Computational and Theoretical Chemistry 1056 (2015) 61–73
Contents lists available at ScienceDirect
Computational and Theoretical Chemistry journal homepage: www.elsevier.com/locate/comptc
A comparison of the Simmons-Smith reaction with carbenoids to nitrenoids and oxenoids Sarah Karbalaei Khani, Thomas R. Cundari ⇑ CASCaM, Univ. of North Texas, Dept. of Chemistry, 1155 Union Circle, #305070, Denton, TX 76203-5017, United States
a r t i c l e
i n f o
Article history: Received 14 October 2014 Received in revised form 9 January 2015 Accepted 10 January 2015 Available online 20 January 2015 Keywords: Computational chemistry Carbenoid Nitrenoid DFT Catalysis
a b s t r a c t A computational study using both density functional and correlated wavefunction methods on the reaction between ethylene and model zinc carbenoid, nitrenoid and oxenoid complexes (L–Zn–E–X, E = CH2, NH or O, L = X = I or Cl) is reported. An investigation at DFT and MP2 levels of theory predicts that the epoxidation of ethylene with ClZnOCl oxenoid proceeds through a two-step mechanism involving an initial oxometalation and subsequent ring-closure, the latter being the rate determining step (RDS). Among all DFT methods applied in this study, BP86/CEP-31G(d) method produced the most similar geometries and energy barrier for the second step to those derived from MP2 simulations with correlation consistent basis sets. Interestingly, the mechanism of the cyclization reaction of ethylene with LZnEX is dependent on both E and X groups. Cyclopropanation of ethylene with IZnCH2I and aziridination of ethylene with IZnNHI proceed via a single-step mechanism with an asynchronous transition state. The reaction barrier for the aziridination with IZnNHI is 5.4 kcal/mol lower than that of cyclopropanation. Changing the leaving group of IZnNHI from I to Cl changes the mechanism of the aziridination reaction to a two-step pathway, with the second step as the RDS. The calculation results from the epoxidation with IZnOI and ClZnOCl oxenoids suggest a two-step mechanism for both oxenoids. The epoxidation reaction barriers for the RDS for both IZnOI and ClZnOCl is 15 kcal/mol, which is 6 kcal/mol less than that calculated for aziridination of ethylene with ClZnNHCl nitrenoid. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Cyclopropanes, aziridines and epoxides are three-member ring moieties found in a wide variety of drugs, natural products and other important chemical compounds [1]. For example, cyclopropane containing molecules are used as starting materials and intermediates in organic synthesis due to their inherent ring strain and also have some pharmaceutical applications as general anesthetics [2,3]. Aziridines, the smallest nitrogen-containing heterocycles, are also widely exploited as building blocks for the synthesis of biologically and synthetically important materials [4]. They are bioactive compounds and form key reactive elements in larger, complex molecules, e.g. antitumor agents, taking advantage of their ring strain [4c,5]. Stereospecific ring opening of aziridines can produce several classes of useful molecules such as amines, amino acids, amino alcohols, and other nitrogen-containing compounds [6]. Epoxides such as ethylene oxide and propylene oxide are produced and employed on an industrial scale to form important intermediates for the industrial synthesis of polymers. Moreover, they form building blocks in a variety of bioactive molecules [7]. ⇑ Corresponding author. E-mail address:
[email protected] (T.R. Cundari). http://dx.doi.org/10.1016/j.comptc.2015.01.005 2210-271X/Ó 2015 Elsevier B.V. All rights reserved.
Numerous synthetic routes have been established for cyclopropanation, aziridination and epoxidation of olefins employing transition metal, main group and organic catalysts. Since the reported reaction of zinc carbenoid with olefins to form cyclopropanes by Simmons and Smith [8,9], many studies have been reported on this important organic name reaction, both experimental and theoretical. In 1964, the reaction of a Li carbenoid with olefins to form arylcyclopropanes was first reported by Closs and Moss [10]; theoretical studies suggested that a concerted [1 + 2] addition is more favored versus a stepwise mechanism [11]. In 1987, the reaction of samarium carbenoids with olefins was reported by Molander and coworkers [12,13], while aluminum carbenoid mediated cyclopropanation was first investigated experimentally by Yamamoto and co-workers a couple of years earlier [14]. Traditionally, epoxides and aziridines have been synthesized by the addition of sulfur ylides to the appropriate ketone, aldehyde, imine, or enone (Corey–Chaykovsky aziridination) [15], by the reaction of imines with diazo-containing compounds via the Aza–Darzens reaction [16], or by the cyclization of amino alcohols through the Wenker synthesis [17]. Recently, transition metalcatalyzed synthesis of aziridines with different choice of nitrene/ nitrenoid source and metal catalyst has received increasing
62
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
research attention [18]. Evans et al. [19] made a critical discovery with the use of bis (oxazoline) copper (I) complexes for the aziridination of alkenes. In 2013, Cordeiro et al. [20] conducted a DFT study on the mechanism of the copper-catalyzed aziridination of styrene. In 1995, Che et al. developed an efficient ruthenium catalyst to mediate nitrene-transfer reactions to C@C bonds and subsequently, the RhII-catalyzed aziridinaton of olefins through a concerted mechanism for the transfer of nitrenoid to the olefin was reported by the group of Müller [21]. In 2011, a reusable iron aziridination catalyst, supported by a macrocyclic tetracarbene ligand, with improved atom economy, was developed by Jenkins and co-workers [22]. The following transition metals have been reported for the conversion of olefins into aziridines: Mn, Fe, Co, Cu, Ru, Rh, Pd, Ag and Au [23]. Among these, rhodium-, ruthenium-, and copper-based catalysts generally show better catalytic performance [18b,24].
a
b
c
Scheme 1. General structure of (a) carbenoids, (b) nitrenoids, and (c) oxenoids with a metal M and a leaving group X and the supporting ligand L. In this study M = Zn, X = L = I or Cl and R1 = R2 = R = H.
