Stereoelectronic Interactions as a Probe for the Existence of the

Jul 12, 2017 - Vera A. Vil' , Gabriel dos Passos Gomes , Oleg V. Bityukov , Konstantin A. Lyssenko , Gennady I. Nikishin , Igor V. Alabugin , Alexande...
2 downloads 0 Views 4MB Size
Article pubs.acs.org/JACS

Stereoelectronic Interactions as a Probe for the Existence of the Intramolecular α‑Effect Eusebio Juaristi,*,†,‡ Gabriel dos Passos Gomes,§ Alexander O. Terent’ev,⊥ Rafael Notario,*,∥ and Igor V. Alabugin*,§ †

Departamento de Química, Centro de Investigación y de Estudios Avanzados, Av. Instituto Politécnico Nacional 2508, 07360 Ciudad de México, Mexico ‡ El Colegio Nacional, Luis González Obregón No. 23, Centro Histórico, 06020 Ciudad de México, Mexico § Department of Chemistry and Biochemistry, Florida State University, Tallahassee, Florida 32306, United States ⊥ N. D. Zelinsky Institute of Organic Chemistry, Russian Academy of Sciences, 47 Leninsky Prosp., 119991 Moscow, Russian Federation ∥ Instituto de Química Física “Rocasolano”, CSIC, c/Serrano 119, 28006 Madrid, Spain S Supporting Information *

ABSTRACT: The first systematic study of the intramolecular α-effect, both in the stable ground-state structures and in the high-energy intermediates, was accomplished using the anomeric effect as an internal stereoelectronic probe. Contrary to the expectations based on the simple orbital mixing model, the lone pairs in a pair of neutral directly connected heteroatoms are not raised in energy to become stronger donors toward adjacent σ- and π-acceptors. Instead, the key n(X‑Y)→σ*C‑F interactions (X,Y = O,N) in the “α-systems” (both acyclic and constrained within a heterocyclohexane frame) are weaker than nX→σ*C‑F interactions in “normal” systems. Surprisingly, polar solvent effects increase the apparent magnitude of α-effect as measured via increase in the anomeric stabilization. This behavior is opposite to the solvent dependence of normal systems where the anomeric effect is severely weakened by polar solvents. This contrasting behavior reflects the different balance of electrostatic and conjugative interactions in the two types of anomeric systems: the αsystems suffer less from the unfavorable orientation of bond dipoles in the equatorial conformer, a destabilizing electrostatic effect that is shielded by the polar environments. A weak α-effect is brought to life when the buttressing α-heteroatom bears a negative charge. However, electrostatic components mask the role of stabilizing orbital interactions. In contrast, the increased electron demand in carbocations and related electron-deficient TS- like structures does not lead to activation of the α-effect. As a consequence, we observed that ethers are better radical- and cation-stabilizing groups than peroxides. The latter finding should have significant implications for understanding the mechanistic complexity associated with the interaction of carbonyl compounds with hydroperoxides and H2O2 in acidic media, as such reactions involve α-cationic intermediates.



enhanced reactivity of HOO− toward methyl halides and methyl formate in the gas phase, relative to the normal nucleophiles.5 Although naked hydroxide anion is an anomalously strong base in the gas phase, it is only slightly more reactive than hydroperoxide anion in the SN2 reaction with methyl fluoride. When the large difference in the thermodynamic stability of anions is removed (e.g., HOO− is compared with AlkylO−), the less basic α-nucleophile reacts faster (Scheme 1). Subsequent experiments in the gas phase extended these studies to micro solvation conditions where reactivity of anionic nucleophiles is moderated in comparison to that of the bare anions.6 This work also clearly demonstrated

INTRODUCTION Based on experimental kinetics of nucleophilic reactions by Jencks and Carriuolo1 (Scheme 1), Edwards and Pearson2 suggested a model for the increased reactivity of nucleophilic sites with a lone pair at an adjacent atom and coined the term “α-effect” to describe this phenomenon. Later, a narrower definition was suggested as “a positive deviation from Brønstedtype nucleophilicity plots in comparison with a reference nucleophile of the same basicity”.3,4 However, such deviations (kα‑Nu/knormal‑Nu) vary significantly in magnitude (in the range of 5−1000) for different reactions and sometimes even puzzlingly disappear. Although earlier reports of the importance of the α-effect in the gas phase reactivity were inconclusive, recent work of Bierbaum and co-workers elegantly and convincingly confirmed © 2017 American Chemical Society

Received: May 24, 2017 Published: July 12, 2017 10799

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society Scheme 1. Selected Examples of Different Kinetic and Thermodynamic Parameters of Reactions of Normal and αNucleophilesa

Scheme 2. Comparison of the Well-Known Effects of αHeteroatoms in Cationic and Cation-Forming Processes with the α-Effect in Nucleophilic Substitutiona

a

a

Top panel shows the greater nucleophilic reactivity of less basic anionic species. Bottom panel shows that the unusually high gas-phase basicity of hydroxide anion leads to only a slight increase in reactivity relative to the much less basic hydroperoxide anion.

(a) Thermodynamic stabilization of cations. (b) Assistance to formation of cations at the anomeric positions. (c) The α-effect in nucleophilic substitution and its possible rationalization through the involvement of single electron transfer (SET) in the transition state.

“normal” analogues. For example, HOO− is less basic than HO−,11 and PhNHNH2 (pKa = 28.8)12 is more acidic than PhNH2 (pKa = 30.6),13 indicating that a negatively charged center is stabilized next to oxygen and nitrogen. In addition, many heteroatomic functionalities can display chameleonic14 stereoelectronic features that allow them to switch from being a donor to being an acceptor at certain geometries (Scheme 3).

the modulation of α-effect for the α-nucleophiles upon coordination with just a few solvent molecules. Interestingly, association of the anions with a single solvent molecule uncovered reactivity trends that were difficult to see in the reactions of the unsolvated anions. Increased contribution of nucleophilic substitution relative to proton transfer supported greater nucleophilicity of HOO− relative to that of CH3O− in reactions with dimethyl methylphosphonate.7 The controversies and differences of opinion regarding the magnitude and origin of the α-effect are not surprising since this phenomenon involves the interplay of kinetic and thermodynamic factors as well as the intricate effects of solvation.8 To complicate matters even further, the list of αnucleophiles sometimes is expanded to large multifunctional molecules that can stabilize the transition states (TSs) by interactions between remote parts of the reactants. Such interactions may be interesting but have little to do with the key premise of the α-effect: the lone-pair effect at the adjacent nucleophilic center. There are several ways for this effect to operate. Considering the well-known stabilizing effects of lone pairs at the adjacent cations, Edwards and Pearson surmised in their 1962 paper that similar delocalizing assistance can be beneficial in alleviating the loss of electron density at the nucleophilic center as the latter attacks the electron-deficient position of an electrophile (Scheme 2). Although the “electron demand” created by a decrease in electron density at a negatively charged nucleophilic center should be much smaller than the electron withdrawal from the positively charged centers, this concern can be alleviated by the involvement of single electron transfer (SET) in the TS9 radicals are well-known to be stabilized with the formation of two-center, three-electron (2c,3e) bonds with the adjacent lone pairs (Scheme 2). However, as we will discuss in detail later, the donor ability of the lone pairs can be offset by other factors, such as inductive and hybridization10 effects. In fact, αnucleophiles have lower basicity in comparison to their

Scheme 3. Chameleonic Properties of RX Groups with a Lone Pair at an Electronegative Heteroatom X

Even though mixing of filled orbitals corresponding to the two lone pairs can produce a higher energy antibonding combination with a greater donor ability,15 such mixing and concomitant “reactant destabilization” are often irrelevant for the ground state of flexible molecules.16 The conformationally unrestricted reactants can avoid ground-state destabilization by adopting a geometry where the lone pairs are misaligned with each other (Scheme 4). Gauche geometries of hydrogen peroxide and hydrazine illustrate this geometric preference that corresponds to a stereoelectronically stabilized arrangement where the donor lone pairs are aligned with σ acceptor orbitals. Because orbital alignment usually associated with the α-effect disrupts this favorable arrangement, such intrinsic penalties can mask stereoelectronic TS stabilization. For example, when analyzing the relative reactivity of NH3/ N2H4/NH2OH and substituted analogues, Mayr and coworkers found acceleration in reactions with benzhydrylium ions and quinone methides due to the change of H to NH2 (NH3 → N2H4) is significant (∼100-fold) but not as large as acceleration by a Me group (NH3 → CH3NH2: >200-fold). Acceleration by an OH group in hydroxylamines (factor of 10800

