Reduction of Nitrogen Oxides by Hydrogen with Rhodium(I)-Platinum(II) Olefin Complexes as Catalysts.

The nitrogen oxides NO2 , NO, and N2 O are among the most potent air pollutants of the 21st century. A bimetallic RhI -PtII complex containing an especially designed multidentate phosphine olefin ligand is capable of catalytically detoxifying these nitrogen oxides in the presence of hydrogen to form water and dinitrogen as benign products. The catalytic reactions were performed at room temperature and low pressures (3-4 bar for combined nitrogen oxides and hydrogen gases). A turnover number (TON) of 587 for the reduction of nitrous oxide (N2 O) to water and N2 was recorded, making these RhI -PtII complexes the best homogeneous catalysts for this reaction to date. Lower TONs were achieved in the conversion of nitric oxide (NO, TON=38) or nitrogen dioxide (NO2 , TON of 8). These unprecedented homogeneously catalyzed hydrogenation reactions of NOx were investigated by a combination of multinuclear NMR techniques and DFT calculations, which provide insight into a possible reaction mechanism. The hydrogenation of NO2 proceeds stepwise, to first give NO and H2 O, followed by the generation of N2 O and H2 O, which is then further converted to N2 and H2 O. The nitrogen-nitrogen bond-forming step takes place in the conversion from NO to N2 O and involves reductive dimerization of NO at a rhodium center to give a hyponitrite (N2 O2 2- ) complex, which was detected as an intermediate.

Introduction

Nitrogen oxides such as N2O (nitrous oxide), NO (nitric oxide), and NO2 (nitrogen dioxide) are potent greenhouse gases and a hazard to human health. Because NO readily oxidizes to NO2 in air, the effect of NO and NO2 on climate and human health is frequently discussed jointly by referring to NOx as toxic pollutants. N2O and NO have high global warming potentials of 298 and 126 CO2 equivalents, respectively. Additionally, N2O has been identified as the dominant ozone‐depleting agent of the 21st century. NO facilitates the formation of certain respiratory diseases and cancers, especially in densely populated areas. Significant amounts of nitrogen oxides in the atmosphere are of natural origin (64 % for N2O and 16 % for NO ). Since the start of the industrial revolution, emissions from non‐natural sources have increasingly contributed to the overall atmospheric concentration. The combustion of fossil fuels is the main source of NO in our atmosphere (64 %). N2O is a waste product from chemical processes (6 %), predominantly from adipic and nitric acid syntheses. Furthermore, denitrification from nitrogen‐containing fertilizers, derived from the Haber‐Bosch process, adds 30 % N2O and 20 % NO to the atmospheric concentration.

Mitigation of these gases is crucial for both, slowing down global warming and improving the health of all living populations. N2O, NO, and NO2 are thermodynamically unstable concerning their decomposition in N2 and O2 (Δf H 0(298 K)=82, 90 and 33 kJ mol−1), but high activation barriers for these processes make them kinetically inert. Various heterogeneous catalysts have been developed to catalyze their conversion to nitrogen and oxygen, but high temperatures and pressures are required. Such catalytic systems were used for decades in after‐treatment systems of exhaust lines of combustion engines. However they are not suited to fit modern‐day regulations, which led to the development of new types of converters in which NO is reduced predominantly by ammonia (Figure 1).

Figure 1

State‐of‐the‐art catalysts for the reduction of the nitrogen oxides NO2, NO and N2O with H2, NH3 or H2O as reagents. Enzymatic, heterogeneous, and homogeneous examples are presented alongside the homogeneous and heterogenized catalysts presented in this work.

