Rapid and selective formic acid dehydrogenation catalysis by molecular ruthenium hydrides supported by rigid PC_carbeneP pincer ligands

Abstract

A series of four molecular ruthenium hydrido complexes supported by previously reported rigid PC_carbeneP pincer ligand frameworks were evaluated as formic acid dehydrogenation (FAD) catalysts. The ligands in the complexes L_RRu(H)X (R = H, NMe₂; X = Cl, κ²-O₂CH) differ in the electron richness by substitution on the aryl groups linking the di-iso-propylphosphine arms to the central carbene donor. We find that only the unsubstituted (R = H) chloro and formato complexes are effective catalyst precursors; the NMe₂ substituted derivatives decompose under catalytic conditions. However, the two compounds L_HRu(H)X are highly active (TOF = 1300–4200 h⁻¹), long lived (TON up to 122 000) and selective (dihydrogen and carbon dioxide are the sole products) at 21 °C with no base additives necessary in 13 M formic acid in water/dioxane. These performance metrics compare well with state of the art catalysts operating under ambient conditions. Mechanistic experiments support a simple two-step mechanism involving rate limiting protonolysis of the Ru–H by formic acid to release H₂ and rapid loss of CO₂via β-elimination from the resulting formato complex.

---

Introduction

In discussions regarding societal transitioning from fossil fuels to less polluting, sustainable fuels, dihydrogen is often presented as an attractive alternative,¹ provided its production can be driven by renewable energy sources and its storage and transport can be done safely and economically. With regard to the latter concern, a major practical limitation in using hydrogen as a fuel is the low volumetric energy of H₂ gas (1.4 kW h L⁻¹ at 70 MPa),² and the attendant difficulties in its widespread distribution. Potential solutions to this issue include H₂ adsorption into porous heterogeneous materials like metal–organic frameworks or storage in the form of metal hydrides, which suffer from low volumetric capacities and poor reversibility, respectively.³

An alternative means of H₂ storage and transport uses reversible hydrogenation reactions and one of the most promising chemical storage systems is formic acid, since it is a relatively non-toxic liquid with low volativity under ambient conditions with a high gravimetric H₂ capacity (4.4 wt% resulting in energy density of 1.77 kW h L⁻¹) and low flammability.² Formic acid is available through biomass degradation or CO₂ hydrogenation.⁴ If formic acid dehydrogenation (FAD) can be made efficient,⁵ this in theory implements a carbon-neutral cycle for storing and releasing H₂ (Fig. 1). However, the high kinetic barrier to dehydrogenation for liberation of H₂ (ref. 6 and 7) from formic acid means that effective catalysts for “hydrogen on demand” from this carrier are needed.

Fig. 1 Homogeneous CO₂ hydrogenation (forward) and formic acid dehydrogenation (reverse) pathways catalysed by a metal-hydride complex, where [M] represents a molecular metal complex. Base additives are usually necessary for both processes.

Although FAD is a thermodynamically favoured reaction, selectivity is also an issue. Thermal decomposition can follow two pathways: the dehydrogenation pathway, giving CO₂ and H₂, and the dehydration pathway giving CO and H₂O.⁸ Therefore, a catalyst is also beneficial in directing the selectivity of the process for production of clean H₂. Accordingly, there has been intense interest in catalysts that can dehydrogenate formic acid with high activity, longevity and selectivity.⁵ These can take the form of heterogeneous systems⁹ or, more germane to the work reported here, homogeneous molecular catalysts.^10,11

While there have been some notable recent advances in the use of molecular catalysts based on earth abundant metals such as manganese^12,13 or iron,^14,15 with few exceptions these catalysts are unable to rival the TON or TOF performance of precious metal-based systems that use ruthenium or iridium.¹⁰ Focusing on the former metal, several ruthenium hydride catalyst precursors exhibit exceptionally high turnover numbers (TONs) and reasonable turnover frequencies (TOFs), Chart 1. Beller et al. reported a ruthenium hydride catalyst formed in situ from [RuCl₂(benzene)]₂ and 1,2-bisdiphenylphosphinoethane¹⁶ that eventually demonstrated a TON over one million with a modest TOF of 1000 h⁻¹ at 25 °C.¹⁷

Chart 1 Recent examples of ruthenium-based molecular catalysts for formic acid dehydrogenation.

