1: \documentclass[preprint,12pt]{aastex}
2: \usepackage{emulateapj5}
3: \usepackage{graphicx}
4: \usepackage{rotate}
5: \usepackage{amsmath}
6: \usepackage{amssymb}
7:
8: \begin{document}
9:
10: \title{On the role of magnetic fields in abundance determinations}
11:
12: \author{J.M.~Borrero}
13: \affil{High Altitude Observatory (NCAR), 3080 Center Green Dr. CG-1, Boulder, CO 80301, USA}
14: \email{borrero@ucar.edu}
15:
16: \begin{abstract}
17: {Although there is considerable evidence supporting
18: an ubiquitous magnetic field in solar/stellar photospheres, its impact
19: in the determination of abundances has never been quantified. In this work
20: we investigate whether the magnetic field plays a measurable role for this kind
21: of studies. To that end, we carry out simulations of spectral line formation in the
22: presence of a magnetic field, and then use those profiles to
23: derive the abundance of several atomic species (Fe, Si, C and O) neglecting the magnetic field.
24: In this way, we find that the derived iron abundance can be significantly biased,
25: with systematic errors up to 0.1 dex. In the case of
26: silicon, carbon and oxygen their role is very marginal (errors smaller than 0.02 dex). We
27: also find that the effect of the magnetic field strongly depends on its
28: inclination with respect to the observer. We show that fields that are aligned
29: with the observer lead to an underestimation of the real abundance, whereas more inclined ones overestimate it.
30: In the case of a mixture of fields with different inclinations these effects
31: are likely to partly cancel each other out, making the role of the magnetic field even less important.
32: Finally, we derive a simple model that can be used to determine the suitability
33: of a spectral line when we wish to avoid the bias introduced by the neglect of the magnetic field.}
34: \end{abstract}
35:
36: \keywords{Sun: magnetic fields, abundances -- Stars: magnetic fields, abundances}
37: \shorttitle{Magnetic field and abundance determinations}
38: \shortauthors{J.M.~Borrero}
39: \maketitle
40:
41:
42: %%
43: \section{Introduction}
44: %%
45:
46: From the very first determinations of solar and stellar abundances
47: using 1D semi-empirical atmospheric
48: models (e.g. Lambert 1968, Lambert \& Warner, B. 1968, Grevesse 1968, Garz et al. 1969)
49: to more recent values obtained from state-of-the-art 3D hydrodynamical
50: simulations (e.g. Asplund et al. 2005a), the role
51: of the magnetic field has rarely been considered.
52: This would be strictly valid only if atomic transitions with zero Land\'e factors
53: are used in the analysis. Unfortunately this has never been the case.
54: The reason for this is that there are few magnetic insensitive spectral
55: lines having accurate oscillator strengths. In the case of the Sun, the role
56: of the magnetic field has been avoided by arguing that the FTS-disk
57: center spectral atlas (Brault \& Neckel 1987; Neckel 1999), which is the
58: most common source to compare observed and simulated line profiles,
59: was recorded around a quiet Sun region.
60:
61: The existence of significant magnetic flux in quiet Sun regions
62: has passed unnoticed because this magnetic field organizes in patches
63: of opposite polarity\footnote{In spectropolarimetry, the term
64: positive polarity refers to magnetic fields pointing towards the observer:
65: inclination angles $\gamma < 90^{\circ}$, while negative polarities refer
66: to magnetic fields pointing away from the observer: $\gamma > 90^{\circ}$}
67: over very small scales, leading to a cancellation
68: of the polarization signals. However, there is now strong evidence
69: supporting the omnipresence of magnetic fields in regions previously
70: thought to be void of them (Trujillo Bueno et al. 2004;
71: Manso Sainz et al. 2004). The details about its actual strength
72: and distribution are subject to debate, however.
73:
74: This is even more critical when disk integrated data is used in
75: stellar abundances studies (e.g. Allende Prieto et al. 2002), in particular if
76: the star is magnetically active (Ap-Bp types), as regions of strong
77: magnetic field (i.e.: starspots) can have a large influence.
78:
79: Although the magnetic field will mainly affect the polarization signals,
80: it also has an impact on the intensity profiles (Stokes $I$), being
81: this particularly important because this effect on the intensity profiles
82: adds up regardless of its polarity.
83: If not accounted for, this might lead to systematic errors
84: in the abundance values derived from the fitting.
85: The goal of this work is to asses the importance of neglecting the
86: role of magnetic fields. To that end we
87: will perform simplistic simulations of intensity profiles of various important
88: atomic elements in the presence of a magnetic field. In Section 2 we describe
89: the synthesis code employed, the spectral lines used and we briefly review
90: the Zeemann effect applied to Stokes $I$. In Section 3 we study how the strength and inclination of the
91: magnetic field taintes the inferred abundances of iron, silicon
92: carbon and oxygen when the existence of this field is neglected. In Section 4 we
93: derive a simple model that is able to quantitatively predict
94: the commited error. Section 5 is devoted to studying if results from
95: our simple modeling could differ significantly if the same investigation
96: is performed using more realistic 3D MHD models. Section 6 summarizes
97: our findings and anticipates possible future work.
98:
99:
100: %%
101: \section{Spectral lines synthesis}
102: %%
103:
104: %
105: \subsection{Synthesis code}
106: %
107:
108: We have employed the SIR code (Ruiz Cobo \& del Toro Iniesta 1992) to
109: produce synthetic spectral lines. This code solves the radiative transfer equation in the presence of
110: a magnetized plasma. Although SIR synthesizes the full Stokes vector, we will
111: restrict ourselves to consider only the total intensity, Stokes $I$, as done in most
112: abundance studies. In addition, SIR allows for the magnetic field vector to be a function
113: of the optical depth, but in this work we will consider it constant.
114:
115: The Harvard-Smithsonian Reference Atmosphere (HSRA; Gingerich et al. 1971)
116: was used in our calculations. Note that using this 1D LTE semi-empirical model
117: means that we study the impact of the magnetic field alone, that is,
118: we assume that other important ingredients of the spectral line formation are
119: already being accounted for. For instance, the magnetic field couples with
120: the energy and momentum equation, resulting in a modification
121: of the temperature stratification. In our analysis we are assuming that the correct temperature
122: stratification and convective velocity fields are known (i.e.: obtained through realistic 3D simulations)
123: and therefore we focus on the magnetic field. If this was not the case, small errors in the
124: temperature or velocities would dominate over the neglect of the magnetic field.
125:
126: To account for the convective broadening of the spectral lines we
127: use, unless otherwise specified, a macroturbulent velocity of 2 km s$^{-1}$ and a microturbulent velocity of 1 km s$^{-1}$.
128: We anticipate (see Section 5) that these values have no particular consequences in our discussion.
