Mechanisms of Carbonyl Activation by BINOL N-Triflylphosphoramides Enantioselective Nazarov Cyclizations Lovie-toon, Joseph P.; Tram, Camilla Mia; Flynn, Bernard L.; Krenske, Elizabeth H.

Published in: ACS

DOI: 10.1021/acscatal.7b00292

Publication date: 2017

Document version Publisher's PDF, also known as Version of record

Citation for published version (APA): Lovie-toon, J. P., Tram, C. M., Flynn, B. L., & Krenske, E. H. (2017). Mechanisms of Carbonyl Activation by BINOL N-Triflylphosphoramides: Enantioselective Nazarov Cyclizations . ACS Catalysis, 7(5), 3466-3476. https://doi.org/10.1021/acscatal.7b00292

Download date: 09. Apr. 2020 This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.

Research Article

pubs.acs.org/acscatalysis

Mechanisms of Carbonyl Activation by BINOL N‑Triflylphosphoramides: Enantioselective Nazarov Cyclizations Joseph P. Lovie-Toon,† Camilla Mia Tram,†,‡,∥ Bernard L. Flynn,§ and Elizabeth H. Krenske*,†

† School of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane, Queensland 4072, Australia ‡ University of Copenhagen, Universitetsparken 5, DK-2100 Copenhagen Ø, Denmark § Medicinal Chemistry, Monash Institute of Pharmaceutical Sciences, Monash University, 381 Royal Parade, Parkville, Victoria 3052, Australia

*S Supporting Information

ABSTRACT: BINOL N-triflylphosphoramides are versatile organocatalysts for reactions of carbonyl compounds. Upon activation by BINOL N-triflylphosphoramides, divinyl undergo rapid and highly enantioselective (torquoselective) Nazarov cyclizations, making BINOL N-triflylphosphoramides one of the most important classes of catalysts for the Nazarov cyclization. However, the activation mechanism and the factors that determine enantioselectivity have not been established until now. Theoretical calculations with ONIOM and M06-2X are reported which examine how BINOL N-triflylphosphoramides activate divinyl ketones and control the torquoselectivity of the cyclization. Unexpectedly, the computations reveal that the traditionally accepted mechanisms for these catalysts (i.e., NH···O C hydrogen bonding or proton transfer) are not the dominant activation mechanisms. Instead, the active catalyst is a less-stable tautomer of the phosphoramide containing a P(NTf)OH group. Proton transfer from the catalyst to the substrate occurs concomitantly with ring closure. The enantioselectivities of Nazarov cyclizations of three different classes of divinyl ketones are shown to depend on a combination of factors, including catalyst distortion, the degree of proton transfer, intramolecular substrate stabilization, and intermolecular noncovalent interactions between the substrate and catalyst in the transition state, all of which relate to how well the cyclizing divinyl fits into the chiral binding pocket of the catalyst. KEYWORDS: BINOL N-triflylphosphoramide, Nazarov cyclization, Brønsted , tautomerism, density functional theory, noncovalent interactions

■ INTRODUCTION cyclization. Some of the most efficient asymmetric Nazarov 4 Chiral phosphoric derived from the enantiomers of cyclizations reported to date have involved 2 as chiral catalysts. BINOL (1, Figure 1) occupy a prominent role in the field of Scheme 1 shows three examples, which showcase the ability of N-triflylphosphoramides 2 to activate divinyl ketones at low catalyst loadings, leading to rapid and highly enantioselective − cyclizations.5 7 Indeed, the versatility and success of phosphor- 2 have significantly boosted the utility of the Nazarov cyclization as a method for the asymmetric synthesis of multistereocenter-containing cyclopentanoids.4 There have been many theoretical studies on BINOL 8,9 Figure 1. BINOL-derived phosphoric acids 1 and N-triflylphosphor- phosphoric acid (1) catalyzed reactions. The stereoselectiv- amides 2. ities of reactions catalyzed by 1 depend on a rather complex combination of covalent and noncovalent interactions which 1 organocatalysis. Their ability to activate basic substrates by occur in the transition state (TS). Noncovalent interactions hydrogen bonding or proton transfer and the ability to tune the ′ that have been found to play a role in BINOL phosphoric acid chiral binding pocket by varying the 3- and 3 -Ar substituents catalyzed reactions include steric crowding, hydrogen bonding, have led to many applications of 1 as chiral catalysts. However, 10   CH−O, π−π,CH−π, cation−π, and electrostatic interactions. certain classes of substrates notably, carbonyl compounds 9 are resistant to activation by 1. For these challenging substrates, While some general predictive rules have emerged, the exact the more acidic BINOL N-triflylphosphoramides 22 have emerged as powerful activating agents.1d,e,3 Received: January 27, 2017 One important application of BINOL N-triflylphosphor- Revised: March 22, 2017 amides 2 to carbonyl activation has been the Nazarov Published: April 17, 2017

© 2017 American Chemical Society 3466 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article

− Scheme 1. Enantioselective Nazarov Cyclizations Catalyzed by BINOL N-Triflylphosphoramides 25 7