Several molecular catalysts with PNP pincer ligands capable of ligand cooperativity¹⁸ have also proven highly effective. Pidko et al. reported a PNP ruthenium hydride that showed high activities towards FAD (257 000 h⁻¹ at 90 °C).^19,20 Huang et al. reported a related PNP ruthenium hydride that achieved a TON of over one million with a TOF of 7333 h⁻¹ at 90 °C.²¹ While these systems were highly active, they required basic additives to avoid decomposition under acidic conditions. Acid sensitivity is a common limitation of FAD catalysts and only a handful of base-free FAD catalysts have been reported.^22,23 Within the context of (PNP)Ru catalyst precursors, a noteworthy base-free catalyst was reported by Milstein et al. using a PNP-supported ruthenium hydride that could dehydrogenate neat formic acid, producing over one million turnovers with a rate of 3067 h⁻¹.²⁴ In order to achieve these performance metrics, most of these and other catalysts require elevated temperature (60–95 °C) and at room temperature, while TONs are maintained or improved, TOFs are significantly dampened. In general, for these systems, the TOF is governed mechanistically by rate limiting elimination of CO₂ from higher energy κ¹ metal formate intermediates.²⁵

Recently we reported ruthenium hydrides supported by an advanced generation of our rigid PC_carbeneP family of pincer ligands²⁶L_RRu(H)X (R = H, NMe₂; X = Cl, κ²-O₂CH, Chart 1).²⁷ These complexes were initially studied in the context of CO₂ hydrogenation and (somewhat to our surprise) we found them to be almost completely inactive for this process. Mechanistic and chemical studies showed this to be related to the extremely facile deinsertion of CO₂ from the formato intermediates presumably formed upon insertion of CO₂ into the Ru–H bond of these complexes, i.e., L_RRu(κⁿ-O₂CH)X (n = 1, X = Cl; n = 2, x = κ²-O₂CH). As we concluded in the previous contribution,²⁷ this suggests that these compounds should facilitate FAD catalysis as opposed to the reverse reaction. Here we show this to be the case, demonstrating that the complex where R = H is an extremely active, stable platform for highly selective, base additive free formic acid dehydrogenation catalysis at room temperature.

Results and discussion

A. FAD catalysis

Formic acid typically contains up to 15 mole% H₂O, therefore it is necessary that the precatalysts are water stable to engage in FAD catalysis. Although quite oxygen sensitive,‡ all four derivatives L_RRu(H)X (R = H, NMe₂; X = Cl, κ²-OOCH) proved completely stable when treated with degassed H₂O. None of the complexes were soluble in aqueous formic acid solutions, even at milligram per 100 mL concentrations, so an organic solvent additive was necessary to solubilize the catalyst precursors in the catalytic medium. Commonly, water-miscible solvents like DMF, DMSO, and 1,4-dioxane are used in FAD catalysis,^21,25,28 although liquid amines can also serve the dual purpose of catalyst solubilizer and Lewis basic additive to drive catalysis.^17,28 For precatalysts L_RRu(H)X (particularly the R = H derivatives), we have found that strongly donating solvents (specifically acetonitrile in this case) can induce a 1,2-hydride to carbene-carbon shift to form a PC_alkylP complex;²⁷ this is a primary mode by which these ligands engage in cooperativity for stoichiometric^29–33 and catalytic^34–38 transformations. In this context, transfer of the Ru–H to the ligand carbene carbon shuts down FAD catalysis (see mechanistic discussion below), so we used minimal amounts of the weakly-coordinating solvent 1,4-dioxane to solubilize the precatalysts in aqueous formic acid.