129:
130: %
131: \subsection{Spectral line selection}
132: %
133:
134: We have decided to focus our study on four important atomic elements: \ion{Fe}{1},
135: \ion{Si}{1}, \ion{C}{1} and \ion{O}{1}. Iron is particularly important because it shows a large
136: number of atomic transitions at visible wavelengths, and therefore it is often used to
137: investigate solar and stellar atmospheres. In addition, it is commonly used to distinguish whether a
138: star is first or second generation since heavier elements are only produced in Supernova
139: explosions (Christlieb et al. 2002). Silicon is important in the solar context as it is used,
140: together with iron, to compare with meteoritic abundances (Asplund 2000). Oxygen
141: and carbon's importance comes from being the third and fourth most abundant elements in the Universe,
142: respectively. In addition, they have both been targeted as being responsible for the
143: lowering of the solar metallicity (Asplund et al. 2005a), that has caused
144: major discrepancies between helioseismic inversions and solar models (Castro et al. 2007).
145: Therefore, it is worthwhile investigating whether magnetic fields could have
146: something to say in this regard.
147:
148: In total we consider 57 spectral lines: 29 of \ion{Fe}{1}, 15 of \ion{Si}{1}, 9 of \ion{C}{1}
149: and 4 of \ion{O}{1}. Their properties are summarized in Table 1. They have been adopted from Asplund (2000) and
150: Asplund et al. (2000, 2004, 2005b). We have rejected those lines for which accurate
151: collisional parameters (under the ABO theory; Anstee \& O'Mara 1995; Barklem \& O'Mara 1997;
152: Barklem et al. 1998) were not found. Note that only one line, \ion{Si}{1} 5665.555 \AA\ has
153: a zero Land\'e factor: $g_{\rm eff}=0$. Several neutral iron and carbon lines are potentially
154: very sensitive to magnetic fields: $g_{\rm eff} \gtrapprox 2$.
155:
156: %
157: \subsection{Intensity profiles under the presence of a magnetic field}
158: %
159:
160: Ignoring off-diagonal elements in the propagation matrix, the main contributor to the intensity
161: profiles is $\eta_I$ (Wittmann 1974; del Toro Iniesta 2003):
162:
163: \begin{equation}
164: \eta_I = 1 + \frac{\eta_0}{2}\left\{\phi_p \sin^2\gamma + \frac{1}{2}[\phi_b+\phi_r](1+\cos^2\gamma)\right\}
165: \end{equation}
166:
167: where $\eta_0$ is related to the abundance of the element, the excitation potential of the lower level
168: and the transition probability of the atomic transition (Landi Degl'Innocenti 1976). $\gamma$ refers
169: to the inclination of the magnetic field vector with respect to the observer. Since it appears
170: as $\cos^2\gamma$ and $\sin^2\gamma$, its contribution is
171: the same regardless of the polarity of the magnetic field (see Footnote 1). Treating the real
172: Zeemann pattern as an effective triplet ($J=1 \rightarrow J=0$), the functions $\phi_p$, $\phi_r$
173: and $\phi_b$ refer to the Voigt profiles for the $\Pi$ ($\Delta M=0$), blue ($\Delta M = -1$)
174: and red ($\Delta M = 1$) components of the Zeeman pattern, respectively. The former is centered at $\lambda_0$
175: (central laboratory wavelength), whereas the latter ones are shifted by an amount $\pm \lambda_B$ with respect
176: to the $\Pi$ component.
177:
178: \begin{equation}
179: \lambda_B = \mathcal{C} g_{\rm eff} B \lambda_0^2
180: \end{equation}
181:
182: \noindent where $C=4.67\times 10^{-13}$ [\AA\ Gauss]$^{-1}$, $B$ is the strength of the magnetic field
183: (measured in Gauss), and $g_{\rm eff}$ is the effective Land\'e factor of the spectral line.
184:
185: A number of remarks are in order. First of all, in the absence of a magnetic field (or if $g_{\rm eff}=0$),
186: $\phi_p=\phi_r=\phi_b$ and therefore $\eta_I=1+\eta_0 \phi_p/2$. Also, it is interesting to see that
187: the blue and red components are always present independently
188: of the orientation of the field. When the field is small enough the splitting is much smaller
189: that the Doppler width of the spectral line, $\lambda_B \ll \Delta \lambda_{\rm D}$,
190: causing the spectral line to broaden. In the opposite case, $\lambda_B \gg \Delta \lambda_{\rm D}$,
191: they appear as two separate spectral lines with half the strength of the spectral line in the absence of
192: magnetic field. Finally, the central $\Pi$ component vanishes for magnetic fields
193: aligned with the observer ($\gamma=0$) causing the line core to desaturate.
194:
195:
196: %%
197: \section{Effects on the magnetic field in the abundance of Fe, Si, C and O}
198: %%
199:
200: We now proceed to calculate synthetic Stokes $I$ profiles of the spectral lines
201: in Table 1. We use the numerical code and the atmospheric model described in Sect~2.1.
202: We consider a magnetic field with a varying strength and inclination: $B=[0,500]$ Gauss,
203: $\gamma=[0,90]$ deg. We consider also standard solar abundances: $\log\epsilon_{Fe}=7.45$ (Asplund et al. 2000),
204: $\log\epsilon_{Si}=7.51$ (Asplund 2000), $\log\epsilon_{C}=8.39$ (Asplund et al. 2005b)
205: and $\log\epsilon_{O}=8.66$ (Asplund et al. 2004). The resulting profiles are then fitted
206: with different abundances but assuming that there is no magnetic
207: field. For \ion{Fe}{1} we use $\log\epsilon_{Fe}=[7.30,7.80]$ in steps of 0.01. For \ion{Si}{1}
208: and \ion{O}{1} we consider $\log\epsilon_{Si}=[7.48,7.54]$ and $\log\epsilon_{O}=[8.62,8.70]$,
209: respectively, both in steps of 0.002. Finally, for \ion{C}{1} we use $\log\epsilon_{C}=[8.32,8.46]$
210: in steps of 0.005.
211:
212: From the comparison of the original profiles with magnetic field
213: and fixed abundance with those where the abundance varies and the magnetic field is neglected,
214: we obtain a $\chi^2-\log\epsilon$ curve for each spectral line. We therefore infer for each
215: line an optimum abundance, as the one that minimizes the $\chi^2$ curve. No other free parameters
216: are considered. An example of this process is presented
217: in Figure 1 using the spectral line \ion{Fe}{1} 6136.994 \AA\, a magnetic field of $B=250$ Gauss and
218: $\gamma=60^{\circ}$. The final inferred abundance is obtained as the mean of the best-fit abundances
219: from all spectral lines of that atomic element.
