Photomechanochemical control over stereoselectivity in the [2+2] photodimerization of acenaphthylene

Abstract

Tuning solubility and mechanical activation alters the stereoselectivity of the [2+2] photochemical cycloaddition of acenaphthylene. Photomechanochemical conditions produce the syn cyclobutane, whereas the solid-state reaction in the absence of mechanical activation provides the anti. When the photochemical dimerization occurs in a solubilizing organic solvent, there is no selectivity. Dimerization in H₂O, where acenaphthylene is insoluble, provides the anti product. DFT calculations reveal that insoluble and solid-state reactions proceed via a covalently bonded excimer, which drives anti selectivity. Alternatively, the noncovalently bound syn conformer is more mechanosusceptible than the anti, meaning it experiences greater destabilization, thereby producing the syn product under photomechanochemical conditions. Cyclobutanes are important components of biologically active natural products and organic materials, and we demonstrate stereoselective methods for obtaining syn or anti cyclobutanes under mild conditions and without organic solvents. With this work, we validate photomechanochemistry as a viable new direction for the preparation of complex organic scaffolds.

Graphical Abstract

Unique and mutually orthogonal photomechanochemical and photosolvochemical stereoselective routes for cycloaddition of acenaphthylene with high yields have been reported. These methods provide scalable and environmentally benign approaches for stereoselective photocycloaddition reactions.

Introduction

Mechanochemistry involves^1–3 the use of force/strain/curvature to make or break chemical bonds, and such methods are attractive because they satisfy all twelve principles of green chemistry.⁴ In addition to the minimal environmental impact, the benefits of mechanochemical reactions can include faster reaction kinetics,⁵ improved stereoselectivities and yields, or access to stereo- and regioisomers, or other chemical products,⁶ that are not accessible by solvothermal methods. Recognizing the advantages of mechanochemistry and the growing suite of tools for performing and studying reactions under force, the diversity of reactions that have been performed mechanochemically is rapidly expanding, and includes the formation of metal-organic frameworks,⁷ synthetic polymers,⁸ and small molecules.^{1, 9} Amongst the reactions that have been most thoroughly studied under mechanochemical activation is the thermally allowed [4+2] Diels-Alder^{3, 10, 11} reaction, and it is now known that this reaction can be driven forward with force because of its negative volume of activation.^12–14 The Diels-Alder reaction is an example of a broader class of reactions, the pericyclic cycloadditions,¹⁵ that also have negative activation volumes, and many of these have been carried out mechanochemically as well.^{10, 16, 17} A class of pericyclic reactions which should also possess negative activation volumes, and thus should be accelerated by the application of force, are symmetry-allowed [4n] photochemical cycloadditions, which are widely exploited in organic synthesis to prepare complex molecular scaffolds.^{14, 18, 19} The potential benefits of performing photochemical cycloadditions mechanochemically include improved yields and stereochemical control, the latter of which is particularly important as a mixture of stereoisomers are often obtained.²⁰ Despite these potential advantages, [4n] cycloadditions have not yet been investigated under photomechanochemical conditions – where both illumination and force are applied to the reactants simultaneously – because of challenges associated with developing instrumentation that is capable of illuminating the reactants during compression. To this end, here we implement a new reactor to drive a [2+2] cycloaddition reaction under compression, and, in doing-so, show that such conditions are highly stereoselective as a result of a unique photomechanochemical reaction pathway.

The symmetry-allowed [2+2] photochemical cycloadditions of alkenes to form cyclobutanes,²⁰ are frequently used to form primary and secondary metabolites^{21, 22} with attractive antimicrobial,^{23, 24} antibacterial,²⁴ and analgesic properties,²⁴ and are increasingly explored as a responsive component of functional materials.^25–27 As a consequence, developing methods for their construction remains a focus of the synthetic community.²³ One challenge that still remains, however, is controlling the stereoselectivity of these photoreactions because an undesired mixture of cyclobutane stereoisomers is often produced.^{20, 28–30} For example, the dimerization of acenaphthylene is considered a model reaction for understanding how experimental conditions affect the yields and stereochemical outcomes of [2+2] photocycloadditions.^31–36 In organic solvents, a mixture of syn and anti isomers of the acenaphthylene dimer is obtained,³⁷ and, generally, the former is the major product.³⁴ Strategies to alter the reaction outcome so that the primary product is anti include the use of heavy atom solvents,³⁴ photosensitizers,³⁴ or a combination of a photosensitizer and a molecular cage,³⁸ but such conditions often require undesirable solvents, use expensive catalysts, or continue to produce a mixture of isomers. As such, further efforts are needed to understand the drivers of stereoselectivity, which could result in design rules that could be deployed to develop simple, scalable, and widely accessible methods for the stereochemical control of [2+2] photochemical cycloadditions that minimize the use of organic solvents and complex and expensive reagents.

