Single-Site Iridium Picolinamide Catalyst Immobilized onto Silica for the Hydrogenation of CO2 and the Dehydrogenation of Formic Acid.

The development of an efficient heterogeneous catalyst for storing H2 into CO2 and releasing it from the produced formic acid, when needed, is a crucial target for overcoming some intrinsic criticalities of green hydrogen exploitation, such as high flammability, low density, and handling. Herein, we report an efficient heterogeneous catalyst for both reactions prepared by immobilizing a molecular iridium organometallic catalyst onto a high-surface mesoporous silica, through a sol-gel methodology. The presence of tailored single-metal catalytic sites, derived by a suitable choice of ligands with desired steric and electronic characteristics, in combination with optimized support features, makes the immobilized catalyst highly active. Furthermore, the information derived from multinuclear DNP-enhanced NMR spectroscopy, elemental analysis, and Ir L3-edge XAS indicates the formation of cationic iridium sites. It is quite remarkable to note that the immobilized catalyst shows essentially the same catalytic activity as its molecular analogue in the hydrogenation of CO2. In the reverse reaction of HCOOH dehydrogenation, it is approximately twice less active but has no induction period.

Introduction

Energy transition from fossil to renewable fuels is the ultimate challenge of our society which might be successfully faced through the implementation of an efficient technology for producing “green” hydrogen, i.e., hydrogen derived from the photoelectrocatalytic splitting of water.[1−6] H2 is surely a clean primary energy carrier having, nevertheless, with its high flammability, low density, and handling, some serious criticalities.[7,8]

An interesting alternative to using H2 as such is to store it in the so-called liquid/solid hydrogen carriers and regenerate it when needed. One of the most promising H2 carriers is formic acid (FA),[9−12] which can be generated by storing H2 into CO2, from which H2 might be reformed by the reverse reaction.

Both forward and back reactions ask for suitable catalysts in order to be viable, and several efficient catalysts have been reported so far.[9,10,12] Supported nanoparticles of noble metals such as Pd and AuPd[13−17] and, especially, organometallic complexes,[18−20] mainly based on Fe,[21−31] Ru,[32−40] and Ir,[41−44] have shown promising catalytic performances in terms of both turnover numbers (TONs) and turnover frequencies (TOFs). Notably, Hazari and Bernskoetter reported a class of iron complexes (a and b in Scheme ) capable of reaching very high values of TON (ca. 106 in FA dehydrogenation and 6 × 104 in CO2 hydrogenation).[21,23,31] Unfortunately, the conditions involve organic solvents and the presence of Lewis acids. Williams reported an IrI-based complex (c in Scheme )[42] with remarkable performance in FA dehydrogenation (TON value over 2 × 106 with a maximum TOF of 3.7 s–1) carried out in neat FA, whereas Nozaki developed an IrIII pincer trihydride catalyst (complex d in Scheme )[43,45] very active in CO2 hydrogenation (TON = 3.5 × 106 and TOF = 1.5 × 105 h–1), operating in a basic aqueous solution, capable of catalyzing also FA dehydrogenation in BuOH, in the presence of NEt3.

Scheme 1

Relevant Organometallic Catalysts for FA Dehydrogenation and CO2 Hydrogenation

Cp*Ir-based (Cp* = pentamethylcyclopentadienyl) complexes have been successfully exploited as efficient catalysts for both CO2 hydrogenation and FA dehydrogenation in water, under mild conditions, without using any additive, except a base in the case of CO2 hydrogenation.[46−51]

Remarkable results have been obtained by Himeda, Fujita, and co-workers using [Cp*Ir(R-pica)X] (pica = picolinamide) complexes (e and f in Scheme )[52,53] in both FA dehydrogenation[52,54,55] and CO2 hydrogenation.[53,56−58] Originally introduced by Watanabe and co-workers for the preparation of amine compounds,[59] [Cp*Ir(R-pica)X] complexes are a successful class of compounds that have been found to efficiently catalyze many other reactions including transfer hydrogenation in cell growth media,[60,61] reductive amination of ketones,[62−64] water oxidation,[65] NADH regeneration,[65,66] hydrogenolysis of halosilanes,[67] and hydrogen peroxide generation.[68] They also exhibited good performance as antimicrobial, antibacterial, and anticancer agents.[69−72]