220:
221:
222: \begin{center}
223: \includegraphics[width=7.5cm]{f1.ps}
224: \figcaption{$\chi^2$ between the intensity profile of \ion{Fe}{1} 6136.994 \AA\ with
225: magnetic field ($B=250$ Gauss; $\gamma=60^{\circ}$) and an iron abundance of $\log\epsilon_{Fe}=7.45$,
226: and the intensity profile of the same spectral line without magnetic field, as a function of the abundance.
227: The abundance that produces the best fit to the former profile is $\log\epsilon_{Fe}\simeq7.49$.}
228: \end{center}
229:
230: Figure 2 presents the errors $\Delta\log\epsilon=\log\epsilon_{\rm fit}-\log\epsilon_{\rm real}$
231: (as a function of the magnetic field strength and inclination) introduced in the abundance of
232: \ion{Fe}{1}, \ion{Si}{1}, \ion{C}{1} and \ion{O}{1}, when the magnetic field is not accounted for. As
233: expected, the larger the field strength the larger the error. Neutral iron,
234: with up to $\Delta\log\epsilon_{\rm Fe} \le 0.1$, presents the largest deviations. For the rest of considered
235: elements the magnetic field seems to have only a marginal effect: $\Delta\log\epsilon_{\rm Si} \le 0.01$ and $\Delta\log\epsilon_{\rm C,O}
236: \le 0.02$.
237:
238: We also find that vertical fields underestimate the correct abundance, whereas the opposite happens for more inclined
239: magnetic fields. This can be explained attending to Sect~2.3 (Equation 1). For vertical
240: magnetic fields, $\gamma=0^{\circ}$, the $\Pi$ component of the Zeemann pattern
241: is absent, leading to a desaturation of the core intensity and therefore requiring a smaller abundance to fit
242: the line profile. An example with \ion{Fe}{1} 5250.209 \AA\ is presented in Figure 3 (left panel).
243: However, when the magnetic field is horizontal, $\gamma=90^{\circ}$, the line mainly broadens and
244: thus yieling a larger abundance. See example for \ion{Fe}{1} 6200.313 \AA\ in Figure
245: 3 (right panel).
246:
247: \begin{center}
248: \includegraphics[width=7.5cm]{f2a.ps}
249: \includegraphics[width=7.5cm]{f2b.ps}
250: \includegraphics[width=7.5cm]{f2c.ps}
251: \includegraphics[width=7.5cm]{f2d.ps}
252: \figcaption{Errors in the inferred abundances $\Delta\log\epsilon=\log\epsilon_{\rm fit}-\log\epsilon_{\rm real}$
253: when the role of the magnetic field is neglected, as a function of the magnetic field strength for 4 different
254: inclinations: $\gamma=0^{\circ}$ (dotted line; vertical magnetic field), $\gamma=30^{\circ}$ (dashed), $\gamma=60^{\circ}$
255: (dashed-dotted) and $\gamma=90^{\circ}$ (dashed-triple dotted; horizontal magnetic field). Solid lines indicate a mixture
256: of field inclinations, in which profiles obtained with the previous 4 inclinations are averaged before inferring an abundance.}
257: \end{center}
258:
259:
260: We must also consider that different regions of solar and stellar atmospheres can posses magnetic
261: fields with different inclinations. For instance, in the case of the solar granulation, where the
262: field is mostly horizontal in the granules (upflowing gas), but is generally vertical in
263: intergranular lanes (downflowing gas). Since the inclination that matters is with respect to
264: the observer, this situation reverses as we move towards higher latitudes or towards the solar limb.
265: Many other low and intermediate-mass stars also posses an outer convective layer. In addition, strong magnetic concentrations
266: also display a variety of inclinations: star-sunspots (umbra and penumbra),
267: pores, network regions etc. Although they are normally avoided in the Sun, in the stellar case they certainly
268: contribute to the observed profiles.
269:
270: Since vertical and horizontal fields seems to have opposite effects in the derived abundance it
271: is appropriate to study whether they can cancel each other out when a mixture of different
272: inclinations is present. To study this effect we have carried out a similar experiment as the ones previously
273: presented, being now the difference that we average the intensity profiles obtained with 4 different inclinations,
274: $\gamma=0,30,60,90^{\circ}$ (with the same field strength) before inferring an abundance for each spectral line.
275: In this case, the errors in the retrieved abundance are much smaller (see solid line in Figure 2), being only
276: perceptible for the case of neutral iron: $\Delta\log\epsilon_{\rm Fe} \le 0.02$.
277: This simulation is very simple in the sense that it does not include different temperature
278: stratifications, different velocity fields, etc. However, it helps to highlight that the aforementioned cancellation effect
279: can indeed take place. This cancellation effect in Stokes $I$ due to vertical and horinzontal magnetic fields, is similar in a way
280: to the cancellation of circular polarization signals due to a mixture of magnetic fields pointing towards and away from
281: the observer.
282:
283: \begin{center}
284: \includegraphics[width=7.5cm]{f3a.ps}
285: \includegraphics[width=7.5cm]{f3b.ps}
286: \figcaption{Filled circles are intensity profiles obtained with the nominal iron abundance of $\log\epsilon_{Fe}=7.45$,
287: a magnetic field strength of $B=500$ Gauss and inclinations of $\gamma=0^{\circ}$ (left panel) and $\gamma=90^{\circ}$
288: (right panel). Those profiles are then fitted without a magnetic field. When the original profile is affected
289: by a vertical magnetic field, the best fit (solid lines) is obtained by underestimating the original abundance. The
290: opposite occurs for horizontal magnetic fields.}
291: \end{center}
292:
293: From these results we conclude that the effect of neglecting the role of the magnetic field in abundance determinations
294: strongly depends on the atomic specie and spectral line under consideration, as well as the strength and inclination of the magnetic field.
295: From the four atomic species considered, it appears to have a measurable effect in the case of \ion{Fe}{1} only. It could be
296: marginally important for \ion{Si}{1}, \ion{C}{1} and \ion{O}{1} only if the magnetic field had a clearly preferred orientation.
297: In stellar atmospheres, where we observe disk-integrated signals, this is unlikely the case. In addition, magnetic fields do not appear
298: as a plausible source of error in the current controversy of the solar carbon and oxygen abundance.
299:
300: %%
301: \section{A phenomenological model}
302: %%
303:
304: The different behavior seen in \ion{Fe}{1} as compared to the other three elements considered cannot
305: be understood in terms of the magnetic sensitivity of the lines used, as most
306: of them have similar Land\'e factors ranging from $g_{\rm eff}=1-2$. The source of these differences must be
307: therefore thermodynamic. In this section we will develop a tool to differentiate whether a given
308: spectral line of a particular atomic element is prone to yield unreliable abundances.
