Cobalt Catalyst Determines Regioselectivity in Ring Opening of Epoxides with Aryl Halides.

Ring-opening of epoxides furnishing either linear or branched products belongs to the group of classic transformations in organic synthesis. However, the regioselective cross-electrophile coupling of aryl epoxides with aryl halides still represents a key challenge. Herein, we report that the vitamin B12/Ni dual-catalytic system allows for the selective synthesis of linear products under blue-light irradiation, thus complementing methodologies that give access to branched alcohols. Experimental and theoretical studies corroborate the proposed mechanism involving alkylcobalamin as an intermediate in this reaction.

Introduction

Driven by high demand for sustainable and efficient reactions, the dis<span class="Chemical">covery of selective reactivity patterns remains a key challenge. As epoxides are crucial building blocks in the synthesis of nonsymmetrical alcohols, their regioselective reactions have been intensively studied.[1−3] In particular, considerable attention has been recently devoted to the utilization of epoxides in cross-electrophile couplings leading, in general, to regioisomeric (linear and branched) products (Scheme ).[4−7] It has been shown that the innately electrophilic epoxides can be transformed into radicals and, as such, be involved in a transition-metal-catalytic cycle.[7,8] Depending on reaction conditions, the initial nucleophilic attack occurs predominantly at either the terminal or internal carbon atom.

Scheme 1

Regioselective Nickel-Catalyzed Cross-Electrophile Coupling of Epoxides with Aryl Halides

In 2014, Weix and co-workers developed a <span class="Chemical">nickel-catalyzed, regiodivergent cross-electrophile coupling of epoxides with various halides and triflates.[6] For aliphatic epoxides (Scheme , upper part), the regioselectivity of the ring-opening step depends on the cocatalyst used. Sodium iodide promotes the formation of a linear product. The nucleophilic attack of the iodide anion at the less substituted carbon atom affords iodohydrin, which in turn undergoes reduction and Ni-catalyzed coupling with an electrophile. On the other hand, in the presence of a titanocene cocatalyst secondary alkyl radicals are generated, facilitating the formation of branched products.[9,10]Aryl epoxides, however, react predominantly at the benzylic position, regardless of the conditions employed. A similar reactivity pattern has been recently reported by the Doyle group, who used organic iodides and the Ti/Ni/photoredox catalytic system in ring-opening reactions of three major classes of epoxides, namely, aryl, aliphatic, and bicyclic (Scheme , lower part).[11] By changing a nickel complex, the authors were able to transform aliphatic epoxides into linear products, while aryl epoxides selectively formed branched ones. Despite the enormous importance of these contributions, the synthesis of linear products from aryl epoxides via cross-electrophile coupling still represents an unsolved challenge.

Our recent work on the alkyla<span class="Chemical">tion of strained molecules showed that cobalt catalysis opens the path to a polarity-reversal strategy for radical couplings.[12] We questioned whether it would be possible to adapt this methodology to achieve selective reactions of epoxides. Herein, we disclose that the nucleophilicity of Co(I) species along with sterically restricted side chains allows generating C-radicals from epoxides in a selective manner and engage them in Ni-catalyzed cross-coupling.

Results and Discussion

Design of the Catalytic System

Vitamin B12 (1, <span class="Chemical">cobalamin) is a natural cobalt complex of remarkable stability and high biological importance.[13−15] Due to the unique ability to form light-sensitive cobalt–carbon bonds, vitamin B12 (1) and its hydrophobic and amphiphilic derivatives 2 and 3 (Scheme A)[16−21] have also been adopted for synthetic chemistry and used as redox mediators for the generation of various radicals.[22,23] We assumed that a nucleophilic Co(I) complex that forms upon the reduction of cobalamin should open electrophilic epoxides, generating alkyl cobalamins (Scheme B). Such intermediates, upon light irradiation, undergo the homolytic Co–C bond cleavage to give alkyl radicals, which can be engaged in a number of both radical reactions and transition-metal-catalyzed cross-couplings. Importantly, from the viewpoint of regioselective design, a bulky vitamin Bcatalyst should attack an epoxide from the less sterically hindered side. This kinetic factor may prevail over the high thermodynamic preference for stabilized benzyl radicals and thus allow the selective formation of primary radicals of type III.

