Catalytic Enantioselective Synthesis of Cyclobutenes from Alkynes and Alkenyl Derivatives

Abstract

Discovery of enantioselective catalytic reactions for the preparation of chiral compounds from readily available precursors, using scalable and environmentally benign chemistry, can greatly impact their design, synthesis and eventually manufacture on scale. Functionalized cyclobutanes and cyclobutenes are important structural motifs seen in many bioactive natural products and pharmaceutically relevant small molecules. They are also useful precursors for other classes of organic compounds such as other cycloalkane derivatives, heterocyclic compounds, stereo-defined 1,3-dienes and ligands for catalytic asymmetric synthesis. The simplest approach to make cyclobutenes is through an enantioselective [2+2]-cycloaddition between an alkyne and an alkenyl derivative, a reaction which has a long history. Yet known reactions of this class that give acceptable enantioselectivities are of very narrow scope and are strictly limited to activated alkynes and highly reactive alkenes. Here we disclose a broadly applicable enantioselective [2+2]-cycloaddition between wide variety of alkynes and alkenyl derivatives, two of the most abundant classes of organic precursors. The key cycloaddition reaction employs catalysts derived from readily synthesized ligands and an earthabundant metal, cobalt. Over 50 different cyclobutenes with enantioselectivities in the range of 86–97% ee are documented. With the diverse functional groups present in these compounds, further diastereoselective transformations are easily envisaged for synthesis of highly functionalized cyclobutanes and cyclobutenes. Some of the novel observations made during these studies including a key role of a cationic Co(I)-intermediate, ligand and counter ion effects on the reactions, can be expected to have broad implications in homogeneous catalysis beyond the highly valuable synthetic intermediates that are accessible by this route.

Graphical Abstract

INTRODUCTION

Recent incisive analyses of reactions¹ and molecules² of interest to medicinal chemists have validated the notion that molecular complexity, measured by fraction of saturated carbons (Fsp3) and the presence of chiral centers, correlate with success as compounds move from discovery, through clinical testing, to drugs. About a third of the compounds whose biological assays were analyzed had at least one chiral center.¹ The improved clinical efficacy has been ascribed to solubility, diminished promiscuity towards receptors, and occasionally, to better transport properties across biological barriers. Practical considerations in the preparation and screening of large and diverse array of structures, and, eventually manufacturing of the successful candidates from readily available precursors, provide strong justification for research into efficient and enantioselective synthetic methods.³ Functionalized cyclobutanes and cyclobutenes are important structural motifs seen in many bioactive natural products and pharmaceutically relevant small molecules (Figure 1, A).⁴ They are also useful precursors for other classes of organic compounds such as other cycloalkane derivatives, heterocyclic compounds, stereo-defined 1,3-dienes and ligands for catalytic asymmetric synthesis.⁵ Even though direct synthesis of suitably functionalized cyclobutane precursors from readily available starting materials has been the subject of a burgeoning area of research,⁶ there is considerable room for improvement with respect to diversity of functional groups on the ring and stereoselectivity associated with the ring formation. In this regard, a class of compounds with enormous potential for diversification are the chiral 3-substituted cyclobutenes (Figure 1, B), which allow further modification of the small ring through a myriad of ways involving the double bond,⁷ the G group,⁸ or through activation of the ring C-H bonds.⁹ The simplest approach to making cyclobutenes is through an enantioselective [2+2]- cycloaddition between an alkyne and an alkenyl derivative, a reaction with a long history, starting with mostly addition of reactive alkynes to bicyclic¹⁰ or activated¹¹ alkenes with the notable exceptions of additions of activated alkynes to cyclopentene (Hilt)^12a and a variety of styreneyl alkenes to enynes (Ogoshi),^12b both giving racemic products. Thus, there are no examples in the literature of a broadly applicable enantioselective version of this [2+2]-cycloaddition process. Reactions that give acceptable enantioselectivities are limited to specialized substrates, most often strained alkenes or activated alkynes (example: Figure 1, C). ¹³ Here we disclose a broadly applicable enantioselective [2+2]-cycloaddition between wide spectrum of alkenyl derivatives and alkynes (Figure 1, D).