combination of interactions that controls the selectivity in any during the conrotatory 4π electrocyclic ring closure for 6 but given example can only be understood through TS modeling. anticlockwise for 3 (see red arrows in Scheme 1). Asymmetric organocatalysis involving BINOL phosphora- Third, we examine the Nazarov cyclization of 8 (reaction C). mides 2 has recently begun to attract the attention of This reaction was developed by Tius as part of an elegant theoretical chemists. Goodman and Grayson11 studied strategy for the construction of vicinal all-carbon quaternary 7 phosphoramide-catalyzed asymmetric propargylations of alde- stereocenters. Cyclization of 8, catalyzed by 2d, gave after loss hydes by allenyl boronates. They modeled the TS as a of the stable diphenylmethyl cation the sterically congested hydrogen-bonded complex containing an NH···O hydrogen cyclopentenone (S,S)-9 in an enantiomeric ratio (er) of 98:2 bond between the catalyst and one of the boronate oxygens. (clockwise conrotation). Catalyst 2e, containing Ph groups t The differing levels of stereoselectivity provided by N- instead of BuC6H4 groups, gave lower selectivity. fl Little is known about the factors that determine the direction tri ylphosphoramide versus phosphoric acid catalysts in the fl propargylation reactions were explained by analyzing how the of conrotatory ring closure in N-tri ylphosphoramide-catalyzed Nazarov cyclizations.14 A recent paper by Tu and co-workers CF3 group altered the shape of the binding pocket. Houk and Rueping12 examined N-triflylphosphoramide-catalyzed cyclo- reported computations on a phosphoramide-catalyzed Nazarov additions of hydrazones and alkenes. They used theoretical cyclization/semipinacol rearrangement cascade which led to spiro[4.4]nonane derivatives.15 On the basis of density calculations to explore the catalyst−substrate binding modes functional theory (DFT) calculations with B3LYP, the authors and to determine whether catalysis involved hydrogen bonding suggested that steric clashing between the substrate and the from the acidic NH group to the substrate or proton transfer. catalyst controlled the stereoselectivity in that case. Equally In this paper we use theoretical calculations to investigate fl importantly, the exact mechanism of activation of divinyl how N-tri ylphosphoramides 2 activate divinyl ketones and ketones by N-triflylphosphoramides 2 is unknown. Different control enantioselectivity, in three distinct Nazarov cyclizations ··· fi authors have depicted these reactions as involving either NH (Scheme 1). The rst of our three case studies focuses on the OC hydrogen bonding or the transfer of a proton from 2 to cyclization of dihydropyranyl vinyl ketone 3 (reaction A). This the to form a chiral contact pair (Scheme 1, reaction was reported by Rueping in his original publication, 1d,e,3−7 5,13 lower right). Here, we use density functional theory which introduced 1 and 2 as Nazarov cyclization catalysts. calculations to address this question and to explain the Both 1 and 2 catalyzed the cyclization of 3, giving mixtures of enantioselectivities of the three reactions depicted in Scheme cis- and trans-cyclopentenones 4 and 5. The electrocyclization 1. Unexpectedly, the calculations reveal that catalysts 2 do not installs the chiral center at C4, favoring C4-S. Enolization of the use either of the traditionally drawn activation modes (Scheme product then gives a mixture of C5 epimers (4 and 5). 1). Instead, the active catalyst is a less-stable tautomer of 2 Phosphoramides 2 gave superior enantioselectivities in containing a P(NTf)OH group. This discovery has comparison to 1, with enantiomeric excesses (ees) as high as significant implications for the understanding and development 95%. of BINOL N-triflylphosphoramide-catalyzed reactions. Our second case study focuses on the Nazarov cyclization of divinyl ketone 6 (reaction B). This reaction was employed by ■ THEORETICAL CALCULATIONS Flynn et al. in a concise formal synthesis of the anticancer Density functional theory calculations were performed with 6 natural product (+)-roseophilin. Nazarov cyclization of 6, Gaussian 09.16 To explore the mechanism of divinyl ketone catalyzed by 2b or 2c, followed by hydrolytic trapping of the activation by phosphoramides 2, we performed computations intermediate oxonium ion gave (R,R)-7 in up to 82% ee. The on a model system consisting of phosphoramide 10 and 1,4- torquoselectivity of ring closure of 6 was opposite to that for pentadien-3-one. The computations on this system employed Rueping’s substrate 3 with the same enantiomer of catalyst. M06-2X/6-311+G(d,p)17 and modeled the solvent (chloro- That is, the termini of the divinyl ketone rotate clockwise form) with the SMD18 continuum model. Vibrational

3467 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article frequency calculations were used to characterize each stationary ■ RESULTS AND DISCUSSION fi point as a minimum or rst-order saddle point and to calculate Activation Mechanism. There are multiple ways in which thermochemical data. The reactant and product to which each an N-triflylphosphoramide can interact with a divinyl ketone. TS was directly connected were determined by means of ff 19 Potentially, the phosphoramide can adopt three di erent intrinsic reaction coordinate (IRC) calculations. tautomeric forms and each tautomer could form either a The procedure for locating TSs for the asymmetric Nazarov hydrogen-bonded complex or a chiral contact ion pair with the cyclizations of 3, 6, and 8 catalyzed by BINOL phosphoramides divinyl ketone. In order to identify the catalyst−substrate 2 commenced with the computation of TSs for the binding modes relevant to the Nazarov cyclization, we corresponding model proton-catalyzed cyclizations. Conformer examined a model system consisting of N-triflylphosphoramide searching of each proton-catalyzed TS was performed with 10 and 1,4-pentadien-3-one (Figure 2). Phosphoramide 10 has B3LYP/6-31G(d).20 Those TSs lying within 3 kcal/mol of the most stable TS were then used to build transition states for the phosphoramide-catalyzed reaction. The relevant catalyst was combined with the TS such that the acidic proton of 2 took the place of the catalytic H+ from the proton-catalyzed TS. The accessible conformations of each catalyst−TS complex were then explored by means of a conformational search in MacroModel 10.6.21 The conformer search employed the MCMM algorithm in conjunction with the OPLS_2005 force field,22 in vacuum, with the atoms of the divinyl ketone held fixed. Catalyst−TS complexes identified as lying within 2.4 kcal/ mol (10 kJ/mol) of the global minimum from each conformer search were then fully optimized as transition states in Gaussian 09 using a two-layer ONIOM23 protocol. The inner (high- level) layer consisted of the entire divinyl ketone together with the NH, SO2CF3, and PO3 groups of the catalyst and was treated with B3LYP/6-31G(d). The outer (low-level) layer consisted of the binaphthyl unit and the 3- and 3′-aryl groups of the catalyst and was treated with UFF.24 Goodman has employed a similar ONIOM(B3LYP/6-31G(d):UFF) protocol to model a variety of BINOL phosphoric acid catalyzed reactions.8b,c,9a,b The catalyst was modeled as either the P( O)NHTf or P(NTf)OH tautomer, and hydrogen-bonding and ion-pairing activation modes were computed for each case. Harmonic vibrational frequency calculations on the optimized TSs confirmed the presence of a single imaginary frequency corresponding to C−C bond formation. A single-point energy calculation was then performed for each TS at the M06-2X/6- 311+G(d,p) level of theory in the solution phase (chloroform) as modeled using the SMD implicit solvent model. Gibbs free energies in solution were calculated by adding the thermo- chemical corrections derived from the ONIOM frequency calculations to the M06-2X solution-phase potential energies and are reported at a standard state of 298.15 K and 1 mol/L. For comparison with the M06-2X energies, calculations were also performed with several other density functionals, including B3LYP-D325 and B97-D3.25,26 Details of the computations Figure 2. Computed energies of (a) the tautomers of model N- performed with these other methods are reported in the triflylphosphoramide 10, (b) a variety of catalyst−substrate complexes Supporting Information. Briefly, computations with B3LYP-D3 formed from 10 and 1,4-pentadien-3-one, and (c) transition states for and B97-D3 were found not to provide as close an agreement the Nazarov cyclization of 1,4-pentadien-3-one catalyzed by 10. The zero of energy is taken as the separated 10a and divinyl ketone. ΔH with the experimentally observed stereoselectivities as was and ΔG values (kcal/mol) and distances (Å) were computed with obtained with M06-2X. All three functionals predicted the M06-2X/6-311+G(d,p)-SMD(CHCl ). correct major enantiomer for reaction B, but some departures 3 from experiment with respect to the major enantiomer were found for reactions A and C. These differences were larger for the three tautomers 10a−c (Figure 2a). Computations with B97-D3 than for B3LYP-D3. All three functionals, however, M06-2X/6-311+G(d,p) in SMD continuum solvent (chloro- were in general agreement with respect to the identity of the form) predict that the most stable tautomer is 10a, which has a active catalyst, predicting that transition states involving the P(O)NHTf group. Tautomer 10b,containingaP( P(NTf)OH tautomer of the phosphoramide were usually NTf)OH group, is 3.9 kcal/mol higher in energy than 10a lower in energy than those involving the P(O)NHTf (ΔG), while tautomer 10c, containing a S(NR)OH group, is tautomer (vide infra). Molecular structures were drawn with 16.4 kcal/mol higher in energy. The X-ray crystal structure of 27 28 CYLview. an H8-BINOL phosphoramide, reported by Rueping, showed

