Several theoretical studies have suggested that a spin–orbitinduced isomer may be found for a molecule of the as-yetunknow superheavy element 118, that is, (118)F4, [1] but there have been no reports of an experimentally observed molecule for which the inclusion of spin–orbit effects is essential for the correct identification of the ground state structure. We report here a molecular ion, [CH2ClI] , for which spin–orbit interactions are crucial for the identification of the structure and vibrational frequencies of the correct ground state. Theoretically, the spin–orbit interaction is part of the relativistic effect. The importance of relativity for the description of heavy atoms is well recognized. Scalar relativistic effects are routinely included in electronic-structure calculations of molecules containing heavy elements through the use of relativistic effective core potentials (RECP), but spin–orbit interactions are usually omitted when deriving optimized structures partly because of the assumption that their influence on the molecular structures is negligible and partly due to computational difficulties. Even when spin–orbit terms are available in RECPs, the usual treatment involves perturbational inclusion of these terms after the variational determination of orbitals and structures. Quantum chemical calculations employing RECPs and spin– orbit operators from the start have been available for some time. The spin–orbit density functional theory (DFT) method available in NWChem is particularly useful for the present purpose of demonstrating spin–orbit effects on geometries since the geometry can be optimized with both electron correlations and spin–orbit interactions included. We have been investigating the reactivity of mixed dihalomethane cations for some time. Hence, we have recorded the vibrational spectra of the cations by massanalyzed threshold ionization (MATI) spectrometry. Figure 1 shows the MATI spectrum of CH2ClI recorded by monitoring [CH2 ClI] in the electronic ground state. The most intense peak at around 78644 cm 1 corresponds to the 0–0 band. The distance of each peak from the 0–0 band in this spectrum corresponds to the vibrational frequency of the cation. CH2ClI has nine nondegenerate normal modes: modes 1–6, with a’ symmetry, and modes 7–9, with a’’ symmetry. All the fundamentals and overtones of the totally symmetric modes, a’, are dipole-allowed, while only the even-numbered overtones of the a’’modes are allowed. Utilizing the selection rule and the frequencies in the neutral species, plausible assignments can be made for the prominent peaks; these are listed in Table 1. Modes 2 and 3 are due to CH2 motion, modes 4 and 5 are C Cl and C I stretchings, respectively, and mode 6 is I C Cl bending. Other peaks in the spectrum are due to overtones and combinations. Our normal routine is to perform quantum chemical calculations for the vibrational frequencies, isotope shifts, and Franck–Condon factors to confirm, improve, or revise the phenomenological assignments. In particular, results from the DFT/B3LYP calculations have been found to be adequate in most cases. When we performed DFT/B3LYP calculations for [CH2ClI] + with Gaussian98, using a well-knownRECP for iodine such as LanL2DZ, we always obtained two distinct stationary states, A’ and A’’, formed by removal of an electron from in-plane and out-of-plane iodine nonbonding orbitals, respectively. Even though their energies are nearly the same, their geometries are noticeably different because of the partial antibonding characteristics contained in essentially iodine nonbonding orbitals. The vibrational frequencies in the two states are also noticeably different, suggesting that the observed spectrum would be a superposition of two spectra. The fact that a rather straightforward phenomenological assignment of the MATI spectrum was possible is not compatible with the calculated results. Moreover, the experimental frequencies could not be correlated with the calculated results with the accuracy expected for the DFT/B3LYP results ( 15 cm ). We thought that neglecting the spin–orbit effect could be one possible cause of this discrepancy, and we therefore performed the spin–orbit DFT calculations implemented in NWChem for neutral and cationic CH2ClI in the electronic ground states using RECPs with the effective spin– [*] M. Lee, Prof. M. S. Kim National Creative Research Center for Control of Reaction Dynamics and School of Chemistry, Seoul National University Seoul 151-742 (Korea) Fax: (+82)2-889-1568 E-mail: myungsoo@plaza.snu.ac.kr