309: To that end we assume that the error in the inferred abundances, that appears as a consequence of
310: neglecting the magnetic field, is directly proportional to the changes in the line profile in the
311: presence of a magnetic field:
312:
313: \begin{equation}
314: \Delta \log \epsilon \propto \max \left\|\frac{\partial I}{\partial B}\right\|
315: \end{equation}
316:
317: \noindent where we use maximum of the absolute value since the derivative
318: changes sign as a function of wavelength. Attending Equation 1, we will
319: model the intensity profile as a combination of three Gaussian, where
320: two of them are shifted by an amount $\pm \lambda_B$ with respect
321: to the central laboratory wavelength $\lambda_0$. The third Gaussian
322: is centered at $\lambda_0$ and possesses twice the strength of the other two.
323:
324: \begin{align}
325: I(\lambda) = 1 - \frac{[1-I_0] \sin^2\gamma}{2} e^{-\frac{(\lambda-\lambda_0)^2}{\Delta\lambda^2}}\\-\frac{[1-I_0](\cos^2\gamma+1)}{4}
326: \left\{e^{-\frac{(\lambda-\lambda_0-\lambda_B)^2}{\Delta\lambda^2}} +e^{-\frac{(\lambda-\lambda_0+\lambda_B)^2}{\Delta\lambda^2}}\right\}\notag
327: \end{align}
328:
329: \noindent where $I_0$ and $\Delta\lambda$ refer to the core intensity and the Doppler width of
330: the spectral line in the absence of magnetic field. That is,
331:
332: \begin{equation}
333: \lim_{B \to 0} I(\lambda) = 1 - (1-I_0) e^{-\frac{(\lambda-\lambda_0)^2}{\Delta\lambda^2}}
334: \end{equation}
335:
336: Note that $\Delta\lambda$ is related to the HWHM of the Gaussian profile by a constant
337: factor: HWHM=$\sqrt{\log 2} \Delta\lambda \simeq 0.832\Delta\lambda$. Table 1 presents
338: the values of $I_0$ and $\Delta\lambda$ obtained from a Gaussian fit to the intensity profiles
339: in the absence of magnetic field for each spectral line. Finally, Equations (2)-(4) can be combined
340: to write:
341:
342: \begin{align}
343: \Delta \log \epsilon \propto \max \bigg\| \frac{1-I_0}{2\Delta\lambda^2} (\cos^2\gamma+1)
344: \mathcal{C} g_{\rm eff} \lambda_0^2 [(\lambda-\lambda_0 +\notag \\ + \lambda_B) e^{-\frac{(\lambda-\lambda_0+\lambda_B)^2}{\Delta\lambda^2}}-
345: (\lambda-\lambda_0-\lambda_B) e^{-\frac{(\lambda-\lambda_0-\lambda_B)^2}{\Delta\lambda^2}}] \bigg\|
346: \end{align}
347:
348: This procedure is very similar to the one used in Cabrera Solana et al. (2005) but using Stokes
349: $I$ instead of the circular polarization, Stokes $V$. Equation 6 allows to predict the effect of ignoring the magnetic field
350: using: properties of the spectral line (regardless of the atomic specie)
351: in the absence of a magnetic field ($I_0$, $\Delta\lambda$, $\lambda_0$), Land\'e factor $g_{\rm eff}$
352: (given by the electronic configurations) and the strength of
353: the magnetic field (used to calculate $\lambda_B$). Note that the inclination of the magnetic field
354: does not play any role. The factor $[\cos^2\gamma+1]$ does not help to explain the fact that for
355: vertical fields $\Delta\log\epsilon<0$, but $\Delta\log\epsilon>0$ for horizontal fields. Therefore
356: it is more appropriate to write:
357:
358: \begin{align}
359: \Delta \log \epsilon = \mathcal{A}(\gamma) \max \bigg\| \frac{1-I_0}{2\Delta\lambda^2}
360: \mathcal{C} g_{\rm eff} \lambda_0^2 [(\lambda-\lambda_0+ \notag \\ + \lambda_B) e^{-\frac{(\lambda-\lambda_0+\lambda_B)^2}{\Delta\lambda^2}}-
361: (\lambda-\lambda_0-\lambda_B) e^{-\frac{(\lambda-\lambda_0-\lambda_B)^2}{\Delta\lambda^2}}] \bigg\|
362: \end{align}
363:
364: \noindent where the calibration constant $\mathcal{A}(\gamma)$ can be determined for different
365: field inclinations using our results in Section 3. This is done in Figure 4, where we plot
366: the left-hand side term of Equation 6 (obtained from Section 3) versus the right hand-side of
367: the same equation (evaluated using Table 1) for two limiting inclinations: $\gamma=0^{\circ}$
368: (upper panel) and $\gamma=90^{\circ}$ (lower panel) for all spectral lines. As it can be seen
369: the correlation is good enough to empirically justify, in first approximation, our assumption in Equation 3.
370: That is, the more a spectral line is affected by the magnetic field, the more unreliable the
371: derived abundance will be if the magnetic field is not considered. In addition, all spectral lines
372: seem to follow a linear relation regardless of the atomic specie. The present model
373: explains why \ion{Fe}{1} is affected by the magnetic field more than
374: \ion{Si}{1}, \ion{C}{1} and \ion{O}{1}. As already mentioned, this is not due to the different Land\'e factors, but rather due
375: to their different sensitivity to thermodynamic parameters: $I_0$ and $\Delta\lambda$. Narrower
376: and deeper spectral lines ($\Delta\lambda$ and $I_0$ small) such as the ones from
377: \ion{Fe}{1}, are more sensitive to the magnetic field than weak and broad spectral lines.
378:
379: It is also interesting to notice that the selected \ion{C}{1} lines are broader and weaker than those of
380: \ion{Si}{1} and \ion{O}{1}. Despite this, carbon is affected by the magnetic field as much as the other two
381: (see Fig.~2). The reason is that the employed \ion{C}{1} lines are located at larger wavelengths, $\lambda_0^{C} > \lambda_0^{Si,O}$,
382: and therefore the Zeeman splitting is larger (Eq.~2 and 7).
383:
384: Since we use the same temperature stratification, the different
385: $I_0$'s and $\Delta\lambda$'s among the selected spectral lines can only be due to differences in
386: the excitation potential of the lower level $\chi_l$, transition probability
387: $\log gf$, and collisional broadening parameters $\alpha$ and $\sigma$.
388: In first approximation, the former two would be related to $I_0$ whereas the latter two determine
389: $\Delta\lambda$.
390:
391: The utility of the procedure described here lies in the fact that whenever another spectral line
392: is considered (regardless of the atomic element), we can use this method to evaluate Equation 7, and therefore
393: calculate its approximate position in Figure 4. This will give us an idea of the error introduced
394: by ignoring the effects of the magnetic field, even if a numerical code that solves the radiative transfer
395: equation in the presence of a magnetic field is not available.