Epoxidation is an important process for both industrial and pharmaceutical purposes. Hence, many transition metal catalysts have been successfully developed to make a variety of epoxides from olefins. In addition to well-known epoxidation catalysts based on Ti, V, Mo and W that can facilitate the epoxidation of a variety of alkenes with an alkyl hydroperoxide oxidant, in the last decade several noble metal catalysts utilizing Ru, Pt, Co or Au have been developed for selective epoxidation of alkenes [25]. Epoxidation of propene with dioxygen in the presence of a supported gold catalyst on titania or titanium silicalites was reported by Haruta and co-workers [26] in 1999. They showed that Au/TiO2 can be used selectively for propene epoxidation in the presence of H2. Beller et al. developed a cobalt catalyst, which can be easily recycled, for epoxidation of olefins with high yields [25e]. Carbenoids are putative intermediates in cyclopropanation reactions, and are qualitatively similar to carbenes, but do not necessarily form the free divalent species [10,30]. Experimental and theoretical studies revealed that carbenoid, nitrenoid and oxenoid species have many properties in common, e.g. they are very electrophilic and react easily with nucleophiles through an SN2-type mechanism (Scheme 1) [32]. The Simmons-Smith (SS) cyclopropanation is a general and efficient method for the conversion of olefins into cyclopropanes [27]. The utility of the Simmons-Smith reaction compared to other cyclopropanation techniques is due to the stereospecificity and the compatibility of this reagent with different functional groups [28]. The SS reaction is proposed to go through a methylene
Table 1 DFT Gibbs free energy of activation DGà and reaction DGr° relative to reactants in kcal/mol for the two-step epoxidation of ethylene by ClZnOCl. Step 1 = oxometalation, step 2 = ring closure (see Scheme 9).
63
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
Table 2 MP2 Gibbs free energy of activation DGà and reaction DGr° relative to reactants in kcal/mol for the two-step epoxidation of ethylene by ClZnOCl. Step 1 = oxometalation, step 2 = ring closure (see Scheme 9).
I
II
III
IV
Scheme 2. Transition states of the epoxidation of ethylene with ClZnOCl oxenoid using BP86/CEP-31G(d) (I, II) and MP2/cc-pVDZ (III, IV) levels of theory, (I, III – oxometalation), (II, IV – ring closure).
transfer mechanism with a concerted [1 + 2] addition via a butterfly transition state [9]. A two-step carbometalation mechanism involving a four-centered transition state was also proposed based on later mechanistic studies [29]. Furthermore, some research has been reported on the influence of carbenoid structure on the cyclopropanation reaction, elucidating that changing the leaving group
dα
dβ
I
∆d = dα - dβ
dα
dβ
II
Scheme 3. Transition states for the two-step epoxidation of ethylene by ClZnOCl (I, II) and the definition of the bond asynchronicity Dd.
64
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
Table 3 Incipient bond distances (da and db in Scheme 4) and bond asynchronicities (Dd) in the transition states for the epoxidation of ethylene with ClZnOCl oxenoid (all distances are given in Ångström). Method
Step 1
BP86/CEP-31G(d) B97D/CEP-31G(d) BLYP/CEP-31G(d) B3P86/CEP-31G(d) B3LYP/CEP-31G(d) M06/CEP-31G(d) BP86/6-31G(d) B97D/6-31G(d) BLYP/6-31G(d) B3P86/6-31G(d) B3LYP/6-31G(d) M06/6-31G(d) BP86/6-311+G(d) B97D/6-311+G(d) MP2/aug-cc-pVDZ//bp86/CEP-31G(d) MP2/aug-cc-pVTZ//bp86/CEP-31G(d) MP2(full)/aug-cc-pVDZ MP2(full)/aug-cc-pVTZ
L-Zn-CH2-X S-S
Step 2
da
db
Dd
da
db
Dd
2.245 2.260 2.284 2.192 2.215 2.140 2.144 2.184 2.172 2.123 2.142 2.080 2.251 2.279 2.245 2.245 2.187 2.167
2.060 1.993 2.014 2.014 1.993 2.017 2.053 1.997 2.018 1.994 1.977 1.980 2.002 1.940 2.060 2.060 1.903 1.900
0.185 0.267 0.270 0.178 0.222 0.123 0.091 0.187 0.154 0.129 0.165 0.100 0.249 0.339 0.185 0.185 0.284 0.267
2.165 2.150 2.217 Not found Not found 2.088 2.113 2.121 2.160 2.063 2.104 2.041 2.164 2.159 2.165 2.165 2.175 2.156
1.536 1.499 1.565 Not found Not found 1.478 1.527 1.512 1.546 1.498 1.516 1.476 1.536 1.508 1.536 1.536 1.485 1.477
0.629 0.651 0.652 Not found Not found 0.610 0.586 0.609 0.614 0.565 0.588 0.565 0.628 0.651 0.629 0.629 0.690 0.679
L-Zn-NH-X
L-Zn-O-X D-S
Scheme 4. Proposed transition state continuum between single-step (S-S) and double-step (D-S) pathways for cyclopropanation, aziridination and epoxidation of ethylene with L–Zn–E–X (E = CH2, NH or O, X = L = I or Cl).
