Abstract
Pressure broadening coefficients of the O3–N2(O2) vibrotational lines at room temperature were computed in the framework of the semiclassical formalism of Robert and Bonamy improved by exact trajectories, and a semiempirical approach based on the Anderson theory. In the first method, a more precise trajectory description enables a good (up to a few percent) reproduction of the experimental data available in the literature, but is characterized by a quite high CPU cost. In the second method, a simplified form of the scattering matrix accounts for the subtle effects of the trajectory curvature and vibrational dependence via effective adjustable parameters and noticeably reduces the CPU time without a loss of precision for further systematic computations. In contrast to the case of the water molecule, these parameters exhibit a quite pronounced dependence on the rotational quantum number J values of the lines used for fitting and should be properly adjusted for transitions from low, middle and high rotational levels.
Acknowledgements
The authors acknowledge financial support from the French national program Les Enveloppes Fluides et l'Environnement—CHimie ATmosphérique (LEFE-CHAT) and the RAS program ‘Optical Spectroscopy and Frequency Standards’.
Notes
Notes
1. The dependence on vibrational quantum numbers is omitted in the following since both isotropic and anisotropic parts of the available interaction potential are vibrationally independent; therefore, the first-order terms proportional to the vibrational matrix elements of the isotropic potential do not appear in the line width expression. This dependence is, however, tacitly accounted for through the O3 energy levels and coefficients of the rotational wave functions in the final state.
2. The non-commutative character of the interaction at two different instants is neglected, so that the imaginary part of the second-order contributions vanishes.
3. These transition probabilities represent the squared reduced matrix elements of the relevant molecular operators of different tensor types: dipole (l = 1), quadrupole (l = 2), etc.
4. The interactions involving the dipole moment of the active molecule are assumed to be dominant, so that only one pair of c1 and c2 parameters corresponding to l = 1 is considered.
5. An additional test realized with a separate fitting for the P and R branches resulted in approximately the same quality of the c1 and c2 parameters and a very slight improvement in the theoretical values, therefore the results are not presented here.