Scheme 2

(A) Structures of Cocatalysts: Vitamin B12 and Derivatives; (B) Proposed Mechanistic Concept

To examine our hypothesis, in the first instance, we theore<span class="Chemical">tically investigated the possible formation of alkyl radicals via a sequence of the epoxide ring opening with reduced vitamin B12 (Co(I) complex) followed by homolytic cleavage of the Co–C bond in alkyl cobalamin II. DFT calculations were performed with Gaussian 16.[24] Geometry optimizations were computed at the BP86/6-31G(d) level of theory with the D3 version of Grimme’s empirical dispersion correction and solvation (acetone) with the SMD model. Frequency analysis was performed at the same level to provide correction to thermodynamic functions and confirm the nature of optimized structures (minima and transition states featured zero or one imaginary frequency, respectively). Single-point energies were computed at the BP86/6-311++G(2df,p) level of theory with the D3 version of Grimme’s empirical dispersion correction and solvation (acetone) with the SMD model. Several hybrid and long-range corrected functionals were tested for the model reaction (see Supporting Information (SI) for details). BP86 was, however, selected for further studies due to good performance reported for both ground and exited state calculations of cobalamin systems.[25−30] We performed calculations approximating the structure of vitamin B12 (1) with a Co-corrin complex bearing 15 methyl groups, reflecting the substitution pattern at the periphery of the macrocyclic ring (Figure ).

Figure 1

Calculated Gibbs free energy profile for the reaction of styrene oxide with the Co(I)-corrin complex.

Calculated Gibbs free energy profile for the reaction of <span class="Chemical">styrene oxide with the Co(I)-corrin complex.

The calculated Gibbs free energy profile for the benchmark reaction of <span class="Chemical">styrene oxide (5a) with the Co-corrin complex is depicted in Figure . Two paths, involving the nucleophilic attack on either side of the epoxide, were considered. In line with our assumptions, the ring opening of the epoxide with the nucleophilic Co(I) complex should proceed at the less hindered terminus with a 43.7 kJ/mol barrier, accessible even under mild conditions (black path). The barrier for the analogous reaction at the more hindered side is ∼8 kJ/mol higher (gray path). Sterically driven differences in the reactivity might be even more pronounced for native vitamin B12 or its derivatives compared to the selected model, due to presence of more sizable substituents at the corrin ring. Then, the resulting Co(III) complex (I) is protonated, providing intermediate II. As expected for alkyl cobalamins, the Co–C(sp3) bond in IIa and IIb is relatively weak and quite vulnerable to homolytic cleavage toward alkyl radical IIIa or IIIb and a Co(II) complex (ΔG = 141.9 and 82.6 kJ/mol, respectively). In particular, IIa could undergo Co–C photodissociation, presumably through the mechanism proposed by Kozlowski, involving generation of the singlet radical pair from the first electronically excited state (S1).[31−33]

The lowest singlet (IIa-S1) vertically excited states of intermediate IIa were found at 2.20 eV (212.5 kJ/mol, TD BP86-D3/6-311++G(2df,p)), while the relaxed S1 state lies 28.2 kJ/mol lower and features elonga<span class="Chemical">tion of the Co–C bond by 0.22 Å. Noticeably, due to the preference for the nucleophilic attack at the less hindered side of the epoxide, the above-described path (black) should provide access to a 2-hydroxy-2-phenyl ethyl radical (IIIa), even though isomeric benzyl radical IIIb is thermodynamically more stable by 47.2 kJ/mol.

To support theoretical studies, the reduc<span class="Chemical">tive photochemical ring opening of styrene oxide (5a) in the presence of a hydrophobic vitamin B12 derivative, HME (3), was performed (Scheme ). The selected Co complex 3 allows convenient monitoring of reactive intermediates by ESI mass spectrometry due to its tendency to undergo facile ionization.

