Preliminary  Capella analysis

(D. Huenemoerder  3 September 1999) 
(Updated 1 October 1999) 

This is still for gawking at, but is now a little bit better starting-point for detailed investigations.
Data have beeen re-processed w/ better calibrations. (but there are still tweaks and improvements to do!)

Data are from the HETG/ACIS-S Focus Test and from the Emission Line Project

Top: obsids 1103, 1318, and 1235, merged MEG +1, -1 (about 90ks worth). Some HEG detail are shown in insets (also merged +1, -1 orders).

These spectra have updated chip-gap positions, a CCD gain table appropriate to the epoch (for 1103 and 1318, and is OK for 1235), and a wide order-sorting table, appropriate to the degraded CCD resolution.

There is still a trend at long wavelengths for wavelengths to be too long (by about 0.015A in the MEG 1st order.) This is still being investigated.

Bottom: The observation with a model of the Capella spectrum overplotted (not broadened via a simulation) based on the EUVE-determined emission measure distribution (Brickhouse 1995) and the SPEX emissivities, and a continuum hacked from NEI code.

Line IDs in the top plots are PRELIMINARY!     They were done via automatic search of the emissivity table for the strongest emitter within the profile region. They need careful review. In the bottom, the lines labeled are thosed actually binned by the model.

The images are in a wide table - scroll to see them all!

The whole thing:

Given the emissivity database, a nominal effective area, exposure time, and counts in each feature, we can sum all the emissivities which contribute to the feature and compute an diagram of the EMT (Emission Mess vs Temperature). Each quasi-parabolic line represents the emission measure vs. temperature of all lines which contribute to the wavelengths of the feature (as present in the SPEX line-list, for zero-density, Solar abundances). The solid, thick histogram is the model distribution, taken  from the EUVE fit (Brickhouse 1995). (The histogram model DEM is scaled by delta log T, which has the value 0.1; the derived curves have been scaled by 0.1 to match this distribution.) Untangling this Emission Mess will take some time, since each curve's lines need to be understood, and correlations between curves (line-ratios) need to be examined. In addition, lines not present (upper limits) also can constrain the emissivity.

Another set of potentially useful diagnostics are line ratios, though one must be careful with temperature distributions, since ratios are often quoted for isothermal plasmas. Such as the following for OVIII/OVII:

Some ratios are density sensitive, particularly the forbidden-line to intersystem-line ratio in Helium-like ions. Here is the ratio as modeled with XSPEC, and the measured ratio in Capella for O VII (f(22.1)/f(21.8)):

Postscript plots (for best viewing, use swap landscape, and turn anti-aliasing off):
Capella detail,
Capella detail, with model lines overlayed,

A simulation: this is the instrumentally broadened version of the discrete line spectrum shown above in the data-model comparison, in counts for 70 ks.

The MEG +- first order counts spectrum, for comparison with the simulation:

The linked data represent the best processing to date, but still require some improvement on gain correction and order-sorting!

Please consult the Emission Line Project site for latest CXC and community approved data and information.

Binned spectra, for three observations (OBSIDs 1103, 1235, and 1318). These are not-quite-OGIP compatible "PHA Type II" files. The specification has been extended to allow generic axes called BIN_LO and BIN_HI, which by default are linear in Angstroms (as specified by FITS header keywords).

Nominal effective areas (Note: "nominal" means nominal aim-point and nominal dither pattern. For careful analysis, the effective area (ARF) should be derived for each observation.)
These are OGIP ARFs (Auxiliary Response Files). The energy columns map to a linear wavelength scale on the same grid as the PHA2 files, but in reverse order. (OGIP ARFs require columns called ENERG_LO, ENERG_HI; for Chandra grating spectra, we prefer a linear wavelength scale. These compatibility issues are being worked.)
HEG, order -1:acisheg-1D1999-09-22arfN0000.fits
HEG, order +1: acisheg1D1999-09-22arfN0000.fits
MEG, order -1: acismeg-1D1999-09-22arfN0000.fits
MEG, order +1: acismeg1D1999-09-22arfN0000.fits


Brickhouse, N.S., 1995, Harvard-Smithsonian Preprint No.4119, and IAU Colloq.152.