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society

focuses on the first-order effects of the α-lone pair on both kinetics and thermodynamics instead of the second-order effects such as the deviations from the Brønsted plots. We will illustrate how basic stereoelectronic considerations can explain why both basicity and nucleophilicity are significantly different in the HOO−/HO−, t-BuOO−/t-BuO−, H2O2/H2O, and NH3/ N2H4 pairs. In the starting materials, the gauche effect18 is utilized. The best donors (lone pairs of X and Y) align antiperiplanar to the best acceptors (σ*Y‑R′ and σ*X‑R orbitals, respectively, Scheme 5). In the TS for nucleophilic attack of X on E+, strong anomeric effects (nY→σ‡*X‑E) develop between the high-energy lone pair and the low-energy antibonding orbital associated with the incipient bond. In the product, the lone pair of Y has the choice to align itself with either the X−E or X−R bonds. If the X−Y reactant was neutral, X is positively charged at this stage and both of these σ*-orbitals are strong acceptors. Depending on the acceptor ability of the X−E bond, a thermodynamic α-effect may be present on the stability of reaction (e.g., SN2) products. The evolution of (nY→σ*X‑H/ nY→σ ‡*X‑H/nY→σ‡*X+‑H) interactions along the protonation path describes additional stereoelectronic activation of this process that can potentially increase the basicity of such compounds. The contrasting findings from computational analysis further illustrate that if stereoelectronic complexity of peroxides is unrecognized, correct modeling of α-effect is difficult. A key example is provided by the re-evaluation of the gas-phase SN2 barriers19 at saturated carbon by Ren and Yamataka,20 who found that TS energy for the gas-phase SN2 reaction of CH3Cl with HOO− at the G2(+) level depends on the orientation of the α-OH group in the nucleophile (Scheme 6). The

Scheme 4. Preferred Gauche Conformation of Hydrazine, Controlled by Generalized Anomeric Effect with Benefits from Stabilizing nN→σ*NH Interactionsa

a Activation of α-effect in this molecule (e.g., in the TS for its nucleophilic substitution) would require a penalty associated with loss of these effects in the less stable anti-conformation.

∼10) is also significantly smaller than the effect of a methyl group (Scheme 4).17 A refined stereoelectronic model of the α-effect is based on the evolution of anomeric interactions in this system along the reaction path.18 It is illustrated below for the simplest case with one lone pair on each atom, X and Y (Scheme 5). This model Scheme 5. (A) α-Effect That Results When a Relatively Weak Stereoelectronic Interaction in the Reactant Is Transformed to a Stronger Anomeric Effect during the Reactiona and (B) Curtin−Hammett Analysis for the Situation in Which the More Stable Ground-State Conformer Does Not Have αEffect Activatedb

Scheme 6. Two Transition States in the Reaction of HOO− + CH3Cla

In the left TS, the properly aligned lone pair contributes to the ∼2 kcal/mol decrease in energy.

a

computations suggested that when the lone pairs of the two oxygen atoms are aligned properly HOO− becomes more reactive than HO−, even though HO− is a much stronger base than HOO−. In the lower energy TS, the lone pairs of the two heteroatoms are aligned as expected from the classic molecular orbital (MO) model for the α-effect. An alternative TS where the OH group is rotated in a way that misaligns the forming O···C bond with the p-type lone pair at the α-oxygen is 2.1 kcal/mol less stable. As often observed in bimolecular gas phase reactions, both of the TSs were well below the energy of reactants and, thus, selectivity of such reactions is expected to be relatively low. Indeed, Bierbaum and co-workers found no strong deviations from the correlation with Brønsted basicity in the reactions of normal and α-nucleophiles with methyl chloride.5a Notwithstanding the above analysis, the fundamental question at the heart of α-effect is whether the lone pairs of

a

Kinetic effects correlate with nucleophilicity, while thermodynamic effects correlate with basicity. The “excessive” α-effects associated with the positive deviations from the Brønsted correlation result from stronger interactions in the TS than in the product. bThis explains how the α-effect may remain kinetically unimportant. For TSa′, TS stabilization does not compensate for the penalty associated with reaching the geometry where the α-effect is activated. 10801

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society two directly connected heteroatoms can combine into a more powerful donor than each of the lone pairs taken separately in a TS for a reaction of interest. Simple MO arguments suggest that mixing of the two lone pairs should lead to formation of a higher energy antibonding MO (Scheme 7). However, one

Scheme 8. (A) Common Situations and Their Preferred Geometries for Intermolecular α-Effect and (B) Scope of Substrates for the Computational Analysis Reported in This Work

Scheme 7. (A) Simplified MO Description of α-Effect and (B) Effect of Electronegativitya

a

Formation of an X−X bond can lower the lone-pair energy and compensate for the expected increase in the donor ability of stereoelectronically coupled lone pairs.

the inherent complexity of this question by providing a way to align or misalign the adjacent lone pairs as a function of structural constraints. We will show below that the answer to this effect of an adjacent lone pair on the donor ability of another lone pair depends on several factors that include the electronegativity effects on hybridization and energy of lone pairs. Furthermore, the situation depends on the number of stereoelectronically active lone pairs present at the α-heteroatom that determines whether the switching of dominant stereoelectronic effects is necessary upon the transition from the most stable ground-state conformation to the TS geometry. These factors are strikingly different for nitrogen and oxygenthe two most common reactive centers in α-nucleophiles.

must keep in mind that the overall MO energy increase can be counterbalanced by the inductive and hybridization effects of the acceptor neighbor. According to Bent’s rule,21 presence of electronegative group Y at atom X, increases the allocation of pcharacter of X in the X−Y orbital, thus making the lone pair of X have more s-character and be less donating.22 Furthermore, such molecules may have to endure an additional penalty for using both lone pairs when, in order to do so, they have to abandon their preferred ground-state conformations. The latter requirement is absolute when the αatom has only one lone pair (e.g., nitrogen). It is less restrictive for α-oxygen nucleophiles that have two lone pairs or when additional lone pairs are created at the nucleophilic center by deprotonation (e.g., for anionic N- or O-nucleophiles). Additionally, the role of the α-effect was suggested to be quite different for reactions at saturated and unsaturated centers suggesting that there may be a connection with the preferred trajectories for the nucleophilic attack (Scheme 8). Furthermore, secondary interactions between parts of the reagents that are not directly involved in bond-breaking/bond-making events can significantly contribute to the stability of bimolecular TS, especially in the gas phase. The importance of remote interactions can vary but it will always remain an additional factor in the traditional α-effect type, the case of intermolecular interactions. Such interactions can be prevented by pre-organizing the donor and acceptor in well-defined mutual arrangement, typical of intramolecular settings. The possible intramolecular consequences of α-effect are still unknown, even though the intramolecular settings are particularly well-suited for probing the key premises of α-effect, e.g., whether the combination of two lone pairs is actually a better donor than a single lone pair. The present work is designed to test, for the first time, the key premise of α-effect by using an intramolecular probe (Scheme 8). More specifically, it takes advantage of one of the well-established stereoelectronic effects (i.e., the anomeric effect)23 as a tool for probing the energetic and structural consequences of donor−acceptor interactions. The intramolecular setting in well-defined rigid cyclic structures reduces



RESULTS AND DISCUSSION Contrasting Effects of α-Heteroatoms on the LonePair Energies. As a starting point for the following discussion, it is helpful to analyze the effect of α-heteroatoms on the lonepair energies. For this purpose, we have combined NBO analysis with the examination of highest occupied MOs (HOMOs) in ammonia, water and two conformers of hydrazine and hydrogen peroxide. Natural Bond Orbital (NBO) analysis provided energies and hybridizations for the individual lone pairs, whereas the canonical MO analysis described the energetic consequences of the lone pairs’ mixing with each other and/or other parts of the molecules. Immediately, interesting differences between oxygen and nitrogen are obvious as a consequence of a shifting balance in the interplay between hybridization and electronegativity effects (Scheme 9). The addition of an α-heteroatom decreases the lone-pair energy in the most stable conformers of N2H4 and H2O2 relative to the lone pairs of NH3 and H2O, respectively. The observed energy lowering is greater for the introduction of oxygen, a more electronegative neighbor. When rotation around the N−N and O−O bond aligns the lone pairs at the two adjacent heteroatoms, an additional large effect on the lone-pair energy in hydrazine is observed. The orbital 10802

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society

Scheme 9. (A) Hybridization for RNH2 and ROH Species, (B) Energies of NBOs Corresponding to Individual Lone Pairs of Oand N-Lone Pairs,a and (C) Anti-symmetric Lone-Pair Combination MOs in the Representative Conformations of Hydrazine and Hydrogen Peroxide

a Values shown in gray for the O-containing molecules correspond to the high-energy p-lone pair; values in black describe σ-type spn hybrid lone pairs.

