Two-center three-electron bonding in ClNH₃ revealed via helium droplet infrared laser Stark spectroscopy: Entrance channel complex along the Cl + NH₃ → ClNH₂ + H reaction

The two-center three-electron (2c–3e) bonding concept was first introduced by Pauling in the early 1930s,¹ and it is commonly invoked to describe the chemistry and molecular orbital (MO) theory of free-radicals.^2–5 Nearly resonant interaction between a doubly occupied orbital on one atomic center with a singly occupied orbital on another generates bonding and anti-bonding molecular orbitals, the latter of which is singly occupied (SOMO). Although longer and weaker in comparison to a typical two-electron bond, 2c–3e hemi-bonds have nevertheless been shown to be significantly stronger than the electrostatic or dispersion binding associated with closed-shell molecular complexes.^6,7 We present a combined experimental and theoretical study of the ClNH₃ complex, which exhibits a strong 2c–3e hemi-bonding interaction that is revealed in the analysis of the NH stretch vibrations measured via helium nanodroplet isolation spectroscopy.

One of the simplest examples of a molecule containing a 2c–3e bond is the helium dimer cation (He⁺ (1s¹) + He (1s²) → He₂⁺ (σ)²(σ^∗)¹), which has a dissociation energy equal to 2.47 eV.⁸ By contrast, neutral He₂ is characterized by a largely repulsive potential with a 1.3 meV equilibrium well-depth.⁹ As another example, Pauling described the bonding in nitric oxide (NO) as consisting of a double bond and a three-electron bond. Pauling postulated that the presence of the three-electron bond precludes NO dimerization at room temperature.¹ In a theoretical study of Cl atom addition to nitrogen bases, Radom and co-workers found several examples of 2c–3e bond formation; in general, the strength of the 2c–3e bond was found to be inversely proportional to the difference between the ionization potential of the nitrogen base and the Cl electron affinity (3.62 eV).⁴ 2c–3e hemi-bonding was also explored in computational work by Guo and co-workers in which the interactions between halogen atoms (F, Cl, and Br) and water were discussed in terms of the aforementioned molecular orbital picture. Rather than electrostatic or dispersive, the interaction between monomers was shown to be electronic in character due to the near resonant mixing of the half-filled p-orbital on the halogen atom with the highest occupied molecular orbital (HOMO) on water.⁷ 

Careful consideration of the long-range interaction potential between halogen atoms and closed shell molecules is essential for an accurate description of the associated chemical reaction dynamics, especially at low temperature.¹⁰ As they approach one another, long-range stereodynamic forces can steer reactants toward or away from the transition state geometry, affecting the reaction rate and/or the branching ratio of the products.^10–13 Electronic structure theory predicts a stable van der Waals complex (CR1; Fig. 1) in the entrance channel of the Cl + NH₃ reaction.⁶ In addition to the van der Waals complex, two minima are predicted that correspond to more strongly bound 2c–3e complexes (CR2 and CR3; Fig. 1).^6,14 CR1 and CR3 are entrance channel complexes along the hydrogen abstraction path Cl + NH₃ → NH₂ + HCl, whereas CR2 lies in the entry valley to the substitution reaction Cl + NH₃ → NH₂Cl + H.¹⁴ CR2 is the most strongly bound adduct of the three, and the barrier to substitution is ten times larger than for hydrogen abstraction, as shown in Figure 2.^6,14

Cl–NH₃ complexes were investigated theoretically by Monge-Palacios and Espinosa-Garcia in an attempt to rationalize the disagreement between computed hydrogen abstraction barrier heights^6,14 and experimental determinations of rate coefficients near room temperature.^15–17 They concluded that CR1 and CR3 indirectly influence the kinetics by increasing the tunneling contribution to reaction. However, the authors noted that CR1 and CR3 complexes were elusive even at the CCSD(T)/cc-pVTZ level, and in fact, the CR3 complex could not be located at the MP2 level. Harmonic frequency computations for CR1 at the CCSD(T)/cc-pVTZ level yield three vibrations with frequencies <90 cm⁻¹, indicative of a very flat potential energy surface in the vicinity of the van der Waals complex.

