An Efficient Photo-driven Retro-Diels-Alder Reaction; Dienes on Demand

Summary:

Skeletal editing enables precise structure modifications by inserting, deleting, or rearranging atoms to achieve diversification. Building on previous computational studies that highlighted the potential for cycloreversion of trans-cyclohexenes as a means of diene synthesis, and motivated by our ongoing interest in molecular transducers, we present the first experimental evidence for the retro-Diels–Alder reaction of trans-4-phenyl-3,6-dihydro-2H-pyran. Energy transfer photocatalysis leads to isomerization to the strained trans-isomer, where the strain energy is exploited to drive cycloreversion, yielding both diene and formaldehyde. Mechanistic studies support a charge-separated transition state, while computational models explain the unique reactivity of this heterocyclic ring. We employed sequential cycloreversion/cycloaddition reactions in a one-pot sequence to convert the parent aryldihydropyrans into [4+2] adducts with dienophiles, in addition to preparing 4-vinylcyclohexenes via self-dimerization, and dihydroterphenyl cores by adding 1,1-diarylethylene. This new photo-driven strategy facilitates use of dihydropyrans as diene synthons, providing routes to arylcyclohexenes via atom-pair swap skeletal editing.

Graphical Abstract

Introduction

Skeletal editing, the strategic modification of a molecule’s carbon framework, has emerged as a transformative tool in organic synthesis.¹ By enabling efficient construction, rearrangement, or functionalization of molecular structures, it provides a powerful means to enhance molecular complexity and properties.^2,3 Unlike traditional synthetic approaches that rely on stepwise assembly, skeletal editing offers rapid and direct transformations, granting access to new structural motifs with minimal synthetic efforts. Among the various methods to achieve this, the retro-Diels-Alder (rDA) -Diels-Alder (DA) reaction sequence stands out as a versatile and impactful approach.

The DA reaction, a [4+2] cycloaddition between a conjugated diene and a dienophile, allows chemists to selectively modify molecular cores by introducing new carbon–carbon bonds, forming six-membered rings with high regio- and stereo-control, making it ideal for synthesis and skeletal editing- should good strategies to reverse the reaction be developed. Notably, the [4+2] cycloaddition and its reverse (rDA), have been employed in creative ways to accomplish skeletal editing, either by ring contraction or expansion,^4,5 atom insertion⁶ or deletion and atom exchange.^7,8 Typically, for most DA reactions, both ΔH and ΔS are negative, favoring the forward reaction at lower temperatures. However, at elevated temperatures, the entropy term dominates, promoting rDA. This property has been exploited in specific cases where high temperatures (often exceeding 500 °C) or volatile byproducts drive the reaction.⁹ To date, temperature remains the primary method for reversing the DA reaction, but such conditions limit its broader applicability.

Our lab has previously investigated the visible light photosensitized generation of trans-cycloalkenes^10–16 and their protonation by weak acids to form carbenium ion species, which can drive further chemical transformations (Scheme 1b).^10,11,13,14 Using an appropriate photosensitizer and blue light, we have demonstrated the synthetic utility of the previously reported photoisomerization of cis-arylcyclohexenes¹⁷ to the strained trans-isomers¹⁷. Despite the short lifetime of trans-phenylcyclohexene (9 μs at room temperature),¹⁸ its high strain energy (ca. 50 kcal/mol)^18–21 enables unique and extreme reactivity.

Scheme 1. Cycloreversion reaction via trans-cyclohexene.

Building on our interest in trans-arylcycloalkenes, we aimed to further investigate their fundamental reactivity. This exploration was inspired by a computational study from the Johnson group, published over two decades ago, which examined the DA reaction and showed that the reaction is stereospecific with respect to diene configurations (Scheme 1a).²²

This study examined the cycloreversion of cyclohexene by both the usual concerted rDA or by initial π-bond rotation to trans-cyclohexene, followed by rDA. Predicted relative free energies of relevant structures and transition states are summarized in Scheme 1a. The key takeaway was that as a consequence of strain, the barrier for cycloreversion from trans-cyclohexene is significantly lower than for cis-cyclohexene (25.1 kcal/mol vs. 61.0 kcal/mol, respectively, for concerted reactions). They also proposed that cycloreversion could proceed via initial cis-to-trans isomerization, facilitated initially by photoexcitation. However, the study noted that experimental observation of this pathway is unlikely due to the competitive and rapid thermal isomerization of trans- to cis-cyclohexene, which has a barrier estimated at 8.7 kcal/mol. We hypothesized (Scheme 1b) that the design of structures with lower barriers to rDA of the strained trans isomer might favor a synthetically useful photo-driven cycloreversion.