396:
397: \begin{center}
398: \includegraphics[width=9cm]{f4a.ps}
399: \includegraphics[width=9cm]{f4b.ps}
400: \figcaption{Error in the determination of element abundances when using individual spectral lines versus
401: the derivative of its intensity profile with respect to the magnetic field. Red crosses
402: show the 29 \ion{Fe}{1} lines from Table 1. Green squares indicate \ion{Si}{1}. \ion{O}{1} is denoted
403: by blue triangles. Yellow circles show \ion{C}{1} lines. A linear fit to the data points is
404: indicated by the black dashed line. The slope of the fitted line is indicated by $\mathcal{A}(\gamma)$.
405: {\it Top panel}: vertical magnetic field. {\it Bottom panel}: horizontal magnetic field.}
406: \end{center}
407:
408: %
409: \section{Convective vs Magnetic broadening}
410: %
411:
412: In our simple 1D model the convective broadening is introduced through the macro and
413: microturbulent velocities, whereas in a more realistic 3D modeling this broadening
414: naturally occurs when adding the profiles emerging from different regions (i.e.:
415: granules and intergranules) in the solar atmosphere. At first glance it seems plausible
416: for this convective broadening to mask the magnetic one, thus making the contribution
417: of the magnetic fields even more negligible. However, we must take into account that the individual
418: intensity profiles emerging from different regions of the solar atmospheres are already affected by the
419: magnetic field. Therefore, when all intensity profiles are added up to produce the corresponding
420: convective broadening, the fingerprints of magnetic field will still be visible.
421:
422: To prove this statement we have carried out a simple simulation of the process previously described.
423: To that end we synthesize intensity profiles of \ion{Fe}{1} 5247.05 \AA\ (g$_{\rm eff}$=2)
424: without magnetic field and no net line-of-sight velocity, but using a macroturbulent velocity of 2 km s$^{-1}$.
425: The resulting profile is plotted in Figure 5 ($I_1$, thick dashed line). We then repeat this synthesis
426: but now including a vertical magnetic field, $\gamma=0^{\circ}$, with a strength of 1000 Gauss.
427: The resulting profile is plotted in Figure 1 ($I_2$, thick solid line).
428:
429: Next we perform a similar synthesis, where: a) we set the macroturbulent velocity to zero;
430: b) we include a vertical field of 1000 Gauss and, c) we use a net line-of-sight velocity that
431: changes from $-$4 km s$^{-1}$ to 4 km s$^{-1}$. Each of the emerging profiles is indicated
432: by one of the thin dashed lines in Fig.~1 (labeled as $I_k$). With this we try to simulate the effect of having
433: different structures affected by convective velocity fields. We then add all those profiles
434: using a weighting function with a Gaussian shape in order to mimic the macroturbulence.
435: The resulting profile is indicated in Fig.~1 as $I_3$ (squares).
436:
437: \begin{eqnarray}
438: I_3 = \sum w_k I_k \\
439: w_k = \frac{v_{\rm mac}}{2} \sqrt{\frac{\pi}{\log2}} e^{-4\log2 \frac{v_{\rm los,k}^2}{v_{\rm mac}^2}}
440: \end{eqnarray}
441:
442: where the FWHM of the Gaussian weighting function is equivalent to a macroturbulent velocity of 2 km s$^{-1}$.
443: As it can be seen, $I_3 = I_2$. Therefore, the broadening due to convective
444: motions does not hide the broadening produced by magnetic
445: fields. Our simulations do not consider that this magnetic field is probably different in granules
446: (upflows) and intergranules (downflows). We also use the same temperature stratification
447: in all cases. Considering a more realistic case (using results from 3D MHD simulations)
448: might yield slightly different results, but the same basic idea will apply also in that case.
449: In this example we have used a relatively strong magnetic field (1000 Gauss) to facilitate
450: a visual comparisons in Fig.~5, but it is worth mentioning that repeating the experiment with smaller fields leads
451: to the same result.
452:
453: \begin{center}
454: \includegraphics[width=8.5cm]{f5.ps}
455: \figcaption{Simulated intensity profiles of \ion{Fe}{1} 5257.050 \AA\. Thick dashed line, $I_1$
456: was obtained using a macroturbulent velocity of $v_{\rm mac}=2$ km s$^{-1}$. In addition to this,
457: we include a magnetic field with: $B=1000$ Gauss and $\gamma=0^{\circ}$, thus obtaining $I_2$ (thick solid).
458: Individual thin dashed lines, $I_k$, were obtained with $v_{\rm mac}=0$, $B=1000$ Gauss and $\gamma=0^{\circ}$
459: and a varying $v_{\rm los}$ from $-$4 km s$^{-1}$ to 4 km s$^{-1}$ to simulate a convective velocity field. All
460: $I_k$ are added up into $I_4$ according to Eq.~8 and 9.}
461: \end{center}
462:
463: To provide further support to our argument we have recalculated the effects of
464: the magnetic fields in the abundance (as done in section 3) but using a smaller
465: macroturbulence: $v_{\rm mac}=1$ km s$^{-1}$. The resulting plots are identical to
466: those in Fig.~2. This result might seem at odds with Eq.~7, since the new
467: macroturbulent velocity changes the values of $\Delta\lambda$ and $I_0$.
468: However, if we repeat the calculation of the calibration curves in Fig.~4 no
469: differences are observed in $\Delta\log\epsilon$ as compared to the case
470: with $v_{\rm mac}=2$ km s$^{-1}$. In order for this to happen, the only possibility
471: is that the horizontal axis changes. This results in a different calibration constant
472: $\mathcal{A}(\gamma)$ but equal errors in the abundance.
473:
474: %%
475: \section{Conclusions}
476: %%
477:
478: Ignoring the broadening caused by the magnetic field when fitting the intensity profiles of spectral lines,
479: may lead to an erroneous determination of atomic abundances in the Sun and other magnetically active stars.
480: Although there have been previous works where the magnetic field has been considered (Kochukhov et al. 2004;
481: Socas-Navarro \& Norton 2007), to our knowledge this is the first systematic study aiming at quantifying the
482: role of the magnetic field. Our results indicate that \ion{Fe}{1} lines are more affected than lines from elements
483: like \ion{Si}{1}, \ion{C}{1} or \ion{O}{1}. We have shown that vertical magnetic fields lead to
484: an underestimation of the real abundance, while horizontal fields tend to overestimate it. In a more real
485: situation, where the spectral lines would receive contributions from magnetic fields with different inclinations,
486: these two effect are likely to cancel each other, making the contribution of the magnetic field almost negligible with
487: the exemption of perhaps \ion{Fe}{1}. It is important to mention that the degeneracy between magnetic fields
488: and abundances occurs only for small fields, since for very large magnetic fields, the spectral line
489: is fully split in its different Zeemann components (see for example Nesvacil et al. 2004).