Experimental investigation of the Cl + NH₃ reaction system is limited to two gas-phase kinetics studies of the hydrogen abstraction reaction near room temperature.^15,16 Related 2c–3e Cl–NR₃ systems were postulated in the analysis of ESR spectra obtained following the radiolysis of alkylammonium salts at 77 K.¹⁸ To the best of our knowledge, however, there are no spectroscopic reports for the isolated entrance or exit channel complexes of the Cl + NH₃ reaction. Helium droplets are advantageous in the study of pre-reactive complexes because of their ability to capture and quickly cool individual “reactants” and quench the associated kinetic energy generated upon molecular complexation. Due to this rapid cooling, He droplets are capable of stabilizing entrance channel complexes involving open-shell reactants.^19–21 In this work, we sequentially add Cl and NH₃ to He droplets and probe with infrared (IR) laser Stark spectroscopy the outcome of the low temperature reaction between these species. The results are rationalized with high-level electronic structure computations of the Cl + NH₃ potential energy surface, revealing facile rearrangement pathways favoring the formation of the 2c–3e CR2 complex. The nature of the bonding in this system and its impact on the infrared spectrum is discussed.

A brief description of the helium nanodroplet methodology follows, although it has been discussed in detail previously.^22–25 Superfluid helium nanodroplets are generated in the high-pressure region of a continuous expansion of He gas (30 bars backing pressure) through a cold nozzle with a 5 ± 1 μm diameter aperture. With the nozzle temperature at 17 K, droplets are produced with an average size of 4500 He atoms.^22,26–28 Nascent droplets evaporatively cool to ≈0.4 K and are collimated into a beam by a skimmer having a 0.4 mm aperture.^23,29 The droplet beam passes into a pickup chamber where droplets collide with and solvate gas-phase Cl atoms and NH₃ molecules and subsequently cool their internal degrees of freedom to 0.4 K.³⁰ Atomic Cl is generated through the thermal decomposition of Cl₂ (Sigma-Aldrich, ≥99.5% purity) as it collides with the walls of a resistively heated SiC tube (ID = 1 mm; T ≈ 1600 K).^20,31 This SiC pyrolysis source is oriented perpendicular to the droplet beam path with the SiC tube adjacent to the droplet beam such that Cl atoms exiting the SiC tube effuse into the droplet path and are picked up. The effusive flow of Cl₂ through the SiC source is adjusted to optimize for the pickup of single Cl atoms. The Cl-doped droplet beam subsequently passes through a 2 cm long, differentially pumped, stainless steel tube containing ≈10⁻⁶ Torr of anhydrous NH₃ (99.99% purity), metered directly from a lecture bottle.

Following dopant pickup, the droplet beam enters a Stark cell consisting of two Au-coated multipass mirrors and two parallel, stainless steel electrodes mounted orthogonal to the mirrors.^32–35 A homogeneous electric field up to tens of kV/cm can be applied to a ≈20 cm portion of the droplet beam, and the interaction of the electric dipole moment of the He-solvated molecule with the applied field lifts the M degeneracy of the rotational energy levels. Droplets subsequently pass into a mass spectrometer where they are ionized by electron impact.²³ The He⁺ + M → He + M⁺ charge transfer reaction ionizes the He-solvated molecular complex, which typically results in molecular fragmentation and the complete evaporation of the droplet.^36,37 Gas-phase ions subsequently enter the quadrupole mass spectrometer (QMS) and are detected by an electron multiplier. The current from the electron multiplier is converted into a voltage, processed by a lock-in amplifier, and recorded by a custom Labview data collection program.

