Miriam
Abán‡
,
J. Marco
Cuenca
,
Irene
Embid
,
Alba
de Toro
,
Pilar
Gómez-Sal
,
Ernesto
de Jesús
,
Marta
Valencia
* and
Cristina G.
Yebra
*
Departamento de Química Orgánica y Química Inorgánica, Instituto de Investigación Química “Andrés M. del Río”, Universidad de Alcalá, Campus Universitario, 28805 Alcalá de Henares, Madrid, Spain. E-mail: cgarcia.yebra@uah.es; marta.valencia@uah.es
First published on 2nd October 2024
The reaction of [Cp*IrCl2]2 with 1-pyridyl-1,2,3-triazoles La–Ld in the presence of sodium triflate yielded the complexes [Cp*Ir(k2-NN)(Cl)][OTf] (1a–1d) in which the k2-NN ligand coordinates to iridium through the pyridyl nitrogen and the N2 nitrogen of the triazole ring, leaving the more basic N3 atom free. These complexes serve as efficient catalytic precursors for the dehydrogenation of formic acid in the presence or absence of an external solvent, achieving turnover frequencies (TOFmax) of up to 10703 h−1 and producing a 1
:
1 mixture of hydrogen and carbon dioxide with no detectable carbon monoxide. The catalysts can be reused after a single reaction cycle, as they retain their activity. This retention of activity has allowed a cumulative turnover number (TON) of 26
876 to be reached after six reloading cycles of neat formic when using sodium formate as a base. Kinetic and 1H NMR studies reveal that the active catalyst, [Cp*Ir(CO)H2] (2), forms in situ from precursors 1a–1d under the reaction conditions. Thus, complexes 1a–1d function as stable precatalysts, simplifying the use of the otherwise unstable complex 2. The reaction mechanism is likely to involve mono- and dihydride species. The decarboxylation of the coordinated formate is the rate-determining step at high formic acid concentrations, according to the kinetic profiles and kinetic isotope effect (KIE) values obtained.
Physical storage of H2 gas requires materials with special properties, resistant to hydrogen corrosion and capable of withstanding high pressures to ensure safety during transport.4 Liquid hydrogen storage is not a competitive alternative due to its high energy demand, involving extremely low temperatures and very high pressures, with daily losses of 3–6% due to vaporization.5 Chemical hydrogen storage offers an alternative approach using materials with high hydrogen content that are easy to transport and store. These materials can release their hydrogen content through a chemical process, often catalyzed.6 Among liquid organic hydrogen carriers (LOHCs), formic acid (FA) has been considered a hydrogen storage material for decades as it is a weak, biodegradable acid, liquid at room temperature, and relatively low in toxicity. It can reversibly convert into hydrogen (H2) and carbon dioxide (CO2) using catalysts (HCOOH(l) → H2(g) + CO2(g), ΔG° = −32.9 kJ mol−1).7,8
Dehydrogenation of formic acid using heterogeneous catalysts typically requires high temperatures and produces carbon monoxide as a result of competitive FA dehydration (HCOOH(l) → H2O(l) + CO(g), ΔG° = −12.4 kJ mol−1).7,9–11 Therefore, the development of catalysts that allow the production of high-purity hydrogen from formic acid, under milder conditions, becomes an important issue.
Homogeneous catalysts are particularly interesting as highly versatile coordination complexes that enable formic acid dehydrogenation under milder conditions and facilitate the proceeding of the reaction and understanding of the reaction mechanisms,12,13 allowing for the isolation and study of their reaction intermediates.14 On the other hand, the dehydrogenation of formic acid in the absence of solvent is an environmentally sustainable process, producing only H2 and CO2 as the final reaction products. However, the number of organometallic complexes capable of efficiently catalysing the dehydrogenation of formic acid under these conditions is limited and the use of additives is generally needed (Chart 1). Investigations employing ruthenium complex A, as detailed by Milstein,15 lead to remarkable turnover numbers (TON) of up to 1.7 million over 24 runs. The [Ir(tBu2PCH2(2-py))(cod)]+ precatalyst (B), as described by Williams, has been examined for both high efficiency (turnover frequency, TOF, up to 13320 h−1) and robustness.16 Analysis of the gaseous mixture in these cases reveals the presence of variable amounts of carbon monoxide. Experiments conducted utilizing the rhodium catalysts stabilized by pincer CNC ligands C, perform with maximum TOF over 10
000 h−1 in the presence of sodium formate and water.17 Research conducted with the iridium system D, indicates that the maximum TOF value is reached at 5 hours, albeit necessitating the addition of 30 mol% sodium formate and 100 mol% water.18 Other iridium systems only provide initial TOF values (lasting few minutes or seconds), lacking information on the process rate over extended durations (E).19 Cp*IrIII-derivatives developed by Fischmeister (F) dehydrogenate neat formic acid with initial TOF of up to 13
292 h−1 at 100 °C.20 Xiao also reports impressive TOF values of up to 147
000 h−1, based on a few seconds of reaction, for the dehydrogenation of neat 5
:
2 azeotropic HCOOH/NEt3 mixtures using their Cp*Ir-systems (G).21
![]() | ||
Chart 1 Efficient organometallic catalysts for solventless formic acid dehydrogenation; H2O, NEt3, or HCOONa are used as additives when specified. |
Catalysts containing the organometallic fragment Cp*Ir (Cp* = pentamethyl-η5-cyclopentadienyl) and k2-N,N ligands, show high activity and stability in aqueous media.12,22–31 Recently, He and co-workers have prepared complexes [Cp*Ir(k2-NN)(OH2)]n+ (k2-NN being pyridyl-1,2,3-triazole and pyridyl-1,2,4-triazole ligands, and n = 1, 2),32,33 in which the 1,2,3-triazole heterocycles coordinate to iridium through their more electron-rich N3 atom.34–38 These complexes efficiently catalyse formic acid dehydrogenation in aqueous solution.
