Probing the Free Energy Landscape of Organophotoredox-Catalyzed Anti-Markovnikov Hydrofunctionalization of Alkenes

Abstract

Experimental ¹³C kinetic isotope effects (KIEs) provide unprecedented mechanistic insight into three intermolecular anti-Markovnikov alkene hydrofunctionalization reactions—hydroesterification, hydroamination, and hydroetherification—enabled by organophotoredox catalysis. All three reactions are found to proceed via initial oxidation of the model alkenes to form a radical cation intermediate, followed by sequential nucleophilic attack and hydrogen-atom transfer to deliver the hydrofunctionalized product. A normal ¹³C KIE on the olefinic carbon that undergoes nucleophilic attack provides qualitative evidence for rate-limiting nucleophilic attack in all three reactions. Comparison to predicted ¹³C KIE values obtained from density functional theory (DFT) calculations for this step reveals that alkene oxidation has partial rate-limiting influence in hydroesterification and hydroamination, while the nucleophilic attack is solely rate-limiting in the hydroetherification reaction. The basic additive (2,6-lutidine) activates the nucleophile via deprotonation and is an integral part of the transition state for nucleophilic attack on the radical cation, providing an important design principle for the development of asymmetric versions of these reactions. A more electron-rich pyridine base (2,6-dimethoxypyridine) exhibits considerable rate enhancements in both inter- and intramolecular hydrofunctionalization reactions.

Graphical Abstract

INTRODUCTION

The Nicewicz lab has pioneered the development of anti-Markovnikov hydrofunctionalization reactions of alkenes that utilize acridinium salts as photocatalysts.¹ The seminal reports describe intramolecular hydroetherification of alkenols² and intramolecular hydroamination of unsaturated amines (Scheme 1A).^3,4 Following these initial reports, the Nicewicz lab also reported intermolecular anti-Markovnikov hydrofunctionalization of alkenes (Scheme 1B). Specific examples include hydroesterification of alkenes using carboxylic acids,⁵ hydroamination of alkenes using triflyl amines,⁶ and hydrohalogenation of styrenes using mineral acids.⁷ Fundamental to the success of these reactions is the use of catalytic amounts of a base (2,6-lutidine) and a thiol co-catalyst in addition to the organic photocatalyst.

Scheme 1.

(A) Intra- and Intermolecular Anti-Markovnikov Hydrofunctionalization of Olefins via Organophotoredox Catalysis Pioneered by the Nicewicz Group and (B) Catalytic Cycle of Intermolecular Hydrofunctionalization Reactions Investigated in This Study

The proposed mechanism of the intermolecular hydrofunctionalization (using anethole (1) as a model alkene) involves the oxidation of 1 via single electron transfer (SET) by a photoexcited acridinium catalyst (Mes-Acr⁺*) to form cation radical intermediate 1^·+ and the reduced acridine radical (Mes-Acr●). This is followed by attack of the nucleophile (Nu–H), resulting in the formation of carbon-centered radical intermediate 4. H-atom transfer (HAT) from the thiol co-catalyst to 4 delivers the anti-Markovnikov product 3 and a thiyl radical (Scheme 1B). The thiyl radical likely oxidizes Mes-Acr● to Mes-Acr⁺ while forming a thiolate anion. The base, 2,6-lutidine, presumably serves as a proton shuttle between Nu–H and the thiolate anion to regenerate the thiol co-catalyst.

Fundamental mechanistic studies by the Nicewicz group using a battery of photophysical probes provide valuable insight into the elementary steps involved in this catalytic cycle.⁴ Stern–Volmer analysis reveals that oxidation of alkenes such as anethole (1, E_P/2 = 1.34 V) can occur via SET to either a singlet excited state of Mes-Acr⁺ (E*_red > +2.0 V) or a locally excited or charge transfer triplet state (E*_red = +1.45/1.88 V). Transient absorption spectroscopy supports formation of 1^·+ as a key intermediate in this reaction. Evidence was also obtained for the turnover of the Mes-Acr● by the thiyl radical as the mechanistic event that unites the photoredox and HAT catalytic cycles.