Currently, hydrogen is still produced mainly by methane steam reforming but due to large research efforts into the development of power‐to‐gas technologies, inexpensive hydrogen from renewable sources will become available soon. Hence, it might become a promising atom‐economic reducing agent for the mitigation of N2O and NO . Several heterogeneous catalysts successfully hydrogenate N2O, but there are only two homogeneous catalysts containing rhodium or ruthenium which promote this reaction (TONs up to 417). To the best of our knowledge, no homogeneous catalytic system has been reported for the catalytic reduction of NO or NO2 with hydrogen. Only stoichiometric reactions with iron, nickel, copper, or ruthenium complexes were reported recently. Nature has developed a set of enzymes that can catalyze the reduction of N2O and NO. NO can be reduced by iron‐containing nitric oxide reductases and the bimetallic active core of one of these is presented in Figure 1. In such enzymes, NO can be reduced to N2O. The reduction of N2O by nitrous oxide reductase, which contains a Cu4S cluster as the active site, also relies on cooperativity between two copper centers: N2O is first chelated between two copper centers followed by N2 extrusion.

Inspired by the high stability of catalytic converters that contain rhodium, palladium or platinum, and the bimetallic cooperativity observed in nature, we synthesized a series of RhI–PtII complexes using a special tropPPh2 (trop=5H‐dibenzo[a,d]cyclohepten‐5‐yl) ligand. These complexes were analyzed by X‐ray diffraction methods and multinuclear NMR spectroscopy in order to determine their structures. Subsequently, these bimetallic complexes were tested in the reduction of N2O with hydrogen. In addition, selected complexes were also examined in the reduction of NO and NO2. Finally, a series of stoichiometric and catalytic hydrogenation reactions were performed which, combined with results from DFT calculations, give insight into a possible reaction mechanism.

Results and Discussion

Synthesis and characterization of heterobinuclear Rh–Pt complexes

The multidentate ligand 1 (Figure 2 a) was synthesized by adapting a recently published protocol (Supporting Information, Figure S1). The ligand is designed in such a way that two transition metals can bind in close proximity within the concave pocket formed by the diphenylphosphino group, Ph2P, the alkene unit of the central seven‐membered ring, and the two alkynyl units. The first metal is introduced by reacting 1 with [Rh2(μ2‐Cl)2(C2H4)4] in benzene which gives the chloro‐bridged dimeric rhodium complex 2 in good yield (Figure 2 a). Subsequent reaction with [Pt2Me4(μ2‐SMe2)2] in dichloromethane cleanly gives the heterobimetallic complex 3 in 85 % isolated yield. Various salts such as M(PF6), M(BArF) {M=Na, Tl; BArF=B[(3,5‐CF3)2C6H3]4 −} or Ag(X) (X=SbF6 −, OTf−=CF3SO3 −) were tested as chloride scavenger agents and reacted with 3. In the presence of coordinating solvents such as acetonitrile or water, complexes 4–6 were obtained (Figure 2 a). The reaction between 3 and AgOTf leads to 7 which contains OTf− as weakly coordinating anion bound to the Rh center. Remarkably, the reaction in a non‐coordinating solvent such as benzene or toluene leads to the dimeric aggregates 8 and 9 in which two [RhPt(Me)2(1)] units are bridged via a Cl− ion or a TlCl2 − ion. All complexes were obtained as deep red crystals in moderate to good yields and their molecular structures were determined by X‐ray diffraction analysis of single crystals (Figures S3–S10).

Figure 2

a) Synthesis of complexes 2–9 starting from ligand 1. b) Plot of the molecular structure of 7 determined by X‐ray diffraction methods (hydrogen atoms and two molecules of tetrahydrofuran are omitted for clarity). c) Contour plot for the Laplacian of the electron density (∇2 ρ) of 7 in the P‐Rh‐Pt plane. Only relevant atoms are shown. Solid lines indicate electron concentration, dotted lines electron depletion. Bond critical points are indicated as green dots. d) 31P{1H} NMR and e) 195Pt{1H} NMR spectra of complex 7.

