Vinylcyclopropane [3+2] Cycloaddition with Acetylenic Sulfones Based on Visible Light Photocatalysis

Abstract

The first intermolecular visible light [3+2] cycloaddition reaction performed on a meta photocycloadduct employing acetylenic sulfones is described. The developed methodology exploits the advantages of combining UV and visible‐light in a two‐step sequence that provides a photogenerated cyclopropane which, through a strain‐release process, generates a new cyclopentane ring while significantly increasing the molecular complexity. Mechanistic studies and DFT calculations indicate an energy transfer pathway for the visible light‐driven reaction step. This strategy could be extended to simpler vinylcyclopropanes.

A new visible light‐mediated intermolecular [3+2] cycloaddition reaction of meta photocycloadducts with acetylenic sulfones catalyzed by iridium complexes has been developed. The strain energy contained in the photogenerated vinylcyclopropane system enables efficient access to a new complex molecular scaffold under very mild conditions. DFT‐calculations and time‐resolved optical spectroscopy studies substantiated a triplet energy transfer mechanism.

Introduction

Functionalized five‐membered carbocyclic rings containing contiguous stereocenters are found in a wide array of natural products with diverse biological activities. ^[1] Therefore, simple and efficient methods to access highly substituted cyclopentanes and cyclopentenes within a polycyclic framework are particularly desirable in current organic and medicinal chemistry research. ^[2] One major general strategy to access them involves [3+2] cycloadditions, a powerful method that in a single step allows the formation of two new σ bonds and vicinal quaternary centers. ^[3] Vinylcyclopropanes (VCPs) represent an important class of reactive cyclopropane that can participate in this cycloaddition as three‐carbon synthons, especially if they bear electron‐withdrawing groups on the cyclopropane ring. ^[4] However, the use of non‐activated VCP as a competent intermolecular cycloaddition partner is still limited. In particular, the use of Rh catalysts ^[5] is the most developed strategy for cycloadditions of non‐activated VCPs and finding other catalyst systems and protocols that can exploit a wider range of VCP reactivity patterns remains a challenge. The merging of visible‐light photocatalysis with the [3+2] cycloaddition involving VCPs already proved useful in strain‐releasing fragmentation reactions in the assembly of larger‐ring systems, but mainly used activated cyclopropanes as the substrates. Most reported examples are based on electron transfer (ET) photoredox processes, ^[6] while transformations proceeding through energy transfer (EnT) are underdeveloped. ^[7] The broad substrate scope, often relatively independent of the electronic nature, makes the energy transfer process attractive for a wide set of applications.

Considering the lack of examples of intermolecular transition metal‐catalyzed visible light [3+2] cycloadditions involving nonactivated VCPs through EnT, we studied the scope of the same UV‐driven strain‐promoted/visible light strain‐releasing sequence employed in our previous work on cyclopropyl ketones ^[8] by using meta photocycloadduct bearing a vinylcyclopropane substructure as substrates in a visible light‐mediated intermolecular [3+2] cycloaddition (Scheme 1). This strategy which has gradually gained attention from the synthetic community, ^[9] uses the inherent ring‐strain of the photogenerated intermediate as a thermodynamic driving force for the subsequent ring expansion. In other words, the first UV light induced transformation provides a complex fused ring skeleton that is otherwise difficult to obtain. ^[10] This meta photocycloadduct is prone to undergo an irreversible ring‐opening due to the intrinsic strain of the three‐membered ring. ^[11] This feature will be exploited in the second, visible light‐induced transformation by reacting with an alkyne in a [3+2] cycloaddition reaction resulting in the formation of a new complex tetracycle. ^[12]

Scheme 1.

UV and visible light activation sequence.

In general, non‐activated alkynes display low reactivity in this type of cycloadditions, initial experiments using the meta photocycloadduct 4 and simple alkynes with different catalytic systems showed no reactions. After much optimization, we found the use of acetylenic sulfones to be an attractive alternative. The sulfonyl group lowers the LUMO energy of the adjacent π‐bond increasing their reactivity as dienophiles while providing the means for controlling the regiochemistry of the cycloaddition. ^[13] Furthermore, the sulfonyl group can be removed by different methods making the acetylenic sulfones useful reagents for a variety of cycloadditions. ^[14]

Results and Discussion

Our test of the new design started with the photo‐induced intramolecular arene‐olefin meta‐cycloaddition of the N‐allyl‐1,1,1‐trifluoro‐N‐(2‐methylbenzyl)methanesulfonamide (3) which was synthesized in two steps: a reductive amination of 2‐methyl‐benzaldehyde (1) with allylamine, followed by the reaction with trifluoromethanesulfonic anhydride. The irradiation of 3 in anhydrous cyclohexane furnished two isomeric meta‐photocycloadducts in 72 % yield (ratio: 9 : 1), with the linear isomer 4 being favored over the angular isomer 5 (Scheme 2).

