Chiral-at-Metal: Iridium(III) Tetrazole Complexes With Proton-Responsive P-OH Groups for CO2 Hydrogenation

A rise in atmospheric CO2 levels, following years of burning fossil fuels, has brought about increase in global temperatures and climate change due to the greenhouse effect. As such, recent efforts in addressing this problem have been directed to the use of CO2 as a non-expensive and non-toxic single carbon, C1, source for making chemical products. Herein, we report on the use of tetrazolyl complexes as catalyst precursors for hydrogenation of CO2. Specifically, tetrazolyl compounds bearing P–S bonds have been synthesized with the view of using these as P∧N bidentate tetrazolyl ligands (1–3) that can coordinate to iridium(III), thereby forming heteroatomic five-member complexes. Interestingly, reacting the P,N′-bidentate tetrazolyl ligands with [Ir(C5Me5)Cl2]2 led to serendipitous isolation of chiral-at-metal iridium(III) half-sandwich complexes (7–9) instead. Complexes 7–9 were obtained via prior formation of non-chiral iridium(III) half-sandwich complexes (4–6). The complexes undergo prior P–S bond heterolysis of the precursor ligands, which then ultimately results in new half-sandwich iridium(III) complexes featuring monodentate phosphine co-ligands with proton-responsive P-OH groups. Conditions necessary to significantly affect the rate of P–S bond heterolysis in the precursor ligand and the subsequent coordination to iridium have been reported. The complexes served as catalyst precursors and exhibited activity in CO2 and bicarbonate hydrogenation in excellent catalytic activity, at low catalyst loadings (1 μmol or 0.07 mol% with respect to base), producing concentrated formate solutions (ca 180 mM) exclusively. Catalyst precursors with proton-responsive P-OH groups were found to influence catalytic activity when present as racemates, while ease of dissociation of the ligand from the iridium center was observed to influence activity in spite of the presence of electron-donating ligands. A test for homogeneity indicated that hydrogenation of CO2 proceeded by homogeneous means. Subsequently, the mechanism of the reaction by the iridium(III) catalyst precursors was studied using 1H NMR techniques. This revealed that a chiral-at-metal iridium hydride species generated in situ served as the active catalyst.


INTRODUCTION
The utilization of fossil fuels as a source of energy has contributed toward human development (Saeidi et al., 2014). However, this dependence on fossil fuels has contributed to the gradual buildup of greenhouse gases such as CO 2 in the atmosphere. This increase in the amount of greenhouse gases is a major contributor to the greenhouse effect that has caused increase in global temperatures and climate change (Huang and Tan, 2014;Saeidi et al., 2014). Increase in the amount of CO 2 as a greenhouse gas in the atmosphere has necessitated the development of economical, safe, and efficient systems to capture and utilize it as a resource instead of disposing it as waste. Thus, CO 2 serves as an attractive single carbon (C 1 ) building block for making various compounds including formic acid, methanol, formamides, formaldehyde to name a few (Ma et al., 2009). Although CO 2 is currently a C 1 source in some industrial processes, like in the production of urea and salicylic acid, its thermodynamic stability still hinders its widespread utilization. Therefore, CO 2 valorization requires electro-or photo-reduction or reacting it with high-energy substrates in the presence of catalysts. Catalytic hydrogenation of CO 2 provides an avenue for CO 2 fixation so as to unlock its full potential as a C 1 source, thereby promoting its valorization (Jessop et al., 1995).
Various homogeneous and heterogeneous catalysts have been reported for the hydrogenation of CO 2 to a variety of products. However, one of the best catalysts for the hydrogenation of CO 2 is the trihydridoiridium P ∧ N ∧ P pincer complex, by Nozaki and co-workers (Tanaka et al., 2009), which served as an excellent catalyst. Despite the impressive catalytic activity shown by the Nozaki iridium catalyst (ca TONs as high as 3,500,000), it requires a strong base to achieve such high catalytic activity and productivity (200,000 h −1 ). Also, a strong base is used during the catalytic process, meaning isolation of the product (i.e., formic acid) requires use of a strong acid in order to convert the product formate to formic acid. Use of acids in chemical processes is not ideal as they are highly corrosive to equipment and they generate unwanted acid effluent. There is also the challenge of forming unwanted inorganic salts as by-products (Yasaka et al., 2010) Therefore, the development of highly active catalysts for the hydrogenation of carbon dioxide that use mild organic bases such as DBU is desirable (Makuve et al., 2019;Malaza and Makhubela, 2020).
