Rather extensive and accurate transition frequencies have been reported by
(1) L. Dore, L. Bizzocchi, C. Degli Esposti, and J. Gauss, 2010, J. Mol. Spectrosc. 263, 44,
and by
(2) L. Dore, L. Bizzocchi, and C. Degli Esposti, 2012, Astron. Astrophys. 544, Art. No. A19.
Additional data were taken from
(3) H. Krause, D. H. Sutter, and M. H. Palmer, 1989, Z. Naturforsch. 44a, 1063.
With respect to the first entry from July 2012, extensive additional data were reported by
(4) Y. Motoki, F. Isobe, H. Ozeki, and K. Kobayashi, 2014, Astron. Astrophys. 566, Art. No. A28.
These data improve the predictions slightly. The authors retained in their fit also data summarized by
(5) W. H. Kirchhoff, D. R. Johnson, and F. J. Lovas, 1976, J. Phys. Chem. Ref. Data 2, 1.
Data from (5) were not merged. This also applies to data from (4) with low values of J below 1 THz.
¹⁴N hyperfine structure (HFS) splitting was resolved in (1) and (3) as well as in part in (2), (4), and (5). Even partial ¹H HFS was resolved in (1). This splitting was removed by intensity-averaging the components. The data from (3) appear to be almost as accurate as those from (1) below 32 GHz. However, the frequencies above 32 GHz are shifted to high freuencies by about 10 kHz. Therefore, larger uncertainties were used for these transitions. Nevertheless, the data from (3) improve the HFS slightly.
Predictions with uncertainties larger than 0.3 MHz should be viewed with caution. Probably all transitions observable by radioastronomical means are predicted well enough.
¹⁴N hyperfine splitting may matter in astronomical observations. Therefore, a separate hyperfine calculation is provided for J' ≤ 30 below 1 THz. NOTE: The partition function does take into account the spin multiplicity of the ¹⁴N nucleus !
The dipole moment components were reported by
(6) M. Allegrini, J. W. C. Johns, and A. R. W. McKellar, 1979, J. Chem. Phys. 70, 2829.