The tunable, mid-IR (3150–4100 cm⁻¹) idler beam from a continuous-wave optical parametric oscillator (cw-OPO) either counter-propagates the droplet beam or intersects the droplet beam at nearly right angles in the Stark/multipass cell. Upon rovibrational excitation, the internal energy of the dopant is quickly quenched by the evaporative loss of many He atoms, with an individual He atom carrying away ≈5 cm⁻¹ of internal energy.³⁸ This geometric cross section change reduces the probability for ionization by electron impact. Ion-signal depletion is therefore observed in mass channels associated with ionization-induced dopant/droplet fragmentation. The IR radiation is mechanically chopped at 80 Hz and continuously tuned from 3150 to 3530 cm⁻¹ with ≈10 MHz resolution,³⁹ enabling background-free measurement of the laser-induced depletion signal. In most of the current work, the quadrupole is set to pass only ions with m/z = 51 u. The relevant ions of this mass correspond to the ³⁵ClNH₂⁺ species; however, droplets containing (NH₃)_n≥3 clusters may be ionized (and fragment), producing laser-induced depletion in the 51 u mass channel (N₃H₉⁺). Nevertheless, the experimental conditions minimize the production of ammonia clusters larger than the dimer. Moreover, the 51 u mass channel largely discriminates against complexes containing the ³⁷Cl isotope. Spectral simulations of experimental rovibrational bands were carried out with PGOPHER software.⁴⁰ 

Following our recent work on hemi-bond complexes,^7,41 structures of NH₃, NH₃Cl, and the NH₄Cl complex were optimized using the coupled-cluster method (unrestricted coupled-cluster method for the open shell molecule NH₃Cl) with singles, doubles, and perturbative triples excitations (CCSD(T)/UCCSD(T)),^42,43 with the augmented correlation consistent polarized valence triple zeta (aug-cc-pVTZ or AVTZ) basis.⁴⁴ The harmonic frequencies of the molecules were further calculated at the same CCSD(T)/AVTZ level of theory. These calculations have been carried out using MOLPRO 2010.1.⁴⁵ For the infrared intensity calculations, the CCSD method⁴⁶ with AVTZ basis was employed in GAUSSIAN 09.⁴⁷ 

Figure 3 is a survey spectrum covering the relevant NH stretching region. The survey scan was carried out with Cl₂ flowing through the hot pyrolysis source, 10⁻⁶ Torr of NH₃ in the pickup cell, and the QMS set to pass ions having m/z = 51 u. As described above, the spectrum corresponds to the laser-induced cross section reduction for droplets producing m/z = 51 u upon ionization in the mass spectrometer. Spectral signatures of NH₃ monomer and dimer do not appear in this mass channel. Several broad, weak features in the survey spectrum are assigned to (NH₃)_n≥3 clusters via comparison to previous work by Vilesov and co-workers.⁴⁸ Three distinct regions near 3350, 3450, and 3500 cm⁻¹ contain intense laser-induced depletion signals. To determine the origin of these features, these regions were re-measured with Cl₂ flowing through the pyrolysis source at room temperature. Only the features straddling 3450 cm⁻¹ are present under these “cold-pyrolysis” source conditions, and because these bands cannot be ascribed to NH₃ or NH₃ clusters alone, they must arise from complexes consisting of at least one Cl₂ molecule and one NH₃ molecule. Spectral features in the 3350 and 3500 cm⁻¹ regions, however, are only present upon heating the SiC pyrolysis furnace (T ≈ 1600 K). These bands likely arise from species produced via the Cl + NH₃ reaction carried out within He droplets. The analysis of partially resolved rotational fine-structure exhibited by bands near 3350 and 3500 cm⁻¹ provides strong evidence for an assignment to symmetric and antisymmetric NH₃ stretches of a ClNH₃ molecular complex (either CR2 or CR3). As described below, Stark spectroscopy reveals the structural identity of this ClNH₃ species.

