Carbonate-promoted C–H carboxylation of electron-rich heteroarenes†

C–H carboxylation is an attractive transformation for both streamlining synthesis and valorizing CO2. The high bond strength and very low acidity of most C–H bonds, as well as the low reactivity of CO2, present fundamental challenges for this chemistry. Conventional methods for carboxylation of electron-rich heteroarenes require very strong organic bases to effect C–H deprotonation. Here we show that alkali carbonates (M2CO3) dispersed in mesoporous TiO2 supports (M2CO3/TiO2) effect CO32−-promoted C–H carboxylation of thiophene- and indole-based heteroarenes in gas–solid reactions at 200–320 °C. M2CO3/TiO2 materials are strong bases in this temperature regime, which enables deprotonation of very weakly acidic bonds in these substrates to generate reactive carbanions. In addition, we show that M2CO3/TiO2 enables C3 carboxylation of indole substrates via an apparent electrophilic aromatic substitution mechanism. No carboxylations take place when M2CO3/TiO2 is replaced with un-supported M2CO3, demonstrating the critical role of carbonate dispersion and disruption of the M2CO3 lattice. After carboxylation, treatment of the support-bound carboxylate products with dimethyl carbonate affords isolable esters and the M2CO3/TiO2 material can be regenerated upon heating under vacuum. Our results provide the basis for a closed cycle for the esterification of heteroarenes with CO2 and dimethyl carbonate.


Introduction
C-H carboxylation (Scheme 1) is a compelling alternative to conventional syntheses of carboxylic acids that utilize oxidative transformations or more functionalized substrates and has attracted attention as a way to expand the use of CO 2 in chemical production. [1][2][3] However, carboxylation faces the challenge of overcoming the low reactivity of C-H bonds and CO 2 , and it lacks the large intrinsic driving force of other C-H functionalizations such as oxidation or amination. The insertion of CO 2 into C-H bonds to form a carboxylic acid is actually endergonic on account of the small DH and negative DS, while C-H carboxylation is exergonic (depending on base strength) because of the driving force from deprotonation. Increased interest in this transformation over the last several years has led to a number of methods that encompass both acid-base 2,4-9 (ionic) and radical mechanisms 2,10-14 for C-H activation. Despite these recent advances, most methods for C-H carboxylation under conventional, solution-phase conditions require highly reactive, resource-intensive reagents to activate C-H bonds. As such, the development of alternatives that use benign, regenerable reagents is critical to create opportunities for scalable CO 2 utilization.
The carboxylation of aromatic substrates is of particular interest for the synthesis of a wide variety of both ne and commodity chemicals. 2,[15][16][17][18][19] Because of the high bond dissociation enthalpy (BDE) of aromatic C-H bonds, acid-base (ionic) activation of the substrate has been the most commonly employed strategy. For some substrates, deprotonation of an X-H bond (X ¼ heteroatom) generates a nucleophilic intermediate that undergoes C-H carboxylation via an electrophilic aromatic substitution (EAS) mechanism. The classic example is the Kolbe-Schmidt reaction used for aspirin synthesis, in which phenol is transformed into salicylate by reaction with hydroxide and CO 2 . 20 While the carboxylation of indoles and pyrroles has been achieved similarly, 6,8,21 these reactions have required the use of superstoichiometric LiO t Bu to deprotonate the N-H bonds.
