N ‑ Heteroaryl Carbamates from Carbon Dioxide via Chemoselective Superbase Catalysis: Substrate Scope and Mechanistic Investigation

: We report a mild superbase-catalyzed and nitrogen-selective carboxylation of N -heteroaryls, with subsequent alkylation enabling the synthesis of drug-like O -alkyl carbamates in good yields (av. 86%). Our findings suggest a partial revision of the current mechanistic understanding as superbases upon mixing with indoles and azoles generally form uncharged hydrogen-bonded complexes and not ionic salts as previously proposed. However, when these complexes are exposed to CO 2 , carbamate salts are formed. These can be categorized into two subgroups, stable and fluxional carbamate salts, where the latter undergo fast and reversible CO 2 exchange, thus being poor substrates for alkylation. Experiments and DFT calculations indicate that the fluxional behavior is primarily caused by substrate-specific electronic destabilization effects. The degree of destabilization depends on the number of nitrogen atoms within and the functional group substitution on the heterocyclic ring structures. Fluxionality can be compensated for by the use of lower temperatures and/or higher CO 2 pressures as both measures stabilize the carbamate salts sufficiently, enabling subsequent alkylation.


■ INTRODUCTION
Nitrogen-containing heteroaryls, such as indoles and azoles, are important constituents of natural products and many FDAapproved small-molecule pharmaceuticals (Scheme 1A). 1−3 Consequently, there have been extensive research efforts on Nheteroaryl functionalization and derivatization, which are essential for the development of new chemical compounds for medicinal and other applications. 4−9 Carboxylation at the nitrogen position leads to the formation of carbamates which are prodrug elements and key motifs in many pharmaceuticals. 10−13 Indeed, N-heteroaryl carbamates are promising medicinal targets due to their potential in improving physiochemical properties and stabilities under biological conditions (Scheme 1B). 13,14 A recent example involves the utilization of N-heteroaryl carbamates as intermediates in the prodrug development of pibrentasvir. 15 Carbamates based on N-heteroaryls have also been applied as degradable polymers. 16 Carbon dioxide (CO 2 ) is a synthetically attractive molecule as it is a non-toxic, inexpensive, and readily available C1source. 12,17 The use of CO 2 has received significant attention in the functionalization of N-heteroaryls, particularly in C−C bond formation. 5−7 The alternative C−N bond formation, resulting in the formation of carbamates, is well-known for saturated N-heterocycles (Scheme 2A). 11,17 However, carbamate formation using CO 2 and unsaturated N-heterocycles (Nheteroaryls) is a much less explored field (Scheme 2B).
Indoles, an important subgroup of N-heteroaryls, can be Ncarboxylated using n-BuLi; however, the use of a hard base severely limits functional group compatibility. 18−21 Additionally, the inherent nucleophilic reactivity of indoles resides at the C3 position, 22 implying that most CO 2 -based carboxylations take place here. 5,23−25 Consequently, directing reactivity to nitrogen N1 is challenging. 26,27 A recent report detailed indole N-carboxylation as part of an intramolecular cyclization. 28 The current state-of-the-art approach to accessing indole O-alkyl carbamates involves a two-pot process (Scheme 2C). 29,30 Alcohols are transformed to imidazole carbamates, and once isolated, these can be used to transfer the carbamate moiety to indoles. It would be advantageous if the desired indole O-alkyl carbamates could be prepared directly from benign CO 2 without the need to prepare and isolate imidazole carbamates. For related azole-based substrates, Nformylated products can be obtained with CO 2 under reductive conditions, whereas carbamates can be formed catalytically using an NHC−copper complex. 31−33 Selective formation of desired products from N-heteroaryls requires a good understanding of their inherent reactivities. N-Heteroaryls have found widespread application in ionic liquids as the anionic component (Scheme 2D). 34−48 The cationic component can be a tetraalkylphosphonium, 36−39 an azolium (HetAr + ), 40,41 or a superbase, such as TMG, DBU, or MTBD, which are proposed to deprotonate N-heteroaryls. 42−48 The resulting salts interact with CO 2 , but the mechanistic details are not thoroughly understood. 49,50 The behavior of ionic liquids is important to understand for rationalizing their role as modulators and enhancers of heterogeneous catalysts. 