490: We have also developed a phenomenological model that can used to determine if a particular
491: spectral line is suitable for abundances studies that do not consider the effect of the magnetic field.
492:
493: Our analysis 1D LTE analysis neglects the effect of having variations in
494: temperature stratifications and/or convective velocities fields. Therefore,
495: our results apply only if all those others possible sources of errors
496: can be eliminated. If that was not the case, the error introduced by the
497: magnetic field would be of second importance. This situation is highlighted
498: by the work of Socas-Navarro \& Norton (2007), who in spite of consistently
499: considering the magnetic field, obtained strong discrepancies between
500: the inferred abundances in quiet solar regions (e.g.: granulation) and magnetic regions (e.g.: pores).
501: The source of those discrepancies is therefore to be ascribed to small ($< 100$ K) errors
502: in the temperature stratification.
503:
504: It would be ideal to repeat this work using realistic 3D MHD simulations (using
505: several initial magnetic fluxes) in order to model more realistically the different
506: temperatures, velocities, field strength and inclinations present in solar and stellar photospheres.
507: However, our work does provide a first hint that magnetic fields are an unlike source
508: of large errors in abundance determinations, unless a very particular spectral line
509: or a very particular magnetic configuration is present.
510:
511: \acknowledgements{I wish to thank Eberhard Wiehr for bringing
512: up the issue, during a talk I gave at the University of G\"ottingen in 2003,
513: of using only spectral lines with zero Land\'e factor in abundance determinations.
514: That warning led to consider the effects of the magnetic field whenever
515: those were not available. Thanks also to Carlos Allende Prieto and an anonymous
516: referee for important suggestions and comments.}
517:
518: \begin{thebibliography}{}
519: \bibitem[2002]{carlos1}
520: Allende Prieto, C., Asplund, M., L\'opez Garc{\'\i}a, R.J. \&
521: Lambert, D. 2002, \apj, 567, 544
522: \bibitem[1995]{anstee95}
523: Anstee, S.D. \& O'Mara, B.J. 1995, \mnras, 276, 859
524: \bibitem[2000]{martin1}
525: Asplund, M., Nordlund, A., Trampedach, R. \& Stein, R.F. 2000,
526: \aap, 359, 743
527: \bibitem[2001]{martin4}
528: Asplund, M. 2001, \aap, 2000, 359, 755
529: \bibitem[2004]{martin5}
530: Asplund, M., Grevesse, N., Sauval, A.J., Allende Prieto, C. \& Kiselman, D. 2004,
531: \aap, 423, 1109
532: \bibitem[2005]{martin3}
533: Asplund, M., Grevesse, N., Sauval, A.J., Allende Prieto, C. \& Blomme, R. 2005a,
534: \aap, 431, 693
535: \bibitem[2005]{martin2}
536: Asplund, M., Grevesse, N., Sauval, A.J. 2005b, in Cosmic
537: Abundances as Records of Stellar Evolution and Nucleosynthesis
538: in honor of David Lambert, ASP Conference Series, 336, 25.
539: Eds: Thomas G. Barnes III and Frank N. Bash.
540: \bibitem[1997]{paul2}
541: Barklem, P.S. \& O'Mara, B.J. 1997, \mnras, 290, 102
542: \bibitem[1998]{paul3}
543: Barklem, P.S., O'Mara, B.J. \& Ross, J.E. 1998, \mnras, 296, 1057
544: \bibitem[2000]{paul}
545: Barklem, P.S., Piskunov, N.E. \& O'Mara, B.J. 2000,
546: A\&AS, 142, 467
547: \bibitem[1987]{brault}
548: Brault, J. \& Neckel, H. 1987, Spectral atlas of the solar
549: absolute disk-averaged and disk center intensity from 3290 \AA\
550: to 12510 \AA\, available at ftp.hs.uni-hamburg.de/pub/outgoing/FTS-atlas
551: \bibitem[2005]{dani}
552: Cabrera Solana, D., Bellot Rubio, L.R. \& del Toro Iniesta, J.C. 2005,
553: \aap, 439, 687
554: \bibitem[2007]{castro}
555: Castro, M., Vauclair, S. \& Richard, O. 2007, \aap, 463, 755
556: \bibitem[2002]{nova}
557: Christlieb, N., Bessell, M. S., Beers, T. C., Gustafsson, B., Korn, A.,
558: Barklem, P. S., Karlsson, T., Mizuno-Wiedner, M. \& Rossi, S. 2002, \nat,
559: 419, 904
560: \bibitem[1969]{garz}
561: Garz, T., Holweger, H., Kock, M., Richter, J. 1969, \aap, 2, 446
562: \bibitem[1971]{hsra}
563: Gingerich, O., Noyes, R.W., Kalkofen, W. \& Cuny, Y. 1971, \solphys, 18, 347
564: \bibitem[1968]{grev}
565: Grevesse, N. 1968, \solphys, 5, 159
566: \bibitem[2004]{doppler}
567: Kochukhov, O., Bagnulo, S., Wade, G. A., Sangalli, L., Piskunov, N.,
568: Landstreet, J. D., Petit, P., Sigut, T. A. A. 2004, \aap, 414, 613