Herein, we present the preparation and characterization of a family of [Cp*Ir(k2-NN)(Cl)]+ complexes, featuring 1-pyridyl-1,2,3-triazole ligands coordinated to the iridium centre through the N atom of the pyridine and the N2 atom of the triazole. These complexes serve as precatalysts for the dehydrogenation of formic acid in the presence or absence of solvent. They exhibit competitive activities under neat conditions, including TOF values of up to 10703 h−1 and accumulated TON exceeding 26
800. 1H NMR analysis of the catalytic residues confirms the in situ formation of the catalytically active species, identified as [Cp*Ir(CO)H2], which forms upon the release of the pyridyl–triazole ligand under the reaction conditions.
![]() | ||
Scheme 1 Synthesis of complexes [Cp*Ir(k2-NN)(Cl)][OTf] (1a–1d) and ORTEP diagram of the cationic unit of 1a (50% probability ellipsoids; H atoms omitted). |
These ligands adopt the less common inverse triazole complex-type k2-(N,N)-coordination mode,34,39 in which the N2 nitrogen of the triazole ring and the pyridyl nitrogen coordinate to iridium, leaving free the more basic N3 atom of the triazole ring (see structure of 1a obtained by X-ray diffraction analysis in Scheme 1). Otherwise, complex 1a displays a typical piano stool geometry with three ledges formed by the chlorido and the bidentate pyridyl–triazole ligand. The Ir–triazole distance (Ir–N(2) = 2.063(2) Å) is shorter than the Ir–pyridine (Ir–N(4) = 2.120(2) Å) distance and also slightly shorter than that previously reported for related inverse triazole iridium complexes (2.07–2.08 Å).36 Complete characterization details for complexes 1a–1d are given in the ESI.†1H NMR spectra of the complexes reveal the highly acidic character of the triazolium proton, that appears as a singlet at around 9.5 ppm for 1a–1c and is shifted to a higher frequency for 1d (9.77 ppm), which agrees with the presence of a p-CF3 substituted pyridine in the latter. The observation of 13C resonances at around 128 ppm for the CH of the triazole moiety confirms the coordination of the triazole moiety through the N2 atom.
Notably, there is no induction period in subsequent reaction cycles, when fresh reactants (FA and NEt3) are added to the residue from a previous catalytic batch. Fig. 2 depicts one such experiment, in which complex 1a was subjected to four reloading experiments with the objective of evaluating the durability of the catalytic system. While the complete dehydrogenation of FA took 1 hour in the initial cycle, this time was reduced to 25 minutes in the first recharge, reaching 50% conversion in only 5 minutes (TOF50 = 664 h−1) due to the absence of an induction period. The catalytic activity decreases slightly in the third cycle and sharply in the fourth cycle (TOF50 of 416 and 36 h−1, respectively; Table S3†).
When using complex 1d, the durability of the catalytic system was further investigated by adding successive fresh charges of HCOOH/NEt3 mixtures. After three hours and five cycles of reaction at 100 °C, a cumulative turnover number of 4350 was reached when a 0.1 mol% of 1d was used in the initial cycle (Fig. 3a). However, there was a noticeable decline in the catalytic activity with each reloading cycle. This was evidenced by the reduction in the measured TOF50 values from 2465 h−1 in the first cycle to 1237 h−1 in the fifth cycle (Table S5†).
