The regioselective addition of water to alkynes presents a straightforward and atom-economic method to obtain versatile carbonyl compounds [1–3]. Since the seminal work on Hg(Ⅱ)-catalyzed hydration of alkynes by Kutscheroff [4–6], many efforts have been devoted to developing transition metals, especially Au [7–9], Ag [10, 11], and Co [12, 13], as efficient catalysts that promote this reaction. Among the hydration of multifarious alkynes, propargylic alcohols with hydroxyl groups in alkynes are generally not appropriate substrates, except for Hg catalysis, presumably owing to their low activity and/or the formation of byproducts resulting from the Meyer-Schuster or Rupe rearrangements (Scheme 1a) [14–16]. In fact, α-hydroxy ketones obtained from the hydration of propargylic alcohols are important structures because of their existence in various drugs and natural products [17, 18]. Therefore, developing a facile and efficient method for preparing α-hydroxy ketones is of great significance and highly desirable in fine chemical synthesis.

α-Alkylidene cyclic carbonates are a kind of functionalized cyclic carbonates that are widely used as versatile intermediates in organic synthesis [19]. Derivatization of α-alkylidene cyclic carbonates is achievable by utilizing their ester groups and/or exocyclic C=C bonds. For example, water as a nucleophile reacts with α-alkylidene cyclic carbonate to produce α-hydroxy ketone through the ring-opening of α-alkylidene cyclic carbonate followed by decarboxylation and tautomerization [20, 21]. In view of the fact that α-alkylidene cyclic carbonates can be easily and efficiently prepared from the commercially available propargylic alcohols and CO₂ [22–30], the development of efficient methods to obtain α-hydroxy ketones from the hydration of propargylic alcohols in the presence of CO₂ will be attractive [20, 21]. As shown in Scheme 1b, propargylic alcohol is first converted into α-alkylidene cyclic carbonate by fixation of CO₂, and α-hydroxy ketone is formed through the addition of water to α-alkylidene cyclic carbonate, which is accompanied by the regeneration of CO₂. In the whole process, CO₂ is not consumed, but acts as a cocatalyst [31, 32].

Recently, the hydration of propargylic alcohols with CO₂ as a cocatalyst has been investigated by using Ag or ionic liquids (ILs) as catalysts (Scheme 2a) [20, 21]. In this regard, silver acetate is found to catalyze this reaction in the presence of 2 MPa CO₂ [20]. This protocol shows good applicability to pyridinyl-substituted internal propargylic alcohols with high reactivity. Subsequently, an excess amount of the IL, [Bu₄P][Im], results in good performance of the various terminal and phenyl-substituted propargylic alcohols [21], even though pressurized CO₂ (1 or 2 MPa) is required to carry out the reaction efficiently in most cases.

Herein, efficient and inexpensive Cu catalysis involving CO₂ as a cocatalyst for the hydration of propargylic alcohols is reported for the preparation of α-hydroxy ketones (Scheme 2b). The high efficiency of Cu catalysts ensures that these reactions proceed smoothly under atmospheric CO₂ pressure even in a gram-scale experiment. This protocol is compatible with various terminal propargylic alcohols, affording the corresponding α-hydroxy ketones in high yields.

CO₂ of purity 99.99% was supplied by Liquefied Air (Tianjin) Co., Ltd. All reagents were purchased from Aladdin or J & K Scientific Ltd. and used as-received. The solvents were dried before use. NMR spectra were recorded on a Bruker 400 MHz spectrometer operating at 400 MHz (¹H) or 101 MHz (¹³C) with CDCl₃ or DMSO-d₆ as the solvent. The chemical shifts were reported relative to CDCl₃ (7.26 ppm) (or DMSO-d₆, 2.50 ppm) for ¹H NMR and CDCl₃ (77.0 ppm) (or DMSO-d₆, 39.52 ppm) for ¹³C NMR. GC-MS spectra were measured at an ionization voltage of 70 eV on a Shimadzu GCMS-QP2010 equipped with a RTX-5MS capillary column.

In a 15 mL Schlenk tube, Cu₂O (0.2 mmol, 28.6 mg), cyclohexyldiphenylphosphine (L1, 0.2 mmol, 53.6 mg), 1, 8-diazabicyclo[5.4.0]undec-7-ene (DBU) (0.5 mmol, 76.1 mg), propargylic alcohol (1 mmol, 84.1 mg), water, and CH₃CN (1 mL) were added sequentially. Then, the Schlenk tube was sealed and attached to a CO₂ balloon (about 1 L capacity). Finally, the reaction mixture was stirred at 60 ℃ for 24 h. After completion of the experiment, the CO₂ balloon was taken off and the reaction system cooled to room temperature. Then, 1, 3, 5-trimethoxybenzene (50 mg) as an internal standard was added for ¹H NMR analysis to determine the yields of the products. The crude reaction mixture was purified by column chromatography on silica gel (200–300 mesh) with n-hexane/ethyl acetate (20:1–5:1) as the eluent to yield the desired α-hydroxy ketones, which were further characterized by ¹H and ¹³C NMR as well as GC-MS or HRMS (ESI).