Multinuclear NMR spectroscopy including 103Rh and 195Pt NMR spectroscopy, was used to characterize these complexes in solution (Figures S18–S64). As an example, a structure plot of the triflate complex 7 is shown in Figure 2 b. Remarkably, complex 9 shows an interaction between the central Tl+ ion and two phenyl groups of the PPh2 units (Tl‐ct= 3.2147(18) Å; ct=centroid of the phenyl groups; Figure S10). This interaction is retained in solution as indicated by an resonance signal at δ 205Tl=−699 ppm in the 205Tl NMR spectrum in [D8]THF (Figure S64), which is comparable with previously reported data for TlI diarene complexes (e.g. [Tl(hexamethylbenzene)2]+: δ 205Tl=−563 ppm, vs. TlCl: δ 205Tl=383 ppm). This low‐frequency chemical shift is likely caused by ring current effects within the arene rings.

The structural data collected for complexes 3–9 (Figures S3–S10) reveal distances between the two metal centers (2.67–2.77 Å; Figure S65) which are in line with previously reported RhI−PtII bond lengths (2.63–2.98 Å). The short Rh–Pt distances, the 2 J PPt coupling constants of 2 J PPt=506–538 Hz, and the direct 1 J PtRh coupling of 1 J PtRh=82–104 Hz observed by 195Pt NMR spectroscopy (Figures S26–S63) are strong indications for a Rh−Pt bond. Although the NMR data are relatively invariant for 3–9, the shorter Rh−Pt bonds in the cationic monomeric complexes 4–6 (Rh−Pt≤2.70 Å) correlate with a slightly larger 1 J PtRh coupling constant (>97 Hz; Supporting Information, Table S1). Theoretical methods were employed to further characterize the Rh−Pt bond. An analysis of the electronic structure of 7 using the quantum theory of atoms in molecules (QTAIM) (Figure 2 c) supports the assumption that the Rh−Pt interaction is a direct metal−metal bond as indicated by the bond critical point on the path of maximum electron density between Rh and Pt (Figure 2 b). In addition, the positive value of the corresponding Laplacian of the electron density computed at such bond critical point (∇2 ρ=+0.13) suggests that the Rh−Pt bond is best described as a closed‐shell (donor–acceptor) interaction. A molecular orbital—linear combination of atomic orbitals (MO‐LCAO) analysis (Figure S66) and an energy decomposition analysis combined with the natural orbitals for chemical valency (EDA‐NOCV, Figure S67) indicate that the Rh−Pt bond is indeed best described as a donor–acceptor interaction in which the d8‐PtII center acts as a donor through its doubly occupied dz2‐orbital (Figure S68, Table S2).

Catalytic reduction of N2O

Complexes 3–9 were tested in the catalytic reduction of N2O in tetrahydrofuran with hydrogen at an overall pressure of 2 bar (room temperature, one bar partial pressure of N2O or hydrogen, respectively, Table 1). The turnover number (TON) was determined by 1H NMR spectroscopy by quantification of the proton signals of water as one of the products (Supporting Information). Nitrogen as the sole further product was detected in the final reaction mixture by gas chromatography with a thermal conductivity detector (GC‐TCD). All catalysts are stable under the reaction conditions and no decomposition was observed by 31P NMR spectroscopy after the catalytic reaction ceased (Figure S70). Catalyst 7 remains active for more than two days without significant loss in catalytic activity (Figure S71). Under these very mild reaction conditions, the complexes showed TONs between 10 and 140 in tetrahydrofuran (Table 1, Entry 1–7). Changing the solvent from tetrahydrofuran to dichloromethane resulted in a lower TON of 12 (using complex 7 as catalyst) while in benzene, hydrogenation of the solvent to cyclohexane instead of N2O was observed. Comparing the catalytic activity of complexes 3–9 shows that the aquo complex 6 and the ClTlCl‐bridged complex 9 are the best performing catalysts (Table 1, entries 4 and 7). Even at temperatures as low as 0 °C, TON of up to 151 were obtained with catalysts 9 (Table S3). These results indicate that a weakly bound ligand at the rhodium center as in complexes 6 (H2O), 7 (OTf−) and 9 (TlCl2 −) ensures efficient catalysis suggesting that an easily accessible coordination site at the Rh center is required. This conclusion is further supported by the enhanced activity when 3 is reacted with a chloride abstracting agent such as AgSbF6 leading to an increase of TON values of 11 to 36 (Table 1, entry 9). On the contrary, the addition of a complexing agent such as PPh3 inhibits the hydrogenation of N2O (Table 1, entry 8) which also indicates that the catalysis proceeds in homogeneous solution. Finally, the mononuclear rhodium complex in which the two‐electron donating PtMe2 fragment is replaced by PPh3, [Rh( tropPPh2)(PPh3)](OTf) 10, has low activity (TON=10, Table 1, entry 10). This result indicates that the PtMe2 fragment is important as structural and electronic modulator as well as cooperative metal center in substrate binding and activation (vide infra).

