A Photochemical Microfluidic Reactor for Photosensitized [2+2] Cycloadditions

Abstract

Here we report a microfluidic system for photochemical cycloadditions fabricated using silicon micro processing technologies. The system was optimized to yield residence times of just a few minutes for a range of photochemical [2+2]-cycloaddition reactions facilitated using high power UV-LEDs at 375 nm and triplet photosensitizers, which removed the need for the low wavelengths typically required for these types of transformations. Adducts using different excitable olefins with different linear, carbocyclic, and heterocyclic coupling partners were explored to demonstrate the feasibility of performing photochemistry in microflow in an academic research environment. Finally, a reaction leading to a novel dihydrooxepin-2(3H)-one scaffold and a mechanistic proposal for its formation are reported.

[2+2]-Cycloaddition reactions are a rapid and atom-economic way to form strained cyclobutane ring systems.[1] While originally promoted by very high energy and destructive UV-C light (200–280 nm), more recent advances have used photosensitizers, allowing these reactions to proceed with lower energy UV-A light (>350 nm) or even visible light.[2] [3] Despite this, many intermolecular [2+2] reactions are still carried out with high energy UV-C light.[4] Photochemical reactions have also benefitted greatly from advances in flow chemistry, which enable a more even exposure to light compared to conventional round-bottomed flask set-ups, a higher surface-area-to-volume ratio, and more controlled reaction conditions.[5] While equipment for carrying out photochemistry on large scales in flow have been reported,[6] microfluidic devices for photochemical reaction screening are less common.[7] [8]

Herein, we report the design and production of a microfluidic reactor for photochemistry and demonstrate its use in the optimization of simple [2+2] cycloadditions using thioxanthone (TX) as a photosensitizer. As compared to similar, previously reported transformations,[9] [10] this enables the use of lower-energy UV-A light (375 nm) and/or short reaction times, while achieving comparable yields. Finally, we uncover an unexpected competing rearrangement pathway for a tetrasubstituted alkene which gives rise to an unprecedented dihydrooxepin-2(3H)-one core.

We initially desired to fabricate an enclosed reactor with high transmittance of UV-A irradiation, limited compliance effects for rapid steady-state conditions, high thermal and chemical resistance for reactor versatility and with high resolution within the microfluidic domain for examination of advantageous scaling relations. We chose to fabricate the base of the reactor from silicon due to its high thermal and chemical stability as well as the many available processing techniques available for producing highly tailored structures with high resolution at the submicron level. Since silicon does not transmit UV irradiation, a lid with the desired properties was required. For this purpose, Borofloat^® 33 was selected, which only allows transmittance of UV irradiation in the UV-A region (>90%). Besides its optical, chemical, and thermal properties, the composition includes an adequate concentration of alkali ions. These allowed for sealing between Borofloat^® 33 and silicon through anodic bonding techniques forming irreversible siloxane bonds between the two materials. To fabricate an enclosed reactor a Borofloat^® 33 wafer was first structured using a laser micromachining tool before anodic bonding.

For the initial reactor design a simple meander structure was chosen with a single inlet and outlet and characteristic channel dimensions of 800 μm × 200 μm (W/H, Figure [1a]). Channels were produced using KOH in a wet etching process resulting in trapezoid channel geometry with slanted side walls [see Figure [1] in the Supporting Information (SI)]. To avoid significant undercutting in convex corners a beam structure along the (100) plane was added to the design for protection of the corners (see Figures [2–4] in the SI). With a final design channel length of 3.71 m, 200.31 μm average channel depth and using the theoretical angle between the (111) and (100) plane of 54.7°, an internal reactor volume of 489 μL was found. In order to protect the user from UV-A radiation a TLC UV chamber was repurposed to fit the UV LED using leadscrews for variable light source height (Figure [1b]). Initially, 10 W 365 nm and 375 nm UV LEDs were tested, however, to increase the irradiance of the light source, 8 × 10 W 375 nm LEDs were subsequently used at 90% duty cycle (Figure [1c] and Figures [5] and 6 in the SI). The current was limited due to heat generation from the LEDs. The UV chamber reached 35 °C at steady state using the LEDs at a power measured to 53 W (ca. 66% of rated value).