Apart from these special cases, carboxylation of (hetero) arene substrates via acid-base chemistry requires direct activation of the C-H bond to generate a reactive carbon-centered nucleophile. Within the last decade, several groups have demonstrated Brønsted-base-promoted carboxylation of (hetero)arenes in organic solvents at near ambient CO 2 pressure. 4,6-9,21 Hu et al. have shown that relatively acidic heteroarenes (pK a up to 28 in organic solvent) can be carboxylated using Cs 2 CO 3 as the base in reuxing DMF. 4,7 The carboxylation of electron-rich heteroarenes beyond this pK a threshold, however, has required much stronger bases. For example, the carboxylation of benzothiophene (pK a of 33 in THF) was not possible under these same conditions. 4 Recently, Kondo et al. have demonstrated the carboxylation of a diverse set of (benzo) thiophenes and (benzo)furans by reaction with excess LiO t Bu, CsF, and crown ether at 160 C under a CO 2 atmosphere. 9 However, these conditions were not able to carboxylate protected indoles, such as 1-methylindole, whose C2 carbon has a pK a near 38. Carboxylation of electron rich heteroarenes functionalized with an amide directing group has been achieved using Ni catalysis with stoichiometric KO t Bu and Mn 0 . 22 Arenes have pK a s that generally lie beyond what can be measured (pK a > 40). Researchers have developed methods to carboxylate arenes that are functionalized with a directing group by using a Rh or Pd species to catalyze C-H activation. [23][24][25] In addition to the directing group and catalyst, these methods also require a strong base (KO t Bu) or Lewis acid activator (AlMe(OMe) 2 ) to engender reactivity. In the absence of a directing group, solution-phase arene C-H carboxylation requires an extremely strong base such as Schlosser's base, 26 or stoichiometric aluminum reagents. 25,27 Apart from acid-base strategies, a very recent report by König et al. has described a photoredox method to carboxylate of (hetero)arenes under mild conditions in which the substrate is activated by one-electron photoreduction and Cs 2 CO 3 serves as the stoichiometric base. 28 This method affords moderate to high yields across a variety of substrates, although it is presently incompatible with some classes of (hetero)arene substrates and uses relatively high loadings of a photocatalyst requiring multistep synthesis.
We previously showed that simple alkali carbonates (M 2 CO 3 ) can promote C-H carboxylation of very weakly acidic substrates in solvent-free, alkali salt media at elevated temperature. [29][30][31][32] This transformation is particularly useful for converting a monocarboxylate substrate into a dicarboxylate product, where the substrate enables the formation of a molten reaction medium. 33 More recently, we demonstrated that M 2 CO 3 dispersed into mesoporous TiO 2 (M 2 CO 3 /TiO 2 , Fig. 1a) promotes the carboxylation of benzene and other aromatic hydrocarbon C-H bonds in gas-solid reactions (Scheme 1b). 34 Dispersion in mesopores disrupts the bulk M 2 CO 3 crystal structure, creating an amorphous material that can attain superbase reactivity, even in the presence of CO 2 . This carbonate-promoted C-H carboxylation of aromatic hydrocarbons takes place at moderate pressures and temperatures of $400 C. In this study, we begin to assess the generality and selectivity of this strategy using electron-rich heteroarenes, which have somewhat more acidic C-H bonds. We show that gas-solid carbonate-promoted C-H carboxylation occurs at substantially lower temperatures for these substrates and that selective reactions are possible in the presence of multiple C-H bonds (Scheme 1c, Fig. 1b). For thiophene substrates, the selectivity and mechanistic studies support a carboxylation pathway that proceeds via C-H deprotonation by the amorphous CO 3

2À
, as seen previously with arenes. For more nucleophilic indole substrates, however, carboxylation proceeds via electrophilic aromatic substitution, which provides a new pathway for CO 2 utilization enabled by dispersed carbonate materials.

Acidity calculations
Thiophene-and indole-based heterocycles were selected as C-H carboxylation substrates to probe the effects of C-H acidity and p-nucleophilicity. The C-H acidities were evaluated by using density functional theory (DFT) to calculate the standard enthalpy change for heterolytic bond dissociation in the gas phase (D acid H , also known as the gas phase acidity) ( Fig. 2 and S1 †). 35,36 D acid H provides a way to compare the thermodynamics of deprotonation irrespective of whether the pK a can be measured. Benzene, which reacts with dispersed carbonates at $400 C, has a D acid H of 401 kcal mol À1 ; its pK a is too large to be measured but has been estimated to be >43. 36 The most acidic C-H bonds in each heterocycle were found to be more acidic (lower D acid H ) than benzene by 15-23 kcal mol À1 , while the separation between the two most acidic C-H bonds in each substrate was 6-11 kcal mol À1 . For comparison, the experimental pK a values of benzothiophene (C2), thiophene (C2), and 1-methylindole (C2) are 32, 33, and 38 according to measurements performed in THF. 36,37 Additional DFT calculations to determine solution state pK a values showed good agreement to these experimental values ( Fig. S2 †).