Figure 1.

A. Cyclobutane and cyclobutene motifs occur in many medicinally important small molecules and natural products. B. Cyclobutenes are excellent precursors for cyclobutanes and other useful intermediates. C. Current methods for their direct preparation via [2+2]-cycloaddition are applicable only to activated alkenes and/or activated alkynes. D. This work documents a practical, catalytic enantioselective (enantiomeric excess: 86–97%, ~50 examples) approach to the most diverse set of cyclobutenes reported to-date.

RESULTS AND DISCUSSION

For the enantiopure, substituted cyclobutenes (Figure 1, B) as the targets, we recognized that none of the existing methods^{11c.11e–g.13a} including our own recently disclosed^7a enantioselective cobalt-catalyzed cycloaddition between an enyne (1) and ethylene (Eq 1a) are suitable. The use of ethylene as the 2p-component necessarily leaves two prochiral carbons of the cyclobutene unfunctionalized. An attractive way to circumvent this problem would be to access precursor cyclobutenes (6) with the use of alkenyl derivatives (5) in place of ethylene (Eq 1b). This further raises the intriguing possibility of preparing these chiral 3-substituted cyclobutenyl derivatives via enantioselective catalysis. However, the application of reaction conditions described in Eq 1a led to complex mixtures with very little of the desired products (Eq 1b), prompting further investigations into a new strategy.

(1a)

(1b)

A New Reaction and Optimization Studies

Following many failed attempts to modify the reaction conditions described in Eq 1a, we wondered if the incompatibility of the Lewis bases in the substrates was the problem, and if so, could it be addressed by activation of the reagents by a cationic Co(I) species, an intermediate that was implicated in our recent 1,2-hydroboration of prochiral 1,3-dienes.¹⁴ Initial scouting experiments (Table 1) were conducted with a prototypical alkyne, 4-octyne (4a) and two acrylates, methyl acrylate (5a) and 1,1,1-trifluoroethylacrylate (5b). Concurrently, we also studied the corresponding reaction between a more reactive enyne and methyl acrylate (Table 2). The most pertinent results from these studies are summarized in the Tables 1 and 2. Details of initial optimization of the reaction conditions, from which the results in Tables 1 and 2 are abbreviated, can be found in the Supporting Information (See Tables S1–S7, pp. S12-S20). These experiments include effects of the following parameters on the efficiency and selectivity of the reactions: (i) cobalt(II) halide (Br or Cl) complexes of 1,n-bis-diphenylphosphinoalkanes of varying bite angles [Ph₂P–(CH₂)_n-PPh₂, n = 1–5)], including chiral variants such as chiraphos, BDPP, BDPH, DIOP, and other chelates with different backbone motifs such as dppf, biaryl, Josiphos ligands and phosphino-oxazoline ligands [see Figure 2 for structures of the ligands and Figure S1 (p. S12) in the Supporting Information, which contains a complete list of all ligands]; (ii) zinc, manganese, and, various 1,4-bis-trimethylsilyldihydropyrazines ^14,15 as reducing agents; (iii) activators such as ZnX₂, AgOTf, AgSbF₆, InBr₃, sodium tetrakis-[3,5-(bis-trifluoromethyl)phenyl]borate (NaBARF); (iv) solvents, dichloromethane (DCM), 1,2-dichloroethane (DCE), toluene, hexanes, THF, ether, ethyl acetate or acetonitrile; (v) reaction temperature. In general, it was found that the reactions proceed best at or near room temperature in non-coordinating solvents such as DCM, DCE or toluene, with NaBARF as the most suitable activator. As expected, none of the solvents containing heteroatoms were suitable for this reaction. For enantioselective reactions, toluene was found to be a better solvent (giving 5–10% higher enantioselectivities vis-à-vis DCM) in several instances.