calculating the thermodynamic consequences of separating the interacting functional groups via isodesmic equations. Finally, the calculated substituent effects were complemented with NBO analysis (a method of choice for quantifying stereoelectronic interactions).26 Conformational Analysis. The conformational behavior of fluorocyclohexane (1-ax/1-eq), 2-fluoro-oxane (2-ax/2-eq), 2fluoro-1,3-dioxane (3-ax/3-eq), 3-fluoro-1,2-dioxane (4-ax/4eq), 3-fluoro-1,2,4-trioxane (5-ax/5-eq), and 3-fluoro-1,2,4,5tetraoxane (6-ax/6-eq) is summarized in Table 1. The structural constraints of the six-membered rings assured that the minimum energy structures for the axial and equatorial conformers of compounds 1-6 adopted similar chair conformations. In particular, the similarity in the calculated dihedral angles F−C(1)-C(2)−C(3) = 66.2° in 1-ax, F− C(2)−O−C(6) = 65.4° in 2-ax, F−C(2)−C(3)−C(4) = 68.7° in 2-ax, F−C(2)−O−C(6) = 66.4° in 3-ax, F−C(3)−O−O = 62.9° in 4-ax, F−C(3)−O−O = 62.2° in 5-ax; F−C(3)−O− C(5) = 69.4° in 5-ax, and F−C(3)−O−O = 63.6° in 6-ax supports the working hypothesis that the overlap of the key orbitals involved in the potential stereoelectronic interactions is similar in these heterocycles. Notable observations are the following: (1) The calculated gas-phase free energy difference for the axial-to-equatorial equilibrium in fluorocyclohexane, 1-ax ⇌ 1eq, ΔG = −0.2 kcal/mol is close to the experimentally measured value of −0.25 kcal/mol.27 The equilibrium is little

alignment effect on the lone-pair energies in H2O2 is much smaller. We attribute the much larger effect in the nitrogen case to the greater rehybridization,24 where increases in pyramidalization and s-character in the lone pair are coupled to each other. For oxygen, rehybridization at the s-type lone pair and the associated energy decrease are much smaller. The p-type lone pair retains its original 100% p-character and changes its energy only slightly and in the opposite direction. When the localized nonbonding orbitals at the heteroatoms mix to form the MOs of N2H4 and H2O2, the resulting HOMOs are raised. For the most stable conformation of hydrazine, the HOMO is raised only moderately relative to that of NH3, presumably because the hydrazine HOMO is stabilized by hyperconjugative mixing with the σ*NH of the adjacent NH2 moiety. On the other hand, the HOMO of the most stable H2O2 conformation was much higher than the HOMO of H2O. This increase was due to the presence of the two lone pairs at the oxygen atoms that makes mixing of the nonbonding orbitals unavoidable. When the nitrogen lone pairs were aligned in the eclipsed conformations of hydrazine and H2O2, the antisymmetric combination of the two lone pairs was significantly destabilized, as expected from the ground-state destabilization model of the α-effect. Two Oxygen Centers. In our search for the intramolecular thermodynamic α-effect, the conformational analysis for the fluorine substituent in several oxygen-containing six-membered rings was studied.25 We have further extended this analysis by 10803

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society

Table 1. Calculated Differences in ΔE and ΔG for the Axial/Equatorial Conformational Equilibria in Compounds 1−6a

a

Calculations, at the MP2/6-311+G(d,p) level of theory, in kcal/mol.

presence of two O−C−F anomeric segments in 2-fluoro-1,3dioxane 3 and only one such segment in 2-fluoro-oxane 2. Again, the magnitude of the anomeric effect (axial predominance) decreases in going from the gas phase (ΔG = +3.5 kcal/mol) to cyclohexane (ΔG = +2.9 kcal/mol) and DMSO (ΔG = +2.0 kcal/mol). This observation suggests an evergreater guarding of the dipole−dipole repulsive interaction, present in 3-eq, by solvents of increasing polarity.18,25,28 (4) The most relevant observation is made in the gas phase conformational equilibrium of peroxide-containing 3-fluoro-1,2dioxane 4. The conformational free energy difference of +2.0 kcal/mol indicates a predominance of the axial conformer4-ax as expected from operation of n(O‑O)→σ*(C‑F)ax stereoelectronic interaction. However, this axial preference is less than that found in 2-fluoro-oxane 2, ΔG = +2.5 kcal/mol. This observation opposes the expectation in terms of the α-ef fect, which would suggest an increased donor ability of the peroxide [n(O‑O)→σ*(C‑F)ax] relative to an ether analogue [n(O)→σ*(C‑F)ax]. (5) Polar solvents, such as DMSO, increase the anomeric preference in peroxides but decrease the anomeric preference in ethers. This unusual feature of the peroxide system leads to the opposite responses of the cyclic ether 2 and peroxide 4 to the change in solvent polarity. In addition it significantly increases the stabilization of the axial peroxide conformer. The switch in the relative magnitude of solvent effect suggests the possible appearance of the α-effect in polar solvents. Considerable solvent effects were reported for the classic intermolecular αeffect in the literature.8 This finding can be attributed to the

affected by the presence of a solvent, as it can be anticipated for a monosubstituted cyclohexane.28 (2) The calculated gas-phase free energy difference for the axial to equatorial equilibrium in 2-fluoro-oxane, 2-ax ⇌ 2-eq, ΔG = +2.5 kcal/mol indicates a large predominance of the axial conformer. This preference is in line with the classic anomeric effect and importance of the n(O)→σ*(C−F)ax stereoelectronic interaction (Scheme 10, eq 1). The decrease of axial preference Scheme 10. Stereoelectronic (Orbital) and Electrostatic (Dipole) Components of the Anomeric Effect

for 2-ax over 2-eq in solution, especially in DMSO, is in line with alleviating the destabilizing effect of repulsive dipole− dipole interactions by the polar solvent (Scheme 10, eq 2).18,25,28 (3) The gas-phase predominance of the axial conformer in 2fluoro-1,3-dioxane 3 in conformational equilibrium is even larger, ΔG = +3.5 kcal/mol. This is in line with expectation the 10804

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society

Scheme 11. NBO Charges for Monofluorinated Systems with Zero, One, and Two Oxygen Atoms in a Six-Membered Saturated Cycle, and Estimated Bond Dipoles in the Axial and Equatorial Conformers

Interaction Energies and Donor/Acceptor Orbital Gaps for n(O)→σ*(C‑F) Interactions. An attractive feature of NBO analysis is that it can evaluate the interaction energies for any given pair of donor and acceptor orbitals. In the NBO formalism, the delocalizing interactions can be readily evaluated either by deletion of the corresponding off-diagonal elements of the NBO Fock matrix and recalculation of the energy (referred to as Edel energies) or by the second-order perturbation approach where ⟨σ|F|σ*⟩, or Fij, is the Fock matrix element between the orbitals (NBOs) i and j, εσ and εσ* are the energies of the σ and σ* NBOs, and nσ is the population of the donor orbital (eq 3).29−31

drastically different bond dipoles for the C−O and O−O bonds at the anomeric center. (6) The predominance of axial conformer for peroxyacetal 5 (ΔG = +2.3 kcal/mol) in the gas phase is significantly lower than the axial predominance for acetal 3 (ΔG = +3.5 kcal/mol). As discussed above, this finding contrasts expectations in terms of the increased donor ability of the peroxide segment relative to the ether segment, owing to the α-effect. In DMSO, this difference is the same for the two compounds (ΔG = +2.0 kcal/ mol). (7) Finally, even though the axial 3-fluoro-1,2,4,5-tetraoxane 6-ax is more stable than equatorial 6-eq as anticipated in terms of the anomeric effect, this preference, in the gas phase, is significantly lower (ΔG = +1.2 kcal/mol) than the axial preference in 2-fluoro-1,3-dioxane 3 (ΔG = +2.3 kcal/mol). However, upon transition from the gas phase to a polar solvent (DMSO), the relative magnitudes of the axial preference for the bis-peroxide 6 and acetal 3 are, again switched (as they were for the 2/4 and 3/5 systems discussed above). It is clear the donor ability of the peroxide segments is smaller than the donor ability of the ether segments in the gas phase, against expectations in terms of the α-effect. However, the effects of the solvent leave room for different interpretations of this observation. In particular, the opposite trend in the solvent effects in ethers and peroxides suggests that the balance of orbital and dipole contributions to the anomeric effect can be quite different in these two functionalities. Inspection of the NBO charges suggests different dipole/dipole interaction patterns for the polar bonds at the anomeric center. In fact, due to the low polarity of the O−O bonds, the interaction of bond dipoles in peroxides should be similar to that in cyclohexanes (Scheme 11). Considering the above, we have investigated orbital and dipole contributions for systems in this study using NBO analysis outlined in the following sections.