The generation of gas-phase Cl atoms yields a low background of HCl within the pickup chamber. To rule out the possibility that the observed bands in Figure 3 are due to formation of ClH–NH₃ complexes, the 3350 and 3500 cm⁻¹ regions were measured on m/z = 51, 52, and 53 u with the pyrolysis source turned off and both HCl and NH₃ (taken from lecture bottles) present in the pickup cell. The (³⁵ClNH₃)⁺ species may contribute to the 52 u mass channel. (Throughout the remainder of this article, references to Cl correspond to the more abundant ³⁵Cl isotope unless otherwise indicated.) There are no observed vibrations of an ClH–NH₃ complex in either the 3350 or 3500 cm⁻¹ regions; however, measurements in the 52 u channel reveal K sub-bands between 3430 and 3460 cm⁻¹ (Fig. 4) consistent with the antisymmetric stretch of the ClH–NH₃ binary complex. This signature of the ClH–NH₃ complex is not observed in the “hot-pyrolysis” survey spectrum (Fig. 3), indicating that the subset of droplets containing ClH–NH₃ are discriminated against when measuring depletion in the 51 u channel. The ClH–NH₃ band origin (3439.75 cm⁻¹) lies very close to the ν₃ band origin of He-solvated NH₃ at 3443.1 cm⁻¹.⁴⁹ Consistent with this, harmonic frequency calculations at the CCSD(T)/AVTZ level predict small complexation-induced shifts of the NH₃ stretching vibrations (Table I). No symmetric NH₃ vibration was located for this complex, which is consistent with the small intensity (0.0042 km/mol at the CCSD/AVTZ level) computed for this mode.

The ClH–NH₃ antisymmetric stretch, perpendicular band (e symmetry) is satisfactorily simulated using a symmetric top Hamiltonian. The observation of three K sub-bands is consistent with the expected nuclear spin statistics associated with this C_3v complex, which contains three equivalent H atoms.⁵⁰ The B rotational constant was set in the simulation to three times less than the ab initio equilibrium value (Table I) to account for coupling of the He environment to b-axis rotation.^22–24,51 In comparison, the coupling of the He solvent to the faster rotational motion about the a-axis is expected to be less efficient.⁵² The A rotational constant and the band origin were adjusted to satisfactorily simulate the observed band, revealing an A constant (5.89 cm⁻¹) in qualitative agreement with the ab initio equilibrium value (6.27 cm⁻¹). Indeed, the contribution of the He solvent (ΔI_a = 0.17 amu-Ų) to the effective moment of inertia for rotation about the a-axis is small in comparison to b-axis rotation (ΔI_b ∼ 240 amu-Ų). The K = 2←1 sub-band is significantly broadened in comparison to the others, an effect observed previously for other He-solvated symmetric tops, such as NH₃⁴⁹ and CH₃.⁵³ The origin of this broadening is related to the three-fold symmetry of the molecule-He interaction potential, which leads to a strong propensity for changes in the K quantum number by ±3.⁵⁴ As a result, rotational dephasing/relaxation via the |K| = 2 → |K| = 1 pathway is expected to be efficient, consistent with the observed homogeneous broadening of the K = 2←1 sub-band.⁵³ 

An expanded view of the intense vibrational band near 3359 cm⁻¹ observed in the “hot-pyrolysis” survey scan is shown in Figure 5. This band is well simulated using a symmetric top Hamiltonian and a₁ vibrational symmetry (bottom of Fig. 5). The simulation requires equal nuclear spin statistical weights for A₁/A₂ and E rotational levels. These weights are expected for a system like ClNH₃, which belongs to the C_3v(M) molecular symmetry group and has three equivalent H atoms.⁵⁰ Upon capture by He droplets, the ClNH₃ population cools into levels with K = 0 (A₁ or A₂ rotational symmetry) and K = 1 (E rotational symmetry). Because nuclear spin conversion is slow on the time scale of the measurement,⁵⁵ the observed spectrum is comprised of two sub-bands obeying the symmetric top, parallel band (a₁ symmetry), ΔK = 0 selection rule.⁵⁰ 

