Urgently Needed Atomic Rates for X-ray and UV Astronomy

John C. Raymond
Harvard-Smithsonian CfA

Recent and impending satellite launches are bringing enormous improvements in spectral resolution, spectral range, sensitivity and intensity calibration to UV and X-ray astronomy. Data from STIS on HST and the CDS, SUMER and UVCS instruments aboard the SOHO satellite provide a glimpse of the diagnostic power of observations with the new generation of satellites. FUSE will extend this level of spectral resolution to the 912-1200 $\rm \AA$ range. The grating spectrometers on the AXAF and XMM satellites will resolve X-ray spectral features that have hitherto been perceived as blends whose components could not be unambiguously separated.

The overall goals of astrophysical spectral analysis at these wavelengths have not changed greatly over the past two decades. The difference is that the new instruments and the accumulating high quality atomic data make it possible to imagine getting reliable, accurate, unambiguous results. From an emission line spectrum, one ought to be able to derive the temperature distribution, the pressure and the elemental abundances. From these physical parameters of the emitting gas, one hopes to understand the physical processes responsible for generating and exciting the emitting gas. If both the data and the atomic rates are good enough, it is possible to go beyond the standard simplifying assumptions and search for indications of non-Maxwellian electron distributions (indicative of heating processes or steep temperature gradients), time-dependent ionization states (as a way to infer the history of the emitting plasma), optical depths in the emission lines (as a way to derive the geometrical structure of unresolved sources), and even electro-magnetic fields (through their effects on processes such as dielectronic recombination).

The current state of the atomic data leaves a great deal to be desired. The uncertainties in excitation ionization and recombination rates will dominate over the measurement errors for many observations of emission lines from astrophysical plasmas-an inefficient use of very expensive satellite observatories. Analysis of stellar continuum spectra will be limited by uncertainties in wavelengths and oscillator strengths.

The following sections list some of the important processes that should be measured in the laboratory or computed in order to make full use of X-ray and UV observations. It should be kept in mind that, with some important exceptions, that the importance of a given rate is more or less proportional to the astrophysical abundance of the element involved. H, He, C, N, O, Ne, Na, Mg, Al, Si, S, Ar, Ca, Fe and Ni account for most of the observable emission lines. Less abundant elements are detected, however, and their abundances can be crucial to understanding things as diverse as supernova explosions or radiative levitation in stellar atmospheres.

Line Identification (top)

Wavelengths of the strong lines of the more abundant elements are known quite accurately, and for many purposes the strong lines provide adequate, easily measured diagnostics. Wavelengths are less reliable for less abundant elements, for complex atoms and ions, and for weak lines of the abundant elements. While oscillator strengths for enormous numbers of transitions are available from Opacity Project or HULLAC calculations, the wavelengths predicted in these calculations are generally inadequate for line identification without considerable additional effort. About 30% of the emission lines detected in the low solar corona by the SUMER instrument have not been identified (Feldman et al 1997). The situation is likely to be worse for the 40-150 $\rm \AA$ band observable with the AXAF LETG. Identification of these features is important for understanding line blends and for the additional diagnostic possibilities the line present.

Absorption lines can be detected for a greater range of elemental abundance and lines strength, particularly for high signal-to-noise ultraviolet spectroscopy of narrow-lined, chemically peculiar and normal B- and A-type stars. Understanding these spectra requires accurate wavelengths and transition probabilities for an enormous number of lines, spanning the wavelength range 1150-3200A. There is an especially acute need for atomic data at the shorter VUV wavelengths (< 1700 A), where the line crowding increases and the accuracy and completeness of existing laboratory work is much lower than at longer wavelengths. These investigations would benefit greatly from expanded large scale studies that could provide wavelengths with sub milli-Angstrom accuracy and improved estimates of transition probabilities (to 10-20% accuracy) for large numbers of lines of the second and third spectra of iron-group and other elements at these VUV wavelengths. There are numerous examples of observable isotope shifts (IS) and hyperfine structure (hfs) in the UV spectrum of chi Lupi. A comprehensive data base of IS and hfs parameters is very important. It is not easy to predict in which ions these nuclear effects will have a discernible influence on stellar line strengths and abundances derived therefrom.

Oscillator Strengths and Photoionization Cross Sections (top)

The OPAL and Opacity Project calculations represent such a large advance in the quantity and quality of the data that it is tempting to regard oscillator strengths and photoionization cross sections as a solved problem, and the general agreement between the two sets of calculations provides confidence that the predictions are basically correct. Howver, it would be extremely useful to have better data for the more complex ions (especially low ionization states of the iron group elements), better estimates of the uncertainties in the numbers, and laboratory benchmark measurements of the computed resonance structure in photoionization cross sections. Resonances may be especially important for questions involving ionization by particular strong emission lines such as Ly$\alpha$ or He II $\lambda$304.