In a 15 mL Schlenk tube, Cu₂O (0.2 mmol, 28.6 mg), cyclohexyldiphenylphosphine (L1, 0.2 mmol, 53.6 mg), DBU (0.5 mmol, 76.1 mg), 2-methylbut-3-yn-2-ol (1a, 1 mmol, 84.1 mg), and CH₃CN (1 mL) were added sequentially. Then, the Schlenk tube was sealed and attached to a CO₂ balloon (about 1 L capacity). Finally, the reaction mixture was stirred at room temperature. After 24 h, DBU (0.5 mmol, 76.1 mg) and water (1 mmol, 18.0 mg) were added, and the resulting solution was stirred at 60 ℃ for another 24 h. Finally, the reaction system was cooled to room temperature. Then, biphenyl (20 mg) as an internal standard was added to determine the yield of 3-hydroxy-3-methylbutan-2-one (2a) by GC analysis.

3-Hydroxy-3-methylbutan-2-one (2a). Colorless oil. ¹H NMR (400 MHz, DMSO-d₆) δ 5.23 (s, 1H), 2.15 (s, 3H), 1.17 (s, 6H) ppm. ¹³C NMR (101 MHz, DMSO-d₆) δ 213.83, 75.74, 26.18, 24.23 ppm. GC-MS (EI, 70 eV) m/z (%) 87 (100), 60 (92), 69 (62), 102 (9).

3-Hydroxy-3-methylpentan-2-one (2b). Yellow oil. ¹H NMR (400 MHz, DMSO-d₆) δ 5.04 (s, 1H), 2.13 (s, 3H), 1.60 (dq, J = 14.9, 7.5 Hz, 1H), 1.46 (dq, J = 14.8, 7.5 Hz, 1H), 1.12 (s, 3H), 0.74 (t, J = 7.5 Hz, 3H) ppm. ¹³C NMR (101 MHz, DMSO-d₆) δ 214.19, 78.55, 31.77, 25.10, 24.14, 7.88 ppm. GC-MS (EI, 70 eV) m/z (%) 43 (100), 73 (25), 87 (21), 45 (19), 57 (12), 101 (10).

3-Hydroxy-3-methylnonan-2-one (2c). Yellow oil. ¹H NMR (400 MHz, CDCl₃) δ 3.88 (s, 1H), 2.13 (s, 3H), 1.69–1.51 (m, 2H), 1.38–1.30 (m, 1H), 1.26 (s, 3H), 1.20 (dd, J = 15.9, 5.9 Hz, 6H), 0.94 (tt, J = 15.9, 8.0 Hz, 1H), 0.79 (t, J = 6.8 Hz, 3H) ppm. ¹³C NMR (101 MHz, CDCl₃) δ 212.04, 78.43, 39.09, 31.24, 29.08, 24.93, 23.30, 22.92, 22.12, 13.57 ppm. GC-MS (EI, 70 eV) m/z (%) 69 (100), 129 (94), 59 (89), 43 (82), 55 (36), 41 (34), 84 (20), 45 (19), 111 (18), 28 (16), 85 (14).

3-Hydroxy-3, 5-dimethylhexan-2-one (2d). Yellow oil. ¹H NMR (400 MHz, CDCl₃) δ 3.83 (s, 1H), 2.18 (s, 3H), 1.69–1.57 (m, 3H), 1.29 (s, 3H), 0.89 (d, J = 6.4 Hz, 3H), 0.79 (t, J = 6.1 Hz, 3H) ppm. ¹³C NMR (101 MHz, CDCl₃) δ 212.71, 78.96, 47.71, 26.56, 24.19, 23.76, 23.41 ppm. GC-MS (EI, 70 eV) m/z (%) 81 (100), 99 (96), 43 (29), 55 (21), 79 (19), 57 (10).

1-(1-Hydroxycyclohexyl)ethan-1-one (2e). Yellow oil. ¹H NMR (400 MHz, CDCl₃) δ 2.24 (s, 3H), 1.75–1.64 (m, 6H), 1.49 (d, J = 6.5 Hz, 2H), 1.28 (dd, J = 15.1, 10.3 Hz, 2H) ppm. ¹³C NMR (101 MHz, CDCl₃) δ 212.74, 77.97, 33.82, 25.28, 23.69, 21.07 ppm. GC-MS (EI, 70 eV) m/z (%) 81 (100), 99 (71), 79 (21).

3-Hydroxy-3-phenylbutan-2-one (2f). Brown oil. ¹H NMR (400 MHz, CDCl₃) δ 7.48–7.42 (m, 2H), 7.38 (t, J = 7.6 Hz, 2H), 7.32 (dd, J = 10.5, 3.8 Hz, 1H), 2.09 (s, 3H), 1.79 (s, 3H) ppm. ¹³C NMR (101 MHz, CDCl₃) δ 209.62, 141.37, 128.68, 128.05, 125.96, 79.84, 23.99, 23.43 ppm. GC-MS (EI, 70 eV) m/z (%) 121 (100), 77 (31), 105 (19).

3-Hydroxy-3-methylpent-4-en-2-one (2g). Yellow oil. ¹H NMR (400 MHz, CDCl₃) δ 5.91 (dd, J = 17.1, 10.5 Hz, 1H), 5.44 (dd, J = 17.1, 0.9 Hz, 1H), 5.22 (dd, J = 10.6, 0.8 Hz, 1H), 3.93 (s, 1H), 2.20 (s, 3H), 1.43 (s, 3H) ppm. ¹³C NMR (101 MHz, CDCl₃) δ 209.38, 138.76, 116.23, 79.31, 24.25, 23.60 ppm. GC-MS (EI, 70 eV) m/z (%) 55 (100), 83 (83), 43 (68), 99 (37), 101 (36), 41 (31), 45 (30), 71 (24), 29 (15), 57 (13), 100 (10), 74 (10), 27 (10).

17-Hydroxy-17βHpregn-4-ene-3, 20-dione (2h). White solid, m.p. 192–193 ℃. ¹H NMR (400 MHz, CDCl₃) δ 5.72 (s, 1H), 2.92 (d, J = 30.2 Hz, 1H), 2.38–2.23 (m, 8H), 2.03–1.95 (m, 1H), 1.91–1.84 (m, 1H), 1.72–1.40 (m, 10H), 1.18 (s, 3H), 1.10–1.04 (m, 1H), 0.98 (s, 3H), 0.92 (dd, J = 9.8, 5.9 Hz, 1H) ppm. ¹³C NMR (101 MHz, CDCl₃) δ 214.22, 199.43, 171.01, 123.82, 90.65, 53.18, 49.06, 38.47, 36.08, 35.55, 34.95, 33.81, 32.95, 32.68, 31.49, 28.19, 24.18, 20.66, 17.27, 14.07 ppm. HRMS (ESI): C₂₁H₃₁O₃ for [M + H]⁺ calcd 331.2268, found 331.2274.

The reactions were initially investigated with 2-methylbut-3-yn-2-ol (1a) as the model substrate that was reacted with water under atmospheric CO₂ pressure to obtain 3-hydroxy-3-methylbutan-2-one (2a), as described in Table 1. Obviously, 1a could not be effectively converted into 2a without a catalyst or when DBU was used alone (entries 1 and 2). When employing AgOAc, which was the optimal catalyst in previous work [20], to promote the hydration, the yield of 2a was only 10%, with the remaining being mostly 1a (entry 3), indicating the low efficiency of Ag catalysis under ambient CO₂ pressure. Mechanically, a tandem process involving the cyclization of propargylic alcohol with CO₂ and the following addition of water to the in-situ generated α-alkylidene cyclic carbonate is involved in the hydration of propargylic alcohol with CO₂ as a cocatalyst to prepare α-hydroxy ketone [20, 21]. In particular, efficient cyclization is crucial to the whole process. Therefore, several Cu compounds were screened for converting 1a to 2a, owing to their excellent performances in the rate-determining cyclization [33–35] (entries 4–10). Excitingly, CuCl as the catalyst allowed the reaction to proceed smoothly, affording 31% of 2a (entry 4). By changing the Cu source to CuI, the yield of 2a was increased to 46% (entry 6). Moreover, a significant increase in the yield of 2a was observed when employing Cu₂O as the catalyst (entry 7). Comparatively, Cu(Ⅱ) showed a lower catalytic activity than Cu(Ⅰ) (entries 7–9). The poor activity of CuOAc should be ascribed to its moisture sensitivity (entry 10). Notably, an increase in the amount of water could obviously decrease the rate of hydration (entries 11 and 12), and no 2a was observed when five equivalents of water was supplied (entry 12).

The effect of other reaction parameters on the reaction outcome was further investigated to enhance the yield of 2a (Table S1, in the Supporting Information). Obviously, a temperature lower than 60 ℃ is deleterious to the hydration and therefore to the yield of 2a, which decreased sharply when the hydration was carried out at 40 ℃; this can probably be ascribed to the inefficiency of cyclization of 1a with CO₂. In addition, after increasing the reaction temperature to 80 ℃, a slight reduction in the yield of 2a was detected. Therefore, the optimal reaction temperature was determined as 60 ℃. The effect of solvents with different polarities was also studied, and the yield of 2a decreased in the order of CH₃CN > DMSO > 1, 4-dioxane. Therefore, solvents with large polarities favor the hydration.

Then, phosphines as additives were examined (Table 1, entries 13–17). To our delight, cyclohexyldiphenylphosphine (L1), with moderate electron-donating capability and proper steric resistance, could largely improve the yield of 2a to 97% (entry 14). Relatively, monophosphines such as PPh₃ and tricyclohexylphosphine (L2) and bisphosphines including 1, 2-bis(diphenylphosphino)ethane (L3) and 1, 4-bis(diphenylphosphino)butane (L4) showed poorer performances than L1 (entries 13, 15-17 vs. entry 14).

The activity of Cu catalysis was subsequently tested with a series of propargylic alcohols, as described in Table 2. In general, this protocol worked well for the hydration of various propargylic alcohols with 70%–97% yields of the corresponding α-hydroxy ketones 2a–g. This system exhibited good activities for substrates with different lengths of substituted alkyl chains such as methyl, ethyl, and hexyl (entries 1–3). However, a slightly reduced yield was observed for 1d (entry 4), which might attributed to the increased steric hindrance that leads to unfavorable activation of the hydroxyl group and CO₂ insertion. In addition, the cyclohexyl-substituted propargylic alcohol 1e is also a suitable substrate, affording 2e in 87% yield (entry 5). Compared with the alkyl chains, substituents with unsaturated bonds such as phenyl and vinyl result in poorer reactivities of propargylic alcohols (entries 6 and 7), indicating that the interaction of unsaturated bonds with Cu may hinder the activation of the C≡C bond and the intramolecular cyclization of the key alkylcarbonate intermediates.

To broaden the application of this catalysis protocol, a larges-cale experiment was performed, as shown in Scheme 3a. By increasing the amount of 1a to the gram scale, 70% yield of 2a was successfully obtained, indicating the scalability of this strategy. Moreover, this method can also be applied to the derivatization of oral progesterone ethisterone [36] (Scheme 3b), highlighting its potential and wide applicability.

In order to clarify the reaction pathway, the role of CO₂ was investigated first (Scheme 4a). The hydration of 1a cannot take place without CO₂, demonstrating the crucial role of CO₂ in this reaction. The tandem process including the carboxylative cyclization and the subsequent reaction of water with α-alkylidene cyclic carbonate involving Cu catalysis was further demonstrated by a two-step reaction (Scheme 4b). Obviously, 66% of 3a could be obtained from 1a and CO₂ through Cu catalysis, which could then be smoothly converted into 2a, which also accounts for high efficiency of Cu catalysis under atmospheric CO₂ pressure.

Insights into the possible interaction between DBU and propargylic alcohol were obtained by using ¹H NMR technique. As shown in Fig. 1, the O–H signal of 1a became broad and showed a downfield shift from 5.32 to 5.36 ppm, indicating the activation of propargylic alcohol by DBU through the formation of hydrogen bonding.

A possible mechanism was proposed, as depicted in Scheme 5, based on our investigation and previous work [20, 21, 33, 37]. Initially, the activated propargylic alcohol in the form of intermediate Ⅰ is generated in the presence of Cu(Ⅰ) catalyst and DBU. After CO₂ insertion, the key intermediate Ⅱ is obtained. Then, intramolecular cyclization of intermediate Ⅱ yields the vinylcopper intermediate Ⅲ. Followed by proto-demetallation, the carboxylative cyclization of propargylic alcohol with CO₂ (Cycle Ⅰ) is completed, and α-alkylidene cyclic carbonate is afforded. With the aid of DBU, water as the nucleophile reacts with the carbonyl of α-alkylidene cyclic carbonate to form the intermediate Ⅳ by ring-opening. Finally, α-hydroxy ketone is produced after the decarboxylation and tautomerization of intermediate Ⅳ (Cycle Ⅱ). In addition, the interaction between CO₂ and DBU in the entire process cannot be excluded.

Cu catalysis for the hydration of propargylic alcohols with CO₂ as a cocatalyst to synthesize α-hydroxy ketones was achieved. Notably, the reaction can be carried out successfully under atmospheric CO₂ pressure [38, 39]. This methodology is highly efficient, showing good performance toward the synthesis of a series of α-hydroxy ketones in good to excellent yields. In addition, this method is scalable and can be applied for the derivatization of complicated bioactive molecules, indicating its practicability and versatility for the preparation of α-hydroxy ketone with CO₂ as a cocatalyst.