(2)

(3)

Table 1.

Optimization of Co-catalyzed enantioselective [2+2]-cycloadditions between acrylates and an alkyne (4a) ^a

	entry	ligandb	conv. (%)	6a (%)	7a (%)	ee (6)
	1	dppp	55	36	15	0
	2	dppf c	40	35	0	0
	3	(S)-BINAP	100	73	23	84
5a	4	(S,S)-BDPP	100	74	22	16
	5	L1	5	5	-	0
	6	L2	55	25	28	-
	7	L8 d	95	87	0	90
	8	L9 e	100	92	0	91
				6b	7b
	9	(R)-BINAP	100	47	42	90
5b	10	L8 d	100	>95	0	84
	11	L9 d	100	>95	0	88

^aSee Eq 2 for a typical procedure.

^bFor structures of ligands, see Figure 2.

^cat 50 °C.

^din toluene.

^ein toluene at 40 °C.

Table 2.

Optimization of Co-catalyzed enantioselective [2+2]-cycloadditions between acrylate 5a and an enyne 8a^a

entry	ligandb	conv. (%)	9a (%) c	10a (%) c	11a (%) c	ee (9a)
Achiral Ligands
1	dppp	100	78	3	19	0
2	dppf	36	23	-	-	0
3	dPEPhos	45	45	-	-	0
4	L1	100	92	4	-	0
Chiral Ligands
5	(S,S)-BDPP	97	80	3	13	52
6	(S,S)-BDPH	100	80	5	10	56
7	Josiphos 1	50	40	40	2	64
8	Josiphos 2	0	0	0	0	0
9	(S)-BINAP	100	80	-	16	18
10	L2	100	97	2	0	62
11	L3	100	93	2	0	54
12	L4	100	94	4	0	80
13	ent-L7	100	82	17	0	10
14	L8	100	95	3	0	82
15 d	L8	100	92	4	0	94
16 d,e	L8	100	98	<1	0	90
17 d	L9	100	90	10	0	90

^aSee Eq 3 for typical procedure.

^bSee Figure 2 for structures of ligands.

^c.Yield determined by internal standard method.

^dToluene used as a solvent.

^eTrifluoroethylacrylate used instead of methyl acrylate.

Figure 2.

Partial list of ligands examined for cobalt-catalyzed [2+2]-cycloaddition between alkynes and alkenyl derivatives

[2+2]-Cycloadditions between an alkyne (4a) and alkyl acrylates (5a and 5b) (Eq 2, Table 1)

The results for the [2+2]-cycloaddition reactions of a prototypical alkyne, 4-octyne show that for the reaction with methyl acrylate, among the 1,n-bis-diphenylphosphino-alkane [Ph₂P-(CH₂)_n-PPh₂] ligands, only dppp (n = 3) gave moderate yield of the expected cyclobutene 6a, the others producing significant amounts of a diene byproduct 7a [Table 1, entry 1, see also: Supporting Information Table S1 (p. S13) in for a more details]. The byproduct contamination was also a major problem with other commonly used chiral bis-phosphines including (S)-BINAP (entry 3) and (S,S)-BDPP (entry 4).

