Spectroscopic measurements of Mira variable stars, as a function of phase, probe the stellar atmospheres and underlying pulsation mechanisms. Modeling  the atmospheres is difficult due to the hydrodynamic nature of the gas as deduced from the large light variations and velocity measurements of various spectral  lines. Many questions still need to be resolved concerning the atmospheres of these stars. Are the depths of formation of the molecular species such as TiO,  VO, and ZrO produced in an extended region above the layers where Balmer line emission occurs or below this shocked region? What is the explanation for the Balmer-line increment, where the strongest Balmer line at phase zero is H-delta and not H-alpha? Furthermore, why is the H-epsilon line virtually absent in the spectra of Miras when the other Balmer lines are strong? Click here for one of our published papers.

I began in January 1996 low resolution (1.08 Angstrom/pixel) spectroscopy of Mira variables from about 6000 Angstroms to 8750  Angstroms. The spectra are taken in a region which includes H-alpha, TiO, VO, ZrO, and the CaII infrared (IR) triplet. Spectra of a dozen Mira variables  observed at more than one phase are presented. We investigate the final question listed above by noting variations in the CaII IR triplet in relationship with  H-alpha variations as a function of phase. These preliminary observations suggest that H- epsilon's observational characteristics result from an interaction of H- epsilon photons with the CaII H line.

The most recent paper (November 2000 issue of the Astronomical Journal) includes 4 years of observations. I use the Optomechanics Model 10C spectrograph. The spectrograph uses a 600 g/mm grating, with aresolution of about 2.8 Angstroms  (1.08 A/pixel, on a 768 x 512 CCD with 9 micron square pixels), blazed at 8500 Angstroms. I take spectrum in the range from 6200 to 8800 Angstroms of Mira stars. The spectrograph has been used on both the SARA 0.9-m telescope and the Appalachian State University Dark Sky Observatory (DSO) 0.45-m telescope. It currently resides at DSO where Dr. Dan Caton (DSO Director) and I have developed a collaboration to do simultaneous spectroscopy and photometry of Miras. 


The  image on the left shows the SARA 0.9-m telescope on the left and the DSO 0.45-m telescope on the right

On the left, the upper image shows the raw spectrum of R CVn as recorded on the CCD camera. The lower image is the sum of columns along the length of the CCD.  Note the H-alpha emission feature.


The figure below shows a time series of R Leo spectra taken over a period almost 2 years.  Note the changes in the absorption features.


 Undergraduates have been important colleagues in this research, through NSF REU Supplements, Independent Study, as students at PARI, or in  collaboration with other mentors/students. Students at ETSU made significant contributions  to this work. Brian Heaton and Kevin Crowe have developed the IDL code necessary for wavelength calibration of the spectra. Eric Lingerfelt, Clayton  Clark, and Rose Patrick have extended the program to include new Mira variables, and enhanced the program with BVRI photometry. Michael Bales has  experimented with a novel spectrograph slit design for spectrophotometry. Erica Messer, from Valdosta State University spent the summer 1998 at ETSU as part of the SARA REU Program. She and I spent the summer acquiring spectra. Dr. Don Luttermoser and his SARA REU student, Rob Piontek, used the   spectra, for the first time to determine log(g) and effective temperatures.  Marie Rinkoski, another SARA REU student during the Summer 2000, took spectra, wrote a phase prediction program, and used the LTE code ATLAS to approximate the stellar atmosphere synthetic spectra.

Marie Rinkoski calculated, using ATLAS, log(g) and plotted Log(g)
vs Phase shown above.  The plot shows minimum log(g) at phase 0.5

Marie Rinkoski also calculated the effective temperature versus phase.
The coolest temperature is at minimum light. Note that ATLAS is
LTE which provides only a rough approximation to Log(g) and Teff.

This research has been supported through an NSF CAREER grant, AST- 9500756, an American Astronomical Society Small  Research Grant, and two NSF REU Supplements.

During the summer 2001, Andreas Schwietzer at the University of Georgis-Athens and his REU student Marcus Woo used our Mira spectra to model the stellar atmospheres.    Woo and Schweitzer created a grid of model atmospheres to reproduce phase-resolved spectra of three Mira variables, R Leo, R CVn, and V CVn.  The models produce good fits to the observed spectra. We have determined effective temperature variations with phase, and they are consistent with independent  studies. However, changes in log(g) are too small to resolve.  They used the AMES-dusty models that considerdust opacity in the atmosphere.   The models confirmed V CVn and R CVn to be hotter stars than R Leo.  The models also reflect the difference in visual magnitude variability in the three stars.  The effective temperature and visual magnitude range for R CVn is largest, while the temperature and visual magnitude range for R  Leo is the smallest. The models reproduce the VO and TiO bands well, with some deviations in regions of less flux. The determined effective temperature variations range from 2400-3100 K in R Leo, 3400-3600 K  in V CVn, and 2800-3600 K in R CVn. Changes in log(g) cannot be resolved.


Above, R Leo Model by Woo and Schweitzer (red line) overlayed on the observed spectrum (black line).


Above, R CVn Model by Woo and Scwheitzer (green line) overlayed on the observed spectrum (black line)

[Home] [Research] [Miras] [IR Polarization] [Instrumentation] [Proplyds] [North Sky] [Masers] [Reference Stars] [Education] [CV]

November 2004. mwc