Initial small-scale experiments were conducted by dissolving the four ruthenium hydride precursors in ≈0.1 mL of dioxane in vial sealed with a septum under argon and adding ≈0.6 mL of 1.15 M aqueous, argon sparged formic acid solutions. For the R = H complexes L_HRu(H)X, immediate, visible evolution of gas was observed. Through analysis of the headspace by gas chromatography (GC), the gases evolved were determined to be H₂ and CO₂; no detectable amounts of CO were observed (detection limit ≈50 ppb). For the two L_NMe2Ru(H)X catalyst precursors, gas evolution was significantly less vigorous, and solutions underwent a rapid orange to purple color change upon which catalysis halted, leaving most of the formic acid substrate intact. Subsequent experiments therefore focussed exclusively on the L_HRu(H)X catalyst precursors.

Larger scale catalytic reactions were performed using a simple system that captures the evolved gas in an inverted graduate cylinder; the set-up is similar to what other researchers have employed for evaluation of molecular FAD catalysts^15,21,24 and is depicted schematically in Fig. S1.† Results of various experiments are summarized in Table 1. Catalyst precursors were dissolved in 1,4-dioxane in the reaction flask and the process initiated by injection of an amount of degassed, commercial grade (88%) formic acid through the flask septum at room temperature (21 °C). Under optimized conditions, both L_HRu(H)Cl and L_HRu(H)-κ²-O₂CH mediate complete, selective conversion (TON = 6900) of formic acid to H₂ and CO₂ (Table 1, entries 1 and 2). The reaction selectivity was confirmed by GC analysis and to confirm the complete conversion of formic acid, the reaction solution was assayed using ¹H NMR spectroscopy, using sodium benzoate as an internal standard. The κ² formato catalyst precursor is ≈4× more active that the chlorido derivative. When determining the minimum amount of dioxane solubilizer required, it was found that formic acid concentrations above the 13 M used for the optimized runs leads to a halting of catalysis. For L_HRu(H)-κ²-O₂CH, gas evolution ceases completely at 15 M formic acid, while L_HRu(H)Cl remained highly active for a few minutes before the rate began to decline (Table 1, entry 3), with gas production stopping completely after 27 minutes. Experiments analogous to that in entry 3 which included basic additives like triethylamine or sodium formate^{19,21,28,39,40} gave the same results. Furthermore, these additives did not enhance the activity of the catalysts as found in entries 1 and 2, indicating that FAD catalysis using L_HRu(H)X do not require basic additives to be effective. The decline or lack of activity observed at [HCOOH] > 13 M appears to be a solubility issue, where catalyst precipitation in more concentrated formic acid stops the catalytic reaction. Consistent with this hypothesis, injection of more 1,4-dioxane to entry 3 solutions leads to a resumption of gas evolution at comparable activities to those observed for entries 1 and 2. As expected, activity can be increased by raising the temperature to 40 °C (entries 4 and 5); we did not explore higher temperature regimes. It should also be noted that no reaction occurred in the absence of catalyst. Furthermore, as shown in entry 8, the complex [Ru(p-cymene)Cl₂]₂ (which is the starting material used to prepare complexes L_HRu(H)X²⁷), is completely inactive under these conditions. This simple precursor has been shown mediate FAD catalysis, albeit under much different conditions and rather poorly.²⁸

