The NH$_2$D hyperfine structure revealed by astrophysical observations

The 1$_{11}$-1$_{01}$ lines of ortho and para--NH$_2$D (o/p-NH$_2$D), respectively at 86 and 110 GHz, are commonly observed to provide constraints on the deuterium fractionation in the interstellar medium. In cold regions, the hyperfine structure due to the nitrogen ($^{14}$N) nucleus is resolved. To date, this splitting is the only one which is taken into account in the NH$_2$D column density estimates. We investigate how the inclusion of the hyperfine splitting caused by the deuterium (D) nucleus affects the analysis of the rotational lines of NH$_2$D. We present 30m IRAM observations of the above mentioned lines, as well as APEX o/p-NH$_2$D observations of the 1$_{01}$-0$_{00}$ lines at 333 GHz. The hyperfine spectra are first analyzed with a line list that only includes the hyperfine splitting due to the $^{14}$N nucleus. We find inconsistencies between the line widths of the 1$_{01}$-0$_{00}$ and 1$_{11}$-1$_{01}$ lines, the latter being larger by a factor of $\sim$1.6$\pm0.3$. Such a large difference is unexpected given the two sets of lines are likely to originate from the same region. We next employ a newly computed line list for the o/p-NH$_2$D transitions, where the hyperfine structure induced by both nitrogen and deuterium nuclei is included. With this new line list, the analysis of the previous spectra leads to linewidths which are compatible. Neglecting the hyperfine structure owing to D leads to overestimate the linewidths of the o/p-NH$_2$D lines at 3 mm. The error for a cold molecular core is about 50\%. This error propagates directly to the column density estimate. It is therefore recommended to take into account the hyperfine splittings caused by both the $^{14}$N and D nuclei in any analysis relying on these lines.