Another class of catalysts for CO 2 hydrogenation are iridium half-sandwich complexes. They have been shown to be effective catalysts for both CO 2 hydrogenation and formic acid dehydrogenation (Leitner et al., 1994;Himeda et al., 2007Himeda et al., , 2008Boddien et al., 2011Boddien et al., , 2012Papp et al., 2011). In fact, Himeda and co-workers have demonstrated that the electron-donating ability of N,N ′ -bidentate ligands in half-sandwich iridium complexes enhances the catalytic activity of these complexes for the CO 2 hydrogenation reaction through electronic effects (Himeda et al., 2007). More recently in 2017, Himeda and co-workers reported (cyclopentadienyl)iridium(III) complexes with pyridyl-diazole ligands that have hydroxyl electron-donating functional groups and an electron-rich diazole ring. The presence of the hydroxyl groups under basic conditions deprotonate to form strongly electron-donating oxyanions that activate dihydrogen through second sphere coordination effects, leading to the formation of the catalytically active iridium hydride in the CO 2 hydrogenation reaction (Rakowski Dubois and Dubois, 2009;Wang et al., 2012;Suna et al., 2014). This is generally referred to as proton-responsive catalytic reactions. Both computational and experimentally observed studies have confirmed the participation of proton-responsive groups in the catalytic cycle of iridium hydride species in CO 2 hydrogenation reactions (Suna et al., 2017). This, therefore, clearly indicates that the choice of ligand plays a vital role in the catalytic activity of complexes in CO 2 hydrogenation reactions.
We postulated, at the beginning of this project, that iridium tetrazole complexes featuring NH groups are likely to display proton-responsive behavior and thus act as highly active CO 2 hydrogenation catalysts. This formed the basis of our investigation and the current report (Ocansey et al., 2020). We are aware of several tetrazole MOFs, with possible proton-responsive NH groups, that have mostly been utilized in the synthesis of MOF for CO 2 capture or CO 2 -assisted cycloaddition reactions, but none, to the best of our knowledge, has been used as CO 2 hydrogenation catalyst (Pachfule and Banerjee, 2011;Cui et al., 2012;Poloni and Smit, 2012;Go et al., 2013;Zhao et al., 2016). Such CO 2 adsorption by the tetrazoles can be attributed to hydrogen bonding between CO 2 molecules and polarizing groups in the tetrazole (Poloni and Smit, 2012;Zhao et al., 2016). We believe that it is this polarizing effect in our (cylopentadienyl)irirdium(III) tetrazole complexes that makes them very good catalysts for CO 2 hydrogenation catalysts.
Another attribute of the (cylopentadienyl)irirdium(III) tetrazole complexes as CO 2 hydrogenation catalysts is their chirality at the iridium center. There are prominent examples of optically active half-sandwich piano stool complexes that are chiral-at-metal complexes where the configuration may be stable or labile in solution. Such compounds allow the monitoring of the stereochemical course of substitution reactions as well as being used in organic synthesis in ligand transformation reactions as a result of their chirality (Brunner, 2001). Complexes with chirality at the metal center have been used as effective catalysts in the enantioselective Friedel-Crafts addition of indoles to α,β-unsaturated 2-acyl imidazoles (Huo et al., 2014) and trichloromethylation of 2-acyl imidazoles and 2-acyl pyridines (Huo et al., 2015). In addition, these chiral complexes have been used as hydrogenation catalysts in the hydrogenation of acyclic aromatic N-aryl imines (Imamoto et al., 2006), α-dehydroamino esters, enamides, dimethyl itaconate (Kurihara et al., 2009), and imines (Murata et al., 1999), and in transfer hydrogenation of β,β ′ -disubstituted nitroalkenes (Chen et al., 2013) and ketones (Tian et al., 2016) with high enantioselectivities and at low catalyst loadings. Despite the impressive array of hydrogenation reactions these complexes are capable of, they have rarely been evaluated as catalysts for CO 2 hydrogenation. Knowing the role of proton-responsive groups on hydrogenation catalysts and the electron-rich nature along with CO 2 affinity of tetrazole compounds led to the present study in search of highly active CO 2 hydrogenation catalysts.

