Hypercoordinate iodine for catalytic asymmetric diamination of styrene: insights into the mechanism, role of solvent, and stereoinduction† †Electronic supplementary information (ESI) available: Cartesian coordinates of all stationary points and other relevant information is provided. See DOI: 10.1039/c9sc01513b

Stereoselectivity in the asymmetric diamination of styrene catalyzed by chiral hypercoordinate iodine originates from the prochiral face recognition when the substrate binds to the catalyst.


Introduction
Vicinal diamines are important structural elements in pharmaceuticals, natural products, and organo-as well as transition metal catalysts. 1 Hence, efficient methods for the synthesis of diamines are of great interest. 2 Palladium-and copper-catalyzed asymmetric diamination of alkenes involving intramolecular C-N bond formation is the most frequently found catalytic diamination reaction. 3 Among the metal-free approaches, hypercoordinate iodine mediated diamination of alkenes has received considerable attention in recent years. 4 Hypercoordinate iodine mediated methods have found widespread acceptance owing to their utility in a diverse set of reactions under milder and environmentally benign conditions. 5 An excellent strategy for diamination of styrene derivatives in the presence of a stoichiometric amount of phenyliodine(III) diacetate (PIDA) with two equivalents of sulfonimide HNMs 2 as the aminating agent (Scheme 1a) was recently reported. 6 This method offers wide substrate scope and functional group tolerance. 7 Replacing PIDA with chiral iodoresorcinol derivatives could provide asymmetric diamination. 6 In a recent catalytic version of this reaction, catalytically active chiral hypercoordinate iodine species was generated in situ from the corresponding Ar-I (4) and HNMs 2 in the presence of two equivalents of meta-chloroperbenzoic acid (mCPBA) as an oxidizing agent (Scheme 1b). 8 Difunctionalization of alkenes using hypercoordinate iodine(III) has found interesting applications in the synthesis of complex target molecules. 9 In particular, iodoresorcinol based catalysts derived from the corresponding Ar-I (similar to 4) have been recently used for stereoselective difunctionalization of various styrenes, in which the S stereogenic center at the benzylic position was formed in the product. 9a-c Different chiral iodoresorcinol derivatives differ in terms of the substituents attached to the carbonyl (-N(iPr 2 ) in the representative example 4) and C a (-Me in 4) positions of the chiral arms, while maintaining the R conguration at the C a position. This observation indicates that the preferred prochiral face of styrene involved in all of these reactions is the same, such that the (S)-congured enantiomer of the product is generated. However, the origin of stereoinduction in these reactions is oen interpreted using qualitative working hypotheses. 10 In view of the impressive developments in asymmetric catalysis using hypercoordinate iodine, 11 we probed the origin of stereoinduction 12 in styrene diamination by using DFT(SMD (diethylether) /M06-2X/6-31G**,SDD(I)) computations.

Computational methods
All geometry optimizations were performed using the Gaussian09 suite of quantum chemical programs. 13 The M06-2X density functional was chosen for geometry optimization which accounts for the potentially signicant dispersion interactions in relatively larger molecules. 14 The 6-31G** basis set was chosen for all atoms, except for iodine. 15 The Stuttgart-Dresden double-zeta valence basis set (SDD) with an effective core potential for 46 inner electrons was employed for iodine. 16 Solvent effects were incorporated using the universal SMD continuum solvation model developed by Truhlar and Cramer. 17 The experimental conditions for the catalytic asymmetric reaction employed a solvent system consisting of a 3 : 1 mixture of methyl tert-butyl ether (MTBE) and hexauoroisopropanol. Since the dielectric continuum eld for MTBE (3 ¼ 4.5) is not available in Gaussian09, implicit solvation calculations for closely related diethyl ether (3 ¼ 4.2) were employed. Hexa-uoroisopropanol was explicitly included in several stationary points to evaluate specic interactions with the solute. All stationary points were veried by vibrational frequency analysis. Minima (reactants, products, and intermediates) were characterized by having zero imaginary vibrational frequencies, while transition states had only one imaginary frequency. Further verication of the transition states was done using intrinsic reaction coordinate (IRC) calculations in order to connect to the corresponding reactants and products. 18 We wish to convey that locating various transition state geometries in hypercoordinate iodine catalyzed reactions could be challenging due to geometry convergence issues. Similar computational methods to those employed in this study have successfully been employed in recent studies of hypercoordinate iodine reactions. 19 Topological analysis of the electron densities, within the Atoms In Molecule (AIM) framework, was carried out using AIM2000 soware to identify the bond paths representing the weak inter-atomic interactions. 20 Natural bond orbital (NBO) 21 analysis of important stationary points was done to understand important electron delocalizations and charge distributions. To examine the origin of energy difference between the stereocontrolling transition states, the activation-strain model was employed. 22 In this model, the activation barrier (DE ‡ ) is considered as the sum of (a) destabilizing distortion energies (DE ‡ d ) in the reactants while going from their ground state geometries to those in the transition states, and (b) stabilizing interaction energy between such deformed reactants (DE ‡ i ) in the TS geometry. This approach would help quantify the relative distortion and interaction energies between the stereocontrolling transition states.
