Supersoft Sources: An X-tra Galactic Study in M31

Introduction

In 1978, the first space-based fully imaging X-ray telescope, HEAO-2, was launched. The telescope, later renamed Einstein, was unique in that its range included some soft X-rays, whereas all previous satellites had concentrated on hard X-rays. We define soft X-rays as in the range 0.09-2.5 keV, and hard X-rays as up to a few hundred keV.

Picture courtesy NASA Goddard Library.

Three years later, Long et al.^[7] published a paper about a survey of the Large Magellanic Cloud that had been carried out with Einstein. In it, they reported the discovery 10-25 unidentifiable objects. These X-ray sources had very soft spectra, but were very luminous (10³⁵ to 10³⁶ ergs s^-1). These objects had never before been seen in the Milky Way, and did not match anything in catalogues.

At first, it was suspected that these supersoft sources (SSSs) were accreting black holes^[2], which can have softer spectra than neutron stars. However, when, in 1990, ROSAT (a German X-ray satellite) was launched, it was discovered that the temperatures of these SSSs were much less than they should have been if indeed the SSSs were accreting black holes. Also, the radius of these objects was much greater than any neutron star could be - of the size (and larger) than typical white dwarfs.

In 1991, van den Heuvel et al.^[11] published a paper discussing and analyzing the various scenarios which could result in the high luminosity, soft spectrum combination observed in these sources. The idea that SSSs are accreting neutron stars or black holes is dissected early on, and comes up short - the peak energies of the aforementioned are between 1 and 10 keV, whereas the SSSs peak between 30 and 50 eV. This indicates that SSSs hav radii ~10²-10⁴ times bigger than a neutron star, again, the size of a white dwarf. The relatively low peak energy also explains why SSSs have not been observed in the Milky Way: an SSS would have to be within 1 kpc in order to be detected, as interstellar absorption would extinguish them at greater distances.

So what is an SSS? What do we think is happening? The current "definition" of an SSS is a "persistent or transient source whose radiation is almost completely emitted in the energy band below 0.5 keV and whose bolometric luminosity is 10^37-38 ergs s^-1" ^[4]. There is evidence that all of these sources may be white dwarfs surrounded by a thin, hot atmosphere, burning hydrogen in an accreted layer on the surface. Within this group, there are single white dwarfs, for instance, planetary nebulae nuclei or hot PG1159 stars, and there are white dwarfs in a binary system. This latter category includes post-nova remnants burning hydrogen, symbiotic stars (where the white dwarf is in a wide orbit around the red giant), and "close binary systems with a white accreting from a 1-2 solar mass main sequence secondary"^[4].

Picture courtesy NASA.

In the above picture, a white dwarf is accreting matter from a red giant, either because the red giant has filled its Roche lobe, or, if the red giant is smaller or has a wider orbit, it can power an SSS with wind. It can also happen that a main sequence star feeds the accretion disk, if it is in a tight orbit and overflows its Roche lobe.

Nuclear burning on an accreting white dwarf will occur when the accretion exceeds a certain rate. This burning of hydrogen on the surface of the white dwarf is the source of the supersoft X-rays. Ordinarily, such soft X-rays are not observed, as nuclear burning occurs deep within a star, and these photons are either The hydrogen burning becomes stable when the accretion rate is 0.4 dM_RG/dt &le dM_WD/dt &le dM_RG. If the accretion rate is less than this lower bound, hydrogen burning occurs in flashes, or the matter simply builds up on the surface of the white dwarf, until it exceeds the Chandresekar limit. If, on the other hand, the accretion rate exceeds said limit, as mass accumulates, the photospheric radius of the accreting star increases, and eventually fills its Roche lobe. At this point, any more matter that accretes would most likely be ejected. Since no more matter can accrete, the fuel on the surface will run out, and nuclear burning will cease, causing the envelope to collapse, and allowing matter to once again accrete. In their 1992 paper, van den Heuvel et al comment that small changes in accretion rate will probably have large effects on the X-ray spectrum^[11]. In general differences in SSSs come from various fields, such as mass of the companion, mass of the white dwarf, the stage of evolution of the companion, the orbital inclination, etc.

We estimate that there are about several thousand SSSs in the Milky Way, with a few created, and a few dying, every thousand years^[6]. What happens to an SSS at the end of its lifetime? The accretion and fusion of matter on the surface of the white dwarf obviously increases its mass. The white dwarf could reach the Chandrasekar limit (1.44 solar masses), and either collapse into a neutron star, or explode in an enormous nuclear fireball, depending on the properties of the white dwarf. Or, the white dwarf may simply reach the mass limit and explode. Or, the helium on the surface of the white dwarf could reach some critical mass, and ignite itself in an explosion. This would create shock waves, which would start carbon fusion at the center of the white dwarf, and explode in the runaway reaction.

S8: An SSS in M31?