Phosphino-oxazoline Ligands

While these studies were in progress, we were also examining the related [2+2]-cycloaddition of the acrylate 5a with the enyne 8a (see: Eq 3, Table 2), where it was found that the phosphino-oxazoline ligands, including the easily synthesized achiral ligand L1 and chiral ligand L2 (Figure 2), were much superior in giving the cyclobutene product (9a) with very high chemo- and regioselectivity. These observations, along with the highly tunable nature of the phosphino-oxazoline ligands prompted us to examine this ligand system for further detailed investigations. Accordingly, several of phosphino-oxazoline ligands (L1-L9) were prepared (Figure 2), and their cobalt (II) complexes studied as catalysts for the cycloaddition reaction at room temperature under conditions described in Eq 2. As can been seen in entries 7 and 8 (Table 1), with the new Cocomplexes of these ligands, the [2+2]-cycloaddition between 4-octyne and methyl acrylate proceeded to give synthetically useful yield in unprecedented levels of enantioselectivity for the cyclobutene products. Most notably, replacing the Pdiarylphosphino group in L2 with a dicyclohexylphosphino group (L8) produced a very active Co-catalyst that gave exclusively the cyclobutene adduct 6a in an enantiomeric excess (ee) of 90% in toluene, with none of the undesired heterodimerization product 7a. With yet another modification, viz., introduction of 4,5-disubstitution on the oxazoline as in the ligand L9 [readily prepared from (1S,2R)-norephedrine], further improvements in both the reactivity and selectivity of the reaction were observed (entry 8). With these new ligands and use of an activated trifluoroethyl acrylate 5b, a quantitative reaction ensued with minimal loss in selectivity (Table 1, entries 10 and 11). For the effects of a more complete set of ligands, see Supporting Information, Table S6 (p. S19).

[2+2]-Cycloaddition between a 1,3-enyne (8a) and methyl acrylate (Eq 3, Table 2)

The cycloadditions of 1,3-enynes presented further challenges, the increased reactivity of the enynes notwithstanding. In the initial investigations, we found that 1- or 2-substituted alkenyl alkynes were the best substrates for this reaction (vide infra). For example, one such substrate, 8a, with a C1-methyl substituent (Eq 3, Table 2), gave varying amounts of codimerization product (11a) along with regioisomeric mixtures, 9a and 10a, of the [2+2]-cycloadducts. Most gratifyingly, these selectivity problems could be readily solved by examination of ligand effects on this reaction. See Supporting Information Tables S2–S5 (p. S14-S18) for complete optimization of reaction conditions. Initially, we found that cobalt-complexes of achiral (entries 1–3) and chiral (entries 5–9) bisphosphine ligands were generally unsatisfactory either because of low reactivity or low selectivity. However, the phosphino-oxazoline (PHOX) ligands, including the achiral ligand L1, are excellent ligands for this transformation giving almost exclusively the expected cyclobutene product 9a. Recall that L1 was surprisingly ineffective in the cycloadditions of simple alkynes (e.g., Table 1, entry 5). Changing the 4-aryl substituent on the oxazoline moiety in the PHOX ligands to alkyl groups (e. g., from Ph in L2 to t-Bu in L7, entries 10–13) led to poorer ligands. In sharp contrast, tuning of the P-aryl substituents led to significant improvements in enantioselectivites (the ee’s improving from 62% to 80%, entries 10–12). The best results were obtained when the P-aryl groups were replaced by cyclohexyl groups as in L8, which resulted in the highest yield (98%), regioselectivity (regioisomeric ratio, rr = >95:5), and enantioselectivity (ee = 94%) for this substrate, especially when the reaction was carried out in toluene as the solvent (entries 15 and 16). For an ORTEP representing the solid-state structure of the pre-catalyst, [(R)-L8]CoBr₂, see Figure 4 (A). The norephedrine-derived ligand L9, also gave high enantioselectivity (ee = 90%), albeit with only modest regioselectivity (9a:10a = 90:10, entry 17). With the identification of a viable protocol for the reaction and the availability of several useful ligands, the stage is now set for a broader investigation of the scope of vinyl derivatives and the alkynes.

Figure 4.

A. Solid-state structure (ORTEP) of the pre-catalyst [(R)-L8]CoBr2 (Cambridge Crystallographic Data Centre CCDC # 1860755). B. Absolute configuration (S) established by X-ray crystallography of a 1,6-dibromo-2-naphthoate corresponding to cyclobutene 9a (Cambridge Crystallographic Data Centre CCDC # 1921037). See Supporting Information for details.