EXPERIMENTAL
All reactions were carried out in air unless otherwise stated. All solvents used were reagent grade, purchased from Sigma-Aldrich, and dried under nitrogen before use. 1-Phenyl-1H-tetrazole-5-thiol, chlorodiphenylphosphine, chlorobis(3,5-dimethylphenyl)phosphine, bis(3,5di(trifluoromethyl)phenyl)chloro phosphine, and sodium methoxide were purchased from Sigma-Aldrich and used without further purification. Iridium(III) chloride was purchased from Heraeus South Africa and used as received. [Ir(C 5 Me 5 )Cl 2 ] 2 (White et al., 1992) was synthesized according to literature procedures. NMR spectra were recorded on a Bruker 400-MHz NMR spectrometer ( 1 H at 400 MHz, 13 C{ 1 H} at 100 MHz, and 13 P{ 1 H} NMR 161.99 MHz). Spectrometer chemical shifts were reported relative to the internal standard tetramethylsilane (δ 0.00 ppm) and referenced to the residual proton and carbon signals at 7.24 and 77.0 ppm, respectively, of CDCl 3 . Infrared spectra were obtained neat using a Perkin Elmer Spectrum BX II fitted with an ATR probe. Melting points were obtained using a Gallenkamp Digital Melting-point Apparatus 5A 6797. Elemental analysis was performed on a Thermos Scientific FLASH 2000 CHNS-O Analyzer. Mass spectrometry was performed using Waters Synapt G2 mass spectrometer with both ESI positive and Cone Voltage 15 V. XRD spectra were obtained from a Bruker APEX-II CCD Diffractometer.

Synthesis of [IrCp
(2) was added to 95 mg (0.12 mmol) of [Ir(C 5 Me 5 )Cl 2 ] 2 using 5.00 ml of methanol as solvent. The reaction mixture was then stirred for 24 h under argon at room temperature. After the reaction time had elapsed, the precipitate formed was filtered off in air, and the filtrate dried with the crude product obtained being washed with 10 ml 1:1 (v:v diethylether: hexane) and subsequently dried in vacuo for 6 h to yield (5

Synthesis of [IrCp
The residue obtained from the procedure above in the synthesis of 5 was washed with 10 ml 1:1 (v:v diethylether: hexane) and dried in vacuo to yield (8

General CO 2 Hydrogenation Reaction
Into a stainless steel reactor, 1 µmol of an appropriate catalyst was dissolved in 10 ml of appropriate solvent. 1.44 mmol of the appropriate base was added and the resulting solution was pressurized with the required CO 2 :H 2 ratios. The reaction mixture was then stirred at the appropriate temperature for the required time at 1,500 rpms. After the reaction time had elapsed, the reactor was cooled and the gasses slowly vented. 0.4 ml of the contents of the reactor was sampled and added to 0.2 ml of D 2 O and 5 µl of DMF (as standard). Amounts of product (formate) obtained was calculated by integral relations between singlet formate peak at 8.3 ppm and the singlet DMF formamide peak at 7.9 ppm in the 1 H NMR spectrum of the sample.

SYNTHESIS AND CHARACTERIZATION OF TETRAZOLE-BASED LIGANDS 1-3
Synthesis of three thio-tetrazole compounds 1-3 (Scheme 1) was achieved by first reacting 1-phenyl-1H-tetrazole-5-thiol with a slight excess of NaOMe, to deprotonate the hydrogen in the thiol functional group, followed by the addition of the appropriate chlorodialkylphosphine and subsequent separation of NaCl. All three compounds were isolated in excellent yields as colorless oil for 2, and as whitish solids for 1 and 3 that are air and moisture sensitive. They are soluble in most organic solvents (such as chloroform and dichloromethane) and were characterized by NMR spectroscopy. The easiest way to monitor this reaction was by 31 P{ 1 H} NMR spectroscopy. The phosphorus peak for chlorodiphenylphosphine is around 85 ppm, while the peaks for compounds 1, 2, and 3 are 22.9 ppm, 18.0 ppm, and 13.3 ppm, respectively. To confirm that compounds 1-3 were unoxidized, we prepared the oxidized forms of 1-3, which have phosphorus peaks at 29 ppm, 31.0 ppm, and 25 ppm, respectively. Mass spectra of the ligands also confirmed their formation for which molecular ions or masses of fragmented moieties are observed (Supplementary Figures 15-17).