In support of this concept, during our exploration of light-induced chemistry of cyclohexene and its analogs, we sometimes observed the formation of dienes. Motivated by this result and the potential utility of the rDA reaction, we explored the possibility of rDA reactions from trans-arylcyclohexenes¹⁷ and heteroatomic analogs.

In this study, we report the first experimental demonstration of photo-driven strain-mediated rDA reactions from heteroatomic analogs of trans-phenylcyclohexene. This process provides access to semi-stable substituted 1,3-dienes, enabling their one-pot transformation to generate valuable cyclohexene cores via net atom-pair swap from C–O to C–C. Herein, we detail the reaction conditions, mechanistic and computational insights, and initial achievements that highlight the potential of this method in advancing skeletal editing.

Results and Discussion

Our study began by subjecting a small library of 2-phenylcyclohexene analogs to blue LED irradiation in the presence of Ir(dFppy)₃, a triplet energy transfer photocatalyst, in DCM at room temperature (Scheme 2) where we qualitatively watched for the formation of products via ¹H NMR. These conditions were very similar to those used by us to generate trans-cyclohexene for the purposes of carbon capture¹¹ and ether formation¹⁰- the only difference being the absence of a weak acid used to protonate the trans-cyclohexene. Our library probed the nature of X in the 4-position of the cyclohexene (Scheme 2a). We did not observe any desired retro-DA reaction for most of the library (X = CH₂, NH, NBoc, NAc, NBz, SO, SO₂). In the case of 3, 5, 6, 7 the cycloalkenes appeared to be photostable, leaving behind the starting alkenes, while 8 was found to be photo-unstable and underwent decomposition under the reaction conditions. However, when X = O (1) and S (2) the desired light driven retro-DA reaction took place leading to the fragmentation of the parent molecules into 2-phenyl-1,3-butadiene and formaldehyde or thioformaldehyde, respectively. These latter two products are observed by ¹H NMR but not isolated. Although, both pyran 1 and thiopyran 2 underwent retro-DA reaction (75% and 26%, respectively), a much cleaner reaction was observed with 1, so we focused on this structure for our subsequent investigation.

Scheme 2. Photocatalytic retro-Diels-Alder reaction.

Although full conversion of 1 was observed under the reaction conditions, a byproduct was found to form (Scheme 2b). Isolation and characterization of the byproduct revealed that it was cyclohexene derivative 1i, which is a dimer of diene 1a. Thus, if we wanted to be able to provide the product of the rDA we needed to understand this process.

After carefully monitoring the reaction, we found that diene 1a was photoreactive and underwent dimerization by photocatalyzed formal [4+2]-cycloaddition under the reaction conditions. Isolated 1a was also prone to thermal [4+2] reaction, albeit at a much slower rate.²³ Due to this subsequent reaction of product 1a, the overall yield of 1a dropped significantly with extended reaction time. After a thorough screening of several reaction conditions (See SI for more details), we found irradiation of a 0.05 M solution of 1 and 0.2 mol% Ir(dFCF₃ppy)₂(dtbpy)PF₆ as photocatalyst with 5 W blue LED light in DCM solvent at 18 °C temperature under inert atmosphere to be the optimized reaction conditions, affording 75% conversion to the diene 1a with 10% conversion to the dimer. The following reaction conditions were important to obtain high yield of 1a: a lower light intensity, a decreased substrate concentration, and the nature of the photocatalyst- along with its concentration. Finally, carefully monitoring the reaction and halting it at the optimal reaction time proved critical.

Reaction without light or photocatalyst failed to provide any reaction- indicating the fragmentation of 1 is a light mediated photocatalyzed process (Scheme 3a). Furthermore, the addition of benzoic acid to the reaction conditions results in the formation of ester 1A with 82% isolated yield, along with a trace amount of rDA products (Scheme 3b). This esterification experiment serves as a probe for ground state trans-cyclohexenes, which have been shown to undergo protonation with weak acids.¹⁰ This supports the formation of a trans-cycloalkene- as such reactivity would be difficult to explain from triplet excited state. To investigate the dimerization process, freshly prepared diene 1a, was subjected to the reaction conditions, and the formation of dimer 1i was observed in 76% yield. In control experiments without light or photocatalyst, the dimerization became extremely slow and was only observed after two days; this supports a conclusion that the dimer formation was also a photoinitiated process (Scheme 3c).

