7.1. What is the source of the elevated temeratures?
Some of the line ratios, including [O III] / H and the largest observed [N II] / H and [S II] / H ratios do not seem to be explained by photoionization, even when hardening of the O star's radiation is considered (Section 5). Do shocks play a role (Collins and Rand 2001, Hidalgo-Gámez 2005)? Or is there some additional source of non-ionizing heat, such as photoelectric heating from interstellar dust particles or large molecules (Reynolds and Cox 1992, Weingartner and Draine 2001), dissipation of turbulence (Minter and Spangler 1997), magnetic reconnection (Raymond 1992), shocks and cooling hot gas (Slavin et al. 1993, Collins and Rand 2001)? Most of these mechanisms seem capable of generating heating rates of the order 10-26 erg s-1 cm-3, the power requirement for the additional heat source (see discussion in Reynolds et al. (1999)). In addition, Reynolds et al. (1999) show that for these non-ionizing sources, their heat can dominate over that from photoionization at low ( 10-1 cm-3) densities, because their heating rates are proportional to the first power (or less) of the density, rather than the second power as with photoionization. This could explain the observed inverse correlation between the line ratios and the H intensity (see Fig. 2). Ways of discriminating between these mechanisms are very much needed, but none is yet forthcoming.
Can photoelectric heating be eliminated as a candidate for the supplemental heat? In neutral gas, the photoejection of electrons from polycyclic aromatic hydrocarbon (PAH) molecules by stellar ultraviolet photons below the Lyman limit is believed to be a major source of heating. However, PAHs may not be abundant in ionized gas, as suggested by both observational and theoretical considerations. Observations and modeling of the Orion Nebula H II region (e.g., Ferland 2001), show that PAHs are not present in the ionized gas there. From a theoretical perspective, if PAHs were present in H II regions their large Lyman continuum opacity (e.g., Weingartner and Draine 2001) would compete with hydrogen for ionizing photons and result in around 90% of the ionizing photons being absorbed by PAHs (Mathis and Wood 2005). PAH opacity peaks around 17 eV, which, for example, would result in He+ / H+ ratios in ionized gas containing PAHs being much higher than is observed in H II regions surrounding late O stars and in the DIG. On the other hand, silicaceous dust particles have a grayer wavelength dependent opacity and only absorb about 5% of ionizing photons in the Mathis and Wood (2005) simulations, not significantly changing the He+ abundance or other line ratios. However, the role they may play in photoelectric heating is unclear.
7.2. What is the spatial distribution of the gas?
Does the wide-spread H emission originate from a low density intercloud gas that occupies large volumes of the interstellar medium? Or is some significant fraction of the H coming from the denser ionized faces of clouds or superbubble walls? Models investigating the different scenarios and comparisons of the resulting emission line and line ratio maps with observations, including electron column densities from pulsar dispersion measures may provide some answers. Hill et al. (2008), for example, have shown that the statistical distribution of emission measures in the WIM is matched very well by the density distribution produced in magnetohydrodynamic (MHD) models of an isothermal, mildly supersonic turbulent medium. Also, new observational methods that sample the H+ at very small scales (Hill et al. 2003) or that access emission lines in a different spectral region (e.g., Kutyrev et al. 2004), may offer new insights about the small scale structure and dynamics of the gas as well as its larger scale distribution within the disk and halo.
7.3. How much ionizing radiation escapes the galaxy?
There have been several theoretical calculations of how much ionizing radiation escapes from galaxies (e.g., Dove and Shull 1994, Wood and Loeb 2000, Ricotti and Shull 2000, Ciardi et al. 2002, Clarke and Oey 2002). The models show that if the interstellar medium is smooth the fraction of escaping photons is small. Clumpy density structures allow for larger escape fractions through low density paths in the interstellar medium (see Fig. 16 above). However, there are few observations to test these models and determine how much ionizing radiation actually escapes. If interface radiation or shocks (Bland-Hawthorn et al. 2007) are not a major source of ionization, then one promising avenue is to study the ionized surfaces of distant high velocity clouds surrounding the Galaxy (Tufte et al. 2002, 1998). Bland-Hawthorn and Maloney (1999), Putman et al. (2003), Tufte et al. (1998) have all estimated that a few percent of the ionizing luminosity from the Galaxy would be required to explain the H emission from high velocity clouds located ~ 10000 pc above the disk (Wakker 2004). As discussed in Section 5, the spectrum of the escaping radiation (e.g., from future observations of He I / H from intermediate and high velocity clouds) could reveal whether photons escape the galaxy through very low density channels or by filtering through density bounded ionized regions.