Carbonate-promoted C-H carboxylation reactions
C-H carboxylation reactions were performed in a sealed vessel containing M 2 CO 3 dispersed on TiO 2 (M 2 CO 3 /TiO 2 , M + ¼ Cs + , K + , Na + ), heterocycle substrate, and CO 2 (see ESI † for detailed experimental procedures). In most cases, the substrate was placed within a glass culture tube in the reactor to ensure that only volatilized substrate would be able to react with the M 2 CO 3 / TiO 2 material (Fig. S3 †). The products were isolated by aqueous extraction from the TiO 2 support and quantied by 1 H NMR ( Fig. S4-S9 †). In all cases, control experiments using M 2 CO 3 without the TiO 2 support showed no reactivity, whereas M 2 CO 3 / TiO 2 promoted C-H carboxylation in varying degrees depending on the identity of M + . Additional control experiments showed minimal reactivity with the mesoporous TiO 2 support alone.
We rst assessed the temperature dependence of C-H carboxylation under a common set of conditions using 1.5 mmol substrate, a CO 2 loading corresponding to 4-5 bar at the reaction temperature, and a reaction time of 3 h ( Fig. 3 and S10 †). The relatively low substrate loading corresponded to a maximum pressure of $2.5 bar at the highest temperature evaluated (320 C). Thus, the overall pressure of the reactor at temperature was <8 bar for all of the reactions in this temperature screen. For benzothiophene (Fig. 3a), the onset of carboxylation reactivity was observed at 200 C. Optimal results were seen at 280 C, where 190 mmol of benzothiophene carboxylation product was obtained per gram of TiO 2 (190 mmol g À1 TiO 2 ) with a 20 : 1 ratio of 2carboxylate to 3-carboxylate isomers for Cs 2 CO 3 /TiO 2 . Using K 2 CO 3 /TiO 2 , 207 mmol g À1 TiO 2 of benzothiophene carboxylate product was obtained with a 25 : 1 product ratio (Fig. 3a). While both the yield and selectivity declined at higher temperatures, the carboxylation selectivity followed the C-H acidities, consistent with a mechanism gated by C-H deprotonation (see below). In contrast to Cs + and K + , much lower reactivity was observed with Na 2 CO 3 /TiO 2 , suggesting that this material is a weaker base in gas-solid reactions.
Comparison of benzothiophene carboxylation with our previous results for benzene carboxylation further highlights the effect of C-H acidity on carboxylation. Whereas >200 mmol g À1 TiO 2 of carboxylate products were obtained for benzothiophene at 280 C and <8 bar total pressure, the maximum yields for benzene carboxylation using the same M 2 CO 3 /TiO 2 materials were $100 mmol g À1 TiO 2 at 420-440 C and $30 bar total pressure. Thus, reducing the C-H bond acidity (D acid H ) by 23 kcal mol À1 enables higher yielding carboxylation reactions under substantially milder conditions (100 C lower temperature, 1/3 the total pressure). Furthermore, the benzothiophene results also demonstrate that a 7 kcal mol À1 separation in C-H acidity (C2 vs. C3 position) is sufficient for selective C-H carboxylation.
Because of its high boiling point (221 C), the vapor pressure of benzothiophene is expected to reach its saturation pressure at T # 240 C under the conditions used for the data in Fig. 3a (see Table S1 † for saturation vapor pressures calculated using the Clausius-Clapeyron equation). As a result, the vapor pressure of benzothiophene varies by $5Â over the 200-320 C range examined. To deconvolute temperature dependence from substrate pressure dependence, a series of carboxylation reactions were performed at 280 C for 3 h using different amounts of benzothiophene corresponding to calculated pressures ranging from 0.5 bar to 3.5 bar, which is approximately the saturation pressure at 280 C. The total benzothiophene carboxylate yield showed a modest variation from 150 mmol g À1 TiO 2 to 210 mmol g À1 TiO 2 over this range (Fig. S17 †). Thus, the temperature dependence of the benzothiophene carboxylation yield in Fig. 3a is primarily a result of the temperature effect on the rate constant.
Phenylthiophene reacted in a very similar manner to benzothiophene. The onset of carboxylation was observed at 200 C with very high selectivity for the 5-phenylthiophene-2carboxylate isomer (derived from the most acidic C-H bond) observed up to 280 C. Comparable yields were observed for Cs 2 CO 3 /TiO 2 and K 2 CO 3 /TiO 2 , while substantially lower yields were seen for Na 2 CO 3 /TiO 2 (Fig. 3b). The carboxylate yield varied by $50% over a 7-fold variation in phenylthiophene pressure (0.5-3.5 bar) at 320 C (Fig. S17 †). The similarity in the temperature-and pressure-dependent yields for both benzothiophene and phenylthiophene is reected in their nearly identical D acid H values for their two most acidic C-H bonds, suggesting that the same mechanism is operative for both substrates. Notably, although separating the substrate with a culture tube in the reactor ensures that it can only interact with the M 2 CO 3 via the gas phase, the carboxylation reactions with low-volatility substrates like benzothiophene and phenylthiophene proceed in comparable or better yield when the two are combined directly (Fig. S11 †).