E(2) = −nσ

Fi , j 2 σ |F |σ * 2 = −nσ ΔEij ε σ * − εσ

(3)

Table 2 collects the E(2) interaction energies and the corresponding energy gaps (ΔEij) for the main hyperconjugative interactions in 2-6-ax and 2-6-eq in the gas phase, cyclohexane and DMSO. Note that in every case, the magnitude of nO→σ*C‑F interactions increases in the more polar environment and that this slight increase parallels a decrease in the energy gap between the interacting donor and acceptor NBOs. This is a clear indication that the opposite solvent effects on the conformational preferences of cyclic ethers and peroxides discussed in the previous sections do not stem from the changes in the main hyperconjugative interactions. Importantly, the NBO energies evaluate donor ability of anomeric oxygens toward the same σ*(C‑F) acceptor in a variety of oxygen heterocycles, both ethers and peroxides (Scheme 12). Most relevant is the observation that the O−C−Fax interaction in 2-ax [E(2) = 22.8 kcal/mol] is greater than the O−O−C−Fax interaction in 4-ax [E(2) = 19.2 kcal/mol]. This is contrary to expectation in terms of the α-effect, which dictates 10805

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society Table 2. nO→σ* ax and 2−6-eqa

C‑F

Scheme 12. Comparison of NBO nO→σ* C‑F Interactions (in kcal/mol) in Fluorotetrahydropyran 2-ax and the Respective Peroxide 4-ax

Hyperconjugative Interactions in 2− 6-

Comparison of the interaction energies for systems with two sets of donor lone pairs directly interacting the C−F acceptor; i.e. the axial 2-fluoro-1,3-dioxane 3-ax and 3-fluoro-1,2,4trioxane 5-ax show the same trend [E(2) = 20.6 vs 19.9 kcal/ mol, respectively]. Introduction of the second peroxide moiety in the axial 3-fluoro-1,2,4,5-tetraoxane 6-ax decreases the energy of the key interaction to 15.6 kcal/mol. Together, these NBO data make evident the lower donor ability of the peroxide segments O−O−C−Fax relative to the ether segments O−C−Fax. The magnitude of changes in the NBO energies qualitatively correlates with the differences between the axial and equatorial conformers from Table 1. However, the conformational energies do not give an accurate comparison of the anomeric effect in peroxides relative to that in ethers because the conformational energies use different reference points (i.e., the peroxide is compared to a peroxide, whereas the ether is compared to an ether). In order to have a more balanced analysis, one need to compare ethers and peroxides directly using the isodesmic approach outlined in the following section. Group Separation Equations. Because reactivity and stability are molecular properties that depend on a variety of factors of which orbital interactions is just a single component, we sought additional insight into the possible manifestations of the α-effect from the classic isodesmic approach that derives accurate thermochemical analysis from error-balancing reactions. These reactions can be constructed to provide information on the magnitude of electronic effects, when their experimental determination is not simple.32 A simple way to evaluate interaction between two functional groups is to construct a “separation” reaction which compares a molecule with the two groups adjacent with the suitable molecules that have the two groups separated. Scheme 13 illustrates the utility of such approach for evaluating the anomeric stabilization (ΔE) in axial 2-fluoro-tetrahydropyran (2-F-THP) (Scheme 13a) and 3-fluoro-1,2-dioxane (Scheme Scheme 13. “Separation” Isodesmic Reactions Allowing the Comparison of n(O)→σ*(C‑F)ax and n(O‑O)→σ*(C‑F)ax Hyperconjugation and NBO E(2) Energies for n(O)→σ* a (C‑F)ax and n(O‑O)→σ* (C‑F)ax Interactions

a

All calculations were carried out at the MP2/6-311+G(d,p) level of theory.

that lone pair−lone pair electron repulsion in the peroxide should increase its donor ability relative to an ether analogue. However, it agrees well with the lower energies of the localized lone pairs in Scheme 9. The overall decrease in the interaction energy comes from the two components: (a) lower energy of the donor oxygen lone pair that increases its energy separation ΔE with the σ*C‑F and (b) smaller Fock matrix (resonance) element between the donor and acceptor orbitals (Scheme 12).

a

10806

Calculations, at the MP2/6-311+G(d,p) level of theory, in kcal mol. DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society 13b). As indicated by the first equation, the axial C−F group is strongly stabilized by the anomeric effect in the 2-F-TFP (ΔE = −6.1 kcal/mol). Contrary to expectation in terms of the αeffect and increased donor ability of the peroxide, the axial C−F group is significantly less stabilized by the two endocyclic oxygen atoms than by one (2.8 kcal/mol). In fact, even the combination of two anomeric interactions in the bis-diaxial difluoroperoxide is still weaker than a single interaction in the respective ether. In this system, the 3.3 kcal/mol decrease in the anomeric stabilization evaluated through these isodesmic equations agrees very well with the 3.6 kcal/mol decrease in the NBO energies for the n(O)→σ*(C‑F)ax interactions in peroxides. Nitrogen-Containing Systems. As we have shown above, the most straightforward way to evaluate the role of the αheteroatom is to use isodesmic equations with reference compounds that have only one endocyclic heteroatom. We will adopt this approach for the subsequent broad search for αeffect in a wider range of heterocycles. We will keep this analysis focused and concentrate on isodesmic equations that separate the donor X−Y part from the C−F acceptor (similar to those given in Scheme 13 for the cyclic peroxides). A particularly useful structural feature of nitrogen is the presence of single lone pair. The precise orientation of this lone pair in space facilitates stereoelectronic analysis and can provide an insight in the relative role of orbital interactions and indirect polarization effects. We will use this feature in two ways: (a) nitrogen as a tool for modifying the donor ability of oxygen and (b) nitrogen itself as the donor atom. Nitrogen as the Buttressing Element. When nitrogen is the buttressing atom adjacent to the donor oxygen (i.e., when the O−O bond is changed to an N−O bond), the weakening of the anomeric effect persists. When the lone pair of nitrogen is axial and, thus, is aligned with the stereoelectronically important lone pair of oxygen, the lowering of the anomeric preference is still observed relative to the tetrahydropyran system, albeit to a slightly lower extent than for the peroxide. Even though neighboring nitrogen does not deactivate donor oxygen as much as neighboring oxygen does, it is clear that presence of any electronegative neighbor decreases the donor ability of a heteroatom relative to the carbon-substituted analogues, piperidine and tetrahydropyran (Scheme 14). Furthermore, a stronger stabilization effect is observed when the N−H bond is axial and thus the lone pair of nitrogen cannot assist in the anomeric stabilization through orbital mixing with the p-type oxygen lone pair. We suggest that such increase in stabilization of the axial N−H moiety can be

attributed to the formation of an N−H···F stabilizing interaction. Nitrogen as the Donor Element. We have also investigated the importance of α-effect at nitrogen created by a neighboring atom with nonbonding electrons. As a starting point of our discussion, it is useful to evaluate the donor ability of a single nitrogen atom using “substituent separation equations” in Scheme 15. Axial lone pair in piperidine stabilizes axial fluorine Scheme 15. “Separation” Isodesmic Reactions Allowing the Comparison of n(N‑Heq)→σ*(C‑F)ax and n(N‑Hax)→σ*(C‑F)ax Hyperconjugation and NBO E(2) Energies for n(N‑Heq)→σ* a (C‑F)ax and n(N‑Hax)→σ*(C‑F)ax Interactions

a

Results in kcal/mol; energies in SMD = DMSO are given in parentheses.

as much as two oxygen lone pairs in tetrahydropyran (5.5 vs 5.7 kcal/mol). The much lower stabilizing effect of the equatorial lone pair (1.4 kcal/mol) illustrates the importance of orbital alignment in the anomeric effect. In fact, a part of this stabilization should be attributed to the exo-anomeric effect (i.e., the nF→σ*C‑N donation).26i The α-effect imposed by oxygen at a donor nitrogen center is similar to what was discussed above. The separation equations in Scheme 16 show that the axial lone pair of endocyclic Scheme 16. “Separation” Isodesmic Reactions Allowing the Comparison of n(O‑N‑Heq)→σ* (C‑F)ax and n(O‑N‑Hax)→σ* (C‑F)ax Hyperconjugation and NBO E(2) Energies for n(O‑N‑Heq)→ σ*(C‑F)ax and n(O‑N‑Hax)→σ*(C‑F)ax Interactionsa

Scheme 14. “Separation” Isodesmic Reactions Allowing the Comparison of n(O)→σ*(C‑F)ax and n(N‑O)→σ*(C‑F)ax Hyperconjugation and NBO E(2) Energies for n(O)→σ* a (C‑F)ax and n(N‑O)→σ* (C‑F)ax Interactions

a

a

Results in kcal/mol; energies in SMD = DMSO are given in parentheses.

nitrogen becomes a less efficient donor when the nitrogen is connected to an oxygen atom. The calculated energies show that the axial conformer is still preferred but the preference is much lower than it was in piperidine (2.1 vs 5.5 kcal/mol). Furthermore, the axial conformer is destabilized when the lone pair of endocyclic nitrogen is equatorial. These results suggest that non-stereoelectronic factors such as orbital hybridization, lone-pair energies, and electrostatics play an important role in the O−N systems. Modulation of calculated energies by solvation effects (data in parentheses) further reinforces this notion.

