We calculate the He i case B recombination cascade spectrum using improved radiative and collisional data. We present new emissivities over a range of electron temperatures and densities. The differences between our results and the current standard are large enough to have a significant effect not only on the interpretation of observed spectra of a wide variety of objects, but also on determinations of the primordial helium abundance. Subject headings: atomic data -atomic processes -ISM: atoms -ISM: clouds -plasmas 1. INTRODUCTION Helium is the second most abundant element in the universe, and its emission and opacity help us determine the structure of any interstellar cloud. Its abundance relative to hydrogen can be measured within a few percent since the emissivities of H i and He i lines have similar dependences on temperature and density. This makes it an indicator of both stellar and primordial nucleosynthesis (Pagel 1997).A good discussion of the history of calculations of the helium recombination spectra is given by Benjamin et al. (1999, hereafter B99), who present new calculations-the current standard in the field. Yet much progress has been made since the work by Smits (1991Smits ( , 1996 on which the B99 results depend. We implement these improvements, present a new set of predictions, and compare our results with those of B99. The differences are large enough to impact continuing attempts to estimate the primordial helium abundance (Peimbert et al. 2002).
THE NEW MODEL HELIUM ATOMThe basic physical processes have been described by Brocklehurst (1972) and B99. Here we describe the differences between B99 and our new numerical representation of the helium atom, which is a part of the spectral simulation code CLOUDY (Ferland et al. 1998). This model resolves all terms, nlS, up to an adjustable maximum principal quantum number n max , followed by a pseudolevel, , in which all lS terms are assumed n ϩ 1 max to be populated according to statistical weight and "collapsed" into one. We set recombinations into the collapsed level equal to the convergent sum of recombinations from n p n ϩ 1 max to . In the low-density limit, the collapsed level increases the ϱ emissivities of our benchmark lines (the same 32 lines given in B99) by 0.4%, on average, with . The decays n p 100 max from states with are most sensitive to this correction l p n Ϫ 1 for system truncation. The strong optical line l5876 is corrected upward by 1.3%. At finite densities collisional processes force the populations of very highly excited states into local thermodynamic equilibrium (LTE). In this case the adequacy of the method used to compensate for truncation is unimportant. We find the corrections negligible for cm Ϫ3 and n p 100 e . Consequently, the uncertainties in the results pren p 100 max sented in § 3 are due to the uncertainties in atomic data, especially the often substantial uncertainties in collisional rates affecting terms not in LTE at given conditions.There are several differences in atomic data for radiative proc...