Considering the versatility and success of [Cp*Ir(R-pica)X] complexes as catalysts, we decided to immobilize them into mesoporous silica, aiming at preparing a hybrid single-site organometallic heterogeneous catalyst, having the distinctive features of the analogous molecular catalysts, adding all the advantages of heterogeneous catalysts in terms of recoverability and process intensification.[73−76] Mesoporous silica was selected as support since it is inert and inexpensive, easily recoverable from the reaction mixtures, and characterized by a high surface area, thus facilitating the exposure of the catalytic species to the reactants and maximizing the exploitation of the noble metal.[77]

Herein, we report the synthesis of a heterogeneous immobilized catalyst (Ir_PicaSi_SiO) and its successful application in CO2 hydrogenation and FA dehydrogenation. The preparation of Ir_PicaSi_SiO involved the initial synthesis of a modified version of the pica ligand (PicaSi), in which R is the (3-triethoxysilyl)propyl moiety, and the immobilization of PicaSi onto mesoporous silica (PicaSi_SiO) via a sol–gel process, which was recently reported as an effective strategy to provide homogeneously distributed ligands on high surface area materials.[78−80] The catalytic iridium-single site was then implanted by the reaction of PicaSi_SiO with [Cp*IrCl2]2 (Ir_PicaSi_SiO) (Scheme ).

Scheme 2

Steps of the Synthesis of Ir_PicaSi_SiO

To obtain a deeper molecular-level understanding of the surface, both materials were investigated via solid-state NMR spectroscopy and XAS. In order to increase the sensitivity of NMR toward surface sites, the dynamic nuclear polarization surface-enhanced NMR spectroscopy (DNP-SENS) approach was exploited. This technique allows the increase of NMR sensitivity by up to 2 orders of magnitude and thereby the recording of natural abundance 13C, 15N, and 29Si solid-state NMR spectra in a reasonable acquisition time.[81−85]Ir_PicaSi_SiO exhibited remarkable catalytic performances in aqueous FA dehydrogenation and CO2 hydrogenation that compare well with those of the analogous molecular catalysts. Extensive kinetic studies revealed that the reaction pathway for supported catalysts differs from the analogous molecular systems, because of the formation of cationic sites stabilized by the surface.

Results and Discussion

We will first discuss the preparation and comprehensive characterization of Ir_PicaSi_SiO (Section ) and then its application as a catalyst (Section ). The first section provides evidence regarding the nature of the surface sites thanks to the use of state-of-the-art characterization, while the second section focuses on the performance of the catalyst and includes detailed catalytic and kinetic tests in both the CO2 hydrogenation and the reverse reaction, FA dehydrogenation.

Preparation and Characterization of Ir_PicaSi_SiO

The hybrid material Ir_PicaSi_SiO was prepared by a two-step procedure involving 1a) the synthesis and immobilization onto mesoporous SiO2 of a pica-modified ligand (PicaSi_SiO) via a sol–gel process and 1b) a postfunctionalization to incorporate the iridium organometallic moiety (Ir_PicaSi_SiO), taking advantage of the coordination ability of PicaSi_SiO (Scheme ).

Synthesis and Heterogenization of PicaSi onto SiO2

A modified version of the picolinamide ligand (PicaSi, Scheme ) having a (3-triethoxysilyl)propyl group on the nitrogen atom of the amide moiety was prepared, following reported procedures (SI).[86,87]PicaSi was successively immobilized via a sol–gel procedure onto commercially available mesoporous silica beads (PicaSi_SiO), that are easy to handle and to recover due to a convenient particle diameter (from 60 to 200 μm). This support is also characterized by high surface area (738 m2 g–1), which is ideal for catalysis.[79,80]

Specifically, TEOS (tetraethyl orthosilicate), PicaSi, and silica beads were contacted in an acidified solution in THF, at 343 K for 1 h. The collected solid was washed with water/THF, ethanol, and diethyl ether and dried at 408 K under high vacuum. The thus-obtained material (PicaSi_SiO) was characterized by means of IR spectroscopy, low-temperature N2 adsorption (BET), elemental analysis (EA), and DNP-SENS. In order to assign the NMR resonances, DFT calculations were also conducted.

