The "warm" (or, ionized) absorbing gas that is common in the X-ray spectra of low-redshift AGN is a major new component of their near-nuclear structure. About half of all low-redshift AGN show absorption by ionized gas (Reynolds 1997; George et al. 1998), and a similar fraction show associated UV absorption in highly ionized species such as C IV (Crenshaw et al. 1999). For AGN that have been observed at moderate to high spectral resolution in both the X-ray and the UV, there is a one-to-one correspondence between objects showing X-ray and UV absorption, suggesting that the phenomena are related in some way (Crenshaw et al. 1999). The gas has a total mass exceeding ~ 103 M (greater than the broad-line region, or BLR), and is outflowing at a rate > 0.1 M yr-1 (10 x the accretion rate in some objects) (Reynolds 1997).
The role played by the ionized absorbing gas in AGN at the moment is unclear. The UV-absorbing gas lies exterior to the BLR, as revealed by the depth of the absorption lines in many cases (Crenshaw et al. 1999). This gas could either arise in winds from the molecular torus (Weymann et al. 1991; Emmering et al. 1992), the accretion disk (Königl & Kartje 1994; Murray et al. 1995), or stripped stellar envelopes in the BLR (Alexander & Netzer 1994; Scoville & Norman 1995). Or, the X-ray heated winds that provide the reflecting medium in Type 2 AGN (Krolik & Begelman 1986) may be seen as the warm absorbing medium in Type 1 AGN (Krolik & Kriss 1995). Understanding the absorbing gas in nearby AGN may help us to understand how radiation is collimated (Kriss et al. 1997). This goes not only to the heart of understanding the structure of AGN, but is also a crucial element in understanding the ionizing radiation field at high redshifts and links between AGN and the X-ray background. If the obscuration/orientation paradigm for AGN unification is correct, then the solid angle subtended by unobscured lines of sight from an active nucleus is a crucial parameter in determining the fraction of ionizing radiation that illuminates the surrounding IGM. Obscured Type 2 AGN may also comprise a significant fraction of the point sources producing the X-ray background. Understanding the intrinsic structures that lead to the Type 1/Type 2 distinction is a vital clue in our overall understanding of these phenomena.
A key question for understanding warm absorbers is how the X-ray and UV spectral domains are related. In some cases, UV absorbing gas may be directly associated with the X-ray warm absorber (3C 351: Mathur et al. 1994; NGC 5548: Mathur et al. 1995). In other cases, however, the UV gas appears to be in an even lower ionization state, and there is no direct relation between the X-ray absorption and the UV absorption (NGC 4151: Kriss et al. 1995; NGC 3516: Kriss et al. 1996a, 1996b; NGC 7469: Kriss et al. 2000a). High spectral resolution observations of O VI absorption in AGN with X-ray warm absorbers is proving to be an invaluable tool for definitively analyzing these issues.
The disparity between the UV and X-ray absorbers is illustrated most clearly by FUSE observations of the Seyfert 1 galaxy Mrk 509 (Kriss et al. 2000b). As shown in Fig. 10, this observation resolves the UV absorption into seven kinematic components. In addition to the O VI and Ly absorption shown here, C III 977 lines are seen in the lowest-ionization components, and Lyman-line absorption up to Ly permits an accurate measure of the neutral hydrogen column density and its covering fraction. Comparison of the O VI, H I, and C III column densities to photoionization models permits one to independently assess the total column density of each kinematic component. Table 1 summarizes the physical properties of each of the seven components. Of the seven systems present, only #5, a system near the systemic velocity of Mrk 509, stands out clearly as having a level of ionization and total column density compatible with an X-ray warm absorber. Note, however, that the total equivalent width of the UV absorption is dominated by the other six components, yet the X-ray absorption that would be associated with them is negligible.