Scope of [2+2]-Cycloadditions between Alkynes and Alkenyl Derivatives

The optimized reaction conditions with minor modifications (Figure 3, Eq 4) were employed in the [2+2]-cycloaddition reactions of a wide variety of alkynes and alkenyl derivatives. The enantioselective reactions were carried out using the ligand L9 unless otherwise noted, and authentic racemic products were synthesized using the ligand rac-L8. In general, excellent yields and enantioselectivities were observed for both sets of ligands for a broad range of cyclobutenes carrying many common organic functional groups, originating either from the alkyne or the alkene. Details of the synthesis of each of the specific adducts can be found in the Supporting Information. For the sake of brevity and clarity, the products are classified according to the types of alkyne precursors (4), and, within each type, by the alkenyl partners. The simplest alkynes are the symmetric ones (R = n-Pr, Ph, CH₂OTMS) leading to products 6a-6j. The alkenyl derivatives include alkyl and trifluoroethyl acrylates (6a-6d), vinyltriethoxysilane (6e), vinyltrimethylsilane (6f), vinylboronic acid pinacolate (6g and 6h), each giving excellent yields (86–97%) and ee’s (88–97%). The trifluoroethyl acrylates were generally more reactive and the reactions proceeded at a faster rate compared to the alkyl acrylates. This permits lower temperatures for the reactions, leading to improved yields but with little change in selectivities. Among the acrylates t-butyl acrylate gave the product 6d with the highest enantioselectivity (94%) with 4-octyne. The hydroxymethyl-bearing product 6i,was formed in very good yield and excellent ee (95%). Products derived from unsymmetrical alkynes bearing an aryl group and an alkyl group are represented by 6k-6t. 1-Phenylpropyne and 1-phenylbutyne gave high yields of the [2+2]-adducts 6k, 6l and 6p upon reaction with the respective acrylates in excellent regioselectivity (rr >95:5) and enantioselectivity (ee >89%), with the ester moiety placed adjacent to the aromatic substituent. Ligand L8 was found to be the most optimal for these aromatic substrates. The regioselectivity in the products from unsymmetrical dialkyl alkynes depends on the difference in the size of the two alkyl groups. Methyl/butyl and methyl/i-propyl combination gave relatively low selectivity (62:38 and 87:13 for 6m and 6n), even though the enantioselectivity reached >91% ee for the major product in both cases. Significantly, the norephedrine-derived ligand L9 gave better regioselectivity compared to L8. t-Butylmethyl acetylene with a significant size difference between the groups underwent a highly regioselective [2+2]-cycloaddition to give the product 6o.

Figure 3. Scope of [2+2]-Cycloaddition between Alkenyl Derivatives and Alkynes.

The versatility of the reaction is illustrated by the diverse functional group present in the two coupling partners. In unsymmetrical alkynes regioselectivities of the products (6) are >95:5 unless otherwise indicated. The structures shown are for the major regioisomers and the corresponding ee’s were determined by using CSP-GC or CSP-HPLC. See the Supporting Information for further details. Yields shown are for isolated products.

Products 6q-6ai derived from functionalized, unsymmetrical alkynes, including some carrying hetero-atoms (N, S) on the alkynyl carbon (6s, 6ag, 6ah), demonstrate the vast functional group compatibility of the reaction. The following additional substituents on the alkynes are tolerated: carbonyl group as in a ketone or ester (6q, 6r, 6t, 6u, 6v and 6ai), electron-rich or electron-poor aryl and heteroaryl groups (6t, 6w, 6x, 6y, 6z, 6aa, 6ab, 6ac, 6ad, 6ae, 6af, 6ai), alkyl sidechains with −OH (6y), −Cl (6z), or phthalimido- (6aa) groups, cyclopropyl groups (6ab). In general, very good to excellent yields of the isolated products were obtained for a wide variety of alkenyl derivatives previously shown to work with simpler symmetric alkyne substrates. The regioselectivity observed was also high (>95:5) except for a few instances (6m, 6ag, 6ai). Where ever we have carried out the enantioselective reactions (6a-6t), the ee’s are very high (86–97%).