Table 1 Formic acid dehydrogenation catalysis with L_HRu(H)X

Entry	Catalyst	T (°C)	TON	TOF (h^−1)
a Optimized conditions, performed in triplicate, room temperature: 0.0015 mmol of catalyst dissolved in 2.0 mL 1,4-dioxane, 4.5 mL of 88% formic acid added (13 M). Deviations in calculated TOFs were no greater than ±5% based on the difference of the slopes in plots of gas volume vs. time. b 0.0015 mmol of catalyst dissolved in 2.0 mL 1,4-dioxane, 5.0 mL 88% formic acid added (15 M). c 0.004 mmol of catalyst in 2.5 mL 1,4-dioxane, 23 mL of 88% formic acid added at a rate of 0.86 mL per hour. d 0.003 mmol catalyst dissolved in 2.0 mL dioxane, 5 sequential additions of 0.45 mL 88% formic acid, TOF reported for first addition.
1^a	L R Ru(H)Cl	21	6900	1300(65)
2^a	L H Ru(H)-κ ^2 -O 2 CH	21	6900	4170(200)
3^b	L R Ru(H)Cl	21	1926	—
4^a	L R Ru(H)Cl	40	6900	1850(90)
5^a	L H Ru(H)-κ ^2 -O 2 CH	40	6900	8000(400)
6^c	L H Ru(H)-κ ^2 -O 2 CH	21	122 735	—
7^d	L H Ru(H)-κ ^2 -O 2 CH	21	17 000	3850
8^a	[Ru(p-cymene)Cl2]2	21	0	0

---

To obtain a better indication as to the longevity of catalyst precursors L_HRu(H)X in comparison to related systems with high TONs (Chart 1), we conducted catalytic runs with substantially lower catalyst loadings (Table 1, entry 6). Being limited by the formic acid concentration as described above, we used a syringe pump to add 23.00 mL of 88% formic acid at a rate of 0.86 mL per hour to 0.004 mmol of the more active catalyst, L_HRu(H)-κ²-O₂CH, in 2.5 mL of 1,4-dioxane. The reaction was stopped once the addition was complete (∼27 hours), at which point no further gas evolution was observed. Based on integration of the ¹H NMR spectrum, ∼93% conversion was reached with a total of ≈120 000 turnovers. The rate of gas production could not be measured with this high volume of gas, although the gas evolution did appear to slow after several hours before stopping after 24 hours. Given the oxygen sensitivity of the catalyst precursor, we attribute the halting of catalysis to eventual degradation of the catalyst due to the seepage of trace oxygen into the system and view these turnover numbers as representing a minimum value under the conditions employed.

Finally, we also probed the question of catalyst longevity by conducting five successive single injection cycles (Table 1, entry 7; Fig. 2) of substrate, using L_HRu(H)-κ²-O₂CH as the catalyst precursor. In this experiment, after each injection of formic acid substrate charge, the evolved gases in the inverted graduated cylinder were removed in vacuo and the reaction vessel recharged with formic acid. As shown in Fig. 2, resumption of gas evolution was instantaneous upon reintroduction of substrate. However, while the first two cycles showed the same rate within error (∼2%) the rate during the third cycle dropped by ≈25%. The 4th cycle showed a further drop in the rate of gas evolution to ≈58% slower than that of the first cycle (note, this 4th charge occurred several hours after the 3rd charge had been fully converted). By the 5th cycle the rate was slowing during the reaction, even as formic acid was present in large excess. These experiments demonstrate that L_HRu(H)-κ²-O₂CH exhibits remarkable stability over long periods of time, likely limited primarily by its oxygen sensitivity.

Fig. 2 Graph of the volume of gas formed over time in 5 cycles of formic acid dehydrogenation catalyzed by L_HRu(H)-κ²-O₂CH.

B. Mechanism

When FAD reactions mediated by L_HRu(H)X are monitored by ³¹P{¹H} NMR spectroscopy, the catalyst precursors are the only species observable during catalysis (Fig. S2†). Stoichiometric reactions of L_HRu(H)X and one equivalent of H¹³CO₂H, when monitored by ¹³C{¹H} NMR spectroscopy (Fig. S3,† X = Cl; S4, X = κ²-O₂CH) show signals only for ¹³CO₂ product, with no intermediates observed in either case; for L_HRu(H)-κ²-O₂CH, small amounts of label are incorporated into the formato ligand of the precatalyst, consistent with previous observations.²⁷