The O VI and O VI edges seen by Reynolds (1997) in ASCA spectra of Mrk 509 imply column densities for the ionized gas of NO7 = (4.6+1.3-1.7) x 1017 cm-2 and NO8 = (3.7+3.7-2.8) x 1017 cm-2. The photoionization model for the UV-absorbing component #5 computed by Kriss et al. predicts NO7 = 2.3 x 1017 cm-2 and NO8 = 0.1 x 1017 cm-2. Given the large uncertainties in the X-ray columns and the temporal difference between the X-ray and UV observations, the agreement is remarkably close, and it is quite likely that component #5 is same gas as that absorbing the X-ray radiation. It is puzzling, however, that this component lies so close to the systemic velocity. Most absorbers in other AGN (including most of the other components in this one) are blue-shifted relative to the host galaxy, and this is true for the high-resolution X-ray spectra obtained so far with Chandra: the X-ray absorption lines in NGC 5548 are blue-shifted by ~ 280 km s-1 (Kaastra et al. 2000) and by ~ 440 km s-1 in NGC 3783 (Kaspi et al. 2000). Clearly a larger sample of objects that would permit us to assess these differences is essential.
|#||v||NOVI||NHI||NOVI / NHI||Ntot||log U|
|( km s-1)||(cm-2)||(cm-2)||(cm-2)|
The multiple kinematic components frequently seen in the UV absorption spectra of AGN clearly show that the absorbing medium is complex, with separate UV and X-ray dominant zones. One potential geometry is high density, low column UV-absorbing clouds embedded in a low density, high ionization medium that dominates the X-ray absorption. This is possibly a wind driven off the obscuring torus or the accretion disk. What would this look like in reality? Detailed modeling of dense molecular gas irradiated by the ionizing continuum of an AGN is likely to miss the complexities that arise as material is ablated from the surface and flows away. We will not soon get a close-up look at this aspect of an AGN, so it is instructive to look at nearby analogies. The HST images of the pillars of gas in the Eagle Nebula, M16, show the wealth of detailed structure in gas evaporated from a molecular cloud by the UV radiation of nearby newly formed stars (Hester et al. 1996). Fig. 11 shows what this might look like in an AGN - a dense molecular torus surrounded by blobs, wisps, and filaments of gas at various densities. It is plausible that the multiple UV absorption lines seen in AGN with warm absorbers are caused by high-density blobs of gas embedded in a hotter, more tenuous, surrounding medium, which is itself responsible for the X-ray absorption. Higher density blobs would have lower ionization parameters, and their small size would account for the low overall column densities.
Figure 11. Artistic rendering of how a molecular torus surrounding an AGN might appear based on HST images of the Eagle Nebula.
At sight lines close to the surface of the obscuring torus, one might expect to see some absorption due to molecular hydrogen. Given the dominance of Type 1 AGN in our observations so far, the lack of any intrinsic H2 absorption is not too surprising since our sight lines are probably far above the obscuring torus. NGC 4151 and NGC 3516 are examples where the inclination may be more favorable since these objects have shown optically thick Lyman limits in the past (Kriss et al. 1992; Kriss et al. 1995; Kriss et al. 1996a), but our FUSE observations do not show such high levels of neutral hydrogen at the current epoch. Molecular hydrogen will not survive long in an environment with a strong UV flux. Given this, one might think that any sightline in an AGN in which a strong UV continuum was visible could not show H2 absorption since the H2 would not be optically thick enough to be self shielding. However, if the H2 is contained in small, dense blobs that have ablated from the obscuring torus, these blobs could be thick enough to shield their cores and small enough that they do not completely obscure the central radiation source. Thus, the H2 absorption, if present, is likely to show up as optically thick, high-column-density components that only partially cover the continuum or broad lines. Our best prospects for seeing such material are in upcoming FUSE observations of NGC 1068.
As I've shown here, FUSE offers numerous advantages for the study of absorbing gas in AGN. First, since the Ly absorption in the HST bandpass is often saturated, it is difficult to deblend the various velocity components and obtain accurate H I column densities. The unsaturated high-order Lyman lines in the FUSE bandpass make accurate measurement of the H I column density straightforward. Second, the O VI resonance lines are a key link between the lower-ionization species seen in the UV and the higher-ionization gas observed with Chandra. The presence of different, adjacent ionization states in the two bandpasses makes it possible to ascertain the ionization structure in a more model-independent way. One need not make assumptions about abundances or use photoionization models, for example, as would be required in comparing C IV in the HST band to O VII and O VIII in the Chandra spectrum. Similarly, the ionization state of the UV-absorbing gas can be obtained more directly without recourse to assumptions about abundances by comparing the C III opacity in the FUSE bandpass at 977 Å to C IV as observed with HST. Finally, after using O VI to identify the kinematic components in the absorbing gas associated with the X-ray warm absorber, the FUSE observations can be used to trace the kinematics of the X-ray absorbing gas at higher velocity resolution than is possible with the Chandra grating observations.