Scope of [2+2]-Cycloadditions between 1,3-Enynes and Alkenyl Derivatives

1,3-Enynes as the alkyne components in the [2+2]-cycloadditions increase the versatility and applicability of this chemistry by placing yet another functionalizable carbon in the form of an alkenyl substituent on the product cyclobutene (9, Eq 5, Figure 5). The (E)-1-propenyl moiety was chosen as the optimal substituent on many of the enynes in this study for three reasons: (a) the substrates bearing this group can be easily synthesized in high yield by cross-coupling chemistry; (b) the propenyl derivative can be readily converted into a more versatile cyclobutenyl aldehyde by chemoselective ozonolysis (Figure 6, A); (c) 1,3-enynes with no alkenyl substituents require three equivalents of the alkenyl partner to promote efficient [2+2]-cycloaddition (for example, to prepare 9ac or 9ad) and to prevent a competitive co-dimerization of such enynes (example, to give 13ad, Figure 6, B) to 1,6-dialkylvinylaromatic compounds. See Supporting Information Table S7 for further details.

Figure 5. Scope of [2+2]-Cycloaddition between Enynes and Alkenyl Derivatives.

A. Typical reaction conditions. Broad scope of the reaction is illustrated by diverse functional groups tolerated in the two coupling partners. Regioisomeric ratio of product 9 was >95:5 unless otherwise indicated. The structures shown are for the major regioisomers and the corresponding ee’s were determined by using CSP-GC or CSP-HPLC. See the Supporting Information for further details. B. Most useful ligands for the [2+2]-additions of 1,3-enynes and alkenyl derivatives.

Figure 6.

A. A propenyl substituent in an enyne (thus in the cycloadduct 9a) is useful as an aldehyde surrogate. B. This group also prevents homodimerization of the enyne to a vinyl aromatic (13ad). C. Allyl derivatives give [2+2]-cycloadducts with ligand L1, but an eneproducts with dppp. D. A gram-scale reaction and a diastereoselective transformation of a cyclobutene.

The most general reaction conditions, applicable to the broadest spectrum of substrates, is shown in Eq 5 (Figure 5), and, the corresponding substrate scope is illustrated by the accompanying structures. The electron-rich phosphino-oxazoline ligand L8 appears to be the best ligand to obtain the highest regio- and enantioselectivity under the previously optimized conditions for this transformation. See also Table 2, and, a more elaborate Table S2 in the Supporting Information (p. S14) for the effect of other ligands, and Table S3 (p. S16) for the effect of solvents. Solid state structure of the most useful CoBr₂-complex is shown in Figure 4 (A). The authentic racemic products were synthesized using ligand L1. The diverse structures of products in Figure 5 validate the broad scope of this [2+2]-cycloaddition reaction with respect to the enynes and the alkenyl derivatives. The absolute configuration of the products is based on the X-ray crystallographic structure of a 1,6-dibromo-2-naphthoate corresponding to the methyl ester 9a (Figure 4, B). Configuration of the other products are assumed by analogy.

For convenience, the products from enynes are classified into six somewhat arbitrary groups, five of them showing the variations in the starting alkenyl components – alkyl acrylates (9a-9e), hetero-functionalized alkenes (9f-9i: starting from trialkoxyvinylsilane, trialkylvinylsilane, vinylboronate and vinylsulfone), allyl derivatives (9j-9n: from allyl ethers, allyl amides, allyl pinacolboronates, allyl sulfones and allyl trialkylsilanes), vinylarenes (9o-9s), and disubstituted alkenes such as dihydrofuran (9t), norbornene (9u) and transanethole (9v). Another group of adducts (9w – 9ag) show the variations possible in the enynes.