Furthermore, the performance of complex 1d was examined in the dehydrogenation of formic acid/sodium formate mixtures under the same conditions (Fig. 3b). This approach has the advantage of significantly reducing the total reaction volume, which in turn increases the energy density. Additionally, the elimination of volatile organic solvents prevents any potential damage to the materials in fuel cell electrodes. The reactor was charged with a 10:
1 formic acid/sodium formate mixture and a 0.1 mol% of 1d in the first cycle, and exclusively with formic acid in subsequent cycles. A rapid evolution of gases was observed when reagents and precatalyst were initially mixed at a temperature of 100 °C (C1; Fig. 3b). This resulted in a turnover number (TON) of 634 and a maximum turnover frequency (TOFmax) of 1749 h−1 within just 16 minutes (entry C1, Table S6†). Following the addition of fresh, neat formic acid, a TOFmax of 4253 h−1 was achieved at the same temperature (entry C2, Table S6†). Subsequently, the temperature was reduced to 90 °C and neat formic acid was added for eight additional runs, with both TOFmax and TOF50 values consistently exceeding 2000 h−1. An accumulated TON of 7702 was achieved after overall ten cycles of formic acid dehydrogenation. By lowering the iridium loading to 0.02 mol% (Fig. 3c), the accumulated TON increased to 26
876 over six runs (entries 1-C1 to 1-C6 in Table S7†). A further reduction in the metal-loading to 0.01 mol% over two runs resulted in an improved TOF value of up to 10
703 h−1 (entries 2-C1 and 2-C2, Table S7; Fig. S4†).
The starting complex 1a is almost completely converted in about 1 h into two hydride complexes, neither of which contains the pyridyl–triazole moiety La (complexes D and E in Fig. S5a†). Complex D, probably a monohydride, is characterized by 1H resonances at 1.71 ppm (Cp*) and at −15.43 ppm (Ir–H). Complex E corresponds to the dihydride carbonyl compound [Cp*Ir(CO)H2] (2) previously reported by Heinekey and co-workers (Scheme 2).40 The latter complex was unambiguously identified by the HMBC correlation found between the carbonyl 13C resonance at about 176 ppm and the 1H resonances of both the Cp* ring at 1.94 ppm and the hydride ligand at −15.91 ppm (Fig. S7†). A new 1H NMR spectrum of the reaction mixture recorded after 18 h at room temperature, shows hydride 2 and free ligand to be the only species in solution (Fig. S6†). Similar observations were made in a related study with precatalyst 1d. In this case, the pyridyl–triazole ligand Ld was completely decoordinated after heating the reaction mixture at 70 °C for less than 20 min, and the complex 1d was fully converted to the hydride 2 after another 18 h of reaction at room temperature (Fig. S8†). The formation of the same complex [Cp*Ir(CO)H2] (2) from 1a and 1d is confirmed by comparison of the corresponding 1H NMR spectra after 18 h of reaction (Fig. S9†).
An open question at this stage was whether complex 2 was also formed under the solvent-free conditions used in the experiments of Fig. 3a. This concern was addressed by analysing the residue obtained from the dehydrogenation of neat HCOOH/NEt3 using 0.1 mol% of 1d at 100 °C, where complex 2 again emerged as the main organometallic species (Fig. S10†).
The formation of the complex [Cp*Ir(CO)H2] (2) involves an initial stage where the pyridyl–triazole ligand is released from the corresponding complex 1. This release is likely driven by the significant nucleophilicity of the N3 atom in the triazole ring (Fig. 4),41 making it more reactive towards electrophiles,36,42–44 particularly the proton of formic acid. Subsequent steps in the formation of 2 probably involve the disproportionation of formate intermediates to carbonyl complexes, as previously documented for transition metal complexes,45–48 including iridium.45,49
Complex 2 was tested in the dehydrogenation of HCOOH/NEt3 (11 mol%) in toluene. The catalytic profile shows no discernible induction period, as anticipated from the results discussed above. Under conditions analogous to those used in Fig. 2 for precatalyst 1a (70 °C and 1 mol%), complex 2 reduced the time to reach 50% conversion from 25 minutes to approximately 2 min. This improvement is primarily due to the elimination of the induction period when comparing the performance of complex 2 with that of 1a in a subsequent catalytic cycle. In that second cycle, 1a has already been converted to 2 (Fig. S2†). However, complex 2 is thermally unstable, releasing dihydrogen and eventually forming the dimer [Cp*Ir(CO)]2. Consequently, and despite the introduction of an induction period, complexes 1a–1d may be useful precatalysts for FA dehydrogenation due to their ability to generate in situ the highly active, albeit thermally unstable, catalyst 2.
