Imaging Reaction Dynamics of F-(H2O) and Cl-(H2O) with CH3I.

The dynamics of microhydrated nucleophilic substitution reactions have been studied using crossed beam velocity map imaging experiments and quasiclassical trajectory simulations at different collision energies between 0.3 and 2.6 eV. For F-(H2O) reacting with CH3I, a small fraction of hydrated product ions I-(H2O) is observed at low collision energies. This product, as well as the dominant I-, is formed predominantly through indirect reaction mechanisms. In contrast, a much smaller indirect fraction is determined for the unsolvated reaction. At the largest studied collision energies, the solvated reaction is found to also occur via a direct rebound mechanism. The measured product angular distributions exhibit an overall good agreement with the simulated angular distributions. Besides nucleophilic substitution, also ligand exchange reactions forming F-(CH3I) and, at high collision energies, proton transfer reactions are detected. The differential scattering images reveal that the Cl-(H2O) + CH3I reaction also proceeds predominantly via indirect reaction mechanisms.

Introduction

Reaction dynamics of the important class of bimolecular nucleophilic substitution (SN2) reactions have been studied extensively for X– + CH3Y model systems with several combinations of halide anions and methyl halides.[1−5] Detailed information on gas phase reaction dynamics is obtained by measuring differential cross sections of bimolecular reactions in crossed molecular beams under single collision conditions. The in-depth gas phase picture obtained from experimental evidence and simulations often provides a good starting point to interpret reaction dynamics in liquids, which has become experimentally accessible by time-resolved infrared spectroscopy.[6]

The gas phase approach to understanding solvent effects in ion–molecule reactions is the addition of single solvent molecules to the nucleophile, which is referred to as microsolvation.[7] Especially in the case of protic solvents, preferential stabilization of the reactants relative to the transition state leads to transition from a double-well potential energy surface (PES) in the gas phase to a higher reaction barrier and unimodal profile in solution.[8] The consequence is smaller reaction rates by several orders of magnitude. If the solvent cannot rearrange to follow charge in the reaction intermediate, desolvation of the reactants and unsolvated products imply a lower exothermicity. Therefore, already the addition of a few solvent molecules will strongly quench the reaction.[9,10] This effect is important in microsolvated SN2 reactions that typically avoid the energetically favored solvated products—as opposed to reactions in solution where different molecules concertedly desolvate the nucleophile and solvate the leaving group.[11] When SN2 reactions become inefficient by stepwise solvation, ligand exchange can become important as has been reported for Cl–(D2O)1–3 and F–(H2O)4–5 reactions with CH3Br.[12,13]

This work focuses on the highly exothermic halide-exchange SN2 reaction F–(H2O) + CH3I with a single water solvent. The reactivity of the F– + CH3I reaction at 302(2) K decreases from 19.4(2) × 10–10 cm3 s–1 for the solvent-free case to 8.64(9) × 10–10 cm3 s–1 upon single H2O solvation of the anion, whereas reactivity is decreased by a factor of 100 in the less exothermic reaction with CH3Cl. The observed I– to I–(H2O) product branching is 9:1.[14]

The PESs of the reactions with F–(H2O)[15] and OH–(H2O)[16] are overall similar to those of the water-free systems. They allow initial association either in a hydrogen-bonded X–(H2O)···HCH2I complex or in an ion–dipole X–(H2O)···CH3I complex with a small barrier in between. According to the energetics, the latter may undergo Walden inversion with the water molecule, driven by H-bonding to the halide atoms, shifting to the iodine side. However, atomistic dynamics deviate considerably from the intrinsic reaction coordinate, as more than 90% of the reactive trajectories avoid water-bound postreaction complexes in low energy collisions.[17] Single water molecules mostly dissociate from F– at the moment of nucleophilic displacement. In the case of higher hydrated F– ions, additional water molecules detach early in the entrance channel and SN2 occurs through the less solvated barrier with higher energy but diminished steric hindrance.[18] Direct dynamics studies on the related reactions of F–(H2O) and OH–(H2O) with CH3Cl have revealed a strong effect of the relative H2O position in the entrance geometry on product channels and reaction probability.[19,20]

Computed reactive cross sections for F–(H2O) + CH3I collisions at 0.32 eV yield 4.4(13)% I–(H2O), 9(6)% CH3F(H2O), and three-body dissociation in most cases.[17] For all three channels, indirect mechanisms are dominant with fractions of about 70%, of which ca. 90% pass through the hydrogen-bonded prereaction complex. Also direct rebound (DR) is reported for all three channels, whereas direct stripping (DS) always leads to I– formation. High level ab initio stationary points suggest additional mechanisms at higher collision energies such as front-side attack, double inversion, and proton transfer from H2O to F– followed by OH– driven SN2 forming methanol.[21] Proton transfer from CH3I opens at much higher collision energies.[22]

At 1.53 eV collision energy, direct dynamics simulations do not observe any I–(H2O) and only 1.2(5)% CH3F(H2O).[23] In about 70% of the indirect reactions, water dissociates at the initial collision. Subsequently, the water-free prereaction complexes are formed such that SN2 resembles ligand exchange followed by fragmentation of the FCH3I– complex. Indirect mechanisms still dominate at higher energy, which is attributed to steric effects. The early dehydration favors nucleophilic attack and therefore attenuates the suppression of reactivity by the solvent. Furthermore, methanol formation is indeed observed in about 10% of the trajectories.