3468 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article

a Scheme 2. Low-Energy Dimerization-Based Mechanism for the Tautomerization Reaction 10a → 10b

aDistances are given in Å and ΔH and ΔG values in kcal/mol. that the phosphoramide existed as the P(O)NHTf tautomer pentadienyl cation characterand not merely polarization to a in the solid state, where it formed a hydrogen-bonded dimer. δ+ stateto be activated to undergo electrocyclization. Six possible catalyst−substrate binding modes for 10 and the Unexpectedly, the TS for the electrocyclization catalyzed by model divinyl ketone are shown in Figure 2b. In these phosphoramide tautomer 10a (TS1) is 3.1 kcal/mol higher in complexes, the divinyl ketone was modeled in the U-shaped (s- energy than that for the reaction catalyzed by the less-stable trans/s-trans) conformation, which is the conformation closest tautomer 10b (TS2). Both barriers are calculated with respect in geometry to that of the transition state for electrocyclization. to the divinyl ketone plus 10a. This result challenges the The three catalyst−substrate complexes 11a−c are derived traditional understanding of phosphoramide-catalyzed Nazarov − from phosphoramide tautomers 10a−c, respectively, and are cyclizations. Quite reasonably, previous studies1d,e,3 7 have bound by a between the acidic proton of 10 represented the active catalyst as the P(O)NHTf tautomer, and the carbonyl oxygen of the divinyl ketone. Transfer of the which is the major tautomer present at equilibrium. However, acidic proton to the carbonyl oxygen in these three complexes TS1 and TS2 indicate that the activation via OH···OC leads to ion pairs 11d−f. Exploration of the potential energy hydrogen bonding or ion pairing with the P(NTf)OH surface of 11 located the four complexes 11a,b,e,f. Complexes tautomer cannot be discounted and is actually preferred. This 11c,d could not be located and instead underwent proton finding reflects the higher acidity of the P(NTf)OH group of transfer upon attempted geometry optimization. Complex 11e 10b in comparison to the P(O)NHTf group of 10a. Proton is formally an ion pair, but the ionic interaction is very short, transfer from 10b is easier because it requires less stretching of − and the pair of complexes 11b/11e together constitute an the O H bond (rO−H = 1.53 Å in TS2 vs rN−H = 1.65 Å in example of low-barrier hydrogen bonding. Complexes 11a,b,e TS1).29,30 are the most stable and have similar energies (ΔG = −0.3 to 0.2 Since the lower-energy transition state contains the higher- kcal/mol relative to the divinyl ketone plus phosphoramide). energy catalyst tautomer, there is the question of whether the The enthalpies of formation of complexes 11a,b,e are favorable tautomerization (10a → 10b) would be fast enough to allow by about 8−10 kcal/mol, but the entropic penalty leads to ΔG electrocyclization to proceed predominantly through TS2 values which are close to 0. It is likely that a mixture of these under the conditions of the reaction. Tautomerization via a three complexes would exist in equilibrium with the divinyl four-centered intramolecular proton transfer is calculated to ⧧ ketone and phosphoramide. Complex 11f is high in energy (8.0 have a high barrier (ΔG = 38.1 kcal/mol, see the Supporting kcal/mol) and is unlikely to be present in significant quantities. Information) and is unlikely to occur under the experimental Even though the most stable tautomer of the catalyst (10a)is conditions. A more likely, low-energy pathway for the favored by ≥3.9 kcal/mol relative to the other two, it does not tautomerization is shown in Scheme 2. This pathway form the most stable complexes with the divinyl ketone. The commences with two molecules of 10a combining to form complexes involving the P(NTf)OH tautomer 10b are dimer 12a. Double proton transfer through an eight-membered slightly (0.2−0.5 kcal/mol) more stable than those involving transition state (TS3) leads to the product dimer 12b, which 10a. A related observation was reported by Houk and Rueping then dissociates to give two molecules of 10b. The proton  Δ ⧧ in their study of the binding of a hydrazonium cation (R2C transfer is facile (TS3, G = 3.5 kcal/mol relative to dimer NH−NHR)+ to an N-triflylphosphoramide anion.11 The cation 12a and −3.0 kcal/mol relative to phosphoramide 10a) and the formed several ion pairs with the phosphoramide anion, in overall ΔG value of the reaction (2 × 10a → 2 × 10b) is 7.7 which the cation interacted with the PO, SO, and/or P kcal/mol. Provided that the dimer−monomer equilibria are also N groups. These ion pairs had energies similar to that of a facile, the equilibration of 10a and 10b would be much faster hydrogen-bonded complex of the neutral species. than electrocyclization, which has ΔG⧧ ≥ 22.6 kcal/mol. Thus, Transition states for the Nazarov cyclization of 1,4- a Curtin−Hammett scenario would be established and the pentadien-3-one catalyzed by the two low-energy tautomers electrocyclization would proceed primarily via TS2. That is, of 10 are shown in Figure 2c. Intrinsic reaction coordinate 10b is indeed the active catalyst.31 calculations indicated that the acidic proton is transferred from For BINOL N-triflylphosphoramides 2, the P(O)NHTf the catalyst to the substrate during the electrocyclization. That group is located deep within a pocket lined by the binaphthyl ′ is, the electrocyclization reaction commences with the hydro- and 3,3 -Ar groups. The reported X-ray structure of an H8- gen-bonded complex of the divinyl ketone with the neutral BINOL phosphoramide28 showed that these groups do not phosphoramide and leads to an ion pair of the hydroxyallyl inhibit the formation of a dimer akin to 12a. It is therefore cation and phosphoramide anion. By the stage the TS is likelythatBINOLN-triflylphosphoramides 2 undergo reached, proton transfer is almost complete (rOH = 1.01 Å). tautomerization through a dimerization-based mechanism in Transition states containing a hydrogen-bonded activation solution. Furthermore, the dimer−monomer equilibria must be mode could not be located for this reaction. This suggests that rapid, because if they were not, the phosphoramide would it is necessary for the divinyl ketone to assume complete become trapped as the dimer (cf. ΔG = −6.5 kcal/mol for 2 ×