Results in kcal/mol. 10807

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society

Scheme 18. “Separation” Isodesmic Reactions Allowing the Comparison of n(N‑Heq)→σ* (C−F)ax Hyperconjugation and NBO E(2) Energies for n(N‑Heq)→σ* (C‑F)ax Interactions in Systems Containing an α-C-Radical, an α-C-Anion, or an αN-Aniona

Finally, it is interesting to analyze the well-controlled N−N systems which can potentially provide useful information on stereoelectronic aspects of the modulation of the α-effect. Depending on the relative orientation of the two lone pairs at the two nitrogen atoms, four stereoisomeric systems are possible. In analyzing their relative energies, one has to bear in mind the greater stability of the diaxial conformer of 3fluoropiperidine, apparently stabilized by the electrostatic N− H···F interaction between two axial groups. Unfortunately, this effect masks the role of adjacent nitrogen on the n(N)→σ*C‑F interactions. In any case, the 6.4 kcal/mol stabilization due to reuniting axial C2−F bond and equatorial N−H moiety (i.e., the axial lone pair) in piperidine (Scheme 15), is decreased to 5.9 kcal/mol by an axial N−H group and to −3.1 kcal/mol by an equatorial N−H group (ax/eq and eq/eq examples in Scheme 17). The latter result is particularly striking because

a

Scheme 17. “Separation” Isodesmic Reactions Allowing the Comparison of n(H‑N‑N‑H)→σ* (C‑F)ax Hyperconjugation and NBO E(2) Energies for n(H‑N‑N‑H)→σ* (C‑F)ax Interactions in 3-Fluoro-1,2-diazacyclohexane with ax/eq, eq/eq, ax/ax, and eq/ax Orientation of the N−H Bondsa

a

Results in kcal/mol.

stabilizing ability toward the adjacent axial C−F bond. Carbanion increases donor ability relative to the radical (6.7 vs 4.2 kcal/mol), but the effect only marginally exceeds the axial C−F stabilization energy in piperidine itself (6.4 kcal/mol). Surprisingly, α-effect comes back to life in the last equation of Scheme 18 that shows that, despite the greater electronegativity of nitrogen, the α-effect in the N-anion is greater than in the carbanion (7.9 vs 6.7 kcal/mol)! A priori, one could suggest that a part of the above increase may originate from the more favorable hybridization of the key lone pair of anionic nitrogen in comparison to the lone pair at the anionic carbon. N-anions are isoelectronic to neutral oxygen compounds and have a high-energy p-type lone pair, whereas pyramidalized carbanions have their negative charge localized in an spn-hybrid lone pair.33 Fortunately, NBO analysis can provide a direct insight into the relative hyperconjugative donor ability of the carbon and nitrogen lone pairs. In contrast to the neutral “α-systems” discussed earlier, the NBO interaction energies for the anionic systems do increase dramatically, suggesting enhanced donor ability of the nitrogen lone pairs in the presence of an adjacent negative charge (Scheme 19). A large part of this increase comes from the lowered gap between the interacting n(N) and σ*(C‑F) Scheme 19. Effect of Lone-Pair Hybridization on Orbital Interactions in the C- and N-Centered Anomeric Anionsa

Results in kcal/mol.

this is the case where the lone pairs of both nitrogens are axial and aligned well with each other and the axially oriented C−F bond. Nevertheless, the much weaker apparent anomeric stabilization is observed in a stark contrast to the expectations based on the α-effect. The effects of adjacent nitrogen atoms in the ax,ax and eq,eq examples in Scheme 17 (where the principal n(N)→σ*C‑F interaction is deactivated) are even more complex. The ax,ax conformer is 1.4 kcal/mol more stabilized by the axial C−F substituent than the analogous piperidine, whereas the eq,ax conformer is 2.5 kcal/mol less stabilized than the analogous piperidine. Considering the complex role of remote axial N−H groups on the apparent donor ability of the anomeric nitrogen atom, we have further expanded our analysis to systems where such polar bonds were not present (Scheme 18). Since the radical is an acceptor, it does not provide a buttressing effect to the nitrogen lone pair. Instead, as expected, it weakens its

a

10808

Interaction energies in kcal/mol. DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society

process that models the TS and the product for an SN1 ionization at the anomeric position. The C−F stretching strongly increases the acceptor ability of this bond, providing increasing “electron demand” for the assistance from the α-heteroatom. The C−F distance scans in the left panel of Scheme 21 show that for the each of the C−F distances, energy penalty for the C−F distance increase is always lower for the ether 2 than for peroxide 4. At 2 Å (C−F distance chosen to approximate the TS geometry for C···F scission), the penalty for C−F stretching is ∼8 kcal/mol higher for the peroxide 4 than it is for the ether 2. In order to test whether constraints imposed by the six-membered cycle play a role in the interaction of peroxide moiety with the breaking C− F bond, we also analyzed the effects of C−F bond stretching in the acyclic α-F peroxide (the right panel of Scheme 21). The latter plot compares C−O−O−C dihedral scans for the ground state of CH3−O−O−CH2F and the system where the C−F bond is elongated to 2 Å. The global minima for the two systems correspond to dihedrals in the vicinity of 100°. Again, the penalty for aligning the lone pairs of the two oxygens (i.e., reaching the COOC dihedral of 180°) is greater when the C−F bond is stretched.34 When the process of C−F stretching is taken to its logical end, i.e., the formation of a carbenium ion, the difference in the cation energies is ∼17 kcal/mol in favor of the oxacarbenium ion over the dioxacarbenium ion. Although both the ether and the peroxide moieties in the ring stabilize potential energy surface for the cation formation relative to that for the cyclohexyl-cation formation, presence of an α-oxygen significantly decreases donor ability of an oxygen group next to the cation (9 vs 26 kcal/mol difference in the ΔE values). We have also compared stabilities of acyclic cations and found that the difference in the relative magnitude of cation stabilization by the ether and peroxide moieties is significantly decreased relative to that in the six-membered cycles (7 vs 17 kcal/mol difference in the ΔE values). These findings suggest that the structural constraints imposed by the cyclic frame restrict geometric reorganization needed for taking the full advantage of the electronic stabilization. However, it is clear that the peroxide moiety is a weaker cation-stabilizing substituent than an ether group even in the acyclic systems. Scheme 22 illustrates the difference the cationic center makes in the PES of dimethyl-peroxide. The cationic example

orbitals that originates from the presence of a negative charge. However, the above increase in the single intramolecular interaction is not translated into the larger overall molecular stability. Although it is clear that the “anomeric” n(N)→σ*(C‑F) interactions represent only a part in a complex jigsaw puzzle of interconnected effects, their increase in the anionic systems along with the small increase in the overall stabilization energy suggests that, under the conditions of increased “electron supply” that is not masked by inductive effects, an intermolecular version of the α-effect is possible. Increasing Electron Demand. In the final part of this work, we will test whether the α-effect that seems to be unimportant in the neutral compounds and weak in the anionic systems can be brought into life by increase in the “electron demand” brought by a stronger acceptor. As a first step, we have replaced the σ-C−F acceptor by a πacceptor (the carbonyl group). Scheme 20 illustrates that, although both the −O− and the −O−O− groups can stabilize the carbonyl, one oxygen provides twice the stabilization as two oxygens of a peroxide. Scheme 20. Carbonyl Stabilization in a Cyclic Ether and a Cyclic Peroxidea

a

Energies in kcal/mol.

Perhaps a much stronger acceptor would have the power to resurrect α-effect? In order to probe this question, one can increase electron demand gradually by stretching the C−F bond en route to an “anomeric” cation via an SN1 path. Let us use transition from a stretched C−F bond to a carbocation, a

Scheme 21. (A) Effect of C−F Bond Stretching on Energy of Ether 2 and Peroxide 4a and (B) C−O−O−C Dihedral Scans for the Ground State of CH3OOCH2F and the System Where the C−F bond Is Elongated to 2 Åb

a

Note that the two lines diverge. Calculations for done at the (SMD = DMSO)/M06-2X/6-311++G(d,p) level of theory. bLevel of theory: B2PLYPD3BJ/6-311++G(d,p). 10809

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society

Scheme 22. (A) Cation Stabilization in a Cyclic Ether and a Cyclic Peroxidea and (B) C−O−O−C Dihedral Scans for the Ground State of CH3−O−O−CH3 and Related Catiobn

a

Energies, in kcal/mol, at the M06-2X/6-311++G(d,p) level of theory. bLevel of theory: B2PLYP-D3BJ/6-311++G(d,p). Resonance structures with positive charge on carbon are shown for oxa- and peroxacarbenium ions to illustrate and contrast the stabilizing role of the oxygen-containing groups.