51−53 Superbases have many important applications as catalysts and stoichiometric reagents. 54 Consequently, understanding superbase-mediated nucleophile activation is critical for designing new reactions. 55 We reasoned that the superbase-mediated activation of N-heterocycles could enable a convenient and complementary CO 2 -based entry to the synthesis of Nheteroaryl carbamates. Herein, we present a one-pot method toward in situ conversion of N-heteroaryls and CO 2 to carbamate anions, followed by a facile reaction with available alkyl bromides (Scheme 2E). Using this approach, structurally diverse indole carbamates are obtained with complete selectivity for nitrogen N1 in high yields and under mild conditions. Our mechanistic analysis reveals clear trends in the reactivity of N-heteroaryls with CO 2 , leading to a partial revision of the currently accepted mechanistic understanding. ■ RESULTS AND DISCUSSION Indole Carboxylations and Reaction Scope. Initially, we investigated the reactivity of unsubstituted indole 1a with 1-bromopentane under 1 bar CO 2 (Scheme 3). After optimization (Supporting Information, Section S11), we found that the reaction could be conducted under stoichiometric conditions (condition A) or alternatively under catalytic conditions (condition B). The latter produced O-alkyl carbamate 2a in a superior yield. Carboxylation occurred exclusively at N1, i.e., no reaction at C3 was observed.
The generality of the reaction was explored for electronically varied indoles. Electron-donating (2b and 2c) and moderately electron-withdrawing substituents (2d−g) are well tolerated. Strongly electron-deficient O-alkyl carbamates 2h and 2i can be formed via conditions A and B but with significant quantities of N-alkylated byproduct 3 (Supporting Information, Section S11.3). Gratifyingly, increasing the CO 2 pressure to 40 bar and using 1-iodopentane (Scheme 3, condition C) suppressed the formation of byproduct 3, thereby providing the desired O-alkyl carbamates in good yields.
The structural variation of the indole ring system was explored next. Introduction of a second nitrogen atom (azaindole 2j) is only compatible with condition A, while adding a fused benzene ring (2k) or removing one (2l) is well tolerated. More complex substrates such as the sleep hormone melatonin (2m) and biologically relevant tryptophan derivates (2n and 2o) are well tolerated, with enantiopurity preserved under catalytic condition B. Curiously, indoles bearing substituents adjacent to the nitrogen center fail to react with CO 2 (1p and 1q). These indoles are also known to be unreactive in carbamate transfer (Scheme 2C). 29 The scope of alkyl bromides utilizing condition B was then explored (Scheme 4). Complex functionalities are well tolerated, including heterocycles and secondary bromides. Drug-like compounds 4a-f and 4r-g are prepared in a facile manner. The structure of 4a-e was confirmed by X-ray crystallography. 56 We conclude that the developed conditions A, B, and C provide facile access to indole O-alkyl carbamates 2 and 4, with sensitive synthetic handles such as aldehyde (2f), boronic ester (2g), Weinreb amide (2n), and alkyne (4a) remaining intact. In the absence of CO 2 , only the N-alkylated byproduct 3 is observed (Supporting Information, Section S11). The superbase is a critical component as its absence dramatically decreases the carbamate yield, leading to the formation of significant quantities of 3. To understand the role of the superbase, we turned toward exploring the reaction mechanism.
Mechanistic Studies on Indole Carboxylations. On the basis of literature results and our previous work, we propose a catalytic reaction pathway for superbase-mediated carboxyla-tion of indoles (Scheme 5). 57,58 Initially, the superbase reacts with CO 2 and indole 1a, forming a mixed carbamate. The latter reacts with the alkyl bromide through an S N 2-type reaction, forming the observed product 2a. The superbase is regenerated by Cs 2 CO 3 . In the absence of a superbase, Cs 2 CO 3 can mediate the reaction; however, cesium carbamate salt 5a seems to be significantly more unstable than the superbase-derived mixed carbamate. We deduce this from the substantially lowered yields of product 2a and the large quantities of byproduct 3a. DFT calculations (PBE0-D3BJ[IEFPCM-(DMF)]) on the reaction of 1a with CO 2 in the presence of the superbase TMG show that the proposed mixed carbamate formation has a feasible barrier of 16.1 kcal/mol (313 K, Supporting Information, Section S1.2). In contrast, the competing N-alkylation has a high barrier of 33.0 kcal/mol, in line with the experimentally observed absence of byproduct 3a.