3469 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article

Figure 3. Transition states for the Nazarov cyclization of dihydropyranyl vinyl ketone 13 (highlighted in light blue) catalyzed by 2c, leading to (S)- 14 or (R)-14. Two views of each TS are shown. Also shown is the lowest-energy TS for the H+-catalyzed Nazarov cyclization of 13 (TS13). Distances are given in Å and energies in kcal/mol.

10a → 12a), thereby preventing the acidic proton from being with an ONIOM protocol (B3LYP/6-31G(d):UFF) similar to available to activate the divinyl ketone. that used by Goodman for many BINOL phosphoric acid Reaction A. Our first case study of BINOL phosphoramide catalyzed reactions.8b,c,9a,b After optimization with ONIOM, controlled enantioselectivity focuses on the Nazarov cyclization single-point energies were calculated with M06-2X/6-311+G- of Rueping’s5 dihydropyranyl vinyl ketone 3 (Scheme 1, (d,p) in SMD chloroform in order to provide a more complete reaction A). We modeled Rueping’s dihydropyranyl vinyl description of quantum mechanical dispersion and solvent ketone as the dimethyl derivative 13 (Figure 3). Our study effects. On the basis of the results of our model study (Figure commenced with computations on the transition state for the 2), we computed transition states derived from both the P( proton-catalyzed cyclization of 13. At the B3LYP/6-31G(d) O)NHTf and P(NTf)OH tautomers of 2c and we explored level of theory, the lowest-energy TS adopts the conformation both hydrogen-bonding and ion-pairing activation modes. shown in Figure 3 (TS13). The OH proton points toward the An extensive conformational search was performed, involving dihydropyranyl oxygen, where it can engage in a stabilizing a total of 106 ONIOM optimizations. Of these, only 4 TS electrostatic interaction. This OH−O interaction provides conformers were found to contribute significantly to the about 3 kcal/mol of stabilization to the transition state, in reaction (≥1% of the population in a Boltzmann distribution). comparison to other TS conformers where the interaction is Figure 3 shows the lowest-energy TSs for clockwise and absent. anticlockwise conrotation. The calculations correctly predict Transition-state modeling of the corresponding phosphor- that anticlockwise conrotation (TS13S) is favored relative to -catalyzed electrocyclization was performed using the clockwise conrotation (TS13R). Experimentally, the Nazarov catalyst 2c. The 3- and 3′-Ar substituents in 2c are 9- cyclizations of 3 with catalysts 2a,b favored anticlockwise anthracenyl groups. Catalyst 2c was not reported in Rueping’s conrotation, giving the C4-S products in 89−95% ee.5 Theory original publication, but the reactions studied therein with predicts that the energy difference between TS13S and TS13R either 2a (Ar = 1-naphthyl) or 2b (Ar = 9-phenanthryl) (ΔΔG⧧) is 1.4 kcal/mol. This value corresponds to a predicted displayed consistently high selectivities across a range of enantiomeric excess of 87% at the experimental temperature of substrates, suggesting that the nature of the 3,3′-Ar groups has 0 °C. A Boltzmann analysis of all calculated TSs predicts an ee only a small effect on the enantioselectivity of this cyclization. of 84%. These values are in good agreement with the reported We chose to use 2c (with symmetrical Ar groups) in order to enantioselectivities. reduce the conformational space needing to be sampled. For comparison, the geometries of selected TSs were Geometry optimizations of the transition states were performed reoptimized with a fully quantum mechanical treatment using

3470 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article

Scheme 3. Analysis of the Enantioselectivity of Nazarov Cyclization of 13 Catalyzed by 2c Using the Distortion/Interaction a Model and Computations on Structural Fragments of the Transition States

aEnergies are given in kcal/mol.

B3LYP-D3/6-31G(d), using Grimme’s D3 correction25 to The total difference in energy between TS13S and TS13R ⧧ capture intra- and intermolecular dispersive interactions. (ΔΔE ) is 1.3 kcal/mol. The difference in energy between the Structural differences between the DFT-optimized and divinyl ketone fragments (13′) in the two TSs is small: 0.8 ONIOM-optimized TSs were mostly small, although the kcal/mol. The OH proton is syn to the dihydropyranyl oxygen DFT-optimized TSs displayed more advanced proton transfer in both TSs, similar to the case for TS13. In contrast, the ′ ff (see the Supporting Information). distorted catalysts (2c )di er by 6.7 kcal/mol, with the catalyst 8,32 R We used the distortion/interaction model to analyze the fragment from TS13 being more distorted than that from S origins of enantioselectivity. In this model, the activation barrier TS13 . Further computations on substructures A and B of the (ΔE⧧) of the Nazarov cyclization is the sum of three catalyst allow the location of this distortion to be pinpointed more precisely. Thus, fragments A, which represent the components: (i) the energy required to distort the divinyl ff ketone from its ground-state geometry to its transition-state aromatic moiety of the catalyst in the two TSs, di er by only geometry (ΔE (DVK)), (ii) the energy required to distort the 0.5 kcal/mol, whereas fragments B, representing the N- dist triflylphosphoramide moiety, differ by a larger 4.8 kcal/mol, catalyst from ground-state to transition-state geometry with TS13R being more distorted. This distortion can be traced (ΔE (Cat)), and (iii) the interaction energy between the dist to the P−O−H bond angle. In order to bind the substrate, the distorted substrate and catalyst in the TS (ΔE ). The results of int P−O−H angle of TS13R has widened to 136° (Figure 3), the distortion/interaction analysis are shown in Scheme 3. The whereas TS13S has a P−O−H angle of 120°, closer to ideal. first structure in Scheme 3 shows how the distortion and ff R ff This di erence suggests that, in the R transition state, the interaction energies in the disfavored TS13 di er with respect cyclizing divinyl ketone is a less ideal fit for the binding pocket S to those in the favored TS13 . The remainder of Scheme 3 of the catalyst. In order to activate the divinyl ketone to cyclize shows a series of structural fragments of the TSs, whose to the R product within the binding pocket, the POH group has distortion energies were computed to help analyze the TS to deviate significantly from its ideal angle. energies. The distortion/interaction analysis is based on M06- Considered together, the distortions of the catalyst and R Δ 2X/6-311+G(d,p)-SMD(CHCl3) single-point energy calcula- substrate in the disfavored TS13 amount to a total Edist that tions. The structural fragments 13, 2c′, and A−E were obtained is 5.9 kcal/mol larger than that in the favored TS13S fi ΔΔ ff by deleting certain portions of the TSs and lling any free ( Edist(total)). However, the total energy di erence between valences with hydrogen atoms. the TSs (ΔΔE⧧) is only 1.3 kcal/mol. This indicates that in