carbenoid, it is surprising that similar reagents [18a,c] have been little explored or exploited in organic chemistry for related zinc-nitrenoid and zinc-oxenoid mediated transformations to form aziridines and epoxides, respectively. Motivated by this, density functional and wavefunction theory (perturbation theory and coupled cluster methods) investigations were performed to help elucidate mechanisms for the conversion of ethylene to aziridine and ethylene oxide by analogous IZnNHI nitrenoid and IZnOI oxenoid, respectively, models. Meanwhile, the influence of leaving group X (X = I, Cl) on the epoxidation of ethylene pathway was also assessed, and found to be sensitive to the level of theory employed. 1.1. Computational methods
X (X = F, Cl, Br, I, OH) and the metal ion can change the nature of XMCH2X carbenoids (M = Zn, Li, Al, Rd, Sm) [30]. Since cyclopropane-containing molecules have an important role as starting materials and intermediates in organic synthesis, considerable effort has been invested in the development of new techniques to form active reagents similar to the proposed zinc carbenoid active species in SS reactions [31]. In this regard, Zhao et al. conducted quantum mechanical studies on cyclopropanation by LiCH2X (X = halogen and OR) carbenoids of CH2@CH2. In different aggregation and solvation states, the methylene transfer and carbometalation pathways were modeled to determine the reaction mechanism. These researchers found that the methylene transfer pathway makes more of a contribution to the lithium carbenoid-promoted cyclopropanation reaction than does the carbometalation mechanism [11]. In 2006, the same group reported DFT calculations for the cyclopropanation of ethylene by an aluminum-carbenoid model (CH3)2AlCH2X (X = Cl, Br, I) via both methylene-transfer and carbometalation pathways and demonstrated that (CH3)2AlCH2Cl is the most reactive carbenoid, while the (CH3)2AlCH2I is the least reactive one through the methylene transfer pathway, which is more favored than carbometalation pathway. The comparison of the cyclopropanation reaction of IZnCH2I, (CH3)2AlCH2I, ISmCH2I and LiCH2I carbenoids suggests that the more carbene-like character, the more reactive the carbenoid is. The computed trend of the free energy barriers for carbenoid reactivity in the cyclopropanation reaction is IZnCH2I (21.2 kcal/mol) > (CH3)2AlCH2I (12.8 kcal/mol) > ISmCH2I (5.5 kcal/mol) LiCH2I (6.8 kcal/mol), which is consistent with experiment [31]. Considering Simmons-Smith reaction as a widely used approach to access cyclopropanes and the IZnCH2I as a reactive
The Gaussian ‘09 software package [35] was used for all calculations reported in this paper. Calculations were performed for molecules in the gas phase. Reaction and activation free energies were calculated using density functional theory (DFT): BP86, B97D, BLYP, B3P86, B3LYP and M06 functionals, each with the CEP-31G(d) and 6-31G(d) basis sets. The BP86 and B97D functionals were also tested with the larger 6-311+G(d) basis set. Single point calculations and full optimizations were also performed at MP2 level of theory with the cc-pVDZ and cc-pVTZ basis sets. Coupled cluster (CCSD(T)) single point computations were also performed in conjunction with correlation consistent basis sets. All species were modeled as neutrals with a spin multiplicity of 1. Preliminary computations suggested that triplet species were much higher than the singlet stationary points discussed here. Free energies are reported in kcal/mol that were calculated at 1 atm and 298.17 K. According to the calculated energy Hessians, the stationary points, minima or transition states, are defined by having 0 and 1 imaginary frequencies, respectively. 2. Result and discussion 2.1. Impact of functional and basis set upon epoxidation of the Cl–Zn– O–Cl oxenoid with ethylene Various computational strategies were used to assess the sensitivity of the reaction of ClZnOCl + C2H4 to different methods and basis sets. Chloride supporting ligands, rather than the iodide ligands that are more commonly used experimentally in the Simmons-Smith reaction, were utilized to provide more basis set
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
options for calibration of the computational methods. Additionally, initial studies of the epoxidation reaction indicated this reaction to proceed by a two-step mechanism – oxometalation followed by ring closure – and hence it was deemed a more rigorous test of the levels of theory than a single step (i.e. concerted via a ‘‘butterfly’’ transition state) group transfer mechanism.
A1
65
Tables 1 and 2 summarize the free energy changes for each class of methods. Comparing the data in Table 1 with Table 2, it can be seen that for the transition state for the first mechanistic step (TS1) in oxometalation to form ClZnOCH2CH2Cl, B3LYP/631G(d) and M06/6-31G(d) predict similar free energy barriers to those predicted with MP2 methods in conjunction with
B1
C1 Fig. 1. BP86/CEP-31G(d) computed reaction coordinate of ethylene and IZnCH2I (A1). Singlet TS for the cyclopropanation of ethylene (B1). Dissociated product (C1). Bond lengths in Å.
Scheme 5. BP86/CEP-31G(d) calculated free energies (kcal/mol) for cylcopropanation of ethylene.
66
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
correlation consistent basis sets. There is an increase in the mean value of DGà1 by 4 kcal/mol in the MP2 barrier height for TS1 as compared to the mean barrier computed with DFT methods cf. Tables 1 and 2. The MP2 computed reaction barriers and reaction free energies for step 2 of the epoxidation mechanism – ring closure – are highly consistent for the basis sets tested, as indicated by the small standard deviation (15.7 ± 1.2 kcal/mol for hDGà2i ± r and 57.3 ± 0.9 kcal/mol for hDGr2°i ± r). On the other hand, there is a more considerable spread in the DFT computed barrier heights for step 2 (ring closure); for example, M06/CEP-31G(d) predicts a barrier of 25.8 kcal/mol, while B97D/6-311+G(d) predicts DGà2 = 9.2 kcal/mol. Interestingly, the coupled cluster results (CCSD(T)/cc-pVDZ// BP86/CEP-31G(d)) are intermediate for most of the computed free energies between the density functional and MP2 computed values. This suggests that one treatment of correlation may be under and the other over correcting the computed free energies. To analyze the dependence of transition state geometries (ground state geometries showed less variation with the level of theory) on the choice of quantum chemical optimization method for the epoxidation of ethylene with ClZnOCl oxenoid, the transition state geometries at the BP86/CEP31G(d) and MP2 levels have been compared. The transition state geometries at BP86/ CEP31G(d), within an acceptable error (±0.09 Å for bond lengths and ±0.05° for angles), are very similar to those produced using the MP2/cc-pVDZ level of theory. One exception is the OAC2 bond length in the first step transition state (TS1), which is 0.157 Å longer at BP86/CEP-31G(d) than MP2/cc-pVDZ level (Scheme 2). These results support the hypothesis put forth above that the sophistication of electron correlation treatment, more than differences in stationary point geometries, are responsible for the differences in wavefunction and density functional treatments.