The key step of mixed carbamate formation was investigated using in situ NMR (Scheme 6A). We employed TMG as the base as it produces simpler NMR spectra than DBU while being catalytically comparable (Supporting Information, Section S11.1). Reference spectra were first measured of indole 1a and TMG under argon (Scheme 6A). Under these conditions, TMG and indole 1a appear to form a hydrogenbonded complex as indole 1a NH is shifted downfield by 0.11 ppm by 1 H NMR in the presence of TMG. Similar bonding is known for anilines. 58 The complex appears to consist of neutral components, as judged from the TMG central carbon 13 C{ 1 H} resonance at 166.3 ppm, which is close to freebase at 166.2 ppm. 58 This finding seems to contradict previous literature, which has proposed that indole 1a is deprotonated by superbases. 43 However, DFT calculations (PBE0-D3BJ-[IEFPCM(DMF)], Supporting Information, Section S1.3) support our experimental findings, predicting a cost of 5 to 8 kcal/mol for forming ionic salts from TMG and indoles 1a, 1h, or 1q. Known pK a values for indole 1a (21.0) and TMG (13.2) in DMSO also support that the indole 1a−TMG complex remains fully neutral (Supporting Information, Section S4.1). 59−61 We proceeded to measure the NMR of indole 1a and TMG under CO 2 (1 bar). Under these conditions, TMG becomes protonated as the central carbon is now observed at 162.0 ppm (Scheme 6A). This is consistent with the formation of a mixed carbamate, which is also supported by the appearance of the indole carboxylate as a new 13 C{ 1 H} resonance at 151.8 ppm. It was further assessed if the change from a neutral complex (under argon) to an ionic salt (under CO 2 ) can be observed with other NMR-active nuclei. The difference between freebase and protonated TMG is ca. 0.3 ppm by 1 H NMR. While this is a meaningful difference, we felt a larger absolute value would be beneficial in determining the protonation state of TMG. In this regard, we found that 15 N NMR at natural isotopic abundance is very sensitive since the difference between freebase and protonated TMG is up to 80 ppm (Supporting Information, Section S4.4). The 15 N NMR results clearly indicate that the indole 1a−TMG complex is neutral under argon but forms an ionic mixed carbamate under CO 2 .
While 15 N NMR is accurate, we recognize the long acquisition times (15−48 h). As a compromise between high sensitivity and short acquisition times, we advocate the use of 13 C{ 1 H} NMR in studying whether N-heteroaryl−superbase complexes are neutral or ionic. This strategy is based on the key observation that the signal from the central carbon of TMG is sensitive to protonation. Under our conditions, goodquality spectra (of TMG) are obtained in 1 min. The total change in the chemical shift of the TMG central carbon, Δδ( 13 C), is provided in Scheme 6 using eq 1. This equation assesses changes in protonation state when changing the atmosphere from argon to CO 2 . For example, if a Nheteroaryl−TMG complex is a fully ionic salt under argon, then switching to CO 2 should not affect the protonation state of TMG, resulting in Δδ( 13 C) = 0. The generality of eq 1 was also demonstrated for other complexes of N-heteroaryls and superbases (vide infra).
( C) ( C CO ) ( C Ar ) 13 13 2 We continued by studying a selection of 5-substituted indoles in order of increasing electron-deficiency by the Hammett constant σ p (Scheme 6B). NMR spectra were recorded first under argon and then under CO 2 (1 bar). In all cases, there was a large change in the protonation state of TMG (Δδ( 13 C) ≈ −4 ppm). This indicates that the studied indole−TMG complexes are neutral under argon but form ionic salts under CO 2 . The electronics of the parent indoles have a significant effect on the carboxylation behavior of the observed as a single set of broad and low-intensity signals due to rapid CO 2 exchange on the NMR time scale. 58 A rapid CO 2 exchange rate is supported by the fact that these indoles require elevated CO 2 pressure (40 bar) to suppress the formation of N-alkylated byproduct 3 (Scheme 3 and Supporting Information, Section S11.3). The higher CO 2 pressures likely stabilize the fluxional behavior and, therefore, the mixed carbamate has a sufficient life time to react with alkyl bromides.