Scheme 2.

Preparation of the meta photocycloadduct 4.

The acetylenic sulfones 6 as reaction partners were prepared in a two‐step synthesis. The first step is the preparation of the intermediates β‐iodovinyl sulfones using either the cerium(IV) ammonium nitrate (CAN) mediated reaction of aryl sulfinates and sodium iodide with alkynes ^[15] or the iodine‐promoted reaction of the same substrates using water as solvent. ^[16] The second step was the dehydroiodination of the β‐iodovinyl sulfones with potassium carbonate under reflux to afford the corresponding acetylenic sulfones in relatively good yields (see the Supporting Information). ^[15]

Reaction optimization

The screening of various photocatalysts for the [3+2] photocycloaddition with visible light was performed on meta photocycloadduct 4 and the acetylenic sulfone 6 a as the coupling partners (Supporting Information, Table S1). The desired product 7 a was obtained in appreciable yields (53 % and 63 %, respectively) when employing [Ir(ppy)₂(dtbbpy)]PF₆ (PC I, 2.5 mol %) and phenanthrene (PC VI, 100 mol%). In further optimization studies (Supporting Information, Tables S2, S3 and S4), the yield of 7 a could be increased to 85 % when employing PC I with concomitant reduction of the catalyst loading to 2 mol % when using 2 equivalents of 6 a. Moreover, the cycloaddition proceeded with higher yields in polar solvents and at higher concentrations (0.15 M in CH₃OH).

Substrate scope

Having optimized the reaction conditions, the scope of the synthesized alkynyl sulfones 6 (Scheme 3) was explored next. Substrates bearing electron‐neutral, electron‐rich, and electron‐withdrawing aromatic rings were found to be suitable substrates. Alkyl groups on the sulfone moiety also provided the desired product in moderate to good yields (7 i and 7 j). However, aromatic groups at the triple bond were required for the reaction to proceed successfully; substrate 6 k featuring a butyl‐substituted alkynyl sulfone gave no reaction.

Scheme 3.

Substrate scope of the formal [3+2] cycloaddition. All yields are those of isolated products.^a PC III and Blue LED 440 nm was used (see Supporting Information for details).

At this point, the essential factors governing the catalyst reactivity used in this reaction were investigated by analyzing the results of the screening (Table 1). To distinguish the two possible mechanisms, ET or EnT, the redox potentials, and the calculated triplet energy of the meta photocycloadduct 4 were compared with those of the catalysts (Supporting Information, Table S7).

Table 1.

Evaluation of photocatalysts in the [3+2] cycloaddition.

Entry	PC	E1/2(M*/M+)/E1/2(M*/M−) (V vs. SCE)	ET (kcal/mol) [a]	Yield (%)[b]
1	VI[c]	−2.10/+1.27	61.9	63
2	III	−0.89/+1.21	60.1	49
3	II	−1.00/+1.32	60.0	51
4	IV	−1.73/+0.31	55.2	49
5	I	−0.96/+0.66	49.2	53
6	V	−0.81/+0.77	46.0	traces

[a] 1 kcal mol⁻¹=4.184 kJ mol⁻¹. [b] isolated yields. [c] 100 mol% of catalyst loading and UV/vis CFL bulb was used.

In the cyclic voltammograms of 4 (Supporting Information, Figure S7), oxidation and reduction features were observed with half‐peak potentials of+1.96 V vs. SCE and −1.65/−1.34 V vs. SCE, respectively. Analyzing the reducing power of the catalyst PC IV (entry 4) ^[17] and PC VI (entry 1) ^[18] after excitation with that of 4 reveals that an oxidative quenching is feasible. However, the reduction potentials of the meta photocycloadduct are not sufficient to be reduced by the photoexcited states of the catalyst PC V (entry 6) ^[19] or the other iridium catalysts (entries 2–3, 5). ^[20] Likewise, reductive quenching could be ruled out based on the oxidation potentials of all the catalysts. On the other hand, there is no clear correlation between the triplet energies of the catalysts or photosensitizers and the reaction yield, making a Dexter energy transfer (sensitization) ^[21] also difficult. The Gibbs free energy of each reaction intermediate was predicted using density functional theory (DFT) as it has already be done in the literature for related reactions (Scheme 4). ^[22] While the reaction follows an overall exergonic pathway, releasing 48.9 kcal mol⁻¹ of energy, the computed adiabatic triplet energy of 4 (59.1 kcal mol⁻¹) is considerably higher than those of PC I and PC IV (entries 4 and 5) but is low enough for the rest of iridium catalysts and PC VI. Although there is no compelling evidence for an energy transfer process based on these results, the direct excitation of the VCP moiety by UV irradiation of 4 also gives rise to 7 a in 27 % yield (see the Supporting Information). ^[23]

Scheme 4.

