No Title

Improved Atomic Physics Parameters for Modelling Gas in Galaxies

Matthew Malkan

UCLA

``When someone says "theoretically...", they usually mean "not really."

Abstract

Emission line spectroscopy, from infrared and optical, to ultraviolet and X-ray wavelengths, is one of the most powerful tools used for studying galaxies. For the sizable minority of them having active galactic nuclei, it is the principal observational tool. New NASA facilities are making this even more true in the future. Many large sophisticated programs have been developed to model emission and absorption properties in the various kinds of gas found in starburst and Seyfert galaxies; the large community of observers and theorists depends heavily on these models. Many of the atomic physics inputs to these models, however, are extremely uncertain and unreliable, including recombination coefficients for heavier elements, and energy levels for "coronal" ions.

It should be admitted that this review is not going to be complete. It will inevitably reveal some of my bias as an astronomical observer, with emphasis on the study of many extragalactic environments where further advances depend critically on laboratory measurements. At least I am aware of our increasing dependence on previous work that is needed for us to interpret our data. Fortunately I gained some appreciation of this when, as an undergraduate I had the opportunity to work in Bill Parkinson's group with Peter Smith. Aside from being a lot of fun, it started building my awareness of the extent to which all areas of astrophysics rely on the difficult laboratory work of experimenters like them. More and more, observers use sophisticated physical models to understand our data, such as codes for photoionization, photo-dissociation regions (PDR's), and shocks. These models are in turn heavily dependent on atomic and molecular parameters, most of which have been calculated theoretically. Both the modellers and the computational theorists ultimately depend on laboratory measurements. Where the experimental data are poorly measured (or, more often, not measured at all), the entire astronomical enterprise is perched on shaky foundations.

As noted by many speakers at this workshop, it is worth repeating the special relevance of this to the NASA Space Science program. Observational astrophysics at NASA is entering an unprecedented ``Golden (Goldin?) Age of Discovery," with the launch of so many powerful new observational facilities. Operating across the entire electromagnetic spectrum, these telescopes are going to be providing astronomers with extraordinary new data, particularly from spectroscopy at wavelengths and resolutions that have not been explored previously. After making such a magnificent effort to obtain these data, the key question is whether astronomers will be able to use them optimally. To answer this first requires some scientific description of how space-based observations can be most fully exploited to advance the field.

WHAT general astrophysical information do we want to infer from spectroscopy?

In typical extragalactic studies spectroscopy is used to learn about the physical conditions of gas, either through its absorption line spectrum, or, more often, through its line emission. The most fundamental (inter-related) physical quantities we want to infer are the gas density, temperature, and ionization state. Knowing these then allows us to calculate other interesting gas parameters, such as its a) mass and size, from its emission measure, or column density and optical depth; b) location from the source of ionization; and c) filling factor and covering fraction with respect to that ionizing source. We need several methods for making these determinations, because the gas in galaxies is generally a mixture of various regions with differing physical conditions. Furthermore, even in studying a single region, it is vital to have a few independent measures of its properties, to check for consistency.

After providing us with this basic information, spectroscopy can determine the elemental abundances of gas in galaxies. These are essential for studying its chemical evolution. Since this concerns the origin of the elements, it is going to be a major component of NASA's exciting and prominent new program on ``Origins" (http://origins.jpl.nasa.gov/).

Improving our ability to obtain that kind of basic astrophysical information would in itself advance virtually every field in astronomy. Nonetheless, deeper understanding of complex astronomical objects-especially those on galactic scales-will increasingly require detailed physical modeling. In these models, the most important parameters specify the density of the emitting gas, and the rate at which it receives energy. For example, in models of PDR's, the second variable is G₀, the intensity of the photon field between 6 and 13.6eV, relative to the solar neighborhood. Similarly, in photoionization models the variable is the "ionization parameter," the ratio of ionizing photons to particles. In shock models the energy-input variable is the shock speed. Models have also been made in which excitation and ionization is produced by relativistic particles. A second-order parameter we would like to constrain with these models is the spectral distribution of the input energy. The best example is in the study of photoionized gas in AGN, which are being used to determine the overal energy budget of the active nucleus, especially the dominant portion which emerges in the unobserved far- to extreme-ultraviolet (e.g. Zheng and Malkan 1993).

All of these increasingly complex kinds of models rely on large atomic and molecular databases.

The short answer is: in virtually every extragalactic environment we know about, and probably some that we don't (yet). Here is a more specific enumeration of the objects and the suspected astrophysical processes happening in them, from the relatively rare to the more common.