The Diffuse Reflectance Infrared Fourier Transform (DRIFT) spectrum of PicaSi_SiO (Figure S1, SI) displays all the characteristic bands of the picolinamide moiety, namely the C=O stretching (1670 cm–1) of the amide and the C=C stretching (1545 cm–1) of the aromatic carbons of the pyridine ring as well as the classic aromatic (3065 cm–1) and aliphatic CH stretching (2964 and 2887 cm–1). EA confirms the presence of nitrogen and carbon on the solid (SI). The incorporation of the ligand onto the surface caused a reduction of the surface area from 738 m2 g–1 (starting mesoporous silica) to 650 m2 g–1 (PicaSi_SiO).

In order to further understand the structure of the ligand on the surface, PicaSi_SiO was characterized via DNP-SENS. Using a typical sample formulation (PicaSi_SiO with an aqueous solution of AMUPol biradical to hyperpolarize the surface)[88,89] and DNP-SENS acquisition conditions (100 K, constant microwave (MW) irradiation and cross-polarization DNP cross effect),[85] we recorded spectra with DNP solvent enhancement of the NMR sensitivity reaching 120. The DNP-enhanced 13C cross-polarization magic angle spinning (CPMAS) NMR spectrum exhibits (i) three different aliphatic resonances (at 43, 23, and 9 ppm) that can be assigned to the three CH2 of the propyl moiety that bonds the ligand onto the surface, (ii) five resonances in the aromatic region, and (iii) a resonance centered at 168 ppm consistent with the quaternary carbon of the amide moiety (Figure a). Interestingly, the pattern of resonances of the 13C NMR spectrum of PicaSi_SiO matches that of the PicaSi ligand in CD2Cl2 solution (Figure S3), indicating structural similarities between the immobilized and molecular organic moieties.

Figure 1

DNP-enhanced MAS NMR spectra of PicaSi_SiO and Ir_PicaSi_SiO materials. a) 13C CPMAS NMR spectrum of PicaSi_SiO (MAS 10 kHz, 100 K), b) 13C CPMAS NMR spectrum of Ir_PicaSi_SiO (MAS 10 kHz, 100 K), c) {1H}13C HETCOR NMR spectrum of Ir_PicaSi_SiO (MAS 10 kHz, 100 K, black = positive, red = negative), d) 15N CPMAS NMR spectrum of PicaSi_SiO (MAS 10 kHz, 100 K), e) 15N CPMAS NMR spectrum of Ir_PicaSi_SiO (MAS 10 kHz, 100 K), and f) 29Si CPMAS NMR spectrum of PicaSi_SiO (MAS 10 kHz, 100 K). Asterisks mark spinning sidebands.

To determine the binding mode of the ligand to the surface, the 29Si DNP-SENS NMR spectrum of PicaSi_SiO was also recorded. In addition to the classical resonances of the amorphous silica at ca. −90 to −110 ppm, two peaks centered at −55 and −63 ppm are observed (Figure f), showing that the ligand is bound to the surface of the material predominantly in T2 and T3 fashion.[90] This result is in agreement with the findings of DFT computational studies conducted for various binding modes of the ligand to the SiO2 surface (summarized in Table S2; detailed descriptions are in the SI).

To further understand the structure of the immobilized ligand, 15N NMR spectroscopy was also carried out. DNP-enhanced spectroscopy was used in order to enable such studies at 15N natural abundance. The spectrum (Figure d) shows a resonance at 122 ppm assigned to the amidic nitrogen, while the resonance of the pyridinic nitrogen was not detected. The absence of the latter is likely due to the larger chemical shift anisotropy of the signal that decreases sensitivity.[91]

Overall, all pieces of information confirm that PicaSi is present on the surface of the mesoporous SiO2, and particularly, DFT calculations in combination with 29Si DNP-SENS allow establishing that the ligand is linked to the silica surface through the silane moiety predominantly in T2 and T3 fashion.