Table 1 Photochemical Reactions Carried Out in Flow^a

Entry	Enone	Alkene/yne	Product	Yield (%)b
1	1a			42
2	1b			41
3	1b			18
4	1b			33
5	1b			53
6	1b			45
7	2			44
8	2			46

^a Unless otherwise specified, the retention time was 3 min; TX = thioxanthone.

^b Yields are of pure compound isolated after column chromatography.

To test the suitability of our developed system as a photochemical reactor we applied it to the [2+2] cycloaddition of N-methylmaleimide and diversely functionalized alkenes and alkynes. While similar reactions have been explored,[9] [11] [12] they require expensive iridium-based photocatalysts or high-energy UV light, typically 250 nm in wavelength.[13,14] As an alternative, we sought to identify a readily available photosensitizer (PS) which was both cheap and enabled the use of lower energy light, reducing operational risks of carrying out the transformation. As thioxanthone (TX) had been recently used successfully in the photosensitized [2+2] cycloaddition of citraconic anhydride and ethylene, we focused our initial efforts on this PS.[15] Although no examples of intermolecular [2+2] cycloadditions involving simple maleimides catalyzed by TX had been reported, the triplet energies (65.5 kcal/mol for TX and 62.5 kcal/mol for maleimide)[15] [16] suggested that this transformation should be possible. Pleasingly, we were able to achieve appreciable conversions when reacting maleimide with propargyl alcohol in the presence of TX (Table [1], entry 1). However, initial reactions with one 10 W LED produced residence times (40–50 min) that were not in line with our expectation for a small-scale laboratory set-up. We addressed this by increasing the number of LEDs to eight, giving a maximum power output of 80 W, although typically 53 W was applied. This decreased residence times to 1–3 min, enabling rapid production and isolation of the desired product (Table [1], entry 1). In contrast, the batch reaction required >12 h to achieve comparable conversions. However, productivity is unlikely to significantly differ, according to previous reports by Elliott et al.[4] The moderate isolated reaction yield was also in line with previous reports for similar substrates and can be explained by competing formation of the maleimide dimer.[9]

With the optimized reaction conditions in hand, we explored the scope of the reaction conditions. A range of cyclic and linear alkenes were well tolerated in combination with N-methyl maleimide, which was used in further experiments due to its increased solubility (Table [1], entries 2–6).[17] We subsequently switched the enone to citraconic anhydride, which also delivered products in moderate yields, although ring opening was observed upon purification by column chromatography (entries 7 and 8). Despite this, the suitability of the reactor set-up for rapid production of [2+2] cycloadducts in flow was demonstrated.

Surprisingly, the reaction of citraconic anhydride with 2,3-dimethylbut-2-ene produced a complex mixture of products, the main one of which was not the expected cyclobutane 10, but a highly substituted dihydrooxepin-2(3H)-one 11 (Scheme [1]). This was the only isolable product from a complex mixture; both when the reaction was carried out in flow and batch, resulting in isolated yields of 5% and 15% yield after chromatography, respectively. The structure was confirmed by NMR and IR analysis (see the Supporting Information for full data). A tentative reaction mechanism is depicted in Scheme [1], where diradical intermediate 12 undergoes a homolytic bond cleavage to produce the carboxylate diradical 13. Recombination of radicals produces ketene 14, which may undergo a photochemical [1,3]-sigmatropic rearrangement to deliver 15.[18] [19] This reacts with water upon purification to produce the ketal 10. Interestingly, alternative reaction pathways to the more classical [2+2] cycloaddition, such as the photoene reaction, have recently been reported for citraconicimides.[20] Furthermore, although extensive ring rearrangements through competing pathways are also known,[21] the preparation of the dihydrooxepin-2(3H)-one core by photochemical or indeed any other methods has not been reported.

In conclusion, a microfluidic photochemical reactor with characteristic dimensions of 200 × 800 μm (H/W) and an internal volume of 489 μL was designed and developed using silicon wet-etching techniques. A UV LED light source was developed for triplet-sensitized photochemical cycloadditions and housed in a repurposed TLC UV box for operational simplicity. The system was optimized using the model [2+2] cycloaddition of maleimide with propargyl alcohol and the required residence time for full conversion could be reduced from 73 to 3 min. To examine the scope of the [2+2] cycloadditions, a small N-methyl maleimide and citraconic anhydride based compound set of cycloaddition adducts with carbocyclic, heterocyclic, and linear unsaturated coupling partners were synthesized in moderate yields. Most intriguingly, a novel dihydrooxepin-2(3H)-one scaffold was discovered, accessible through the reaction of citraconic anhydride and tetramethylethene, adding to the repertoire of complexity-generating photochemical transformations.