SYNTHESIS AND CHARACTERIZATION OF TETRAZOLYL IRIDIUM COMPLEXES
Reacting compounds 1-3 with [Ir(C 5 Me 5 )Cl 2 ] 2 in a 2:1 ratio yielded interestingly complexes 4-9 (Scheme 2), all of which contain portions of the original compounds used to ligate (C 5 Me 5 )Ir fragment, thus suggesting that the original compounds expected to be the ligand in a product had hydrolyzed. For example, monitoring the reaction between compound 2 and 0.5 eq [Ir(C 5 Me 5 )Cl 2 ] 2 by 31 P{ 1 H} NMR spectroscopy showed that the reaction yielded Iridium 5 within 1 h, and only after 4 h was 8 observed as a second product (Figure 1). It was only after 24 h that the reaction mixture showed complexes 5 and 8 formed in nearly the same amounts, clearly demonstrating that the rate of formation of 8 is much slower than that of 5. All iridium complexes were obtained as yellow solids in excellent yields and readily soluble dichloromethane and chloroform, but some of these compounds are insoluble in methanol. The structures of these complexes were confirmed by mass spectrometry for which peaks pertaining to [M-Cl] + in addition to the corresponding isotopic peaks are observed. For example, complex 5 is soluble in methanol while complex 8 is insoluble in methanol. Therefore, carrying out the complexation reaction in methanol provides an easy route to separate the different iridium complexes that are formed.
To confirm that the products from the reactions of 1-3 with [Ir(C 5 Me 5 )Cl 2 ] 2 are the results of a water-assisted hydrolysis SCHEME 1 | Synthesis of thio-tetrazole ligands 1-3. SCHEME 2 | Synthesis of tetrazolyl iridium complexes. of the P-S bonds present in the ligands, the progress of the reaction between 2 and [Ir(C 5 Me 5 )Cl 2 ] 2 was monitored in situ with 31 P{ 1 H} NMR spectroscopy in D 2 O. We observed distinct 31 P{ 1 H} NMR peaks that can be attributed to deuterated and non-deuterated analogs of 5 and 8 (Supplementary Figure 1).
Interestingly, there is very little oxidation of the phosphorus atom in 2 as observed from the in situ 31 P{ 1 H} NMR studies in the presence of D 2 O (Supplementary Figure 1). This indicates that the presence of adventitious water in these reactions only hydrolyzes the P-S bond in compound 2 leading to the formation of 5 and subsequently the reaction of 5 with in situ generated 1phenyl-1H-tetrazole-5-thiol to form 8. This reaction route was also confirmed by the reaction of isolated 5 with phenyl-1Htetrazole-5-thiol that forms 8 in <5 min.
Changing the metal precursor from [Ir(C 5 Me 5 )Cl 2 ] 2 to [Ir(C 5 Me 5 )(H 2 O) 3 ][SO 4 ] 2− in the reaction with 2 resulted in the formation of new species where 13 P{ 1 H} NMR peaks at 77 ppm and 57 ppm are observed. Considering 5 and 8 have 13 P{ 1 H} NMR peaks around 77 ppm and 57 ppm, respectively, we propose the formation of aqua/sulfato-5 at 77 ppm and aqua/sulfato-8 at 57 ppm where the chloride ligands in 5 and 8 are substituted with aqua or sulfato ligands in each complex. Observing similar P-S heterolysis with different iridium precursors indicated that P-S heterolysis may also be assisted by the presence of the metal. Similar metal-assisted bond hydrolysis has been reported in literature (Ahmad et al., 2019). In order to confirm what effects substituents on the phenyl groups bonded to the phosphorus atom have on coordination mode for the monodentate Iridium(III) complexes, two new ligands with electron-withdrawing (3) and electron-donating groups (1) on the phenyl groups were reacted with [Ir(C 5 Me 5 )Cl 2 ] 2 (Scheme 2).