When the nucleophile is changed to Cl–(H2O), only an ion–dipole complex is found in the entrance channel.[24] The predominant mechanism at high collision energies is direct rebound instead of indirect mechanisms, and a roundabout mechanism dominates indirect reactions instead of complex formation. At 1.9 eV, simulations found early water loss, which leads to similar reaction probabilities, mechanisms, and energy and angular distributions as in the unsolvated reaction.[24] At lower collision energies, different solvated dynamics are expected, as also the Cl– + CH3I mechanisms depend significantly on collision energy.

Previous crossed beam experiments and direct dynamics simulations on the solvent-free F– + CH3I reaction revealed the importance of a hydrogen-bonded prereaction complex that leads to deviations from the traditional Walden inversion pathway.[25] Also a halogen-bonded front-side complex is important at low collision energies.[26,27] Complex formation results in a large contribution of indirect dynamics to the scattering images which was unexpected and is still important at rising collision energies.[28] Agreement of energy and angular distributions with simulation results also permitted identification of direct stripping as a third mechanism. Furthermore, two retention pathways have been found as minor pathways based on trajectories on an accurate analytical PES[29] and competing reaction channels at higher collision energies have also been experimentally investigated.[30] In the Cl– + CH3I system, complex formation does not occur and dominant indirect dynamics is only observed at low collision energies. Scattering images at 1.1 eV reveal direct rebound as the predominant mechanism,[31] indicative for a collinear approach with Walden inversion. An indirect roundabout mechanism appears at 1.9 eV and has been characterized using direct dynamics simulations.[32]

Simulations of reaction dynamics under microsolvation have been extended to non-halide anions[33] and reactions with ethyl halides[34] that involve intricate competition effects with the additional E2 elimination channel.[35,36] At the same time, extensive experimental results on reaction dynamics of these systems have only been reported for the OH–(H2O) + CH3I reaction. Previous crossed beam studies with our setup cover different solvation levels n = 0, 1, 2 in the 0.5–2 eV collision energy range[37,38] and variation of the anion water cluster temperature.[39] Most interestingly, the hydrogen-bonded complex, similar to the isoelectronic reaction with F–, avoids the traditional collinear nucleophilic attack—but addition of a single water molecule sterically facilitates the collinear approach and therefore enhances direct rebound at intermediate and high collision energies.[37] Upon addition of a second water molecule, direct mechanisms are completely suppressed.

In order to test the different theoretical simulations, insight from reactive scattering experiments is needed. In the present work, the differential scattering cross sections have been studied for the reactions of F–(H2O) and Cl–(H2O) with CH3I using crossed beam ion imaging. This allows for detailed information on the atomistic dynamics and the interplay of different reaction mechanisms. In particular, it provides a test of the product branching ratios and sheds new light on the effect of the solvent molecule on the atomistic reaction dynamics. Improved data for the unsolvated reaction F– + CH3I allows us to extract the relative contributions of different atomistic mechanisms in the SN2 reaction. Furthermore, the Cl–(H2O) data provide insight into the role of the nucleophile and complement recent direct dynamics simulations with experimental evidence at lower collision energies.

Experimental and Theoretical Methods

Crossed Beam Imaging

Angle and energy differential cross sections of the charged products of ion–molecule reactions are measured by three-dimensional velocity map imaging[40] (VMI) in a crossed beam setup.[41] Precursor gases are ionized by plasma discharge in a supersonic expansion from a pulsed piezo cantilever valve. Ions are then thermalized in a radio frequency trap[37] with room temperature buffer gas, in the present work typically argon. After 40 ms, ions are accelerated to the desired kinetic energy and cross a molecular beam that is seeded with the neutral reactant in the center of the VMI spectrometer. Position and flight time are recorded by a combination of multichannel plates with a phosphor screen, digital camera, and photomultiplier tube. Product ions are discriminated by their flight time, and each pair of position and time is converted into the velocity vector in the center-of-mass frame. Symmetry about the collision axis permits mapping of the velocity vectors to two components parallel (v) and perpendicular (v) to the axis. To mimic slice distributions in the scattering plane, ion counts are weighted by v–1. Background from the ion beam is recorded with asynchronous timing of the neutral beam and subtracted. Collision energy and center-of-mass velocity in the scattering plane are determined from ion and neutral beam distributions recorded by two-dimensional VMI. For this, the neutral beam is ionized by electron impact.

For comparison with measured arrival times on the VMI detector, flight times of ions with different masses were determined from ion trajectory simulations using SIMION.[42] In particular, simulations have been performed for metastable product ion complexes that may dissociate during acceleration in the field of the VMI spectrometer. Apparent masses as a function of the moment of dissociation were calculated in steps of 0.2 μs. Details about the specific VMI geometry and potentials have been published in ref (43).

F– anions were formed from NF3 diluted in argon. The same mixture was bubbled through distilled water in a gas washing bottle with filter plate in order to have F–(H2O) clusters form in the dense part of the supersonic expansion. As other ions (FHO–, F2–, and HF2–) with close-lying masses were also present in the plasma, mass separation was required to prepare the F–(H2O) reactant. Time-of-flight separation together with pulsed opening of the radio frequency trap allowed for the manipulation of the reactant ion composition and suppression of the unwanted coreactants. The reactant branching ratios were estimated from time-of-flight traces obtained by velocity mapping of the ion beam at the same moment at which product ions were imaged in reactive scattering. Overshoots in the time-of-flight traces were approximately corrected.[44] At 2.6 eV collision energy, two measurements were performed with H2 instead of Ar buffer gas to further suppress FHO– contamination. At the same time, the amount of HF2– could be reduced by earlier timing of the trap entrance. The remaining amount of coreactants has been found to have only a minor relevance for the results on the F–(H2O) reactions (see the Supporting Information).