Scheme 3. Control Studies.

Stern-Volmer quenching experiments revealed that the ambient temperature phosphorescence emission intensity of the excited photocatalyst was effectively quenched by the addition of both pyran 1, diene product 1a and the dimer 1i. The rate of quenching was almost four-fold greater in the case of 1a compared to 1 and 1i (Fig. 1). Despite the greater quenching rate by the diene, dimerization could be largely mitigated by dilution which had a greater impact on the bimolecular dimer formation than the unimolecular fragmentation of E-1.

Figure 1.

Photocatalyst phosphorescence quenching by diene (1a) and alkene (1) and dimer (1i).

Based on these experiments and literature precedent^18–21 which shows that both S₁ and T₁ of phenylcyclohexene 3 can lead to the trans isomer, we proposed the following mechanism for this novel reaction (Scheme 4). The photocatalyst is excited by the absorption of a blue photon and leads to a long-lived triplet. This triplet serves as a photosensitizer, and in turn undergoes a Dexter energy transfer (EnT) with 1 to generate its excited triplet, represented as biradical B. On the triplet surface, B undergoes relaxation by rotation about the double bond to give the orthogonal biradical (Ortho-B), before returning to the ground state via inter-system crossing (ISC) which presumably occurs at a geometry similar to the transition state for rotation about the double bond. Some of this high energy species will relax to give E-1. The substantial energy stored in this highly strained intermediate (E-1) can be released either by π-bond rotation and return to 1, or by [4+2] cycloreversion to give 1a and dienophile. As shown in Scheme 4, the relative energetics of π-bond rotation vs rDA will determine the net reaction outcome. Increasing heteroatom-induced polarization in the rDA transition state was expected to favor this pathway.

Scheme 4. Potential reaction pathaways: p bond rotation vs [4+2] cycloreversion.

To probe the proposed reaction mechanism, we first performed a Hammett study, examining the relative rates of different para-substituted derivatives of 1 (orange line, Scheme 5). To probe the proposed reaction mechanism, we first performed two Hammett studies, examining the relative rates of different para-substituted derivatives of 1 (orange line, Scheme 5) as well as 2-aryl-4-phenyl-3,6-dihydro-2H-pyrans bearing different para-substituents in the second aryl ring (blue line, Scheme 5). The initial rates of the reactions were determined by proton NMR (See SI for details) and the relative initial rates with respect to un-substituted substrate (k_X/k_H value) of each reaction were calculated. These studies assume a reproducible light intensity as well as high efficiency in triplet sensitized cis to trans isomerization.²⁴

Scheme 5. Hammett Study.

In the first case, the plot of log(k_X/k_H) versus substituent constant σ gave a straight line with negative slope (−1.2) which suggested that positive charge buildup was stabilized by the substituents in the transition state. Moreover, a better fit was obtained using the classic σ values (pKa of benzoic acid derivatives) instead of σ+ values which are based on the solvolysis of benzyl chloride derivatives (R₂ = 0.99 for σ versus R₂ = 0.83 for σ+) indicating that the positive charge might not be directly felt at the benzylic site. In this case, the reaction rate increases with an electron donating substituent on the liberated diene component.

The electronic consequence of substituents on the liberating dienophile component was inspected by a similar Hammett study and showed the opposite effect. In this case, a straight line with a clear positive slope (0.39) was obtained. Here, the reaction rate increases with an electron withdrawing substituent on the liberated dienophile component.

These opposing Hammett plots have a straightforward explanation. For a “normal” electron demand Diels-Alder cycloaddition such as these, forward reaction is favored by an electron rich diene and electron poor dienophile. According to microscopic reversibility, the same should be true for retro-Diels-Alder reactions.^9,25 As with cycloaddition, making the diene more electron rich or the dienophile more electron poor, both lower the barrier to cycloreversion. The observation of a steeper slope for diene modification follows from the usual frontier MO argument that the stronger orbital interaction is between the diene HOMO and dienophile LUMO. Taken together, our observations support a polarized transition state but not an ionic intermediate. Additionally, initial rate studies in various solvents showed an increased rate in polar protic solvents (such as MeOH, EtOH), further supporting charge development along the reaction coordinate. (see SI for details). In addition to the mechanistic insight these experiments provide, it also provides clear guidance in terms of reaction design and expected rates for various substrates, based on the electronics of the substituents.

