Univ. of Wisconsin - Madison
Madison, WI 53706
Just for comparison, the following are pictures of H alpha emission (an optical wavelength) from these two galaxies. I got these images from the NASA Extragalactic Database.
Clearly, the Level 1 X-ray data files need to be processed. The first step in cleaning up the data is to create a level 2 event file, applying the most recent instrument calibration files. We also need to filter the data to include only a certain energy range. At energies higher than about 8 keV, the detector is not accurate. We get a very increased count rate, but that data is not real. The response of the detector is also faulty at energies lower than about 0.3 keV. Knowing this, we filtered the data to include only the meaningful energy range.
Not only must we filter over energy, we also need to filter over time. Chandra is known for having times of background flares. During these times, the count rate is much higher than average, so we can't trust the data that we get. To remove them from the data set, we extracted a light curve from a background region. This region has no source in it, so the count rate should be uniform.
In this lightcurve, we can clearly see three spikes in the count rate, which represent the times of background flares. Once we have isolated these times, we can remove them from the data.
The cleaned up version of the data looks like this:
In NGC 3351 we notice a bright central source. This emission is coming from the central bulge of the galaxy. In NGC 4736 we can see a central bulge as well as some emission coming from the disk. However, in both images, we see several point sources. We needed to remove these sources before we could continue our analysis. To do this we used a CIAO package called wavdetect. Wavdetect finds potential point sources, and creates elliptical regions around them. However, the point source detection is not perfect; we still went through the regions that were created and used our own judgment as to whether or not the sources detected were actual point sources or simply clumps of diffuse gas. We ended up with an image that looked like this:
We used a CIAO package called Sherpa to display and fit the spectra. The fits for NGC 3351 contain a powerlaw, two thermal components and a Gaussian. The Gaussian was used to fit the emission line which can be seen on the right. The power law fits the spectra from any unresolved point sources, like low mass X-ray binaries, which were not removed. Although Chandra resolution is good, it is still impossible to identify all of the point sources.
Apart from the power law and Gaussian, we fit two thermal components to the spectra. These components find the temperature of the hot gas (or rather, they find kT for each component). In fitting two thermals, we assumed that the hot gas is only at two different temperatures. This is clearly not physically accurate- the gas in the galaxies should be at a continuum of temperatures. However, it is impossible to fit a thermal continuum to the spectra, because we do not know much about the physical parameters of the gas, such as pressure and density. Without this a priori knowledge, we cannot find a continuum of temperatures. We chose to fit two thermal components because this gives a better fit than fitting only one component. However, fitting three does not drastically change the statistical accuracy. For NGC 3351, we had three different data sets. We fit all three of these independently, and found that the results were consistent between them. Moreover, the statistical values of the three fits were 24.6, 24.5, and 38.6, with 38, 40, and 45 degrees of freedom, respectively. So we can be fairly confident that the kT's for the two thermal components are 0.28 and 0.77 keV. We also found an average flux of 1.01e-13 ergs/cm^2/s from the bulge of NGC 3351.
NGC 3351 ObsID 5929
NGC 3351 ObsID 5930
NGC 3351 ObsID 5931
The fits for the bulge and disk of NGC 4736 were a little more difficult than those for NGC 3351. In the disk of NGC 4736, the data was not very accurately fit by two thermal components and a power law. We attempted several different combinations of two power laws and three thermal components, but all of the fits had approximately the same statistical error. More analysis should be done to truly determine the components of the spectra for the disk of NGC 4736. However, the bulge of NGC 4736 was fit seemingly accurately by a power law and two thermal components. So, we were able to say with some confidence that the bulge contained two thermal components with kT's of 0.39 and 0.71 keV. However, more analysis must be done to compare the thermal emission from the disk with these results. We also found the flux of the bulge of NGC 4736 to be 2.44e-13 ergs/cm^2/s.
NGC 4736 Bulge emission
NGC 4736 Disk emission
In this equation, I is the intensity in Raleighs, and T is the temperature of the gas in Kelvin. This formula applied to emission from the H alpha line, but we used it here as a first approximation for our data.
Here, f is the filling factor, and h is the path length. For the purpose of our calculations, we assumed a filling factor of 1 and a path length of 200pc. In actuality, gas at two different temperatures should not have the same path length, but at high temperatures the filling factor can be approximated as 1.
In this equation, k is Boltzmann's constant.
Results for NGC 3351
at 0.28 keV:
T=3.25 million Kelvin
at 0.77 keV:
T=8.94 million Kelvin
Results for NGC 4736
at 0.39 keV:
T=4.53 million Kelvin
at 0.71 keV:
T=8.24 million Kelvin
Doane, et al. "The Origin and Distribution of Diffuse Hot Gas in the Spiral Galaxy NGC 3184". The Astronomical Journal. 128:2712-2723. 2004 December.
Bellm, et al. "Origins of the 1/4 keV Soft X-ray Background". The Astrophysical Journal. 622:959-964. 2005 April.
Swartz, et al. "Chandra Observations of Circumnuclear Star Formation in NGC 3351". The Astrophysical Journal. 647:1030-1039. 2006 August.
CIAO 3.4 Here one can learn everything they need to know about calibrating Chandra data.