Communication: A hydrogen-bonded difluorocarbene complex: Ab initio and matrix isolation study Free

Carbenes are the key intermediates in a wide variety of chemical processes.^1–4 The electronic structure, spin dynamics, and reactivity of these species attract increasing interest in both theoretical and experimental aspects. Typically, simple carbenes are highly reactive, so they can be trapped only in an inert environment at low temperatures. The properties and reactivity of carbenes in condensed phases may be essentially controlled by non-covalent interactions as demonstrated in a number of recent studies on aromatic substituted carbenes.^5–7 In particular, it was shown that 1,3-dimethylimidazol-2-ylidene yielded strongly hydrogen bonded adducts with methanol and water.⁵ Moreover, complexation switches the ground spin state of diphenylcarbene from triplet to singlet.⁶ 

Meanwhile, the metal-free complexes of simple triatomic carbenes (methylene and halocarbenes) are much less studied from both experimental and computational points of view, and the hydrogen-bonded complexes of such species are unknown. One of the important species of this kind is difluorocarbene (CF₂). Unlike the aromatic substituted carbenes, it has a singlet ground state and the singlet-triplet gap is quite large (ca. 50 kcal/mol⁸), so hydrogen bonding cannot lead to spin state switching but should result in further stabilization of the singlet state.

In fact, it is not easy to produce the complexes of CF₂ and other simple carbenes under the conditions of matrix isolation. One way is co-deposition of carbene produced by dissociation of an appropriate precursor with a complexing agent in a matrix gas flow. Using such an approach, Davis and Liu⁹ prepared the CF₂⋯O₃ complex by co-deposition of pyrolized C₂F₄/Ar with O₃/Ar gas flows. To the best of our knowledge, it is the only metal-free complex of CF₂ reported up to now. Another common route to preparation of weak intermolecular complexes is photo- or radiation-induced dissociation of a suitable parent molecule, which intrinsically contains both fragments of a complex, in solid noble gas matrices.^10,11 Following this approach, one may consider preparation of stabilized CF₂⋯HX complexes from the CHF₂X molecules, and the simplest candidate is fluoroform (CHF₃, X = F). Indeed, it is known that the vacuum ultraviolet (VUV) photolysis of CHF₃ in the gas phase at room temperature yields CF₂ and HF.¹² 

We are unaware of any data on VUV photolysis of matrix isolated fluoroform in the solid phase below the ionization threshold. The matrix photoionization experiments in solid argon¹³ revealed formation of various ionic and neutral fragments, including CF₂, but did not provide clear evidence for a CF₂⋯HF complex. It should be noted that in such experiments, ionization of a target molecule occurs in the gas phase leading to multiple reaction channels. Meanwhile, in the case of solid-state radiolysis, the energy is primary absorbed by matrix followed by charge and excitation transfer to a guest molecule in the solid phase.^14,15 This means that the isolated guest molecule is not ionized, if the ionization energy value of matrix (IE_M) lies below the IE value of the guest molecule. In our recent study on the X-ray radiolysis of fluoroform in different noble gas matrices,¹⁶ it was found that ionic processes dominate in neon and argon (IE_M > IE_CHF3) and partially occur in krypton (IE_M ∼ IE_CHF3), but only neutral dissociation channels are observed in xenon (IE_M < IE_CHF3). Thus, the radiolysis of fluoroform in solid xenon may provide a good chance for stabilization of the CF₂⋯HF complex produced by dissociation of a neutral excited state.

In this work, we report characterization of structure and stability of a hydrogen-bonded CF₂⋯HF complex on the basis of quantum-chemical calculations and its spectroscopic identification in a solid xenon matrix.

Molecular geometries have been fully optimized (tolerance on gradient: 10⁻⁶ a.u.) at the unrestricted CCSD(T) (only the valence electrons are correlated) level of theory. Type of stationary points on the potential energy surface (PES) was checked by vibrational analysis.¹⁷ The augmented valence correlation consistent basis sets of type L3a were used.¹⁸ This basis is analogous to the aug-cc-PVQZ basis set but contains a bigger number of primitive Gaussian functions (see the supplementary material for details). Using the optimized geometry, we calculated harmonic zero-point vibrational energy (ZPVE), harmonic vibrational frequencies, and IR intensities at the same level of theory. All calculations were performed using a program developed by Laikov and Ustynyuk.¹⁹ For complexes, the interaction energy was found as a difference between energies of the complex and the monomers and corrected for basis set superposition error (BSSE)^20,21 and zero-point vibrational energy (ZPVE).

CHF₃ (99.99%) was purchased from HORST and used without further purification. Xenon (99.99%) was used as received. An original closed-cycle helium cryostat on the basis of a SHI RDK-101E cryocooler was described elsewhere.²² The CHF₃/Xe gas mixtures with different concentrations of CHF₃ were slowly deposited onto a cooled KBr substrate at 25 K. The matrix sample thickness was typically 70-100 μm. Irradiation of the samples was carried out at 5 K using a 5-BKhV-6(W) X-ray tube with a tungsten anode (operated at 33 kV and 70 mA). The irradiated samples were annealed in the temperature range from 5 to 45 K. All the IR spectra were recorded at 5 K using a Bruker Tenzor II FTIR spectrometer with a cooled MCT detector (resolution 1 cm⁻¹, averaging by 144 scans).