Results and discussion

Here we study the [2+2] photodimerization of acenaphthylene under four different conditions, (i) illumination of the solid-state acenaphthylene under ball milling, (ii) illumination of the solid-state crystal in the absence of ball milling, (iii) illumination in organic solvents (soluble phase), and (iv) illumination in H₂O (insoluble phase). For (i), we find that primarily the syn dimer is selectively formed, whereas (ii) and (iv) selectively form the anti dimer. Interestingly, we do not observe any noticeable selectivity for (iii). First-principles Density Functional Theory (DFT) calculations reveal that the extent of solubility changes the cycloaddition pathway: for insoluble and solid-state conditions (b and d), the cycloaddition proceeds via the formation of a covalently bonded excimer, which is geometrically and energetically closer to the anti transition state, thereby explaining the anti selectivity observed in (ii) and (iv). Additionally, we reveal that the noncovalently bound syn dimer is more mechanosusceptible, as it experiences more destabilization upon mechanical compression, which underpins the formation of syn product under ball milling (a). Therefore, in addition to providing mild conditions for the selective production of both syn and anti cyclobutanes that avoid the use of organic solvents, this work lays the foundation for understanding photomechanochemical stereoselectivity, which in turn, could lead to environmentally benign methods for the production of important chemical reagents.

Solid-state [2+2] photocycloaddition of acenaphthylene

We first investigated whether the [2+2] dimerization of acenaphthylene can be driven photomechanochemically and how these conditions affect reaction yield and stereoselectivity. To do so, a ball-mill reactor (SPEXSamplePrep, 8000M) was modified (Figure S1) with a blue LED (HepatoChem, DX Series light 30 W, λ_max = 450 nm), and the reaction was run in a glass vial (20 mL) with two methacrylate balls (SPEXSamplePrep, 8006A, 9.5 mm diameter, 350 mg) so that light could reach the reagents during milling. In addition, we found it necessary to fluorinate the vial³⁹ and add silica gel (SILICYCLE, SilicaFlash^® GE60, 70–230 mesh) to the reaction to prevent the reagents from adhering to the side walls. Acenaphthylene (1.2 g) and silica gel (1.2 g) were added to the vial under inert Ar atmosphere, unless otherwise noted, and milled for 20 h at a frequency of 17.7 Hz. Upon completion of the milling, yields and anti:syn ratios were determined by ¹H NMR spectroscopy by dissolving a portion of the crude mixture in CDCl₃ followed by filtration to remove silica from the CDCl₃ solution. We found (Table 1, entries 1 – 2) that, photomechanochemically, the dimerization reaction proceeded with yields up to 96 % with 6:94 anti:syn selectivity. Stereochemical assignments were confirmed by ¹H NMR spectroscopy and microcrystal electron diffraction^{40, 41} on the crystallites isolated from the reactions (Figure S26 and Table S2). This syn selectivity and yield were retained in ambient atmosphere. In the absence of the illumination, no product was observed. To put this result in the context, in organic solvents, this reaction favors the syn product but with poor selectivity, and high syn selectivity has only been obtained using expensive and complex reagents or undesirable organic solvents.^42–45 For example, Kaanumalle et al.⁴⁶ reported 99 % syn product (99 % yield) in borate buffer using an organometallic molecular cage, while Cowan et al.³⁴ obtained 98 % syn (39 % yield) selectivity in O₂-saturated benzene. Thus, we find that photomechanochemical conditions provide the mildest route towards the syn product of the acenapthylene dimer yet reported.

Table 1.

Solid-state [2+2] photocycloaddition of acenaphthylene.