Key questions regarding the overall mechanism of the intermolecular hydrofunctionalization of alkenes remain unclear (Scheme 1B). (1) What is the overall rate-limiting step? (2) Does the nature of the nucleophile affect the free energy landscape of these reactions? (3) What is the exact nature and timing of the role of 2,6-lutidine in the catalytic mechanism? (4) What are the key features of the transition states of the nucleophilic attack and HAT step? We sought to answer these questions utilizing a combination of experimental ¹³C kinetic isotope effects (KIEs) and computational studies. Since the olefinic carbon atoms undergo bonding change or rehybridization in each step of the catalytic cycle, experimental ¹³C KIEs can provide vital insight into the details of the free energy landscape of the reaction. We report, herein, a combined experimental and theoretical study that investigates the general mechanism of intermolecular hydrofunctionalization of alkenes catalyzed by an organophotoredox catalyst.^8–12

RESULTS AND DISCUSSION

We initiated our study with the prototypical reactions of anethole (1) with two different nucleophiles—benzoic acid (2a, hydroesterification) and triflylamine (2b, hydroamination)—catalyzed by Mes-Acr⁺ for the determination of ¹³C KIEs. Both reactions utilize 20 mol % thiophenol as a co-catalyst and 25 mol % 2,6-lutidine as a basic additive. The ¹³C KIEs of 1 were measured under catalytic conditions in a standard way (Figure 1).⁸ The key result is a modest ¹³C KIE on the carbon–nucleophile (C–Nu) bond-forming carbon atom (C2) for both reactions—1.013 for hydroesterification and 1.020 for hydroamination. The KIE on a carbon atom (C1) in both reactions is near unity—1.003 for hydroesterification and 1.005 hydroamination. The qualitative interpretation of these results is that nucleophilic attack is the first irreversible step and has significant rate-limiting influence in both hydroesterification and hydroamination reactions for anethole. The absence of a significant normal KIE on C1 suggests that HAT has a lower barrier than C–Nu bond formation and has no rate-limiting influence.

Figure 1.

Experimental ¹³C KIEs for three intermolecular hydrofunctionalization reactions of anethole determined via recovered starting material analysis. Key olefinic KIEs are highlighted in orange, and standard deviation in the last digit of the measurement is indicated in parentheses. Also shown are the transition states for nucleophilic attack on the radical cation for each reaction along with predicted ¹³C KIEs at the olefinic carbon atoms for comparison with the experimental values.

For the quantitative interpretation of experimental KIEs, we modeled transition structures for both C–Nu bond formation and HAT steps using density functional theory (DFT) calculations. Geometry optimizations were carried out at three different levels of theory. We observed little functional and basis set dependence on the geometries and therefore chose M062X/6-31 + G(d)^13,14 based on efficiency and precedent in the literature.¹⁵ For the energies, we benchmarked five levels of theory against experimental redox potentials of the system and identified ωB97X-D/aug-cc-pVTZ^16–18 as the most suitable level of theory. An integral equation formalism variant polarizable continuum model (IEFPCM) implicit solvation model^19,20 was used for all computations to account for solvent effects. Predicted ¹³C KIEs were obtained from the scaled vibrational frequencies, and a Wigner tunneling correction was applied.^21–23 A thorough potential energy scan was performed to explore various binding modes for the nucleophiles, lutidine, thiol, and the catalyst. Alternate pathways such as the formation of a thioether adduct followed by S_N2 displacement by the nucleophile were also investigated.²⁴

The lowest energy transition structures for C–Nu bond formation for both nucleophiles involve attack of the lutidinium salt of the nucleophile on the radical cation of 1 (TS-Nu_2a and TS-Nu_2b). Intrinsic reaction coordinate (IRC) calculations confirm that deprotonation of 2a/2b by lutidine occurs prior to the nucleophilic attack. Free energy barriers ΔG^‡) for the nucleophilic addition TSs are 12.9 and 14.8 kcal/mol for TS-Nu_2a and TS-Nu_2b, respectively (barriers are relative to 1^·+ + 2a/2b). These structures are lower in energy than TSs calculated with no lutidinium counterion present: TS-Nu_2a and TS-Nu_2b are lower in energy by 9.7 and 3.1 kcal/mol, respectively, than the corresponding TSs without the lutidinium counterion (see Supporting Information Figure S14). The lowest energy transition structures for the HAT step from thiophenol to the carbon-centered radical are significantly lower in energy than nucleophilic addition for both hydrofunctionalizations ΔG^‡ for TS-HAT_2a is 6.9, and TS-HAT_2b is 10.3 kcal/mol, Figure 2). The lower energy barrier combined with a nominal experimental KIE at C1 indicates that HAT is a facile process. This is consistent with the qualitative interpretation of experimental ¹³C KIEs, indicating that C–Nu bond formation is likely the rate-determining step in both reactions. If this interpretation is accurate, the predicted KIEs at C1 and C2 for TS-Nu_2a and TS-Nu_2b should agree with the experimental KIEs determined in the hydroesterification and hydroamination reaction, respectively. However, analysis of the predicted KIEs (Figure 1) reveal values at C2 that are significantly larger than the experimental ones. In the hydroesterification reaction, the experimental KIE at C1 and C2 represents ~40% expression ((experimental KIE/predicted KIE)*100) of the predicted KIEs for TS-Nu_2a. The discrepancy is slightly less pronounced in the hydroamination reaction, where the experimental KIE at C1 and C2 corresponds to ~60% expression of the predicted KIEs for TS-Nu_2b.