Scheme 3

Vitamin B12-Catalyzed Ring Opening of Epoxide 5

Indeed, the formation of intermediate <span class="Chemical">alkyl-cobalt(III) complex II was observed by HR-MS (m/z = 1157.5149 [M + H]+, see the SI, Section 6.2), which is in good agreement with previous reports by Scheffold[34−38] and Rusling,[39] who used vitamin B12 (1) for isomerization of symmetrical epoxides to allyl alcohols. Satisfyingly, alcohol 9a with a −OH group at the benzylic position formed just after 30 min. These results corroborate the proposed mechanistic concept in which vitamin B12 opens the aromatic epoxide from the less hindered side at the thermodynamic expense of forming the less stable radical III in the subsequent light-induced cleavage step.

Knowing that B12 catalysis can be merged with <span class="Chemical">metal-catalyzed reactions,[12] we next evaluated the feasibility of incorporating the generated alkyl radicals in the Ni catalytic cycle. Adding electrophilic aryl halides should enable cross-electrophile coupling and thus provide a convenient method for the carbon–carbon bond formation.[40] The plausible mechanism for the reaction of epoxides with aryl halides in the presence of the B12/Ni catalytic system based on literature reports is outlined in Scheme B.[7,41−43] The coupling requires the cooperation of both transition metal complexes (Co and Ni) that are activated by Zn/NH4Cl.[44,45] The oxidative addition of aryl halide to Ni(0) produces aryl nickel(II) species IV, which undergoes subsequent alkylation with radical III and generates intermediate V. Alternatively, the same Ni(III) species can originate from the interception of alkyl radical III by Ni(0), preceding the oxidative addition, as has been recently proposed by Molander and Kozlowski.[46] Both these possible pathways are followed by irreversible reductive elimination, leading to the regioselective formation of a linear product. Cobalt(II) and nickel(I) complexes are regenerated to Co(I) and Ni(0) with Zn, thereby closing the cycles.

Styrene oxide (5a), when subjected to the reac<span class="Chemical">tion with p-iodotoluene (6a) in the presence of HME (3) and NiCl2(DME), generated desired linear product 7aa as a single regioisomer in 16% yield (Scheme ). The replacement of HME (3) with native vitamin B12 increased the yield up to 41%. Noteworthy, the reaction without any cobalt complex added not only was lower-yielding but also led to a mixture of two regioisomers with a predominance of the branched product. This result clearly shows the decisive influence of the cobalt cocatalysis on the selectivity of this transformation. Control experiments confirmed the dual-catalytic and light-induced nature of the process, while the addition of a radical trap (TEMPO) supported its radical character (see SI).

Scheme 4

Proof-of-Concept Experiments

Conditions: epoxide (5a, 0.2 mmol), aryl halide (6a, 1.5 equiv), NiCl2(DME) (20 mol %), Zn (3 equiv), NH4Cl (3 equiv), dtbbpy (40 mol %), dry NMP (c = 0.1 M), blue LED (single diode, 10 W), 30 min.

Condi<span class="Chemical">tions: epoxide (5a, 0.2 mmol), aryl halide (6a, 1.5 equiv), NiCl2(DME) (20 mol %), Zn (3 equiv), NH4Cl (3 equiv), dtbbpy (40 mol %), dry NMP (c = 0.1 M), blue LED (single diode, 10 W), 30 min.

Noteworthy, the reaction without any <span class="Chemical">cobalt complex added not only was lower-yielding but also led to a mixture of two regioisomers with a predominance of the branched product. This result clearly shows the decisive influence of the cobalt cocatalysis on the selectivity of this transformation. Control experiments confirmed the dual-catalytic and light-induced nature of the process, while the addition of a radical trap (TEMPO) supported its radical character (see SI).

Optimization

Next, we turned our attention toward the synthe<span class="Chemical">tic utility of the developed method. The reaction was optimized with respect to cobalt and nickel catalysts, solvent, ligand, and reducing system, providing the desired product 7aa in 60% yield (Table , entry 1).