We have found three structures corresponding to the CF₂⋯HF complexes on the potential energy surface (PES). The minima on the PES were verified by analysis of the harmonic vibrations which confirmed no imaginary frequencies.

As revealed by calculations, strong interaction between the lone electronic pair of carbene and H atom of hydrogen fluoride leads to formation of complex 1 with the stabilization energy of 3.58 kcal/mol (taking into account ZPVE and BSSE corrections). Geometrical parameters of different CF₂⋯HF complexes are presented in Fig. 1. As a result of complexation, the H–F distance elongates from 0.919 to 0.930 Å, whereas both C–F bonds are shortened from 1.303 to 1.289 Å, and the F–C–F angle increases from 104.7° to 106.2°.

Another expected hydrogen bonding may exist between the H atom of HF and fluorine atoms of CF₂. It should be noted that the fluorine atom of CF₂ species have two lone pairs located inside and outside the F–C–F angle, respectively. Thus, one may consider two structures of the CF₂⋯HF complexes with FH⋯FCF hydrogen bonding. As shown by calculations, the corresponding complexes 2 and 3 were also found on the PES, but they demonstrate substantially weaker bonding between the components in comparison with complex 1 (ZPVE + BSSE corrected interaction energies are 0.30 and 0.24 kcal/mol, respectively). This situation is characterized by the alteration of the bond lengths in CF₂, so one bond stretches out, whereas another one shrinks in both complexes as compared to the free carbene (see Fig. 1). The F–C–F valence angle decreases slightly to 104.4° in complex 2 and in complex 3. The effect of complexation on the H–F bond length in complexes 2 and 3 is actually negligible. The ZPVE and BSSE corrected relative energies of complexes are presented in Table I.

In addition, we have found the transition state (TS) for thermal transformations from complex 1 to the fluoroform molecule. The corresponding activation barrier was calculated to be 23.85 kcal/mol (accordingly, the barrier for reverse reaction is 72.94 kcal/mol). Thus, complex 1 should be completely stable to reaction restoring the parent molecule under our experimental conditions.

The harmonic frequencies and IR intensities calculated for stable structures of the complexes and free molecules are presented in Table II. The strong bonding occurring in complex 1 results in very large shifts of the CF₂ and HF fundamentals. In particular, a strong red shift (−293 cm⁻¹) is predicted for the H–F stretching mode. The computed intensity of this mode increases remarkably in the complex as compared to the HF monomer. As mentioned above, both C–F bonds become markedly shorter, so the corresponding stretching modes exhibit a significant blue shift (ca. 50 cm⁻¹ for both symmetric and asymmetric C–F vibrations). The F–C–F bending mode in this complex is predicted to be blue-shifted (ca. 11 cm⁻¹). It should be noted that appearance of the new fundamentals, corresponding to librations of HF at 649.0 and 649.9 cm⁻¹ is typical for hydrogen bonded complexes of hydrogen halides.²³ 

On the other hand, in the cases of 2 and 3, bonding is relatively weak, so their fundamentals should be closer to the corresponding features of the fragments. Due to complexation, the H–F stretching vibration exhibits a relatively small red shift (ca. 20 cm⁻¹). Similarly, the C–F asymmetric stretching is red-shifted by 37 cm⁻¹ in both complexes. A different effect is predicted for the C–F symmetric stretching, which are blue-shifted by 11 and 4 cm⁻¹ from the corresponding features of free CF₂ for complex 2 and complex 3, respectively. The F–C–F bending frequency and IR intensity remain virtually unchanged upon complexation in structures 2 and 3. It is worth noting that these weak complexes have computationally strong IR absorptions in the region of 250-270 cm⁻¹, which correspond to librations of the HF fragment.

As shown previously, irradiation of fluoroform in the xenon matrix leads to the formation of CF₃, CF₂, CHF₃, CF₄, XeF₂ and some other products.¹⁶ In addition to the known bands of these species, we observed a number of relatively intense unidentified absorptions. In particular, the bands at 630, 1175, and 3471 cm⁻¹ (marked as 1 on Fig. 2) exhibit correlated behavior with increasing absorbed dose showing the build-up of the corresponding species upon radiolysis (see Fig. S1 in the supplementary material). Annealing of the sample results in simultaneous decrease in intensities of these bands in the temperature range of 35–45 K, which is characteristic of the global mobility of hydrogen atoms in solid xenon^24,25 and probably indicates the reaction with thermally mobilized H atoms (see Fig. S2 in the supplementary material).

In addition to these rather prominent features, irradiation leads to appearance of several weaker bands in the regions of CF₂ antisymmetric and symmetric vibrations (denoted as 2 and 3 in Fig. 3). As can be seen in this figure, the bands denoted as 3 (1069 and 1227 cm⁻¹) result directly from irradiation, whereas the absorptions 2 (1076 and 1231 cm⁻¹) appear only after annealing the sample at 33 K.