Together, these observations point to a rate limiting protonolysis of the hydride ligands in L_HRu(H)X by HCOOH (pK_a = 3.75 (ref. 41)), to release H₂ and form the unobserved intermediates I (X = Cl) and II (X = κ²-O₂CH) depicted in Scheme 1. In our previous contribution, we attempted to prepare these compounds but found them to be prone to rapid loss of CO₂, producing L_HRu(H)X cleanly and quantitatively. These two steps constitute a simple mechanism for formic acid dehydrogenation. We note here that when more strongly donating solvents like acetonitrile are used to solubilize the catalyst, catalysis is shut down because of the 1,2-hydride transfer to the carbene carbon induced by such solvents.²⁷ Thus, maintaining the Ru–H moiety is essential for catalysis. The enhanced activity observed for the formato precatalyst L_HRu(H)-κ²-O₂CH likely stems from its more rapid reaction with formic acid in the first rate determining protonolysis step in comparison to the chlorido version L_HRu(H)Cl. In the stoichiometric reactions with H¹³CO₂H mentioned above (Fig. S3 and S4†), the degradation of formic acid is essentially instantaneous for the former while for the latter it takes 15–20 minutes to fully convert.

Scheme 1 Proposed mechanism for FAD catalysis mediated by L_HRu(H)X.

The products of rate determining protonolysis, intermediates I and II, Scheme 1, rapidly release CO₂ to complete the cycle. Noting that the hydrido formates previously reported undergo loss of CO₂ at a much slower rate, the question arises as to why elimination of CO₂ from I and II are so facile in comparison. We propose that the disparity is related to differing trans influence for a hydrido ligand vs. the chlorido and κ¹-formato ligands in I and II. Presumably, in order to undergo β-elimination to release carbon dioxide, slippage of the κ²-formato ligands in I and II to a κ¹ structure must occur in order to access the proposed β-elimination type transition states TS-Cl and TS-O₂CH shown in Scheme 1. (Although II already has one κ¹-formato ligand, it is not able to β-eliminate due to coordinative saturation). Indeed, both structural and computation evidence suggests that the O_trans bonds (highlighted in pink in Scheme 1) are shortened significantly in in silico or chemical model structures of I and II in comparison with the formato hydride L_RRu(H)-κ²-O₂CH structures.

For example, the β-elimination immune benzoate complex L_NMe2Ru(Cl)-κ²-O₂CPh was prepared and structurally characterized for comparison to the published structure of L_NMe2Ru(H)-κ²-O₂CH.²⁷ The benzoate was readily prepared via reaction of L_NMe2RuCl₂ and NaO₂CPh; further details can be found in the ESI.† Interestingly, this compound can also be cleanly prepared by treatment of L_NMe2Ru(H)Cl with benzoic acid (pK_a = 4.20 (ref. 42)), a weaker acid than formic acid. This observation supports the initial protonolysis step proposed in Scheme 1. In any case, crystals of model complex L_NMe2Ru(Cl)-κ²-O₂CPh were obtained from benzene solutions layered with pentane, and the molecular structure of this compound is shown in Fig. 3. In comparison to the reported structure of L_NMe2Ru(H)-κ²-O₂CH,²⁷ where the difference between the Ru–O(1) and Ru–O(2) is only 0.081 Å, this disparity in L_NMe2Ru(Cl)-κ²-O₂CPh is substantially larger at 0.222 Å. This is mainly due to an elongation of Ru–O(2)_trans bond in the hydrido formate in comparison to the same bond in the chloro benzoate, consistent with the greater trans influence expected for the hydrido ligand in the former complex. With a larger disparity between the Ru–O(1)/Ru–O(2) distance, the benzoate ligand in L_NMe2Ru(Cl)-κ²-O₂CPh is more energetically inclined to access the κ¹ mode than the formato ligand in L_NMe2Ru(H)-κ²-O₂CH and provides some circumstantial support for the notion that weaker trans influence ligands favor κ¹-formato structures and more facile CO₂ loss from intermediates like I and II.

