Ligand assisted carbon dioxide activation and hydrogenation using molybdenum and tungsten amides

Abstract

The hepta-coordinated isomeric M(NO)Cl₃(PN^HP) complexes {M = Mo, 1a(syn,anti); W, 1b(syn,anti), PN^HP = (iPr₂PCH₂CH₂)₂NH, (H_N atom of PN^HP syn and anti to the NO ligand)} and the paramagnetic species M(NO)Cl₂(PN^HP) (M = Mo, 2a(syn,anti); W, 2b(syn,anti)) could be prepared via a new synthetic pathway. The pseudo trigonal bipyramidal amides M(NO)(CO)(PNP) {M = Mo, 3a; W, 3b; [PNP]⁻ = [(iPr₂PCH₂CH₂)₂N]⁻} were reacted with CO₂ at room temperature with CO₂ approaching the M=N double bond in the equatorial (CO,NO,N) plane trans to the NO ligand and forming the pseudo-octahedral cyclic carbamates M(NO)(CO)(PNP)(OCO) (M = Mo, 4a(trans); W = 4b(trans)). DFT calculations revealed that the approach to form the 4b(trans) isomer is kinetically determined. The amine hydrides M(NO)H(CO)(PN^HP) {M = Mo, 5a(cis,trans); W, 5b(cis,trans)}, obtained by H₂ addition to 3a,b, insert CO₂ (2 bar) at room temperature into the M–H bond generating isomeric mixtures of the η¹-formato complexes M(NO)(CO)(PN^HP)(η¹-OCHO), (M = Mo, 6a(cis,trans); M = W, 6b(cis,trans)). Closing the stoichiometric cycles for sodium formate formation the 6a,b(cis,trans) isomeric mixtures were reacted with 1 equiv. of Na[N(SiMe₃)₂] regenerating 3a,b. Attempts to turn the stoichiometric formate production into catalytic CO₂ hydrogenation using 3a,b in the presence of various types of sterically congested bases furnished yields of formate salts of up to 4%.

---

Introduction

Carbon dioxide is one of the main carbon sources and although it is abundant and inexpensive, its utilization as feedstock for commodity chemicals is insufficiently explored.¹ CO₂ being a thermodynamically and kinetically stable molecule a promising approach to overcome its often too low reactivity would be its metal mediated “activation”.^2,3a In fact recent years have seen tremendous efforts to activate and reduce CO₂ and to employ it as a C₁ building block for chemical synthesis.³ Activation by binding to a metal center and subsequent reaction modes of the CO₂ depend on the type of the CO₂ binding. There are three possible ways of CO₂ coordination as depicted in Fig. 1.^1b

Fig. 1 Possible binding modes of CO₂ to a metal centre.

Lately metal ligand cooperation (MLC) has been explored as a way to activate H₂ or small molecules with C–H, N–H, O–H moieties.^4,5 MLC would be a promising approach for CO₂ activation, but reported examples of such CO₂ activation are scarce. Only recently Sanford and coworkers have demonstrated that (PNN)Ru(H)(CO) pincer systems take up CO₂ at room temperature involving C–C bond formation between the unsaturated arm of the pincer ligand and the electrophilic carbon atom of CO₂ (Fig. 2).⁶ Milstein and coworkers investigated related types of reversible CO₂ binding employing unsaturated PNP pincer complexes of Ru and Re (Fig. 2).⁷

Fig. 2 Metal–ligand cooperation in CO₂ activation.

“Metal-free” CO₂ activation was accomplished using Frustrated Lewis Pair (FLP) type chemistry.⁸ In this context it seems also worth mentioning that a transition metal based L_nRe-H/B(C₆F₅)₃ systems can activate CO₂ in a FLP type manner, where the metal hydride species took the role of the FLP Lewis base component.⁹ Activation of CO₂ is anticipated to be preceding its hydrogenation, which may then lead to formic acid, but also can continue in the reduction sequence forming products with higher hydrogen contents, which like formic acid could in principle function as reversible H₂ storage materials.¹⁰ Though the reaction of H₂ with CO₂ to obtain formic acid is exothermic (ΔH = −8 kcal mol⁻¹), it is endergonic with ΔG = 8 kcal mol⁻¹ due to gaseous starting materials. Therefore an additional thermodynamic driving force has to be provided, like for instance pressure and the addition of a base inducing salt formation or stabilizing acid–base hydrogen bonding interactions³ causing a shift of the equilibrium towards the formic acid side.