Figure 2.

Reaction coordinate diagram depicting the relative barriers of alkene oxidation, nucleophilic attack, and HAT steps for all three hydrofunctionalization reactions—hydroesterification (blue), hydroamination (purple), and hydroetherification (black). SET is co-rate-limiting in the hydroesterification and hydroamination reactions. Carbon-nucleophile bond formation is solely rate-liming for intermolecular hydroetherification. The barrier for TS-SET is an average of the estimated SET barriers for hydroesterification and hydroamination as determined from the relative contribution of TS-SET and TS-Nu_2a/TS-Nu_2b to the experimental KIEs in the respective reactions.

A possible explanation for this mismatch between experimental and predicted KIEs at C2 could be that a step prior to nucleophilic attack has partial rate-limiting influence in these reactions. In other words, electron transfer from 1 to Mes-Acr⁺* to form the radical cation of 1 is kinetically competitive with C–Nu bond formation.²⁵ A similar discrepancy in experimental and theoretical KIEs was recently observed by Singleton in the photoredox-promoted [2 + 2] cycloaddition of enones.¹⁰ In such a scenario, the experimental KIEs would be a weighted average of the predicted KIEs for SET and the C–Nu bond-forming step, based on the relative free energies of these steps. Predicted KIEs for electron transfer to form 1^·+ are near unity (1.003) for both C1 and C2; therefore, a weighted average that includes partial rate-limiting electron transfer will have the effect of lowering the KIE for C2.

Our results thus far are consistent²⁶ with a partially rate-limiting mechanistic scenario—the higher barrier for TS-Nu_2b (14.8 kcal/mol) compared to TS-Nu_2a (12.9 kcal/mol) makes nucleophilic attack more rate-limiting for hydroamination as compared to hydroesterification, consistent with the increased KIE expression for hydroamination (~60%) relative to hydroesterification (~40%). To further validate this interpretation of our experimental KIEs, we decided to determine ¹³C KIEs for the intermolecular hydroetherification of 1 with methanol (2c) as the nucleophile. We hypothesize that since 2c is an inferior nucleophile compared to deprotonated 2a/2b, C–Nu bond formation would become solely rate-limiting, resulting in full expression of experimental KIEs. Nucleophilic addition of methanol was originally reported by Nicewicz as a sole example of intermolecular hydroetherification in the seminal paper described earlier.² We made one modification to the originally reported procedure by adding 25 mol % 2,6-lutidine to this reaction to prevent the formation of thiol-ene side products as well as to maintain consistency with the experimental conditions used for hydroesterification and hydroamination.²⁷

Gratifyingly, ¹³C KIEs of 1.028 and 1.006 were observed at C2 and C1, respectively, for the reaction of 1 with 2c—values that correspond to >90% expression of the ¹³C KIE predictions for TS-Nu_2c, the transition structure for lutidine-activated nucleophilic attack of methanol on 1^•+ (Figure 1, ΔG^‡ = 21.1 kcal/mol). Unlike 2a/2b, where deprotonation by lutidine occurs prior to the nucleophilic attack step, TS-Nu_2c is a transition structure where the two steps—deprotonation and nucleophilic attack—occur as a concerted event (confirmed by IRC calculations).