Fig. 5a delineates two hypotheses regarding the mechanism of action of 2 in the dehydrogenation of formic acid.7,50 The depicted IrIII–IrI mechanism is initiated by the reductive elimination of dihydrogen from 2, which is then followed by the oxidative addition of formic acid to the iridium(I) intermediate A. The subsequent decarboxylation of the resulting formate B regenerates the initial dihydride complex 2. However, it is anticipated that intermediate A will readily dimerize in this cycle, thereby generating the more stable complex [Cp*Ir(CO)]2 (C).40 This off-cycle species might re-enter the catalytic cycle by hydrogenation, but as discussed above, this process is exceedingly slow (Fig. S31†). Therefore, the catalyst should accumulate rapidly in the form of the off-cycle species C, leading to a decrease in catalytic activity, which is inconsistent with the observed data. In addition, dimer C was found to be inactive as a catalyst for the FA dehydrogenation at 90 °C under the otherwise standardized conditions that were used in this work. The mechanism shown on the right-hand side of Fig. 5a, which exclusively involves IrIII species, seems more plausible. The first step in such cycle is the release of hydrogen by the protonation of 2 with formic acid to form the monohydride B.
![]() | ||
Fig. 5 Formic acid dehydrogenation catalysed by complex [Cp*Ir(CO)H2] (2). (a) Simplified mechanistic proposals for an IrIII/IrI cycle and a cycle based on IrIII species. (b) The experimental reaction profile obtained at 60 °C using a 0.6 mol% loading of 2 (red plot) is compared with that calculated using the COPASI software and a chemical model based on the IrIII species (blue plot).51 |
We have conducted a preliminary kinetics study to gain further insight into the reaction mechanism. As shown in Fig. 5b, the reaction profile is nearly linear up to approximately 80% conversion. Consequently, the reaction rate remains largely unaffected by the formic acid concentration up to that point. Assuming the IrIII cycle as the operating mechanism, the decarboxylation of B should be the rate-determining step. As illustrated in Fig. 5b, a satisfactory fit was obtained between the experimental and calculated reaction profile using a chemical model based on this IrIII cycle and a least squares regression analysis of the complete kinetic profile using the COPASI software (see ESI† for details).51 The analysis yielded precise values for the rate constant of the decarboxylation step across a temperature range of 40 to 90 °C (Table S12†). These rate constants, in conjunction with the Eyring equation, enabled the determination of the activation energy parameters for the decarboxylation step: ΔH‡ = 13.3 ± 1.0 kcal mol−1 and ΔS‡ = −21.8 ± 3.1 cal mol−1 K−1.
Additionally, we performed H/D Kinetic Isotope Effect (KIE) measurements to support a mechanistic proposal involving a rate-determining decarboxylation step. These experiments used complex 1d as the precatalyst for the dehydrogenation of neat formic acid, employing sodium formate (10 mol%) at 100 °C with no solvent. The observed maximum turnover frequency (TOFmax) decreased from 1749 h−1 to 641 h−1 when the HCOOH/HCOONa mixture was replaced with DCOOH/DCOONa (Table S8†). This change corresponds to a KIE value of 2.7, which is consistent with the proposed rate-determining step, as decarboxylation involves the cleavage of a C–H bond in the formate ligand. Furthermore, as observed experimentally, a negative variation of the activation entropy can be expected from the changes in the geometry of the complex that must occur to activate the C–H bond.
Finally, we analysed the gas mixtures present in the reactor headspace in four representative experiments, two using species 1a and 2 for the dehydrogenation of HCOOH/NEt3 solutions in toluene, and two using precatalyst 1d for the dehydrogenation of neat HCOOH/HCOONa mixtures (Table S9†). The CO content was below the detection limit of the GC analyser in all four cases. This result is consistent with observations made by other authors that the presence of a CO ligand in the structure of Cp*Ru active species prevents the competitive dehydration of formic acid, which produces carbon monoxide.52
Complexes 1a–1d are not directly active in the dehydrogenation of formic acid. The catalytic profile of the reactions shows an induction period during which the pyridyl-1,2,3-triazole ligand is released and the dihydride complex [Cp*Ir(CO)H2] (2) is formed. This complex is responsible for the observed catalytic activity regardless of the catalytic precursor used. The handling of complex 2 is challenging due to its inherent instability towards dehydrogenation. This highlights the practicality of using complexes 1a–1d as stable precatalysts for the dehydrogenation of formic acid.
Crystallographic data for 1a has been deposited at the CCDC under 2362559 and can be obtained from https://www.ccdc.cam.ac.uk.
Footnotes |
† Electronic supplementary information (ESI) available: Synthesis of ligands and complexes; description of catalytic experiments, including tables, reaction profiles, and kinetic studies; NMR and mass spectra; crystallographic data and computational details. CCDC 2362559. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4qi01818d |
‡ Present address: IMDEA Energy, Avda. Ramón de la Sagra 3, 28935 Móstoles (Madrid), Spain. |
This journal is © the Partner Organisations 2024 |