3471 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article

Figure 4. Transition states for the Nazarov cyclization of divinyl ketone 15 (highlighted in pink) catalyzed by 2c, leading to (R,R)-16 or (S,S)-16. Two views of each TS are shown. Also shown is the lowest-energy TS for the H+-catalyzed Nazarov cyclization of 15 (TS15). The distortion/ interaction analysis in the center of the figure shows how the catalyst distortion energy, substrate distortion energy, and TS interaction energy in TS15S differ from those TS15R. Distances are given in Å and energies in kcal/mol.

TS13R there are stronger stabilizing interactions between the where the N-triflylphosphoramide moiety of the catalyst has substrate and catalyst, which reduce the difference between the been removed, indicate that the divinyl ketone interacts more ΔΔ fl energies of the TSs by 4.6 kcal/mol ( Eint). Insights into the strongly with the N-tri ylphosphoramide moiety in the favored specific interactions involved are provided by the calculations TS13S than in TS13R. Short CH−F (2.34 Å) and CH−O on the lower set of fragments in Scheme 3.33 When the lower (2.79 Å) interactions and a stabilizing electrostatic interaction anthracenyl group of the catalyst is removed (fragment C), the between the CF3 group and the cyclizing pentadienyl cation remaining portion of TS13R is 3.5 kcal/mol higher in energy (3.22 Å) are found in TS13S. than the corresponding fragment of TS13S. This indicates that Consistent with the results reported above for the model the lower anthracenyl group interacts more strongly with the system (Figure 2), all of the low-energy TSs for the cyclization divinyl ketone in TS13R than in TS13S. Figure 3 (red arrows) of 13 involve the P(NTf)OH tautomer of catalyst 2c. indicates that in TS13R there is a strong cationic CH−π Transition states containing the P(O)NHTf tautomer are at interaction (2.92 Å) between the proton at C3 of the least 4.2 kcal/mol higher in energy than those discussed here. pentadienyl cation and the central aromatic ring of the lower The energy difference is smaller when the TSs are reoptimized anthracenyl group, which stabilizes the TS. The closest similar with B3LYP-D3 (2.2 kcal/mol), but the OH tautomer is still interaction in TS13S is longer (3.01 Å) and involves a less more active. Indeed, a correct prediction of the stereoselectivity acidic proton (a CH2 proton from the dihydropyranyl ring), of this reaction requires that the OH tautomer be included in consequently providing less stabilization. the model. A Boltzmann analysis based on only the TSs derived When the upper anthracenyl group of the catalyst is removed from the P(O)NHTf catalyst predicts that the cyclization (D), an opposite result is obtained: the remaining portion of would favor the R product with 40% ee, opposite from the TS13R is 0.7 kcal/mol lower in energy than that of TS13S. experimental result. This indicates that the upper anthracenyl group interacts more Reaction B. Our second case study examines the Nazarov strongly with the divinyl ketone in the favored TS13S. This can cyclization of divinyl ketone 6, employed by Flynn et al.6 in be traced to a weak CH−π interaction (3.67 Å) between the their roseophilin synthesis (Scheme 1, reaction B). Divinyl divinyl ketone and upper anthracenyl group in TS13S (in ketone 6 was modeled as 15 (Figure 4). First, transition-state comparison, in TS13R the substrate lies ≥4 Å from this modeling of the proton-catalyzed cyclization of 15 revealed an anthracenyl group). Finally, computations on structure E, interesting stabilizing feature, not previously observed in the

3472 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article

Figure 5. Transition states for Nazarov cyclization of divinyl ketone 8 (highlighted in gold) catalyzed by 2d, leading to (S,S)-17 or (R,R)-17. Two views of each TS are shown. Also shown is the lowest-energy TS for the H+-catalyzed Nazarov cyclization of 8 (TS8). The distortion/interaction analysis in the center of the figure shows how the catalyst distortion energy, substrate distortion energy, and TS interaction energy in TS8R differ from those in TS8S. Distances are given in Å and energies in kcal/mol.

Nazarov cyclization.34 In the most stable TS (TS15), the a Boltzmann analysis of all TSs is also 71%, almost identical ketone carbonyl group is located above the cyclizing with the value obtained experimentally for 6. pentadienyl cation, 3.06 Å from the center of the ring and A distortion/interaction analysis of the two TSs is shown in ff ΔΔ ⧧ 2.61 Å from the proton on the terminus of the pentadienyl the center of Figure 4. The energy di erence ( E ) between − cation. The resulting cation−lone pair and CH−O interactions the TSs is slightly negative ( 0.4 kcal/mol), i.e. opposite in ΔΔ ⧧ ff contribute about 3 kcal/mol of stabilization to the TS. sign to the value of G , indicating that entropic e ects are To model the cyclization of 15 catalyzed by BINOL N- important for the calculated enantioselectivity of this reaction. triflylphosphoramide 2c, ONIOM optimizations were per- Nevertheless, the distortion/interaction analysis reveals two significant contributions to the TS energies. First, the catalyst is formed on a total of 127 transition state conformers. Of these, 3 S fi much more distorted in the disfavored TS15 than in favored were found to contribute signi cantly to the reaction. The R ΔΔ TS15 ( Edist(Cat) = 37.1 kcal/mol), but this penalty is lowest-energy TS for each direction of conrotation is shown in offset by a much stronger catalyst−substrate interaction Figure 4 (TS15R and TS15S). Experimentally, in the ΔΔ − ( Eint = 42.5 kcal/mol). These large values can be traced cyclization of 6 catalyzed by 2c, the R,R product is preferred, to the different degrees of proton transfer in the two TSs. In in 72% ee (room temperature). The calculations correctly TS15R, the breaking O−H bond of the catalyst has stretched to match this outcome, predicting that TS15R is 1.0 kcal/mol 1.66 Å, in comparison with 1.36 Å in TS15S, explaining the lower in energy than TS15S. The ee predicted on the basis of much higher distortion energy in TS15R. The large difference this value of ΔΔG⧧ is 71%, and the ΔΔG⧧ value on the basis of in interaction energies reflects the strong ion pairing interaction