The difference in energy between TS-SET and TS-Nu_2a/2b (Figure 2) was obtained by weighting the predicted ¹³C KIE at C1 and C2 for both SET and the nucleophilic attack steps appropriately to obtain the experimental ¹³C KIE at those carbons (see Supporting Information page 3, Table C2 for details). This difference in the barriers was then used to estimate the free energy barrier for TS-SET by averaging the values obtained for both the hydroesterification and hydroamination reactions. Since alkene oxidation is a common step for all three hydrofunctionalizations, we estimated the effective redox potential for the catalyst species to be ~1.83 V based on the percent expression of the ¹³C KIEs at C2 and the calculated SET barrier. We emphasize that the calculation of this free energy surface is only possible by estimation of the effective redox potential of the photoexcited catalyst species from the expression of the KIEs, highlighting the importance of a combination of kinetic experiments and computations to probe the energy surface of these reactions.²⁸ It must be noted that SET can occur from different combinations of photoexcited states of the catalyst as long as the average redox potential is ~1.83 V (see the Supporting Information for detailed discussion). Finally, we applied Marcus theory to estimate the barrier for electron transfer from 1 to Mes-Acr*⁺ to generate the radical cation 1^•+.^29–37 The four-point approximation was used to calculate the inner sphere reorganization energies, and a modified two sphere model was applied to the outer sphere reorganization energies. From this analysis, we found that TS-SET is ~0.2 kcal/mol uphill from 1 + Mes-Acr*⁺ (~13.9 kcal/mol relative to 1^•+ + Mes-Acr●, the arbitrary zero in Figure 2). This barrier is in remarkable agreement with the experimentally obtained free energy barrier for TS-SET, further supporting our approach and therefore serving as a benchmark for future efforts in the area of computational investigation of SET in photoredox catalysis.

A composite free energy profile (Figure 2), for all three hydrofunctionalization reactions of anethole, is consistent with (a) partial rate-limiting alkene oxidation (TS-SET, ΔG^‡ ~ 13.9 kcal/mol) and nucleophilic attack for both hydroesterification (TS-Nu_2a, ΔG^‡ = 12.9 kcal/mol) and hydroamination (TS-Nu_2b, ΔG^‡ = 14.8 kcal/mol) and (b) sole rate-limiting influence of TS-Nu_2c (ΔG^‡ = 21.1 kcal/mol) in the hydroetherification reaction, as previously hypothesized. The HAT step is facile in all three hydrofunctionalization reactions—lending strong support to the qualitative interpretation of our experimental KIEs. This free energy profile presents an experimentally validated unified model that provides comprehensive insight into the rate- and selectivity-determining factors for intermolecular hydrofunctionalization reactions enabled by organophotoredox catalysis.

To further probe the energy surface of this reaction, we envisioned that a reaction involving an alkene with a lower oxidation potential or a more efficient photocatalyst could alter the rate-limiting influence of alkene oxidation, resulting in an increased expression of KIEs in the hydroesterification reaction (Figure 3). Gratifyingly, determination of ¹³C KIEs for isosafrole (1a, an alkene with .14 V lower oxidation potential than 1) in the hydroesterification reaction resulted in a ¹³C KIE at C2 of ~1.023. This corresponds to ~70% expression of the predicted KIE at C2 for 1a (predicted C2 KIE = 1.031), suggesting a reduced rate-limiting influence of TS-SET for this alkene. A second experiment involved the hydroesterification of 1 with the more robust, second-generation Nicewicz photocatalyst (t-Bu-Mes-Acr⁺).^1,38 Despite the faster rate of reactions utilizing t-Bu-Mes-Acr⁺, there is no statistical difference in the experimental KIEs (Figure 3) when compared with corresponding reactions ran with Mes-Acr⁺ (Figure 1). These results demonstrate that the free energy profile for both photocatalysts remains the same, and the faster kinetic profile observed for t-Bu-Mes-Acr⁺ must be due to the prolonged viability of the catalyst as originally hypothesized by Nicewicz.^38,39

Figure 3.

Experimental ¹³C KIEs for intermolecular hydroesterification reaction of an alkene with lower redox potential (isosafrole, top) using Mes-Acr⁺ as the catalyst and of anethole using a more sterically hindered photocatalyst (t-Bu-Mes-Acr⁺, bottom).

As additional experimental validation of the hypothesis that SET and C–C bond formation are co-rate-limiting, we measured intramolecular KIEs for the hydroesterification reaction of stilbene (1b) with 2a. Intramolecular KIEs probe the first step in a catalytic cycle that irreversibly desymmetrizes a symmetric reactant (in this case 1b).^11,12 Since oxidation of 1b to 1b^·+ does not result in desymmetrization, intramolecular KIEs determined for 1b will probe the first irreversible step after alkene oxidation (even if alkene oxidation is partially rate-limiting). Accordingly, intramolecular KIEs for 1b were determined at natural abundance using standard methodology. The experimentally measured ratio of KIEs for carbons undergoing functionalization and HAT is 1.025 (Figure 4), which is in excellent agreement with the predicted KIE ratio for the two carbons in TSNu_1b-2a (the TS analogous to TSNu_2a calculated for 1b). This experiment further validates our hypothesis that the mismatch between experimental and predicted KIEs in hydroesterification reactions is a consequence of co-rate-limiting alkene oxidation (Figure 2). In addition, we have demonstrated herein that a combination of inter- and intramolecular KIEs can be utilized to probe the mechanism of photoredox reactions.