In the late 1990s, P. Kahabka screened the 1991 ROSAT PSPC M31 X-ray point source catalogue in hopes of finding candidate supersoft sources^[5]. M31, the Andromeda galaxy, makes a good choice for such studies, as the amount of interstellar matter between us and Andromeda is relatively low, and Andromeda is relatively close by. He used a hardness ratio as a basis for his selection, and checked to make sure that the observed count rate was reasonable for the expected steady-state luminosity of a source with such a hardness ratio. From his list of candidates, we chose to look at SWh, what we call S8, located at 00:41:49.9 +40:59:21. Unfortunately, the error margins for the coordinates given by ROSAT were large, so there are three candidates:

S8 was detected as an SSS, so we want to see if it is a binary system. We do this by checking the optical variability of the source. If regular optical variability is observed, it is an excellent indicator of a binary system. The optical data was taken at WIYN with the Mini-Mosaic Imager. This particular object really pushes the limits of the telescope, it has M_B&asymp 24, and is located in a dense starfield of M31.

So why do we care? Why do we want to study SSSs? First of all, SSS are interesting objects in and of themselves. We certainly do not understand them completely, and continuing analysis in multiple wavelengths will help to chip away at the riddles posed by SSSs and the evolution of close binaries. Further, by identifying the optical counterparts (by way of spectra, color indices, and orbital periods), we hope to ultimately find out the basic characteristics of these systems, and use these properties to figure out their final fate. These characteristics include the mass of the white dwarf, effective temperature of the white dwarf when it is not too bright in X-rays, mass transfer rate, whether or not there can be thermonuclear flashes without mass ejection...

At this point, we suspect a fundamental connection between SSSs and type Ia supernovae. Type Ia supernovae are important in calibrating the cosmic distance scale and the rate of expansion of the universe.

Data Reduction and Photometry

As stated above, the optical data came from the Mini-Mosaic Imager at WIYN. I received raw data from ten nights which spanned over three years (2004-2006). The first step was to use IRAF (Image Reduction and Analysis Facility) to reduce the images. The original .fits files have to be corrected for a number of different things, such as bad pixels, dust, noise, etc. I used a reduction recipe written by Aaron Steffen to trim the images, and combine and subtract the biases and dome flats.

The left hand is clearly raw, whereas the right hand image is reduced and, hence, useful.

After the reduction is done, we can perform basic photometry. For this, I used DAOphot, originally written by Peter Stetson, along with a worksheet developed by Laura Chomiuk^[1]. Basically, DAOphot performs PSF photometry on the images. It finds stars in the .fits file based on the parameter set by the user, such as full-width half-max, minimum and maximum photon count, background level, and generates a files with the coordinates of the stars, their magnitudes, etc.

The original .fits image.


The PSF subtracted image.

DAOmatch is the next step. It uses a reference frame and the the method of matching triangles to "match up" the frames. It picks the thirty brightest stars in each image, and attempts to cross-identify them. DAOmatch can match up frames from different filters and even different telescopes. Lastly, DAOmaster uses a reference image to correct/scale the magnitudes of the objects. It also requires that the user set certain criteria, such as the minimum number of frames that the object must be detected in, and the maximum sigma. For the latter, it is useful to plot the magnitudes against the errors from the .als file produced by DAOphot. The plot looks something like:

The turnoff point, where the plot starts to slope very steeply, is a good choice for the maximum sigma. In the end, DAOmaster produces a file with the corrected magnitude for each of the detected objects - the file I needed to perform my analysis.

Once I had the file with the corrected magnitudes, I picked five reference stars of similar (and, of course, constant) magnitude to our candidates. I then used IDL to plot the data in a number of different ways. The most useful is displayed below. It shows the average difference in magnitude of each of the candidates with respect to the five reference stars. The error is the standard deviation.

Results

Data from September 2004.


Data from October 2005.


Data from October 2006.

As we can see, the middle is variable! This is an accomplishment, but, unfortunately, we still do not know if S8 is even an SSS. S8t was not detected by any other X-ray missions, and thus there is some suspicion about its identification.

References and Links

[1] Chomiuk L., "AY257 Problem Set 3 - Point Source Photometry."

[2] Cowley A. P., Crampton D., Hutchings J. B., et al. 1984, ApJ 286, 196

[3] Crampton D., Hutchings J.B., Cowley A.P., Schmidtke P.C., 1997, ApJ 489, 903

[4] Greiner J., Orio M., Schwarz R., 2000, A&A 355, 1041

[5] Kahabka P., 1999, A&A 344, 459

[6] Kahabka P., van den Heuvel E.P.J., Rappaport S., 1999, Sci. Am. 280, 46

[7] Long K.S., Helfand D.J., Grabelsky D.A., 1981, ApJ 248, 925

[8] Orio M., Della Valle M., Massone G., Ogelman H., 1994, A&A 289, L11

[9] Orio M., Della Valle M., Massone G., Ogelman H., 1997, A&A 325, L1

[10] Pakull M. W., Beuermann K., van der Klis M., van Paradijs J., 1988, A&A 203, L27

[11] van den Heuvel E.P.J., Bhattacharya D., Nomoto K., Rappaport S.A., 1992, A&A, 269, 97

Alphabetical List of IDL Routines

Einstein (HEAO-2) Observatory

Mini-Mosaic Imager

ROSAT

WIYN Observatory