Mechanistic proposal for the [3+2] photocycloaddition. The reported adiabatic energy differences obtained with different functionals are given in kcal mol⁻¹ (no bracket: (U)B3LYP, parentheses: (U)CAM‐B3LYP, square bracket: (U)M06‐2X). The spin densities (yellow, isovalue=0.04) were plotted for the triplet intermediates I–III.

Mechanistic studies

Several control experiments were performed to gain insight into the reaction mechanism. First, radical trapping using (2,2,6,6‐tetramethylpiperidin‐1‐yl)oxyl (TEMPO) completely inhibits the reaction, indicating it may proceed through a radical pathway. To determine whether the reaction involves an efficient ^[24] radical chain mechanism, we performed a light on‐off experiment showing no product formation in the dark phases (Figure 1). Initial intensity‐based Stern‐Volmer quenching experiments on PC I indicated that the phosphorescence of the catalyst could be quenched by 4, suggesting an electron or energy transfer from the excited triplet state of the catalyst (see the Supporting Information). To get more reliable kinetic results, we turned to lifetime‐based Stern−Volmer studies using laser flash photolysis (LFP). We selected PC III (the Ir catalyst with the highest triplet energy), which efficiently catalyzed our initial test reaction under optimized conditions (86 % yield of 7 a) indicating the very same reaction mechanism as with PC I. Furthermore, PC III is well‐suited for mechanistic investigations by LFP as it is known to possess clearly observable spectroscopic signatures in all states that might occur during photocatalytic cycles. ^[25] Linear Stern‐Volmer plots were obtained clearly demonstrating that quenching by the meta photocycloadduct 4 is much faster than with the acetylenic sulfone 6 a (Figure 2, top). Despite the high triplet energy of PC III (Table 1), the quenching rate constant for the reaction between the excited catalyst and 4 (4×10⁶ M⁻¹ s⁻¹) is slower than the diffusion limit by almost 4 orders of magnitude. However, under typical conditions used for lab‐scale photocatalysis with substrate concentrations on the order of 0.1 M, quenching can be quite efficient.

Figure 1.

Visible light irradiation on/off experiment.

Figure 2.

Mechanistic studies upon excitation with 355 nm laser pulses. Top: emission decay kinetics of excited PC III (32 μm) in the presence of varying concentrations of 4 in Ar‐saturated MeCN. Inset: Stern‐Volmer plots of PC III and 4 (purple fit, the colored symbols correspond to the same quencher concentration as in the main plot), and PC III combined with 6a (dark blue). Bottom: transient absorption spectra of the triplet‐excited state (³ PC III), the reduced species (PC III ⁻), the oxidized species (PC III ⁺), as well as in the presence of 4 and 4/6a (80 mM) in Ar‐saturated MeCN obtained by LFP. Apart from the triplet spectrum, all spectra have been recorded after the photoluminescence decay of PC III (see Supporting Information for details).

Next, we tried to directly identify the quenching product(s) by measuring transient absorption (TA) spectra with our LFP setup ^[26] upon selective excitation of PC III. Reference spectra (Figure 2, bottom; see Supporting Information for details) of the photocatalyst triplet, its one‐electron‐oxidized as well as its reduced species were recorded in the reaction medium (MeCN) prior to experiments with the substrates of this study. TA spectra with quencher (4 and/or 6 a) concentrations ensuring about 50 % ³ PC III quenching were measured. These spectra did not show any noticeable TA signals (green and purple spectra in Figure 2), but intense photocatalyst‐derived bands in the transient spectra are expected for both oxidative and reductive quenching pathways, based on our reference measurements. Hence, our TA experiments summarized in Figure 2 (bottom) allow us to exclude quenching via ET. ^[27]

