Air-Water Interface Effects on the Regioselectivity of Singlet Oxygenations of a Trisubstituted Alkene

Abstract

The regioselective synthesis of allylic hydroperoxide sulfonates by singlet oxygenation at the air-water interface has been found to depend on the concentration of the alkene sulfonate and added calcium salt. The regioselectivity is proposed to originate from an orthogonal alkene relative to the water surface for preferential methyl hydrogen abstraction by airborne singlet oxygen in an ene reaction. The findings hint that the air-water interface is a locale for synthetic reactions.

Graphical abstract

Introduction

Organic chemists have yet to exploit the air-liquid interface for synthetic reactions. Nature—on the other hand—has synthesized complex molecules with prebiotic chemistry at the air-water interface¹ or in appropriate dynamical micellar systems.² Indeed, reactions at phase boundaries (e.g., air-water,³ liquid-liquid,⁴ solid surfaces⁵) can be unique and provide selectivity relative to the bulk phase. Such selectivity was recently seen at the air-water interface by airborne ¹O₂ with a trisubstituted alkene surfactant, 8-methylnon-7-ene-1 sulfonate (1) (Figure 1).^6,7 Two hydroperoxides were formed in an ene reaction,⁶ one was a secondary hydroperoxy sulfonate (2) and the other was a tertiary hydroperoxy sulfonate (3) with the preferred formation of regioisomer 2. In contrast, the homogeneous ¹O₂ ene reaction with trisubstituted alkenes such as 2-methyl-2-pentene occur without selectivity for methyl and methylene hydrogens,⁸ producing secondary and tertiary hydroperoxides in a ~1:1 ratio. Scheme 1 shows an exception⁹ where a sterically hindered allylic group leads to methylene H abstraction and preference for the tertiary regioisomer due to an unfavorable 1,3-interaction of the proximal perepoxide oxygen with the large allylic group.

Figure 1.

Schematic of the three-phase reaction for the delivery of airborne ¹O₂ to the water surface. Red laser light passes through a sensitizer-bound glass wafer which is not in contact with the water. A small air space between the wafer and solution is bridged before ¹O₂ gas reaches the liquid for alkene oxidation.

Scheme 1.

The purpose of this paper is to extend our study on the regioselectivity of airborne ¹O₂ with 1, which we now report depends on reagent concentration and added calcium salt. Hydroperoxides 2 and 3 are the only products with percent conversions reaching as high as 100%, where lower alkene 1 concentrations increased the regioselectivity for 2, but concentrations above the critical micelle concentration (CMC), or with the addition of Ca²⁺ ions or use of acetonitrile-water mixtures showed no regioselective preference for 2 over 3.

Results and Discussion

Figure 1 shows our device, which resembles a flow system¹⁰ and a device developed by Midden¹¹ for the clean and pure production of ¹O₂. Our device irradiated samples from above with red diode light through a silica wafer. The silica wafer had a green color as it was coated with aluminium (III) phthalocyanine (Pc) chloride tetrasulfonic acid. Pc sensitizers often are not maintained as monomers in H₂O,¹² which was not of concern for us because the sensitizer wafer was not in contact with water. In Figure 1, ¹O₂ gas was formed at the bottom of the wafer, and crossed an air gap to reach 1 at the water layer underneath.

Included in Table 1 are data for four reaction conditions that indicate the importance of environmental factors in the alkene sulfonate 1 photoperoxidation: (i) with a concentration of 1 (1 mM) that covered the water with a single layer of alkene molecules, a 7:3 regioselective preference for hydroperoxide 2 over 3 was seen (Table 1, entries 1 and 2). However, the selectivity of hydroperoxide products (2 and 3) was lost (ii) with a 25-mM concentration of 1 (i.e., above its CMC, 9.7 mM)⁶ (entries 2 and 3), (iii) with added Ca²⁺ ions (CaCl₂, 1–10 mM) (entries 4–6), and (iv) in an acetonitrile/water mixture to dissolve 1 (entries 7 and 8). The efficiencies of the reaction by airborne ¹O₂ transfer to solution were also investigated.

Table 1.

Yields and Ratios of Hydroperoxides 2 and 3 Formed by Air-Liquid ¹O₂-Ene Reaction^a

entry	interface	[1], (mM)	CaCl2 added (mM)	% yield b	product ratio 2:3 b,c
1	air-H2O	1	0	54 ± 4	70:30 (±5)
2	air-D2O	1	0	92 ± 2	69:31 (±1)
3	air-D2O	25	0	25 ± 6	47:53 (±2)
4	air-D2O	1	1	25 ± 7	52:48 (±2)
5	air-D2O	1	5	20 ± 7	49:51(±1)
6	air-D2O	1	10	20 ± 8	41:59 (±5)
7	air-liquid [CH3CN:H2O (9:1)]	1	0	46 ± 4	43:57 (±3)
8	air-liquid [CD3CN:D2O (9:1)]	1	0	100 ± 1	49:51 (±1)

^aSamples (0.6 mL) were illuminated with 669-nm diode laser light exiting a fiber optic positioned above the sensitizer wafer for 1 h at 26 °C. Airborne ¹O₂ was produced at the sensitizer wafer and crosses an air gap to reach the solution. The distance between the wafer and the liquid surface in water was ~0.4 mm (at the walls of the vessel) to 1.5 mm (center of meniscus), and was ~1 mm for the acetonitrile-water samples.

^bThe values shown here are an average of 2 or more measurements.

^cThe ratio of 2 and 3 were calculated from the integration of ¹H NMR signals at 4.8 and 5.5 ppm, respectively.

Singlet oxygen reactions are usually very sensitive to solvent isotope effects;¹³ for example, in homogeneous solution the ¹O₂ lifetime (τ_Δ) is 20-fold lower in H₂O (3.5 µs) compared to D₂O (65 µs). We found an interesting contrast that the product yields from our ¹O₂ reaction on a protio surface is only 2-fold lower than on a deutero surface (compare entries 1 with 2, and 7 with 8). This modest solvent isotope effect in quenching of ¹O₂ by 1 is rationalized by ¹O₂ not transferring deep into the solution. It may penetrate into the solution distances of 880 nm in D₂O and in 150 nm in H₂O,¹⁴ where the surroundings and location of hydrophobic group of 1 play a role in its reactivity.

Based on our findings, the orientation of the alkene is proposed to be important (Figure 2). In Figure 2i, the regioselectivity at sub-micellar concentrations is proposed to arise from a perepoxide transition state (TS) oriented with methyl groups pointed up in a less solvated state for easier abstraction by airborne ¹O₂. With an orthogonal sheet surfactant, the hexyl sulfonate chain in 1 is not acting as a bulky allylic group as mentioned in the Introduction, or else 3 and not 2 would be the anticipated major product.⁹ A product dependence on alkene wetting would connect to an ongoing discussion in ¹O₂-ene chemistry, the idea of the perepoxide as a transition state vs an intermediate in gas and solution phases,^15–17 which has been of substantial interest. In Figure 2 (ii–iv), the methyl and methylene hydrogens appear to be equally abstractable in 1 and the regioselectivity is lost. In Figure 2ii, above the CMC, ¹O₂ can oxidize but also proceed through the surfactant monolayer and reach the micelle, where micelles can assist in O₂ solubilization compared to that in the surrounding aqueous solution¹⁸ and ¹O₂ likely collides with several micelles during its lifetime.¹⁹ In Figure 2iii, the Ca²⁺ additive induces an accumulation of 1 at the air-water interface due to its suspending power. Thus, the orthogonal geometry is thought not to be preserved in the presence of Ca²⁺ ions. In Figure 2iv, the lack of selectivity in acetonitrile/water can be explained where the surfactant is solubilized and not expected to adsorb at the interface or aggregate well for micellization,²⁰ analogous to 2-methyl-2-pentene in homogeneous solution.⁸

Figure 2.

Mechanistic summary of airborne ¹O₂ reaction of alkene 1 at sub-micellar concentrations (i), micellar concentrations (ii), in the presence of Ca²⁺ ions (iii), and dissolved in acetonitrile/water (iv).

Our data in Table 1 also led us to seek alternative explanations for the regioselectivity seen in entries 1 and 2. Notably, there are reports where secondary and tertiary hydroperoxides can decompose at different rates when encapsulated in zeolites.²¹ Control reactions showed that 2 and 3 did not decompose under the reaction conditions prior to quantification. The hydroperoxides were stable for several days at room temperature, where after 1 h, 2 decomposed completely at 100±3 °C, and 3 decomposed completely at 185±5 °C. Because allylic hydroperoxides can rearrange thermally by O₂ shift in a radical rearrangement, i.e. the Schenck mechanism,²² we further examined the stability of 2 and 3 toward isotope exchange of ¹⁸O₂ for introduction of the ¹⁸O-labels. However, oxygen sparging with ¹⁸O-labeled O₂ in solution produced no hydroperoxide mass increase of 4 amu based on LCMS data. Thus, the regioselectivity is inconsistent with a mechanism that involves a hydroperoxy radical rearrangement.

Conclusion

Thus, we have developed a phase-separated reaction, where regioisomer 2 is favored at the air-water interface by the orientation of the alkene methyls toward incoming (airborne) ¹O₂ in dilute solution of 1. The origin of the selectivity is an unequally wetted alkene in this synthetic process. The delivery of airborne ¹O₂ to the alkene site aggregated at the interface or in a micelle or dissolved in acetonitrile/water showed no selectivity. Future studies will be needed for a deeper understanding of conformational properties of this ¹O₂ ene reaction at the interface, such as dynamics of the alkene group and tilting,²³ as well as possible selectivity dependence on the alkene sulfonate chain length.