Immobilization of an Iridium Organometallic Moiety

The immobilization of Ir was next carried out by reacting a solution of [Cp*IrCl2]2 with PicaSi_SiO in anhydrous dichloromethane, under inert atmosphere, for 48 h at RT, in the presence of a base (NEt3) (SI). The resulting material Ir_PicaSi_SiO shows a small decrease of surface area from 650 m2 g–1 (PicaSi_SiO) to 595 m2 g–1 (Ir_PicaSi_SiO). The C:N:Ir:Cl experimental ratio in Ir_PicaSi_SiO (19.4:2.0:1.0:0.3), obtained by EA, is close to the expected one (19.0:2.0:1.0:1.0), albeit with a slight chlorine deficiency that could be due to a partial replacement of Cl– by OH– or HSO4– (SI, page S7). Based on the EA and BET measurements, the surface density of Ir is ≈1 per 10 nm2. This result suggests that organometallic sites are well separated from each other, decreasing the probability of having multimetallic processes during catalytic reactions.

The DRIFT spectrum of Ir_PicaSi_SiO (Figure S4, SI) shows all the expected bands of the organic moiety on the surface with appreciable shifts with respect to the spectrum of PicaSi_SiO consistent with the binding of iridium on the immobilized ligand. In particular, C=O and C=C stretchings shift from 1670 to 1625 cm–1 and from 1545 to 1595 cm–1, respectively.

The DNP-enhanced 13C NMR spectrum shows all resonances of the Cp* and pica moieties confirming the presence of the iridium complex onto the material (Figure b). Interestingly, the resonance belonging to the quaternary carbon of the amide moiety shifts at an appreciably higher frequency, from 168 to 173 ppm, consistent with the coordination of the metal center to the immobilized organic ligand. The dipolar coupling between the resonances of pica and Cp*, observed in the DNP-enhanced 2D {1H}13C HETCOR NMR spectrum, further confirms the spatial proximity of the two moieties (Figure c). Additionally, the 15N chemical shift of the amide moiety decreased from 122 to 57 ppm (Figure e), pointing to the coordination of the ligand to iridium.

In order to gain further information about the local surrounding of Ir in Ir_PicaSi_SiO, X-ray absorption spectroscopy (XAS) studies were performed at the iridium L3-edge. X-ray absorption near-edge structure (XANES, Figure ) edge energies of Ir_PicaSi_SiO and the molecular analog [(Cp*)Ir(N-propyl-pica)Cl] (1) (1 is a new complex, and its synthesis and characterization are reported in the SI from page S11 to page S13) are identical and equal to 11214.75 eV (Table S3, entries 1 and 2) and very close to that of Ir(acac)3 (acac = acetylacetonate) and [Cp*IrCl2]2 (2) (11214.75 and 11214.5 eV, respectively). Notably, for IrCl3, the edge energy (11213.75 eV) is shifted to a lower value than the other references found in the +3 oxidation state, likely as a result of the decreased covalency of the Ir–Cl bonds. Further comparison of the edge energy to reference compounds in the +1 (Ir(cod)(acac), 11215.0 eV, cod = 1,5-cyclooctadiene) and +4 (Na2IrCl6, 11213.75 eV) oxidation states suggests that the average oxidation state is most likely +3 for the supported system. Comparison of the white line intensity for the series of compounds measured indicates that, as expected, a more intense white line feature corresponds to higher average oxidation states (Table S3), while 2 possesses an anomalously high white line intensity (2.80).

Figure 2

XAS data of Ir_PicaSi_SiO and selected reference compounds. a) Normalized Ir L3 edge XANES and b) the first derivative of Ir L3 edge XANES. From bottom to top: Ir(COD)(acac) (dark gray, dash), Ir(acac)3 (gray, dot), Na2IrCl6 (light gray, short dash), 2 (dark blue), IrCl3 (red), 1 (purple), and Ir_PicaSi_SiO (turquoise). c) k2-weighted R-space EXAFS data (turquoise) and fitting (gray) with the corresponding fitting parameters summarized in the table (for full details of fit see Table S5, SI).