In contrast to the heterocycle pressure dependence, increasing CO 2 pressures were found to signicantly inhibit C-H carboxylation for both substrates (Fig. S18 †). The CO 2 pressure dependence was evaluated for benzothiophene and phenylthiophene at 280 C and 320 C, respectively. Interestingly, inspection of the culture tubes for both substrates postreaction revealed increasing amounts of un-vaporized This journal is © The Royal Society of Chemistry 2020 heterocycle with increasing CO 2 partial pressure (Fig. S18 †). While the calculated saturation pressures indicate that all of the 1.5 mmol of substrate should be vaporized at these temperatures, this observation suggests that CO 2 dissolves in the substrate upon melting and lowers its vapor pressure substantially. An additional contributing factor may be that higher CO 2 pressure results in the formation of polycarbonate species (e.g. C 2 O 5 2À ) on the M 2 CO 3 /TiO 2 material, which are weaker bases than CO 3

2À
, thereby reducing the rate of C-H deprotonation. C-H carboxylation was also possible with thiophene itself. In the temperature screen performed with 1.5 mmol substrate (Fig. 3c), all of the thiophene is expected to be volatilized over the 200-320 C range because of its relatively low boiling point (84 C). The corresponding thiophene pressures range from 2-2.5 bar. In contrast to benzothiophene and phenylthiophene, no thiophene carboxylates were observed at 200 C, which is consistent with the 6 kcal mol À1 higher D acid H for its C(2)-H bond (Fig. 2). The formation of thiophene-2-carboxylate was observed beginning at 240 C, with optimal results at 280 C, where 96 mmol g À1 TiO 2 was formed along with 9 mmol g À1 TiO 2 thiophene-2,5-dicarboxylate when using Cs 2 CO 3 /TiO 2 . Comparable yields were obtained with K 2 CO 3 /TiO 2 , while Na 2 CO 3 /TiO 2 was much less effective. The observation of thiophene-2,5dicarboxylate indicates that the initially formed monocarboxylate product undergoes a second C-H carboxylation on the support. In addition to the thiophene carboxylates, $25 mmol g À1 TiO 2 of propionate was produced across the temperature range of 240-320 C (Table S12 †). This product arises from an unknown decomposition pathway starting from thiophene or a thiophene carboxylate. The yield of thiophene carboxylates was improved by increasing the thiophene pressure to 5 bar, with a comparable proportion of propionate byproduct (Fig. S17 †). In contrast to benzothiophene and phenylthiophene, essentially no CO 2 pressure dependence was observed for thiophene at 280 C. Given the much higher volatility of thiophene, CO 2 has a negligible effect on its vapor pressure at this temperature.
We next investigated the effects of increasing the nucleophilicity of the heterocycle by switching from thiophene to indole substrates. 38 To avoid the complication of an acidic N-H bond, we rst evaluated 1-methylindole. The most acidic C-H position of this substrate is C(2)-H, whose D acid H (384 kcal mol À1 ) is very close to the C(2)-H bond of thiophene (Fig. 2). The most nucleophilic position, however, is C3, 39 which has a much less acidic C-H bond (D acid H of C(3)-H is 11 kcal mol À1 higher than C(2)-H). Surprisingly, C-H carboxylation occurred readily at 200 C with a strong preference for the C3 position (Fig. 3d). Moreover, the yield increased substantially as the alkali cation size was decreased, resulting in the highest yields for reactions using Na 2 CO 3 /TiO 2 . Optimal results were obtained using Na 2 CO 3 /TiO 2 at 200 C, with a yield of 250 mmol g À1 TiO 2 for the C3-carboxylate ( Fig. 3d and S8 †). At higher temperatures (T > 240 C) the C2carboxylate was observed as an additional minor product. The selective formation of the C3 carboxylate is consistent with an EAS mechanism in which C-C bond formation precedes C-H deprotonation. Further support was found in the kinetic isotope effect for C-H carboxylation and DFT calculations (see below). Previously reported methods have achieved selective C3 carboxylation of 1-methylindole with CO 2 , but have required the use of stoichiometric organoaluminum reagents. 25,40,41 Na 2 CO 3 /TiO 2 provides a benign and much less resource-intensive alternative.