Table 1

Catalytic reduction of N2O by complexes 3–10.[a]

Entry	[RhPt]	t [h]	P[b] [bar]	TON[c]
1	3	22	1	11
2	4	22	1	10
3	5	22	1	11
4	6	22	1	135
5	7	22	1	62
6	8	22	1	34[d]
7	9	22	1	140[d]
8	9 + 50 equiv. PPh3	22	1	9
9	3 + AgSbF6 [e]	22	1	36
10	10 [f]	22	1	10
11	6	48	2	475
12	9	22	2	250[d]
13	9	48	2	587[d]

[a] Conditions: [cat.] (3.5 mM), tetrahydrofuran (2 mL), total volume (120 mL). [b] Partial pressure of both gases. [c] Water formed during the reaction determined by 1H NMR spectroscopy with mesitylene as an internal standard. [d] With respect to one RhPt core. [e] Generated in situ (see the Supporting Information for details). [f] Monomeric rhodium complex [ TropPPh2RhPPh3]OTf (Figure S69).

The reaction conditions were optimized using complexes 6 and 9 as catalysts. Increasing the pressure of the reactant gases to a total pressure of 4 bar (2 bar of each reactant gas) and the reaction time to 48 h resulted in an increase of the TONs to 475 and 587, respectively (Table 1, entries 11–13). These results set a new benchmark for the hydrogenation of N2O to N2 and H2O (previously a TON of 417 at 3.5 bar H2/3.5 bar N2O at 70 °C were achieved with the Ru complex shown in Figure 1).

Catalytic reduction of NO

The high stability and efficiency of the Rh–Pt complexes prompted us to investigate the hydrogenation of nitrogen oxide NO (Table 2, Figure 3). The chloride complex 3, as well as the three best‐performing complexes 6, 7, and 9 in the hydrogenation of N2O were investigated as catalysts using tetrahydrofuran as solvent. The progress of the reaction was monitored by quantification of the amount of water, which forms during the reaction. The combined TON′=(TON + TON ) for this transformation is given in Table 2. Surprisingly, the aquo complex 6 and ClTlCl‐bridged dimer 9, the most active N2O reduction catalysts so far, performed poorly in the conversion of NO/H2 to N2O/H2O (TON′ of ≈10, Table 2, entries 2 and 4). The neutral complexes 3 and 7 exhibit a higher activity with 3 affording a TON′ of 16 and 7 reaching a TON′ of 36 after 19 h (Table 2, entry 3). In all cases, N2O was detected as one of the reaction products by GC‐TCD.

Table 2

Catalytic reduction of NO by selected [RhPt] complexes.[a]

Entry	[RhPt]	TON′[b]
1	3	16
2	6	11
3	7	36
4	9	9[c]

[a] Conditions: [cat.] (3.0 mM), tetrahydrofuran (2 mL), total volume (50 mL). [b] Water formed during the reaction determined by Karl‐Fischer titration. [c] With respect to one RhPt core.

Figure 3

Optimized conditions for the catalytic reduction of a) NO and b) NO2 with hydrogen and rhodium platinum triflate complex 7. Reaction conditions for reduction of NO: 7 (6.5 mM), tetrahydrofuran (3 mL), total volume (50 mL). TON′ determined by GC‐TCD and Karl‐Fischer titration. Reaction conditions for reduction of NO2: 7 (14 mM), [D8]THF (0.5 mL), total volume (2.5 mL). TON′′ was determined by 1H NMR spectroscopy with mesitylene as an internal standard. NO and N2O were detected by GC‐TCD. The hyponitrite complex 11 was characterized as a key intermediate by c) 31P/103Rh HMQC and d) 31P/15N HMQC of 15N‐labeled 11.