Subsequently, analysis of the extended X-ray absorption fine structure (EXAFS) was performed to ascertain the identity of the nearest neighbors of Ir in the supported system (Ir_PicaSi_SiO). It is noteworthy that EXAFS provides information related to the average coordination of Ir in the material, rather than a specific coordination geometry, or indeed a single, unified coordination environment.[92,93]

Fitting of the k2-weighted EXAFS for the supported catalyst Ir_PicaSi_SiO is consistent with a material containing Ir sites with C/N/O scattering paths (8.1 ± 0.8) and Cl scattering paths (1.2 ± 0.4) (Figure S14, parameters summarized in Table S5). Since the light atoms (C/N/O) are similar in mass, they cannot be easily distinguished by EXAFS.[94,95] As such, the C/N/O scattering paths are separated into two groups for fitting–those arising from the Cp* ligand, which have constrained path degeneracy (N = 5, R = 2.14 ± 0.04 Å), and those assigned to the picolinamide moiety and C,N,O-based ligands (H2O, SO42–) (N = 3.1 ± 0.8, R = 2.08 ± 0.03 Å), giving a total C/N/O coordination of (8.1 ± 0.8). To obtain a reasonable fit, it was necessary to also include an Ir–Cl path, the degeneracy of which was found to be (N = 1.2 ± 0.4, R = 2.34 ± 0.02 Å) from the fit obtained. The presence of an Ir–Cl path is consistent with the data from elemental analysis, which suggests that some chloride is retained after synthesis.

For Ir_PicaSi_SiO, the combination of elemental analysis and the assignment made on the basis of the obtained EXAFS fits suggests that there are two distinct Ir species present in the material, both of which contain both Cp* and picolinamide moieties bound to Ir, as well as an additional ligand which can be either a chloride anion or a water molecule.

In order to address this issue further, reaction of Ir_PicaSi_SiO with 15N-pyridine (Py) was carried out, and the recovered material was studied by means of DNP-enhanced 15N CPMAS NMR (Figure S13, SI).[96] Two resonances were observed at 219 and 287 ppm corresponding to the interaction of Py with iridium and silanol groups, respectively. Whereas the latter one is commonly observed in silica-supported materials,[97] the presence of the former interaction indeed further suggests that iridium sites in Ir_PicaSi_SiO have a cationic character.[98]

The general picture emerging from such an in-depth characterization indicates that Ir_PicaSi_SiO is a hybrid material in which rather dispersed cationic iridium sites are coordinated at PicaSi moieties anchored onto the silica surface.

Catalytic Applications of Ir_PicaSi_SiO

FA Dehydrogenation

Ir_PicaSi_SiO was tested as a catalyst in the FA dehydrogenation to CO2 and H2 at different pH values (1.4–8.2 range, entries 1–5, Table ), catalyst concentrations (25–500 μM range, entries 6–11, Table ), [HCOOH]+[HCOO–] concentrations (0.5–5 M range, entries 3, 9, and 12–14, Table ), and temperatures (288–353 K range, entries 7 and 15–18, Table ). The progress of the reactions was followed by means of differential manometry and solution NMR spectroscopy. The experiments were conducted by adding formic acid to a suspension of Ir_PicaSi_SiO in a HCOO– solution. Conversion was evaluated by measuring the amount of residual formic acid/formate via 1H NMR spectroscopy. The complex [Cp*Ir(N-Me-pica)Cl] (3) was used as a literature benchmark.[54,55] All the results are summarized in Table .

Table 1

Summary of the Performances of Ir_PicaSi_SiO in the FA Dehydrogenation Reaction

	[Cat] (μM)	[HCOOH]+[HCOO–] (M)	[HCOOH] (M)	pH	T (K)	TOF (h–1)	TON	convn (%)
1	250	3	3	1.4	298	254	12000	>99
2	250	3	2.8	2.4	298	480	11200	>99
3	250	3	1.5	3.7	298	636	5640	94
4	250	3	0.8	4.2	298	505	3370	>99
5	250	3	0	8.2	298	2	120
6	25	1	0.5	3.7	298	425	17600	88
7	50	1	0.5	3.7	298	448	9200	92
8	100	1	0.5	3.7	298	354	5100	>99
9	250	1	0.5	3.7	298	469	2160	>99
10	350	1	0.5	3.7	298	450	1490	>99
11	500	1	0.5	3.7	298	359	1020	>99
12	250	0.5	0.25	3.7	298	212	1000	>99
13	250	0.75	0.375	3.7	298	235	1410	94
14	250	5	2.5	3.7	298	890	10400	>99
15	50	1	0.5	3.7	288	105	5600	56
16	50	1	0.5	3.7	313	1070	7200	72
17	50	1	0.5	3.7	333	5400	7400	74
18	50	1	0.5	3.7	353	11200	5800	58