Advances in the efficiency of homogeneous hydrogenation catalysis of CO₂ were recently reported with late transition metal complexes,^11,12 but somewhat different from these developments it is one of the aims of this research to replace the platinum group metals ruthenium,^13–15 rhodium^16,17 and iridium^18,19 with non-precious metals.²⁰ In this regard molybdenum and tungsten based reductions of CO₂ are challenging objectives. Despite some recent achievements in molybdenum and tungsten catalysed homogeneous hydrogenations²¹ and related hydrosilylations,²² and our group's recent accomplishments in bifunctional imine hydrogenations,^23a the stepwise ionic hydrogenation of imines,^23b,c Osborn type olefin hydrogenations²⁴ and nitrile hydrogenations,²⁵ the homogeneous hydrogenation of carbon dioxide could not be achieved with compounds of these metals till date.

In this paper a carbon dioxide activation mode supported by MLC could be established as shown in Fig. 2. Furthermore, stoichiometric hydrogenation of CO₂ turned out to be accessible using the previously reported molybdenum and tungsten amides 3a,b.^23a

Results and discussion

Preparation of M(II) and M(I) (M = Mo, W) complexes bearing (ⁱPr₂PCH₂CH₂)₂NH ligand

Reaction of (iPr₂PCH₂CH₂)₂NH (PN^HP) ligand with the M(NO)Cl₃(NCMe)₂ complexes (M = Mo, W) led at room temperature in THF to precipitation of the isomeric mixtures of Mo(NO)Cl₃(PN^HP), 1a(syn,anti) and W(NO)Cl₃(PN^HP), 1b(syn,anti)) in moderate yields (1a(syn,anti), 55%; 1b(syn,anti), 62%) (Scheme 1).

Scheme 1 Preparation of the isomeric mixture of M(II)(NO)Cl₃(PN^HP), 1a,b(syn,anti)) and reduction to M(I)(NO)Cl₂(PN^HP), 2a,b(syn,anti) complexes (M = Mo,W).

The syn and anti notation for the isomers of 1a,b(syn,anti) correspond to the relative position of the H_N atom at the internal N atom of the PN^HP ligand with respect to the NO ligand. The spectroscopic yields of syn and anti isomers of both 1a,b(syn,anti) are given in Scheme 1. The isomers of 1a(syn,anti) and 1b(syn,anti) were found to be reasonably soluble only in CH₂Cl₂. The ³¹P{¹H} NMR spectra of the isomeric mixture of 1a(syn,anti) revealed two singlets at δ 69.3 ppm (1a(syn)) and 71 ppm (1a(anti)) in a ratio of 7 : 3 indicating equivalence of both phosphorus atoms of each isomer of 1a(syn,anti) (Fig. 3, left). The ³¹P{¹H} NMR spectra of the isomeric mixture of 1b(syn,anti) exhibited also two singlets, but in this case along with tungsten satellites at δ 56.8 ppm (s, ¹J_P–W (d, satellite) = 184.8 Hz, 1b(syn)) and 58.3 (s, ¹J_P–W (d, satellite) = 189.3 Hz, 1b(anti)) assigned to the equivalent phosphorus atoms of each syn and anti isomer (Fig. 3, right).

Fig. 3 ³¹P{¹H} NMR spectra of the isomeric mixture of Mo(NO)Cl₃(PN^HP), 1a(syn,anti) {left, 1a(syn), 70%; 1a(anti), 30%} and the isomeric mixture of W(NO)Cl₃(PN^HP), 1b(syn,anti) {right, 1b(syn), 70%; 1b(anti), 30%} at room temperature in CD₂Cl₂.

