Identifying Unknown Solar Absorption Lines

Todd F. (Grosse Pointe North High School) & Regan W. (Grosse Pointe South High School)

Introduction

Despite one hundred years of research into the solar spectrum, there are still many absorption features that have not been identified. These absorption features directly correspond to identities of elements. Since the last time (1970’s) that major identifications were made on these lines, much work has been done on all the of the atomic spectrum, especially recently with new line identifications being made in the rare earth and transition elements. This project used a synthetic spectrum to identify unknown absorption lines in the real spectrum of the sun.

Solar Spectrum with Black Absorption Lines

Previously unknown absorption lines of transition metals and rare earth elements in the sun can now be identified. In addition, all solar models currently factor out unidentified lines. These new identifications will affect the elemental abundance estimates so future solar models will have to be revised.

Procedure

The first step of the process was to meet with Professor Donald Bord of the University of Michigan and learn generally about the solar spectrum and how to identify elements by their absorption lines. Professor Bord then explained how to submit a query to the Vienna Atomic Line Database (or the VALD).

The VALD is a collection of atomic line parameters of astronomical interest and it also provides tools for selecting subsets of lines for typical astrophysical applications: Line identification, chemical composition and radial velocity measurements, model atmosphere calculations, etc. In addition, the French BASS 2000 web site, and the Photometric Atlas of the Solar Spectrum could be used to examine the lines in the current spectrum.

After examining the solar spectrum the query starting point was placed in the ultraviolet range, because the line density is greater and there are more unidentified features. This area also has less telluric lines, absorption bands that are caused by molecules in the Earth’s atmosphere. Since these aren’t caused by the photosphere of the sun, using them in the data is out of the question because they are distinctly a foreign piece of information. Also, they are molecular or polyatomic bands, so their activity clearly differs from that of the atomic lines in the sun.

With all this in mind a query to the Vienna Atomic Line Database (VALD) was made through the use of email. Using "Extract Stellar" mode, the default parameters were solar, so few changes needed to be made. VALD would create a synthetic spectrum for the portion queried, with all lines identified, according to their known behavior on Earth. This would allow the unidentified features in the solar spectrum to be identified.

For each VALD query, the synthetic spectrum was compared to the actual spectrum. The features in the observed spectrum were matched to their corresponding lines in the virtual spectrum. To identify the unknown lines, a graph of that portion of the spectrum would be obtained using the Photometric Atlas of the Solar Spectrum, a reference book that plotted the entire spectrum on paper and the BASS 2000 web site. This web site had the entire spectrum online. The position and range of the portion in question could be changed in seconds, saving valuable time.

If the lines had double-peaks, round tips, misshapen slopes, fat bottoms, or were asymmetrical at all, the VALD identification could not be taken as absolute. These blends, or features that are more than one line near the same position, that add together to produce an asymmetrical feature. Blends were ruled out and taken as unidentifiable.

The position of the VALD line would need to have an acceptable agreement in position with the observed line, and the unknown observed line would need to have the correct placement between two known features, to assure it was a single line, and be good-looking. If these checks had passed, the VALD’s identification of the line could be assured as proof positive.

The differences between the position of the lines in the query and the actual spectrum were recorded on the "known lines" spreadsheet.

There were lines in the observed spectrum already identified in 1966, but the VALD definition of the line disagreed. In some cases, the observed feature would have a question mark after it. This indicated that the line was thought to be of that identification, but it wasn’t an absolute. The lines where this occurred were Fe 1 or Fe 2. In the past decade, there have been intensive studies on the iron lines in the solar spectrum in Sweden. Since the VALD is updated to this, and the observed features book is rather outdated, the VALD definition was taken. On lines that didn’t agree, and didn’t have a question mark after them, the lines bracketing that feature were examined. If they were identified in both the VALD and the observed spectrum, and, were an acceptable distance away from the line in disagreement, then the VALD definition was taken. If the line didn’t meet the above parameters, the observed definition was taken, because the VALD could not be used unless it was without a reasonable amount of error. The solar spectrum research was dated 1966, and the VALD model has revisions made, as of the 1993 solar model. This is the reasoning behind favoring the VALD definition during disagreements.