The band origin and rotational constants used for the simulated spectrum are given in Table I alongside values computed for both isomers at the CCSD(T)/AVTZ level of theory. Although the rotational constants for the CR2 and CR3 isomers are too similar to settle on a definitive assignment on the basis of the zero-field rotational fine structure, the constants used in the simulation are in agreement with the computed values for both, given the above discussion of expected He contributions to effective moments of inertia. The rotational line positions are not sensitive to the absolute magnitude of the A′ or A″ rotational constants, but they are rather sensitive to the difference between the two. The best agreement is obtained when ΔA (A′–A″) = − 0.045 cm⁻¹, indicating an increase in I_a, consistent with the expectation of elongated NH bonds upon vibrational excitation. The experimental B constant (0.075 cm⁻¹) is reduced by a factor of ≈3 relative to the computed B constant (0.24 cm⁻¹) due to the aforementioned contribution of the He solvent to the moment inertia about the b-axis. The observed band origin is blue-shifted by ≈22 cm⁻¹ relative to the NH₃ ν₁ band origin, and this is consistent with a shortening of the N–H bond distances, as predicted for the geometries of the CR2 and CR3 isomers. Using an average scale factor of 0.96, derived from comparing the experimental and computed harmonic ν₁ and ν₃ vibrations of NH₃, the experimental bands best agree with the scaled harmonic frequencies of the CR2 isomer (hydrogen atoms pointing away from the Cl atom).

This partially rotationally resolved band was measured in both 51 and 53 u mass channels to verify the formation of both ³⁵ClNH₃ and ³⁷ClNH₃ complexes. Although the fine structure between the two spectra remains essentially unchanged at our resolution, the reduced depletion signal observed on m/z = 53 u is qualitatively consistent with the ³⁵Cl : ³⁷Cl ≈ 3 : 1 ratio associated with the natural abundance of Cl isotopes.

Several features are observed in the survey spectrum near 3500 cm⁻¹, which is expanded in Figure 6. Three of these are assigned to ΔK = ± 1 rotational sub-bands of a perpendicular vibrational excitation (e symmetry), in accordance with the symmetric top rotational selection rule.⁵⁰ This perpendicular band allows for the determination of a vibrationally averaged A rotational constant, in addition to the band origin (3501.05 cm⁻¹). Assuming this band is derived from the same carrier associated with the parallel band at 3358 cm⁻¹, it is reasonable to use the B and D_J extracted from the parallel band in a simulation of the perpendicular band. With these constants fixed in the simulation, best agreement is obtained when the A″ and A′ constants are set to 5.45 cm⁻¹. This is in good agreement with the computed A constants for both the CR2 and CR3 isomers, where only a small renormalization is induced by the He environment (ΔI_a = 0.24 amu-Ų). As observed and described above for the perpendicular band of the ClH–NH₃ complex, the K = 2←1 sub-band is significantly broadened beyond the rotational contour of the K = 0←1 and K = 1←0 sub-bands.⁵³ The line width of this homogenously broadened transition implies an upper state lifetime of ∼1.6 ps. The remaining bands observed in this region, marked by * in Figure 6, were optimized at higher background pressures and are likely due to unpyrolyzed Cl₂ effusing through the pyrolysis source at higher flow rates. Given the droplet pickup process obeys Poisson statistics,^22–24,56 these bands are likely due to the antisymmetric NH₃ vibrations of (Cl₂)_nClNH₃ isomers, although no effort to further characterize their origin was undertaken.

The features in the spectrum near 3500 cm⁻¹ (Fig. 6) are consistent with the large blue-shifts predicted for the antisymmetric NH₃ vibrations (relative to the NH₃ monomer) of the CR2 and CR3 isomers. The aforementioned scale factor of 0.96 applied to the frequency predictions for the antisymmetric NH₃ vibrations again yields slightly better agreement between the observed band origin and that predicted for CR2. Comparisons to computations strongly suggest that the 3358 and 3501 cm⁻¹ bands should be assigned to the symmetric (a₁) and antisymmetric (e) stretches of the CR2 ClNH₃ complex formed via the He-mediated Cl + NH₃ reaction.