In the IR spectra of 1a,b(syn,anti) the isomers could not be distinguished, which showed only one sharp band for the molybdenum and tungsten complexes at 1653 and 1596 cm⁻¹, respectively, attributed to the ν_NO vibrations. The ¹H NMR spectra of the isomeric mixtures of 1a,b(syn,anti) exhibited several doubled signals for the methyl, methyne and methylene protons appearing in the expected region with strong overlap of the signals of the syn and anti isomers impeding their distinct assignment. In the ¹H NMR spectra a conspicuous signal appeared at δ 4.4 ppm, which was assigned to the H_N atom of 1a(syn). However, the H_N atom of 1a(anti) could not be assigned in the ¹H NMR spectrum presumably owing to the quite low concentration of this isomer and overlap of the signal with the signal of the CH₂Cl₂ solvent. The H_N atoms of the isomeric mixture of 1b(syn,anti) appeared as broad singlets at δ 4.47 ppm and 5.50 ppm in the ¹H NMR spectra and based on the different intensity these could be assigned to the 1b(syn) and 1b(anti) isomers, respectively. The virtual triplet signal in the ¹³C{¹H} NMR at δ 55.9 (t, ^vJ_C–P = 3.6 Hz) ppm for isomeric mixture of 1a(syn,anti) was attributed to the carbon atoms of 1a(syn) isomer adjacent to the nitrogen atom of the PN^HP ligand and the related signals of 1b(anti) could not be detected. On the other hand the virtual ¹³C{¹H} NMR triplets at δ 55.9 (t, ^vJ_C–P = 5.9 Hz) ppm for 1b(syn) and at 55 ppm (t, ^vJ_C–P 4.8 Hz) for 1b(anti) could be assigned to the N-adjacent carbon atoms. Several attempts failed to grow suitable single crystals for X-ray diffraction studies of any isomer of the 1a,b(syn,anti) mixture, but 1a,b(syn,anti) could be fully characterized by ¹H NMR, ³¹P{¹H} NMR, ¹³C{¹H} NMR, IR and mass spectrometry, and the compositions were established by elemental analyses.

1a,b(syn,anti) are heptacoordinated diamagnetic species with the metal centers in +II oxidation states. Our initial attempts to deprotonate the N–H moiety of 1a(syn,anti) applying NaN(SiMe₃)₂ to obtain 16 e⁻ unsaturated amido complex^23a failed, which led to inseparable mixtures of products and could not be identified. Attempts to reduce 1a,b(syn,anti) by 1-electron reducing agents led to paramagnetic M(+I) species (M = Mo, W). Further reduction to M(0) species seemed difficult with the given strongly electron donating coordinative environment.

The reaction of 1a,b(syn,anti) with approximately one equivalent of 1% Na/Hg in THF at room temperature resulted in the formation of green isomeric mixtures of Mo(I)(NO)Cl₂(PN^HP) complexes (2a(syn,anti)) or of W(I)(NO)Cl₂(PN^HP) (2b(syn,anti)) in 74% and 78% yields, respectively (Scheme 1). 2a,b(syn,anti) were found quite soluble in THF and CH₂Cl₂, but were sparingly soluble in toluene and benzene. In the ³¹P{¹H} and ¹H NMR spectra of the isomeric mixtures of 2a,b(syn,anti) no signals could be observed due to the paramagnetic nature of these complexes. The inaccessibility of ¹H and ³¹P{¹H} NMR spectra impeded estimates of spectroscopic yields of the syn and anti isomers in solution. Thus, the isomeric mixtures of 2a,b(syn,anti) were characterized by IR and EPR spectroscopy, elemental analyses and X-ray diffraction studies. Sharp bands at 1595 and 1547 cm⁻¹ for the isomeric mixtures of 2a,b(syn,anti) were assigned to ν_NO vibrations, respectively, which appeared at lower wave numbers than those of 1a,b(syn,anti) indicating a higher degree of back-donation from the metal center to the nitrosyl ligand. Single crystals suitable for X-ray diffraction studies were grown from the isomeric mixtures of 2a,b(syn,anti) from concentrated toluene–pentane mixtures at −30 °C. The crystal structure determination revealed preferential crystallization of the 2a(anti) and 2b(syn) isomers. Both of them crystallized in the orthorhombic space group P2₁2₁2₁. The molecular structures revealed neutral pseudo-octahedral complexes. Perspective views with selected bond distances and bond angles of 2a(anti) and 2b(syn) are shown in Fig. 4. The two phosphorus atoms and the nitrogen atom of the PN^HP ligand and the two chloride ligands attached to the metal center occupy “equatorial” positions in both cases. The H_N proton and the nitrosyl ligand were found to be disposed anti indicating the presence of the anti2a(anti) isomer. No NO/Cl disorder was found in the structure of 2a(anti), but despite this, the presence of isomeric mixtures in solution with one prevailing component could not fully be ruled out.