The progress of each reaction was monitored by 31 P{ 1 H} NMR spectroscopy. Figure 2 shows the changes that occurred in the course of the reaction between 3 and [Ir(C 5 Me 5 )Cl 2 ] 2 , where a new Iridium phosphorus intermediate (6) with a peak around 65 ppm was converted in less than 2 h to complex 9 with a peak 55 ppm. However, in the case of the reaction of ligand 1 with [Ir(C 5 Me 5 )Cl 2 ] 2, even after heating the reaction mixture at 50 • C for 24 h, we observed various phosphine species (Supplementary Figure 2). The phosphorus peak at 77 ppm was assigned to 4, the peak at 58 ppm was assigned to 7, but the peak at 33.8 ppm remains unidentified. We were, however, unable to separate the complexes in the reaction mixture as they are all soluble in similar solvents. It is, nevertheless, evident that ligands 1, 2, and 3 react with [Ir(C 5 Me 5 )Cl 2 ] 2 in the following order: ligand with electron-withdrawing substituent > ligand with no substituent > ligand with electron-donating substituent. This indicates that the electron-withdrawing ability of substituents attached to the phosphorus atom plays an important role in the heterolysis and in subsequent stabilization of iridiumphosphorus in bonds 7, 8, and 9. The latter can be explained by the fact that the π-acid strength of the phosphorus atom follows the order 3,5-(CH 3 ) 2 -C 6 H 3 )P(OH) < (C 6 H 5 )P(OH) < 3,5-(CF 3 ) 2 -C 6 H 3 )P(OH). The electron-withdrawing (CF 3 ) group concentrates electron density away from the phosphorus atom, which favors increased π back-donation to the phosphorus, thereby reinforcing the Ir-P bond more in 9 than in 8 and 7.

SINGLE CRYSTAL X-RAY STRUCTURES OF COMPLEX 8
Enantiopure crystals for 8 were grown by slow evaporation from a solution of 8 in CHCl 3 , and the crystallographic data of this complex are shown in Supplementary Table 2. Complex 8 crystallizes in the monoclinic crystal system with space group P2 1 /n with piano-stool geometry around the iridium center. The iridium center is an iridium(III) center with a negatively charged sulfur moiety as ligand. Interestingly, hydrogen bonding is observed between the hydroxyl protons bonded to the phosphorus atom in complex 8 and one of the nitrogen atoms in the tetrazole ring (Figure 3).
It is interesting to note that re-growing crystals of 8 results in racemic co-crystallization of enantiomers of 8 that have chirality at the metal center in the triclinic space group P-1. As expected, the complexes have piano-stool geometry around both iridium centers with a negatively charged sulfur containing moiety as ligand (Figure 4). Similarly, hydrogen bonding is FIGURE 2 | Reaction between 3 and [Ir(C 5 Me 5 )Cl 2 ] 2 monitored by 31 P{ 1 H} NMR spectroscopy. observed between the hydroxyl protons and a nitrogen atom in the tetrazole ring. Supplementary Tables 1, 2 show selected bond lengths and angles and crystallographic data, respectively, for enantiopure complex 8 and racemic complex 8. The bond angles and distances obtained for racemic and enantiomer 8 are in similar ranges for reported Ir-S or Ir-P complexes (Hughes

EVALUATION OF CHIRAL-AT-METAL IRIDIUM COMPLEXES IN CO 2 HYDROGENATION
Using either ethanol or DMSO as solvents for CO 2 hydrogenation, there was only trace amounts of formate detected in the 1 H NMR spectrum of the reaction mixture using 5, 8, or 9 (the product obtained is a DBU salt of formic acid and was not separated). However, changing the solvent to water resulted in the formation of a modest amount of formate (0.027 mmol) using 8. This increase in TONs is probably because water has been shown to enhance the rate of CO 2 hydrogenation possibly due to hydrogen-bonding interaction between water and an oxygen atom in CO 2 during the insertion of CO 2 in the catalytically active metal hydride species (Suna et al., 2017). However, adding THF resulted in an eight-fold increase in TON that indicates the importance of THF as solvent for the reaction. Increasing the temperature of the reaction also resulted in an increase in the amount of formate formed. At 160 • C, 0.656 mmol of formate was produced compared to 0.197 at 90 • C ( Table 1). Varying the ratios of both CO 2 and H 2 gave the ideal ratio of CO 2 :H 2 at 40 bar. Changing the ratio of H 2 :CO 2 from 3:1 to 1:1 generally resulted in an improvement in TONs. This suggests that the rates of formation of the catalytically active iridium hydride species and the insertion of CO 2 into the iridium-hydride bond proceed at similar rates. Furthermore, addition of different bases affect the TONs of the reaction ( Table 1). The best base for the reaction is DBU (Table 1, entries 11 and 12). We also observed that the amount of product formed is higher than the amount of base used. This is possible because 1 eq of DBU is known to stabilize multiple equivalences of formate via homoconjugation during the CO 2 hydrogenation reaction (Munshi et al., 2002;Getty et al., 2009;Jeletic et al., 2014;Zhang et al., 2015).