Incipient bond distances (da and db in Scheme 3) are reported for transition states of the epoxidation of ethylene with ClZnOCl oxenoid in Table 3. While DFT computed bond asynchronicities (Dd) decrease by changing the basis set from CEP-31G(d) to 631G(d) for both TS1 and TS2, the 6-311+G(d) geometries are more asynchronous than CEP-31G(d) in TS1 and are similar for CEP31G(d) in TS2. Among all DFT methods applied in this study, BP86/CEP-31G(d) gives the same synchronicity as MP2/CEP31G(d) (Table 3 and Scheme 2). Perhaps most interestingly in terms of the potential to use DFT methods for catalyst design, the mechanism of the reaction is changed for B3P86/CEP-31G(d) and B3LYP/CEP-31G(d) hybrid methods. The corresponding all-electron DFT computations with the 631G(d) basis set favor a two-step mechanism, which is consistent with the MP2 calculations. This highlights the difficulties in assigning a preferred pathway to these late-metal group transfer reactions from theory, and suggests that from a synthetic viewpoint there is likely a continuum of transition states between fully concerted and fully step-wise. Furthermore, the position a particular catalyst system may occupy on this TS continuum is likely quite sensitive to the olefin substrate, metal (M), its supporting ligands (L) and the substituent (X) used to protect the ‘‘oid’’ reagent (Scheme 4). While the (hDG°i ± r)DFT values in Table 1 are different from (hDG°i ± r)MP2 values in Table 2, in terms of the averages and standard deviations, most DFT/basis set computations as well as MP2 predict a two-step mechanism for the reaction of ClZnOCl + C2H4, in which the second step is the rate determining step, with the exception of B97D/6-311+G(d), where the first step is predicted to be the rate determining step by a mere 0.6 kcal/mol. Given the aforementioned results, BP86/CEP-31G(d) was chosen as the best compromise for further simulations for cyclopropanation, aziridination and epoxidation calculations, because it predicts a similar
B2
A2
C2 Fig. 2. BP86/CEP-31G(d) computed stationary point geometries for reaction of ethylene and IZnNHI (A2). Singlet TS for the aziridination of ethylene (B2). Product (C2). Bond lengths in Å.
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
free energy barrier for the second step (15.1 kcal/mol), which is the predicted rate determining step, to that of the more expensive MP2(full)/aug-cc-pVDZ+ZPE (15.4 kcal/mol) level of theory. Moreover, the BP86 functional has been used with success in previous theory-experiment studies of late transition metal group transfer [36]. Finally, the computed data in Tables 1 and 2, determined with a variety of density functional and wavefunction base methods, provide some insight into the expected ‘‘error bars’’ for the predicted thermodynamics and kinetics of these and related group transfer reactions mediated by zinc and other late transition metal carbenoid systems. 2.2. Cyclopropanation of ethylene by the I–Zn–CH2–I carbenoid The optimized geometry for the Zn carbenoid I–Zn–CH2–I is shown in Fig. 1 along with the optimized geometry of the reactant complex (A1) and the transition state (B1) for methylene transfer to produce cyclopropane (c-C3H6) and ZnI2. The methylene transfer is
67
a concerted [2 + 1] addition in which the Zn carbenoid approaches ethylene in an asymmetric way to add the methylene group of the carbenoid to the ethylene p-bond and form two new CAC bonds. The asymmetry is evident in the transition state geometry; the C1AC2 distance in the cyclopropanation TS is 2.38 Å, which is 0.29 Å shorter than the C1AC3 distance. In comparison to the CAC bond lengths in the cyclopropane product, the relatively large C1AC2 distance indicates that the TS is closer to the reactant complex than products (i.e. an ‘‘early’’ transition state). The C2AC3 and C1AI2 bond lengths are respectively increased by 0.01 and 0.50 Å, and the ZnAI2 distance is decreased by 0.72 Å upon going from reactant adduct A1 to transition state B1 (Fig. 1). The latter structural changes are attributed to the incipient formation of ZnI2 product in the transition state. This cyclopropanation reaction is predicted to be highly exergonic by 47.4 kcal/mol with a modest barrier of 13.7 kcal/mol (Scheme 5). According to previous calculations, the reaction of IZnCH2I + C2H4 to give c-C3H6 and ZnI2 has a barrier of 21.2 kcal/mol, relative
A3
B3
C3
D3
E3 Fig. 3. BP86/CEP-31G(d) computed reaction coordinate for the reaction of ethylene and ClZnNHCl (A3). Singlet four-centered TS1 (B3). Intermediate (C3). Singlet ring-closing TS2 (D3). Product (E3). Bond lengths in Å.