Fig. 4 Molecular structures of M(NO)Cl₂(PN^HP), {M = Mo; 2a(anti) (left); W, 2b(syn) (right)} Thermal ellipsoids are drawn at the 50% probability level. Selected bond distances (Å) and bond angles (°) for 2a(anti): Mo1–N2 2.255(2), Mo1–N1 1.855(3), Mo1–Cl1 2.5021(7), Mo1–P1 2.5372(7), N2–H2 0.91, P2–Mo1–P 1156.25(2), P2–Mo1–Cl2 100.64(2), P2–Mo1–Cl1 87.04(2), Mo1–N1–O1 178.7(3). Selected bond distances(Å) and bond angles (°) for 2b(syn): W1–N2 2.232(4), W1–N1B 1.811(19), W1–P1 2.5123(13), W1–Cl2B 2.381(6), N2–H 0.79 (4), P2–W1–P1 157.23(4), P2–W1–N2 78.82(13), W1–N1B–O1B 178(3), N1B–W1–Cl2B 172.2(8).

On the other hand the crystal structure analysis of 2b(syn) revealed that the structure had trans NO/Cl disorder over two positions with site occupancy factors of 0.698(8) and 0.302(8). This NO/Cl disorder again supports the presence of the isomeric mixtures with respect to the H_N position and the nitrosyl ligand and the presence of the syn and anti isomer in an approximately 7 : 3 ratio in the solid state. An inversion at the N atom with prototopic rearrangement at this atom seemed at least not to occur in the solid state. This would be also in accord with the assignment of the same syn/anti ratio for the precursors 1a,b(syn,anti). The isomeric mixtures of 2a,b(syn,anti) are rare paramagnetic complexes with the metal centers in +I oxidation state and of low-spin d⁵ configurations. To probe the paramagnetic nature, solution EPR measurements were carried out on 2a,b(syn,anti) in toluene at room temperature. The spectra are displayed in the ESI section (Fig. S1†).

We then attempted deprotonation of the H_N proton of the PN^HP ligand backbone of 2a,b(syn,anti) employing Na[N(SiMe₃)₂], which led to decomposition and the solution turned black immediately. Moreover, attempts were also undertaken to prepare the previously reported M(NO)(CO)(PNP) {M = Mo, 3a; W, 3b} amides^23avia a new synthetic route starting from the 2a,b(syn,anti).

Reduction of 2a,b(syn,anti) in the presence of 1 bar CO and 1% Na/Hg at room temperature showed the formation of several unidentified species instead of the expected unique complex of an isomeric carbonyl chloride M(NO)(CO)Cl(PN^HP) (M = Mo, W), which prevented concomitant deprotonation to obtain the requested M(NO)(CO)(PNP) (M = Mo, 3a; W, 3b).

Nevertheless M(NO)(CO)(PNP) 3a,b could be prepared according to our previously published procedure^23a starting from M(NO)(CO)₄(ClAlCl₃) and PN^HP ligand followed by deprotonation with NaN(SiMe₃)₂ and tested their reactivity with CO₂ and their hydrogenation capability.

Reactivity of the M(NO)(CO)(PNP) {M = Mo, 3a W, 3b} amide complexes with CO₂

Reaction of M(NO)(CO)(PNP) {M = Mo, 3a W, 3b; PNP = N(CH₂CH₂PiPr₂)₂} with CO₂ (2 bar) at room temperature in THF led to formation of the 2 + 2 addition products across the M=N bonds, the carbamate complexes Mo,W(NO)(CO)(PNP)(OCO) 4a,b(trans) (Scheme 2). The specification “trans” refers to the orientation of the CO₂ approach transoid to the NO ligand of 3a,b. The orange 4a,b(trans) complexes could be isolated in pure form in quantitative yields. Monitoring the CO₂ addition reactions visually, a purple color appeared indicating the presence of short-lived intermediates, which were not isolable and are proposed to be charge transfer complexes. This type of reaction of CO₂ with 3a,b to form carbamates 4a,b(trans) is closely related to the 2 + 2 cycloaddition reactions of early transition metal imido complexes with alkynes, olefins and ketones.²⁶ Hessen and coworkers demonstrated an example of a thermodynamically favored insertion of alkynes into the V–amide bond of cationic amido imido vanadium complexes rather than a 2 + 2 cycloaddition reaction.²⁷ Therefore, an insertion reaction of CO₂ across the M–amide bond could alternatively be envisaged. Nevertheless, this CO₂ activation product is unique in its structure and is therefore in its way of formation presumed to be distinct from the insertion reactions of early transition M–N bonds.