Conflict of Interest

The authors declare no conflict of interest.

Acknowledgment

We acknowledge DTU DanChip for access to clean-room facilities.

• References and Notes

• 1 Poplata S, Tröster A, Zou YQ, Bach T. Chem. Rev. 2016; 116: 9748
  CrossrefPubMedSearch in Google Scholar
• 2 Elliott LD, Kayal S, George MW, Booker-Milburn K. J. Am. Chem. Soc. 2020; 142: 14947
  CrossrefPubMedSearch in Google Scholar
• 3 Zheng J, Swords WB, Jung H, Skubi KL, Kidd JB, Meyer GJ, Baik MH, Yoon TP. J. Am. Chem. Soc. 2019; 141: 13625
  CrossrefPubMedSearch in Google Scholar
• 4 Elliott LD, Knowles JP, Koovits PJ, Maskill KG, Ralph MJ, Lejeune G, Edwards LJ, Robinson RI, Clemens IR, Cox B, Pascoe DD, Koch G, Eberle M, Berry MB, Booker-Milburn KI. Chem. Eur. J. 2014; 20: 15226
  CrossrefPubMedSearch in Google Scholar
• 5 Cambié D, Bottecchia C, Straathof NJ. W, Hessel V, Noël T. Chem. Rev. 2016; 116: 10276
  CrossrefPubMedSearch in Google Scholar
• 6 Elliott LD, Berry M, Harji B, Klauber D, Leonard J, Booker-Milburn KI. Org. Process Res. Dev. 2016; 20: 1806
  CrossrefSearch in Google Scholar
• 7 Pimparkar K, Yen B, Goodell J, Martin V, Lee WH, Porco J, Beeler A, Jensen K. J. Flow Chem. 2011; 1: 53
  CrossrefSearch in Google Scholar
• 8 Kreis LM, Krautwald S, Pfeiffer N, Martin RE, Carreira EM. Org. Lett. 2013; 15: 1634
  CrossrefPubMedSearch in Google Scholar
• 9 Skalenko YA, Druzhenko TV, Denisenko AV, Samoilenko MV, Dacenko OP, Trofymchuk SA, Grygorenko OO, Tolmachev AA, Mykhailiuk PK. J. Org. Chem. 2018; 83: 6275
  CrossrefPubMedSearch in Google Scholar
• 10 Ahuja S, Jockusch S, Ugrinov A, Sivaguru J. Eur. J. Org. Chem. 2020; 1478
  Crossref
• 11 Demchuk OP, Hryshchuk OV, Vashchenko BV, Kozytskiy AV, Tymtsunik AV, Komarov IV, Grygorenko OO. J. Org. Chem. 2020; 85: 5927
  CrossrefPubMedSearch in Google Scholar
• 12 He J, Liu Q. Synthesis 2022; 54: 925
  Thieme ConnectSearch in Google Scholar
• 13 Nettekoven M, Püllmann B, Martin RE, Wechsler D. Tetrahedron Lett. 2012; 53: 1363
  CrossrefSearch in Google Scholar
• 14 Lin Y, Kouznetsova TB, Craig SL. J. Am. Chem. Soc. 2020; 142: 2105
  CrossrefPubMedSearch in Google Scholar
• 15 Williams JD, Nakano M, Gérardy R, Rincón JA, De Frutos Ó, Mateos C, Monbaliu JC. M, Kappe CO. Org. Process Res. Dev. 2019; 23: 78
  CrossrefSearch in Google Scholar
• 16 Dalton C, Montgomery FC. J. Am. Chem. Soc. 1974; 2: 6230
  CrossrefSearch in Google Scholar
• 17 General Procedure: [2+2] Cycloadditions with N-Methyl Maleimide A standard solution of thioxanthone (0.1 equiv.) and N-Me maleimide (1 equiv.) dissolved in MeCN at 0.1 M was prepared. To 20 mL of the standard solution an alkene (1.5 equiv.) or alkyne (1.5 equiv.) was added, and the solution degassed for 30 min under nitrogen. The solution was transferred to a syringe and pumped through the chip at 6 mL/h under 375 nm irradiance at 35 °C using a Model 11 syringe pump from Harvard Apparatus. 2 mL of the solution was pumped through the chip to ensure steady-state conditions. A 1 mL aliquot was sampled for analysis of the crude reaction mixture by NMR and TLC. The reaction was sampled for 2.75 h followed by concentration under reduced pressure to yield a residue that was further purified by flash chromatography. (1R*,5S*)-1,3,5,6,6,7,7-Heptamethyl-3-azabicyclo [3.2.0]heptane-2,4-dione (6) Transparent oil (143.3 mg, 53% yield). ¹H NMR (400 MHz, DMSO-d ₆): δ = 2.88 (s, 2 H), 2.83 (s, 3 H), 1.15 (s, 6 H), 0.87 (s, 6 H). ¹³C NMR (101 MHz, DMSO-d ₆): δ = 178.3, 46.8, 40.4, 26.6, 24.5, 21.6. HRMS-ESI [M+H]⁺: m/z calcd for C₁₁H₁₈NO₂ ⁺: 196.1332; found: 196.1331 [M+H]⁺ (Δ ppm = –0.51). ATR-FTIR: 2960, 2888, 2873, 1763, 1686, 1482, 1453, 1428, 1372 cm^–1.
• 18 Singh V, Porinchu M. J. Chem. Soc., Chem. Commun. 1993; 134
  Crossref
• 19 Wilsey S, Bearpark MJ, Bernardi F, Olivucci M, Robb MA. J. Am. Chem. Soc. 1996; 118: 176
  CrossrefSearch in Google Scholar
• 20 Ahuja S, Raghunathan R, Kumarasamy E, Jockusch S, Sivaguru J. J. Am. Chem. Soc. 2018; 140: 13185
  CrossrefPubMedSearch in Google Scholar
• 21 Kollenz G, Terpetschnig E, Sterk H, Peters K, Peters EM. Tetrahedron 1999; 55: 2973
  CrossrefSearch in Google Scholar

Corresponding Author

Publication History

Received: 28 October 2021

Accepted after revision: 14 February 2022

Accepted Manuscript online:
14 February 2022

Article published online:
14 March 2022

© 2022. Thieme. All rights reserved

Georg Thieme Verlag KG
Rüdigerstraße 14, 70469 Stuttgart, Germany

• References and Notes

• 1 Poplata S, Tröster A, Zou YQ, Bach T. Chem. Rev. 2016; 116: 9748
  CrossrefPubMedSearch in Google Scholar
• 2 Elliott LD, Kayal S, George MW, Booker-Milburn K. J. Am. Chem. Soc. 2020; 142: 14947
  CrossrefPubMedSearch in Google Scholar
• 3 Zheng J, Swords WB, Jung H, Skubi KL, Kidd JB, Meyer GJ, Baik MH, Yoon TP. J. Am. Chem. Soc. 2019; 141: 13625
  CrossrefPubMedSearch in Google Scholar
• 4 Elliott LD, Knowles JP, Koovits PJ, Maskill KG, Ralph MJ, Lejeune G, Edwards LJ, Robinson RI, Clemens IR, Cox B, Pascoe DD, Koch G, Eberle M, Berry MB, Booker-Milburn KI. Chem. Eur. J. 2014; 20: 15226
  CrossrefPubMedSearch in Google Scholar
• 5 Cambié D, Bottecchia C, Straathof NJ. W, Hessel V, Noël T. Chem. Rev. 2016; 116: 10276
  CrossrefPubMedSearch in Google Scholar
• 6 Elliott LD, Berry M, Harji B, Klauber D, Leonard J, Booker-Milburn KI. Org. Process Res. Dev. 2016; 20: 1806
  CrossrefSearch in Google Scholar
• 7 Pimparkar K, Yen B, Goodell J, Martin V, Lee WH, Porco J, Beeler A, Jensen K. J. Flow Chem. 2011; 1: 53
  CrossrefSearch in Google Scholar
• 8 Kreis LM, Krautwald S, Pfeiffer N, Martin RE, Carreira EM. Org. Lett. 2013; 15: 1634
  CrossrefPubMedSearch in Google Scholar
• 9 Skalenko YA, Druzhenko TV, Denisenko AV, Samoilenko MV, Dacenko OP, Trofymchuk SA, Grygorenko OO, Tolmachev AA, Mykhailiuk PK. J. Org. Chem. 2018; 83: 6275
  CrossrefPubMedSearch in Google Scholar
• 10 Ahuja S, Jockusch S, Ugrinov A, Sivaguru J. Eur. J. Org. Chem. 2020; 1478
  Crossref
• 11 Demchuk OP, Hryshchuk OV, Vashchenko BV, Kozytskiy AV, Tymtsunik AV, Komarov IV, Grygorenko OO. J. Org. Chem. 2020; 85: 5927
  CrossrefPubMedSearch in Google Scholar
• 12 He J, Liu Q. Synthesis 2022; 54: 925
  Thieme ConnectSearch in Google Scholar
• 13 Nettekoven M, Püllmann B, Martin RE, Wechsler D. Tetrahedron Lett. 2012; 53: 1363
  CrossrefSearch in Google Scholar
• 14 Lin Y, Kouznetsova TB, Craig SL. J. Am. Chem. Soc. 2020; 142: 2105
  CrossrefPubMedSearch in Google Scholar
• 15 Williams JD, Nakano M, Gérardy R, Rincón JA, De Frutos Ó, Mateos C, Monbaliu JC. M, Kappe CO. Org. Process Res. Dev. 2019; 23: 78
  CrossrefSearch in Google Scholar
• 16 Dalton C, Montgomery FC. J. Am. Chem. Soc. 1974; 2: 6230
  CrossrefSearch in Google Scholar
• 17 General Procedure: [2+2] Cycloadditions with N-Methyl Maleimide A standard solution of thioxanthone (0.1 equiv.) and N-Me maleimide (1 equiv.) dissolved in MeCN at 0.1 M was prepared. To 20 mL of the standard solution an alkene (1.5 equiv.) or alkyne (1.5 equiv.) was added, and the solution degassed for 30 min under nitrogen. The solution was transferred to a syringe and pumped through the chip at 6 mL/h under 375 nm irradiance at 35 °C using a Model 11 syringe pump from Harvard Apparatus. 2 mL of the solution was pumped through the chip to ensure steady-state conditions. A 1 mL aliquot was sampled for analysis of the crude reaction mixture by NMR and TLC. The reaction was sampled for 2.75 h followed by concentration under reduced pressure to yield a residue that was further purified by flash chromatography. (1R*,5S*)-1,3,5,6,6,7,7-Heptamethyl-3-azabicyclo [3.2.0]heptane-2,4-dione (6) Transparent oil (143.3 mg, 53% yield). ¹H NMR (400 MHz, DMSO-d ₆): δ = 2.88 (s, 2 H), 2.83 (s, 3 H), 1.15 (s, 6 H), 0.87 (s, 6 H). ¹³C NMR (101 MHz, DMSO-d ₆): δ = 178.3, 46.8, 40.4, 26.6, 24.5, 21.6. HRMS-ESI [M+H]⁺: m/z calcd for C₁₁H₁₈NO₂ ⁺: 196.1332; found: 196.1331 [M+H]⁺ (Δ ppm = –0.51). ATR-FTIR: 2960, 2888, 2873, 1763, 1686, 1482, 1453, 1428, 1372 cm^–1.
• 18 Singh V, Porinchu M. J. Chem. Soc., Chem. Commun. 1993; 134
  Crossref
• 19 Wilsey S, Bearpark MJ, Bernardi F, Olivucci M, Robb MA. J. Am. Chem. Soc. 1996; 118: 176
  CrossrefSearch in Google Scholar
• 20 Ahuja S, Raghunathan R, Kumarasamy E, Jockusch S, Sivaguru J. J. Am. Chem. Soc. 2018; 140: 13185
  CrossrefPubMedSearch in Google Scholar
• 21 Kollenz G, Terpetschnig E, Sterk H, Peters K, Peters EM. Tetrahedron 1999; 55: 2973
  CrossrefSearch in Google Scholar

• Supplementary Material