Selective C3 carboxylation was also observed with indole at 200 C using M 2 CO 3 /TiO 2 (Fig. 3e). The M 2 CO 3 dependence followed the same trend as for 1-methylindole, with optimal results obtained using Na 2 CO 3 /TiO 2 . Because the N-H functionality on indole is much more acidic than the C-H bonds ($35 kcal mol À1 difference in D acid H ), it is likely that indole is rapidly deprotonated by M 2 CO 3 /TiO 2 to form indolide, which can react reversibly with CO 2 to form indole-1-carboxylate (N-CO 2 À ). Given the very low acidity of the C-H bond at C3 (D acid H ¼ 397 kcal mol À1 , Fig. 2), the selectivity for C3 carboxylation at 200 C is consistent with an EAS mechanism in which deprotonated indole is the reactive nucleophile. 39 Beyond 200 C, however, the reaction yielded a mixture of C3, C2, and C7 carboxylation products. At 280 C, C2 carboxylation accounted for 44-64% of the total carboxylation products depending on the choice of M 2 CO 3 . Substitution at C2 is commonly seen alongside C3 in solution-phase EAS reactions with indole. 42 Both methylindole and indole showed very similar pressure dependences on both heterocycle and CO 2 partial pressure ( Fig. S17 and S18 †). Finally, 3-methylindole (skatole) was evaluated to assess the effects of blocking carboxylation at C3. Carboxylation was observed at C2 with a similar temperature dependence as seen for C2 carboxylation of indole (Fig. 3f). Using Na 2 CO 3 /TiO 2 , 94 mmol g À1 TiO 2 of the C2-carboxylate was obtained at 200 C. Increasing the temperature to 240 C boosted the yield to 138 mmol g À1 TiO 2 , although minor amounts of additional carboxylates were observed at this temperature, including the product of methyl carboxylation. To our knowledge, C2 carboxylation of 3-methylindole with CO 2 has not previously been achieved.

Kinetic isotope effects and DFT calculations to probe C-H carboxylation mechanism
To better understand the differences in thiophene-vs. indolebased heteroarene C-H carboxylation, kinetic isotope effects (KIEs) were measured using intermolecular competition experiments. 43 C-H carboxylation reactions were performed for 1-methylindole (200 C, 1.5 h) and benzothiophene (260 C, 0.5 h) using various ratios of protiated and deuterated substrate (Fig. 4). 34 KIE values of 2.0 and 1.7 were observed for C2 carboxylation of benzothiophene using K 2 CO 3 /TiO 2 and Cs 2 CO 3 /TiO 2 , respectively. These values are consistent with a mechanism in which C-H deprotonation is slow and the resulting carbanion reacts rapidly with CO 2 (Scheme 2) and does not support an EAS mechanism. In addition, previous studies of benzothiophene substitution with strong electrophiles have shown selective substitution at C3, indicating that this is the preferred position for EAS reactivity. 39,44,45 The KIE values for benzothiophene are similar to what we have previously observed for benzene C-H carboxylation using the same M 2 CO 3 /TiO 2 materials, 34 as well as solid base-catalyzed reactions that feature ratedetermining deprotonation. 46 In contrast to benzothiophene, a KIE value of 1.1 was observed for C3 carboxylation of 1-methylindole, which is within NMR quantication error of 1.0. The disparity in KIE values for these two substrates indicates distinct mechanisms for their C-H carboxylation reactivity. The lack of a KIE for 1methylindole is consistent with an EAS mechanism at 200 C in which attack of the p system on CO 2 precedes C-H deprotonation (Scheme 2). To our knowledge, an EAS reaction between CO 2 and a neutral substrate has not previously been reported. DFT calculations were performed to assess the feasibility of such a pathway with 1-methylindole. Calculations performed using either vacuum or low dielectric solvents (3 < 9) failed to identify a transition state or putative EAS intermediate, suggesting that a gas-phase reaction between 1-methylindole and CO 2 is unlikely. With a higher dielectric (3 > 20), however, an EAS transition state was identied that is $30 kcal mol À1 higher in energy than the substrates (Fig. S19 †). Interestingly, the zwitterionic intermediate resulting from CO 2 addition was very close in energy to the transition state, indicating that the reverse reaction is extremely rapid. Together, the KIE and DFT results suggest that the carboxylation of methylindole takes place via an EAS mechanism with substrate that is adsorbed onto the M 2 CO 3 /TiO 2 material. The amorphous carbonate provides a dielectric to stabilize the transition state for CO 2 addition and a proximal base that can immediately deprotonate the putative zwitterionic intermediate. The higher yield for Na 2 CO 3 /TiO 2 may reect a stronger adsorption of 1-methylindole because of the higher charge density for Na + . Further studies incorporating atomistic modeling of the amorphous carbonate surface are needed to assess this pathway more thoroughly. Nonetheless, the DFT results indicate that an EAS-like mechanism is possible.