In order to better understand the different reactivity observed for phenylcyclohexene and its derivatives, we created computational models for these reactions. Structures were optimized with M05–2X/6–311+G(d,p) density functional theory (DFT), followed by CCSD(T)/aug-cc-pvdz single point calculations to more accurately assess energetics. Larger basis set calculations were not feasible for these structures and do not seem to be necessary. CCSD(T) methods provide a higher level of electron correlation and are expected to give the most accurate results. Relative free energies were estimated by adding DFT thermal corrections (mostly ZPVE + entropy) to the CCSD(T) total energies. In some cases, the effect of polar solvents was estimated through DFT computations which included implicit solvation.

Earlier experimental studies on 1-phenylcyclohexene by Caldwell²⁰ and Peters²¹ provide critical benchmarks for comparison to computations. The transient trans-isomer is easily observed by laser flash photolysis using both direct and triplet sensitized irradiation. An activation energy of 12.1 ± 0.12 kcal/mol was measured for trans to cis isomerization on the ground state surface. Pulsed photoacoustic spectroscopy gave an energy for the thermalized triplet of 56.0 ± 3.4 kcal/mol and a cis to trans energy difference of 44.7 ± 5 kcal/mol.

Computational results are summarized in Table 1 (below), Table SI-1 and Scheme 7. The predicted structures for the trans-stereoisomers are best described as a chair conformation, with the phenyl group and vinyl hydrogen in an axial orientation relative to the ring.^[16] A twist-boat conformer lies at higher energy, with a low barrier to interconversion. This is similar to our earlier results for the parent structure.²⁶

Table 1.

Relative Energies (kcal/mol) from Computations on 4-X-Phenylcyclohexenes

CCSD(T) corr Results (a)
	3 (X=CH2)	1 (X=O)	4( X=NH)	2 (X=S)
Cis-trans energy difference	46.4	46.0	43.3	36.3
Barrier to π bond rotation from trans isomer	10.6	11.7	14.8	17.1
Barrier to retro Diels-Alder (rDA) from trans	30.0	13.1	21.4	17.0
rDA energetics from trans isomer	−14.5	−32.7	−24.0	−6.9
(a) CCSD(T) corr = CCSD(T)/cc-pvdz//M052X + DFT thermal correction (298 K)

Scheme 7. Substitution steric effect at 6-position.

One important observation is that the predicted DFT barrier (ΔG^‡) of 4.2 kcal/mol for trans to cis isomerization in phenylcyclohexene (Table SI-1)clearly is too low, presumably because of spin contamination. The CCSD(T) value of 10.6 kcal/mol (Table 1) is in better agreement with experiment. We also find excellent agreement between predicted and observed triplet energies (56.6 vs 56.0 kcal/mol) and the cis to trans energy difference (46.4 vs 44.7 kcal/mol). This high level of agreement provides confidence in our ability to model these reactions.

Computational results for phenylcyclohexene (3) and heteroatom substituted analogues, summarized in Scheme 6 and Table 1 (see SI for details), provide a straightforward explanation of our overall results. With phenylcyclohexene, the barrier to rDA (30 kcal/mol) is too high to compete with π-bond rotation (10.6 kcal/mol). For the oxo-analog (1), energies for the triplet state, π-bond rotational barrier and the cis-trans energy difference are essentially unchanged relative to the parent structure (3). Notably, strategic placement of an oxygen in the ring opens the second reaction channel by lowering the barrier to cycloreversion (13.1 kcal/mol for 1, X=O, vs 30 kcal/mol for 3, X=CH₂), thus making it energetically feasible for 1 to undergo rDA, while 3 reverts to the cis reactant.

Scheme 6. Estimated free energy surface (CCSD(T)//M052X) for 4-X-phenylcyclohexenes.

The aza-analog (4) also has a substantially lowered barrier (21.4 kcal/mol) to cycloreversion, but apparently not close enough to be competitive with π-bond rotation (14.8 kcal/mol). Interestingly, we see that the thia-analog (2) also has a low barrier for rDA (17.0 kcal/mol). However, the transition state for its π-bond rotation is also increased as a result of the longer C-S bonds and less strained ring to 17.1 kcal/mol, making these processes competitive. For the computational details on other analogs, see SI. Overall, oxygen substitution represents the “sweet spot” in this heterocyclic series where rDA is competitive with π-bond rotation. This is likely because formation of the strong C=O bond makes this the most exergonic cycloreversion (Scheme 6).