As mentioned above, the radiation-induced decomposition of fluoroform in the xenon matrix leads to non-charged products. Moreover, the formation of CF₂⋯HF complexes in condensed phase is expected due to the cage effect. Comparison between computational predictions for the complexes and experimental data allows us to suggest that the strongest unidentified bands belong to complex 1. The absorption at 1175 cm⁻¹ is attributed to C–F antisymmetric stretching, and the absorption at 3471 cm⁻¹ corresponds to the H–F stretching mode. From this comparison, we can conclude that the feature at 630 cm⁻¹ is due to the C–F intermolecular vibration mode (libration), which is predicted to be particularly strong in this case. As expected, the F–C–F bending mode is too weak to be observed.

The complexation-induced experimental and computed shifts of the observed absorptions are presented in Table III. Because of lack of uncomplexed HF in our system, we used the experimentally known frequency of free hydrogen fluoride in solid Xe (3954 cm⁻¹)²⁶ to calculate the shift of this feature upon complexation. The absorption of symmetric C–F vibration overlaps with a strong band of CF₄ at 1268 cm⁻¹, so it is observed as a poorly defined shoulder at 1270 cm⁻¹.

Generally, there is a good agreement between experimental and computed shifts. Meanwhile, it is worth noting that the observed shift of the H–F stretching mode (−483 cm⁻¹) is significantly higher than that predicted by theory 297 cm⁻¹. This is, however, non-surprising since the H–F frequency is strongly dependent on the environment and even a slight change of the complex geometry due to matrix perturbation may cause a strong shift. Such large shifts in the hydrogen-bonded complexes have been observed in some other systems, such as diphenylcarbene/H₂O and 1,3-dimethylimidazol-2-ylidene/H₂O, indicating strong interaction between HF and CF₂ fragments.^5–7 

Additional weaker features denoted as 2 and 3 (see above) could be attributed to the less stable complexes. It should be noted that absorptions 3 appears immediately after irradiation, whereas bands 2 result from annealing (see Fig. S3 in supplementary material). However, exact assignment of these complexes is questionable. As mentioned in Table III, the calculations predicted almost equal shifts for antisymmetric C–F stretching modes (−35 cm⁻¹ in complex 2 and −36 cm⁻¹ in complex 3) and somewhat different shifts for symmetric ones (+11 and +4 cm⁻¹, respectively). Experimentally observed shifts of CF₂ modes are −18 and −26 cm⁻¹ for asymmetrical stretching and +16 and +12 cm⁻¹ for symmetrical stretching. Based on the computational results, the spectral features 2 and 3 could be tentatively assigned to complex 2 and complex 3, respectively. One may speculate that the former species results from thermal relaxation upon annealing, although it is not clear in view of very small computed interaction energies suggesting the predominating role of the matrix effects. Generally speaking, the effect of matrix on energy and vibration frequencies on the caged species can be simulated in the frame of the super-cell approach, if the calculations are performed at a reasonable level of theory (e.g., DFT).²⁷ Application of this approach to hydrogen-bonded complexes may be a challenge for future studies.

In summary, we have characterized CF₂⋯HF complexes by ab initio calculations and matrix isolation FTIR spectroscopy. To the best of our knowledge, it is the first example of the hydrogen-bonded complex of difluorocarbene. An interesting feature of the system clearly revealed by calculations is concerned with the existence of different kinds of complexes corresponding to H⋯C (carbene-type) and H⋯F (fluorine-type) bonding. The carbene-type hydrogen bonding is much stronger (interaction energy 3.58 kcal/mol). Rather intense absorptions of the corresponding species were well defined in the IR spectra observed after irradiation of fluoroform in the solid xenon matrix and the experimental vibration frequencies are in good agreement with the computational prediction. Possible manifestations of two weaker complexes corresponding to the H⋯F bonding (predicted interaction energies 0.30 and 0.24 kcal/mol) were also found in the IR spectra; however, their assignment is tentative.

Regarding chemical implications of the results, it should be noted that the complex is protected from intrinsic reaction yielding a fluoroform molecule by a remarkably high energy barrier (23.85 kcal/mol). The observed decay of this complex at 35-45 K is due to bimolecular reactions with mobilized H atoms and not due to intrinsic instability. From this point of view, it would be of interest to explore other hydrogen-bonded complexes of CF₂ (e.g., with water and methanol) and to test the effect of complexation on their chemistry. Another challenge is probing the structure and stability of the hydrogen-bonded complexes with heavier dihalocarbenes with respect to the effect of complexation on their spin state and reactivity.

See supplementary material for the description of the basis set used, primary data on calculated harmonic frequencies, reduced masses, IR intensities, molecular geometries, and energies; experimental curves showing the radiation-induced build-up and thermal decay of the CF₂⋯HF complex.

This work was supported financially by the Russian Science Foundation (Project No. 14-13-01266). The technical assistance of I. V. Tyulpina is acknowledged. We also acknowledge the Joint Supercomputer Center of the Russian Academy of Sciences (Moscow) for granting computation resources.