If fluxional behavior was to be caused by rapid CO 2 exchange, the equilibrium would also be expected to be sensitive to thermal effects. 58 In this regard, we performed variable-temperature NMR studies (VT-NMR) in DMF-d 7 , similar in properties to DMSO-d 6 but melting at lower temperatures. Electron-deficient and highly fluxional 5cyanoindole 1h was chosen as the substrate (Figure 1). Once CO 2 (1 bar) is applied, the resonances of protons H7 and H2 become very broad. In the carbon spectrum, there is significant line broadening, particularly for carbons adjacent to N1. This is accompanied by N1 no longer being detectable by 15 N NMR. Lowering the temperature to −25°C stabilizes the mixed carbamate, as is evident from the increased sharpness and intensity of H7 and H2, as well as the overall sharpening of the 13 C{ 1 H} spectrum. Moreover, N1 is again detectable by 15 N NMR. In other words, when the rate of CO 2 exchange is fast, the reactive nitrogen atom is not detectable by 15 N NMR. The NMR behavior of 5-cyanoindole 1h indicates that low temperatures can stabilize fluxional mixed carbamates by reducing the rate of CO 2 exchange. Finally, the effect of pressure was evaluated by in situ NMR at 25°C (Supporting Information, Section S14.3). Increasing pressure from 1 bar CO 2 to 5 bar has no observable effect on 5-cyanoindole 1h by 1 H or 13 C{ 1 H} NMR. However, a slight increase in stabilization is observed at 10 bar CO 2 by 1 H NMR. These changes are of lesser magnitude in comparison to thermal stabilization, i.e., when the sample is cooled from 25 to 0°C. Detailed insights into the operative carboxylation pathways were obtained from computational modeling of different indoles (PBE0-D3BJ[IEFPCM(DMF)], 298 K). Our analysis shows that TMG pre-associates to the indole through hydrogen bonding. The subsequent attack on CO 2 occurs concertedly with proton transfer from indole 1a to TMG (Figure 2). 58 The computed Gibbs free energies at 298 K show a favorable barrier of 15.7 kcal/mol for this step, with a reaction energy of −2.8 kcal/mol ( Table 1, entry 1), in line with the experimentally observed reactivity. Electron-deficient 5-cyanoindole 1h also provides a favorable computed barrier and reaction energy (entry 2). Interestingly, indoles 1p and 1q, which failed to produce any carbamate in our experiments (Scheme 3), display feasible barriers but endergonic reaction energies (2.6 and 4.0 kcal/mol), indicating that decarboxylation is faster than carboxylation (Table 1, entries 3−4). The optimized geometries of products 1a and 1h show that the carboxylate is nearly in-plane (θ ≈ 15°, entries 1−2). In contrast, in the products of 1p and 1q, the adjacent methyl and phenyl substituents force the carboxylate out-of-plane (θ ≈ 40°, entries 3−4), reducing non-covalent charge stabilization compared to 1a and 1h (see Supporting Information, Section S1.4).
Differing Reactivity of Azoles. Having successfully reacted CO 2 with indoles and pyrrole, we proceeded to investigate structurally related azoles 1s−w (Scheme 7). Generally, complex product mixtures or N-alkylated byproducts 3 were obtained, which gave us the initial impression that azoles do not react with CO 2 . We turned our attention toward understanding this apparent lack of reactivity from a mechanistic perspective.