Fig. 3 Single crystal XRD determined structure of L_NMe2Ru(Cl)-κ²-O₂CPh, ellipsoids shown at the 50% probability level and H atoms omitted for clarity. Selected bond lengths [Å] and angles [°]: Ru–C1: 1.924(3), Ru–O1: 2.342(2), Ru–O2: 2.120(2), Ru–Cl: 2.4001(7), O1–C33: 1.276(4), O2–C33: 1.250(4), C1–Ru–O1: 109.69(11), C1–Ru–O2: 168.09(10), O1–C33–O2: 120.0(3), C1–Ru–Cl: 94.11(9).

A more direct comparison between L_RRu(X)-κ²-O₂CH (here, X is H vs. Cl) was obtained computationally using DFT methodology (bp3w91-SDD for Ru; def2TZVP for Cl, P, O, C_carbene; def2SVP for the remaining atoms) to compute the ground state structures of the four complexes. Full computational details and coordinates can be found in the ESI.† The calculated structures of L_RRu(H)-κ²-O₂CH once again showed differences in Ru–O_cis and Ru–O_trans bond lengths of less than 0.1 Å and a more stabilized κ² structure; the metrical parameters obtained for the R = NMe₂ complex are in good agreement with the experimentally determined structure.²⁷ As found for the model benzoate structure discussed above, the calculated structures of the chlorido derivatives L_RRu(Cl)-κ²-O₂CH had a much greater difference between these two bond lengths: Δ (Ru–O_cis/Ru–O_trans) = 0.282 Å for R = H and 0.272 Å for R = NMe₂ (Fig. S5†). This further substantiates the notion that the replacement of the hydride with a chloride pushes the formate closer to a κ¹ coordination mode, which would be expected to lower the barrier for β-H elimination assuming this requires slippage to a κ¹ formate to allow access to the TS-Cl or TS-O₂CH transition structures. It should also be noted that since the X ligand in II is a second formato ligand, CO₂ elimination from this species is statistically favored in comparison to I and may also contribute to the overall higher activity observed for this catalyst relative to the chlorido derivative.

Conclusions

In this study, we evaluated four PC_carbeneP pincer complexes L_RRu(H)X as formic acid dehydrogenation catalysts. While the R = NMe₂ derivatives undergo rapid stoichiometric reactivity with HCOOH, they are not stable under typical catalytic conditions. However, the unsubstituted derivatives (R = H) are highly effective catalysts for the selective conversion of formic acid to dihydrogen and carbon dioxide at room temperature, without the need for Lewis base additives. Both for X = Cl and κ²-O₂CH, the catalysts exhibit high turnover numbers and reasonable turnover frequencies, with the latter being about 3–4 times as active. Mechanistic studies support a two-step mechanism where the resting state of the catalysts are the hydrido precursors. Rate limiting protonolysis followed by rapid CO₂ elimination completes the simple catalytic cycle.

In comparison to the state-of-the-art molecular catalysts based on ruthenium depicted in Chart 1, compounds L_HRu(H)X perform very well. In most instances, the TOF values observed here at 21 °C are comparable to those observed for other systems at >90 °C, the exception being the system reported by Pidko et al.^19,20 While we report somewhat lower TON values than the systems listed in Chart 1, the value of ≈122 000 we obtain is a lower limit; the somewhat rudimentary nature of our experimental set up, coupled with the oxygen sensitivity of our catalyst systems limits our ability to more accurately evaluate the true longevity of these catalysts. The system most comparable to the catalysts L_HRu(H)X reported here is the chemically more simple catalyst of Beller et al.¹⁶ While it has demonstrated high turnover numbers, this was evaluated in a more sophisticated apparatus than ours;¹⁷ in terms of TOF under similar conditions, the performance of L_HRu(H)-κ²-O₂CH is about 4× faster than the Beller system, and without the need for amine base additives.

