Crossed Intermolecular [2+2] Cycloaddition of Styrenes by Visible Light Photocatalysis

Abstract

We report a method for the crossed [2+2] cycloaddition of styrenes using visible light photocatalysis. Few methods for the synthesis of unsymmetrically substituted cyclobutanes by photochemical [2+2] cycloaddition are known. We show that careful tuning of the electrochemical properties of a ruthenium photocatalyst enable the efficient crossed [2+2] cycloaddition of styrenes upon irradiation with visible light. We outline the logic that enables high crossed chemoselectivity, and we also demonstrate that this reaction is remarkably efficient; gram-scale reactions can be conducted with as little as 0.025 mol% of the photocatalyst.

Introduction

Cyclobutane-containing natural products have been isolated from a surprisingly diverse range of plant, insect, and microbial species.¹ Yet methods for the synthesis of four-membered carbocycles² are relatively undeveloped compared to those available for the construction of five- and six-membered rings, impeding the synthesis of these compounds and the exploration of a vast domain of bioactivity-rich chemical diversity space.³ Photochemical [2+2] cycloaddition reactions would be a very efficient approach to the synthesis of complex cyclobutane-containing structures,⁴ but few methods for the chemoselective heterodimerization of isolated olefins by [2+2] photocyclization have been reported.⁵ For the past several years, we have been investigating the ability of Ru(bpy)₃²⁺ and related transition metal chromophores to promote cycloaddition reactions initiated by one-electron photoredox processes,^6,7,8 and in 2009, we reported the highly chemoselective crossed intermolecular [2+2] cycloadditions of electron-deficient enones.^5d In this communication, we describe a complementary approach to photochemical heterodimerization of electron-rich olefins. This process is initiated by the one-electron oxidation of styrenes, and its success relies upon the availability of ruthenium poly(pyridyl) photocatalysts with highly tunable electrochemical properties.⁹

Results and discussion

As a prerequisite for the development of a crossed [2+2] heterodimerization, we first investigated the intermolecular homodimerization of styrene 1. We had previously shown that tethered bis(styrene)s undergo efficient intramolecular [2+2] cycloaddition when irradiated in the presence of Ru(bpy)₃²⁺ and the oxidative quencher methyl viologen (MV²⁺).^6b,10 These conditions, however, were ineffective for intermolecular cyclodimerization of 1. We were able to achieve modest yields of the desired cycloadduct using more strongly oxidizing conditions (Ru(bpz)₃²⁺,¹¹ air) that we found to be effective in photocatalysis of radical cation Diels–Alder cycloadditions.^6e However, after extensive optimization, we observed that these reactions do not proceed to complete consumption of starting material even with extended reaction times, higher loadings of catalyst, or multiple additions of catalyst.

When the cycloadduct 2 is isolated and re-subjected to the reaction conditions, the monomeric styrene 1 slowly appears in the reaction medium, suggesting that the product of the cycloaddition is prone to oxidation to and subsequent cycloreversion (Scheme 1).¹² Indeed, we measured a peak oxidation potential of +1.27 V vs SCE for cycloadduct 2, confirming its suceptibility to oxidation by Ru*(bpz)₃²⁺, which has a reported oxidation potential of +1.45 V. Consistent with this interpretation, we find that the putative radical cation intermediate 1^•+ resulting from cycloreversion can be trapped with isoprene to afford Diels–Alder cycloadduct 6 in good yield. These observations indicate that the incomplete conversions in the cycloaddition process are the result of an oxidatively induced cycloreversion process that balances the rate of the desired cycloaddition at high conversions.

Scheme 1.

Reversibility of the [2+2] cycloaddition.

We reasoned that the use of a photocatalyst with a less strongly oxidizing excited state might promote the desired cycloaddition of 1 without oxidizing 2 and triggering its undesired cycloreversion. A systematic screen of ligand-modified ruthenium complexes with varied excited state oxidation potentials revealed that conversion to 2 could be increased when Ru(bpz)₃²⁺ was replaced with the less electron-deficient tris(bipyrimidine) complex Ru(bpm)₃²⁺ (Table 1, entry 5).¹³ The reported oxidation potential of Ru*(bpm)₃²⁺ is +1.20 V vs SCE,¹⁴ which should be sufficient to promote the one-electron oxidation of 1 (+1.1 V)¹⁵ but not of 2. Interestingly, we discovered that complete consumption of 1 could be achieved by conducting the reaction at 0 °C.¹⁶ Under these conditions, the desired cycloadduct 2 could be isolated in 81% yield.¹⁷

Table 1.

Photocatalytic homodimerization of 1^a

entryb	Ru complex	co-oxidant	time	yieldc
1d	Ru(bpy)3(PF6)2	MV(PF6)2	24 h	9%
2	Ru(bpy)3(BArF)2	air	8 h	1%
3	Ru(bpz)3(BArF)2	air	2 h	57%
4	Ru(bpz)3(BArF)2	air	16 h	65%
5	Ru(bpm)3(BArF)2	air	2 h	72%
6e	Ru(bpm)3(BArF)2	air	2 h	83% (81%)f

^aExcited state oxidation potentials reported in MeCN vs. SCE. Values from Ref. 9.