First, we pondered if the interaction between azoles and superbases differs from that between indoles. Previous literature has proposed that azoles and indoles are deprotonated by superbases, forming ionic liquids. 42−48 However, our computational studies show that azole−TMG complexes, just as indole−TMG complexes, prefer to be neutral, with the exception of tetrazole−TMG and 4-nitroimidazole−TMG (Supporting Information, Section S1.3). Additionally, the predicted extent of salt formation was estimated based on Scheme 6. NMR Assessment of 5-Substituted Indoles with TMG (1 equiv)

ACS Catalysis pubs.acs.org/acscatalysis
Research Article known pK a values, and the results were aligned with the computed energies (Supporting Information, Section S4.3). 59−61 Our DFT-based analysis of the carboxylation reactions indicates that pyrrole 1l has a favorable barrier and reaction energy ( Table 2, entry 1), being similar in value to those indoles that yield isolable O-alkyl carbamates ( Table 1, entries 1 and 2). This is in line with the experimentally obtained product 2l (Scheme 3). Azoles 1s−w contain multiple nitrogen atoms in the ring structure, assigned as pyridine-like or pyrrolelike (Table 2 graphic). For both nitrogen types, all azoles show feasible carboxylation barriers; however, there is a clear preference for the pyridine-like nitrogen atom. An exception is 1w, likely due to the proximity of the nitro-group. The computed reaction energy of pyrazole 1s (entry 2) is favorable toward the formation of carbamate, although slightly less when compared to pyrrole 1l, whereas the other azoles (1t, 1u, 1v, and 1w, entries 3−6) display endergonic reaction energies. The lack of isolable products in experiments (Scheme 7) can be explained by the endergonic reaction energies, which imply that the reverse decarboxylation barriers are lower than the carboxylation barriers.
To obtain additional experimental support for our hypothesis of fast carboxylation-decarboxylation, we performed in situ NMR studies. Pyrrole 1l (pK a 23.0), which forms a neutral complex with TMG (pK a 13.2) under argon (Scheme 8A), becomes ionic when CO 2 (1 bar) is applied (Δδ( 13 C) = −4.0 ppm). 60 The mixed carbamate forms in 80% yield by 1 H NMR, while the carboxylate Δδ( 13 C) signal at 150.5 ppm is sharp and of high intensity (Figure 3), indicating that pyrrole forms a stable mixed carbamate.
The different carboxylation behaviors of pyrrole and pyrazole were intriguing and suggested that, as shown for indoles above (Scheme 6), 5-membered N-heteroaryls can be categorized as two distinct subgroups, stable or fluxional mixed   carbamates. To explore the generality of this idea, we conducted a more systematic NMR study of 5-membered Nheteroaryls ( Figure 5). 62 Imidazole 1t produces a very similar 13 C{ 1 H} spectrum to isomeric pyrazole 1s. Both azoles show a broad low-intensity carboxylate, implying similar interaction with CO 2 . For azoles containing a third nitrogen, 1u and isomeric 1,2,3-triazole 1u′, the carboxylates are located somewhat more upfield than for 1s and 1t, retaining the low intensity, but being sharp in shape. This suggested to us that triazoles have a weaker interaction with CO 2 . Tetrazole 1v contains four nitrogen atoms, and only free CO 2 (124.2 ppm) is detected, suggesting very little reactivity with CO 2 (Supporting Information, Section S15.9). Compared to imidazole 1t, 4-nitroimidazole 1w has a significant upfield shift of the carboxylate, suggesting that −NO 2 induces potent electronic destabilization. The NMR results in Figure 5 indicate that the introduction of a second, pyridine-like nitrogen into the heterocycle induces f luxional behavior. Any additional nitrogen or strongly electron-withdrawing group seems to weaken the interaction between CO 2 and any given azole. This observation was evaluated by measuring the CO 2 binding capacity of neat azole−superbase complexes using thermogravimetric analysis (TGA, Figure 5). 63 We found that fluxional mixed carbamates with two nitrogen atoms in the ring (1s and 1t) bind ca. 0.8 equiv CO 2 , which is associated with a large change in protonation state (Δδ( 13 C) ≈ −4 ppm). This amount is approximately halved with the introduction of a third nitrogen atom (1u and 1u′) or a strongly electron-withdrawing group (1w). A fourth nitrogen (1v) results in no measurable binding of CO 2 (for details, see Supporting Information, Section S4.3). Our findings are in line with computationally predicted CO 2 absorption 34 and previous experimental studies applying azole−superbase complexes as CO 2 capture materials. 