Data availability

The data supporting this article have been included as part of the ESI.†

Author contributions

Experiments, computations and data analysis were carried out by LJD. The manuscript was prepared with contributions from LJD and WEP. BSG performed X-Ray crystallographic analysis on L_NMe2Ru(Cl)-κ²-O₂CPh.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This research was funded by the Natural Sciences and Engineering Research Council of Canada through a Discovery Grant to WEP. WEP also acknowledges and thanks the Canada Research Chairs program for a Tier I CRC (2020-2027). LJD thanks NSERD for a PGS-D award and the Province of Alberta for and Alberta Graduate Excellence Scholarship (AGES).

Notes and references

1. I. P. Jain, Int. J. Hydrogen Energy, 2009, 34, 7368–7378.
2. J. Eppinger and K.-W. Huang, ACS Energy Lett., 2017, 2, 188–195.
3. M. Yadav and Q. Xu, Energy Environ. Sci., 2012, 5, 9698–9725.
4. M. Tawalbeh, R. M. N. Javed, A. Al-Othman, F. Almomani and S. Ajith, Environ. Technol. Innovation, 2023, 31, 103217.
5. A. Wang, P. He, J. Wu, N. Chen, C. Pan, E. Shi, H. Jia, T. Hu, K. He, Q. Cai and R. Shen, Energy Fuels, 2023, 37, 17075–17093.
6. S. Satyapal, J. Petrovic, C. Read, G. Thomas and G. Ordaz, Catal. Today, 2007, 120, 246–256.
7. G. Sievi, D. Geburtig, T. Skeledzic, A. Bösmann, P. Preuster, O. Brummel, F. Waidhas, M. A. Montero, P. Khanipour, I. Katsounaros, J. Libuda, K. J. J. Mayrhofer and P. Wasserscheid, Energy Environ. Sci., 2019, 12, 2305–2314.
8. W. L. Nelson and C. J. Engelder, J. Phys. Chem., 1926, 30, 470–475.
9. R. Xu, W. Lu, S. Toan, Z. Zhou, C. K. Russell, Z. Sun and Z. Sun, J. Mater. Chem. A, 2021, 9, 24241–24260.
10. N. Onishi, R. Kanega, H. Kawanami and Y. Himeda, Molecules, 2022, 27, 455.
11. C. Guan, Y. Pan, T. Zhang, M. J. Ajitha and K.-W. Huang, Chem. – Asian J., 2020, 15, 937–946.
12. N. H. Anderson, J. Boncella and A. M. Tondreau, Chem. – Eur. J., 2019, 25, 10557–10560.
13. A. Léval, H. Junge and M. Beller, Catal. Sci. Technol., 2020, 10, 3931–3937.
14. A. Boddien, D. Mellmann, F. Gärtner, R. Jackstell, H. Junge, P. J. Dyson, G. Laurenczy, R. Ludwig and M. Beller, Science, 2011, 333, 1733–1736.
15. E. A. Bielinski, P. O. Lagaditis, Y. Zhang, B. Q. Mercado, C. Würtele, W. H. Bernskoetter, N. Hazari and S. Schneider, J. Am. Chem. Soc., 2014, 136, 10234–10237.
16. A. Boddien, B. Loges, H. Junge, F. Gärtner, J. R. Noyes and M. Beller, Adv. Synth. Catal., 2009, 351, 2517–2520.
17. P. Sponholz, D. Mellmann, H. Junge and M. Beller, ChemSusChem, 2013, 6, 1172–1176.
18. J. R. Khusnutdinova and D. Milstein, Angew. Chem., Int. Ed., 2015, 54, 12236–12273.
19. G. A. Filonenko, R. van Putten, E. N. Schulpen, E. J. M. Hensen and E. A. Pidko, ChemCatChem, 2014, 6, 1526–1530.
20. G. A. Filonenko, E. J. M. Hensen and E. A. Pidko, Catal. Sci. Technol., 2014, 4, 3474–3485.