The discussion is presented on the basis of the Gibbs free energies at 298 K and as obtained at the SMD (diethylether) /M06-2X/6-31G**,SDD(I) level of theory. Gibbs free energies are corrected using Truhlar's quasi-harmonic approximation to address the issues arising due to the harmonic oscillator approximation for frequencies lower than 100 cm À1 . 23 Gibbs free energies were also calculated using the quasi-rigid rotor harmonic oscillator (RRHO) approximation for the purpose of comparison, which is provided in the ESI. † 24 The energies obtained at the SMD (diethylether) /M06-2X/6-31G**,SDD(I) level of theory were rened using a larger basis set using the same functional to learn that the results were similar. Geometries were also optimized at the B3LYP-D3 level of theory. Energy renements at the B3LYP-D3 level of theory using the geometries at the same level of theory were also carried out to check whether our results were consistent across different functionals. 25

Results and discussion
We have investigated all of the likely elementary mechanistic steps to develop an improved understanding of the key intermediates and transition states (TSs) involved in the diamination reaction, as shown in Scheme 2. An achiral hypercoordinate iodine species (3 0 ) was considered rst to develop some basic understanding about the key mechanistic features. The factors responsible for stereoinduction are then delineated by considering the chiral hypercoordinate iodine species that is derived from 4.
revealed a prominent p electron delocalization (45.1 kcal mol À1 ) from the styrene double bond to the I(III)-N antibonding orbital. 29 The coordination of the styrenyl double bond to the electrophilic I(III) center is found to induce a polarization such that the benzylic carbon C 1 becomes relatively more positive (natural charges on C 1 ¼ À0.12 and on C 2 ¼ À0.54 in C1 whereas that on an unbound free styrene is C 2 ¼ À0.43 and C 1 ¼ À0.25), making the C 1 benzylic carbon more prone to nucleophilic addition. The polarization of the styrene p orbital in this catalyst-substrate complex is evident from a relatively higher p orbital coefficient (0.75) on the C 2 carbon than on C 1 (0.65), which was the same for a free styrene (orbital coefficient of 0.70 on both C 1 and C 2 ).