TOF vs pH exhibits a volcano-shaped trend (entries 1–5, Table , Figure a). The highest TOF value (636 h–1) was observed around pH = 3.7, corresponding to the pKa of the formic acid, as already reported for other Ir-based catalysts.[54,55] This can be explained considering the simplified reaction mechanism illustrated in Scheme involving two steps. The first step is the activation of the C–H bond of FA, leading to a LIr–H intermediate, generating CO2 and H3O+; in the second step, the protonation of the LIr–H species liberates H2. Indeed, low pH makes the deprotonation of HCOOH and the consequent formation of Ir–H complicated, whereas at higher pH, the protonation of Ir–H and the consequent evolution of H2 become slightly probable.[99]

Figure 3

Kinetic trends of the performances of Ir_PicaSi_SiO in the FA dehydrogenation reaction. a) TOF (h–1) vs pH trend for formic acid dehydrogenation catalyzed by Ir_PicaSi_SiO ([cat] = 250 μM, [HCOOH]+[HCOO–] = 3 M, 298 K). b) log(d(nO2)/dt) vs log[cat] for formic acid dehydrogenation catalyzed by Ir_PicaSi_SiO ([HCOOH]+[HCOO–] = 1 M, pH = 3.7, 298 K). c) log[d(nO2)/dt] vs log([HCOOH]+[HCOO–]) for formic acid dehydrogenation catalyzed by Ir_PicaSi_SiO ([cat] = 250 μM, pH = 3.7, 298 K). d) ln(TOF) vs 1/T for formic acid dehydrogenation catalyzed by Ir_PicaSi_SiO in a temperature range 288–353 K ([cat] = 50 μM, [HCOOH]+[HCOO–] = 1 M, pH = 3.7).

Scheme 3

Simplified Reaction Mechanism of FA Dehydrogenation Mediated by Ir-Based Catalysts

L = ancillary ligand.

TOF values were found to be slightly dependent on the concentration of Ir_PicaSi_SiO (354–469 h–1, entries 6–11, Table , Figure b), indicating a first-order dependence on catalyst concentration. On the other hand, the trend of TOF vs [HCOOH]+[HCOO–] (entries 3, 9, and 12–14, Table , Figure c) suggests a noninteger reaction order (0.6) on FA concentrations. In all cases, quantitative consumption of HCOOH was achieved.

Catalytic tests at different temperatures (288–353 K, entries 7 and 15–18, Table ) show an increase of TOF from 105 h–1 at 288 K to 11200 h–1 at 353 K. The apparent activation energy (Ea = 14 ± 1 kcal mol–1), evaluated from the Arrhenius plot (Figure d), is considerably lower than that of the previously determined value for 3 (Ea = 20 kcal mol–1).[54] This can be explained considering an active role of the Si–OH functionalities, which might facilitate the deprotonation of FA and/or the protonation of Ir–H, under the assumption that the thermodynamics of HCOOH adsorption on the surface is not affecting the apparent activation energy.

The activity and stability of the heterogenized catalyst Ir_PicaSi_SiO were compared to that of 3 by performing catalytic experiments under the same conditions ([Ir] = 25 μM, an equimolar concentration of [HCOOH] and [HCOO–] amounting at 1 M, pH = 3.7 and 298 K, TONexpected = 20000). Gas evolution was monitored by differential manometry for 7 h (Figure a). TON vs time trends were interpolated using a composite mathematical function developed by Peters and Baskin.[100] The first derivative of these “smooth” trends allows obtaining the evolution of TOF versus time (Figure b).

Figure 4

Kinetic trends of the performances of Ir_PicaSi_SiO and 3 in the FA dehydrogenation reaction. a) TON vs t (h) ([HCOOH]+[HCOO–] = 1 M, pH = 3.7, [Cat] = 25 μM, T = 298 K) for 3 and Ir_PicaSi_SiO. b) TOF (h–1) vs t (h) ([HCOOH]+[HCOO–] = 1 M, pH = 3.7, [Cat] = 25 μM, T = 298 K) for 3 and Ir_PicaSi_SiO.