DFT computations with implicit solvation also show that the rDA barrier in 1 is lowered 2–3 kcal/mol by polar media, presumably because of the more polar transition state structure. This is consistent with our experimental observations (see SI for rate studies in various solvents) and support polarity in the cycloreversion transition state.

Next, a kinetic study was performed to realize the effect of steric hindrance (Et, ⁱPr, ^tBu) at the 6-position of 4-phenyl-3,6-dihydro-2H-pyran (Scheme 7).

The study showed that the initial rate increased with increasing steric demand of the substituents (k_tBu>k_iPr>k_Et>k_H). One explanation for this trend is a minor variation in energy barriers that is observed with a slight decrease in the retro-Diels–Alder (rDA) barrier as substitution size increases, accompanied by a slight increase in the π-bond rotation barrier (see SI for details). Additionally, increased steric crowding in the product diene may reduce the co-planarity of the conjugated double bonds making it less excitable, preventing product competition for excitation.²⁷

Next, we assessed the synthetic utility of the rDA reaction by screening differently substituted dihydropyrans for their ability to form dienes-on-demand (Figure 2). We have developed a one pot process to directly convert the pyrans into new cyclohexenes via subsequent Diels-Alder reaction with different dienophiles. Thus, in a one-pot process the photocatalyzed rDA of 1 was performed and followed by the addition of N-butylmaleimide, kept in the dark at room temperature for 24 h, to yield the thermal DA product cyclohexene 1b in 71% yield.²⁸ Although minimal, there was still some amount of dimerization which explained the remaining mass balance. Accounting for the recovered pyran, we obtained a 96% BRSM yield of 1b. In case of thiopyran analog 2, the rDA reaction was found to be less efficient yielding in low conversion to diene and affording only 26% yield of 1b, with the mass balance going primarily to ill-defined products and dimerization. Despite differences in their rates, different electron donating and withdrawing groups (i.e. Me, OMe, F, CF₃) at the para position of the pyran phenyl cleanly underwent the photocatalyzed rDA/DA reaction providing good yields of the corresponding cyclohexenyl-adduct (10b-13b) with >90% BRSM yield for all but 11. Meanwhile, ortho substitution on the pyranyl arene was not tolerated (14) and failed to react- (even at higher temperatures at 70 °C), likely due to allylic strain that prevented conjugation of the arene and double bond which is necessary for the initial excitation.^24,29,30

Figure 2.

Synthetic scope of the photocatalyzed retro-Diels-Alder reaction

Next, we synthesized a series of substrates with different alkyl substituents on the dihydropyran ring (15–20, Figure 2). Both ethyl and isopropyl substitution worked well and gave 68% and 63% of the corresponding cyclohexenyl-DA product (15b, 16b). The 6-tert-butyl group also performed well, providing 65% yield of the isolated diene³¹ In order for the initial excitation of the alkene to be feasible using visible light (rather than UV), an arene substituent is required, however, for the first time we show that phenylacetylene could also serve to activate cis-trans-isomerization and also led to the corresponding DA adduct in good yield (71%, 18b). This is a general and valuable expansion of the scope for EnT catalysis. Meanwhile, the styrenyl compound (25), which serves as a free rotor, did not undergo rDA, but instead underwent E-Z isomerization of the exocyclic alkene.

Surprisingly, in case of 6,6-disubstitution, instead of the expected dienes (19a, 20a), we isolated the products of a 1,5-hydrogen atom migration (19b, 20b)- also in good yield. As described in the Supporting Information (page 71), we attribute this to a triplet sensitized intramolecular hydrogen atom abstraction which has a predicted barrier of 11.1 kcal/mol. This process is similar to a Norrish Type II ketone photoreaction, in which hydrogen is transferred through a six membered ring transition state. NMR analysis of the reaction mixture with 19 revealed that, initially 19a was being formed in observable amounts, but quickly underwent isomerization to 19b under the reaction conditions. Although such conversion of conjugated dienes to more substituted isomer is reported under thermal-, photochemical-, and metal-catalyzed conditions, at this time it is not clear to us whether the migration is photocatalyzed.^32–35 In case of aromatic substitution at the 6-position, the reaction takes place but leads to an intractable mixture for both 21 and 22. This is not entirely unexpected, given that the dienes would be expected to be highly photoactive- as both of the alkene sites are activated with an arene unit. This is further supported by 23a and 24a, which contain ortho-substituted arenes at the 6-position, and due to allylic strain, are unable to conjugate with the alkene, and ultimately afforded dienes (23b, 24b) in 50% and 41% yield respectively.