showcases the influence of the α-effect: by electronic-demand, the transition structure geometry of the neutral system becomes a shallow minimum in the cationic system.35 The existence of energy minima for the flat cis and trans geometries suggests that interaction of the cationic center and the oxygen lone pairs does follow the expected stereoelectronic pattern. However, comparison with the stabilization energies in the left part of Scheme 22 suggests that stereoelectronics has to operate on top of a destabilizing inductive effect that is exacerbated by placing positive charge on the oxygen in the oxacarbenium resonance structures. The penalty for the partial positive charge localization at oxygen is illustrated by comparison of stabilization energies in the respective radicals (Scheme 23). Stabilization of a radical by

single oxygen atom provides more stabilization to a radical than the two adjacent oxygen atoms of a peroxide group. Experimental Consequences. The difference in stabilization provided by the two oxygen-containing functionalities to radical and cationic centers has significant consequences for experimentally observed reactivity and suggests new mechanistic features important for the future of reaction design. Radical Reactions. Intramolecular addition of carbon centered radicals to the CO bonds usually occurs at the carbon atom in the favorable exo fashion. However, addition at oxygen is also possible.36,37 This is a relatively unusual but a useful process because it allows access to the less accessible endo-trig cyclization path.38 One of the reasons why the Ocentered cyclization of carbon-centered radicals is thermodynamically feasible is the anomeric stabilization of a radical center in the cyclic product by the lone pair of oxygen.39 In contrast, analogous cyclizations of O-centered radicals are thermodynamically uphill (Scheme 24). In the latter case, the inefficiency of anomeric stabilization in the cyclic peroxide complements the weakness of the O−O bonds. As the result, the reverse reactions (fragmentations of cyclic radicals with the formation of carbonyl groups) are much more favorable for the peroxide-containing α-radicals. Cationic Reactions. The dramatically different cation stabilization by the OR and OOR groups has far-reaching experimental implications for the synthesis of organic peroxides, an important class of organic compounds with rapidly expanding role in drug design.40 This lower stabilization in peroxide-bearing cation should shape the potential energy surface for the equilibrium between the cationic species involved in the acid catalyzed formations of endo- and exocyclic peroxides and hydroperoxides from carbonyl compounds and H2O2.41 These processes open synthetic access to a surprising diversity of oxygen-containing functionalities including geminal bis-hydroperoxides,42 geminal bisperoxides,43 tetraoxanes,44 cyclic triperoxides,45 tricyclic monoperoxides,46 and ozonides.47 Scheme 25 illustrates how the mechanistic complexity of such reactions can be understood better once the different donor abilities of the OR and OOR groups are taken into account. In the classic study of the reaction of 1,3-diketones with hydrogen peroxide by Milas et al.,48 the initial conversion

Scheme 23. Radical Stabilization in a Cyclic Ether and a Cyclic Peroxidea

a

Energies in kcal/mol.

an oxygen atom in the cyclic ether is on par (Scheme 23, ΔG = −23.0 kcal/mol) with the stabilization of cations in a similar setting (Scheme 22). Interestingly, geometry optimization of a respective peroxide led to the fragmentation of the weak O−O bond with the formation of a carbonyl moiety and an Ocentered radical. The balance between opened and closed structures is sensitive to structural effectsproviding additional stabilization to the cyclic radical by a methyl substituent prevented the in silico fragmentation. Although the difference between peroxide and ether stabilization is smaller for radicals than it is for cations (ΔG = +2.2 vs +16.4 kcal/mol), the same trend is observed. Opposite to expectations based on α-effect, a 10810

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society Scheme 24. Differences in Radical Stabilization in a Cyclic Ether and a Cyclic Peroxide Account for the Different Balance in O-Centered Radical Cyclizations and Fragmentations for These Functionalitiesa

a

Scheme 26. Differences in Cation Stabilization by Ether and Peroxide Groups Prevent Displacement of the Second OR Group in the I2-Catalyzed Reactions of Bis-hydroperoxides and Acetals

facilitated by anomeric assistance by the remaining OR group, the loss of a second OR group has to be assisted by the newly introduced OOR′ groupa much weaker donor. The reaction approach outlined in Scheme 26 uses a selective activation route where iodine acts as a mild Lewis acid, coordinating with the methoxy group of the acetal and assisting to its departure. This soft activation mode is apparently insufficient for the activation of the peroxoacetal analogue. This result is a striking example of possible control or reaction selectivity that arises from the difference in the apparent donor ability of OR and OOR groups. We have to leave a more detailed and extensive discussion of the myriad of pathways involved in the acid-catalyzed equilibrium between carbonyl compounds, ethers, and peroxides for the future. The field is complex and, in most cases, many questions, even as fundamental as the interplay between kinetic and thermodynamic control, remain open. In addition, our preliminary results suggest that the high energy stored in H2O2 may overcome some of the issues associated with the relative stability of peroxacarbenium ions. We hope that the ability to include these factors in the future reaction design will help in expanding the frontiers of synthetic peroxide chemistry.

Energies in kcal/mol.

Scheme 25. Differences in Cation Stabilization by Hydroxyl and Peroxy-Containing Groups Lead to Different Kinetic Regimes (Fast and Slow Stages) in the Multistep Reaction of β-Diketones with Hydrogen Peroxide

of carbonyl groups (or, alternatively, ketals) to mixed monocyclic hydroxyl/peroxy derivatives proceeded quickly. These intermediate cyclic structures can be isolated, suggesting their relatively low reactivity. Note that the route to monocyclic peroxides proceeded via the formation of oxacarbenium ions, relatively stable species that are expected to be formed readily. On the other hand, the subsequent conversion of monocyclic peroxides into bicyclic bis-peroxides (Scheme 25) is slower and requires an excess of H2O2. This observation is consistent with the lower stability of the peroxycarbenium ions (or the respective SN2 TSs) as a consequence of “reverse α-effect”, discussed above. The difference in the relative stability of the two types of cationic species can be used in a variety of ways. For example, Terent’ev and co-workers reported that ketal in Scheme 26 undergoes substitution quickly to give a mixed peroxy/ether ketal that does not react further. This observation was leveraged to develop a synthetically useful approach to unusual oxygenrich compounds that combine hydroperoxy, peroxy, and ether functionalities.41c Again, the computational insights outlined earlier offer an explanation to the different rates for the departures of the two OR groups. Whereas the loss of a first OR group can be



CONCLUSIONS The magnitudes of the X−C−F and the Y−X−C−F anomeric effects (X = NH, O; Y = NH, O, CH•, CH−) were compared from the calculated free energy differences of conformationally constrained heterocycles. The results are contrary to expectation in terms of the α-effect, which would suggest increased donor ability of the system with two adjacent heteroatoms [n(Y‑X)→σ*(C‑F)ax] relative to an analogous system with a single heteroatom [nX→σ* (C‑F)ax] (e.g., peroxides vs ethers). NBO analysis suggested the energies of the delocalizing orbital interactions that stabilize the conformations with axial C−F bond is lower for the α-systems. In particular, the stabilization gained from n(X)→σ*(C−F)ax interaction is actually larger than the apparent stabilization estimated for the n(Y‑X)→ σ*(C‑F)ax interaction for the X−Y neutral fragments. The isodesmic “separation” equations further support the absence of intramolecular α-effect. Increasing electronic “push” by making the α-position negatively charged led to a small to moderate increase in anomeric stabilization, suggesting the activation of the α-effect. 10811

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Journal of the American Chemical Society



However, the overall stabilization is weaker than expected from the large increase in the stereoelectronic component, indicating that, in the anionic systems, the electrostatic components mask the role of stabilizing orbital interactions. On the other hand, the greater electronic “pull” of electrondeficient reactive species (radicals, cations, and stretched C−F bonds) did not activate the α-effect. Reactions that form carbocations and radicals get greater thermodynamic assistance from an α-ether than from an α-peroxide moiety (i.e., peroxacarbenium ions are less stable than oxacarbenium ions). The relative inefficiency of peroxides in positive charge stabilization should have significant implications for understanding the mechanistic complexity associated with the interaction of carbonyl compounds with hydroperoxides and H2O2 in acidic media, as such reactions involve α-cationic intermediates. The “inverse” α-effect should play an important role in the design of new synthetic approaches to organic peroxides that benefit from seemingly anomalous stabilities of some of the peroxy-containing organic functionalities under acidic conditions.