Under optimized conditions, NO can be fully reduced to N2 over the course of a week by the Rh–Pt triflate complex 7 (Figure 3 a) as indicated by GC‐TCD. The addition of PPh3 poisons the catalyst showing that it proceeds in homogeneous phase. The generation of N2O as an intermediate (GC‐TCD, Figure S73) indicates a stepwise reduction, where NO is first converted to N2O, followed by reduction to N2 (GC‐TCD, Figure S72).

To verify that this reaction can also be carried out under solvent‐free conditions, complex 7 was deposited on activated carbon (7@C Figure 3 a; Figures S74 and S75). The immobilized catalyst 7@C, in presence of molecular sieves (pore size 4 Å) to trap water, was found to be as active as its homogenous counterpart (7 in solution), and a combined TON′=34 was achieved (GC‐TCD, Figure S76).

Remarkably, the more challenging substrate NO2 can also be reduced by catalyst 7 (Figure 3 b). The formation of NO and N2O as reaction products were observed by GC‐TDC, supporting again a stepwise reduction of NO2 (Figure S77). For this transformation, a combined TON′′=TON + TON′ of 8 was observed. Therefore, we rank qualitatively the reaction rates for the individual conversions in the order > > . Very likely this reflects the stability of catalyst 7 against N2O, NO, and NO2. No decomposition of 7 was observed when a solution was stored under a mixture of N2O and hydrogen by 31P{1H} NMR spectroscopy (vide supra). However, 7 decomposes within several days in an atmosphere of NO and hydrogen and within a day under NO2 and hydrogen. The phosphine oxide of ligand 1 was the main decomposition product observed by 31P{1H} NMR spectroscopy.

Mechanistic insight: stoichiometric experiments

The rhodium‐platinum complexes 3–9 seemingly do not react with hydrogen (up to 5 bar) in tetrahydrofuran solution for weeks and neither rhodium nor platinum hydride species were detected by 1H NMR spectroscopy. However, scrambling experiments using a H2/D2 mixture, or H2 and D2O, with complex 7, led to the detection of HD within minutes (Figures S78–S80). This demonstrates the capability of the rhodium‐platinum triflate complex 7 to activate hydrogen and transfer hydrogen centers to a substrate molecule. Next, we investigated the interaction between 7 and NO by following the reaction using multinuclear NMR spectroscopy (Figures S81–S85) with the particular aim to better understand the nitrogen−nitrogen bond‐forming step that must occur at some stage to generate N2O from NO. A fast and quantitative formation of a new complex 11 is observed when 7 is placed under an atmosphere of NO in tetrahydrofuran. Complex 11 is remarkably stable in tetrahydrofuran solution and no decomposition was detectable by 31P{1H} NMR spectroscopy over several days. With 15NO gas, the isotope‐labeled complex 11‐ is formed, allowing a detailed spectroscopic analysis using 31P/103Rh and 31P/15N heteronuclear multiple quantum correlation (HMQC) experiments (Figure 3 c,d). The 103Rh NMR spectrum shows a resonance signal split into a triplet which is caused by coupling between the 103Rh nucleus and two equivalent 15N nuclei (2 J RhN=41.7 Hz). Note that neither 195Pt satellites could be observed nor a 195Pt NMR spectrum could be measured. We attribute this to a larger 195Pt chemical shift anisotropy (CSA) tensor in 11 compared to complexes 3–9. It is a well‐known phenomenon in 195Pt NMR spectroscopy that larger CSA tensors may cause very fast relaxation resulting in resonance lines too broad to be detected. 31P{1H} NMR spectroscopy of 11‐ shows a doublet of a triplet (1 J PRh=171.4 Hz, 3 J PN=4.5 Hz) indicating two chemically equivalent 15 N centers. The observed coupling constant of 3 J PN=4.5 Hz was utilized for polarization transfer in a 31P/15N HMQC experiment to prove the coupling between the nuclei (Figure 3 d). The 15 N NMR spectrum of 11‐ shows a doublet (2 J NRh=41.6 Hz) of virtual triplets centered at δ15N=452.6 ppm because the 3 J NRh and 3 J NP are by chance of the same value (3 J NRh=3 J NP=4.6 Hz). In combination these results strongly indicate that complex 11 contains a cis‐hyponitrite dianion, N2O2 2−, as ligand as a result of a reductive coupling reaction between two 15NO molecules at the Rh center which is formally oxidized from RhI to RhIII in this process (for comparison an aqueous solution of sodium trans‐hyponitrite shows a resonance at δ15N=498.8 ppm ). Pressurizing an NMR tube containing the hyponitrite complex 11 with 3 bar of hydrogen led to the formation of a triflate RhI–PtII complex within 21 hours postulated as 6‐OTf, with a water molecule coordinated to the rhodium center (Figure S86). The formation of hyponitrite complexes by reductive coupling of two equivalents of NO and the decomposition of these to N2O has been intensively investigated but to the best of our knowledge, none of these reactions are catalytic. The decomposition of hyponitrites to N2O, especially in aqueous acidic solution in dependence of the pH, is well‐established.[ , ]