Interestingly, albeit TOFmax of Ir_PicaSi_SiO (303 h–1) is ca. two times slower than that of 3 (620 h–1), at t = 0, the former exhibits a higher TOF (303 h–1) with respect to 3 (248 h–1) (Figure b). The trend of TOF versus time of 3 clearly shows an induction period of ca. 30 min, which is absent in that of Ir_PicaSi_SiO (Figure b). This induction period might be due to the possible formation of poorly active (or inactive), out-of-cycle, dinuclear species.[101] As discussed above, most of the Ir sites in Ir_PicaSi_SiO are well separated (≈1 Ir every 10 nm2), making the associative process unlikely. After 6 h, the activity of Ir_PicaSi_SiO decreased to 231 h–1, whereas that of 3 decreased down to 418 h–1. This decrease of activity is more accentuated than that expected based on the 0.4 and 0.6 dependence of the reaction rate on the concentration of formic acid in homogeneous and heterogeneous catalysis, respectively, and of similar entities, suggesting that catalyst transformation/deactivation processes occur for both catalysts. The main degradative pathway of complex 3 is the reductive deoxygenation of the C=O moiety of the ligand to form the corresponding amino species;[55] the same degradation mechanism might also take place in the case of Ir_PicaSi_SiO. Nevertheless, after 7 days, both Ir_PicaSi_SiO and 3 reached a TON of 17600, over the 20000 expected cycles based on thermodynamics, corresponding to 88% of conversion.

The recoverability of Ir_PicaSi_SiO was evaluated by performing successive tests in which the catalyst was separated and reused in the dehydrogenation of fresh FA aqueous solutions ([HCOOH]+[HCOO–] = 1 M, pH = 3.7, 298 K) (see the SI for experimental details). After each catalytic run, Ir leaching from the solid was determined by means of Inductively Coupled Plasma-Atomic Emission Spectroscopy (ICP-AES) analysis, and the activity of the separated supernatant solution was tested. The activities of Ir_PicaSi_SiO and those of the respective supernatant solutions along with Ir content and leaching for each run are reported in Table S6.

The percentage of Ir leaching in each run, as detected by ICP-AES analysis, is quite low (Ir concentration in the material passed from 0.115 μmol/mg to 0.100 μmol/mg after four cycles, Table S6). The recovered supernatant solution after each run was not active. On the other hand, Ir_PicaSi_SiO can be reused successfully for four successive runs, albeit a progressive decrease (from 385 h–1 to 34 h–1) of TOF is observed (Table S6).

Overall, the supported catalyst Ir_PicaSi_SiO showed remarkable performance in FA dehydrogenation comparable to those of its homogeneous analogue 3 and of the best heterogenized iridium catalysts reported so far. Particularly, Ir_PicaSi_SiO reached a TOF value up to 11200 h–1 ([HCOOH]+[HCOO–] = 1 M, pH = 3.7, [Ir_PicaSi_SiO] = 250 μM, T = 353 K) and a TON value up to 17600 ([HCOOH]+[HCOO–] = 1 M, pH = 3.7, [Ir_PicaSi_SiO] = 25 μM, T = 298 K) in aqueous solution and in the absence of any additives. The catalyst was reused four times, and ICP-AES analysis of the supernatant solutions showed a low Ir leaching for each run. However, from the kinetic data, it is possible to observe a decrease of activity over time consistent with the presence of active catalyst transformation/deactivation processes.[55]

CO2 Hydrogenation

Ir_PicaSi_SiO was also tested as a catalyst for the selective CO2 hydrogenation to formate under batch conditions. Typically, the reaction was carried out for 24 h at 423 K and 50 atm (CO2:H2 = 1:1) in the aqueous solution, in the presence of an organic base. The amount of produced formate was quantified by 1H NMR spectroscopy using 3-trimethylsilylpropanesulfonate sodium salt as the standard; in all the experiments, no other product was observed. Complex 1 was used as the molecular benchmark of the reaction. In the absence of the catalysts, no appreciable formation of formate was observed. The catalytic performances of Ir_PicaSi_SiO were evaluated at different catalyst loadings (0.057–1.16 μmol range, entries 1–7, Table ) and in the presence of different bases (entry 2, Table , and entries S1 and S2, Table S7).