In most of the adducts derived from alkyl acrylates, the hetero-functionalized alkenes, and the allyl derivatives (9a-9n), the major product carries the alkenyl substituent adjacent to the 2-propenyl group of the enyne (>95:5 rr). The enantioselectivities in these instances are also excellent (generally >90% ee), except for the adducts from allyl N-tosylamide (9k, 87%) and allyl phenyl sulfone (9m, 55%). Cycloadditions with allyl derivatives also revealed a remarkable ligand effect. While the cycloaddition reactions carried out using the PHOX ligands (L1 or L8) gave excellent yields of the [2+2]-cycloaddition (9j-9n) with outstanding regioselectivity, the (dppp)CoBr2 complex gave adducts corresponding to a formal ene reaction (Figure 6, C).^12a See Supporting Information for other substrates showing similar behavior (Supporting Information Table S8, p. S21). Simple internal alkynes give only ene-products with 1-alkenes irrespective of the ligands.^12c The vinylarenes (corresponding to the products 9o-9s) are excellent alkenes for the cycloaddition, even though the enantioselectivities seen are unacceptably low, except for the benzofuranyl compound 9s (ee 86%). Compatibility of functional groups in the cyclization of enynes is similar to what was previously observed for simple alkynes (Figure 3). Thus, cyclopropyl (9w), alkyl (9x – 9z), chloroalkyl (9aa), phthalimido-alkyl (9ab) or trimethylsilylvinyl (9af) groups presented no complications in the reaction. In general, very good to excellent regioselectivities (rr >90:10) and enantioselectivities (> 89% ee) were observed for the corresponding substrates.

As alluded to earlier, 1,3-enynes lacking substituents on the C_sp2-carbons (R² or R³) gave only modest yields (e.g., 67% for 9ac; 47% for 9ad), even though the observed enantioselectivity was very high (92% ee and 94% ee respectively). A competing homodimerization of these enynes to give 1- vinyl-2,6-dialkylbenzenes (Figure 6, B) can be avoided by using excess of the alkenyl derivative (up to 3 equivalents in the case of 9ac and 9ad). As illustrated by the high yields and selectivities observed for examples 9ae, 9af, and 9ag (under near stoichiometric ratios of the starting materials), substitution at the internal C_sp2 carbon of the enynes also promotes facile [2+2]-cycloaddition without any complication from the homodimerization of the enyne.

Role of Cationic Cobalt (I) in the [2+2]-Cycloaddition Reaction and a Possible Mechanism of the Reaction

In the initial studies we have recognized a prominent role for a cationic cobalt (I) species in these reactions. Control experiments (Eq 6 and the accompanying Table 3) confirm that no reaction ensues in the presence of isolated Co(I) complex (dppp)₃Co₂Cl₂¹⁴ or a Co(I) complex formed by in situ reduction of LCoBr2 (L = dppp or L1) using Zn as a reducing agent (See Supporting Information Table S5 for further details) in the absence of NaBARF (entry 4 and 9 versus 1 and 2). A possible unified mechanism that accounts for all products formed and the various experimental observations is shown in Figure 7. The presence of NaBARF, which presumably generates a cationic Co(I) species (15), under these reaction conditions is essential for the success of the reaction. Oxidative dimerization of the alkyne and the alkenyl reactant gives the metallacycle 17, which could undergo reductive elimination to give the cyclic product 6 or 9. Alternately, β-hydride elimination followed by reductive elimination could give either 7 or 14 depending upon the substrate. Allylic derivatives with an exo-cyclic CH₂ group (17, G = CH₂Y) gives the formal ene-product via 19. That coordinating solvents such as THF, acetonitrile and even ethyl acetate are not suitable for the reaction also supports the intermediacy of the cationic Co(I) species 15 as a viable catalyst. Thus, this cycloaddition reaction joins a growing list of other alkene functionalization reactions such as dieneacrylate dimerization¹⁶ and HBPin-mediated hydroboration¹⁴ and hydrogenation¹⁷ of prochiral alkenes where cationic Co(I) have been invoked as intermediates.

