Francis, Hooper, and Impey (1993) have shown that BAL QSOs tend to be found closer to the radio-loud/radio-quiet divide than typical optically selected QSOs. They estimated that BAL QSOs are overabundant by a factor of 10 in the ``radio-moderate'' (0.2 < log(R) < 1.0) population compared to the radio-quiet population, and one of the brightest radio-moderates in their sample is a Lo-BAL, 1235+1807B. The low radio flux limit of the FIRST Survey leaves it sensitive to radio-quiet and radio-moderate objects with z 2, so the presence of ordinary Lo-BALs in the FIRST quasar samples would not be surprising. Out of 170 newly identified quasars in the FIRST Survey searches, only two Lo-BALs have been found so far and both have highly unusual spectra placing them in the rare iron Lo-BAL class, defined by 0059-2735 (Hazard et al. 1987). In the current census of 8600 quasars of Véron-Cetty & Véron (1996), 0059-2735 is the only other such recognized object. The FIRST Survey is apparently more effective at finding these unusual quasars than previous surveys.
The rarity of iron Lo-BALs up until now can to some extent be attributed to the absence of UV excess or strong emission lines in their optical spectra as well as their relatively low radio flux densities. Even multicolor high-z quasar surveys (e.g., Warren, Hewett, and Osmer 1994) have not detected this class of quasar. Although Hazard et al. (1987) point out that the appearance of 0059-2735 in their discovery objective prism data drew attention because of its strong Al II and Al III absorption, objects as heavily attenuated as 1556+3517 probably masquerade effectively as M stars in such surveys. The FBQS has a relatively weak color selection, including objects as red as O-E = 2, while the high redshift survey intentionally targets even redder stellar objects, and neither rely on objective prism data, so objects like 0840+3633 and 1556+3517 are not selected against in the optical.
The spectra of the three iron Lo-BALs (Figure 1) suggest a correlation between column density of low-ionization clouds and radio emission. Based on the Fe II absorption lines and continuum attenuation, 0059-2735 has the lowest column density, followed by 0840+3633, and then 1556+3517; ranking by radio luminosity yields the same order. The severe attenuation of the continuum shape of 1556+3517 also suggests the presence of large amounts of dust. Although derived from a very small sample, this correlation suggests that high column densities in these BAL QSOs are associated with stronger radio emission. The preponderance of known BAL QSOs have much lower column densities of low-ionization clouds, consistent with the relatively low radio power of all BAL QSOs discovered to date. If the correlation between extinction in the optical and radio luminosity is correct, then as the iron Lo-BALs become harder to detect in the optical, they become more prominent in the radio and will naturally be more plentiful in radio-selected samples. Based purely on these empirical results, we propose that most BAL QSOs are radio-quiet and will not show up in a radio-selected sample, while the iron Lo-BALs are a special subpopulation which occupy the transition region between radio quiet and radio loud objects. The discovery of two iron Lo-BALs out of < 200 quasars suggests that they make up 1-2% of the FIRST quasars, comparable to the rate of Lo-BALs in the general quasar population as currently understood.
A possible additional high-redshift (z = 2.33) member of the iron Lo-BALs is Hawaii 167, analyzed by Cowie et al. (1994) and Egami et al. (1996). The spectrum of Hawaii 167 (Figure 2 of Cowie et al.) does show marked similarities with 0059-2735 but lacks strong Mg II 2800 emission and does not have particularly broad absorption troughs. The discovery of the two new iron Lo-BAL objects eases the transition between 0059-2735 and Hawaii 167; 0840+3517 is intermediate while 1556+3517 has even less Mg II emission and a more attenuated continuum than Hawaii 167, though its absorption features appear broader. If our suggestion of a link between radio power and column density of absorbing material is correct, then Hawaii 167 should have a radio luminosity between that of 1556+3517 and 0840+3633. Hawaii 167 has not been detected in the 20cm VLA NVSS Survey (Condon et al. 1996), implying an upper limit of 2.5 mJy and log(L) < 32.8, leaving room for it to fit the trend.
Hawaii 167 is quite faint in the optical, B = 23.0 (Cowie et al.). It was found in a complete spectroscopic survey of IR-selected objects in a region covering only 77 arcmin2. Cowie et al. (1994) argue that finding such an object in a small area survey indicates that these objects may be quite common, albeit at faint optical magnitudes, consistent with the results of the FIRST quasar samples.
Voit, Weymann, & Korista (1993) have proposed a model for Lo-BALs in which a nascent QSO is embedded in a region with a high rate of massive star formation, enshrouded with dust. In their scenario, these objects then evolve into more ordinary Hi-BALs with time, as their dust shrouds dissipate. If indeed 1556+3517 were to evolve into a more ordinary BAL, its radio luminosity would have to diminish. Egami et al. (1996) extend this model to Hawaii 167 and 0059-2735, explaining the iron Lo-BALs as a combination of heavily reddened QSO + starburst. They point out that such objects would likely be missed by optical surveys because the high dust content suppresses the rest frame UV (observed optical) and because these objects lack strong emission lines. They further note that if the starburst activity has not begun, these objects will be ``completely dark'' in the optical, implying that there may be a large, undetected population of these hybrid starburst/quasar objects.
This model helps to explain the FIRST Survey sensitivity to iron Lo-BALs: our radio-selected samples are at least partly immune to the optical selection effects that may cause these objects to be overlooked. If so, the FIRST Survey is now picking up the brightest, and possibly lowest redshift, tip of the iceberg. These objects have not yet acquired strong emission lines, but have evolved to the point where they can be seen through the surrounding dust. Additional arguments for populations of undetected, dusty quasars have been put forth by Low et al. (1989), Sprayberry & Foltz (1992), and Webster et al. (1995). Just how numerous the iron Lo-BALs are will be determined as the FIRST QSO surveys progress and IR surveys become more common.