In the next step, we considered a rebound mechanism wherein the displaced sulfonamide anion adds to the polarized styrenyl double bond. Due to effective charge delocalization from nitrogen to oxygen in the imidate ion, the nucleophilic addition can occur either through the oxygen or through the nitrogen. The natural charges on the nitrogen (À1.32) and the oxygen (À1.04) of the imidate indicate that a nucleophilic addition through the nitrogen is more likely. However, attempts to locate a transition state for the nucleophilic addition through the nitrogen remained unsuccessful, and instead resulted in an addition through the oxygen (TS1). 30, 31 The comparison of nucleophilicity indices of 0.61 and 0.08, respectively, for the nitrogen and oxygen suggests that a nucleophilic addition through the soer sulfonyl oxygen is likely to be more preferred at the benzylic carbon of the iodine-bound styrene resulting in the iodonium ion intermediate IM1. 32 In the next step, a concerted intramolecular migration converts IM1 to IM2 via TS2. 33 Formation of the second iodonium ion intermediate IM2 is exergonic by $16.0 kcal mol À1 (Fig. 1). An intramolecular nucleophilic addition by the sulfonyl oxygen at C 2 can now provide a cyclic oxazolidine oxide intermediate, IM3. The barrier to this elementary step for the expulsion of PhI is 11.4 kcal mol À1 . 34 A subsequent nucleophilic ring opening by imidate can furnish the vicinal diamine product 2. Interestingly, an intermediate with a C 1 -O bond similar to IM1 was observed during the oxidation of 1,1-diphenyl ethylene using PIDA in the absence of HFIP. 30 Earlier reports suggested that uorinated solvents such as hexauoroisopropanol (HFIP) are more benecial in hypercoordinate iodine mediated reactions. 35 For example, the improved reactivity of PIDA in the presence of HFIP is attributed to hydrogen bonding between the ligands on PIDA and HFIP 36 which makes the ligands more labile. 37 We have considered a similar scenario wherein the displaced imidate nitrogen is hydrogen bonded in C1 0 as shown in Fig. 2. Interestingly, the Scheme 2 Mechanism for catalytic vicinal diamination of styrene using hypercoordinate iodine. ensuing nucleophilic addition (TS1 0 ) on the styrenyl carbon is noted to occur through the imidate nitrogen. The natural charge on the HFIP-bound imidate nitrogen is higher (À1.34) than that on the unbound (À1.32). However, the nucleophilicity index of the HFIP-bound nitrogen is lower (0.58) than that of the unbound (0.61), indicating relatively soer nitrogen in the former case. Thus, explicit involvement of HFIP enables direct formation of the N-C bond in C1 0 leading to IM2 0 . An alternative nucleophilic addition through the imidate oxygen is found to be 4.0 kcal mol À1 higher in energy when imidate is associated with HFIP. The HFIP-assisted nucleophilic addition exhibits a barrier of 19.5 kcal mol À1 , which is 3.5 kcal mol À1 lower than that of the unassisted nucleophilic addition (Fig. 1). 38 Having understood that the HFIP-assisted pathway is energetically preferable, we calculated the energetics for the HFIP-assisted formation of diamine product 2 as shown in Fig. 2. 39 Aer shedding light on the mechanism of vicinal diamination of styrene using PhI(NMs 2 ) 2 , we have extended these insights to an asymmetric diamination catalyzed by C 2 symmetric iodoresorcinol based chiral catalysts to establish the factors controlling enantioselectivity. As described earlier, nucleophilic addition at the benzylic carbon (TS1 0 ) is more preferred than at the terminal carbon of the catalyst-substrate complex C1 0 . The stereochemical outcome of the reaction will therefore be dictated by which of the prochiral faces of styrene binds to the iodine center. It is likely that the chiral environment offered by the catalyst could make the binding through one prochiral face energetically more preferred over the other in the ion-pair catalyst-substrate complex C1 0 . In other words, the efficiency of this prochiral face recognition is expected to impact on the extent of enantioselectivity.
The process of stereoinduction is investigated by examining the stereoelectronic features of the in situ generated catalytically active hypercoordinate iodine species 4 0 . In the most preferred conformer of the C 2 -symmetric 4 0 , the chiral arms fold in a right-handed P helical fashion and generate a chiral environment around the iodine. The (R)-congured stereogenic center at the C a is the source of stereoinduction, which facilitates the fold with right-handed P helicity. The corresponding M helical fold is found to be of 8.0 kcal mol À1 higher energy. 12,40 A series of weak non-covalent interactions (NCIs) between the imidate ligands and the chiral arms of the resorcinol backbone (Fig. 3) are found to arrange the chiral arms in a helical assembly. This folded assembly of the catalyst provides an asymmetric environment for the incoming styrene. How the binding of the si and re prochiral faces of styrene to the iodine is recognized differently in the catalyst-substrate complexes is analyzed using the cationic complexes (C1 + ) devoid of the displaced nucleophile imidate associated with HFIP. 41 It is noticed that the catalyst-substrate complex C1 si + wherein styrene interacts through its si face is lower by 1.8 kcal mol À1 than the corresponding re face binding (C1 re + ).