42,43,46 Next, the stability of azole-derived mixed carbamates under lower temperatures was investigated. First, we computed thermal effects on 1,2,4-triazole 1u ( Figure 6). The results show that the pyrrole-like nitrogen N4 has the lowest carboxylation barrier at all tested temperatures, yet its reaction energy is consistently the least favorable. In contrast, carboxylation at N2 is favorable at 248 K and even more so at 223 K, suggesting that this carbamate is the major species at low temperatures. These results are supported by VT-NMR studies (Supporting Information, Section S5.2). At 25°C, two 15 N resonances are observable, assigned to pyridine-like N4 and pyrrole-like N1/N2. As the temperature is lowered, the 15 N signal corresponding to N4 decreases in intensity, nearly disappearing at −25°C. This suggests that the equilibrium is shifted toward carboxylated 1,2,4-triazole 1u, in agreement with the computed energies.  We then computed the reaction energies for all in situ NMR-studied N-heteroaryls at different temperatures (Supporting Information, Section S1.5). The results show that of the azoles 1s−1w, pyrazole 1s has more favorable reaction energies at lower temperatures (−2.9 kcal/mol at 248 K). Indeed, VT-NMR shows that when CO 2 (1 bar) is applied at 25°C, the 15 N resonance of pyrazole 1s is not detected. However, when cooled to −25°C, two sharp multiplicities are detected, suggesting that the fluxional behavior of pyrazole 1s mixed carbamate is fully stabilized, i.e., the rate of CO 2 exchange is slow at −25°C (Supporting Information, Section S5.1). Imidazole 1t behaves similarly but appears to be more fluxional in nature as no 15 N signals assigned to imidazole 1t are detectable under CO 2 even at −25°C (Supporting Information, Section S5.3). This is in agreement with the computational results (Supporting Information, Section S1.4). To exclude the possibility that the deuterated solvent affects the reactivity of azoles, we measured NMR of the imidazole 1t−TMG complex under neat conditions (Supporting Information, Section S5.4). In the absence of solvent, the imidazole 1t−TMG complex is neutral under argon, but forms an ionic salt (mixed carbamate) under CO 2 . Under neat conditions, carboxylated imidazole 1t is detectable by 15 N NMR as a broad resonance, consistent with fast CO 2 exchange and implying that the observed reactivity in solvents (DMSOd 6 , DMF-d 7 , and CDCl 3 ) is extendable to neat conditions. Considering that azole-derived mixed carbamates are stabilized at low temperatures, it would be beneficial to develop

ACS Catalysis pubs.acs.org/acscatalysis
Research Article subsequent functionalization reactions that proceed at such low temperatures. Finally, fused derivatives of azoles and indoles were examined in carboxylation reactions to evaluate if they behave similarly to the monocyclic counterparts ( Figure 7). Indeed, also for these substrates, more than one nitrogen in the ring system results in fluxional mixed carbamates when CO 2 is applied, as is seen from the broad carboxylate resonances. Large Δδ( 13 C) values (≈−4 ppm) are observed, indicating that fused systems form neutral complexes with TMG, which react with CO 2 to form ionic salts. Consequently, the behavior of fused systems is similar to that of monocyclic N-heteroaryls. For details, see Supporting Information, Section S6.

Role of Bases.
We questioned whether the observed reactivity patterns of N-heteroaryl−TMG complexes could be generalized to other superbases. A selection of N-heteroaryls was examined with DBU and MTBD, which yield largely the same carboxylation results as with TMG (Scheme 9). This implies that N-heteroaryl carboxylation behavior (stable or fluxional) is minimally affected by the employed superbase. However, the CO 2 binding of the N-heteroaryl is affected by the employed superbase. With TMG (pK a 13.2), indole 1a is carboxylated in 80% yield, which is increased with the stronger bases DBU (87%, pK a = 13.9) and MTBD (94%, pK a = 14.8). 64 We find eq 1 to be well-suited for evaluating whether DBU and MTBD form neutral complexes or ionic salts with Nheteroaryls. We identified DBU C7 and MTBD C4 to be sensitive toward protonation, whereas MTBD central carbon C6 is not sensitive (Supporting Information, Section S7). DBU and MTBD appear to form neutral complexes with non-acidic N-heteroaryls 1a and 1t under argon since large Δδ( 13 C) values of ca. 5 ppm (DBU) or −5 ppm (MTBD) are observed (Scheme 9). In contrast, 1,2,4-triazole 1u-based complexes have a Δδ( 13 C) of 1.1 (DBU) and −0.9 (MTBD), indicating that partial deprotonation (salt formation) occurs before the introduction of CO 2 . The DBU or MTBD complexes with tetrazole 1v have Δδ( 13 C) ≈ 0, meaning that tetrazole 1v is nearly fully deprotonated by both superbases under argon.