68
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
to the separated reactants of IZnCH2I and C2H4, and is exergonic by 21.2 kcal/mol at the B3LYP/6-311G⁄⁄ level of theory [33]. These values are 7.5 kcal/mol higher (for the reaction barrier) and 26.2 kcal/mol lower (for the formation of product) than the BP86/CEP-31G(d) calculated results presented here. Similar to the calibration reported above, in a previous study, a calibration of DFT in relation to MP2 and CCSD(T) methods for the cyclopropanation reaction of methoxymethyllithium with ethylene was reported by Ramachandran et al. [34]. Their results suggested that the reaction of methoxymethyllithium with ethylene gives the cyclopropane through a carbolithiation pathway. 2.3. Aziridination of X–Zn–NH–X (X = I and Cl) ethylene by zinc nitrenoids Interestingly, the aziridination of ethylene by I–Zn–NH–I proceeds through a concerted pathway while, the aziridination with Cl–Zn–NH–Cl proceeds by a two-step mechanism (Figs. 2 and 3, respectively). The aziridination of IZnNHI via a concerted mechanism (Scheme 6) involves the electrophilic attack of the NHAI r⁄
bond on the p bond of ethylene [11], leading to the butterfly-type transition state. A lower free energy barrier of 8.3 kcal/mol is computed for aziridination versus that computed for cyclopropanation (13.7 kcal/mol, Schemes 5 and 6). The C2AN distance is 2.145 Å, which is 0.52 Å shorter than the C1AN distance in the aziridination TS. Hence, the transition state geometry is asynchronous with respect to the formation of the two carbon–nitrogen bonds of aziridine and that asynchronicity is greater than computed for cyclopropanation (0.29 versus 0.52 Å). The interactions of the IZnNHI with the p-olefin orbital are mainly responsible for the slight lengthening of C1AC2 and NAZn bonds [11] from reactant adduct A2 to transition state B2, where the C1AC2 bond length is elongated by 0.02 Å and the NAZn bond length is elongated by 0.02 Å, respectively. There are large changes associated with I–NH–Zn and I–Zn– NH angles from 115.2° to 156.8° in ground state A2 to 88° and 172° in transition state B2, respectively. Particularly noteworthy in the transition state B2 is that the I2AN bond is nearly broken (2.36 Å compared to 2.11 Å in the reactant complex) and the I atom is attracted by the metal center to result in a ZnAI2 bond (Fig. 2) in the ZnI2 product.
Scheme 6. BP86/CEP-31G(d) calculated free energies (kcal/mol) for concerted aziridination of ethylene.
Scheme 7. BP86/CEP-31G(d) calculated free energies (kcal/mol) for two-step aziridination of ethylene.
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
As shown in Scheme 6, the NH transfer pathway has a reaction barrier of 8.3 kcal/mol and is exergonic by 58.9 kcal/mol. In an effort to understand the role of supporting halide ligand in NH transfer, the I atoms in IZnNHI were substituted with the more electronegative Cl atom. Changing the ligands I to Cl leads to a two-step mechanism for the reaction of ClZnNHCl and C2H4. A one-step pathway could not be found for the iodo-ligated zinc nitrenoid with the various methods employed in this research. The first step from ClZnNHCl + C2H4 to a four-centered transition state (B3) has a barrier of 6.8 kcal/mol (Scheme 7). For the transition state B3, the p orbital of the C1AC2 bond is polarized toward C1 atom, leading to the interaction between C1 and Zn atoms, and an appreciable overlap of the NAZn r bond and the polarized p⁄ orbital of ethylene forms a NAC2 r bond [11]. The intermediate C3 that is lower in free energy than A3 by 14 kcal/mol, goes up to D3 in which the interaction of NAZn r bond with p ethylene orbital [11] leads to the formation of a ring-closing transition state (Fig. 3). Hence, there is a difference between the electronic features of transition states B3 and D3. The transition state energies (relative
69
to A3) of B3 and D3 are calculated to be 3.7 and 4.9 kcal/mol, respectively. Energy analysis shows that the energy barrier of the second step (21.1 kcal/mol) is higher than that of the first step (6.8 kcal/mol), Scheme 7, so the second step is the rate determining step. 2.4. Epoxidation of X–Zn–O–X (X = I and Cl) oxenoids with ethylene The epoxidation of ethylene by both I–Zn–O–I and Cl–Zn–O–Cl was computed to proceed through a two-step mechanism including initial oxometalation and subsequent ring-closure as shown in Figs. 4 and 5. Thus, unlike the zinc nitrenoid models, for the zinc carbenoid and zinc oxenoid, the supporting ligand and substituent do not alter the preferred pathway. A concerted, single-step epoxidation transition state, upon starting the geometry optimizations from the corresponding stationary points for the NH congeners, could not be found for either the chloride or iodide reagents at the BP86/CEP-31G(d) level of theory. The first step from IZnOI + C2H4 to transition state B4 has a free energy barrier of 6.4 kcal/mol,
A4
B4
D4
C4
E4 Fig. 4. BP86/CEP-31G(d) computed epoxidation reaction coordinate of ethylene and IZnOI (A4). Four-centered TS1 (B4). Intermediate (C4). Singlet ring-closure TS2 (D4). Product (E4). Bond lengths in Å.
70
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
B5
A5
C5
D5
E5
Fig. 5. BP86/CEP-31G(d) computed epoxidation reaction coordinate for ethylene and ClZnOCl (A5). Singlet four-centered TS1 (B5). Intermediate (C5). Singlet ring-closure TS2 (D5). Product (E5). Bond lengths in Å.
which is very close to that found (6.5 kcal/mol) for the first step of the ClZnOCl + C2H4 reaction, Schemes 8 and 9. As depicted in Figs. 4 and 5, although two pathways show similar transition state shapes for the first step, the distances of OAC2 (2.09 Å) and ZnAC1 (2.27 Å), the key geometric parameters for TS B4 of IZnOI oxenoid are slightly longer than those of TS B5 for the ClZnOCl oxenoid reaction (values of 2.06 and 2.25 Å, respectively). Both transition states are similarly asynchronous. The initial insertion reaction of the ethylene with the ZnAO bond produces either intermediate C4 (X = I) or C5 (X = Cl) through four-centered transition states, B4 or B5. For both pathways, the second step from C to D is the rate determining step, again with similar energy barriers of 15 kcal/mol (Figs. 4 and 5) for either halide. The reaction of ClZnOCl and C2H4 is more exergonic (51 kcal/mol) than the reaction of IZnOI and C2H4 (37 kcal/mol) (Schemes 8 and 9). Hence, the current models suggest only a small impact on computed barriers for zinc oxenoid mediated epoxidation by changing the halide ligands.