The study of active galactic nuclei (AGN) occupies about a tenth of all current research in astronomy, judging from the proportion of Astrophysical Journal articles devoted to it. The ``classical" AGN have several (sometimes overlapping) classifications, including quasar, blazar, Seyfert 1 and Seyfert 2. In spite of this apparent confusing diversity, all AGN are suspected to have several quite different regions in common:

1) Broad Emission Line Region is named because of its high ( $\ge$ 1000 km/sec) Doppler velocities of its permitted lines. This gas is known to be very dense ( $n \sim 10^{9--12} cm^{-3}$ ), and, from its variability, close to the central engine ( $r \sim 10^{16--18}$ cm). Even though this gas is close to an active galactic nucleus, it emits stronly in a wider range of permitted lines. Highly ionized species are not the only important ones present. For example, a ``forest" of optical and ultraviolet transitions from Fe II is a distinctive feature of BELR spectra, which probably arises at extremely high optical depths (Boroson and Green 1992).

2) Narrow Emission Line Region is named for its strong forbidden lines (which imply $n \le 10^6$ ) with Doppler widths $\le$ 1000 km/sec. These arise from a much larger volume than the BELR, and are observed from ultraviolet to far-infrared wavelengths.

3) Intermediate Line Region is suspected to lie in between the BELR and NELR, in density and distance from the nucleus. The original separation into BELR and NELR was probably an artifact of the ranges of parameter space probed by the strongest optical emission lines. In reality, the gas density around an AGN may well have a continuous distribution, with each cloud emitting mostly emission lines characteristic of its particular physical conditions.

4) Coronal Line Region is observable through forbidden transitions of highly ionized atoms such as those seen in the solar corona. It is still not in general known whether most CLR emission comes from low-density gas far from the nucleus, or high-density gas much closer in (e.g., Spinoglio and Malkan 1992, Ferguson, Korista and Ferland 1997).

5) Relativistic Jets are a spectacular defining feature of AGN. Their distinctive bi-polar morphology is often matched by closely correlated bi-polar emission line nebulosity. These "ionization cones" may be energized by several possible physical mechanisms.

6) Black Hole Accretion Disks are hypothesized to produce much of the continuum energy in AGN (Malkan, 1983), and perhaps even some of the broad emission lines. Although the densities are much higher than in the other AGN enviornments discussed above, very significant deviations from LTE populations are expected. A great deal of atomic physics goes into making realistic model predictions of the spectra of AGN accretion disks (e.g., Hubeny and Hubeny 1997).

7) Winds/Scattering Mirrors. In some Seyfert 2 nuclei, spectropolarimetry indicates that we have only an indirect view of the central engine, which is seen via scattering. The nature of this scattering ``mirror" is unclear, but it appears to be hot, highly ionized, and perhaps in a larger-scale outflow from the nucleus (Mathews, Miller and Goodrich 1991).

8) Warm Absorbers/Associated Absorbers are further probes of hot, highly-ionized gas around the centers of active galaxies. The former are seen by their absorption of soft X-rays (e.g., George, et al. 1998); the latter are detected by their absorption in UV resonance-line transitions (e.g., Foltz, et al. 1986). The relation of these two regions to each other, and to 7) is not yet clear.

Most of these observable regions are usually modelled with some kind of photoionization codes, based on the assumption that the gas is primarily energized by the central non-stellar continuum source, supposed to arise in a black-hole accretion flow.

A far larger fraction of galaxies harbor Low-Ionization Emission-line Regions (``LINER's") in their centers. Like the Seyferts, these galactic centers have emission-line spectra that are not readily explained by photoionization from normal hot stars. One possibility is that the emitting gas is ionized and excited by a hard, Seyfert-like spectrum, but with a substantially lower ionization parameter (Ferland and Netzer 1983). Another possibility is that the line emission is produced by high-speed shocks (e.g., Allen, Dopita and Tvetanov 1998). The predicted emission-line spectra of these two paradigms are both investigated with detailed physical modelling

It is not merely the study of ``active" galaxies which relies heavily on physical models. Photoionization models also play a crucial role in the study of starburst galaxies and dwarf ``extragalactic HII regions". In these cases, the ultimate energy source is actually known: populations of recently formed massive O and B stars. Nonetheless, accurate modelling is vital to infer abundances in these systems. The most famous example is the measurement of ⁴ He abundances in the most metal-deficient (and therefore most nearly ``primordial") dwarf galaxies (Pagel et al. 1992). This is another classic ``Origins" problem, with cosmological implications that go right back to Big Bang Nucleosynthesis.