7.4. Do hot, pre-white dwarf stars play a role?
In their late stages of evolution, low-mass stars pass through a hot photospheric phase after shedding their outer envelopes. Once the stellar envelope (i.e., the planetary nebula) has expanded sufficiently and is optically thin to the ionizing radiation from the stellar core, the Lyman continuum photons are available to ionized the ambient interstellar medium. Such stars may be responsible for the ionization of small, localized regions in the low density interstellar medium (Reynolds et al. 2005), but it is not known whether they also contribute significantly to the more diffuse ionizing radiation producing the WIM. Their luminosities and their lifetimes in this phase are orders of magnitude smaller than that of massive O stars; however, in comparison to O stars, their numbers are enormous, and they are much more uniformly distributed throughout the the Galactic disk. Early calculations by Hills (1972) indicated that the ionizing radiation from these hot pre-white dwarf stars could significantly influence the interstellar medium. Since then, there has been progress in understanding the late stages of evolution of low mass stars, but little additional work has been carried out on their collective, large-scale influence on the interstellar medium.
7.5. Is missing atomic data important?
Presently photoionization models are unable to make accurate predictions of the [S II] emission from the DIG because the dielectronic recombination rates for sulfur are unknown (see discusion in (Ali et al. 1991). The determination of these rates is important because they impact the inferred S+ and S++ abundances in the DIG, and, because sulfur is an important coolant, influencing the predicted strengths of other lines. While efforts are underway to calculate these dielectronic recombination rates, there may also be a handle on the rates observationally via modeling the emission line strengths for large, low surface brightness H II regions.
7.6. What insights will new global models provide?
A next step in modeling would be to test global dynamical models of the interstellar medium (e.g., de Avillez and Berry 2001, Kowal et al. 2007) to see whether their density structures can allow for O star radiation to produce the observed ionization (H emission) and temperature structure (line ratios) of the DIG. This will be a formidable task, combining large-scale 3D dynamical and radiation transfer simulations. A unique solution for the structure of the interstellar medium may not even be possible without also incorporating observations of the other gas phases. However, combining dynamical and photoionization models would provide observational signatures that could be searched for. Perhaps progress could be made in testing various scenarios - ruling out classes of models, determining what conditions (density and dynamics) are required to allow radiation to escape to the halo, and determining which models best fit the observed distribution and kinematics of the H over the sky.
While much progress has been made in understanding the nature of the warm ionized medium and the basic physical processes occurring within it, the future promises to be a very exciting time, when the advances in computing ability are combined with high spatial and spectral resolution observations of this gas and other major phases of the interstellar medium. There is much still to be learned.
We thank Carl Heiles for his contributions, enthusiasm, encouragement, and insights over the years. We also thank the meeting organizers, Y.-H. Chu and T. Troland, without whom this paper would not have been written. We thank two anonomous referees, whose comments and criticisms resulted in an improved paper. We also thank A. Ferguson, C. Hoopes, and J. Rossa for access to high-quality versions of their figures as well as their permission to include them in this review. We are also very grateful to our students and collaborators for their support and contributions. RJR thanks Don Cox for many valuable conversations over the years regarding the nature of the WIM. RJR & LMH were supported by the National Science Foundation through grants AST 02-04973 and AST 06-07512, with assistance from the University of Wisconsin's Graduate School, Department of Astronomy, and Department of Physics. GJM acknowledges support from the University of Sydney Postdoctoral Fellowship Program and the National Science Foundation through AST 04-01416. R-JD's work at Ruhr-University Bochum in this field is supported through DFG SFB 591 and through Deutsches Zentrum für Luft- und Raumfahrt through grant 50 OR 9707. JEB, AZ, & CG acknowledge support through grants AYA2001-0435 from the Spanish Ministry of Science and Technology and AYA2004-08251-C02-01 from the Ministry of Education and Science. A. Zurita thanks the Consejería de Educación y Ciencia de la Junta de Andalucía, Spain, for support.