We have previously shown that aniline carboxylation yield correlates with the pK a of the organic base, except for sterically hindered bases. 58 Phosphazenes, such as tert-butyliminotri(pyrrolidino)phosphorane (BTPP, pK a = 17.4), are reported to be more basic than DBU or MTBD. 64 Based on pK a alone, we expected BTPP to carboxylate indole 1a in a high yield. However, the observed yield is only 63%. Despite BTPP being one of the least hindered phosphazenes, it seems its sterics are sufficient to partially inhibit the carboxylation of indole 1a. 65 Mechanistic Overview. Our overall experimental and computational results allow us to construct a comprehensive understanding of N-heteroaryl carbamate formation (Scheme 10). We find that under argon, superbases (TMG, DBU, and MTBD) and most N-heteroaryls associate by hydrogen bonding to form neutral complexes A. Only acidic N-  . The number of nitrogen atoms in the ring structure determines whether the forming mixed carbamate will be stable or fluxional, as exemplified by the carboxylate 13  When CO 2 is introduced, neutral complexes A are in equilibrium with mixed carbamates C, favoring the latter. Ionic complexes B have a significantly reduced affinity to CO 2 , implying that they are in an unfavorable equilibrium with C. The number of nitrogen atoms in the ring structure and the presence of electron-withdrawing groups (EWG) determine the rate of CO 2 exchange. Stable mixed carbamates exchange CO 2 slowly, whereas fluxional mixed carbamates have a fast exchange. We conclude that combinations of superbases and protic N-heteroaryls should be considered CO 2 -triggered switchable-polarity solvents (SPSs) 61,66,67 as they become ionic mixed carbamates first when exposed to CO 2 .

■ CONCLUSIONS
In summary, we report a comprehensive study on the superbase-mediated reactivity of indoles and related azoles with CO 2 . We have shown that the resulting mixed carbamates can be categorized as either stable or fluxional on the basis of experimental and computational results.
The stable mixed carbamates exchange CO 2 slowly and, therefore, the carbamate anion has a sufficient lifetime to readily react with alkyl bromides. The resulting O-alkyl carbamates are isolated in high yields (av. 86%) using mild and catalytic conditions with complete carboxylation selectivity for N1. A broad range of functional groups are tolerated, which are incompatible with conventional methods employing organometallic reagents.
Fluxional mixed carbamates undergo rapid CO 2 exchange, a previously not thoroughly understood phenomenon. In this regard, we demonstrate that 13 C{ 1 H} and 15 N NMR are highly informative in the characterization of this compound class. Experimental and computational results suggest that fluxional behavior is caused by electronic destabilization. This is induced by two or more nitrogen atoms in the ring structure or a strongly electron-withdrawing group. These effects can be compensated by using elevated pressures of CO 2 or low temperatures.
Finally, our results suggest a partial revision of the currently accepted view on how superbases react with protic Nheteroaryls. Under inert gas, we find that superbases and the majority of N-heteroaryls form neutral complexes and not ionic salts as previously thought (Scheme 10). The neutral complexes are better described as CO 2 -triggered switchable-polarity solvents as exposure to CO 2 results in ionic mixed carbamates. We anticipate our work to be significant not only for CO 2 -based synthesis but also in the field of CO 2 capture. ■ ASSOCIATED CONTENT
Optimized coordinates (XYZ) Additional computational details and optimized structures, general information, general experimental procedures, reaction optimization, synthetic methods, compound characterization, details of thermogravimetric curves, HPLC results, and NMR spectra (PDF)