Table 1

Optimization Studies of the Cross-Electrophile Ring Opening of Epoxidesa

entry	deviation from the standard conditions	yield (%) 7aaa
1	none	60
2	HME instead of B12	57
3	Co(acac)3 instead of B12	5
4	CoCl2 instead of B12	7
5	Co(dmgH)2Cl(py) instead of B12	8
6	Co(dmgH)2iPr(py) instead of B12	11
7b	Mn instead of Zn	30
8	NiCl2 instead of NiCl2(DME)	36
9	Ni(acac)2 instead of NiCl2(DME)	33
10	Ni(OTf)2 instead of NiCl2(DME)	39
11	1,10-phenanthroline instead of dtbbpy	24
12	terpyridine instead of dtbbpy	13
13	no water added	53

Conditions: epoxide (5, 0.2 mmol), aryl halide (7, 1.5 equiv), B12 (5 mol %), NiCl2(DME) (20 mol %), Zn (1.5 equiv), NH4Cl (3 equiv), dtbbpy (40 mol %), H2O (1.1 equiv), dry NMP (c = 0.1 M), time 30 min, blue LED (single diode, 10 W) (for more details see SI).

Mn (1.5 equiv), TMSCl (0.2 equiv), dmgH = dimethylglyoxime, dtbbpy = 4,4′-di-tert-butylbipyridine.

Condi<span class="Chemical">tions: epoxide (5, 0.2 mmol), aryl halide (7, 1.5 equiv), B12 (5 mol %), NiCl2(DME) (20 mol %), Zn (1.5 equiv), NH4Cl (3 equiv), dtbbpy (40 mol %), H2O (1.1 equiv), dry NMP (c = 0.1 M), time 30 min, blue LED (single diode, 10 W) (for more details see SI).

Mn (1.5 equiv), TMSCl (0.2 equiv), <span class="Chemical">dmgH = dimethylglyoxime, dtbbpy = 4,4′-di-tert-butylbipyridine.

We found that the addition of <span class="Chemical">water (1.1 equiv) improved the yield of the reaction, while kinetic studies allowed us to determine the optimal reaction time (30 min, for details, see SI). The use of hydrophobic HME (3) instead of the parent vitamin B12 had little impact on the optimized model reaction (entry 2), while other commonly utilized cobalt complexes (Co(acac)3, CoCl2) led to a decrease in the yield of alcohol 7aa (entries 3, 4). We have also examined cobalt dimethylglyooximate (dmg) complexes, which have been used by Pattenden[47] and Morandi[48] in regioselective cobalt-catalyzed coupling of aliphatic epoxides with alkenes. In our system, however, both catalysts afforded the desired product 7aa only in low yields (entries 5, 6). Evaluation of reducing agents ruled out manganese or tetrakis(dimethylamino)ethylene (TDAE) as an efficient alternative to the Zn/NH4Cl system (entry 7). It also allowed establishing the optimal ratio of the two components at the 1.5 equiv: 3 equiv level. The reaction outcome did not improve in the presence of NiCl2, Ni(acac)2, or Ni(OTf)2 as well as other ligands (entries 8–12). Finally, various solvents were tested (for more details, see SI), but NMP with the addition of water (1.1 equiv) assured the highest yield (entry 13).

Detailed analysis of the reaction mixture revealed the forma<span class="Chemical">tion of byproducts aside from desired product 7aa under the optimized conditions (Scheme B). Acetophenone (8a, a side-product originating from epoxide 5a) formed in 5% yield presumably via β-hydride elimination, while styrene (10a) is obtained in 30% yield from intermediate alcohol 9.[49] Finally, the reductive elimination in the nickel cycle may account for the observed small amount of biphenyl 11.[50−52] In order to gain more insight into the reaction mechanism, we carried out the reaction with enantioenriched styrene oxide (5a) under the optimized conditions. The expected coupling product 7aa was obtained without any erosion of the stereocenter, which further supports the premise of the formation of the radical at the terminal position.