Because the CR2 and CR3 isomers are both symmetric top molecules having similar computed rotational constants, it is not possible to arrive at a definitive assignment of the observed bands on the basis of rotational fine structure analysis alone. Again, the comparison between computed band origins is strongly suggestive of the aforementioned assignment to CR2. We employ Stark spectroscopy to measure the permanent electric dipole moment, which is predicted to be quite different for the two isomers in question. Indeed, the dipole moments for CR2 and CR3 are 4.08 D and 2.21 D, respectively, at the CCSD(T)/AVTZ level (Table I). Given this large difference and the relatively good rotational resolution of the parallel band at 3358 cm⁻¹, the Stark effect can be used to determine which of the two isomers is formed following sequential addition of Cl and NH₃ to He droplets. When comparisons are available, the dipole moments of He-solvated molecules are close to their gas-phase values (within a few percent) as a result of the small net polarization of the He solvent.³² The spectra shown on the left of Figure 7 were obtained at the indicated electric field strengths with the polarization of the laser beam oriented perpendicular to the applied electric field, resulting in ΔM = ± 1 selection rules. The excellent agreement between the simulations (μ = 4D) and the experimental spectra confirms the assignment of the 3358 cm⁻¹ band to the symmetric NH₃ stretch of the CR2 complex. Indeed, simulations with μ = 2.2D yield Stark spectra that qualitatively disagree with the observed spectra at all field strengths, ruling out the possibility that the band arises from the CR3 isomer. Because the presence of an intense CR2 symmetric stretch band implies the presence of an intense antisymmetric stretch band (on the basis of intensity calculations at the CCSD/AVTZ level), we also take the Stark measurement as confirmation of the 3501 cm⁻¹ band assignment to the antisymmetric NH₃ stretch of the CR2 complex.

All spectroscopic evidence points to the exclusive formation of the CR2 complex upon Cl + NH₃ association within He droplets. The formation and stabilization of entrance channel complexes is expected due to the dissipative nature of He droplets. Indeed, the computed barrier heights to hydrogen abstraction and Cl substitution are both well above the asymptotic reactant energy (Fig. 2).^6,14 It is interesting, however, that while CCSD(T) computations predict three stable isomers for ClNH₃, there is no spectroscopic evidence for either the C_3v “hydrogen-bound” complex (CR3) or the C_s van der Waals complex (CR1). Although it was noted in earlier work that CR1 and CR3 are entrance complexes along the hydrogen abstraction path, whereas CR2 lies in the entry valley to the substitution reaction path,^6,14 interconversion barriers between these complexes were not considered. CR3 lies ∼1000 cm⁻¹ higher in energy than CR2.⁶ Because the He atom evaporation process is expected to preclude the uphill CR2 → CR3 isomerization upon direct formation of CR2, a facile interconversion path leading from CR3 to CR2 could explain the lack of the former. With CCSD(T) computations, we find the interconversion pathway to be largely composed of NH₃ inversion and Cl-NH₃ intermolecular stretching. Despite many attempts, however, it was not possible to locate a transition state between CR3 and CR2, apparently due to the potential being rather flat in the vicinity of the CR3 minimum. We estimate the interconversion barrier by computing a one-dimensional potential curve along a distinguished reaction path. The potential shown in Figure 8 is obtained as the ClNH angle is scanned between the CR3 and CR2 geometries with other coordinates fixed at the geometry of CR3 while maintaining C_3v symmetry. Along this distinguished reaction path, the computed NH₃ inversion barrier is estimated to be <53.5 cm⁻¹, revealing the facile interconversion pathway necessary for exclusive formation of CR2. The interconversion path is found to be barrierless at the MP2/AVTZ level, and we estimate on the basis of frequency calculations that the CCSD(T)/AVTZ barrier is somewhat smaller upon zero-point correction. Provided the He evaporative cooling is insufficiently rapid to trap the CR3 species, the CR2 minimum can be accessed indirectly via interconversion of CR3. Moreover, we also note that Cl + NH₃ approach trajectories should favor the direct formation of CR2 in He droplets, because the long-range intermolecular interaction for Cl-NH₃ approach is somewhat stronger (e.g., larger dipole moment) in comparison to NH₃–Cl approach. Similar arguments can be invoked to explain the absence of the van der Waals complex, CR1. Despite many attempts, the barrier to CR1 → CR3 interconversion was not found, again due to the floppy nature of the system in the vicinity of the CR1 minimum.