Reducing the amount of catalyst used and reaction time resulted in lower amounts of formate produced. As such, the optimum reaction conditions for catalyst 8 in CO 2 hydrogenation were as follows: 10 ml of a 4:1 mixture of H 2 O:THF, temperature of 160 • C, 1:1 mixture of CO 2 :H 2 at 40 bar, DBU as base, 1 µmol catalyst loading, a 24-h reaction time, and stirring rate of 1,500 rpm. The progress of the reaction could be followed by sampling the reaction mixture after 24 h and running the 1 H NMR (Supplementary Figure 3) and the 13 C{ 1 H} NMR (Supplementary Figure 4) spectra. The NMR spectra indicated the formation of bicarbonate ion during the course of the CO 2 hydrogenation, an observation that has also been reported in the literature as a product from CO 2 and wet DBU (Heldebrant et al., 2005). Hydrogenation of NaHCO 3 using 8 also resulted in the production of formate ( Table 2). This indicates that the hydrogenation of bicarbonate, in addition to CO 2 , is an additional viable route to the production of formate in CO 2 hydrogenation under the conditions used in our study. Comparing other iridium half-sandwich catalysts with P-OH groups to catalyst 8 (TON 1 561) for CO 2 hydrogenation, it is evident that catalyst 5 (TON 1 878) performs better than catalyst 8 ( Table 2). The performance of these two catalysts indicates that electron-donating substituents on the tetrazolyl results in lower catalytic activity. Nevertheless, it is also possible that increased catalytic activity could be the result of the ease with which chlorides dissociate from the iridium center and the subsequent iridium dihydride species formation. Such secondary hydrido species formation for 8 would be more unfavorable for the following reasons. Since 8 exists as an enantiomer, the P-OH group could play the role of a proton-responsive group, where the enantiomer of 8 with P-OH and chloride ligands oriented away from each other would require a significant amounts of energy in forming the iridium hydride species, but the other enantiomer with both P-OH and Cl groups oriented toward each other would require considerably less energy. Interestingly, catalyst 9, which has a more electron-deficient iridium center, because of the electron-withdrawing nature of the CF 3 substituents, also has a similar TON (1,391) to that of 8. This suggests that the CF 3 groups are too far from the metal center to significantly influence catalytic activity. This assertion is further supported by 31 P{ 1 H} NMR peaks for both 8 and 9 around 55 ppm, in contrast to more electron-deficient phosphine metal complex peaks that are expected more downfield. When the catalytic activity of 5-9 are compared to other highly active proton-responsive iridium(III) half-sandwich precatalysts for CO 2 hydrogenation (Siek et al., 2017;Suna et al., 2017;Gunasekar et al., 2018), it is evident that these chiral-at-metal catalysts with proton-responsive P-OH groups are equally as active for CO 2 hydrogenation as shown by their high TONs. For example, more concentrated formate solutions (around 180 mM) are produced from 5, 8, and 9 catalyzed  (Suna et al., 2017) pyrazol-3-yl) (Siek et al., 2017). Reactions were performed in 25 ml of an aqueous solution of 0.3 mM catalyst and 1 M NaHCO 3 at 115 • C and 300 psig of H 2 /CO 2 (1:1) for 18 h.
CO 2 hydrogenation in our study compared to 26-91 mM for N ∧ N bidentate iridium half-sandwich catalysts reported in literature. These concentrated formate solutions are obtained due to higher temperatures and pressures used. However, these catalysts do not require highly basic solutions and/or significantly extended reaction times in order to obtain similar formate solutions.