3473 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article S R in TS15 in comparison to the weaker hydrogen bond found in divinyl ketone CO2Ph group in TS8 (2.74 Å). Experimentally, TS15R. the tBu groups of the catalyst were found to be important for Second, the divinyl ketone distortion in the disfavored obtaining high levels of enantioselectivity (Scheme 1). This TS15S is 5.0 kcal/mol larger than that in the favored TS15R.In appears to be mainly due to the steric bulk of the tBu groups TS15R, the substrate has a conformation similar to that of the creating a tighter binding pocket rather than engaging in any proton-catalyzed TS (TS15), with the ketone oxygen specific interactions that would stabilize the TS leading to the positioned near the cyclizing pentadienyl cation, where it can major product. In this reaction, transition states derived from − − engage in stabilizing CH O and cation lone pair interactions. the P(O)NHTf tautomer of the catalyst lie ≥15 kcal/mol S ff In TS15 , the side chain has a di erent conformation which higher in energy than the P(NTf)OH transition states and places the ketone oxygen farther away from the CH proton are predicted not to contribute significantly to the reaction. (3.42 Å in comparison to 2.67 Å; see red arrows in Figure 4). R fi −π Considered together, a multitude of covalent and non- Furthermore, TS15 also bene ts from a stabilizing CH covalent interactions are present in the transition states of interaction between the CH proton of the pentadienyl cation BINOL N-triflylphosphoramide catalyzed Nazarov cyclizations. and the lower anthracenyl group (2.68 Å). Intermolecular CH−π, cation−π,CH−O, CH−F, and cation− In this reaction, the preferred TSs are derived from the P( NTf)OH tautomer of the catalyst. The P(O)NHTf tautomer lone pair interactions play a role in the three examples is predicted to play a small role, most notably in the formation discussed here. In these three cases, the degree of proton of the major product. Thus, while TS15R leads to 77% of the transfer from the catalyst to the substrate is an important total product according to a Boltzmann analysis, a further 7% of stereodiscriminating factor, which depends on how well the fi the product comes from an ion-paired R-configured NH substrate ts into the binding pocket and is manifested in both transition state which lies 1.4 kcal/mol above TS15R (see the the catalyst distortion energy and the interaction energy. We Supporting Information). have also observed that, in the favored TS for each reaction, the Reaction C. Our final case study involves the Nazarov divinyl ketone has a conformation similar to that of the TS for cyclization of 8 (Scheme 1, reaction C), developed by Tius for the corresponding proton-catalyzed cyclization. In the dis- the construction of vicinal quaternary stereocenters.7 The favored TS, the divinyl ketone sometimes departs from this substrate was modeled in its entirety without truncation. The conformation and loses the benefit of intramolecular stabilizing preferred conformation of the TS for the H+-catalyzed interactions such as the ketone−pentadienyl cation interaction cyclization of 8 is shown in Figure 5 (TS8). Experimentally, found in reaction B. + the product (9) is isolated after loss of the CHPh2 cation. An + IRC calculation indicates that the loss of CHPh2 occurs in a ■ CONCLUSIONS separate step, after the cyclization. Modeling of the transition states for the 2d-catalyzed We have used theoretical calculations to explore the mechanism fl Nazarov cyclization of 8 considered a total of 90 TS and enantioselectivities of BINOL N-tri ylphosphoramide- conformers. Three conformers were found to contribute >1% catalyzed Nazarov cyclizations. The calculations have revealed to the total population in a Boltzmann distribution. The lowest- the unexpected discovery that it is the P(NTf)OH tautomer energy TSs for each conrotatory mode are shown in Figure 5. of the phosphoramide, and not the more stable P(O)NHTf The calculations predict that clockwise conrotation leading to tautomer, that is the most active catalyst in these reactions. the S,S product is favored by 1.2 kcal/mol. This value Interconversion between tautomers is calculated to be fast, corresponds to an er of 89:11, while a Boltzmann analysis of relative to electrocyclization. This discovery challenges the all TSs predicts an er of 84:16. These values are slightly lower previously accepted models for carbonyl activation by N- than the experimental er (98:2), but they correctly predict the triflylphosphoramides, which proposed that the P(O)NHTf configuration of the major product (and have been computed catalyst activates the substrate through either NH···OC − without including any special treatment of the contribution of hydrogen bonding or proton transfer.1d,e,3 7 Houk and low-frequency vibrational modes to the vibrational entropy). Rueping, in their recent study of phosphoramide-catalyzed  Inclusion of the P( NTf)OH tautomer in the model is hydrazone cycloadditions,12 showed that the cationic substrate essential for the correct prediction of stereoselectivity. A interacted with the catalyst PO and SO groups in the TS, Boltzmann analysis based solely on the TSs derived from the  but not with the atom. In that case, however, proton P( O)NHTf tautomer predicts almost 100% selectivity in transfer from the catalyst preceded cycloaddition in a separate favor of the R,R product, opposite from experiment.35 chemical step. This meant that the stereoselectivity did not A distortion/interaction analysis of the two TSs (Figure 5, ⧧ depend on which tautomer of the catalyst protonated the center) indicates that, similar to reaction B (Figure 4), the ΔE substrate. The Nazarov cyclizations considered here represent a value for the favored TS in reaction C is higher than that for the ff disfavored TS (ΔΔE⧧ = −2.3 kcal/mol); that is, entropic effects fundamentally di erent scenario, where proton transfer occurs concomitantly with electrocyclization in the rate- and stereo- are important for the observed enantioselectivity. The favored ff TS8S has a larger catalyst distortion energy than the disfavored selectivity-determining step. In this scenario, the di erent R ΔΔ − tautomers of the phosphoramide react via different mechanistic TS8 ( Edist(Cat) = 17.5 kcal/mol) and a more stabilizing ΔΔ fl pathways. It is likely that P(NTf)OH tautomers of BINOL interaction energy ( Eint = 14.9 kcal/mol), re ecting the more advanced proton transfer from catalyst to substrate in N-triflylphosphoramides are also important even in reactions TS8S. The most important noncovalent interactions between that only involve hydrogen bonding, not proton transfer from the catalyst and the substrate are a CH−π interaction between a the phosphoramide to the substrate. These discoveries, which t S  divinyl ketone Me group and the lower BuC6H4 group in TS8 expose the importance of P( NTf)OH tautomerism, will (3.18 Å, see red arrow in Figure 5) and a CH−π interaction inform the future understanding, modeling, and development t fl between the lower BuC6H4 group of the catalyst and the of N-tri ylphosphoramide-catalyzed reactions.