Cl– anions were formed from CH3Cl diluted in argon with a pulsed discharge stabilized by a static electron source. Dissolution of air in the distilled water bottle was found to lead to the formation of the unwanted coreactant O2–(H2O). To avoid this, a pure Cl–(H2O) reactant beam was obtained without the washing bottle after pure water was accumulated in the mixing bottle of the precursor gas.

Electronic Structure Calculations and Dynamical Simulations

Ab initio calculations at the coupled cluster singles and doubles (CCSD) level using the aug-cc-pVTZ-PP basis set for iodine and the aug-cc-pVTZ basis set for all other elements, further denoted as CCSD/aug-cc-pVTZ(-PP), have been performed to obtain a coherent set of exothermicities for the reactive channels of the CH3I reaction with the anions of interest F–, F–(H2O), and Cl–(H2O) using the Gaussian 16 program.[45] Further computations have been performed for the unwanted coreactants FHO–, F2–, and HF2–. Wave function stabilization was performed for every structure. Zero-point energy is included in all reported energies. Energetics and structural data for all anions is given in the Supporting Information. Analysis of the experimental data was based on the following exothermicities of the most probable channels for the respective charged product, which were computed at the CCSD(T) level. The energy levels of the reactants and products of the reaction pathways with F– and F–(H2O) are summarized in Figure .

Figure 1

Energy level diagram relating the reactants and products of the monohydrated and solvent-free F– + CH3I reactions. Only the SN2 and proton transfer channels, for which the product velocity images are discussed in the present work, are shown. Different collision energies of the reactive scattering measurements are denoted by arrows above the energy level of the respective reactants. Dotted lines indicate the threshold for the proton transfer channel and collision energies for the two reactions that correspond to similar excess energies above threshold.

For a detailed comparison, we constructed a global PES for the reaction of F–(H2O) and CH3I, using the fundamental invariant neural network (FI-NN[46]) fitting method. With the use of Gaussian 09,[48] 141,921 XYGJ-OS[47]/aug-cc-pVTZ (aug-cc-pVTZ-PP for iodine atom) electronic energies were calculated. XYGJ-OS is a fast doubly hybrid density functional method, whose overall accuracy is close to chemical accuracy.[47] The geometries used in the fitting were properly selected based on direct simulations and further quasiclassical trajectory (QCT) calculations using the preliminary PESs iteratively. The final PES is well-converged with respect to the fitting errors and the results of dynamical simulations. The final fitting root-mean-square error (RMSE) is 13.8 meV for energies up to 3.0 eV relative to the reactant F–(H2O) + CH3I.

The QCT simulations were carried out at the collision energies of 0.3, 1.0, and 1.5 eV for the F–(H2O) + CH3I reaction on the new PES described above. Quasiclassical vibrational ground states were prepared for initial reactants F–(H2O) and CH3I by normal mode sampling. The total angular momentum was set to zero by initial momentum adjustments. The initial distance between the centers of mass of F–(H2O) and CH3I was (x2 + b2)1/2, where b is the impact parameter and x is set to 30a0. The orientation of F–(H2O) was randomly sampled with respect to CH3I and b was sampled uniformly between 0 and bmax, where the bmax values were 18a0 at 0.3 eV, 14a0 at 1.0 eV, and 12.6a0 at 1.5 eV. One million trajectories were computed at each energy. The trajectories were terminated when the distance between the collisional product species (for the channel with three products, the value is defined as the minimum distance between three species) reached 20a0 or the maximum distance among all nine atoms reached 30a0.

Results

Product Channels for F–(H2O) Reactions

Product ion mass spectra for the reactions of F– and F–(H2O) with CH3I at 0.3 eV collision energy in Figure show I– as the only product of the unsolvated reaction. In the singly solvated reaction, the unwanted coreactants FHO–, F2–, and HF2– were present at a 3–14% level compared to 100% F–(H2O). The contribution of the different reactants to the observed product ions is analyzed in the Supporting Information. Altogether, 6(3)% of the products stem from FHO–, and the contributions by F2– and HF2– are bound by upper limits of 1.2(3) and 7(7)%. F–(H2O) accounts for at least 86(8) and 94(3)% of the I– and I–(H2O) products. A derived difference mass spectrum for the pure F–(H2O) + CH3I reaction is shown as inset in Figure . Besides the main I– peak, the solvated SN2 product I–(H2O) and the ligand exchange reaction forming FCH3I– are observed. In the hydrated reaction, the I– peak is accompanied by a broad tail of apparently higher masses ranging up to 142 u, which is also assigned to I– and accounts for 11(3)% of the total I– signal area. Including the tail contribution, the extracted product branching ratios of the F–(H2O) + CH3I reaction are 89.3(11)% for I–, 4.5(9)% for I–(H2O), and 6.2(7)% for FCH3I– at 0.32 eV collision energy. Values in parentheses are statistical errors in units of the last digit. The peak integration procedure involves additional systematic uncertainties at the percent level.

Figure 2

Observed product mass spectra for reactions of CH3I with F– and with F–(H2O) in the presence of coreactants. A 8× zoom is shown as inset including a mass spectrum for pure F–(H2O) derived from three observed spectra with different reactant compositions.

The tail of apparently higher masses up to 142 u can be explained by the dissociation of an intermediate complex, candidates being I–(H2O) and FCH3I–, in the acceleration region of the VMI spectrometer, which results in a flight time between those of the heavier and lighter species. Further evidence is given by mass spectra at 2.6 eV collision energy with VMI pulses first, as usual, 0.4 μs after the ion beam peak, and second, delayed by another 2.2 μs after the ion packet has passed the VMI center, in order to avoid acceleration of unstable complexes. The late pulse preserves the total I– intensity and the width of the high mass tail, but reduces the tail fraction from 38(5) to 25(6)%. We conclude that fewer complexes are accelerated before their final dissociation.

To test this, we have computed apparent masses as a function of dissociation time with respect to the start of VMI acceleration. The result is shown in Figure S3.2. For FCH3I– dissociating to I–, the apparent mass increases by 4 u μs–1 from 127 to 143 u at 4 μs and then, in a region of strong acceleration, by 16 u μs–1 up to 161 u near 5 μs. Mass loss during the onset of strong acceleration in the 3–4 μs range corresponds to apparent masses of 139–144 u. The observed I– tail vanishes near 142 u, which corresponds to lifetimes near 4 μs.

The fraction of solvated product ions I–(H2O) relative to total SN2 reactivity, derived from the measured product mass spectra, is plotted as a function of collision energy in Figure . All product mass spectra at 1.1, 1.6, and 2.6 eV collision energies, reactant compositions, and integrated product branching ratios are provided in the Supporting Information. With increasing collision energy, the I– tail to peak fraction rises continuously, roughly doubling from 0.3 eV to the highest energy. At 1 eV, the solvated SN2 product I–(H2O) amounts to about 2%, which is near the detection limit, as is visible from the error bar in Figure . In turn, proton transfer opens up at 2 eV and is discussed in section . Ligand exchange forming FCH3I– is slowly suppressed for larger collision energies. Dihalide formation becomes energetically accessible near 2 eV. The formation of FI– or FHI– by F–(H2O) is compatible with an intensity increase in this mass range at 2.6 eV. An unequivocal assignment to these channels is impeded by the coreactants. However, large differences by factors of 3–6 of the proportion of HF2– and FHO– in the reactant ion beam composition at 2.5 eV collision energy do not have a noticeable effect on the I– and I–(H2O) product ion velocity images, energy, and angular distributions. This justifies interpreting them as results of the F–(H2O) + CH3I reaction not only at 0.3 eV, but also up to the highest investigated collision energy.

Figure 3

Fraction of I–(H2O) from the total SN2 reactivity in comparison of experiment and theory. Only I– in the peak at the nominal mass is considered for the experimental value, because products from long-lived intermediates were not traceable in simulations.

Figure also shows the fraction of solvated SN2 reaction products obtained from the QCT simulations. The trajectory simulations were terminated at a chosen distance between the products. In parts, the obtained FCH3I– product complexes may therefore be metastable. For comparison, the experimental I–(H2O) fractions in Figure were computed without including the I– tails that stem from late complex dissociation.

Nucleophilic Substitution of F– and F–(H2O)

The reactions F– + CH3I and F–(H2O) + CH3I have been measured under similar conditions at four different collision energies ranging from 0.3 to 2.6 eV. Velocity images of the I– product ion are presented in Figure in the center-of-mass frame that is depicted by the Newton diagram above. Left and right half-planes correspond to forward and backward scattering of the product ion relative to the neutral reactant. Outer circles indicate the maximum ion kinetic energies (kinematic cutoffs) that correspond to the maximum kinetic energy Erel′ of relative motion between the ion and the center of mass of the neutral products. It is given by the sum of the average collision energy Erel plus the computed exothermicities Eexo (see section ). Inner circles correspond to 1 eV differences in terms of lower Erel′. Using the energy differencewhere Erelneutral denotes the kinetic energy of relative motion between the neutrals, the kinematic cutoff is described by ΔE = 0. Erelneutral is zero in case of a single neutral product such that ΔE = Eint – Eint0 gives a direct measure, while in general ΔE ≥ Eint – Eint0 gives an upper bound to the internal excitation Eint of all products. Eint0 is dominated by the thermal energy of the anion–water cluster trapped with buffer gas at room temperature. The initial internal excitation energy of reactants Eint0 is estimated to be about 0.1 eV under the room temperature conditions of the ion preparation.

Figure 4

I– velocity images in center-of-mass frame as depicted by the above Newton diagram. (a–d) F– + CH3I reaction at 0.27(4), 1.06(8), 1.56(8), and 2.55(10) eV collision energies. (e–h) F–(H2O) + CH3I reaction at 0.32(5), 1.07(7), 1.55(7), and 2.57(11) eV collision energies. Circles indicate the kinematic cutoff and 1 eV steps in terms of higher ΔE that in the case of F– is identical to Eint. (h) Instead of the Newton rings, cuts to different mechanisms are marked in orange.

The F– + CH3I → I– + CH3F reaction has been studied before by our group[28] and is reinvestigated here to allow for a direct comparison with the hydrated reaction. At low energy it exhibits two distinct features in the product ion image (Figure a) that indicate different reaction mechanisms. A central and rather isotropic distribution indicates one or several indirect reaction mechanisms with intermediates that live long enough such that initial orientation becomes irrelevant. Energy is redistributed into internal degrees of freedom and peaks at maximum internal excitation. The second pronounced distribution is characterized by forward scattering of the product ion relative to the incoming neutral CH3I which is indicative for a direct mechanism. Product ion images at higher collision energies have been theoretically described by three distinct mechanisms and compared to time-sliced product ion images before.[28]

The improved resolution obtained in the present work now allows us to quantify the fractions of the different mechanisms experimentally. The upper bound fraction of indirect mechanisms is estimated as the isotropic area below the minimum (averaged over a small angular range) of the angular distributions (see Figure S3.1). The remaining area is split into the forward and backward hemisphere and attributed as a lower bound to the direct stripping and direct rebound fractions. In this way, the angular distribution in Figure a gives a 12:19:69 ratio for direct rebound, direct stripping, and indirect mechanisms at 0.27 eV. To obtain an improved estimate for the comparison with the computational results, we set an upper limit of 1.2 eV for the internal energy of indirect mechanisms, which yields an isotropic area of 52% relative to total reactivity. At 1.56 eV, the cuts marked with orange lines in Figure c are integrated and give 6% forward, 43% sideways, 41% backward, and 10% low energy isotropic scattering. In this case, the 10% gives a lower bound to indirect mechanisms.

Figure 5

Scattering angle distributions for nucleophilic substitution in reactions with CH3I. Legends specify reactant and product ions. (a) F–, F–(H2O), and Cl–(H2O) at 0.3 eV collision energy. The lower solid black line shows the difference of the I–(H2O) and I– distributions from F–(H2O) scattering. (b) F–(H2O) at different collision energies. The lower solid black line shows the difference of the 2.6 and 1.6 eV distributions. (c) Distributions for F–(H2O) scattering from quasiclassical trajectory simulations. Product ions are specified in the legend. The lower solid black line shows the difference of the I–(H2O) and I– distributions at 0.3 eV.

For F–(H2O) + CH3I reactions, only indirect mechanisms are evident in the images at collision energies up to 1.6 eV. At 2.6 eV, the visible appearance of a backward scattered distribution in Figure h indicates the increasing importance of a direct rebound mechanism at higher energies. This is reflected by the backward tendency of the difference histogram in Figure b. The angular distributions at the lower collision energies are very similar to each other. Corresponding scattering angle distributions from QCT simulations are presented in Figure c. They show a stronger intensity of small and large scattering angles and a more pronounced shift toward stronger backward scattering already at 1.0 and 1.5 eV collision energies. Except for the outermost bins, the close similarity of angular distributions of I– and I–(H2O) is captured by the QCT simulations (see Figure c). The experimental upper bound fraction of indirect mechanisms is again estimated as the isotropic area below the minimum of the angular distributions. Forward and backward contributions are separated near cos(θ) = 0.2 (see Figure S3.3) and percentages summarized in Table .

Table 1

Fractions of Mechanisms in the F–(H2O) → I– Channelsa

	energy (eV)
	0.3	1.1	1.6	2.6
direct rebound (%)	13 (9)	9	13	19
direct stripping (%)	4 (7)	4	4	2
indirect (%)	83 (84)	87	83	78

Fractions for I–(H2O) products are given in parentheses. Uncertainties of about 2% apply to the indirect fractions that are upper bound values.

Images for the solvated SN2 product I–(H2O) and the proton transfer product ion CH2I– are shown in Figure . At the lowest collision energy, an isotropic image of the solvated SN2 product I–(H2O) is observed; see Figure a. I– and I–(H2O) velocity distributions are mostly indistinguishable as is illustrated by the difference histogram in Figure a. The stronger forward tendency of the hydrated product gives a slightly larger estimate of the direct stripping fraction, as shown in Table .

Figure 6

Product ion velocity images in center-of-mass frame of (a) I–(H2O) from nucleophilic substitution and (b–d) CH2I– from proton transfer in the F–(H2O) + CH3I and F– + CH3I reactions at different collision energies. Reactants and collision energies are specified in the images.

The internal energy of the reaction products is quantified by the average total internal excitation, ⟨Eint⟩, divided by the total available energy, Erel + Eexo. For the two-body product channel of I–(H2O) + CH3F, this quantity can be experimentally determined and is given by ⟨ΔE⟩/(Erel + Eexo). One obtains a fraction of 0.76(4) at Erel = 0.32 eV. Average fractions and absolute internal energies at all collision energies are given in the Supporting Information. For the three-body product channel involving I–, which is expected to be the dominant channel,[17] one can also compute the fraction ⟨ΔE⟩/(Erel + Eexo) = 0.74(6). Here, it provides an upper bound to the true internal excitation fraction, due to the relative translational energy between the two neutral products Erelneutral (see eq ).

Solvation of the I– product brings an energy gain of 0.45 eV.[49] Inspecting the product relative kinetic energy Erel′ (extracted from the measured product velocity vectors) for unsolvated and solvated product ions, we obtain a shift of the mean value from 0.29 to 0.36 eV. Thus, most of the I–(H2O) solvation energy, at least 85%, is not partitioned into translational motion, but is retained in product internal excitation.

Proton Transfer Reactions

The proton transfer product anion CH2I– from reactions of hydrated F–(H2O) is not observed up to 1.6 eV and appears weakly at 2 eV collision energy. The image at 2.6 eV shows a single broad distribution with a tendency to forward scattering in Figure d. It is attributed to the proton transfer to F–(H2O) as is detailed in Supporting Information. The observed CH2I– branching is 3%, and no solvated CH2I–(H2O) could be detected. The F– + CH3I proton transfer reaction at 2.6 eV collision energy in Figure c features strong forward scattering near the maximum kinetic energy (see also ref (30)). It is less pronounced at 1.6 eV collision energy in Figure b, which actually resembles the 2.6 eV image from F–(H2O) in terms of angle and energy distribution.

Reactions of Cl–(H2O)

A product mass spectrum and I– velocity images for Cl–(H2O) + CH3I scattering at 0.3 eV collision energy are shown in Figure . The only products are I– from nucleophilic substitution and ClCH3I– from ligand exchange in a 54:46 ratio. I– formation gives rise to a single isotropic distribution without forward or backward flux as is seen for F–(H2O) in Figure a, which points toward indirect mechanisms. Interestingly, a ring-shaped structure in Figure b reveals a lower bound to the product kinetic energy of about 30 meV. A similar signature is observed in the backward hemisphere of the ClCH3I– image in Figure c, which contains a second broader distribution in the forward direction.

Figure 7

(a) Product mass spectrum and (b) I– and (c) ClCH3I– velocity images for the Cl–(H2O) + CH3I reaction at Erel = 0.31(4) eV.

Additional measurements of the Cl–(H2O) + CH3I reaction at 0.6 and 1.1 eV collision energies in the presence of O2–(H2O) show single isotropic I– distributions with increasing tendency to forward scattering. The presence of ClCH3I– at these higher collision energies proves that Cl–(H2O) contributes significantly to the observed reactions. However, we do not observe signatures of direct backward scattering. It can therefore be ruled out as a dominant mechanism in the Cl–(H2O) + CH3I reaction at the investigated energies.

Discussion

Product Solvation

Addition of a water molecule to the F– + CH3I reaction opens new reaction channels including solvation of the neutral or the ionic product, methanol formation following charge transfer in F–(H2O), and ligand exchange. The experiment resolves different product ions with 89(1)% I–, 4.5(9)% I–(H2O), and 6.2(7)% F–(CH3I) at 0.32 eV collision energy. Despite being energetically favored (see Figure ), I–(H2O) comes up for only 4.8(9)% of SN2 reactivity in excellent agreement with 4.4(13)% from direct dynamics simulations.[17] For comparison, in the isoelectronic reaction of OH–(H2O) with CH3I, only 2.5% of the SN2 products are solvated.[38] This may be attributed to the larger fraction of direct dynamics, which may not leave enough time for the water molecule to interact with the leaving group. In contrast, Cl–(H2O) + CH3I exclusively forms the unsolvated I– product and Cl–(CH3I) by ligand exchange in a 54:46 ratio. The large fraction of Cl–(CH3I) is indicative of a strong suppression of the SN2 pathway at the lowest collision energy. As solvent transfer in the gas phase SN2 reactions is inefficient and even more suppressed by early water loss at higher collision energies, the absence of solvated products in the Cl–(H2O) + CH3I reaction might be explained by the significantly lower solvation energy (0.64 eV), favoring early water loss, as opposed to F– (1.01 eV) and OH– (1.2 eV).[49]

A strong influence of the nucleophile has also been observed in the X–(H2O) + CH3Br reaction at thermal energies in the 200–500 K range. With X– = OH–, the product Br–(H2O) accounts for about 10% of the SN2 products at the studied temperatures,[50] and for 7–4% in a beam experiment probing collision energies from 0.3 to 1 eV.[11] With F–, up to 20% solvated product ions are observed at 200 K, but they are completely suppressed at 500 K.[13] With Cl–, the SN2 pathway is completely negligible and there is experimental evidence that Cl–(CH3Br) is formed in the Cl–(D2O) + CH3Br reaction.[12]

At higher collision energies, the F–(H2O) + CH3I → I–(H2O) + CH3F pathway is increasingly inhibited. This is in good agreement with the presented trajectory simulations and stems from the fact that higher collision energy leads to water loss at the initial collisional encounter before nucleophilic displacement takes place.[23] The slightly larger solvated fraction found in our trajectory simulations compared to the experiment may be caused by different upper bounds on the lifetime of metastable reaction complexes, which amounts to submicrosecond and picosecond time scales for experiment and trajectory simulations, respectively. Increasing the collision energy in the unsolvated reaction opens proton transfer and dihalide formation as competing pathways to SN2, e.g., in the F– + CH3I[30] and Cl– + CH3Br[51] reactions. The observed product mass spectra of the F–(H2O) + CH3I reaction are compatible with dihalide formation, but a clear assignment is precluded by the coreactants. Proton transfer could be attributed to the title reaction (see the Supporting Information) and is further discussed below.

Long-Lived Intermediates

The solvent-free reactions with F–, OH–, and Cl– involve different minima of the F–(CH3I), OH–(CH3I), or Cl–(CH3I) complexes[27,32,52] as reaction intermediates on a picosecond time scale.[53] A solvent molecule may leave the intermediate complex with a sufficient amount of kinetic energy to stabilize it and form very long-lived or stable ligand exchange products. Experimental evidence for the latter is absent in the product mass spectra of the monohydrated reactions with Cl– (Figure ) and OH– (ref (39)). However, for reactions with F–(H2O) it becomes significant in the measured 11(3)% fraction of I– products that are observed at apparently higher masses due to dissociation of a complex during acceleration. The fraction is noticeably suppressed by a 2 μs imaging delay. The observed range up to 144 u corresponds to maximum lifetimes of 3–4 μs.

The metastable complex leading to delayed I– product formation may either be the solvated product anion or the ligand exchange complex. The I– fraction from metastable complexes is larger than the I–(H2O) branching. In direct dynamics simulations only 12% of the I–(H2O) products have internal energies above the dissociation threshold at the end of trajectories.[17] This implies the alternative option of FCH3I– complex formation by ligand exchange with subsequent SN2 reaction and dissociation in line with the new QCT trajectories with FCH3I– complexes that are in parts still reactive at 20a0 separation from the neutral. The reaction may be trapped in the halogen-bonded FICH3–, hydrogen-bonded F–···HCH2I, or ion–dipole F–···CH3I prereaction complexes.[27] The latter undergoes Walden inversion with a transition state 1.29 eV above the I– + CH3F asymptote. Trapping in the postreaction complex I–···CH3F thus requires an unlikely late water loss to absorb most of this energy after the nucleophilic displacement. We speculate that metastable complexes are either susceptible to stimulated dissociation by moderate forces during acceleration (see Figure S3.3), or the likelihood of dissociation before 5 μs is too small to detect I– products in the 144–161 u range. Transient trapping in the water-free reaction intermediate in 70% of the indirect F–(H2O) + CH3I reactions at 1.53 eV[23] further supports that metastable complexes formed by ligand exchange are responsible for the delayed I– production.

Trapping in the halogen-bonded front-side complex plays a role in the solvent-free F– + CH3I reaction[26] and is a candidate for a ligand exchange intermediate. However, a high barrier impedes dissociation to I– via the front-side attack mechanism. In the OH– system, energetics moreover preclude trapping in a prereaction complex and, at higher collision energies, HO–···HCH2I is expected to pass the SN2 transition state quickly and dissociate.[16] This should be similar in the case of the ion–dipole complexes with Cl– and F–. For the reaction with F–(CH3I), we single out the hydrogen-bonded F–···HCH2I as the most likely long-lived intermediate that gives rise to the observed dissociation at microsecond time scales—in line with about 90% of the indirect mechanisms passing through this complex in direct dynamics simulations.[17] Also the front-side complex may be important, if I– shifts toward a hydrogen atom and SN2 occurs via the hydrogen-bonded complex.

Mechanisms

Improved resolution of the I– images from F– + CH3I scattering permits us to quantify direct rebound (DR), direct stripping (DS), and indirect mechanisms in a 12:19:69 ratio at 0.3 eV. Selecting a minimum internal excitation of 1.2 eV in this analysis gives a smaller fraction of 52% indirect reactions, in better agreement with chemical dynamics calculations. Direct dynamics simulations using DFT/B97-1 electronic structure theory obtained 15(2), 25(3) and 60(4)%[28] for DR, DS, and indirect mechanisms, while QCT simulations on a PES at the CCSD(T) level yield 17:23:60.[29] B97-1 results at 1.53 eV collision energy also found a large fraction of 59% indirect mechanism. However, this has been refuted by more recent calculations, which yield 46:43:11 using direct dynamics at the MP2 level[54] and 49:31:20 from the CCSD(T) level simulations at 1.53 eV collision energy.[29] Our experimental results at 1.56 eV give 41% DR, 43% sideways stripping, 6% forward, and 10% low energy isotropic scattering. This indirect fraction compares well with the two recent computational results, in particular when considering that the CCSD(T) value includes indirect events with high product ion velocities that have not been captured by the experimental analysis. Overall, this now confirms the transition from dominantly indirect reactions at low collision energies to two different direct mechanisms at higher collision energy, in contrast to earlier findings.[28]

In the angular distributions, the minima for the I– and I–(H2O) products in the F–(H2O) + CH3I reaction shift to the forward direction. This is in contrast to the sideways minima in the bare F– case. In recent simulations, DR extends to forward scattering angles near cos(θ) = 0.3 (see Figure S2 in the Supporting Information of ref (23)). This supports our procedure to separate forward and backward scattering at the minimum of the angular distributions. At the same time, it shows that DR and DS sum up to a non-negligible part of the isotropic area in the angular distribution such that the indirect fractions in Table are upper bound estimates.

The derived 13(2)% DR of I– products at 0.3 eV (see Table ) match well with the 14% from direct dynamics simulations.[17] The 4(2)% DS contribution is smaller than the direct dynamics value of 16%, which hints at an underestimation of DS due to imperfect separation in the angular distributions. In the I–(H2O) channel, the simulations predict 32% DR and the absence of DS. This should lead to stronger backward scattering as for the I– products, which is contradicted by the almost identical angular distributions of the two products. Below 1.6 eV the DR and DS fractions stay roughly constant. The transition from dominantly indirect to direct mechanisms that is found for the unsolvated reaction is therefore strongly suppressed by the solvent molecule. Toward 2.6 eV the derived DR contribution grows, but only from 13 to 19%, while the DS fraction remains small. This is explained by early water loss in high energy collisions.[23] However, this does not affect the DS fraction, which may have to do with different impact parameter ranges for the two mechanisms. Theoretical calculations seem to overestimate the decrease of the indirect fraction with 71 and 57% as opposed to constant 83% observed near both 0.3 and 1.6 eV collision energies.[23] This discrepancy may be caused by the ambiguity that occurs when the fraction of indirect reactions is estimated from product ion angular distributions in the three-body dissociation into I–, CH3F, and H2O. Future experiments using coincident detection of two products may overcome this.

The angular distributions from the presented trajectory simulations agree with the close similarity for I– and I–(H2O) at the lowest collision energy, but there are also differences. Due to larger forward and backward scattering intensities, the simulation results are less isotropic. The observed asymmetry with a minimum intensity at forward directions only appears at higher energies. Finally, the observed 2.6 eV tendency to stronger backward scattering is much stronger and is already seen at 1 eV.

There is no sign of the appearance of DR in the Cl–(H2O) + CH3I reaction between 0.3 and 1.1 eV collision energies. Direct dynamics simulations at 1.9 eV predict dominant DR and an important indirect roundabout mechanism.[24] Crossed beam imaging of the solvent-free Cl– + CH3I reaction showed a transition to dominant DR already near 0.5 eV collision energy.[31,36] DR is nearly the only pathway at 1.1 eV and is complemented by the roundabout mechanism at 1.9 eV.[31] Our measurements therefore confirm the expectation of efficiently suppressed direct mechanisms up to relatively high collision energies. Theory predicts internal energy and angular distributions similar to the unsolvated system due to early water loss in higher energy collisions, which demands further experiments for verification. At 0.3 eV collision energy, a ring-shaped I– product velocity image is reminiscent of the ClCH3I– image of the ligand exchange pathway. In the latter case, an additional broader distribution in the forward direction is observed and attributed to high impact parameters. We speculate that lower impact parameters could allow transfer of the collision energy into internal excitation and nucleophilic substitution could occur only indirectly via the ligand exchange intermediate.

Energy Partitioning

Besides the strong similarity of I– and I–(H2O) angular distributions in the F–(H2O) reaction at 0.32 eV, also the average fraction of product energy partitioned into internal excitation is found to be very similar for both ions, provided the relative energy between the neutral products CH3F and H2O is included in the case of I– formation (see Results). The fraction of product excitation differs by the fraction that is channeled into relative kinetic energy between the neutral products in the case of I– formation. At least 85% of the solvation energy (0.45 eV) gained by I–(H2O) formation is retained in product internal excitation. This suggests that CH3F departs quickly once H2O attaches to I– such that little of the solvation energy is channeled into kinetic energy. As I–(H2O) excitation must be smaller than the dissociation energy, we can attribute a minimum fraction of 0.44 to CH3F internal excitation in comparison to 0.67(2) in the unsolvated reaction. The latter is 10% smaller than the total internal energy fraction 0.74(6) in the solvated case, in agreement with a stronger contribution of direct mechanisms.

At 1.56 eV, the internal energy fraction of the solvent-free reaction drops to 0.58(2), which is not fully reproduced by theory with fractions of 0.69(2) and 0.66(1).[23] In the solvated reaction it stays almost constant with 0.74(6) and then drops to 0.69(3) at 2.6 eV due to the increased DR fraction. These fractions correspond to internal energies, including relative motion of the neutrals, of 0.70(2) and 1.62(2) eV at 0.32 and 1.55 eV collision energy, respectively. These values are comparable with the respective quantities of 0.77(1) and 1.71(4) eV, obtained from direct dynamics simulation.[23,55]

Proton Transfer

Without a barrier[22] and with 1.89 eV endothermicity, proton transfer is expected to open up at collision energies about 1.1 eV higher than in the F– + CH3I reaction with an endothermicity of 0.71 eV (see Figure ). This is confirmed by the observed appearance of a small fraction at 2 eV collision energy and above. The CH2I– product image at 2.6 eV, 0.7 eV above the predicted threshold, gives a similar dynamical fingerprint in terms of angle and energy distributions as the unsolvated proton transfer at 1.6 eV, about 0.9 eV above threshold, which corresponds to a similar excess energy as is indicated in Figure . This agrees well with very similar stationary point geometries of the unsolvated and solvated systems[22] along the proton transfer pathway. Due to the high collision energy, the fluorine anion and water molecule dissociate easily at the initial collision and then propagate separately on opposite sides of the methyl group. In this way, the additional collision energy is consumed by dissociation, and otherwise, the reaction occurs similar to the unsolvated one. The energetically favored solvated product ion has not been detected. Considering the noise level, it is suppressed by roughly a factor of not less than 5 relative to CH2I– formation.

Conclusions

Product branching ratios as well as angle and energy-differential cross sections have been obtained using crossed beam imaging and quasiclassical trajectory calculations. The F–(H2O) + CH3I reaction in comparison to F– + CH3I and the Cl–(H2O) + CH3I reaction were investigated at different collision energies starting at 0.3 eV. At the lowest collision energy, the F–(H2O) + CH3I reaction forms the SN2 products I– and I–(H2O) as well as the ligand switching product FCH3I– in a 89:5:6 branching. The suppression of the energetically favored product I–(H2O), which we have also seen for OH–(H2O), grows at higher collision energies, in very good agreement between experiment and simulation.

Indirect reaction dynamics, evidenced by isotropic scattering with large product internal excitations, dominate the reaction dynamics for F–(H2O) as well as for Cl–(H2O). This differs from the reactions of the unsolvated nucleophiles, where direct reaction mechanisms are more important. The quasiclassical trajectory simulations show overall good agreement with the measured angular distributions. They show a larger probability for forward and backward scattering than observed in the experiment, which still needs an explanation.

Further analysis of the QCT results will allow us to investigate the relative energy partitioning into the different translational and internal degrees of freedom of the reaction products. This will be particularly interesting for the nonsolvated product channels, which dominate the reactivity and lead to three different reaction products. The experiment can only detect the charged product, but together with the simulations the three-particle correlations and the amount of translational and internal energy partitioned into the two neutral molecules become accessible.

The present experiments have been hampered by coreactants that have complicated the analysis significantly. Future experiments are planned with an improved suppression of small fractions of nearby reactant masses. This will allow us to also study reactions with two or even more solvent molecules attached to the nucleophile. Furthermore, it will be interesting to compare the reaction dynamics for very different solvent molecules, such as carbon dioxide or acetonitrile.