The AXAF transmission gratings (HETG and LETG) will revolutionize the
field of X-ray astrophysics, allowing us to observe high resolution
X-ray spectra from a wide variety of sources. To date, only a
handful of astrophysical X-ray sources have been observed with
energy resolution 
and all of those, except for the
Sun, were observed in only a few narrow energy bands centered on
specific lines. In contrast, AXAF grating spectra will cover the
spectral range from 0.1-10 keV with energy resolution 
, revealing this important spectral band in unprecedented
detail. For the first time, a wide range of spectral diagnostics will
be available to X-ray astronomers; emission line diagnostics can be
used to determine temperatures, densities, elemental abundances and
other parameters of non-solar collisionally ionized plasmas,
recombination-driven lines and continuum emission from X-ray
photoionized plasmas can be used to map out the ionization parameter
distribution around the ionizing source, and intrinsic X-ray absorption
features can be used to characterize the environments of AGN.
Figure 38: The effect of source extent on spectral
resolution. The input spectrum (panel a) for the MARX simulations shown
in panels b-e is a SPEX non-equilibrium plasma with an electron
temperature of 1 keV. The count rate normalizations in panels a-e are
arbitrary. Panel b is a simulated HEG observation of a point source.
Panels c and d are simulated HEG observations of a uniform surface
brightness disk with radius 4" and 8" respectively. The spectral
resolution obtained in these grating observations may be compared with
the spectral resolution expected from the ACIS CCD (panel e).
Figure 38 illustrates the spectral resolution of grating spectra -- the strongest emission features appear well isolated or easily separable by standard multi-component deconvolution techniques. Although the full energy resolution of the diffraction gratings will be realized only in observations of point sources (Figure 38b), it will also be possible to observe slightly extended sources at reduced energy resolution. Figures 38c and 38d illustrate the loss of energy resolution in observations of extended sources compared with the point source spectrum of Figure 38b. Roughly speaking, the gratings will be most useful for sources smaller than about 10 arcsec in diameter, although it may be possible to apply deconvolution techniques to improve the energy resolution obtainable from observations of more extended sources. Detailed simulations using MIT-MARX (Wise et al. 1997) can be used to judge the usefulness of grating observations in specific cases. For comparison, Figure 38e shows the spectral resolution of the ACIS CCDs without the gratings, reminiscent of the current generation of X-ray missions.
High spectral resolution over a broad X-ray energy band allows a variety of approaches to X-ray spectral analysis. While the traditional global fitting techniques used in non-solar X-ray spectroscopy will continue to be necessary for non-grating data, their application to grating observations remains to be tested on realistic spectra. The chief advantages of this approach are that the entire spectrum is used, thereby avoiding gross mistakes, and that computer automation obviates the need for tedious line-by-line analysis. Furthermore, a consistent treatment of the count statistics may be built into the method. The method has distinct disadvantages as well, in that it can be difficult to understand what drives the fit, and that a purely statistical comparison of model and data (e.g. on a bin-by-bin basis) may not reflect the information content of the data, particularly if the model parameters are not independent.
The application of line ratio diagnostics represents a major change in the way X-ray spectra can be analyzed. By focusing on strong, high signal-to-noise lines, often the best understood from the atomic physics point of view, one can avoid modeling the entire spectrum in order to extract the plasma quantities of interest. On the other hand, different line ratios can give inconsistent results and thus over-reliance on a few line ratios may lead to errors. Unsuspected line blending, complex emitting regions in the line of sight, and other systematic effects plague line ratio diagnostic analysis unless redundant results are obtained.
An iterative approach, beginning with a set of relatively secure lines and followed by progressively more complete comparisons of predictions with observations, builds in the checks and balances of the global fitting approach and at the same time maximizes the information content. This approach requires both long exposures to sufficiently sample weaker spectral features and patience in data analysis -- reducing the tedium through computer automation will be a software challenge. Finally, we mention physical approaches to spectral modeling, such as explicitly used in X-ray photoionization and non-equilibrium ionization models of supernova remnants, but also of interest for coronal magnetic loop structures. Where radiative transfer plays a major role in the underlying physical processes, as in X-ray photoionized plasmas, interpretation of early grating observations is likely to be driven by empirical evidence, such as the existence of particular emission lines in the soft X-ray spectrum of AGN.
The complexity of X-ray spectra means that their analysis will rely on high-quality atomic data and detailed models of the atomic processes leading to line formation. Unfortunately, the atomic data currently available is likely to fall short of what is required. In July 1995 the AXAF Science Center sponsored a workshop on Rates, Codes, & Astrophysics, with the goal of bringing together researchers from atomic physics, plasma spectroscopy, and astrophysics. It became clear at that workshop that the needs of astrophysics must drive advances in atomic data. Improvements in basic atomic rate data -- for collisional ionization and photoionization, radiative and dielectronic recombination, charge exchange, collisional excitation, and radiative decay -- are necessary if the potential of the AXAF transmission grating data is to be fully exploited.
Much available atomic data is based solely on theoretical calculations because laboratory measurements are difficult and expensive. Because theoretical calculations are also difficult and expensive, many approximations must be made and, without laboratory measurements, the absolute accuracy of the calculations is difficult to judge. Theoretical calculations for critical rate coefficients, e.g. dielectronic recombination rates, can disagree by factors greater than two, while many wavelengths are known only to about 2 to 3% accuracy. The uncertainty of theoretical atomic data and the shortage of laboratory benchmarks means that we must rely on astrophysical sources to provide benchmarks for our plasma emission codes.
In capturing detailed X-ray spectra of astrophysical sources, observations with the AXAF gratings will reveal X-ray lines that have never been seen before, either in observations of the sun or in laboratory experiments. Accounting for this line emission will greatly improve the accuracy and completeness of existing plasma emission codes. In this way, the gratings will have a major impact on our understanding not just of point sources, but of all X-ray sources, even those extended sources for which CCD resolution spectra are the best available. At CCD resolution, spectral features are blends of many hundreds of emission lines and, if the underlying spectral model is incomplete, misleading results can be obtained. By improving the completeness and accuracy of our plasma emission codes, AXAF grating spectra of point sources will clarify our understanding of even lower resolution spectra of a wide range of X-ray sources.
Meanwhile, the plasma spectrum modelers are opening up the black boxes that have been used to generate X-ray spectral models. The available models are being checked by quantitative comparisons between independently developed plasma codes. Public distribution of codes and databases along with modular design of computer programs will eventually allow users of spectral models to adapt the models to their needs, spanning applications from cooling rate calculations used iteratively in hydrodynamical modeling to prediction of detailed spectral data. Common atomic data standards will allow users to make their own sensitivity studies of critical atomic rates.
The public data sets of observations of different types of X-ray sources made available by the AXAF Science Center will surely inspire broad efforts to understand the spectra. Meanwhile, the ASC will continue to facilitate the cooperation, collaboration, and healthy competition among atomic physicists (both theoretical and experimental), plasma modelers, and astrophysicists.
Wise, M. W., Davis, J. E., Huenemoerder, D. P., Houck, J. C., Dewey, D., Flanagan, K. A., and Baluta, C. 1997, The MARX User's Guide, ASC Publication.
John Houck (MIT) and Nancy Brickhouse (CfA)