Substrate Scope

With the optimized <span class="Chemical">conditions in hand, we explored the scope and limitations of the developed method (Scheme ).

Scheme 5

Vitamin B12-Catalyzed Ring-Opening Cross-Electrophile Coupling of (a) Aryl Epoxides and (b) Alkyl Epoxides,

Conditions: epoxide (0.2 mmol), aryl halide (1.5 equiv), B12 (5 mol %), NiCl2(DME) (20 mol %), Zn (1.5 equiv), NH4Cl (3 equiv), dtbbpy (40 mol %), H2O (1.1 equiv), dry NMP (c = 0.1 M), blue LED (single diode, 10 W), 30 min (for more details see SI).

Blue LED (single diode, 3 W), 16 h.

HME (5 mol %), acetone (c = 0.1 M), blue LED (single diode, 3 W), 16 h.

Determined by GC.

Condi<span class="Chemical">tions: epoxide (0.2 mmol), aryl halide (1.5 equiv), B12 (5 mol %), NiCl2(DME) (20 mol %), Zn (1.5 equiv), NH4Cl (3 equiv), dtbbpy (40 mol %), H2O (1.1 equiv), dry NMP (c = 0.1 M), blue LED (single diode, 10 W), 30 min (for more details see SI).

HME (5 mol %), <span class="Chemical">acetone (c = 0.1 M), blue LED (single diode, 3 W), 16 h.

Monosubstituted <span class="Chemical">aryl epoxides 5a-f, bearing both electron-withdrawing and electron-donating substituents, are, in general, well-tolerated and give corresponding products 7aa–fa in 50–58% yields. However, 2-(4-methoxyphenyl)oxirane (5e) does not afford the desired product, as it decomposes rapidly under the present conditions. For disubstituted epoxides, the substitution pattern determines their reactivity. 1,1-Disubsituted epoxide 5f leads to product 7fa in 44% yield, while 1,2-disubstituted epoxide 5g remains unreactive. As far as aryl halides are concerned, under standard conditions, both electron-donating and electron-withdrawing substituents are well tolerated, giving desired products 7ab–ao in good to moderate yield (28–63%). Substitution at the 3- or 4-position of an aryl halide does not affect the reaction. In contrast, the more hindered halide, 2-iodotoluene (6c), undergoes coupling with styrene oxide (5a) in reduced reaction yield (compare 7aa, 7ab, and 7ac). Although vitamin B12 exhibits exquisite reactivity in dehalogenation reactions,[11] which often precludes the use of halogenated substrates, in our conditions product 7ad forms in 44% yield. Importantly from the standpoint of possible further functionalizations, other functional groups (hydroxyl, carbonyl, protected amine) remain unaffected. Moreover, the representative heteroaryl halide, 5-iodo-(4-methylphenylsulfonyl)indole (6l), proves to be a viable substrate in the studied reaction without any further optimization needed. The developed method is also suitable for epoxides with aliphatic substituents (Scheme ). The chain length does not impact the transformation’s outcome; the reaction with 1,2-epoxyhexane (5h) and 1,2-epoxydodecane (5i) gives products in 74% and 77% yield, respectively. We also found that aliphatic epoxide 5j, possessing a protected primary hydroxyl group, could be converted into secondary alcohol 7ja in 60% yield. The reaction with 4-(phenylsulfonyl)-1,2-epoxybutane (5k) gives corresponding product 7ka in 73% yield. The potential use of aziridines as substrates was also investigated under the developed conditions, but only low yields of the respective products were obtained (see SI). Further studies on extending our methodology to other classes of heterocycles are currently ongoing in our laboratory.

Subsequently, the scope of <span class="Chemical">aryl halides for the reaction with 1,2-epoxyhexane (5h) was explored. Substrates with both types of substituents—electron-rich and electron-deficient—on the aromatic ring afford the corresponding products 7hd–ol in satisfactory yields. The N-Boc-protected amine, alkoxy, and carbonyl functionalities are well tolerated. The reaction with 1-chloro-4-iodobenzene (6d) leads to anticipated alcohol 7hd in 51% yield. Similar to the reaction with aryl epoxides, indole-derived halide 6l proved also a competent substrate, affording 1-(1-tosyl-1H-indol-5-yl)hexan-2-ol (7hl).