COMPUTATIONAL METHODS



ASSOCIATED CONTENT

REFERENCES

(1) Jencks, W. P.; Carriuolo, J. J. Am. Chem. Soc. 1960, 82, 1778. (2) Edwards, J. O.; Pearson, R. G. J. Am. Chem. Soc. 1962, 84, 16. (3) Gold, V. Pure Appl. Chem. 1979, 51, 1731. (4) Hoz, S.; Buncel, E. Isr. J. Chem. 1985, 26, 313. (5) (a) Villano, S. M.; Eyet, N.; Lineberger, W. C.; Bierbaum, V. M. J. Am. Chem. Soc. 2009, 131, 8227. (b) Garver, J. M.; Gronert, S.; Bierbaum, V. M. J. Am. Chem. Soc. 2011, 133, 13894. (6) (a) Thomsen, D. L.; Reece, J. N.; Nichols, C. M.; Hammerum, S.; Bierbaum, V. M. J. Am. Chem. Soc. 2013, 135, 15508. (b) Thomsen, D. L.; Reece, J. N.; Nichols, C. M.; Hammerum, S.; Bierbaum, V. M. J. Phys. Chem. A 2014, 118, 8060. (c) Thomsen, D. L.; Nichols, C. M.; Reece, J. N.; Hammerum, S.; Bierbaum, V. M. J. Am. Soc. Mass Spectrom. 2014, 25, 159. (d) Zhao, W.-Y.; Yu, J.; Ren, S.-J.; Wei, X.-G.; Qiu, F.-Z.; Li, P.-H.; Li, H.; Zhou, Y.-P.; Yin, C.-Z.; Chen, A.-P.; Li, H.; Zhang, L.; Zhu, J.; Ren, Y.; Lau, K.-C. J. Comput. Chem. 2015, 36, 844. (e) Singh, N.; Karpichev, Y.; Sharma, R.; Gupta, B.; Sahu, A. K.; Satnami, M. L.; Ghosh, K. K. Org. Biomol. Chem. 2015, 13, 2827. (7) McAnoy, A. M.; Paine, M. R. L.; Blanksby, S. J. Org. Biomol. Chem. 2008, 6, 2316. (8) Buncel, E.; Um, I. H. Tetrahedron 2004, 60, 7801. (9) (a) Hoz, S. J. Org. Chem. 1982, 47, 3545. (b) Patterson, E. V.; Fountain, K. R. J. Org. Chem. 2006, 71, 8121. (10) Alabugin, I. V.; Bresch, S.; dos Passos Gomes, G. J. Phys. Org. Chem. 2015, 28, 147. (11) Relevant pKa values of representative O-centered “normal” and α-systems in water: H2O, 15.7; H2O2, 11.6; MeOOH, 11.5; MeOH, 15.5. Note, however, that MeCO2H (4.8) is a stronger acid than MeCO3H (8.2). The latter fact cannot be attributed to α-effect; the acetate anion is simply more stabilized by conjugation because it is directly connected to a carbonyl. (12) (a) Bordwell, F. G.; Algrim, D.; Vanier, N. R. J. Org. Chem. 1977, 42, 1817. (b) Bordwell, F. G.; Algrim, D. J. J. Am. Chem. Soc. 1988, 110, 2964. (13) Zhao, Y.; Bordwell, F. G.; Cheng, J.-P.; Wang, D. J. Am. Chem. Soc. 1997, 119, 9125. (14) Vatsadze, S. Z.; Loginova, Y. D.; dos Passos Gomes, G.; Alabugin, I. V. Chem. - Eur. J. 2017, 23, 3225. (15) (a) Hoz, S.; Speizman, D. J. Org. Chem. 1983, 48, 2904. (b) Buncel, E.; Hoz, S. Tetrahedron Lett. 1983, 24, 4777. (16) It also can be compromised by electronegativity and hybridization effects, as we will discuss below. (17) Nigst, T. A.; Antipova, A.; Mayr, H. J. Org. Chem. 2012, 77, 8142. (18) Alabugin, I. V. Stereoelectronic Effects: the Bridge between Structure and Reactivity; John Wiley & Sons Ltd: Chichester, UK, 2016. (19) Evanseck, J. D.; Blake, J. F.; Jorgensen, W. L. J. Am. Chem. Soc. 1987, 109, 2349. (20) (a) Ren, Y.; Yamataka, H. Org. Lett. 2006, 8, 119. (b) Ren, Y.; Yamataka, H. J. Org. Chem. 2007, 72, 5660. (21) Bent’s rule states that s-character concentrates in orbitals directed toward electropositive substituents or, alternatively, that atoms direct hybrid orbitals with more p-character towards more electronegative elements. (a) Bent, H. A. J. Chem. Phys. 1960, 33, 1258. (b) Bent, H. A. J. Chem. Educ. 1960, 37, 616. (c) Bent, H. A. Chem. Rev. 1961, 61, 275. (d) See ref 10. (22) Alabugin, I. V.; Bresch, S.; Manoharan, M. J. Phys. Chem. A 2014, 118, 3663. (23) The anomeric effect, historically defined as the axial preference for the acceptor groups at the anomeric position of carbohydrates, defines stability of many X−C−Y compounds with two geminal heteroatoms X and Y (i.e., the generalized anomeric effect). The stereoelectronic component of the anomeric effect is generally attributed to hyperconjugative nX→σ*C‑Y interactions. (24) Alabugin, I. V.; Manoharan, M.; Buck, M.; Clark, R. J. J. Mol. Struct.: THEOCHEM 2007, 813, 21. (25) This approach is useful for the understanding of steric, electrostatic, and stereoelectronic interactions in organic molecules.

Calculations were carried with the Gaussian ‘09 software package,49 using the (U)M06-2X/6-311++G(d,p)50 or the MP2/6-311+G(d,p)51 levels of theory. For the relaxed PES scans, we also used the doublehybrid B2PLYP functional,52 with the D3 version of Grimme’s dispersion with Becke−Johnson (BJ) damping.53 Unless otherwise noted, all calculations were performed at the (U)M06-2X/6-311+ +G(d,p) level of theory. Delocalizing interactions were evaluated with NBO analysis method. Chemcraft 1.754 and CYLView55 were used to render the orbitals and three-dimensional molecules.

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b05367. Information on the calculated structures studied in this work (PDF)



Article

AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] *[email protected] ORCID

Eusebio Juaristi: 0000-0003-0936-7020 Gabriel dos Passos Gomes: 0000-0002-8235-5969 Alexander O. Terent’ev: 0000-0001-8018-031X Igor V. Alabugin: 0000-0001-9289-3819 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Financial support from CONACYT-Mexico, grant number 220945, is gratefully acknowledged. A.O.T. acknowledges the financial support from the Russian Science Foundation (grant 14-50-00126). I.V.A. is grateful to V. Vil’ for helpful discussions of peroxide chemistry. I.V.A. and G.d.P.G. appreciate for partial support of this research from the National Science Foundation (CHE-1465142) and the allocation of computational resources from FSU RCC and the NSF XSEDE (TG-CHE160006). 10812

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813

Article

Journal of the American Chemical Society Cf.: Juaristi, E.; Cuevas, G. The Anomeric Effect; CRC: Boca Raton, FL, 1995. (26) (a) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899. For selected applications of NBO method for analysis of chemical bonding, see: (b) Reed, A. E.; Weinhold, F. Isr. J. Chem. 1991, 31, 277. (c) Goodman, L.; Pophristic, V. T. Nature 2001, 411, 565. (d) Podlech, J. J. Phys. Chem. A 2010, 114, 8480. (e) Freitas, M. P. Org. Biomol. Chem. 2013, 11, 2885. (f) Juaristi, E.; Notario, R. J. Org. Chem. 2015, 80, 2879. (g) Gomes, G. P.; Vil’, V.; Terent’ev, A.; Alabugin, I. V. Chem. Sci. 2015, 6, 6783. (h) Vidhani, D.; Krafft, M.; Alabugin, I. V. J. Am. Chem. Soc. 2016, 138, 2769. (i) Juaristi, E.; Notario, R. J. Org. Chem. 2016, 81, 1192. (j) dos Passos Gomes, G.; Alabugin, I. V. J. Am. Chem. Soc. 2017, 139, 3406. (27) Chu, P.-S.; True, N. S. J. Phys. Chem. 1985, 89, 5613. (28) (a) Eliel, E. L.; Wilen, S. H. Stereochemistry of Organic Compounds; Wiley: New York, 1994. (b) Juaristi, E. Introduction to Stereochemistry and Conformational Analysis; Wiley: New York, 1991. (29) Weinhold, F.; Schleyer, P. v. R. Encyclopedia of Computational Chemistry; Wiley: New York, 1998; Vol. 3, p 1792. (30) Reed, A. E.; Weinhold, F. J. Chem. Phys. 1985, 83, 1736. (31) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899. (32) Hehre, W. J.; Ditchfield, R.; Radom, L.; Pople, J. A. J. Am. Chem. Soc. 1970, 92, 4796. (33) For a general discussion of the interplay between electronegativity and hybridization on the donor ability of lone pairs, see: (a) Alabugin, I. V.; Manoharan, M.; Zeidan, T. A. J. Am. Chem. Soc. 2003, 125, 14014. (b) Alabugin, I. V. General Stereoelectronic TrendsDonors, Acceptors, and Chameleons. In Stereoelectronic Effects: the Bridge between Structure and Reactivity; Alabugin, I. V., Ed.; John Wiley & Sons LtdChichester, UK, 2016; Chapter 5, DOI: 10.1002/9781118906378.ch5. (34) The left part of the energy curve, where the methyl group and the breaking C−F bond are close in space (negative COOC dihedral values), is stabilized relative to the right part of the plots, where the methyl and group and the C−F bond are away from each other (negative COOC dihedral values), most likely due to through-space CH···F interactions. Examples of CF···H interactions: (a) Pitts, C. R.; Siegler, M. A.; Lectka, T. J. Org. Chem. 2017, 82, 3996. (b) Gold, B.; Shevchenko, N.; Bonus, N.; Dudley, G. B.; Alabugin, I. V. J. Org. Chem. 2012, 77, 75. (35) Similar conversion of TSs to potential energy minima by stabilizing electronic factors has been observed in pericyclic reactions: (a) Gilmore, K.; Manoharan, M.; Wu, J.; Schleyer, P. v. R; Alabugin, I. V. J. Am. Chem. Soc. 2012, 134, 10584. (b) See ref 26h. (c) See ref 26j. (36) See, for example: Menapace, L. W.; Kuivila, H. G. J. Am. Chem. Soc. 1964, 86, 3047. (37) A recent review: Liu, D.; Liu, C.; Lei, A. Chem. - Asian J. 2015, 10, 2040. (38) Gilmore, K.; Mohamed, R. K.; Alabugin, I. V. WIREs: Comput. Mol. Sci. 2016, 6, 487. (39) Note also the unusual TS for this reaction, where the carbon radical interacts mostly with the in-plane lone pair of the carbonyl oxygen (via a 2c,3e bond) instead of a more common attack at the πsystem. (40) (a) Jefford, C. W. Adv. Drug Res. 1997, 29, 271−325. (b) O’Neill, P. M.; Posner, G. H. J. Med. Chem. 2004, 47, 2945. (c) Chung, A.; Miner, M. R.; Richert, K. J.; Rieder, C. J.; Woerpel, K. A. J. Org. Chem. 2015, 80, 266. (d) Gilmore, K.; Kopetzki, D.; Lee, J. W.; Horvath, Z.; McQuade, D. T.; Seidel-Morgenstern, A.; Seeberger, P. H. Chem. Commun. 2014, 50, 12652. (e) Vil’, V. A.; Yaremenko, I. A.; Ilovaisky, A. I.; Terent’ev, A. O. Molecules 2017, 22, 117. (41) Review: (a) Terent’ev, A. O.; Borisov, D. A.; Vil’, V. A.; Dembitsky, V. M. Beilstein J. Org. Chem. 2014, 10, 34. Recent examples: (b) Terent’ev, A. O.; Krylov, I. B.; Chernyshev, V. V.; Nikishin, G. I.; Platonov, M. M. Org. Biomol. Chem. 2008, 6, 4435. (c) Terent’ev, A. O.; Yaremenko, I. A.; Chernyshev, V. V.; Dembitsky, V. M.; Nikishin, G. I. J. Org. Chem. 2012, 77, 1833. (d) Terent’ev, A.