Mechanistic insight: DFT calculations

According to DFT calculations (SMD/PBE0‐D3/def2‐SVP) (Figure 4), the formation of the hyponitrite complex 11 from complex 7 and two equivalents NO occurs in a stepwise manner. In the first exergonic addition of NO to 7, a bent nitroso complex 7‐NO is obtained (−4 kcal mol−1) in which NO coordinates with its N‐center to Rh. The NO unit in 7‐NO retains largely its radical character as a plot of the calculated spin density shows (Figure 4 b). In the second endergonic step (+11 kcal mol−1), the second NO molecule attacks the Rh‐bound NO at the N‐center to form complex 7‐(NO) with an unusual Rh‐NO‐NO unit reminiscent of the π*‐π* dimer formed when NO radicals are cooled below −160 °C or concentrated in confined spaces. This is nicely illustrated by the plot of the HOMO of complex 7‐(NO) which is strongly localized on the N2O2 unit and corresponds to the bonding interaction between the π*‐orbitals of each NO unit (Figure 4 c). Subsequently, 7‐(NO) rearranges to the hyponitrite complex 11. The overall reaction 7 + 2 NO → 11 is exergonic by ΔG=−8 kcal mol−1. No transition states could be located along the path 7 → 7‐NO → 7‐(NO) → 11 indicating that this part of the minimum energy reaction path (MERP) may be an almost barrierless process. The presented MERP (Figure 4 a) is in agreement with the experimental observations but at present still remains one possibility and as such corresponds to a simplified model.

Figure 4

a) Minimum Energy Reaction Path (MERP) calculated with DFT for the conversion of 2 NO + H2 to N2O and H2O using complex 7 as a catalyst. b) Plot of the calculated spin density in 7‐NO. c) Plot of the highest molecular orbital (HOMO) in 7‐(NO).