MECHANISM OF ACTION OF CHIRAL-AT-METAL COMPLEXES FOR CO 2 HYDROGENATION
A mercury drop test indicated that CO 2 was hydrogenated via a homogeneous/cocktail of catalysts approach ( Table 2). As such, plausible reaction mechanisms for the CO 2 hydrogenation reaction involving the chiral-at-metal Iridium complexes were investigated. In the presence of excess base and water, complex 8 would be expected to undergo an aquation reaction, with the chloro ligand being replaced with a water ligand while deprotonating the OH group on the phosphorus donor occurs to yield A (Scheme 3). In situ 1 H NMR experiments involving the reaction of complex 8 with two equivalence of DBU demonstrated the disappearance of the hydroxyl protons at 11.6 ppm indicating deprotonation (Supplementary Figure 5). This indicated that during the hydrogenation of CO 2 , the P-OH group could possibly function as a proton-responsive group to assist in the heterolytic cleavage of H 2 .
Pressurizing 8 in a THF/water mixture with two equivalence of DBU and 5 bar of D 2 at 100 • C for 1 h resulted in the SCHEME 3 | Plausible reaction mechanism for chiral-at-metal Iridium complexes in CO 2 hydrogenation. (A) Deprotonated iridium-aqua complex, (B) Iridium-hydride complex, (C) Deprotonated iridium-hydride species, (D) Iridium-carbonato species, and (E) Iridium-formato species.
appearance of doublet hydride peaks around −15.5 ppm as well as corresponding shifts in peaks in the aromatic region ( Figure 5). This was due to the exchange of hydrogen and deuterium present in solution yielding proton signals in the 1 H NMR spectrum. The observed splitting of hydride signal was as a result of the proximity of the NMR active phosphorus atom. Similar splitting for such hydride protons have also been observed in literature (Feller et al., 2016). In addition, there was a corresponding shift in the 31 P{ 1 H} NMR from around 57 ppm to 72 ppm upon formation of the iridium hydride species (Supplementary Figure 6).
Due to the presence of excess base in solution, the iridium hydride species could be expected to exist as C (Scheme 3) since the proton on the P-OH group is deprotonated. Subsequently, insertion of in situ generated bicarbonate ions into the iridiumhydride bond yields intermediate D, which, upon dissociation of a hydroxyl group, yields the iridium-formato species E. However, direct insertion of CO 2 into the metal-hydride bond is also possible during the formation of the iridium-formato species as a result of the observation of formate when other non-carbonate bases are used for the reaction. Introducing CO 2 to a solution of C results in the disappearance of the hydride signal. Such direct CO 2 insertion in metal-hydride bonds have been reported in mechanisms involving interactions between CO 2 and the metal-hydride bonds prior to the formation of the metal-formato species (Wang et al., 2012;Suna et al., 2017). The final step of the reaction is then reductive elimination of formate using the base in order to regenerate species A (Scheme 3).

CONCLUSION
Unintended iridium complexes (4-9) were obtained as a result of the heterolysis of P-S tetrazolyl ligands (1-3) during complexation with [Ir(C 5 Me 5 )Cl 2 ] 2 . The heterolysis is ascribed to the presence of water and electron-withdrawing groups on the phosphorus moiety, which were found to facilitate the heterolysis of the P-S bond in the ligand during complexation with [Ir(C 5 Me 5 )Cl 2 ] 2 . The iridium complexes hydrogenate CO 2 as well as NaHCO 3 in a 4:1 water:THF solvent mixture, and DBU as abase. A key factor in the activity of these iridium complexes as excellent CO 2 hydrogenation catalysts is the ability to be proton responsive. We observed such proton-responsive enhanced catalytic ability from chiral-at-the-metal complexes that have the P-OH proton-responsive functional groups. Complex 8 was used to establish that the mechanism of the CO 2 hydrogenation reaction for the iridium complexes proceeded with the iridium precatalysts first undergoing aquation, followed by H 2 heterolysis that is driven by proton-responsive P-OH groups to form an iridium hydride intermediate. This iridium hydride species is the catalytically active species that hydrogenates either bicarbonate species or CO 2 to yield the formate product.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. The crystallographic datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: Cambridge Crystallographic Data Center, CCDC numbers 1840199 and 1993265.
Electronic supporting information (ESI) is available free of charge. CCDC number 1840199 contains the supplementary crystallographic data for racemic complex 8. CCDC number 1993265 contains the supplementary crystallographic data for enantiopure complex 8. These data can be obtained free of charge from the Cambridge Crystallographic Data Center from the Director, CCDC, 12 Union Road, Cambridge, CB2 1EZ, UK (fax: +44 1223 336063; deposit@ccdc.cam.ac.uk or https://www.ccdc.cam.ac. uk/structures/).