3. Summary and conclusions In this paper, the potential energy surfaces for the reactions between ethylene and model zinc carbenoid, nitrenoid and oxenoid complexes (L–Zn–E–X, E = CH2, NH or O, L = X = I or Cl) have been studied with both density functional and correlated wavefunction methods. Several results from this research may provide insight into group transfer by late transition metal ‘‘oids.’’ The most essential among these are summarized below. (1) The epoxidation of ethylene with ClZnOCl oxenoid at DFT and MP2 levels of theory was modeled to assess how this might affect the computed geometries and free energy barriers. Several DFT functionals applied in this study predict a two-step mechanism for the epoxidation of ClZnOCl, which is consistent with MP2/cc-pVDZ calculations, excepting B3P86/CEP-31G(d) and B3LYP/CEP-31G(d), in which the
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
71
Scheme 8. BP86/CEP-31G(d) calculated free energies (kcal/mol) for two-step epoxidation of ethylene by IZnOI.
Scheme 9. BP86/CEP-31G(d) free energies (kcal/mol) calculated free energies (kcal/mol) for two-step epoxidation of ethylene by ClZnOCl.
mechanism of the reaction is predicted to possess a singlestep mechanism. According to Table 3, it can be noted that BP86/CEP-31G(d) method produced the most similar geometries to those derived from MP2 for both the initial oxometalation and subsequent ring-closure. In terms of energy, BP86/CEP-31G(d) produces similar energy barriers for the second step (the rate determining step) to those derived from MP2 calculations. It was noted that coupled clusters simulations – CCSD(T) with cc-pVDZ and cc-PVTZ basis sets – predicted energies intermediate of those from MP2 and DFT calculations. This suggests that MP2 was overcorrected and DFT was undercorrected relative to couple clustered theory. Moving forward to larger model systems for catalytic group transfer by late transition metals, with more realistic supporting ligands L and leaving groups X, the present research suggests that it would be desirable to employ MP2//DFT or, if feasible, CCSD(T)//DFT levels of theory. (2) The results from the present calculations predicted that the conversion of ethylene to cyclopropane with LZnCH2X occurs by a single-step reaction, which is in agreement with
previous computational studies of related cyclopropanation reactions [8,9,11]. It is interesting that the mechanism is system-dependent. Changing the ‘‘oid’’ transfer group from LZnCH2X carbenoid to LZnNHX nitrenoid (L = X = I or Cl) leads to group transfer to ethylene through a single-step mechanism with IZnNHI nitrenoid and a two-step mechanism with ClZnNHCl nitrenoid. The LZnOX oxenoid (L = X = I, Cl) promoted epoxidation reaction was computed to occur through a two-step mechanism. (3) Cyclopropanation with a IZnCH2I carbenoid and aziridination by a IZnNHI nitrenoid proceed through a concerted pathway via a butterfly transition state, which is more asynchronous for the latter. The energy analysis for ethylene aziridination by IZnNHI shows that the reaction barrier is 5.4 kcal/mol lower than that for the corresponding cyclopropanation reaction and is 11.5 kcal/mol more exergonic than cyclopropanation, as well. On the other hand, the reaction of C2H4 + ClZnNHCl nitrenoid goes through a two-step mechanism in which the first TS is synchronous and the second TS is asynchronous; the second step is the rate determining step. The calculations predict the epoxidation of
72
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73
ethylene with IZnOI and ClZnOCl proceeds via a two-step mechanism for both oxenoid models entailing initial oxometalation and subsequent ring closure. In both epoxidation reactions, oxometalation seems to be more synchronous than ring closure (Table 3). The reaction pathways also confirm the second step as the rate determining step with the energy barrier of 15 kcal/mol for both epoxidation reactions with IZnOI and ClZnOCl oxenoids, which is 6 kcal/mol less than that of the aziridination with ClZnNHCl nitrenoid. The epoxidation of ethylene with ClZnOCl is 14 kcal/mol more exergonic than with IZnOI oxenoid. (4) Considering the effect of the leaving group X and supporting ligand L on the cyclization mechanism, the aziridination of ethylene with IZnNHI nitrenoid (L = X = I) operates via a concerted pathway. As noted above, changing L and X from iodide to chloride changes the mechanism from a concerted one-step to a two-step nitrenoid transfer. Therefore, additional tests were carried out for ‘‘mixed’’ nitrenoid models, ClZnNHI and IZnNHCl, for which BP86/CEP-31G(d) calculations predicted a two-step mechanism for the aziridination of ethylene with both mixed nitrenoid models. The latter calculations suggest that the leaving group and supporting ligand are equally important in terms of impacting the preferred mechanism of group transfer for the nitrenoid case. Taken together, the various pieces of evidence point to group transfer by zinc nitrenoid complexes that occupies a position that is intermediate between its oxenoid and carbenoid congeners on the transition state continuum depicted in Scheme 4. The sensitivity of the reaction pathway – both in terms of the synchronicity and the computed activation barriers – to the level of theory as well as modification of the supporting ligand (L) and leaving group (X) suggests that similar catalysts are intriguing candidates for aziridination of olefins under mild conditions, and certainly worthy of additional scrutiny above and beyond the few experimental studies that have been reported [18a,c].
[5]
[6]
[7]
[8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
[19] [20] [21]
Acknowledgments The authors acknowledge the NSF – United States for support of this research through grant CHE-1057785. The authors also thank Dr. George Schoendorff (UNT Chemistry/CASCaM) for helpful discussions.