3474 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article ■ ASSOCIATED CONTENT S.; Gridnev, I. D.; Terada, M. Chem. Sci. 2014, 5, 3515−3523. (o) Calleja, J.; Gonzalez-Pé rez,́ A. B.; de Lera, Á. R.; Álvarez, R.; *S Supporting Information Fañanas,́ F. J.; Rodríguez, F. Chem. Sci. 2014, 5, 996−1007. The Supporting Information is available free of charge on the (9) For reviews, see: (a) Simon,́ L.; Goodman, J. M. J. Org. Chem. ACS Publications website at DOI: 10.1021/acscatal.7b00292. 2011, 76, 1775−1788. (b) Reid, J. P.; Simon,́ L.; Goodman, J. M. Acc. Computational data including calculations with different Chem. Res. 2016, 49, 1029−1041. (c) Sunoj, R. B. Acc. Chem. Res. − DFT methods (PDF) 2016, 49, 1019 1028. (10) For reviews of noncovalent interactions in transition states and their roles in stereoselectivity, see: (a) Krenske, E. H.; Houk, K. N. Acc. ■ AUTHOR INFORMATION Chem. Res. 2013, 46, 979−989. (b) Wheeler, S. E.; Seguin, T. J.; Guan, Corresponding Author Y.; Doney, A. C. Acc. Chem. Res. 2016, 49, 1061−1069. (c) Walden, D. *E-mail for E.H.K.: [email protected]. M.; Ogba, O. M.; Johnston, R. C.; Cheong, P. H.-Y. Acc. Chem. Res. 2016, 49, 1279−1291. ORCID (11) Grayson, M. N.; Goodman, J. M. J. Am. Chem. Soc. 2013, 135, Camilla Mia Tram: 0000-0003-4503-9155 6142−6148. Elizabeth H. Krenske: 0000-0003-1911-0501 (12) Hong, X.; Basp̧ınar Kücü̧k, H.; Sudan Maji, M.; Yang, Y.-F.; Present Address Rueping, M.; Houk, K. N. J. Am. Chem. Soc. 2014, 136, 13769−13780. ∥ University of Copenhagen, Universitetsparken 5, DK-2100 (13) For further applications of these phosphoramide-catalyzed Copenhagen Ø, Denmark. Nazarov cyclizations, see: (a) Rueping, M.; Ieawsuwan, W. Adv. Synth. Catal. 2009, 351,78−84. (b) Rueping, M.; Ieawsuwan, W. Chem. Notes Commun. 2011, 47, 11450−11452. (c) Raja, S.; Ieawsuwan, W.; The authors declare no competing financial interest. Korotkov, V.; Rueping, M. Chem. - Asian J. 2012, 7, 2361−2366. (14) For theoretical studies of other catalytic asymmetric Nazarov ■ ACKNOWLEDGMENTS cyclizations, see: (a) Kitamura, K.; Shimada, N.; Stewart, C.; Atesin, A. ̧ 2015 − We thank the Australian Research Council for financial support C.; Atesin, T. A.; Tius, M. A. Angew. Chem., Int. Ed. , 54, 6288 6291. (b) Asari, A. H.; Lam, Y.-h.; Tius, M. A.; Houk, K. N. J. Am. (FT120100632 to E.H.K. and DP150103131 to B.L.F. and Chem. Soc. 2015, 137, 13191−13199. (c) Raja, S.; Nakajima, M.; E.H.K.) and the Australian National Computational Infra- Rueping, M. Angew. Chem., Int. Ed. 2015, 54, 2762−2765. structure National Facility and University of Queensland (15) Yang, B.-M.; Cai, P.-J.; Tu, Y.-Q.; Yu, Z.-X.; Chen, Z.-M.; Wang, Research Computing Centre for computer time. S.-H.; Wang, S.-H.; Zhang, F.-M. J. Am. Chem. Soc. 2015, 137, 8344− 8347. ■ REFERENCES (16) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; (1) For reviews, see: (a) Akiyama, T. Chem. Rev. 2007, 107, 5744− Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, 5758. (b) Kampen, D.; Reisinger, C. M.; List, B. Top. Curr. Chem. B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. 2009, 291, 395−456. (c) Terada, M. Synthesis 2010, 2010, 1929− P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; 1982. (d) Rueping, M.; Nachtsheim, B. J.; Ieawsuwan, W.; Atodiresei, Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, I. Angew. Chem., Int. Ed. 2011, 50, 6706−6720. (e) Parmar, D.; T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Sugiono, E.; Raja, S.; Rueping, M. Chem. Rev. 2014, 114, 9047−9153. Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, (2) Nakashima, D.; Yamamoto, H. J. Am. Chem. Soc. 2006, 128, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; 9626−9627. Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, (3) Rueping, M.; Kuenkel, A.; Atodiresei, I. Chem. Soc. Rev. 2011, 40, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; 4539−4549. Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; (4) For reviews of catalytic asymmetric Nazarov cyclizations, see: Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; − Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Shimada, N.; Stewart, C.; Tius, M. A. Tetrahedron 2011, 67, 5851 ̈ 5870. Vaidya, T.; Eisenberg, R.; Frontier, A. J. ChemCatChem 2011, 3, Dannenberg,J.J.;Dapprich,S.;Daniels,A.D.;Farkas,O.; 1531−1548. Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, (5) Rueping, M.; Ieawsuwan, W.; Antonchick, A. P.; Nachtsheim, B. Revision E.01; Gaussian, Inc., Wallingford, CT, 2013. J. Angew. Chem., Int. Ed. 2007, 46, 2097−2100. (17) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215−241. (6) Kerr, D. J.; Flynn, B. L. Org. Lett. 2012, 14, 1740−1743. (18) Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B (7) Jolit, A.; Walleser, P. M.; Yap, G. P. A.; Tius, M. A. Angew. Chem., 2009, 113, 6378−6396. Int. Ed. 2014, 53, 6180−6183. (19) (a) Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154− (8) (a) Yamanaka, M.; Itoh, J.; Fuchibe, K.; Akiyama, T. J. Am. Chem. 2161. (b) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 94, 5523− Soc. 2007, 129, 6756−6764. (b) Simon,́ L.; Goodman, J. M. J. Am. 5527. Chem. Soc. 2008, 130, 8741−8747. (c) Grayson, M. N.; Pellegrinet, S. (20) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B: Condens. Matter C.; Goodman, J. M. J. Am. Chem. Soc. 2012, 134, 2716−2722. Mater. Phys. 1988, 37, 785−789. (b) Becke, A. D. J. Chem. Phys. 1993, (d) Seguin, T. J.; Lu, T.; Wheeler, S. E. Org. Lett. 2015, 17, 3066− 98, 1372−1377. (c) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652. 3069. (e) Seguin, T. J.; Wheeler, S. E. ACS Catal. 2016, 6, 2681−2688. (d) Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. J. (f) Seguin, T. J.; Wheeler, S. E. ACS Catal. 2016, 6, 7222−7228. Phys. Chem. 1994, 98, 11623−11627. (g) Jindal, G.; Sunoj, R. B. Angew. Chem., Int. Ed. 2014, 53, 4432− (21) (a) MacroModel, version 10.6; Schrödinger, LLC, New York, 4436. (h) Li, G.-Q.; Gao, H.; Keene, C.; Devonas, M.; Ess, D. H.; 2014. (b) Maestro, version 9.9; Schrödinger, LLC, New York, 2014. Kürti, L. J. Am. Chem. Soc. 2013, 135, 7414−7417. (i) Maity, P.; (22) Banks, J. L.; Beard, H. S.; Cao, Y.; Cho, A. E.; Damm, W.; Farid, Pemberton, R. P.; Tantillo, D. J.; Tambar, U. K. J. Am. Chem. Soc. R.; Felts, A. K.; Halgren, T. A.; Mainz, D. T.; Maple, J. R.; Murphy, R.; 2013, 135, 16380−16383. (j) Meng, S.-S.; Liang, Y.; Cao, K.-S.; Zou, Philipp, D. M.; Repasky, M. P.; Zhang, L. Y.; Berne, B. J.; Friesner, R. L.; Lin, X.-B.; Yang, H.; Houk, K. N.; Zheng, W.-H. J. Am. Chem. Soc. A.; Gallicchio, E.; Levy, R. M. J. Comput. Chem. 2005, 26, 1752−1780. 2014, 136, 12249−12252. (k) Rodríguez, E.; Grayson, M. N.; Asensio, (23) (a) Dapprich, S.; Komaromi,́ I.; Byun, K. S.; Morokuma, K.; A.; Barrio, P.; Houk, K. N.; Fustero, S. ACS Catal. 2016, 6, 2506− Frisch, M. J. J. Mol. Struct.: THEOCHEM 1999, 461−462,1−21. 2514. (l) Yamanaka, M.; Hirata, T. J. Org. Chem. 2009, 74, 3266− (b) Vreven, T.; Byun, K. S.; Komaromi,́ I.; Dapprich, S.; Montgomery, 3271. (m) Shibata, Y.; Yamanaka, M. J. Org. Chem. 2013, 78, 3731− J. A., Jr.; Morokuma, K.; Frisch, M. J. J. Chem. Theory Comput. 2006, 2, 3736. (n) Kanomata, K.; Toda, Y.; Shibata, Y.; Yamanaka, M.; Tsuzuki, 815−826.