The MO diagram and wavefunctions obtained at the B3LYP/AVTZ level for the ClNH₃ complex (CR2) and its constituents are shown in Figure 9. Because the Cl half-filled p-orbital and the NH₃ HOMO are similar in energy, orbital mixing is efficient and leads to the bonding and antibonding MOs shown in the center of the figure. The bonding MO is more stabilizing than the anti-bonding MO is destabilizing, revealing a partial covalent character to the bonding consistent with the ∼8 kcal/mol binding energy predicted at the CCSD(T)/AVTZ level. Covalent mixing of the NH₃ lone-pair orbital with the singly occupied p-orbital on Cl can be viewed as affecting a rehybridization of the NH₃ moiety, as the p-orbital contribution to the three NH bonds is diminished in favor of its contribution to N–Cl bonding. This effect leads to an opening of the ∠HNH angle (∼4°) and a shortening of the N–H bond lengths (0.005 Å). Indeed, this prediction is completely consistent with the complexation-induced blue shifts observed for both the symmetric (+22.6 cm⁻¹) and antisymmetric (+58.0 cm⁻¹) NH₃ stretching vibrations. Furthermore, this rehybridization makes the NH₃ moiety in the ClNH₃ complex more “D_3h–like,” consistent with the small CR3 to CR2 interconversion barrier discussed above, which in comparison to NH₃ inversion (∼1770 cm⁻¹),^57–59 represents a rather significant change.

Sequential addition of a Cl atom and NH₃ to a helium droplet leads to the formation of a C_3v complex, ClNH₃ (CR2), which exhibits a strong two-center three-electron (2c–3e) Cl–N hemi-bond. This ClNH₃ species is an entrance channel complex leading into the substitution reaction (Cl + NH₃ → ClNH₂ + H). Assignment of the infrared spectrum in the NH stretch region to this entrance complex is confirmed via comparisons to high-level electronic structure theory vibrational band origin computations, the analysis of rotational fine structure that reveals the presence of both parallel (a₁ symmetry) and perpendicular NH₃ stretching bands (e symmetry) and Stark measurements of the complexes’ permanent electric dipole moment (4 D). Another C_3v complex, NH₃Cl (CR3), and a weakly interacting Cl–HNH₂ complex (CR1) are predicted as stable minima in the entry valley to the hydrogen abstraction reaction (Cl + NH₃ → HCl + NH₂). Neither of these species are observed upon cluster formation in helium droplets. This observation is rationalized via computations of the interconversion pathways between the three entrance channel complexes. Orbital mixing between the Cl atom’s half-filled p-orbital and the NH₃ HOMO leads to the 2c–3e bond and a rehybridization of the NH bonds (loss of p-character). This rehybridization results in a more planar NH₃ moiety, which affects the inversion barrier and therefore the CR3 → CR2 interconversion barrier. The interconversion between these entrance channel complexes is predicted to be both downhill and almost barrierless (∼50 cm⁻¹), consistent with the absence of CR3 upon cluster formation in helium droplets.

G.E.D. and H.G. acknowledge support from the Office of Energy Research, Office of Basic Energy Sciences, Chemical Sciences, Geosciences and Biosciences Division of the US Department of Energy (DOE) under Contract Nos. DE-FG02-12ER16298 (to G.E.D.) and DE-FG02-05ER15694 (to H.G.) for support. The calculations were performed at the Center for Advanced Research Computing (CARC) at UNM and National Energy Research Scientific Computing (NERSC) Center.