3.1.2. Transition probabilities, collision strengths and effective recombination coefficients
The atomic data to compute the emissivities of optical forbidden lines have been recently recomputed in the frame of the IRON project (Hummer et al. 1993). The expected accuracies are typically of 10% for second row elements, however, the uncertainty is difficult to determine internally. Comparison with laboratory data is scarce, and actually, PNe are sometimes used as laboratories to test atomic physics calculations. For example, van Hoof et al. (2000) studied 3 PNe in detail and concluded that the [Ne V] collision strenghts computed by Lennon & Burke (1994) should be correct within 30 %, contrary to previous suggestions by Oliva et al. (1996) and Clegg et al. (1987). Another example is the density derived from [OIII] 52µm / [OIII] 88µm, which is significantly lower than derived from [S II] 6731/6717 and [Ar IV] 4711+4740 for a large sample of PNe observed by ISO (Liu et al. 2001). These authors argue that [O III] IR lines can be emitted from rather low density components but it could just be that the atomic data are in error.
Concerning recombination lines, the effective recombination rates for lines from hydrogenic ions have been recomputed by Storey & Hummer (1995) and by Smits (1996) for He I 5876. For C, N, O, estimates for all important optical and UV transitions are given by Péquignot et al. (1991). Detailed computations of effective recombination coefficients are now available for lines from several ions of C, N, O and Ne (see e.g. a compilation in Liu et al. 2000). Note however that these do not include dielectronic recombination for states with high quantum number, which may have important consequences for the interpretation of recombination line data (see Sect. 3.6)