Accordingly, we propose a catalytic cycle as shown in Scheme 4 with an energy transfer as initial photochemical activation step. First, PC I (or PC III) acts as a sensitizer, transferring energy from its ³MLCT state to the vinylcyclopropane moiety in 4, generating the triplet state of the latter, which undergoes a rearrangement followed by a radical addition of the acetylenic sulfone to generate the intermediate III. A subsequent ring‐closure reaction affords the final tetracycle 7. The occurrence of radical intermediate III is supported by the finding that aromatic substituents at R¹ are required for the reaction to proceed. The absence of new signals in the TA spectra right after significant energy transfer quenching can be explained as follows: The catalyst triplet signals completely return to the baseline for EnT mechanisms and the substrate‐derived intermediates II and III (compare, Scheme 4) are expected to absorb very weakly (and predominantly in the UV region), owing to the lack of π‐conjugation and adjacent functional groups. The light absorbing properties of the triplet diradical intermediates II and III were predicted by means of time‐dependent DFT. For the excitation of II and III, an energy of at least 3.67 eV (oscillator strength f=0.0002) and 2.71 eV (f=0.0093) is required for the first excited state. Local absorption maxima were predicted to be at 239 nm and 282 nm, respectively (see Supporting Information). Moreover, the exact factors governing the lifetimes of diradicals are not fully understood ^[28] and these species might be too short‐lived for their efficient detection. Ring‐opening of intermediate I in its triplet state is most likely faster by orders of magnitudes than for the well‐known (doublet) alkyl cyclopropane radical clock (ring‐opening rate constant, 1.3×10⁸ s⁻¹), ^[29] considering the accelerating effects of substituents and/or fused rings at the cyclopropane moiety. ^[30] We speculate that strain release and the associated substantial energy gain could potentially occur on a similar timescale as molecular collision and the actual Dexter energy transfer from the excited iridium complex. Such a concerted reaction mechanism (without the presence of intermediate I) could explain the observed quenching and energy transfer product when PC I is used, for which the generation of intermediate I via conventional mechanisms would be endergonic by about 10 kcal mol⁻¹ (9.9 kcal mol⁻¹). This is clearly beyond the current limit for uphill energy transfer reactions, ^[31] rendering this stepwise pathway highly unlikely. Similar counter‐thermodynamic reactions are well‐established in the PCET community for concerted proton‐coupled electron transfer reactions, but they seem underexplored for energy transfer catalysis. Moreover, concerted ring‐opening to yield the low‐energy triplet intermediate II would block back‐EnT reaction channels, which frequently reduce the overall performance of uphill energy transfer reactions.

The mechanistic alternative of an EnT from the PC to the alkynylsulfone appears less likely as the triplet energy of the latter is higher (62.2 kcal mol⁻¹), no gain from a follow‐up reaction such as the ring‐opening of the VCP can occur and the reaction of the triplet alkynylsulfone to the VCP should occur at the distal carbon of the C=C double bond, which contradicts the observed regiochemistry. To further corroborate that only the vinylcyclopropane substructure of 4 is involved in the mechanism of this reaction, experiments with different substrates were performed. LFP experiments with photocatalyst PC III and (1‐cyclopropylvinyl)benzene (8) as quencher gave a quenching rate constant of 5×10⁸ M⁻¹ s⁻¹ (see Supporting Information), i. e. about two orders of magnitude faster than obtained with substrate 4 (Figure 2). We again excluded electron transfer quenching pathways by transient absorption measurements and the observed higher rate constant is consistent with the expectations for a Dexter‐type energy transfer, given that the in silico triplet energy of 8 (49.3 kcal mol⁻¹) is lower than that of 4. The preparative [3+2] cycloaddition with 8 and the meta photocycloadduct 11 (used as an inseparable mixture of isomers without further purification) yielded the expected products in 46 and 39 % yield, respectively (Scheme 5). The procedure is thus likely applicable to other VCP‐type products of intra‐ or intermolecular photocycloadditions and permits the construction of complex skeletons in only two consecutive photochemical transformations. Two further commercial and naturally ocurring VCPs, ethyl trans‐chrysanthemate and (1S,3R)‐cis‐4‐carene, could also successfully be reacted, albeit only in moderate yields, see the Supporting Information.

Scheme 5.

Alternative substrates in the [3+2] photocycloaddition under standard conditions. The predicted energy differences are given in kcal mol⁻¹ (no bracket: (U)B3LYP, parentheses: (U)CAM‐B3LYP, square bracket: (U)M06‐2X). ^a Inseparable mixture.

To extend the synthetic utility of the method, tetracyle 7 a was exemplarily subjected to a selective reductive desulfonylation, producing a styrene moiety suitable for further follow‐up reactions (Scheme 6).

Scheme 6.

Selective desulfonylation of tetracycle 7 a.

Conclusions

A visible light‐mediated [3+2] cycloaddition of meta‐photocycloadducts and acetylenic sulfones has been developed. Catalytic amounts of iridium complexes act as triplet sensitizers as indicated by DFT calculations and time‐resolved optical spectroscopy, which allowed us to exclude photoinduced electron transfer reactions. The methodology highlights the power of a consecutive UV/visible light activation sequence involving the generation of a strained intermediate and subsequent catalytic strain release.

Crystallographic data

Deposition number 2112151 (for 7 a) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

A. Luque, J. Groß, T. J. B. Zähringer, C. Kerzig, T. Opatz, Chem. Eur. J. 2022, 28, e202104329.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.