^bReactions conducted using 0.5 mol% of the photocatalyst (with respect to product) in CH₂Cl₂ using a 20 W compact fluorescent lightbulb unless otherwise noted.

^cYields determined by ¹H NMR analysis against an internal standard unless otherwise noted.

^dConducted under the conditions of Ref 6b: 5 mol% Ru(bpy)₃(PF₆)₂, 15 mol% methyl viologen hexafluorophosphate (MV(PF₆)₂), in MeNO₂.

^eReaction conducted at 0 °C.

^fIsolated yield in parentheses.

Having identified a photocatalyst with the appropriate excited state redox potential to enable high-yielding intermolecular cyclodimerizations, we next explored the feasibility of chemoselective crossed cycloadditions. We began by investigating the reaction between 1 and 4-methylstyrene 7a. The design of this experiment took two strategic considerations into account. First, we selected a coupling partner less electron-rich than 1 in order to ensure that 1 would be selectively oxidized by Ru(bpm)₃²⁺ and that 1^•+ would the only alkene radical cation generated in the reaction medium.¹⁸ Second, a variety of studies have suggested that the mechanism of the [2+2] cycloadditions of styrene radical cations is a concerted asynchronous process in which formation of the β,β′ bond is significantly more advanced in the transition state than the α,α′ bond.¹⁹ We hypothesized that the crossed cycloaddition of 1 with β-unsubstituted styrene 7a might be sterically favored over homodimerization of 1.

In the experiment, when 1 is irradiated with 2 equiv of 7a in the presence of 0.25 mol% Ru(bpm)₃²⁺, only the crossed cycloadduct 8a is produced (Table 2, entry 1). We observe a noticeable increase in rate and yield when the reaction is conducted at −15 °C (entry 2). However, we found that other alkenes sometimes gave poorer crossed selectivity; for example, when styrene (7b) is used as the reaction partner, a 4:1 mixture of crossed:homodimerization cycloadducts is obtained (entry 3). In these cases, however, the selectivity could be increased by adding 1 slowly by syringe pump over the course of the reaction to minimize the rate of homodimerization (entry 4).

Table 2.

Optimization of crossed [2+2] cycloaddition

entrya	R	T	time	8:2b	yield 8bc
1	Me (7a)	23 °C	2 h	>10:1	77%
2	Me (7a)	−15 °C	1 h	>10:1	86%
3	H (7b)	−15 °C	1 h	4:1	55%d
4e	H (7b)	−15 °C	1 h	>10:1	79%

^aReaction conducted using 0.25 mol% Ru(bpm)₃(BArF)₂ in CH₂Cl₂ and a 20 W compact fluorescent lightbulb.

^bRatio determined by ¹H NMR analysis of the unpurified reaction mixture.

^cIsolated yield unless otherwise noted.

^dYield determined by ¹H NMR analysis against an internal standard.

^e1 was added by syringe pump over 1 h.

Table 3 summarizes experiments exploring the range of olefins that could participate in successful crossed [2+2] cycloadditions with 1 under these conditions. A wide variety of styrene derivatives are suitable coupling partners in this reaction; substitution at all positions of the aryl ring is tolerated (entries 1–4). Substitution at the α position of the styrene is well tolerated without loss of diastereocontrol, enabling the construction of unsymmetrically substituted cyclobutanes possessing quaternary carbon stereocenters with high diastereoselectivity (entry 9). On the other hand, reactions involving β-substituted styrene acceptors exclusively gave homodimerization of 1 (entry 10), consistent with our reaction design plan. The functional group compatibility of this reaction is good; halides, esters, and unprotected alcohols are tolerated (entries 5–7), although very electron-deficient styrenes reacted sluggishly and could not fully outcompete homodimerization (entry 8). Finally, we found that other nucleophilic alkenes such as vinyl ethers and allyl silanes could also produce the crossed cycloadduct in good yield (entries 11 and 12). 1,1-Disubstituted aliphatic olefins, however, are not competent coupling partners and are a current limitation of this system (entry 13).

Table 3.

Scope studies in [2+2] cycloadditions involving 1.

entrya	alkene	cycloadduct	time	yieldb
1	Ar = 4-MePh		1 h	86%
2	Ar = 3-MePh		2.5 h	86%
3	Ar = 2-MePh		1 h	81%
4c	Ar = Ph		1.5 h	79%
5d	Ar = 4-FPh		2.5 h	78%
6	Ar = 4-AcOPh		3 h	78%
7d	Ar = 4-(HOCH2)Ph		2.5 h	56%
8e	Ar = 4-(MeO2C)Ph		4.5 h	31%
9			2 h	87%f
10d			3.5 h	0%
11			3.5 h	67%
12d			2.5 h	48%g
13			4.5 h	0%

^aReactions were irradiated with a 20 W compact fluorescent lightbulb at −15 °C using 0.25 mol% Ru(bpm)₃(BArF)₂, 1 equiv of 1, and 2 equiv of coupling partner.