With the unquestionable utility of the DA reaction in synthesis and a new approach to the generation of dienes, we wanted to ensure that our conditions were compatible with traditional DA-dienophiles. Delving deeper into the scope of dienophiles (Figure 2), symmetric dienophiles like maleic anhydride and dimethyl fumarate, delivered DA adducts (10c, 1c) in impressive yields of 73% and 70%, respectively, with >90% BRSM yields. Dimethyl acetylenedicarboxylate also showed excellent reactivity under inert conditions, providing the 1,4-cyclohexadiene moiety (1g) in high yields (72% and 95% BRSM), which was subsequently oxidized with DDQ to generate the biphenyl adduct (1g’, 90%). Even more intriguing, diisopropyl azodicarboxylate gave outstanding yields, showcasing the remarkable atom-swapping potential from C–O to N–N in a one-pot process (1h, 70% and 93% BRSM), enabling the efficient synthesis of substituted tetrahydropyridazines from aryl dihydropyrans. 1,4-Benzoquinone also performed well under these conditions, yielding the DA adduct in 70% (1f), which was easily isolated by simply washing the reaction mixture with hot hexane, bypassing the need for chromatographic purification. This adduct was then subsequently utilized as a dienophile with diene 1a, yielding DA adducts with two regioisomers (1fa and 1fb, 3:1), both as a single diastereomer.³⁶ This offers a convenient modular route to anthraquinone derivatives which are an important class of molecules.³⁷ We also tested unsymmetric dienophiles like methyl vinyl ketone and acrylonitrile, which provided excellent yields of the products (1d, 1e), exclusively forming the 4-substituted cyclohexenes.

Next, in terms of utility,³⁸ we wanted to briefly explore the dimerization process- we found that the dimer could be intentionally isolated in high yields by simply extending the reaction times, concentration and light intensity.

Of the four different regio-isomeric possibilities of the [4+2] adducts, we only observed a single product via the retro-DA dimerizing DA sequence.

In our hands, only one axially chiral dimerization product (±−4-vinylcyclohexene) was obtained. Simply, utilizing higher concentrations and running this reaction for extended time gave excellent yields across the range of electron donating and withdrawing groups (Scheme 8). Initial mechanistic experiments suggest that this is an excited state reaction and not an electron transfer reaction (see SI for details).³⁹

Scheme 8. Synthesis of arylvinylcyclohexene moieties via diene dimerization.

Additionally, we aimed to explore whether this new vinyl-cyclohexene was photoactive and could be sensitized and undergo subsequent isomerization to trans-cyclohexene, and subsequently be protonated by an acid. The addition of benzoic acid under the photocatalytic reaction conditions resulted in the formation of ester with 90% isolated yield (dr = 55:45) (Scheme 9). This suggests that they may be useful as molecular transducers⁴⁰- capable of converting photochemical energy into potential energy.

Scheme 9. Hydroesterification of 1i.

Again, illustrating the versatility of the diene-on-demand approach, we intercepted the diene with an appropriate alkene coupling partner by placing it in the reaction medium from the beginning of the rDA reaction. After screening several alkenes (including unactivated alkenes, conjugated alkenes, simple styrenes), we observed that by introducing 1,1-diphenylethylene derivatives into the reaction the photoexcited dienes reacted preferentially with the added alkene to yield new cyclohexene derivatives along with small amounts of homodimers (Scheme 10). Key to this reaction was the identification of alkenes that suitably stabilize the incipient radical and do not significantly outcompete the pyran for photoquenching (see SI for more details). For more discussion of the mechanism see the SI.

Scheme 10. Intercepting the dienes with 1,1-diphenylethylene derivatives.