Compared to monosubs<span class="Chemical">tituted substrates, bicyclic epoxide 5o was converted to the desired coupling product 7oa with a significantly lower yield.[5] Therefore, to gain a better understanding of how the reaction conditions affect the cross-electrophile coupling of disubstituted epoxides with aryl halides, additional experimental and theoretical studies were performed.

The use of hydrophobic analogue 3 instead of vitamin B12 (1) does not bring any substan<span class="Chemical">tial improvement (Table , entries 1, 2). However, with the simultaneous replacement of NMP with acetone, a 2-fold increase in the yield of 7oa was observed (entry 3). A similar trend was also present for bicyclic epoxide 5n and 1-oxaspiro[2.5]octane (5m), which provide considerably higher yields of desired alcohols 7na and 7oa in the presence of the HME/acetone system compared to vitamin B12/NMP. The main feature by which the studied Co catalysts differ is the presence/absence of the so-called “nucleotide loop” (the axial ligand located at the α face of the corrin ring with a 5′,6′-dimethylbenzimidazol (DMB) moiety) in addition to the replacement of amide into ester groups. Halpern et al. reported that in methyl malonyl-coenzyme A rearrangement switching between base-on and base-off forms of (CN)Cbl (1) changes the strength of the Co–C bond and hence the rate of its homolytic cleavage.[53] To assess if the presence of this structural element impacts opening of bicyclic epoxides, we used cobalester 2 as a Co complex. This catalyst bears a nucleotide loop in its structure, but unlike parent vitamin B12, it dissolves well in both NMP and acetone, allowing for a direct comparison. A decisive solvent’s dependence was observed, with acetone assuring a higher yield than NMP (entries 4, 5), which corroborates the sole influence of the reaction medium. Likewise, the reaction catalyzed by cobinamide 4 (amide groups, no nucleotide loop) in NMP gives similar results to reactions catalyzed by other B12 derivatives in this solvent (entry 6).

Table 2

Influence of the Co Complex Structure on the Opening of Bicyclic Epoxidesa

entry	solvent	catalyst		yield (%) 7oa
1	NMP	B12		31
		CONH2	loop
2	NMP	HME		26
		CO2Me	no loop
3	acetone	HME[12]		56
		CO2Me	no loop
4	acetone	cobalester[17]		55
		CO2Me	loop
5	NMP	cobalester		32
		CO2Me	loop
6	NMP	cobinamide[54]		35
		CONH2	no loop

Conditions: epoxide (5o, 0.2 mmol), aryl halide (6a, 1.5 equiv), Co catalyst (5 mol %), NiCl2(DME) (20 mol %), Zn (1.5 equiv), NH4Cl (3.0 equiv), dtbbpy (40 mol %), solvent (c = 0.1 M), blue LED (single diode, 3 W), 16 h.

Condi<span class="Chemical">tions: epoxide (5o, 0.2 mmol), aryl halide (6a, 1.5 equiv), Co catalyst (5 mol %), NiCl2(DME) (20 mol %), Zn (1.5 equiv), NH4Cl (3.0 equiv), dtbbpy (40 mol %), solvent (c = 0.1 M), blue LED (single diode, 3 W), 16 h.

Performed kinetic studies <span class="Chemical">contributed to a better understanding of the observed differences. The rate of the bicyclic epoxide (5o) ring opening was found to vary significantly depending on the conditions applied (Chart ). It takes 6 h to fully convert epoxide 5o in both vitamin B12- and HME-catalyzed reactions as long as NMP is used as a solvent (compare fields A and C). On the other hand, the reaction in acetone provides full conversion in less than 3 h, which, presumably, translates to greater availability of alkyl radicals at a particular time (compare fields C and D).