(6)

Table 3.

Role of cationic Co(I) in [2+2]-cycloaddition ^a

entry	catalyst	activator	reductant	solvent.	time (h)	9a (%)
1	(L1)CoBr2	NaBARF	Zn	CH2Cl2	15	96
2	(dppp)CoBr2	NaBARF	Zn	CH2Cl2	6	>80
3	(L1)CoBr2	NaBARF	none	CH2Cl2	24	0
4	(L1)CoBr2	none	Zn	CH2Cl2	24	0
5	(L1)CoBr2	NaBARF	Zn	THF	24	0
6	(L1)CoBr2	NaBARF	Zn	CH3CN	24	0
7 b	(dppp)CoBr2	NaBARF	Zn	CH2Cl2	6	16
8 c	(dppp)3Co2Cl2	NaBARF	none	CH2Cl2	6	9
9	(dppp)3Co2Cl2	none	none	CH2Cl2	24	0

^aSee Eq 6 for a typical procedure and SI for further details. Solvent and additive effects support a key role for cationic Co(I)-species in [2+2]-cycloadditions.

^bAdditionally, 2.5 mol% dppp was used. Note that addition of extra ligand inhibits the reaction (Table entry 2 vs 7).

^c2.5 mol% catalyst loading (5 mol% in Co).

Figure 7.

A possible mechanism that accounts for the observed ligand, counter ion and solvent effects.

CONCLUSIONS

Cyclobutanes are important structural motifs in many biologically relevant compounds. We report a practical approach to diverse array of nearly enantiopure cyclobutenes from which these compounds can be easily accessed. Two of the most abundant organic precursors, alkynes and alkenyl derivatives are used in the key enantioselective [2+2]-cycloaddition reaction, which employs catalysts derived from readily available amino-alcohols and an earth-abundant metal, cobalt. These enantioselective cycloaddition reactions are easily scaled up to gram scale. With the diverse functional groups present in the cyclobutenes, further diastereoselective transformations are easily envisaged for synthesis of other useful intermediates including highly functionalized cyclobutanes, cyclopentanes and stereo-defined 1,3-dienes. Experimental observations implicate a cationic cobalt(I)-species in the mechanism of this reaction. Such species may have broader applications for other carbon-carbon and carbon-heteroatom bond-forming reactions.

EXPERIMENTAL SUMMARY

In an N₂-filled glovebox, an 8-mL vial equipped with a septum screw cap was charged with a magnetic stirrer bar, [LCoBr₂] (0.01 mmol, 0.05 equiv), activated zinc-dust (0.1 mmol, 0.5 equiv), NaBARF (0.02 mmol, 0.1 equiv), and toluene (0.20 – 0.25 M). The vial was capped and after stirring the mixture for 5–10 minutes, the alkyne (0.20 mmol, 1 equiv) was added neat using microliter syringe via the septum, followed by the alkenyl derivative (0.22 – 0.30 mmol, 1.1–1.5 equiv). The resulting mixture was stirred at rt. The progress of the reaction was monitored by taking an aliquot using a glass pipette, diluting with ether and filtering through a pad of silica before analyzing via GC-FID or TLC. Upon completion of the reaction (4 h-24 h), the vial was taken out of the box and quenched with ether (5 mL). The resulting mixture was filtered over a short pad of silica eluting with ether. Concentration on a rotary evaporator, and subsequent purification by column chromatography eluting with hexane/ethyl acetate (0 to 20%) afforded the products. Full experimental details, spectroscopic and analytical data including chiral stationary phase chromatograms can be found in the Supporting Information under each new compound.