The differential recognition of the styrenyl prochiral faces can be understood as follows. In the si face catalyst-substrate complex, styrene is dispositioned closer to the iodine center, as evident from the shorter C 2 -I distance (2.61Å) than in the re complex (2.74Å), indicating enhanced catalyst-substrate interaction (Fig. 4). Further, the styrene double bond is noted to remain staggered and orthogonal with respect to the C 4 -I 3 bond of the catalyst (q (C1-C2-I3-C4) ¼ 90.3 ) in C1 si + whereas it is eclipsed in C1 re + (q (C1-C2-I3-C4) ¼ 8.8 ). 42 The arrangement of styrene in these catalyst-substrate complexes facilitates a C-H/p interaction between the C a -H of the chiral side chain and the phenyl group of styrene (shown as a in Fig. 4). A larger deviation from the eclipsed alignment of the styrene double bond in the re complex (measured in terms of dihedral q (C1-C2-I3-C4) ) will result in loss of this critical interaction. Additional weak noncovalent interactions are noticed in the si mode of binding, which are absent in the re complex. Natural bond orbital (NBO) analysis revealed an enhanced p electron delocalization from the styrene double bond to the I-N antibonding orbital in C1 si + (46.1 kcal mol À1 ) compared to that in C1 re + (30.7 kcal mol À1 ). Comparison of a free styrene with a bound styrene as noted in the catalyst-substrate complexes is carried out to understand the effect of iodine coordination on the reactivity and selectivity.
To examine how the relative orientation of the styrene double bond and the Ar-I bond inuences their energies, catalystsubstrate complexes with staggered and eclipsed orientations of the substrate are analyzed for an achiral catalyst (Fig. 5a). The staggered arrangement (C1 sta + ) is found to be 1.0 kcal mol À1 lower in energy than the eclipsed geometry (C1 ecl + ). In the case of the achiral catalyst, binding of both the prochiral faces of styrene is found to be energetically similar, suggesting no facial discrimination.
To gather more insights, the p and p* natural bond orbitals (NBOs) in a free styrene as well as those in an iodine-bound styrene are analyzed. These are critical NBOs, as the donor p and acceptor p* are, respectively, involved in the catalystsubstrate complex formation and in the nucleophilic addition. Both benzylic (C 1 ) and terminal carbon (C 2 ) atoms of the free styrene contribute equally to the p orbital as evident from the corresponding orbital coefficients (C 1 ¼ C 2 ¼ 0.70, Fig. 5b). Interestingly, in the catalyst-substrate complexes, the p orbital is polarized toward the terminal carbon as indicated by a higher orbital coefficient on C 2 than that on C 1 whereas the corresponding p* orbital is more polarized toward the benzylic Fig. 3 Optimized geometry of catalytically active species 4 0 . All distances are inÅ and relative energies are in kcal mol À1 . Important noncovalent interactions (a-h) are shown in blue with the corresponding electron densities (r Â 10 À2 au in parentheses) at the bond critical points along the bond paths. Only selected hydrogen atoms are shown for improved clarity. The helical axis (red dotted line) formed due to the folding of the chiral side chain (ribbon-like depiction) is shown in the structure on the left. Fig. 4 Optimized geometries of cationic catalyst-substrate complexes C1 si + and C1 re + as well as the corresponding qualitative representations. All distances are inÅ and relative energies are in kcal mol À1 . Important noncovalent interactions (a-c) are shown in blue with the corresponding electron densities (r Â 10 À2 au in parentheses) at the bond critical points along the bond paths. Only selected hydrogen atoms are shown for improved clarity. carbon C 1 . In the achiral catalyst-substrate complexes, the p orbital polarization is noted to be higher in the case of the staggered geometry of styrene in C1 sta + (coefficients on C 1 and C 2 are 0.65 and 0.75, respectively) compared to that in C1 ecl + (C 1 ¼ 0.64 and C 2 ¼ 0.76). 43 More interestingly, a differential polarization of the p and p* is noted in the chiral catalystsubstrate complexes depending on the prochiral face through which styrene binds to the iodine center. The styrene double bond in the si binding mode, which has a staggered arrangement, is found to be more polarized than in the re complex (Fig. 5c). The orbital coefficients for lled p in C1 si + are C 1 ¼ 0.64 and C 2 ¼ 0.76 and those in C1 re + are C 1 ¼ 0.66 and C 2 ¼ 0.74. The chirality induced differential polarization of the styrenyl double bond, as described here, can have a pronounced impact on the prochiral face recognition and hence the stereochemical course of the reaction. Akin to the orbital polarization noted here, the natural charge on C 1 and C 2 exhibited similar features. 43 In summary, the enhanced polarization of the styrene p orbitals, particularly that of the p* orbital in the lower energy staggered catalyst-substrate complex, improves the propensity toward the nucleophilic addition of the imidate ion.