Chart 1

Kinetic Profile of the Opening of Bicyclic Epoxides (6o) (A, C, D) and Aliphatic Epoxide 6h (B)a,b

Condi<span class="Chemical">tions: epoxide (5, 0.2 mmol), aryl halide (6a, 1.5 equiv), Co catalyst (5 mol %), NiCl2(DME) (20 mol %), Zn (1.5 equiv), NH4Cl (3.0 equiv), dtbbpy (40 mol %), solvent (c = 0.1 M), blue LED (single diode, 3 W), 16 h, dodecane as an internal standard.

Measurements at t = 0 min refer to concentra<span class="Chemical">tions of compounds before mixing two solutions; see SI.

The reactivity of <span class="Chemical">aryl iodide toward a Ni catalyst is assumed to be at a similar level, regardless of the conditions applied. Therefore, in NMP an insufficient concentration of alkyl radicals derived from bicyclic epoxides may promote Ni-catalyzed homocoupling of aryl iodide 6a, an unproductive pathway, leading to biphenyl (11a) (fields A and C).[50,52] This side-product is observed only in the presence of the Ni complex. The higher reactivity of monosubstituted aliphatic epoxide 5h is reflected by its faster conversion as compared to bicyclic epoxide 5o (compare fields A and B). In their case, the catalyst and the solvent do not affect the reaction; yields are almost identical (74%) in both cases.

The observed reactivity pattern <span class="Chemical">corresponds well with the calculated barriers for the nucleophilic opening of the epoxides with the Co(I)-corrin complex (Figure ). In general, the Gibbs free energy of activation for the reaction of aryl- and alkyl-monosubstituted epoxides (TS1a and TS2a, 43.7 and 47.1 kJ/mol for Ph- and Me-substituted, respectively) is smaller than for more sterically demanding 1,2-disubstituted epoxides (TS3 and TS4, >70 kJ/mol). Nevertheless, bicyclic substrates 5n,o provide desired products 7na and 7oa in good yields, while epoxide 5g, for which the activation barrier is ∼3 kJ/mol higher, remains unreactive (TS3 versus TS4a) regardless of the Z/E configuration of the epoxide (TS4a vs TS4b, ΔG⧧ = 74.4 vs 78.9 kJ/mol). Additionally, the observed regioselectivity is well reflected by the energetically favored attack of the Co(I)-corrin on the less hindered side on propylene oxide (a model used for an alkyl epoxide, TS2a vs TS2b, ΔG⧧ = 47.1 vs 67.4 kJ/mol).

Figure 2

Gibbs free energy barriers for the opening of epoxides with the Co(I)-corrin complex calculated at the BP86-D3/6-311++G(2df,p)/SMD(acetone)//BP86-D3/6-31G(d)/SMD(acetone) level of theory.

Gibbs free energy barriers for the opening of epoxides with the <span class="Chemical">Co(I)-corrin complex calculated at the BP86-D3/6-311++G(2df,p)/SMD(acetone)//BP86-D3/6-31G(d)/SMD(acetone) level of theory.

Conclusions

We have developed a highly regioselective, <span class="Chemical">Co/Ni-catalyzed ring-opening reaction of epoxides with aryl halides. The scope of our method has been demonstrated in a broad range of aliphatic and aromatic epoxides. Gratifyingly, these include cyclic and disubstituted epoxides even though the Gibbs free energy of activation for their reactions are higher than for alkyl- and aryl-monosubstituted substrates. Due to the mild reaction conditions, a wide range of functional groups is well tolerated.

Only the coopera<span class="Chemical">tion of vitamin B12 as a Co catalyst with Ni catalysis assures high regioselectivity of the cross-electrophile coupling. The crucial ring opening by the Co(I) complex occurs from the less hindered side, leading to linear products.

This new methodology complements the exis<span class="Chemical">ting approaches providing access to a diverse array of substituted alcohols, which are valuable feedstock chemicals in synthetic and medicinal chemistry. Consequently, it closes the gap in the synthesis of linear and branched alcohols via cross-electrophile coupling; they are now accessible from both alkyl and aryl epoxides.