Ionization Rate Coefficients (top)

The ionization state of a plasma is often used as an indicator of the electron temperature, but this is only as accurate as the ionization and recombination rates used in computing the ionization balance. The level of agreement among current ionization balance calculations suggests that the temperature inferred from the ionization state is only reliable to about 0.1 dex in log T. Moreover, some problems that have persisted for decades, such as the inconsistency among the emission measures derived from Li-like, Be-like and B-like ions, are likely to be caused by errors in the computed ionization states. It is necessary to reach about 10% accuracy in the ionization rates to match the uncertainties in instrumental calibration. This should be achievable with storage ring experiments and crossed beams experiments. Measurements for complex ions are important, as the theoretical rates are especially uncertain. Inner shell excitation followed by autoionization is important for several isoelectronic sequences. The Oak Ridge measurements of ionization cross sections of iron ions through $\rm Fe^{15+}$ are exemplary, and it is important to extend them to higher ions. It is important to measure ionization rates from metastable levels for interpretation of plasmas such as the solar transition region. The metastable levels of Be-like and Mg-like ions have statistical weights much larger than those of the ground states, so they have an especially large effect.

Recombination Rate Coefficients (top)

Radiative recombination is the inverse of photoionization, and is therefore covered by the photoionization cross section discussion above. The difference is that determining the recombination rates requires photoionization cross sections from many excited levels, particularly if one is interested in the contribution to emission lines. In general, the radiative recombination rate dominates for ions that have low dielectronic recombination rates due to the lack of low-lying excited levels; H-like, He-like, Ne-like and Ar-like ions.

Dielectronic recombination (DR) is the dominant recombination path for most ions in collisionally ionized plasmas. The standard uncertainty estimate for calculated DR rates is 30%, but there are larger discrepancies than that even among sophisticated calculations for simple ions. The problems become worse for complex ions (e.g. Fe II-Fe VIII) and for temperatures where kT is small compared with the excitation energies of the resonance lines (as found in photoionized plasmas). Dielectronic recombination by way of forbidden lines has been recently shown to be important for some highly charged ions at low temperatures. The dependence of the DR rate on density and electric or magnetic field is seldom included in astrophysical model calculations, and DR from metastable levels is not usually considered. The DR rates under the density and field conditions of astrophysical plasmas appear to be the largest contributor to the uncertainty in ionization balance, and hence to the plasma parameters inferred from the ionization state. They are therefore among the most important quantities to measure.

Charge transfer, especially with neutral hydrogen, is an important contribution to the recombination rates of ions in photoionized plasmas, especially if the ionizing radiation has a hard spectrum. Laboratory measurements for collision energies of the order a few eV are valuable.

Collisional Excitation Rates (top)

If we are to avoid having uncertainties in the collisional excitation rates dominate the uncertainties in derived physical parameters, we need collision strengths accurate to about 10%. This level is at the limits of the capabilities of crossed beams and storage ring experiments. Typical uncertainties quoted are 10% for Distorted Wave calculations and perhaps 10% for the Close Coupling method. Particular attention must be paid to resonance structure in the collision cross sections, especially for fairly complex ions. For most astrophysical applications, the cross section will be integrated over a Maxwellian distribution, so the individual resonances are not as important as the overall contribution of the resonances. In some cases cascades from a large number of higher lying levels, each of which has a fairly small cross section, can be very important. Many weak lines add together to form a quasi continuum which can be very important for abundance determinations. EBIT measurements are especially promising for investigating large numbers of transitions. Excitations from metastable levels are challenging from the laboratory perspective, but they are important in ions where the metastable population exceeds the ground state population, such as Be-like and Mg-like ions at high densities. Collisions with protons have been studied in the laboratory, but mostly at energies well above those important for astrophysical spectroscopy. Proton (and alpha particle) collisions are determine the fine structure populations in many ions under coronal conditions, and in fast shock waves they can excite resonance lines in the UV.

Other Laboratory Astrophysics (top)

The NASA laboratory astrophysics program has traditionally provided atomic data needed for the interpretation of data collected from satellites. Recent laboratory efforts to investigate the physics of shock waves, particularly the growth of instabilities and the interaction of shock waves with density inhomogeneities, show considerable promise. Strong shocks of a variety of types (purely hydrodynamic, high Mach number, low Mach number, radiative, MHD, in fully ionized plasmas) can be generated and well diagnosed on large inertial confinement laser facilities. Such work does not find a natural niche in the current programs (as exemplified by the ROSS NRA structure). Consideration should be given to the question of whether to broaden the NASA laboratory effort to include investigations of plasma physics, hydrodynamics, or other disciplines outside the usual scope of the NASA laboratory astrophysics program.



 
next up previous
Next: About this document ...
Peter L. Smith
5/15/1998