In the next step, stereoselective addition of the imidate on the benzylic position of styrene bound to chiral catalyst 4 0 is considered (for the geometry of the catalyst, see Fig. 3). The C-N bond formation via TS1 0 leads to the formation of an iodonium ion intermediate (IM2 0 ). Nucleophilic addition of the imidate to the catalyst-substrate complexes C1 0 si and C1 0 re is separately examined. 44 The transition state for the imidate addition to the open re face of the bound styrene in C1 0 si is denoted as TS1 0 S as the new stereogenic center created in this mode would be of S conguration. The relative Gibbs energy of TS1 0 S is found to be 6.5 kcal mol À1 lower in energy than that of TS1 0 R . Similarly, the elementary step barrier for the nucleophilic addition through TS1 0 S is 3.1 kcal mol À1 lower than that for TS1 0 R . A staggered arrangement of the styrene double bond (q (C1-C2-I3-C4) ¼ 89.2 ) with respect to the Ar-I bond is noted in the lower energy TS1 0 S whereas it remains in an eclipsed (q (C1-C2-I3-C4) ¼ 21.4 ) geometry in the higher energy TS1 0 R . 45 Analysis of TSs shows the presence of certain types of noncovalent interactions (denoted as h and i in Fig. 6) between the reactants (styrene and imidate) and the catalyst in the case of TS1 0 S which are absent in TS1 0 R . However, these weakly differentiating interactions are rather inadequate toward rationalizing an overwhelming preference for imidate addition via TS1 0 S . Hence, we have performed activation-strain analysis on these diastereomeric TSs by fragmenting them into three parts such as the catalyst, styrene and nucleophile. All three fragments in TS1 0 R exhibited relatively higher distortion energy than in TS1 0 Sthe sum of which amounts to 14.4 kcal mol À1 in the former. 46 The stabilizing interaction energy between these three fragments is also higher (À6.9 kcal mol À1 ) in TS1 0 R . On balance, the distortion appears to be a key contributor of the energy difference between the diastereomeric TSs. The difference in the elementary step barriers for the nucleophilic addition of the imidate on the re face of the catalyst-substrate complex and that on the si face is found to be 3.1 kcal mol À1 , exhibiting a preference toward the formation of the S enantiomeric product. 47 While the difference in barriers corresponds to an ee of 98% which is in good agreement with the experimental observation of 94%, the relative energies of the diastereomeric transition states result in a modest overestimation of ee.

Conclusions
The energetically favorable mechanism and the origin of enantioselectivity in hypercoordinate iodine catalyzed vicinal diamination of styrene have been identied. Styrene coordination to the electrophilic iodine center polarizes the double bond making the benzylic carbon susceptible to nucleophilic addition by imidate derived from the active catalyst ArI(NMs 2 ) 2 . Hexauoroisopropanol-assisted nucleophilic addition for the C-N bond formation proceeds with a lower energy barrier and results in an iodonium ion intermediate. Under the chiral environment provided by the iodoresorcinol catalyst, the coordination of the si prochiral face of styrene is more preferred due to the relatively more effective catalystsubstrate interaction than in the re face complex. This prochiral face recognition is found to have a direct impact on the enantioselectivity of this catalytic asymmetric reaction. The kinetic advantage for nucleophilic addition of the imidate is more for the si-face of the catalyst-substrate complex, which could be traced to a relatively lower distortion in the reactant partners in the stereocontrolling transition state. The computed % ee is in good agreement with the experimental observation.

Conflicts of interest
The authors declare no competing nancial interest. Fig. 6 Optimized geometries of the stereoselective transition states TS1 0 S and TS1 0 R . Relative Gibbs free energy is given in parentheses. Important noncovalent interactions (a-i) are shown in blue along with the corresponding electron densities (r Â 10 À2 au in parentheses) at the bond critical points along the bond paths. All distances are inÅ.