The displacement of the coordinated triflate anion by H2 in the next reaction step to give the hydrogen complex 11‐H is rather endergonic (ΔG=27 kcal mol−1) while the oxidative cleavage of H2 proceeds via a low barrier at TS1 (ΔG ≠=4 kcal mol−1) to give 11‐HH. In this dihydride complex one hydride is bound terminally to Rh and the other in a bridging μ2‐fashion over the Rh−Pt bond. The following step involves the transfer of the μ2‐H center to one oxygen center of the coordinated hyponitrite ligand to give the activated complex TS2 (ΔG ≠=2 kcal mol−1) which decomposes to give N2O and a hydrido hydroxo complex 11‐HOH in which the OH ligand takes a bridging position between Rh and Pt in such a fashion that O is bound to Rh and the H center has a contact to Pt. It is in these two steps, 11‐H → 11‐HH and 11‐HH → 11‐HOH, that the Pt center intervenes as a cooperative reaction center. The reaction 11‐HH to 11‐HOH + N2O is strongly exergonic (ΔG=−54 kcal mol−1). The second product molecule, H2O, is formed via transfer of the remaining hydride ligand to the hydroxo group, both bound at the Rh center to give the aquo complex 6‐OTf which was experimentally isolated as SbF6 − salt (Figure 2). Displacement of the coordinated water molecule in 6‐OTf by the counter anion OTf is exergonic and regenerates 7 (ΔG=18 kcal mol−1) whereby the catalytic cycle is closed. The overall calculated reaction energy for 2 NO + H2 → N2O + H2O of ΔG=−75 kcal mol−1 is in very good agreement with the one obtained from experimental data (ΔG°=−71.13 kcal mol−1).

Conclusion

A new family of stable heterodinucelar complexes was prepared using a ligand platform that was specifically designed to host two low‐valent metal centers in close proximity. The compounds 3–9 contain a directly bonded RhI−PtII unit where both metal centers have a d8 valence electron configuration. Quantum mechanical calculations indicate that this bond is best described as a dative Pt→Rh bond.

Among these dinuclear complexes, those which may liberate a free coordination site at the Rh center are active as catalysts in the hydrogenation (“reduction”) of the nitrogen oxides NO2, NO and N2O which are converted to water and dinitrogen. Following the reaction between the catalyst and NO by NMR spectroscopy allowed us to unambiguously identify a hyponitrite complex as a key intermediate for the N−N bond forming‐step which is crucial for the conversion of NO to N2. Taking the experimental results into account, DFT calculations were preformed and allow us to make a proposition for a possible mechanism leading from 2 NO and H2 to N2O and H2O. Although certainly oversimplified (for example, because specific interactions between the intermediates and solvent molecules and/or water, which forms during the reaction, were neglected) a MERP profile can be proposed which is roughly in accord with the experimental observations: NO is rapidly converted to hyponitrite in a reductive coupling step mediated at one metal center followed by slower hydrogenation in which both metal centers cooperate. A detailed investigation of the interaction between the Rh–Pt complexes and H2 or the hydrogenation of N2O to N2 and H2O is beyond the scope of this work and requires additional experimental results. In a previous study we demonstrated that rhodium hydride species may convert N2O via a [Rh]‐H⋅⋅⋅O‐N=N interaction directly into [Rh]‐OH and N2 and similar mechanisms may apply here. Overall the insight gained from this study may be used to prepare significantly more active catalysts which may be achieved by (i) designing a ligand which is more robust against NO and in particular avoids phosphorus‐containing binding sites and (ii) the preparation of neutral complexes in order to avoid poisoning of the active centers by the counter ion. Remarkably, immobilization of the molecular catalyst on a support did not lower its activity. It may therefore be possible to (i) prepare an active catalyst using the principles of molecular design, (ii) optimize the catalyst components and reaction conditions in homogeneous solution using high‐resolution spectroscopy, and (iii) immobilize the molecular catalyst on an inert support material to give a heterogeneous catalyst with well‐defined active sites.

Acknowledgement

We thank Frank Krumeich for the STEM picture of 7@C and Christoph Neff, Bodo Hattendorf and Detlef Günther for the ICP‐MS analysis of 7@C. P.J., A.A., J.J.G.C., G.L., E.F., F.M., S.G., M.T., T.L.G. and H.G. acknowledge funding from the Swiss National Science Foundation (SNF 162437, 181966 and 192106) and Eidgenössische Technische Hochschule Zürich (ETH). I.F. acknowledges the Spanish Ministerio de Ciencia e Innovación for financial support (RED2018‐102387‐T and PID2019‐106184GB‐I00). Deposition Number(s) 2050091, 2050092, 2050093, 2050094, 2050095, 2050096, 2050097, 2050098, 2050099 contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service www.ccdc.cam.ac.uk/structures.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.