The differences in line positions of the observed and virtual spectra were figured next. This was easily accomplished by subtracting the observed from the virtual. Using this, the standard deviation was calculated for the observed and virtual spectrum’s known and newly identified line differences. Standard deviations of the differences from the known versus the newly identified lines were compared. The standard deviations of the new identifications were smaller than and in good agreement with the standard deviations of the known lines.

Next, the line depths of the absorption features were calculated. The line depths are expressed in decimals, which reflect a percent of absorption in the solar spectrum. For example, 0.20 is equal to 20% absorption at that frequency of the solar spectrum. The VALD returned line depth data for every feature. To measure line depth in the actual solar spectrum, the BASS 2000 website was used. A five-angstrom portion was printed and line depths of matching absorption features were measured.

For some of the newly identified elements in this research, new solar abundance estimates were calculated. The VALD was used to determine the new abundance by sending multiple queries, each time revising the element’s abundance estimate until the line depth in the query matched the line depth in the sun.

Data

Ten separate queries were made to the VALD, three of which were returned to be unusable. The data were returned without any lines and a message that that portion of the spectrum was unusable. The first queries were made in the lower regions, or ultraviolet, and the queries would go higher up each time, ending with a spot just below the infrared range. The areas in the green portion of the visible light were determined unusable. There were too many large absorption features. The large absorption features block out the smaller unidentified lines, making them truly unidentifiable.

The specific queries covered these ranges of wavelengths: 3300-3505, 5700-5703, 6067-6072, 6209-6217, and 6265-6270 Angstroms. In total, 2849 lines of data were processed.

The spreadsheet titled "Known Matched Absorption Lines," has all of the pairs of lines that matched between the observed spectrum and the query. The first column has the observed line position in angstroms. The second column has the line position of the created spectrum in angstroms. And the third column has the difference between the two. The next is the elemental identification column. This gives the element and its ionization number. The ionization number of one is neutral, number two would be missing one electron, number 3 would be missing two electrons, etc. The fifth and sixth columns are the line depths for the observed spectrum and the query. Following those two columns is the line depth difference. The next column is the indication column. This shows all the new lines that have been identified, and shows all the revisions of lines that were wrong in the old model.

Analysis

After counting the number of unidentified lines in the sun, they totaled 6126 lines. Factoring in the 26 new lines which were identified by this research, the change in the number of unidentified lines in the solar spectrum is 0.042%. In addition, 16 lines were revised from their previous identities.

Calculated Abundances of Elements

Conclusions

Twenty-six new lines were identified: rare earth elements (erbium, dysprosium, samarium, cerium) and transition metals (vanadium, osmium, manganese, nickel, iron, cobalt, titanium, chromium, and tungsten) and the element calcium were found. This caused a change in the number of unidentified lines in the solar spectrum of 0.042%.

Sixteen line identifications were revised involving transition metals (cobalt, chromium, titanium, nickel, ruthenium and iron) and rare earth elements (samarium, thorium, gadolinium, and thulium). Because of the discoveries of these new and revised lines, changes will have to be made to the current solar model.

New & Revised Lines

Bibliography

Delbouille, L., Neven, L., Roland, G. Photometric Atlas of the Solar Spectrum from 3000 to 10000. The Institut d’Astrophysique de l’Universite de Liege, Conite-Ougree, Belgium: 1973.

Moore, Charlotte E., Minnaert, M.G.J., Houtgast, J. The Solar Spectrum 2935 A to 8770A. National Bureau of Standards Monograph 61, Dec 1996.

Online Solar Line Database: astro.lsa.umich.edu/hub/get/bord

Vienna Astrophysical Line Database: http://www.astro.uu.se/~vald

BASS 2000: High Resolution Solar Spectrum: http://mesola.obspm.fr/form_spectre.html