3475 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476 ACS Catalysis Research Article

(24) Rappe,́ A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A., III; Skiff, W. M. J. Am. Chem. Soc. 1992, 114, 10024−10035. (25) (a) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104. (b) Grimme, S.; Ehrlich, S.; Goerigk, L. J. Comput. Chem. 2011, 32, 1456−1465. (26) Grimme, S. J. Comput. Chem. 2006, 27, 1787−1799. (27) Legault, C. Y. CYLview, 1.0b; Universitéde Sherbrooke, Sherbrooke, Canada, 2009; http://www.cylview.org. (28) Rueping, M.; Nachtsheim, B. J.; Koenigs, R. M.; Ieawsuwan, W. Chem. - Eur. J. 2010, 16, 13116−13126. (29) This idea is supported by comparing the energies of the distorted catalysts in the electrocyclization TSs to the energies of the closest fully relaxed conformers of 10a,b. It takes 3.0 kcal/mol more energy (ΔE) to distort 10a into the geometry found in TS1 than it does to distort 10b into the geometry found in TS2. This is almost the same as the total difference in energy between the two transition states (ΔΔG⧧ = 3.1 kcal/mol). (30) Transition states TS1 and TS2 were also computed with B3LYP-D3 and B97-D3. These two functionals predicted that TS2 is 1.5−2.0 kcal/mol lower in energy than TS1 (see the Supporting Information). (31) A transition state for the Nazarov cyclization catalyzed by the S(NR)OH tautomer 10c is computed to lie 3.1 kcal/mol higher in energy than the favored TS2 (see the Supporting Information). (32) (a) Ess, D. H.; Houk, K. N. J. Am. Chem. Soc. 2007, 129, 10646−10647. (b) Ess, D. H.; Houk, K. N. J. Am. Chem. Soc. 2008, 130, 10187−10198. (33) For a similar fragment-based analysis of the selectivity of a BINOL phosphoric acid catalyzed reaction, see ref 8f. (34) Magnus et al. reported an AcBr-activated Nazarov cyclization where the electrocyclization was proposed to be accompanied by cyclization of the C1-OAc group onto C5 to form a 5,5-bicyclic structure. In that case, however, the OAc group was directly bound to the pentadienyl cation and would be unlikely to be able to achieve the geometry needed to engage in stabilizing intramolecular CH−O and cation−lone pair interactions. See: Magnus, P.; Freund, W. A.; Moorhead, E. J.; Rainey, T. J. Am. Chem. Soc. 2012, 134, 6140−6142. (35) In comparison with the M06-2X data, calculations of TS energies with B3LYP-D3 and B97-D3 predict 1−3 kcal/mol smaller differences in energy between the lowest-energy P(NTf)OH and P(O)NHTf TSs for reactions A−C. With each of these three functionals, the lowest-energy P(O)NHTf TS leads to the minor enantiomer of the product in reactions A and C but to the major enantiomer in reaction B. Details are provided in the Supporting Information.

3476 DOI: 10.1021/acscatal.7b00292 ACS Catal. 2017, 7, 3466−3476