Entry	Atm.	Total Product Yield (%)	Anti : Syn
1 a	Ar	95	6:94
2 a	air	87	8:92
3 b	Ar	65	70:30
4 b	air	99	46:54

All reactions were performed at 20 °C. Acenaphthylene samples were irradiated with a blue LED (HepatoChem, DX Series light 30 W, λ_max = 450 nm) for 20 h. Yields and selectivities were obtained from ¹H NMR in CDCl₃.

^[a]Ball-mill reactions performed in the presence of silica as an additive to prevent the reagents from adhering to the side walls of the reaction vessel.

^[b]Reactions were performed in a petri dish with ground acenaphthylene crystals and no milling during illumination.

All reactions were performed at 20 °C. Acenaphthylene samples were irradiated with a blue LED (HepatoChem, DX Series light 30 W, λ_max = 450 nm) for 20 h. Yields and selectivities were obtained from ¹H NMR in CDCl₃. [a] Ball-mill reactions performed in the presence of silica as an additive to prevent the reagents from adhering to the side walls of the reaction vessel. [b] Reactions were performed in a petri dish with ground acenaphthylene crystals and no milling during illumination.

From these experiments it was unclear whether the syn selectivity was the result of carrying out the reaction in the solid state or whether force played an active role in dictating stereoselectivity. To determine whether force has a role in stereoselectivity, the reaction was carried out on acenaphthylene crystals by illuminating them in the absence of force. The acenaphthylene crystals (100 mg) were ground with a mortar and pestle, sealed in a petri dish under either inert or ambient atmosphere, and the crystals were then irradiated with the same blue LED for 20 h (Table 1, entries 3 – 4). Under Ar, we found that the reaction favored the anti isomer (70:30 anti:syn), while ambient atmosphere provided a mixture with a 46:54 anti:syn ratio. So, in the absence of O₂, more anti product forms, but as O₂ is introduced, such selectivity is diminished. This was also observed by Haga et al.,³⁶ who noted that the presence of a triplet quencher (here O₂) reduces the anti-selectivity. These results show that the absence of force provides substantially different products than when the products were milled during illumination. Therefore, we find that running the [2+2] photodimerization of acenaphthylene under Ar while milling results in the opposite stereoselectivity than when the reaction is carried out without milling. These observations confirm that a unique photomechanochemical reaction pathway exists for the dimerization of acenaphthylene that is distinct from the pathway followed during irradiation in the absence of force.

[2+2] photocycloaddition of acenaphthylene in solution-state

We then studied the photodimerization of acenaphthylene in different organic solvents with variable H₂O:organic solvent ratios at the room temperature. Here we have examined the photodimerization with 5 different organic solvents that are miscible with H₂O – DMSO, EtOH, MeOH, DMF, and MeCN – along with H₂O itself, under blue light illumination. As acenaphthylene is insoluble in H₂O, these experiments provide insight into the role of solubility on the stereoselectivity. To maximize solvent/acenaphthylene interaction (especially for the solvent mixture where acenaphthylene is poorly soluble), 5 M solution of acenaphthylene was prepared in DMSO, and 4 μL of this stock solution was then added to 1 mL of the desired solvent system to prepare 20 mM solutions of acenaphthylene. The solution was then sonicated for 1 min. The summary of the stereoselectivity observed in these experiments is given in Figure 1C and Table S1. Although, we find that the absolute value of the anti:syn ratio varies depending on the nature of the organic solvent, we observe the conspicuous trend that the anti selectivity increases with the amount of H₂O, where using just H₂O selectively provides the anti product. Because of the insolubility of the acenaphthylene in H₂O, the solution was sonicated to disperse the reactant, which led to a cloudy, opaque solution. Following irradiation, the anti isomer was recovered as the major product (84:16 anti:syn) in 84 % yield. In other words, the insolubility of acenaphthylene favors the anti product. This observation is in good agreement with the solid-phase reactions on the crystal in the absence of milling, which also resulted in anti stereoselectivity (anti:syn 70:30). Thus, we conclude that solubility also plays an important role in determining the stereoselectivity. For photocycloadditions, it is known that lowering the reaction temperature can improve regio- and stereoselectivity.⁴⁷ As such, the reaction temperature was then lowered to 10 °C, and running the reaction for 16 h in H₂O resulted in quantitative conversion, with anti:syn reaching 91:9. The anti selectivity observed here is similar to those reported by Guo et al.³⁸ who used a molecular cage composed of a Ru²⁺ photocatalyst to achieve 89:11 anti:syn. So a major challenge in stereocontrolled photochemical cycloadditions – obtaining the anti isomer of acenaphthylene as the major product – was resolved by simply running the reaction in H₂O, which, in addition to providing an otherwise elusive product, is both atom-efficient and environmentally benign.

Figure 1.

(A) [2+2] Photocycloaddition of acenaphthylene, with the most stereoselective conditions for syn and anti dimers noted. (B) ¹H NMR (300 MHz, CDCl₃) of the products of the [2+2] photocycloaddition of acenaphthylene in a binary solvent mixture composed of different ratios of MeCN and H₂O. (C) Change in selectivity with changing solvent composition for the [2+2] photocycloaddition of acenaphthylene in binary solvent mixtures.

DFT calculations for the [2+2] photocycloaddition of acenaphthylene in solution-state

We performed DFT calculations⁴⁸ to understand why liquid and solid phase reactions result in different stereoisomers. In doing so, we explored the potential energy surfaces by considering the photodimerization pathways from three different initial states– syn and anti conformers, which are noncovalently bound supramolecular dimers, and a state that consists of two separated monomers (Figure 2). On the other hand, the noncovalently bound syn and anti conformers correspond to the insoluble solid-state acenaphthylene (and state present in H₂O), when the monomers are close to each other and experience a net stabilizing interaction as compared to separated monomers. We find that the thermal activation barrier for the anti product formation (179 kJ/mol) is lower than the syn product formation (269 kJ/mol) (Figure 2). Such high barriers suggest that the reactions cannot be performed at room temperature without photoexcitation, which we also observe experimentally. In the soluble-phase scenario, we find that the excited triplet (T) and singlet (S) states of the separated monomers (yellow dotted lines in Figure 2) are above the syn and anti transition states (TS). Thus, when completely solvated, photoexcitation injects sufficient energy to access both the anti and syn TS; this in turn, leads to both syn and anti product formations, explaining the loss of selectivity for this reaction in organic solvents. In the solid state or in H₂O, where the poor solubility keeps the acenaphthylene aggregated, the molecules exist as noncovalently bound syn and anti supramolecular conformers. We find that the spin-flipped T states for both the syn and anti supramolecular conformer lead to the formation (shaded cyan region in Figure 2) of a covalently bonded excimer (excimer in Figure 2). Formation of excimers in photochemical reactions have been reported previously^49–51 to have a role in determining the selectivity of the products. Therefore, as the photoexcitation occurs in the insoluble phase, the supramolecular dimers form a covalent bond in the excimer, gaining a stabilization of nearly 117 kcal‧mol⁻¹ per dimer compared to the vertically excited T state. We find that the geometry of this bonded excimer resembles more closely to the anti transition state than the syn. Additionally, as the anti TS has lower energy than the syn TS, energetically it is also easier to access the anti TS from the bonded excimer. Thus, the formation of the bonded excimer preferentially leads to the anti product. This explains the selective formation of the anti product in the insoluble phase, i.e. in H₂O and in crystals in the absence of ball milling.

Figure 2.

Potential energy surface obtained using the geometries at the B3LYP-D3/6-31G(d) level of theory showing the pathways for syn and anti product formation. The abbreviations are as follows: T: Triplet; S: Singlet; TS: Transition state. Dotted arrows represent vertical photoexcitation, and cyan shading illustrates relaxation of the triplet conformer states to the excimer.

DFT calculations for the [2+2] photocycloaddition of acenaphthylene in solid-state

To understand the origin of syn selectivity that occurs under the photomechanochemical conditions, we have applied force (F) along the C–C bond forming vector of the syn and anti conformations using the external force explicitly induced (EFEI) method.^{52, 53} In this simulation, the intertwined effect of F and photoexcitation is not incorporated, rather only the effect of compressive F on stability is studied. We envisage that ball milling exerts compression by bringing the monomers in the syn and the anti supramolecular conformers closer to each other. Thus, applying forces along the C–C bonds in the simulation (as shown in Figure 3A) can have an effect similar to what occurs in a ball mill. We find that the syn conformer undergoes bond formation at lower F (between 8 nN and 9 nN), compared to the anti dimer (between 12 nN and 13 nN), which is shown by the grey shaded regions in Figure 3B. To understand why the syn supramolecular conformer is more susceptible to F, we examined the F range of 5–8 nN with more datapoints. The reaction coordinate (RC) in Figure 3C is defined based on the C–C distance, wherein the bond distances at 5 nN and 8.55 nN correspond to RC of 0 and 1, respectively. We find that the slope of RC vs. F (black dashed lines in Figure 3C) is higher for the syn supramolecular conformer. This shows that a more significant change in RC is attained for syn than anti at the same applied F. In other words, the results directly confirm that syn supramolecular conformer is more mechanosusceptible, meaning that its RC has a stronger dependence on F. In addition, the syn supramolecular conformer experiences more destabilization for the same applied F than the anti (filled markers in Figure 3C). To analyze these effects, we implement the energy decomposition analysis (EDA)⁵⁴ to understand the origin of the mechanosusceptibility by quantifying the interaction energy between the acenaphthylene monomers (Figure 3D). EDA consists of three terms: frozen density (E_FZ), charge transfer (E_CT), and polarization (E_Pol).⁵⁴ We find that the E_FZ dominates in the higher F regime, and it increases drastically near 7–8.55 nN for the syn, but anti does not exhibit such behavior (Figure 3D). As E_FZ captures the interaction between the unrelaxed electron density between the monomers, we conclude that the face-to-face orientation of the π-clouds of the syn supramolecular conformer, and the resulting repulsion at smaller distances, make it more mechanosusceptible, whereas for the anti supramolecular dimer, π-clouds interact to a lesser extent. Additionally, it is worth noting that the TS for the syn product formation corresponds solely to compression along the C–C bonds participating in the [2+2] cycloaddition (syn TS in Figure 2). On the other hand, for the anti product formation, shear force also plays an important role along with compression as the monomers need to slide to form the anti product (anti TS in Figure 2). We conjecture that as attaining shear and compression simultaneously is statistically less probable under ball milling than just compression, syn selectivity is boosted. Thus, these results explain qualitatively the stereoselectivity of solid-state photomechanochemical conditions.

Figure 3.

Effect of force on the dimerization of syn and anti supramolecular conformers. (A) The directions of the applied forces in simulations using the external force explicitly included (EFEI) method are shown. (B) The effect of applied force on C–C bond formation for syn and anti conformers. The shaded regions indicate where the syn and/or anti dimers undergo covalent C–C bond formation from a noncovalent conformer. (C) The change in reaction coordinate (RC) and energy (E) as a function of the applied force. The RC is defined based on the C–C distance, wherein distances at 5 nN and 8.55 nN correspond to RC 0 and 1, respectively. The filled circles/diamonds correspond to the E values, whereas unfilled markers indicate the change in the RC. (D) Energy decomposition analysis showing the effect of force on different inter-action terms between the monomers. The abbreviations are: FZ: frozen density, CT: charge transfer, Pol: polarization.

Conclusion

In conclusion, we have studied the [2+2] dimerization of acenaphthylene and explored how light, force, and solvent can be combined to achieve both syn and anti stereoselective photodimerization. We find that syn stereoselectivity is observed under photomechanochemical conditions because of the greater mechanosusceptibility of the noncovalently bound syn supramolecular conformer compared to the anti conformer. In studying how solvents and solubility affect yields and stereoselectivity, we have shown that simply running the photodimerization in H₂O provides the anti dimer in high yields, and these conditions provide among the highest anti stereoselectivity yet observed because of the formation of a bonded excimer in insoluble phases. Thus, we report scalable and mild conditions for obtaining both syn and anti [2+2] cycloaddition products without necessitating organic solvent or complex additional reagents. More importantly, we demonstrate that photomechanochemical pathways exist that are distinct from solvothermal or solid-state reaction pathways, and we explain stereochemistry based upon mechanosusceptibility, a concept that may be used to predict stereoselectivity for other mechanochemical reactions as well. In doing so, have shown that photomechanochemistry and photosolvochemistry can provide mutually orthogonal stereoselective routes for making complex organic scaffolds. Therefore, photomechanochemistry merits further investigation as a scalable and environmentally benign approach to stereoselective photochemistry.

Experimental Section

Experimental procedures, synthesis, and characterization of all new compounds, NMRs, Microcrystalline electron diffraction are provided in the supporting information file.