Table 2

Summary of the Performances of Ir_PicaSi_SiO and 1 in the CO2 Hydrogenation Reaction

	cat.	Ir content (μmol)	nFormate (mmol)	[Formate] (M)	base	t (h)	TON
1	Ir_PicaSi_SiO2	0.057	0.4	0.080	DABCO	24	6983
2	Ir_PicaSi_SiO2	0.115	0.67	0.134	DABCO	24	5836
3	Ir_PicaSi_SiO2	0.230	0.88	0.176	DABCO	24	3829
4	Ir_PicaSi_SiO2	0.461	1.07	0.215	DABCO	24	2327
5	Ir_PicaSi_SiO2	0.693	1.19	0.238	DABCO	24	1720
6	Ir_PicaSi_SiO2	0.925	1.16	0.232	DABCO	24	1253
7	Ir_PicaSi_SiO2	1.16	1.36	0.273	DABCO	24	1178
8	1	0.120	0.63	0.126	DABCO	24	5235

Ir_PicaSi_SiO showed comparable performance to its molecular counterpart 1 (entries 2 and 8, Table ). The effect of catalyst loading on the catalytic performances was explored performing the reaction in 1 M 1,4-diazabicyclo[2.2.2]octane (DABCO) aqueous solutions (entries 1–7, Table , Figure ). An increase of formate production was observed up to ca. 4.5 × 10–7 mol of Ir content (Figure ); after that, a plateau is reached indicative of an equilibration between reagents and products, which precluded the possibility of determining a kinetic order on catalyst (Figure ). However, the TON observed with the lowest Ir loading (TON = 7 × 103, entry 1, Table ) is comparable to that of other already reported supported Ir complexes under similar experimental conditions.[48,49,79,80]

Figure 5

Trend of the nFormate vs Ir_PicaSi_SiO content. Experimental conditions: t = 24 h, T = 423 K, P = 50 atm with CO2:H2 = 1:1 ratio, [DABCO] = 1 M.

In summary, Ir_PicaSi_SiO is an active catalyst for the selective hydrogenation of CO2 to formate with catalytic performance comparable to those of the molecular complex 1 and to those of the best heterogenized iridium catalysts reported so far.

Conclusions

A hybrid catalyst consisting of the [Cp*Ir(R-pica)X] complex immobilized onto mesoporous silica (Ir_PicaSi_SiO) was prepared by means of a sol–gel procedure and characterized by a battery of instrumental techniques. The latter allowed the understanding that Ir_PicaSi_SiO maintains a very high surface area (595 m2 g–1) and has a rather dispersed single-site Ir(III) center (≈1 Ir every 10 nm2), still bearing Cp*- and pica-ligands but with a coordination vacancy, generated by the substitution of a chloride ligand by a water molecule or a Si–OH moiety. These features are essential in determining the remarkable catalytic performance of Ir_PicaSi_SiO in both CO2 hydrogenation and FA dehydrogenation. CO2 is hydrogenated to formate with comparable performance to its molecular counterpart, under the same conditions (Table ). Interestingly, a strict comparison of the catalytic performance of Ir_PicaSi_SiO and its molecular analogue in FA dehydrogenation shows that, whereas the former has a TOF about 2 times lower, the latter does not exhibit any induction time. This might be due to having Ir species in its cationic form or, most likely, by the inhibition of any associative deactivation process in Ir_PicaSi_SiO.

Kinetic studies (effect of pH, catalyst and FA concentration, temperature) further reveal a strict analogy between Ir_PicaSi_SiO and its molecular counterpart also in terms of the reaction mechanism, strongly suggesting that the species involved in the catalytic cycle are the same. The parallelism appears to be also applicable to the degradation of the catalytic center that occurs through the reductive deoxygenation of the C=O moiety of the ligand to form the corresponding amino species in Ir_PicaSi_SiO, as previously observed for analogous molecular catalysts.[55] Having observed that such a degradation pathway is active also in the immobilized catalyst allowed the understanding that it occurs intramolecularly, indicating that only an inhibition of the amide moiety rotation might avoid it.[49] Minimizing catalyst degradation might pave the way to the development of catalysts with TON large enough to be really applied in storing green hydrogen into CO2 and releasing it from FA when needed.