Scheme 2 CO₂ activation by molybdenum and tungsten amides 3a,b and hydrides 5a,b(cis,trans).

The ³¹P{¹H} NMR spectra of 4a,b(trans) at room temperature exhibited sharp singlets at δ 61.9 and 56.4 ppm, respectively, indicating equivalence of the phosphorus atoms of the metal attached PNP ligand and under the given conditions also the absence of the theoretically possible isomeric products 4a,b(cis) with the CO₂ approach from the NO side of 3a,b. When the reactions of 3a,b with CO₂ (2 bar) were carried out at −60 °C and pursued by ³¹P{¹H} NMR spectroscopy, additional weak signals were seen at δ 62.7 and at 57.9 ppm besides those of 4a,b(trans) indicating formation of kinetic mixtures of the 4a,b(trans,cis) isomers (7 : 3 ratios in both cases of compounds). These mixtures equilibrated at room temperature into the thermodynamically more stable 4a,b(trans) as the sole products based on the reversibility of the CO₂ additions to 3a,b from the NO side. 4a,b(cis) could thus not be observed by ³¹P{¹H} NMR spectroscopy at room temperature in solution and could also not be isolated (Scheme 2).

The ¹H NMR spectra of 4a,b(trans) showed several distinct signals for the methyl, methylene and methyne protons in the expected region supporting also the absence of isomers. The IR spectra of 4a,b(trans) displayed ν(NO) and ν(CO) bands at 1593, 1554 cm⁻¹ and at 1902, 1879 cm⁻¹, respectively. Additional strong absorptions appeared for 4a,b(trans) at 1732 and 1741 cm⁻¹, respectively, and were assigned to the ν_CO2 vibration of the attached CO₂ molecule. In the ¹³C{¹H} NMR spectra of 4a,b(trans) sharp singlets at δ 156.4 and 159.3 ppm were attributed to the C_CO2 atoms along with multiplets at δ 247 and 250.6 ppm for the C_CO atoms of the carbonyl ligands.

The C_CH2 atoms adjacent to the nitrogen atoms of the PNP ligand of 4a,b(trans) were observed as virtual triplets at δ 53.2 (t, ^vJ_CP = 3.6 Hz) and at 57.7 (t, ^vJ_C–P = 3.6 Hz) ppm, respectively. Single crystals of 4b(trans) suitable for an X-ray diffraction study were obtained by layering pentane over a concentrated THF solution. The structural model obtained from the diffraction study is depicted in Fig. 5, which revealed a pseudo octahedral geometry. The W1–N2 bond distance was found to be 2.2603(19) Å. Upon CO₂ addition it became significantly elongated by about 0.2 Å with respect to the approximate W=N bond of 3b.^23a

Fig. 5 Molecular structure of 4b(trans). Thermal ellipsoids are drawn at the 50% probability level. All hydrogen atoms are omitted for clarity. Selected bond distances [Å] and bond angles [°]: W1–N1 1.789(2), W1–C1 1.969(3), W1–N2 2.2603(19), W1–O3 2.2155(18), C1–O1 1.159(3), C2–O3 1.283(3), C2–O4 1.208(3), N1–W1–C1 88.61(10), N2–W1–O3 172.47(8), N2–C2–O3 108.0 (2) P1–W1–P2 157.60(2).

The transition state of the reaction between W(NO)(CO)(PNP) 3b and CO₂ could be modelled by DFT calculations (see ESI†).