Conclusion

In conclusion, we have introduced a new strategy for photochemically-driven retro-Diels-Alder reaction from 4-phenyl-3,6-dihydro-2H-pyran (1) and its derivatives in which we efficiently harvest the photochemical energy to generate a strain-loaded trans-isomer. This strain drives fragmentation to diene and dienophile units in competition with π-bond rotation. Although the photoreactivity of the newly generated diene can be a liability, it can be curtailed by modification of reaction conditions and careful monitoring of the reaction progression. Furthermore, we have shown that the rDA reaction can be a useful prequel to other cycloaddition reactions; generating the temporal diene on demand. Mechanistic and computational studies support a concerted but asynchronous charge separated transition state made possible by potential energy of the trans-isomer of the dihydropyran. In understanding our results, the overall correlation between theory and experiment is gratifying, but required computations beyond the usual DFT methods. Further, the studies illustrate that changes in the structure differentially impact the relative rates of the competing processes, and that oxygen inclusion is ideal to accomplish rDA. This ‘diene-on-demand’ strategy is triggered by visible light stimulation in the presence of an appropriate photosensitizer and enables the controlled generation of quasi-stable 1,3-dienes that can be utilized in diverse transformations- including cycloaddition with various dienophiles, homodimerization, and reactions with 1,1-diarylethylenes. Finally, we have demonstrated that visible light can drive a one-pot, atom-pair swap, reaction sequence that facilitates otherwise challenging two-atom skeletal editing.

Methods

A. General experimental procedure for photocatalyzed retro Diels Alder reaction

Light-promoted reactions were set up in fraction tubes (13×100 mm) charged with substituted 4-aryl-3,6-dihydro-2H-pyran (1 equiv) and photocatalyst (0.002 equiv) in dry dichloromethane (0.05 M) solvent. A rubber septum was used to seal the tube, which was then degassed by sparging with argon for 5 minutes. The tube was submerged in an ice-bath throughout the sparging process to avoid evaporation of the solvent. The degassed tubes were then placed in the photoreactor at 18 °C and irradiated with blue LED for indicated time. After reaching ~75% to the diene (checked by proton NMR), the reaction mixture was extracted with DCM and washed with water, brine, and dried over anhydrous MgSO₄ and the solvent was evaporated under reduced pressure. The crude product was purified by flash column chromatography (silica, eluting with pentane/ethyl acetate or hexane/ethyl acetate ) to afford the desired product.

B. General experimental procedure for one-pot conversion of the diene to [4+2]-adduct

After completion of the initial reaction (general procedure A), the reaction mixture was protected from light by wrapping the tube in foil and placed in a dark hood, then the dienophile (3 equiv) was added directly via syringe to the same reaction vessel and kept 24 hours at the indicated temperature (if the temperature is not specified, assume rt). For reactions at elevated temperatures, the reactions were performed in a sealed microwave vial (Thomas-scientific, Part number: 1145N29). Then, the reaction mixture was extracted with DCM, and washed with water, and brine. The organic layer was dried over anhydrous MgSO₄, filtered, and the solvent was evaporated under reduced pressure. The crude product was purified by flash column chromatography (silica, eluting with hexane/ethyl acetate) to afford the desired product.

C. General experimental procedure for tandem conversion to arylvinylcyclohexenes

General Procedure A was followed to set up the reactions with the exception being that the reaction concentration was increased to 0.4 M and the reaction times extended to yield the dimer product. The crude product was purified by flash column chromatography (silica, eluting with hexane/ethyl acetate) to afford the desired product.

D. General experimental procedure 4,4 diarylcyclohexenes

These reactions were set up in dry NMR tubes charged with substituted 4-aryl-3,6-dihydro-2H-pyran (1 equiv) and photocatalyst (0.002 equiv), and 1,1-diarylethylene (3 equiv) in dry dichloromethane (0.4 M) solvent. A rubber septum was used to seal the fraction tube, which was then degassed by sparging with argon for 5 minutes. The tube was placed in an ice-bath throughout the sparging process to avoid DCM evaporation. The degassed fraction tubes were then placed in the photoreactor at 18 °C and irradiated with blue LED. Upon diene consumption (monitored by proton NMR or TLC), the reaction mixture was extracted with DCM, washed with water, brine, and the organic layer was separated and dried over anhydrous MgSO₄. The solvent was evaporated under reduced pressure. The crude product was purified by flash column chromatography (silica, eluting with hexane/ethyl acetate) to afford the desired product.

Further details regarding the methods can be found in the supplemental information.

Resource availability

Lead contact.

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Jimmie Weaver (jimmie.weaver@okstate.edu).

All unique/stable reagents generated in this study are available from the lead contact without restriction.

Data: Copied of NMR and cartesian coordinate data are reported in the supplementary information. The raw FID files will be shared by the lead contact upon request.

Code: This paper does not report original code.

Additional information: No additional information is needed to analyze the data.