[22] [23]
Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.comptc.2015.01. 005. References [1] C.J. Thibodeaux, W. Chang, H. Liu, Chem. Rev. 112 (2012) 1681–1709. [2] (a) C. Rodriguze-Garcia, A. Oliva, R.M. Ortuno, V. Branchhadell, J. Am. Chem. Soc. 123 (2001) 6157–6163; (b) C.M. Che, J.S. Huang, F.W. Lee, Y. Li, T.S. Lai, H.L. Kwong, P.F. Teng, W.S. Lee, W.C. Lo, S.M. Peng, Z.Y. Zhou, J. Am. Chem. Soc. 123 (2001) 4119–4129; (c) J. Salaun, In Small Ring Compounds in Organic Synthesis, VI, in: A. de Meijere (Ed.), vol. 207, Springer, Berlin, 2000, pp. 1–67; (d) D.L. Boger, M.W. Ledeboer, M. Kume, Q. Jin, Angew. Chem., Int. Ed. 38 (1999) 2424–2426; (e) M.P. Doyle, M.A. McKervey, T. Ye, Modern Catalytic Methods for Organic Synthesis with Diazo Compounds, Wiley, New York, 1998. [3] R.M. Waters, E.R. Schmidt, J. Am. Med. Assoc. 103 (1934) 975–983. [4] (a) J.B. Sweeney, In Science of Synthesis, in: E. Schaumann, D. Enders (Eds.), vol. 40a, Georg Thieme Verlag, Stuttgart, Germany, 2008, p. 643; (b) A. Padwa, S.S. Murphee, Prog. Heterocycl. Chem. 15 (2003) 3792–3798; (c) J.B. Sweeney, Chem. Soc. Rev. 31 (2002) 247–258; (d) B. Zwanenburg, P. ten Holte, Top. Curr. Chem. 216 (2001) 93–124;
[24] [25]
[26] [27]
(e) U.M. Lindstrom, P. Somfai, Synthesis 1 (1998) 109–117; (f) P. Somfai, J. Ahman, Targets Heterocycl. Syst. 3 (1999) 341–367. (a) C. Botuha, F. Chemla, F. Ferreira, A. Pérez-Luna, Aziridines in natural product synthesis, in: K.C. Majumdar, S.K. Chattopadhyay (Eds.), Heterocycles in Natural Product Synthesis, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2011, http://dx.doi.org/10.1002/9783527634880.ch1; (b) K. Harada, K. Tomita, K. Fujii, K. Masuda, Y. Mikami, K. Yazawa, H. Komaki, J. Antibiot. 57 (2004) 125–135; (c) T. Tsuchida, R. Sawa, Y. Takahashi, H. Iinuma, T. Sawa, H. Naganawa, T. Takeuchi, J. Antibiot. 48 (1995) 1148–1152; (d) F.M.D. Ismail, D.O. Levitsky, V.M. Dembitsky, Eur. J. Med. Chem. 44 (2009) 3373; (f) J.O. Metzger, S. Furmeier, Eur. J. Org. Chem. 26 (2003) 649–659. (a) W. McCoull, F.A. Davis, Synthesis 10 (2000) 1347–1365; (b) X.E. Hu, Tetrahedron 60 (2004) 2701–2743; (c) M. Pineschi, Eur. J. Org. Chem. (2006) 4979–4988; (d) DM. Hodgson, P.G. Humphreys, S.R. Hughes, Pure Appl. Chem. 79 (2007) 269–279. (a) R.A. Sheldon, J.K. Kochi, Metal Catalyzed Oxidations of Organic Compounds, Academic Press, New York, 1981; (b) C.L. Hill, Advances in Oxygenated Processes, JAI, Lond. 1 (1988) 1–30; (c) M. Hudlucky, Oxidations in Organic Chemistry; ACS Monograph Series, American Chemical Society, Washington, DC, 1990; (d) K.A. Jørgensen, Transition Metals For Organic Synthesis, in: M. Beller, C. Bolm (Eds.), vol. 2, Wiley-VCH, Weinheim, 1998, p. 157; (e) R.C. Larock, Comprehensive Organic Transformations: A Guide to Functional Group Preparations, second ed., Wiley-VCH, New York, 1999. H.E. Simmons, R.D. Smith, J. Am. Chem. Soc. 80 (1958) 5323–5324. H.E. Simmons, R.D. Smith, J. Am. Chem. Soc. 81 (1959) 4256–4265. G.L. Closs, R.A. Moss, J. Am. Chem. Soc. 86 (1964) 404–4053. Z. Ke, C. Zhao, D.L. Philips, J. Org. Chem. 72 (2007) 848–860. G.A. Molander, J.B. Etter, P.W. Zinke, J. Am. Chem. Soc. 109 (1987) 453–463. G.A. Molander, L.S. Harring, J. Org. Chem. 54 (1989) 3525–3532. K. Maruoka, Y. Fukutani, H. Yamamoto, J. Org. Chem. 50 (1985) 4412– 4414. A.K. Yudin, Aziridines and Epoxides in Organic Synthesis, Wiley-VCH, Weinheim, 2006. S.E. Larson, L.G. Li, G.B. Rowland, D. Junge, R.C. Huang, H.L. Woodcock, J.C. Antilla, Org. Lett. 13 (2011) 2188–2191. X.Y. Li, N. Chen, X.J. Xu, Synthesis 20 (2010) 3423–3428. (a) B. Kalita, A.A. Lamar, K.M. Nicholas, Chem. Commun. (2008) 4291–4293; (b) MM. Abu-Omar, Dalton Trans. 40 (2011) 3435–3444; (c) J.W.W. Chang, T.M.U. Ton, P.W.H. Chan, Chem. Rec. 11 (2011) 331–357; (d) A.A. Lamar, K.M. Nicholas, J. Org. Chem. 75 (2010) 7644–7650. D.A. Evans, M.M. Faul, M.T. Bilodeau, J. Org. Chem. 56 (1991) 6744–6746. M.M. Montero-Campillo, M.N.D.S. Cordeiro, Int. J. Quantum Chem. 113 (2013) 2002–2011. (a) P.-H. Ko, T.-Y. Chen, J. Zhu, K.