several interesting discussions on
353: atomic spectroscopy and Jan Olof Stenflo for suggesting improvements
354: to the original version of this letter.
355: 
356: \begin{thebibliography}{}
357: \bibitem[1999]{cat} Cattaneo, F. 1999,
358: 	ApJ, 515,L39
359: \bibitem[1995]{fa1} Faurobert-Scholl et al. 1995, 
360:         A\&A, 298, 289
361: \bibitem[2001]{fa2} Faurobert-Scholl et al. 2001, 
362:         A\&A, 378, 627
363: \bibitem[2002]{gandorfer2} Gandorfer, A. 2002, 
364: 	The Second Solar Spectrum Volume {\sc ii}: 3910\AA\ to 4630\AA\ 
365: 	(Z\"urich: vdf Hochschulverlag AG an der ETH)
366: \bibitem[2003a]{rms1} Manso Sainz, R. \& Landi Degl'Innocenti, E. 2003a,
367: 	A\&A, 394, 1093 (Paper {\sc i})
368: \bibitem[2003b]{rms2} Manso Sainz, R. \& Landi Degl'Innocenti, E. 2003b,
369: 	in ASP Conf. Ser. 307,
370: 	Solar Polarization~3, eds. J. Trujillo Bueno \&
371: 	J. S\'anchez Almeida (San Francisco: ASP), 425 (Paper {\sc ii})
372: \bibitem[1982]{stenflo2} Stenflo, J. O. 1982, Solar Phys., 80, 209
373: \bibitem[2000]{st} Stenflo, J. O., Keller, C. U., \& Gandorfer, A. 1998,
374:         A\&A, 329, 319
375: \bibitem[1985]{sugar}
376: Sugar, J. \& Corliss, C. 1985, J. Phys. Chem. Ref. Data, 14, 
377: 	Suppl. 2
378: \bibitem[2003]{javier03} Trujillo Bueno, J., 2003
379: in ASP Conf. Ser. 307,
380: 	Solar Polarization~3, eds. J. Trujillo Bueno \&
381: 	J. S\'anchez Almeida (San Francisco: ASP), 407
382: \bibitem[2004]{javier04} Trujillo Bueno, J., Shchukina, N., \& Asensio Ramos, A. 2004,
383:          Nature, 430, 326	
384: \end{thebibliography}
385: 
386: 
387: \clearpage
388: \begin{figure}
389: \centering
390: \includegraphics[width=7cm]{f1.eps}
391: \caption{Transitions (wavelengths in \AA) 
392: 	between the $a\,{}^5\!F$ and $y\,{}^5\!F^\circ$ terms of Ti {\sc i}.
393: 	Upper labels indicate the total angular momentum of the levels.}
394: \end{figure}
395: 
396: \clearpage
397: \begin{figure}
398: \centering
399: \includegraphics[width=8cm, bb=0 110 255 595]{f2.eps}
400: \caption{Fractional polarization $Q/I$ in the thirteen spectral lines 
401: 	allowed between the $a\,{}^5\!F$ and $y\,{}^5\!F^\circ$ 
402: 	terms of Ti {\sc i}.
403: 	{\bf a)} observations near the solar limb as reported 
404: 	in Gandorfer (2002).
405: 	{\bf b)} theoretical $Q/I$ calculated as explained in the text.
406: 	{\bf c)-d)} as panel {\bf b)} assuming depolarizing rates
407: 	10$^8$ s$^{-1}$ and $5{\times}10^8$ s$^{-1}$, respectively, for all
408: 	Ti {\sc i} levels except the $a{\,}^5\!F_1$ and 
409: 	$y{\,}^5\!F^\circ_1$ones. 
410: 	Note the change of the $Q/I$ scale between panels.
411: 	The meaning of the dotted lines is explained in the text, while
412: 	the symbol * indicates a Ti {\sc i} line that we do not consider
413: 	in our analysis because heavily blended.}
414: \end{figure}
415: 
416: \end{document}
417: 
418: 
419: