Deuterium isotope effect on gas phase ion–molecule hydrogen-bonding interactions: multiply solvated fluoride, chloride, and alkoxide ions☆
Introduction
There has been a heightened interest very recently in so-called “low barrier hydrogen bonds” [1]. Such hydrogen bonds are said to be characterised by their significant bond strengths, short bond distances between the heteroatoms participating in the hydrogen bond, and low intermediate barriers for proton motion between these atoms [2]. The renewed interest in hydrogen bonds of this type is associated with the significant role that they may play in catalysis involving enzyme bound intermediates or transition states [3a], [3b]. The properties attributed to these low barrier hydrogen bonds are exactly those found for the class of gas phase entities known as proton bound dimers. These charged species, involving two bases, either anionic (I) or neutral (II), bound together by a proton can have hydrogen bond strengths [4] on the order of 30 kcal mol−1, heteroatom bond distances [5] as short as ∼2.4 Å, and may frequently have barriers to proton motion between the two heteroatoms [6] less than 1 kcal mol−1
Many studies of the energetics of solvation of gaseous ions by a single molecule of common solvent species have been carried out [7] that aid significantly in the understanding of the role that solvent effects might play in masking the intrinsic nature of ionic entities. Similarly, the determination of the thermochemistry of selected multiply solvated ions has led to an even greater insight into the nature of solvent effects [8].
Hiraoka and Mizuse [9] studied the gas phase clustering of Cl− with several alcohols, (ROH)n · Cl−, for n up to 11. They suggested, based on small but discernible entropy changes between n = 6 and n = 7, that for methanol and ethanol the sixth ligand completes the first solvation shell and ligands with n ≥ 7 belong to the second solvation shell. The bond energies were found to increase in the order H2O < CH3OH < C2H5OH < i-C3H7OH < t-C4H9OH < n-C3H7OH. The stronger bond observed for n-C3H7OH ⋯ Cl− was explained by the fact that both the acidic hydrogen atoms in the −OH and in the terminal −CH3 interact with Cl− with the most favourable configuration, as illustrated in Fig. 1. These “chelate” or “multiple site” interactions may also occur for the other alcohols, however, in these cases the C-H-Cl distance is greater, or the Cl− can only interact with one of the methyl hydrogens for structural reasons. With an increasing number of ligands, the geometry of the ligands interacting with Cl− may gradually change from the “chelate” mode to the “open” mode in which the Cl− ⋯ H-O is more linear in order to reduce the mutual repulsion between ligands. By so changing the mode of interactions, the entropy changes would become more favourable (less negative) owing to the increased degree of freedom of motion. Such chain length effects on chelation are consistent with recent data on the acidity of alcohols as a function of chain length and on the change in binding energies of Cl− to carboxylic acids as a function of chain length.
Recent measurements from this laboratory [10] have extended the studies of halide ion binding to alcohols, X−(ROH)n, to fluoride, bromide, and iodide adducts of methanol, ethanol, i-propanol, and t-butanol for n ≤ 3. Data obtained for the chloride adducts were in very good agreement with those of Hiraoka and Mizuse whereas the qualitative trends observed for chloride were, in general, reproduced for the adducts of the other three halide ions. As expected, the overall bond energies increased in the order I− < Br− < Cl− < F−.
In contrast to the relatively abundant data available for halide ion–alcohol clusters there is little data for alkoxide–alcohol adducts. The bond energy for methoxide methanol has been independently determined both in this laboratory and by Paul and Kebarle using high pressure mass spectrometry (HPMS) equilibrium measurements with excellent agreement between the three values obtained [11]. In addition, Caldwell et al. [12] have examined alcohol exchange equilibria using ion cyclotron resonance (ICR) techniques to obtain a set of relative free energies of binding in symmetric alkoxide–alcohol dimers. For example, the t-butoxide-t-butanol dimer is found to be less strongly bound in terms of free energy by 1.5 kcal mol−1 relative to methoxide methanol. No data are available for the clustering energetics of alkoxide ions with more than one molecule of alcohol.
In addition to these energetics determinations, rate constant measurements for ion–molecule reactions as a function of the number of solvent molecules associated with the ionic reagent also permit an insight into the subtle effects solvent might play in modifying chemical reactivity. Work by MacKay, Rakshit, and Bohme [13] measured the rate constants for the reactions of several acids with methoxide ion solvated with up to three molecules of methanol. These experiments provided insight into the perturbation of the intrinsic reactivity and basicity of the methoxide ions by stepwise methanol solvation. For some reagent acids, the addition of one molecule of methanol to the methoxide ion essentially stopped the reaction. For other acids, the addition of three methanol molecules did not reduce the rate by more than a factor of ten. For yet other acids, an intermediate behaviour was observed, in which the sharp drop in rate was delayed until the addition of two molecules of methanol to the methoxide ion. This divergence of observed rate constants can be accounted for by a consideration of the degree of stabilisation of the reactant and product ions by solvation, and their influence on overall reaction energetics. The rate of proton transfer can be expected to remain high upon solvation if the reaction remains exoergic. Conversely, a sharp drop in rate can be anticipated if solvation renders the reaction endoergic. The solvent-free reactions are all exoergic, and exoergicity is preserved if the free energy of solvation of the conjugate base produced is comparable to or greater than the corresponding free energy of solvation of CH3O−, or if a lower free energy of solvation of the product base is offset by the exoergicity of the unsolvated reaction.
Bowie et al. [14] observed that the rates of reaction of CH3O−(CH3OH)n with acetone decreased dramatically as the extent of solvation of the nucleophile increased. This result is not surprising based on the trends observed by Bohme [13].
In similar work, McIver, Scott, and Riveros demonstrated the effect of a single solvent molecule on the intrinsic relative acidities of methanol and ethanol [15]. They observed, by a comparison of the free energy changes for reactions (1) and (2), that the reversal of relative acidities is already substantial with the first molecule of solvation where ΔG0 = −1.9 and −1.2 kcal mol−1, respectively. Thus, the first molecule of methanol solvent produces a decrease of about 0.7 kcal mol−1 in the apparent relative acidity of methanol and ethanol. Caldwell et al. similarly obtain a value of −1.4 kcal mol−1 for ΔG20. This is consistent with the expected trend in going from the gas phase to the solution phase, but obviously more than one solvent molecule is required to reverse the relative acidity order.
Because a reversal in reactivity order occurs upon going from the nonsolvated ion to the bulk solution, there must be some point in the stepwise solvation where such a reversal occurs. MacKay and Bohme [16] investigated reactions of the form of Eq. (3), using flowing afterglow, where X− is OH−, Y− is CH3O−, and S is H2O or CH3OH
The magnitude of the measured equilibrium constant was found to decrease dramatically upon solvation by one molecule, and continue to drop, but at a decreased rate, for higher solvent molecule additions. This means that water becomes more acidic than methanol upon solvation, and more so for excess methanol than excess water. The results implied a higher solvent (H2O and CH3OH) affinity for OH− than for CH3O−, and more so for CH3OH than H2O. Apparently, solvation by more than approximately two methanol molecules actually results in a reversal of the relative acidity of water and methanol.
Thus, in general, in some cases, the first molecule of solvent makes the behaviour of the ion very solution-like, whereas in others, many solvent molecules are necessary for the reactivity and solvation thermochemistry to approach that of the bulk-solvated form.
Several other studies have demonstrated the effect of extent of solvation on the rate of reactions, such as the gas phase SN2 reaction [17a], [17b]. However, because the focus of the present study is on hydrogen bonding interactions, these systems will not be discussed further.
Baltzer and Bergman [18] determined values of the fractionation factor for the methoxide ion in solutions of methanol and dimethyl sulfoxide (DMSO). The fractionation factor, which is effectively the equilibrium constant defining the preference for deuterium to be found in the ionic species in solution as opposed to being present in the bulk solvent, was found to decrease from 0.74 in pure methanol to 0.38 in 75 mole percent DMSO. A rough extrapolation of this data to pure DMSO gives a fractionation factor in the neighbourhood of 0.3. This drop of fractionation factors seems to be a common type of behaviour for hydrogen-bonded complexes on transfer from hydroxylic solvents to dipolar aprotic ones. It was suggested that in the limit when the mole fraction of DMSO tends to unity, the methoxide ion exists as the monosolvate, and that the trisolvate is a correct representation of the true nature of the solvated methoxide ion in methanol. Baltzer and Bergman also suggested that the gradual desolvation of the ion leads to more symmetric hydrogen bonds for those methanol molecules that are still in a hydrogen-bonded position, possibly through an inductive effect. Thus, in DMSO, the adduct takes the form of a centrosymmetric ion, which changes to an increasingly asymmetric ion as the solvent hydrogen bonding to the anion weakens the hydrogen bond of the proton-bound dimer with increasing methanol content. Larson and McMahon [19] suggested that little additional change occurs upon change from aprotic media to the gas phase.
Recent contributions from this laboratory have discussed the deuterium isotope effect or fractionation factor in mono alcohol solvated fluoride [20] and chloride adducts, as well as symmetrical alkoxide-alcohol adducts [21]. These latter species have also been studied by Dixon and co-workers [22a], [22b]. In the present work the deuterium isotope effect for the interaction of the three anions (fluoride, chloride, and alkoxide) with up to three molecules of solvent alcohol are presented. It was possible in a few cases to compare the equilibrium constant measured from ions generated in the external source with that measured from ions created internally. For the systems CH3OH · F−, CH3OH · Cl−, CH3OH · OCH3−, and (CH3OH)2 · OCH3−, the equilibrium constants measured from ions created in both sources matched extremely well. Most of these higher cluster experiments used the multiple ion monitoring mode of operation to determine accurate intensities of the ions for the determination of the equilibrium constant. As we have demonstrated previously, these deuterium isotope effects, also called deuterium fractionation factors in solution, can provide a qualitative view of the nature of the potential in which the hydrogen bonding hydrogen finds itself. Thus, the transition from the isotope effect found for naked gas phase ions to that found in solution can be studied on a solvent molecule-by-molecule basis.
Section snippets
Experimental
All experiments were carried out on a Spectrospin CMS 47 Fourier transform ion cyclotron resonance (FTICR) spectrometer that has been modified by the addition of a high pressure external ion source. Multiple exchange reactions and nonreactive collisions in the FTICR cell ensure that ions are at thermal equilibrium with the neutral gas. Details of the design and operation of the basic FTICR apparatus and our modification have been described extensively elsewhere [23], [24].
The external high
(CH3OH)n · F− clusters
A typical spectrum showing the distribution of clusters obtained from the external source for F− and methanol is shown in Fig. 3. Peaks due to the small amount of deuterium exchange that has already occurred during the ionization pulse and detection cycle are also evident. The (CH3OH)4 · F− cluster ion decreases in intensity very rapidly after it is trapped in the FTICR cell and thus no equilibrium experiments could be performed with this species. Similarly, the (CH3OH)3 · F− ion was found to
Conclusions
It has been clearly demonstrated that the value of the deuterium isotope effect differs significantly for the fluoride ion, chloride ion, and alkoxide ion adducts of aliphatic alcohols. These differences can be directly linked to the hydrogen bond strength in the adduct ions, and the capability of the chloride ion to undergo multiple site interactions.
The hydrogen bond strengths of alkyl alcohols to fluoride and chloride have been shown to differ significantly. This observation is found to be
References (30)
- et al.
Int. J. Mass Spectrom.
(1998) - et al.
Chem. Phys.
(1987) - et al.
Int. J. Mass Spectrom.
(1999) - et al.
J. Am. Chem. Soc.
(1978) - et al.
Int. J. Mass Spectrom. Ion Processes
(1992) - et al.
Int. J. Mass Spectrom. Ion Processes
(1998) - et al.
Int. J. Mass Spectrom. Ion Phys.
(1983) - et al.
Int. J. Mass Spectrom. Ion Processes
(1990) - et al.
J. Am. Chem. Soc.
(1997) - et al.
J. Am. Chem. Soc.
(1994)
J. Am. Chem. Soc.
J. Am. Chem. Soc.
J. Am. Chem. Soc.
Cited by (0)
- ☆
Dedicated to Henri Edouard Audier on the occasion of his 60th birthday.