Figure 4.

Experimental and predicted intramolecular ¹³C KIEs determined for trans-stilbene in the hydroesterification reaction with benzoic acid catalyzed by the Mes-Acr⁺ photocatalyst.

To demonstrate the importance of nucleophile activation by the base in TS-Nu, we found that an electron-rich base (2,6-dimethoxypyridine) exhibits a rate acceleration over 2,6-lutidine for both hydroetherification and hydroesterification (Figure 4, full results are available in the Supporting Information). When 2,6-lutidine was used as a base in the intermolecular hydroetherification of 1 with methanol, the reaction was 26% complete in 24 h. However, when 2,6-dimethoxypyridine was employed as the base, the reaction was 94% complete in 24 h. A rate acceleration was also observed for ethanol, which was 11-fold faster with the electron-rich base. This also allowed for the expansion of the alcohol scope to include isopropanol and benzyl alcohol, which give no product with 2,6-lutidine in 36 h but result in high NMR yields with 2,6-dimethoxypyridine of ~97 and ~93%, respectively (Figure 5). Finally, a 10-fold rate acceleration was similarly observed in the intermolecular hydroesterification of pivalic acid.

Figure 5.

Effect of basicity of the pyridine base on the rate of intermolecular hydrofunctionalization reactions of anethole.

A rate acceleration was also observed for the intramolecular hydroetherification and hydrolactonization originally published by Nicewicz (Figure 6). In the original report, the hydroetherification of 1c was performed without any added base. Addition of a catalytic amount of a base resulted in a 3–5-fold acceleration in the rate of this reaction (Figure 6A). Similarly, Nicewicz reported the intramolecular hydrolactonization of 1d using 2,6-lutidine to activate the carboxylic acid toward intramolecular nucleophilic attack on the cation radical. Once again, utilization of the more electron-rich 2,6-dimethoxypyridine resulted in a fivefold acceleration in the rate of this reaction (Figure 6B). Based on these results, we hypothesized that use of a chiral pyridine base could render the intramolecular hydroetherification of 1a enantioselective. As proof-of-principle for this hypothesis, when aminoindanol-derived pyridine oxazoline (Pyox) and pyridine bisoxazoline (Pybox) were employed as chiral bases, enantioinductions of 6% and 10% ee were observed (Figure 6C). Further studies are currently underway in our labs to develop mechanism-guided catalytic enantioselective versions of both inter- and intramolecular hydrofunctionalization reactions using chiral bases.

Figure 6.

Effect of basicity of the pyridine base on the rates of (A) intramolecular hydroetherification and (B) intramolecular hydrolactonization reaction. (C) Preliminary results indicating that chiral pyridine derivatives can induce enantioselectivity in intramolecular hydroetherification reactions (absolute configuration of 3c not established).

In conclusion, we have conducted a mechanistic evaluation of intermolecular alkene hydrofunctionalization reactions developed in the Nicewicz lab. The key finding from experimental ¹³C KIEs determined for anethole is that alkene oxidation and nucleophilic attack are kinetically competitive in the hydroesterification and hydroamination reactions, while nucleophilic attack is solely rate-limiting in the hydroetherification reaction.

The 2,6-lutidine base is likely involved in the transition state for nucleophilic attack, which is the selectivity-determining step of these reactions. The electron-rich base (2,6-dimethoxypyridine) exhibits a rate acceleration in both inter- and intramolecular hydrofunctionalization reactions. As proof-of-principle, a chiral base was found to give 10% ee in an intramolecular hydroetherification reaction. This is a rare example of experimental validation of the atomistic features of the transition state geometry of a photoredox reaction. In addition, the combination of ¹³C KIEs and theoretical calculations provides unique insight into the complexities of the mechanistic landscape involved in photoredox catalysis by simultaneously probing SET as well as steps involving bonding changes at carbon. We anticipate that our findings will lead to similar investigations into other areas of photoredox catalysis and provide a valuable complement to existing photophysical tools for mechanistic evaluation of photoredox reactions.