Another example of star-forming galaxies which require good photoionization modelling are the ``Ultraluminous Infrared Galaxies" made famous by the enormously successful IRAS sky survey. These objects are so enshrouded in dust that very little is known about their ultimate energy source(s). In some cases, there may be a role for a buried (nonstellar) AGN, and in other cases galaxy collisions could be important (Fischer et al. 1996). Therefore the possible physical mechanisms that need to be considered encompass ``all of the above" possibilities.

One final application of these models is to understanding the line-emitting gas in the growing population of young (proto-)galaxies being discovered at high redshifts (e.g., Malkan, Teplitz, McLean 1996). We want to measure the cosmic production of heavy elements, and also understand the possible contribution of nuclear activity in these galaxies. Since we are now tracing the evolution of galaxies back from the present to the first 15% of the history of the Universe, understanding these objects is another key component of the Origins program.

One final constituent which is likely to accompany the gas in many of the astrophysical settings described above is Warm Dust Grains. These are the source of much of the strong infrared continuum that is observed in many of these objects, and it also produces absorption (silicates at 10um) and emission (PAH) features. Therefore we need good estimates of its emissivity, scattering and absorption cross sections as a function of wavelength. The size and composition of the grains is of great interest, especially since it may be modified by the intense high-energy radiation field impinging on the grains (e.g., Voit 1992).

We have now covered the Why, What, and Where of modelling gas in extragalactic systems. The last question to consider is:

The short answer is: by incorporating more and better atomic and molecular information into them. We will conclude by discussing some of the most urgent needs. This list is not particularly well organized by wavelength region, which is reasonable given the growing proportion of extragalactic studies that are multi-wavelength in scope.

One of the greatest current deficiencies is that reliable dielectronic recombination rates are not available:

Of course there are reasons for this incompleteness. Determining these rates depends on knowing several things better than we do currently, such as:

1) where are the positions of resonances, if possible to an accuracy better than 1%?

To get better answers to these questions basically requires experimental measurements of energies for autoionizing levels, particularly for those close to the threshold. These are the key inputs which are then used to specify the configurations which are used in computations of the dielectronic recombination rates. There is currently enormous room for improvement. In some cases, uncertainties in the rates are giving us possible errors of a factor of three in predicting the ionization balance.

Improving measures of dielectronic recombination rates is related to another priority: more accurate collision strengths for coronal emission lines. The reason is that the thermally-averaged collision strengths often have large contributions from resonances, which are close relatives of dielectronic recombinations. (In the latter process the electron in an autoionizing state decays to a real bound level, while in a collisional resonance, the electron in the autoionizing state returns to the continuum, after assisting a transition of the outer electron.) In both processes, we need to know the energies of those autoionizing states. The ions with the most important transitions observed at optical and infrared wavelengths, which therefore have the greatest need for accurate collision strengths are, in a very rough order of descending priority: Fe X, Fe XI, Si VII, Si IX, Ca VIII, Ne VI, Si X, Fe VII, Si VI, S VIII, Fe XIV, Fe XIII, Ne V, and S XI. Fe X and Si IX merit special attention because of the critical importance of resonances in their collision strengths.

A key laboratory measurement is accurate determination of bound levels including doubly excited autoionized levels, via spectroscopy. To make the most accurate possible mathematical representation of the ``target" ion, we need to know these energy levels. This information is then input into the theoretical calculations (e.g. by the R-matrix method) of collision strengths.

Electron excitation rates are also needed for less ionized heavy elements, such as Fe II, Fe III, Ni II, Ni III, etc. A-values for these ions could also be improved.

Finally we can add to our ``shopping list" of important atomic parameters that we need to know more accurately the following:

1) Collisional excitation and ionization rates for excited states of hydrogen and helium. This perhaps rates a lower priority, since the resulting changes in models may not be as large as for other atomic data.

3) Radiative recombination rates for nearly all ions, from light low-ionized atoms such as O II, to highly ionized heavies such as Fe XXV.

4) Charge transfer collsion rates. In some cases, such as iron, some of these may be uncertain by factors of two to ten, leading to large corresponding errors in the ionization distributions for these atoms.

5) Fe L-shell transitions. These will be observed in detail with the current generation of high-resolution X-ray spectrometers in orbiting telescopes. It will be essential to compare the astronomical observations with laboratory measurements, such as the ones being done by Liedahl and co-workers.

7) Collision strengths and transition probabilities for highly-ionized iron, particularly Fe IX through XIII.

Acknowledgments: This work benefited from helpful discussions with: Gary Ferland, Tino Oliva, Tim Kallman, Hagai Netzer, Francis Keenan, and Manuel Batista.