569: \bibitem[2005]{japo}
570: Ichimoto, K. \& Solar-B Team 2005, Journal of Korean Astronomical Society, 38, 307
571: \bibitem[1968]{lambert1}
572: Lambert, D.L. 1968, \mnras, 138, 143
573: \bibitem[1968]{lambert2}
574: Lambert, D.L. \& Warner, B. 1968, \mnras, 138, 181
575: \bibitem[1976]{egidio}
576: Landi Degl'Innocenti, E. 1976, A\&AS, 25, 379
577: \bibitem[2001]{bruce2}
578: Lites, B.W., Elmore, D.F. \& Streander, K.V. 2001, ASP Conf. Ser., 236, 33
579: \bibitem[2004]{rafa}
580: Manso Sainz, R., Landi Degl'Innocenti, E. \& Trujillo Bueno, J. 2004,
581: \apjl, 614, L89
582: \bibitem[1999]{neckel}
583: Neckel, H. 1999, \solphys, 184, 421
584: \bibitem[2004]{nesvaril}
585: Nesvacil, N., Hubrig, S. \& Jehin, E. 2004,
586: \aap, 422, 225
587: \bibitem[1992]{basilio}
588: Ruiz Cobo, B. \& del Toro Iniesta, J.C. 1992, \apj, 398, 375
589: \bibitem[2007]{hector}
590: Socas-Navarro, H. \& Norton, A.A. 2007, \apjl, 660, 153
591: \bibitem[2006]{bobstein}
592: Stein, B. \& Nordlund, \AA\ 2006, \apj, 642, 1246
593: \bibitem[2003]{jclibro}
594: del Toro Iniesta, J.C. 2003, {\it Introduction to Spectropolarimetry},
595: Cambridge University Press, ISBN 0-521-81827-3
596: \bibitem[2002]{javier}
597: Trujillo Bueno, J., Shchukina, N. \& Asensio Ramos, A. 2004,
598: \nat, 430, 326
599: \bibitem[2005]{alex1}
600: V\"ogler, A., Sheylag, S., Sch\"ussler, M., Cattaneo, F., Emonet, T. \& Linde, T. 2005,
601: \aap, 429. 335
602: \bibitem[1974]{axel}
603: Wittmann, A. 1974, \solphys, 35, 11
604: \end{thebibliography}
605:
606: \begin{onecolumn}
607: \begin{center}
608: \tabcaption{Summary of employed spectral lines; temperature
609: parameter $\alpha$ and cross section for collisions with neutral hydrogen
610: (in units of Bohr's radius $a_0$) from ABO theory (Barklem et al. 2000);
611: Land\'e factor g$_{eff}$ is calculated under LS approximation from the
612: electronic configurations of the upper and lower levels.}
613: \begin{tabular}{ccccccccccc}
614: \tableline
615: Atom & $\lambda_0$ [\AA] & $\chi_l$ [eV] & $\log(gf)$ & $\alpha$ & $\sigma/a_0^2$
616: & $g_{eff}$ & I$_0$ & $\Delta\lambda$ [m\AA] & Lower Level & Upper Level\\
617: \tableline
618:
619: %\begin{deluxetable}{ccccccccccc}
620: %\tablecolumns{11}
621: %\tablewidth{0pc}
622: %\tabletypesize{\footnotesize}
623: %\tablecaption{Summary of employed spectral lines; temperature
624: %parameter $\alpha$ and cross section for collisions with neutral hydrogen
625: %(in units of Bohr's radius $a_0$) from ABO theory (Barklem et al. 2000);
626: %Land\'e factor g$_{eff}$ is calculated in LS approximation from the electronic
627: %configurations of the upper and lower levels.}
628: %\tablehead{
629: %\colhead{Atom} & \colhead{$\lambda_0$ [\AA]} & \colhead{$\chi_l$ [eV]} & \colhead{$\log(gf)$} & \colhead{$\alpha$} & \colhead{$\sigma/a_0^2$}
630: %& \colhead{g$_{eff}$} & \colhead{I$_0$} & \colhead{$\Delta\lambda$ [m\AA]} & \colhead{Lower Level} & \colhead{Upper Level}}
631: %\startdata
632: \ion{Fe}{1} & 4389.245 & 0.052 & $-$4.583 & 0.249 & 217 & 1.50 & 0.338 & 52.30 & 5D3 & 7F2\\
633: \ion{Fe}{1} & 5247.050 & 0.087 & $-$4.946 & 0.253 & 206 & 2.00 & 0.464 & 59.36 & 5D2 & 7D3\\
634: \ion{Fe}{1} & 5250.209 & 0.121 & $-$4.938 & 0.253 & 207 & 3.00 & 0.471 & 59.21 & 5D0 & 7D1\\
635: \ion{Fe}{1} & 5412.798 & 4.434 & $-$1.761 & 0.280 & 971 & 0.97 & 0.845 & 60.04 & 3G4 & 5H4\\
636: \ion{Fe}{1} & 5525.544 & 4.230 & $-$1.084 & 0.238 & 748 & 1.50 & 0.559 & 66.65 & 5D0 & 5D1\\
637: \ion{Fe}{1} & 5701.544 & 2.559 & $-$2.216 & 0.237 & 361 & 1.12 & 0.439 & 70.38 & 3F4 & 3D3\\
638: \ion{Fe}{1} & 5784.658 & 3.396 & $-$2.530 & 0.244 & 796 & 1.87 & 0.781 & 64.04 & 5F3 & 5D4\\
639: \ion{Fe}{1} & 5956.694 & 0.859 & $-$4.605 & 0.252 & 227 & 0.70 & 0.621 & 64.91 & 5F5 & 7P4\\
640: \ion{Fe}{1} & 6082.710 & 2.223 & $-$3.573 & 0.271 & 306 & 2.00 & 0.745 & 65.48 & 5P1 & 3P1\\
641: \ion{Fe}{1} & 6136.994 & 2.198 & $-$2.950 & 0.265 & 280 & 2.00 & 0.545 & 70.27 & 5P2 & 5D1\\
642: \ion{Fe}{1} & 6151.618 & 2.176 & $-$3.299 & 0.263 & 277 & 1.83 & 0.639 & 67.74 & 5P3 & 5D2\\
643: \ion{Fe}{1} & 6173.335 & 2.223 & $-$2.880 & 0.266 & 281 & 2.50 & 0.535 & 71.16 & 5P1 & 5D0\\
644: \ion{Fe}{1} & 6200.313 & 2.608 & $-$2.437 & 0.235 & 350 & 1.50 & 0.522 & 73.08 & 3F2 & 3F3\\
645: \ion{Fe}{1} & 6219.280 & 2.198 & $-$2.433 & 0.264 & 278 & 1.60 & 0.436 & 77.74 & 5P2 & 5D2\\