O.; Yaremenko, I. A.; Vil’, V. A.; Moiseev, I. K.; Kon’kov, S. A.; Dembitsky, V. M.; Levitsky, D. O.; Nikishin, G. I. Org. Biomol. Chem. 2013, 11, 2613. (e) Terent’ev, A. O.; Yaremenko, I. A.; Vil’, V. A.; Dembitsky, V. M.; Nikishin, G. I. Synthesis 2013, 45, 246. (42) (a) Ž mitek, K.; Zupan, M.; Stavber, S.; Iskra, J. Org. Lett. 2006, 8, 2491. (b) Terent’ev, A. O.; Platonov, M. M.; Ogibin, Y. N.; Nikishin, G. I. Synth. Commun. 2007, 37, 1281. (c) Ghorai, P.; Dussault, P. H. Org. Lett. 2008, 10, 4577. (d) Schwartz, C.; Dussault, P. H. In Patai’s Chemistry of Functional Groups; John Wiley & Sons, Ltd: Chichester, UK, 2009. (e) Li, Y.; Hao, H.-D.; Zhang, Q.; Wu, Y. Org. Lett. 2009, 11, 1615. (f) Bunge, A.; Hamann, H.-J.; Liebscher, J. Tetrahedron Lett. 2009, 50, 524. (g) Azarifar, D.; Khosravi, K.; Soleimanei, F. Molecules 2010, 15, 1433. (h) Tada, N.; Cui, L.; Okubo, H.; Miura, T.; Itoh, A. Chem. Commun. 2010, 46, 1772. (i) Liu, Y.-H.; Deng, J.; Gao, J.-W.; Zhang, Z.-H. Adv. Synth. Catal. 2012, 354, 441. (j) Azarifar, D.; Najminejad, Z.; Khosravi, K. Synth. Commun. 2013, 43, 826. (k) van Tonder, J. H. Synlett 2014, 25, 1629. (l) Surya Prakash, G. K.; Shakhmin, A.; Glinton, K. E.; Rao, S.; Mathew, T.; Olah, G. A. Green Chem. 2014, 16, 3616. (m) Azarifar, D.; Khosravi, K.; Soleimanei, F. Synthesis 2009, 2009, 2553. (n) Das, B.; Krishnaiah, M.; Veeranjaneyulu, B.; Ravikanth, B. Tetrahedron Lett. 2007, 48, 6286. (43) (a) Ž mitek, K.; Zupan, M.; Stavber, S.; Iskra, J. J. Org. Chem. 2007, 72, 6534. (b) Terent’ev, A. O.; Platonov, M. M.; Krylov, I. B.; Chernyshev, V. V.; Nikishin, G. I. Org. Biomol. Chem. 2008, 6, 4435. (c) Kyasa, S.; Puffer, B. W.; Dussault, P. H. J. Org. Chem. 2013, 78, 3452. (d) Kandur, W. V.; Richert, K. J.; Rieder, C. J.; Thomas, A. M.; Hu, C.; Ziller, J. W.; Woerpel, K. A. Org. Lett. 2014, 16, 2650. (44) (a) Terent’ev, A. O.; Borisov, D. A.; Chernyshev, V. V.; Nikishin, G. I. J. Org. Chem. 2009, 74, 3335. (b) Terent’ev, A. O.; Yaremenko, I. A.; Vil’, V. A.; Moiseev, I. K.; Kon’kov, S. A.; Dembitsky, V. M.; Levitsky, D. O.; Nikishin, G. I. Org. Biomol. Chem. 2013, 11, 2613. (c) Ghorai, P.; Dussault, P. H. Org. Lett. 2009, 11, 213. (d) Novikov, V. L.; Shestak, O. P. Russ. Chem. Bull. 2013, 62, 2171. (e) Yadav, N.; Sharma, C.; Awasthi, S. K. RSC Adv. 2014, 4, 5469. (f) Terent’ev, A. O.; Borisov, D. A.; Vil’, V. A.; Dembitsky, V. M. Beilstein J. Org. Chem. 2014, 10, 34. (g) Klapötke, T. M.; Stiasny, B.; Stierstorfer, J.; Winter, C. H. Eur. J. Org. Chem. 2015, 2015, 6237. (45) (a) Rieche, A.; Bischoff, C.; Prescher, D. Chem. Ber. 1964, 97, 3071. (b) McCullough, K. J.; Morgan, A. R.; Nonhebel, D. C.; Pauson, P. L.; White, G. J. J. Chem. Res., Miniprint 1980, 2, 601. (c) Ž mitek, K.; Stavber, S.; Zupan, M.; Bonnet-Delpon, D.; Iskra, J. Tetrahedron 2006, 62, 1479. (d) Terent’ev, A. O.; Platonov, M. M.; Sonneveld, E. J.; Peschar, R.; Chernyshev, V. V.; Starikova, Z. A.; Nikishin, G. I. J. Org. Chem. 2007, 72, 7237. (e) Terent’ev, A. O.; Platonov, M. M.; Tursina, A. I.; Chernyshev, V. V.; Nikishin, G. I. J. Org. Chem. 2008, 73, 3169. (46) (a) Terent’ev, A. O.; Yaremenko, I. A.; Chernyshev, V. V.; Dembitsky, V. M.; Nikishin, G. I. J. Org. Chem. 2012, 77, 1833. (b) Terent’ev, A. O.; Yaremenko, I. A.; Glinushkin, A. P.; Nikishin, G. I. Russ. J. Org. Chem. 2015, 51, 1681. (47) dos Passos Gomes, G.; Yaremenko, I. A.; Radulov, P. S.; Novikov, R. A.; Chernyshev, V. V.; Korlyukov, A. A.; Nikishin, G. I.; Alabugin, I. V.; Terent’ev, A. O. Angew. Chem., Int. Ed. 2017, 56, 4955. (48) Milas, N. A.; Mageli, O. L.; Golubović, A.; Arndt, R. W.; Ho, J. C. J. J. Am. Chem. Soc. 1963, 85, 222. (49) Frisch, M. J.; et al. Gaussian 09, Revision D.01; Gaussian Inc.: Wallingford, CT, 2009; (complete reference in the SI). (50) (a) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215. (b) Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157. (51) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett. 1988, 153, 503. (52) Grimme, S. J. Comput. Chem. 2006, 27, 1787. Grimme, S. J. Chem. Phys. 2006, 124, 034108/1−034108/16. Schwabe, T.; Grimme, S. Phys. Chem. Chem. Phys. 2007, 9, 3397. (53) Grimme, S.; Ehrlich, S.; Goerigk, L. J. Comput. Chem. 2011, 32, 1456. (54) ChemCraft 1.7, build number 405; http://www.chemcraftprog. com (accessed February 2015). (55) Legault, C. Y. CYLview, 1.0b; Université de Sherbrooke: Quebec, 2009; http://www.cylview.org. 10813

DOI: 10.1021/jacs.7b05367 J. Am. Chem. Soc. 2017, 139, 10799−10813