Carboxylate esterication and M 2 CO 3 /TiO 2 regeneration
In our previous study of arene C-H carboxylation, we showed that arene carboxylates could be isolated as methyl esters with concomitant regeneration of the M 2 CO 3 /TiO 2 material by subjecting the carboxylation product to owing CO 2 and methanol at elevated temperatures. 34 The same procedure was unsuccessful for isolating heteroarene carboxylate esters because their high boiling points (>300 C) necessitated temperatures that led to decomposition under the reaction conditions. The use of owing CO 2 and dimethyl carbonate enabled isolation of methyl esters, but the yields were <50% (Fig. S20 †). Instead, it was found that we could isolate the ester at near quantitative yields by heating the supported heteroarene carboxylate ((RCOOM)/TiO 2 ) in neat dimethyl carbonate at 160 C within a stainless-steel batch reactor (Fig. S21 †). Subsequent heating of the support material under vacuum resulted in regeneration of M 2 CO 3 /TiO 2 . Aer establishing optimal carboxylation and methylation conditions, we assessed the ability of M 2 CO 3 /TiO 2 to catalyze a closed heteroarene esterication cycle over multiple iterations (Fig. 5). When a single sample of Cs 2 CO 3 /TiO 2 was used for 5 cycles, methyl benzothiophene-2-carboxylate was isolated as the only detectable product by NMR (Fig. S16 †) from each cycle with an average yield of 150 mmol g À1 TiO 2 . In each cycle following the methylation step, an aliquot of the support ($50 mg) was analyzed by aqueous extraction and 1 H NMR to detect unreacted, supported carboxylate. In all cases, no supported carboxylates were observed, indicating complete methylation. Over the ve cycles, no indication of catalyst degradation was observed (Fig. 5). These results support previous observations of the ability for dispersed carbonates to catalyze a closed esterication cycle, 34 and extend the substrate scope to include heteroarenes.

Conclusion
Conventional solution-phase methods for C-H carboxylation of aromatic substrates with low C-H acidity have relied on the use of highly reactive and resource-intensive organic bases. Our results show that CO 3 2À can serve as a benign, regenerable base for C-H carboxylation via a gas-solid reaction utilizing a dispersed, amorphous carbonate material. Compared to reactions with benzene and other arenes using the same M 2 CO 3 /TiO 2 materials, the heteroarene carboxylations investigated here reach higher yields (up to 250 mmol g À1 TiO 2 ) under substantially milder conditions (200 C lower temperature, 1/3 the total pressure). Thiophene-based heterocycles react preferentially at the most acidic C-H bond. The temperaturedependent selectivity and KIE measured for benzothiophene are consistent with a mechanism in which C-H deprotonation is followed by C-C bond formation. In contrast, indole-based heterocycles react preferentially at the most nucleophilic position (C3). DFT calculations and the absence of a signicant KIE support an EAS mechanism for the carboxylation of 1-methylindole, which nonetheless requires dispersed carbonate. The combination of CO 3 2À -promoted C-H carboxylation and methylation with dimethyl carbonate provides a two-step cycle to convert aromatic heteroarenes into methyl esters with regeneration of M 2 CO 3 /TiO 2 . Ongoing work seeks to improve the efficiency of this cycle by using alternative supports to increase the loading of reactive carbonate and access reactivity at lower temperatures.

Conflicts of interest
The authors declare no competing nancial interests.