646: \ion{Fe}{1} & 6240.646 & 2.223 & $-$3.230 & 0.272 & 301 & 1.00 & 0.634 & 68.97 & 5P1 & 3P2\\
647: \ion{Fe}{1} & 6265.133 & 2.176 & $-$2.550 & 0.261 & 274 & 1.58 & 0.456 & 76.98 & 5P3 & 5D3\\
648: \ion{Fe}{1} & 6271.278 & 3.332 & $-$2.703 & 0.247 & 720 & 1.50 & 0.821 & 69.31 & 5F5 & 7D5\\
649: \ion{Fe}{1} & 6280.618 & 0.859 & $-$4.387 & 0.253 & 223 & 1.45 & 0.565 & 70.22 & 5F5 & 7F5\\
650: \ion{Fe}{1} & 6297.793 & 2.223 & $-$2.740 & 0.264 & 278 & 1.00 & 0.507 & 74.31 & 5P1 & 5D2\\
651: \ion{Fe}{1} & 6322.685 & 2.588 & $-$2.426 & 0.238 & 345 & 1.50 & 0.520 & 75.02 & 3F3 & 3F4\\
652: \ion{Fe}{1} & 6481.870 & 2.279 & $-$2.984 & 0.243 & 308 & 1.50 & 0.587 & 73.46 & 3P2 & 5D2\\
653: \ion{Fe}{1} & 6498.939 & 0.958 & $-$4.699 & 0.253 & 226 & 1.37 & 0.698 & 69.81 & 5F3 & 7F3\\
654: \ion{Fe}{1} & 6581.210 & 1.485 & $-$4.680 & 0.245 & 254 & 1.30 & 0.859 & 68.69 & 3F4 & 5F4\\
655: \ion{Fe}{1} & 6593.870 & 2.433 & $-$2.422 & 0.247 & 321 & 1.15 & 0.496 & 79.91 & 3H5 & 5G5\\
656: \ion{Fe}{1} & 6609.110 & 2.559 & $-$2.692 & 0.245 & 335 & 1.15 & 0.589 & 75.30 & 3F4 & 3G4\\
657: \ion{Fe}{1} & 6750.152 & 2.424 & $-$2.621 & 0.241 & 335 & 1.50 & 0.542 & 79.42 & 3P1 & 3P1\\
658: \ion{Fe}{1} & 6945.205 & 2.424 & $-$2.482 & 0.243 & 331 & 1.50 & 0.519 & 84.01 & 3P1 & 3P2\\
659: \ion{Fe}{1} & 6978.851 & 2.484 & $-$2.500 & 0.241 & 337 & 1.50 & 0.536 & 83.43 & 3P0 & 3P1\\
660: \ion{Fe}{1} & 7723.208 & 2.279 & $-$3.617 & 0.242 & 304 & 1.17 & 0.794 & 82.56 & 3P2 & 3D3\\
661: \ion{Si}{1} & 5645.613 & 4.93 & $-$2.04 & 0.223 & 1791 & 1.75 & 0.748 & 74.38 & 3P1 & 3S1\\
662: \ion{Si}{1} & 5665.555 & 4.92 & $-$1.94 & 0.222 & 1772 & 0.00 & 0.708 & 75.76 & 3P0 & 3P0\\
663: \ion{Si}{1} & 5684.484 & 4.95 & $-$1.55 & 0.221 & 1797 & 1.25 & 0.570 & 82.41 & 3P2 & 3S1\\
664: \ion{Si}{1} & 5690.425 & 4.93 & $-$1.77 & 0.222 & 1772 & 1.50 & 0.646 & 78.42 & 3P1 & 3P1\\
665: \ion{Si}{1} & 5701.104 & 4.93 & $-$1.95 & 0.222 & 1767 & 1.50 & 0.716 & 76.06 & 3P1 & 3P0\\
666: \ion{Si}{1} & 5708.400 & 4.95 & $-$1.37 & 0.222 & 1787 & 1.50 & 0.509 & 87.37 & 3P2 & 3P2\\
667: \ion{Si}{1} & 5772.146 & 5.08 & $-$1.65 & 0.207 & 2036 & 1.00 & 0.660 & 81.16 & 1P1 & 1S0\\
668: \ion{Si}{1} & 5780.384 & 4.92 & $-$2.25 & 0.228 & 1691 & 0.50 & 0.819 & 74.28 & 3P0 & 3D1\\
669: \ion{Si}{1} & 5793.073 & 4.93 & $-$1.96 & 0.228 & 1703 & 1.00 & 0.723 & 77.01 & 3P1 & 3D2\\
670: \ion{Si}{1} & 5797.856 & 4.95 & $-$1.95 & 0.223 & 1755 & 1.16 & 0.727 & 77.27 & 3P2 & 3D3\\
671: \ion{Si}{1} & 5948.541 & 5.08 & $-$1.13 & 0.222 & 1845 & 1.00 & 0.491 & 96.52 & 1P1 & 1D2\\
672: \ion{Si}{1} & 7680.266 & 5.86 & $-$0.59 & 0.495 & 2106 & 1.00 & 0.631 & 126.0 & 1P1 & 1D2\\
673: \ion{Si}{1} & 7918.384 & 5.95 & $-$0.51 & 0.232 & 2933 & 0.75 & 0.643 & 141.8 & 3D1 & 3F2\\
674: \ion{Si}{1} & 7932.348 & 5.96 & $-$0.37 & 0.235 & 2985 & 1.00 & 0.610 & 150.3 & 3D2 & 3F3\\
675: \ion{Si}{1} & 7970.307 & 5.96 & $-$1.37 & 0.232 & 2927 & 0.92 & 0.883 & 117.0 & 3D2 & 3F2\\
676: \ion{C}{1} & 7111.469 & 8.640 & $-$1.074 & 0.313 & 1842 & 0.75 & 0.941 & 121.1 & 3D1 & 3F2\\
677: \ion{C}{1} & 7113.179 & 8.647 & $-$0.762 & 0.314 & 1857 & 1.12 & 0.903 & 125.4 & 3D3 & 3F4\\
678: \ion{C}{1} & 9603.036 & 7.480 & $-$0.895 & 0.236 & 561 & 2.00 & 0.752 & 175.4 & 3P0 & 3S1\\
679: \ion{C}{1} & 10753.976& 7.488 & $-$1.598 & 0.238 & 532 & 2.00 & 0.875 & 174.5 & 3P2 & 3D1\\
680: \ion{C}{1} & 11777.546& 8.643 & $-$0.490 & 0.271 & 746 & 0.92 & 0.854 & 214.1 & 3D2 & 3F2\\
681: \ion{C}{1} & 12549.493& 8.847 & $-$0.545 & 0.293 & 863 & 1.50 & 0.880 & 228.0 & 3P0 & 3P1\\
682: \ion{C}{1} & 12562.124& 8.848 & $-$0.504 & 0.293 & 863 & 1.50 & 0.874 & 230.6 & 3P1 & 3P0\\
683: \ion{C}{1} & 12569.042& 8.848 & $-$0.586 & 0.293 & 863 & 1.50 & 0.885 & 226.3 & 3P1 & 3P1\\
684: \ion{C}{1} & 12581.585& 8.848 & $-$0.509 & 0.293 & 862 & 1.50 & 0.874 & 230.9 & 3P1 & 3P2\\
685: \ion{O}{1} & 6158.176 & 10.74 & $-$0.30 & 0.322 & 1915 & 1.58 & 0.976 & 98.30 & 5P3 & 5D3\\
686: \ion{O}{1} & 7771.944 & 9.15 & 0.37 & 0.234 & 452 & 1.33 & 0.734 & 134.04 & 5S2 & 5P3\\
687: \ion{O}{1} & 7774.166 & 9.15 & 0.22 & 0.234 & 452 & 1.92 & 0.761 & 129.76 & 5S2 & 5P2\\
688: \ion{O}{1} & 7775.388 & 9.15 & 0.00 & 0.234 & 452 & 1.75 & 0.800 & 124.15 & 5S2 & 5P1\\
689: \tableline
690: \end{tabular}
691: \end{center}
692: \end{onecolumn}
693:
694: %\enddata
695: %\end{deluxetable}
696:
697:
698:
699:
700: \end{document}
701: