Introduction

The upgrade of the greenhouse gas carbon dioxide (CO2) into valuable products represents an attractive objective in the realm of modern sustainable, catalytic, and synthetic chemistry. Most research around the use of CO2 in catalytic conversions focuses on two related but distinct objectives: improvement of existing processes, or the discovery of transformations that help to further valorize this carbon feedstock using a “bottom-up” approach mostly involving downstream processing1,2,3,4,5,6,7. The synthesis of heterocyclic compounds known as cyclic carbonates/carbamates has been among the more prevalent activities in this area and widely investigated over the past decade8,9,10,11,12,13,14,15,16,17. A series of applications for these CO2-based heterocycles have been developed including their use as non-protic media18,19,20, precursors to fine chemicals21,22, and as polymerizable monomers23,24,25,26,27. In addition, functionalized cyclic carbonates can also serve as building blocks in transition metal-catalyzed decarboxylative stereoselective cyclizations, allylic and propargylic chemistry21,28,29. While efficient routes towards both five- and six-membered cyclic carbonates (Fig. 1a; 5MCCs and 6MCCs) have been described through highly efficient pathways6,7,8,10,11,13,24,25,26,27, a major limitation within the area is the easy access to larger ring cyclic carbonates such as 7MCCs. As far as we are aware, there have been only sporadic reports on the preparation of elusive, thermodynamically disfavored seven-membered carbonates albeit via stoichiometric approaches in the context of polymer development30,31,32,33.

Fig. 1: State of the art and current challenges.
figure 1

a Current state for the catalytic (cat.) synthesis of five- (5MCCs), six- (6MCCs) and seven-membered cyclic carbonates (7MCCs) from CO2. b Selected bioactive compounds with a single (left) and bicyclic oxazolidinone core (right). c Catalytic strategy towards 7MCCs and their application towards bicyclic pharmacores. Cat stands for catalyst.

Full size image

Therefore, the development of effective strategies that allow for the straightforward preparation and isolation of such medium-sized heterocycles from CO2 creates incentives thereby shifting the current limitations in the catalytic conversion of this carbon feedstock. Furthermore, the development of synthetic concepts coupled with the potential to transform larger-ring carbonates into pharmaceutically interesting scaffolds will open up additional avenues for CO2 valorization. The use of CO2 toward the preparation of various intermediates having potential as pharma-focused synthons has caught increasing attention over the years34,35,36,37,38,39,40,41,42,43. In this realm, the preparation of structurally simple oxazolidinones and related compounds using CO2 has been reported by various groups, though overall with limited structural scope12,14,15,17. More advanced bicyclic scaffolds, as shown in Fig. 1b remain inaccessible by any known method despite their inherent potential as bioactive compounds. Here we present an approach that overcomes these current limitations and represents a catalytic process for a wider series of seven-membered cyclic carbonates from 1,4-alkyne-diols through a successive Ag/base-promoted CO2 coupling/ring-expansion sequence (Fig. 1c).

The utilization of this stepwise strategy avoids the generation of thermodynamically more stable five- or six-membered byproducts. The seven-membered cyclic carbonates exhibit unique reactivity illustrated in a domino process that propels their structural rearrangement into bicyclic oxazolidinones (Fig. 1c). The latter compounds have biological relevance but their scope remains tremendously limited in diversity. The developed protocol will thus significantly expand the access and use of larger-ring carbonates (and related carbamates) in drug-related research programs.

Results and discussion

Screening and optimization studies

The initial hypothesis for our approach was based on the recent success attained in the preparation of acyl-functionalized five-membered cyclic carbonates prepared via an Ag-promoted cascade sequence44. This reported one-pot catalytic approach was only useful to extend the pool of five-membered heterocycles using 1,2-alkyne diols and CO2 as substrates but was inadequate toward the easy preparation and isolation of six- and seven-membered analogs from 1,3- and 1,4-alkyne diols, respectively. We envisaged that the intermediacy of various reactive species with multiple functional (OH, C≡C) groups44 would limit our options for a chemo-selective transformation of 1,4-alkyne diols and CO2 towards seven-membered heterocycles. As an alternative strategy (Table 1), we used an O-protected 1,4-alkyne-diol (1a, 1.0 mmol) in the presence of catalytic AgOAc/JohnPhos (both 2 mol%) at 75 °C and 10 bar CO2 pressure in ACN (CH3CN) as medium. This first step (STEP 1) afforded typically, when combined with in situ O-deprotection under acidic conditions, the free-alcohol, five-membered α-alkylidene carbonate 2a in > 95% yield (see Supplementary Fig. 1 in the Supporting Information, SI, for details). The conditions in Table 1, however, specifically refer to STEP 2 under base catalysis, giving rise to the target product 3a, while previously reported 4a was also identified in some cases as a by-product44 underlining the intrinsic, thermodynamic challenge of this protocol.

Table 1 Screening and optimization of the reaction conditions for the conversion of semi-protected 1,4-alkyne diol 1a into targeted seven-membered acyl cyclic carbonate 3aa
Full size table

The use of DIPEA as a base catalyst to transform intermediate carbonate 2a into seven-membered cyclic carbonate 3a was not very productive (12%). Other bases, such as NMM and DMAP provided somewhat better results (entries 2 and 3; 3a up to 46%), and increasing the amount of DMAP (entry 4; 20 mol%) resulted in more productive catalysis (3a, 57%). The strong N-heterocyclic base DBU gave full substrate conversion but with low chemo-selectivity towards 3a (entry 5, < 10%). When DABCO was used, a significant improvement was noted (entries 6–8), with the highest yield for 3a obtained when 20 mol% of this base was present (entry 7; 3a 75%, 73% isolated). Other N-heterocyclic bases including DBN, TBD, hydroquinine (HQ) and quinine (QUI) were all less efficient compared to DABCO (entries 9–12). The solvent effect was then further probed (entries 13–15), showing that the use of DCM (entry 14; 3a 79%, 78% isolated) was slightly more productive. Notably, prolonging the reaction time of the second step from 14 to 24 h decreased the yield of the desired product to 56% (entry 16), indicating some decomposition of the initially formed seven-membered cyclic carbonate 3a. When 1H NMR solutions were followed over time, we found indications for the presence of oligomeric species caused by ring-opening polymerization (ROP) of the 7MCC likely initiated by the base in line with previous reports45,46. These observations also emphasize the lower thermodynamic stability of 3a compared to their highly stable, five-membered analogs (i.e., 5MCCs in Fig. 1a).

It should be emphasized that the formation of the thermodynamically more stable bicyclic derivative 4a (Table 1) and other products could be largely suppressed under the optimized reaction conditions. This shows the value of developing a catalytic procedure towards 7MCC, with the chemoselectivity control being under kinetic (i.e., catalyst) control. As a result, parasitic substrate conversion pathways such as ROP can be minimized or even prevented. The basicity (pKb) of the catalyst plays a crucial role as shown by previous work on the functionalization of alcohol-derived 6MCC47. Finally, efforts to prepare eight-membered analogs of 3a failed probably due to both thermodynamic and kinetic reasons (see Supplementary Table 1 in the SI).

Scope and further optimization of the developed protocol

Next, we further investigated the scope of the two-step, catalytic transformation of various 1,4-alkyne-diols and CO2 into acyl-functionalized seven-membered cyclic carbonates (Fig. 2) using the conditions in entry 14 of Table 1 as a starting point. Both the yields of the five-membered precursors (compounds 2; see the SI for details, note a different synthetic approach for precursor 2q: DavePhos as ligand, 20 bar and at 40 °C) and their seven-membered acyl carbonates (compounds 3) are provided. Variations in both R1 and R2 in the 1,4-alkyne diol 1 were feasible providing both compounds 2 and 3 with a structural diversity, and typically in good to excellent isolated yields. Various linear, branched, cyclic alkyls and functional alkyls (allyl, vinyl), and substituted aryl-substituents (including larger ones such as 2-naphthyl) were tolerated in this protocol. Noteworthy examples include seven-membered acyl carbonates 3p (65%) with an additional vinyl group and 3r (68%, derived from linoleic acid), illustrating that more complex/functional scaffolds are also accessible. Overall, appreciable yields of seven-membered acyl carbonates 3 could be attained (most of them in > 60% yield) using the two-step, catalytic approach, which significantly expands the larger-ring cyclic carbonate chemical space compared to the reported examples in the literature30,31,32,33. Substrates of type 2 with a substituted C=C bond proved to be unproductive towards ring expansion (see the SI for details). The atom-connectivity and identity of the 7MCCs was supported by a combination of diagnostic features, including 13C NMR analysis (δ = 152.8 ppm), IR spectroscopy (ν = 1744 cm−1), and X-ray diffraction (Fig. 2, inset; 3a as exemplary case). As far as we know, crystallographically characterized seven-membered cyclic carbonates are unknown.

Fig. 2: Carbonate diversity.
figure 2

Scope of seven-membered acyl-carbonates 3 using various 1,4-alkyne diols 1 and CO2 as reagents. All products 3 were obtained under the conditions reported in entry 14 of Table 1 using 0.2 mmol of 2. All reported yields are of the column-purified, isolated products. The yields and analytical data for intermediates 2 are reported in the SI. Cat stands for catalyst.

Full size image

In order to further address the efficiency of this catalytic approach towards seven-membered cyclic carbonates, we examined whether it would be feasible to design a one-pot catalytic protocol (Fig. 3). In this approximation, we first considered a different and more rigid substrate design (compounds 5, Fig. 3a) that have both ends of the 1,4-alkyne diol tethered via an aryl fragment at the 2- and 3-position. This alternative substrate design allowed to combine the initial carboxylation of the 1,4-alkyne diol, the subsequent O-deprotection, and isomerization steps to take place without the need to isolate the intermediate α-alkylidene carbonate species. Bicyclic seven-membered acyl carbonates 6a−6c were thus directly isolated from the reaction mixtures in good yields up to 70%. Inspired by these encouraging results, we then also attempted a one-pot approximation for selected examples (3a, 3b, 3 f, 3 g and 3 s; see Fig. 3b) of the scope of seven-membered acyl carbonates reported in Fig. 2, and found that also in these cases a fairly similar isolated yield of the target product could be attained. This clearly shows that one-pot catalytic approaches, if properly designed, can further increase the overall process efficiency.

Fig. 3: One-pot approaches to seven-membered acyl carbonates.
figure 3

a Use of a more rigid substrate design. b Using the original but further optimized approximation with the comparative yield for the two-step approach provided in parentheses. Cat stands for catalyst.

Full size image

Synthetic explorations and domino conversion of 7MCCs

We then set out to design synthetic applications for cyclic carbonates 3 using the intrinsic acyl functionality as a key enabling fragment (Fig. 4a; full experimental details are provided in the SI) taking 3a (scaled up to 2.4 mmol: 79%) and 3 g as exemplary cases. Treatment of 3 g with NaBH4 afforded five-membered cyclic carbonate 7 in 85% yield. This result may seem unexpected, but the reduction of the acyl fragment would initially result in secondary alcohol, which can then induce a Payne-type isomerization based on backbiting at the seven-membered ring and causing a thermodynamically driven isomerization13,48.

Fig. 4: Product diversification and synthetic applications.
figure 4

a Transformations of seven-membered cyclic carbonate 3a and 3 g. b Various strategies to access functional bicyclic oxazolidinones. Cat stands for catalyst.

Full size image

Bicyclic (5 + 6) cyclic carbonate 8 was formed in 23% yield in the presence of catalytic DBU (see Supplementary Fig. 298 in the SI for a mechanistic proposal), but we found that under Cu-catalysis, the yield could be significantly improved to 80%. Such a bicyclic, partially strained carbonate49,50 is potentially useful in the context of ring-opening polymerization providing polycarbonate architectures24,25,26,27.

We further examined the use of hydroxyl-amine reagents as a means to transform the acyl into a ketimine fragment, and treatment of 3a with either the HCl salt of H2NO-allyl or H2N-OBn smoothly provided the ketimine based seven-membered cyclic carbonates 9 (79%) and 10 (71%) in good yields. Contrary to what is observed in these latter two cases, when 3a is treated with the unprotected reagent NH2-OH·HCl, bicyclic oxazolidinone 11 was produced in 80% yield. The envisioned mechanism through which 11 is generated involves the formation of the ketimine group that subsequently acts as an intramolecular nucleophile triggering a domino sequence involving carbonate ring-opening and an oxa-Mannich type addition to forge a bicyclic carbamate as presented in Fig. 4a.

Given the success of the formation of 11a, we then evaluated a wider synthetic scope of such structures by variation of the starting seven-membered acyl carbonate (3) and assessing the potential of transformations at other sites in the scaffold to maximize diversity (Fig. 4b). First, we expanded the hydroxylamine-initiated domino cascade using different seven-membered acyl carbonates 3 allowing to prepare 11a-f in consistent yields of up to 80% (Fig. 4b, top). Both aryl and (functional) alkyl groups are tolerated in this domino sequence with vinyl-based 11 d providing a synthetic handle to further structurally modify the backbone. Compound 11 f (65%, relative stereochemistry determined by GOESY NMR, see the SI) required a longer reaction time (48 h) due to an increased steric impediment in the ring closure leading to the pyran ring. Then, we turned to a different strategy (Fig. 4b, lower part) focusing on the modification of the pendent N-OH fragment in 11a. Protection of the latter with 3,5-di-nitrophenol in the presence of NaH produced carbamate ether 12 (not isolated) followed by a photochemically based reduction of 12 promoted by Eosin B giving the N-unprotected bicyclic oxazolidinone 13a in an overall yield of 51%, which could be subsequently alkylated using BnBr/NaH (13b, 65%). Conjugation of 11a with Indometacin (a nonsteroidal anti-inflammatory drug) proceeds smoothly under standard esterification conditions using DCC/DMAP and furnished 14 in 83%. Formal etherification is also feasible, and a Michael addition of the N-OH bond to a propargylic ester produced 15 in 85% yield. Compounds 16 (85%) and 17 (67%) were obtained using slightly different etherification procedures, but exemplify that N-functionalization/substitution in the bicyclic oxazolidinones is fairly simple.

We then sought to expand the synthetic potential of the formal domino synthesis of the seven-membered heterocycles and designed the corresponding alkyne-1,4-aminoalcohols 18 (Fig. 5a). Using a similar one-pot, catalytic approach as developed for the 7MCCs (Fig. 3b), an easy entry towards the preparation of seven-membered carbamates 20 via intermediate five-membered cyclic carbonates 19 was achieved. Compounds 20a-g were typically isolated in > 70% yield (except for 20 d: 47%) with various substituents on the heterocyclic ring. In addition, attempts were made to extend this protocol to the synthesis of eight-membered cyclic carbamates (20h-j), however, in these cases, only low product yields were attained, marking currently a limitation for the developed protocol. Nonetheless, the combined results point out that other types of products can also be attained by changing the nature of the heteroatoms in the substrates. In order to take advantage of the free –NH group in bicyclic oxazolidinone 13a (Fig. 5b), it was treated with an acyl chloride after activating it by n-BuLi delivering compound 21 (61%), which is a close mimic to oxazolidinone OSL07, a known bioactive modulator51,52. Thus, with the present work, bicyclic analogs of OSL07 are now accessible providing potential to study their bioactivity and pharmaceutical value.

Fig. 5: Extending the protocol to other heterocycles and further uses.
figure 5

a Formation of 7-membered carbamates. b Preparation of a structural mimic of the bioactive compound OSL07. c Intermolecular ring-opening of 9. d Derivatization of vinyl-substituted 3p. Cat stands for catalyst.

Full size image

As opposed to the intramolecular chemistry that is involved in the formation of bicyclic oxazolidinones 11 (Fig. 4a), we carried out ring-opening reactions between 7MCC 9 and various nucleophiles (see Fig. 5c)53. We first focused on the synthesis of nonsymmetrical linear carbonate 22 (47%), which could be prepared from 9 and BnOH using TBD as a catalyst under mild reaction conditions. Replacing BnOH for its sulfur analog (BnSH) under comparable conditions provided smooth access to thiocarbonate 23 in 66% yield. Amine nucleophiles such as morpholine also are effective reagents for the ring-opening of 9 giving easy access to 24 (78%) that was isolated as a mixture of regio-isomers (rr = 2:1)53. As the final endeavor, vinyl-substituted seven-membered cyclic carbonate 3p was subjected to Pd-catalyzed decarboxylative amination21, resulting in the formation of α-β-unsaturated ketone 25 (58%) as a single stereoisomer (E). The vinyl group in 3p could also be used in a formal radical-promoted, decarboxylative difunctionalization process using a TMS-based carbazole as a radical precursor furnishing dicarbazole-substituted 26 (30%)54.

Mechanistic considerations

Finally, we carried out some control experiments (Fig. 6) that demonstrate several features and the unique reactivity of the seven-membered acyl carbonates compared to their five/six-membered analogs. When a 1,3- instead of a 1,4-alkyne diol (Fig. 5a, 27) was used as a substrate in a one-pot approach similar to the one presented in Fig. 3b, we observed the near quantitative formation of intermediate 29. Raising the temperature to 50 °C induced product formation, and in this case, the target and known six-membered acyl carbonate 3044 could be produced. It contained about 16% of its precursor 29, which could not be separated. Nonetheless, this outcome suggests that one-pot approaches to other types of acyl-carbonates are feasible. The thermal stability of 3a was tested at 70 °C (Fig. 5b), showing that after 18 h most of the initial material was still intact. Next, silyl-protected 2a´ was subjected to typical desilyation conditions (Fig. 5c, TBAF) to produce the free-alcohol species and to initiate isomerization. This afforded bicyclic carbonate 8 in 47% yield and established that a completely different chemo-selectivity is attained under these conditions. When ketimine-based, seven-membered carbonate 9 was thermally activated (Fig. 5d) to see if it would isomerize at elevated temperature, we observed a low conversion but no desired bicyclic oxazolidinone 17 was formed. At this stage, the more sterically demanding character of the ketimine fragment is likely responsible for this. A larger ketimine fragment reduces the conformational potential of this heterocycle for productive isomerization akin to what is observed in the synthesis of compounds 11. Finally, we attempted to use six-membered cyclic carbonate 30 (Fig. 5e) as a starting point for the envisioned synthesis of a (5 + 5) bicyclic oxazolidinone, but in this case, we could only observe (by 1H NMR) the in situ formation of ketimine-derived 31, which proved to be unstable under these conditions. This is an important observation and suggests that only the seven-membered acyl carbonates of type 3 possess sufficient reactivity enabling the formation of the bicyclic oxazolidinone pharmacores of type 11.

Fig. 6: Control experiments.
figure 6

a Use of 1,3-alkyne diols furnishing 6 M acyl carbonate 30. b Thermal stability check on seven-membered carbonate 3a. c Utilization of a different O-protecting group in the 5MCC precursor. d Attempted thermal activation of the pendant ketimine to force the domino synthesis of 17. e Attempted synthesis of a (5 + 5) bicyclic oxazolidinone from 31.

Full size image

In summary, we report here a catalytic protocol for the formation of thermodynamically disfavored acyl-functionalized seven-membered cyclic carbonates (note the exemplary formation of thermodynamically stable five-membered carbonate 7 from seven-membered 3 g in Fig. 4a as support for this notion) that can be achieved either via a two-step or one-pot sequence using 1,4-alkyne diols and CO2 as precursors. This work significantly expands the portfolio of larger-ring heterocycles that can be derived from CO2 and offers a versatile entry towards underrepresented bicyclic oxazolidinone pharmacores. Control experiments demonstrate the unique reactivity of the seven- versus five/six-membered cyclic carbonates in the latter context, making them thus privileged synthons for the developed domino process that involves the intermediacy of hydroxyl-ketimine functional groups. We believe that the cascade process with properly designed substrates having built-in pro nucleophilic sites55 holds great future promise for the creation of complex synthons derived from carbon dioxide as feedstock.

Methods

Experimental procedure for the one-pot synthesis of carbonates 3 from substrates 1

In a typical experiment, 0.2 mmol of propargyl alcohol 1, AgOAc (0.7 mg, 2 mol%), JohnPhos (1.2 mg, 2 mol%), and ACN (0.4 mL) were added in a stainless steel reactor. The reactor was purged twice with CO2 (10 bar) and then charged with CO2 (10 bar). The mixture was stirred at 75 °C for 22 h, then cooled with an ice/water bath and carefully depressurized. The solvent was removed in vacuo, and the residue was dissolved in 1 mL of THF/10% HCl (THF:10% HCl = 5:2 v/v). The mixture was stirred at room temperature (r.t.) for 0.5 h then diluted with water and extracted with ethyl acetate. The combined organic layers were dried over Na2SO4, filtered, concentrated under reduced pressure, and dissolved in 0.4 mL of DCM, then DABCO (4.5 mg, 0.04 mmol, 0.2 equiv) was added. After stirring at r.t. for 14 h (note that for 2r, the reaction time was 1.5 h; extending the reaction time will cause the product to decompose), the mixture was transferred to a round-bottom flask, concentrated, and purified by flash column chromatography with ethyl acetate/hexane as eluent. Note: For 1r, AgOAc (0.7 mg, 2 mol%) and DavePhos (1.6 mg, 2 mol%) were used under 20 bar of CO2. The mixture was stirred at 40 °C for 24 h. Full details are provided in the SI.

Experimental procedure for the synthesis of oxazolidinones 11 from carbonates 3

To a stirred solution of hydroxylamine hydrochloride (13.9 mg, 0.2 mmol, 2 equiv), pyridine (15.8 mg, 0.2 mmol, 2 equiv) in ethanol (2 mL) at r.t. was added carbonate 3 (0.1 mmol, 1 equiv). After the substrate was completely consumed (determined by TLC), the solvent was removed under reduced pressure. To the residue was added water and the product was extracted twice with methylene chloride (20 mL). The organic layer was dried over anhydrous Na2SO4, filtered, and concentrated in vacuo. The residue was purified by flash chromatography employing ethyl acetate/hexane as a gradient eluent to obtain the pure product. Full details are provided in the SI.