^bValues reflect the average isolated yields from two reproducible experiments. Products isolated in >10:1 d.r. unless otherwise noted.

^cAlkene 1 added over 1 h.

^dAlkene 1 added over 2 h.

^eAlkene 1 added over 4 h.

^f9:1 d.r.

^g3:1 d.r.

The range of oxidizable electron-rich styrenes that could initiate the cycloaddition was also probed (Table 4). Substitution at the allylic position is well-tolerated; substrates bearing silyloxy, acetoxy, chloride, hydroxide, and sulfonamide groups proceed smoothly (entries 1–5). Acid-labile acetal groups are also readily tolerated (entry 6). Surprisingly, substitution with electron-donating groups at the meta position of the arene results in a decrease in yield (entries 7 and 8), perhaps due to over-oxidation and decomposition of the more electron-rich cycloadduct. Consistent with this hypothesis, the cycloadduct in entry 8 exhibited a peak oxidation potential of +1.15 V vs SCE, which lies within the range accessible by Ru*(bpm)₃²⁺. Positioning the methoxy group at the ortho position resulted in no reaction (entry 9). Styrenes activated with a silyloxy or alkoxy group at the para position undergo cycloaddition efficiently (entries 10 and 11). However, acetoxy groups are evidently insufficiently electron-donating to activate the styrene towards oxidation and subsequent cycloaddition (entry 12).

Table 4.

Scope studies of crossed cycloadditions involving 7a.

entrya	substrate	cycloadduct	time	yieldb
1	R1 = OTBS		2.5 h	63%
2	R1 = OAc		4 h	91%
3	R1 = Cl		3 h	37%
4	R1 = OH		3 h	74%
5	R1 = NHTs		24 h	74%
6			1.5 h	79%
7			1.5 h	59%
8			24 h	28%
9			24 h	0%
10	R2 = TBSO		2.5 h	88%
11	R2 = BnO		1.5 h	79%
12	R2 = AcO		24 h	0%

^aReactions were irradiated with a 20 W compact fluorescent lightbulb at −15 °C using 0.25 mol% Ru(bpm)₃(BArF)₂, 1 equiv of 1, and 2 equiv of coupling partner.

^bValues reflect the average isolated yields from two reproducible experiments.

Control studies confirmed that no reaction occurs in the absence of light or the photocatalyst. In rigorously degassed solvent and under an atmosphere of argon, the reaction proceeds but to dramatically lower conversion (7% yield after 1 h). From this observation, we conclude that the cycloaddition is a chain process, consistent with Ledwith’s seminal studies on radical cation [2+2] cycloadditions,²⁰ and that the role of the aerobic atmosphere is to turn over a photoreduced Ru(I) complex. We therefore propose the mechanism outlined in Scheme 2. The Ru(bpm)₃²⁺ chromophore is photoexcited by visible light, and the resulting photoexcited Ru*(bpm)₃²⁺ complex participates in a one-electron oxidation of 1. The resulting alkene radical cation (1^•+) undergoes [2+2] cycloaddition with 7a to afford a cyclobutane-substituted arene radical cation (8a^•+), which undergoes one-electron reduction to afford the observed cycloadduct 8a. The reductant in this process could either be another equivalent of unreacted 1, in a chain propagation step, or the photoreduced Ru(bpm)₃⁺ complex. However, given the low concentration of the ruthenium catalyst and the rapid rate of its quenching by oxygen, the rate of this chain termination process is presumably quite slow.

Scheme 2.

Proposed mechanism of crossed cycloaddition

As a demonstration of the scalability of this process, the crossed cycloaddition between 1 and 7a was performed on gram scale (eq 1). This reaction required only 0.025 mol% of Ru(bpm)₃²⁺ and produced the expected cycloadduct 8a in good yield and without loss of chemo- or diastereoselectivity. Notably, direct UV irradiation of 1 and 7a in a Rayonet reactor (254 nm) in the absence of a photocatalyst failed to produce any observable cycloadducts, consistent with the known inefficiency of intermolecular [2+2] cycloaddition of simple acyclic styrenes under standard photochemical conditions.²¹ Thus, visible light photocatalysis provides a uniquely effective and scalable method for the chemoselective synthesis of a wide variety of unsymmetrical styrene dimers inaccessible using traditional photochemical techniques.

(1)

Conclusions

In summary, we have devised a highly efficient method for the chemoselective radical cation-mediated crossed [2+2] cycloaddition under photocatalytic conditions. Key to the success of this strategy was the ability to tune the electrochemical properties of the photocatalyst in order to minimize an unexpected oxidatively triggered cycloreversion of the product. Unsymmetrically substituted cyclobutanes can thus be efficiently and chemoselectively prepared on gram scale using as little as 0.025 mol% Ru(bpm)₃²⁺. These results, together with our previously reported method for crossed [2+2] cycloaddition of enones, suggest that the construction of a wide range of cyclobutane-containing structures are accessible via reactions of photogenerated radical ions.