-F. Cheng, S.-M. Peng, C.-M.J. Che, Chem. Soc. Dalton Trans. (1995) 2215–2219; (b) P Müller, C. Baud, Y. Jacquier, Can. J. Chem. 76 (1998) 738–750. S.A. Cramer, D.M. Jenkins, J. Am. Chem. Soc. 133 (2011) 19342–19345. (a) L. Maestre, W.M.C. Sameera, M.M. Díaz-Requejo, F. Maseras, P.J. Pe´rez, J. Am. Chem. Soc. 135 (2013) 1338–1348; (b) D. Mansuy, J.-P. Mahy, A. Dureault, G. Bedi, P. Battioni, J. Chem. Soc., Chem. Commun. (1984) 1161–1163; (c) J-P. Mahy, G. Bedi, P. Battioni, D. Mansuy, J. Chem. Soc., Perkin Trans. 2 (1988) 1517–1524; (d) S.-M. Au, J.-S. Huang, W.-Y. Yu, W.-H. Fung, C.-M. Che, J. Am. Chem. Soc. 121 (1999) 9120–9132; (e) J.V. Ruppel, J.E. Jones, C.A. Huff, R.M. Kamble, Y. Chen, X.P. Zhang, Org. Lett. 10 (2008) 1995–1998; (f) P. Múller, C. Baud, Y. Jacquier, Tetrahedron 52 (1996) 1543–1548; (g) D.A. Evans, M.M. Faul, M.T. Bilodeau, J. Am. Chem. Soc. 116 (1994) 2742– 2753; (i) Z. Li, K.R. Conser, E.N. Jacobsen, J. Am. Chem. Soc. 115 (1993). 5326–5317; (j) D.A. Evans, M.M. Faul, M.T. Bilodeau, B.A. Anderson, D.M. Barnes, J. Am. Chem. Soc. 115 (1993) 5328–5329; (k) Y. Cui, C. He, J. Am. Chem. Soc. 125 (2003) 16202–16203; (m) Z. Li, X. Ding, C. He, J. Org. Chem. 71 (2006) 5876–5880; (l) D.T. Smith, J.T. Njardarson, Angew. Chem. Int. Ed. 53 (2014) 4278–4280. (a) J.A. Halfen, Curr. Org. Chem. 9 (2005) 657–669; (b) P. Múller, C. Fruit, Chem. Rev. 103 (2003) 2905–2919. (a) A. Corma, I. Domínguez, A. Doménech, V. Fornés, C.J. Gómez-García, T. Ródenas, M.J. Sabater, J. Catal. 265 (2009) 238–244; (b) M. Colladon, A. Scarso, P. Sgarbossa, R.A. Michelin, G. Strukul, J. Am. Chem. Soc. 129 (2007) 7680–7689; (c) H. Tanaka, H. Nishikawa, T. Uchida, T. Katsuki, J. Am. Chem. Soc. 132 (2010) 12034–12041; (d) M. Tada, S. Muratsugu, M. Kinoshita, T. Sasaki, Y. Iwasawa, J. Am. Chem. Soc. 132 (2010) 713–724; (e) D. Banerjee, R.V. Jagadeesh, K. Junge, M.M. Pohl, J. Radnik, A. Brückner, M. Beller, Angew. Chem. Int. Ed. 53 (2014) 4359–4363. M. Haruta, B.S. Uphade, S. Tsubota, A. Miyamoto, Res. Chem. lntermed. 24 (1998) 329–336. (a) F. Bernardi, A. Bottoni, G.P. Miscione, J. Am. Chem. Soc. 119 (1997) 12300; (b) E. Nakamura, A. Hirai, M. Nakamura, J. Am. Chem. Soc. 120 (1998) 5844.
S.K. Khani, T.R. Cundari / Computational and Theoretical Chemistry 1056 (2015) 61–73 [28] H. Lebel, J.-F. Marcoux, C. Molinaro, A.B. Charette, Chem. Rev. 103 (2003) 977– 1050. [29] (a) H.C. Stiasny, R.W. Hoffmann, Chem. Eur. J. 1 (1995) 619–624; (b) A. Hirai, M. Nakamura, E. Nakamura, Chem. Lett. (1998) 927–928; (c) M Nakamura, A. Hirai, E. Nakamura, J. Am. Chem. Soc. 125 (2003) 2341– 2350. [30] (a) L. Friedman, H. Shechter, J. Am. Chem. Soc. 81 (1959) 5512; (b) M.E. Volpin, Y.D. Koreshov, V.G. Dulanova, D.N. Kursanov, Tetrahedron 18 (1962) 107. [31] Z.H. Li, Z. Ke, C. Zhao, Z.Y. Geng, Y.C. Wang, D.L. Phillips, Organometallics 25 (2006) 3735–3742. [32] G. Boche, J.C.W. Lohrenz, Chem. Rev. 101 (2001) 697–756. [33] W.-H. Fang, D.L. Philips, D.–Q. Wang, Y.–L. Li, J. Org. Chem. 67 (2002) 154–160. [34] L.M. Pratt, B.K. Mai, B.R. Ramachandran, J. Org. Chem. 77 (2012) 8605–8614. [35] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery, Jr., J.E. Peralta, F. Ogliaro,
73
M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, Ö. Farkas, J.B. Foresman, J.V. Ortiz, J. Cioslowski, D.J. Fox, Gaussian 09, Revision A.02, Gaussian Inc., Wallingford CT, 2009. [36] (a) S. Wiese, Y.M. Badiei, R.T. Gephart, S. Mossin, M.S. Varonka, M.M. Melzer, K. Meyer, T.R. Cundari, T.H. Warren, Angew. Chem. 122 (2010) 9034–9039; (b) R.T. Gephart, C.L. McMullin, N.G. Sapiezynski, E.S. Jang, M.J.B. Aguila, T.R. Cundari, T.H. Warren, J. Am. Chem. Soc. 134 (2012) 17350–17353; (c) E.S. Jang, C.L. McMullin, M. Káß, K. Meyer, T.R. Cundari, T.H. Warren, J. Am. Chem. Soc. 136 (2014) 10930–10940; (d) Z. Ke, T.R. Cundari, Organometallics 29 (2010) 821–834; (e) H.V.R. Dias, M. Fianchini, T.R. Cundari, C.F. Campana, Angew. Chem. 120 (2008) 566–569; (f) D.M. Ball, C. Buda, A.M. Gillespie, D.P. White, T.R. Cundari, Inorg. Chem. 41 (2002) 152–156.