21. Y. Pan, C.-L. Pan, Y. Zhang, H. Li, S. Min, X. Guo, B. Zheng, H. Chen, A. Anders, Z. Lai, J. Zheng and K.-W. Huang, Chem. – Asian J., 2016, 11, 1357–1360.
22. Y. Himeda, Green Chem., 2009, 11, 2018–2022.
23. Z. Wang, S.-M. Lu, J. Li, J. Wang and C. Li, Chem. – Eur. J., 2015, 21, 12592–12595.
24. S. Kar, M. Rauch, G. Leitus, Y. Ben-David and D. Milstein, Nat. Catal., 2021, 4, 193–201.
25. D. P. Estes, M. Leutzsch, L. Schubert, A. Bordet and W. Leitner, ACS Catal., 2020, 10, 2990–2998.
26. R. A. Manzano and R. D. Young, Coord. Chem. Rev., 2021, 449, 214215.
27. L. J. Donnelly, J.-B. Lin, B. S. Gelfand, C. Y. Chang and W. E. Piers, Dalton Trans., 2024, 53, 1862–1869.
28. B. Loges, A. Boddien, H. Junge and M. Beller, Angew. Chem., Int. Ed., 2008, 47, 3962–3965.
29. R. J. Burford, W. E. Piers and M. Parvez, Organometallics, 2012, 31, 2949–2952.
30. L. E. Doyle, W. E. Piers and J. Borau-Garcia, J. Am. Chem. Soc., 2015, 137, 2187–2190.
31. L. E. Doyle, W. E. Piers, J. Borau-Garcia, M. J. Sgro and D. M. Spasyuk, Chem. Sci., 2016, 7, 921–931.
32. D. V. Gutsulyak, W. E. Piers, J. Borau-Garcia and M. Parvez, J. Am. Chem. Soc., 2013, 135, 11776–11779.
33. E. A. LaPierre, W. E. Piers, D. M. Spasyuk and D. W. Bi, Chem. Commun., 2016, 52, 1361–1364.
34. J. Borau-Garcia, D. V. Gutsulyak, R. J. Burford and W. E. Piers, Dalton Trans., 2015, 44, 12082–12085.
35. J. D. Smith, G. Durrant, D. H. Ess, B. S. Gelfand and W. E. Piers, Chem. Sci., 2020, 11, 10705–10717.
36. W. Weng, S. Parkin and O. V. Ozerov, Organometallics, 2006, 25, 5345–5354.
37. A. V. Polukeev and O. F. Wendt, Organometallics, 2017, 36, 639–649.
38. H. Tinnermann, S. Sung, B. A. Cala, H. J. Gill and R. D. Young, Organometallics, 2020, 39, 797–803.
39. J. F. Hull, Y. Himeda, W.-H. Wang, B. Hashiguchi, R. Periana, D. J. Szalda, J. T. Muckerman and E. Fujita, Nat. Chem., 2012, 4, 383–388.
40. A. Vatsa and S. K. Padhi, New J. Chem., 2022, 46, 15723–15731.
41. J. A. Riddick and W. B. Bunger, Techniques of Chemistry: Organic Solvents, John Wiley and Sons Ltd., New York, 4th edn, 1985.
42. W. M. Haynes, CRC Handbook of Chemistry and Physics, CRC Press Inc., Boca Raton, 91st edn, 2010.

---

Footnotes

† Electronic supplementary information (ESI) available: General Experimental details, synthesis and characterization of all new compounds, catalytic procedures, computational details and crystallographic parameters for CCDC 2400771. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4cy01346h

‡ Unpublished results in the form of a crystal structure and ³¹P NMR data suggest that phosphine oxidation is the primary avenue of decomposition for these compounds in the presence of O₂ and other O atom transfer agents. Further elucidation of these products is beyond the scope of this article, but we note that oxygen poisoning of catalysis is rapid and irreversible in the context of the FAD process reported here.

---