Reasonably low activation energies ΔG^‡ = 14.4 kcal mol⁻¹ and 14.1 kcal mol⁻¹ were calculated for the CO and NO side approaches of CO₂ onto 3b to give 4b(trans) and 4b(cis), respectively. Remarkably, both approaches are kinetically almost not distinct. The free enthalpies of both reactions were determined to be −7.7 kcal mol⁻¹ (4b(trans)) and −5.6 kcal mol⁻¹ (4b(cis)) indicating that 4b(trans) is the thermodynamically more favored and stable product. The calculated lower free energy value for 4b(cis) looks plausible in view of the fact that the CO₂ approaches from the NO side are experimentally equilibria. The calculated activation barrier for the reverse reaction from 4b(cis) to form 3b and CO₂ seems thus to be still in energetic reach for a thermal process at room temperature (ΔG = 19.8 kcal mol⁻¹). Drawings of the TS structure of the CO₂ approach to 3b (Fig. 6 (left)) and the HOMO of the TS (Fig. 6 (right)) are shown in Fig. 6 (see also ESI†). According to the calculations the TS consists of a strongly distorted TS square with W–N, N–C, C–O and O–W contact lengths of 2.144, 2.399, 1.184 and 2.780 Å, respectively. These bond lengths mark an early transition state of the 2e + 2e reaction. The C–N interaction is the main new contact in the transition state, which is also represented in Fig. 6 (right), where large orbital coefficients appear between the C(CO₂) and the N atoms. The other new contact of the O(CO₂) atom to the W center is still very distant. Despite this feature that asymmetry prevails in the bond forming process, the reaction coordinate was found to be continuous and not two-stage without intermediate generated by consecutive bond formations.

Fig. 6 Geometry of the transition state (TS) of the approach of CO₂ to 3b (left) and the TS HOMO−3 orbital for the CO₂ 2e + 2e addition reaction (right).

Reactivity of the hydride complexes M(NO)H(CO)(PN^HP) (M = Mo, 5a(cis,trans); W, 5b(cis,trans)) with CO₂

Then we also tested the CO₂ reactivity of the previously reported^23a isomeric mixtures of the amine hydride compounds M(NO)H(CO)(PN^HP) (5a,b(cis,trans)), which were obtained by H₂ addition to 3a,b from both sides of the M=N bond. The cis and trans notation indicate trans and cis attack of the H₂ molecule with respect to the NO ligand leading to the trans H/NO complexes (5a,b(trans)) or to the trans H/CO complexes (5a,b(cis)). However, the reactions to 5a(cis,trans) with H₂ were found to be equilibria showing at room temperature an approximate 1 : 1 : 1 mixture of 5a(cis)/5a(trans)/3a. Therefore, the reaction with CO₂ (2 bar) had to be carried out using the in situ generated mixture of Mo(NO)H(CO)(PN^HP) (5a(cis,trans)), which led to the formate products Mo(NO)(CO)(PN^HP)(η¹-OCHO) 6a(cis,trans) and the cyclic carbamate species 4a(trans) in an approx. 1 : 1 : 1 ratio as indicated by ¹³C{¹H} and ³¹P{¹H} NMR spectroscopy (Scheme 2). Since the three different kinds of reactions (2 kinds of CO₂ insertions into the M–H bonds and CO₂ addition) did not reveal a change in the ratios from the starting materials to the products, we have to assume that all involved reaction rates are about the same. The formate protons of 6a(trans) and 6a(cis) appeared in the ¹H NMR spectrum at δ 8.1 and 8.5 ppm, respectively (1 : 1 ratio). Signals at δ 170 and 168 ppm in the ¹³C{¹H} NMR spectra were assigned to the C_formate nuclei of the 6a(cis,trans) isomers, respectively. 6a(cis,trans) and 4a(trans) were not in equilibrium, but despite this attempts to separate these complexes turned out to be unsuccessful.

Similarly, the reaction of the 5b(cis,trans) mixture, which is not an equilibrium reaction allowing this mixture to be isolated, with CO₂ resulted at room temperature in the rapid formation of the isomeric η¹-formate complexes W(NO)(CO)(PN^HP)(η¹-OCHO) (6a,b(trans))²⁸ (Scheme 2) appearing in a 9 : 1 ratio. A strong signal at δ 51.3 ppm (with tungsten satellites) in the ³¹P{¹H} NMR spectrum was assigned to 6b(trans) along with a weak signal at δ 52.6 ppm for 6b(cis). In the ¹H NMR spectrum the formate protons of 6b(trans,cis) were observed at δ 8.5 and 8.0 ppm, respectively, and signals at δ 244.7 and 164.3 ppm in the ¹³C(¹H) NMR spectra were attributed to the carbonyl ligand and the C_formate atom of the major isomer 6b(trans). Supposedly due to a too low concentration of 6b(cis), the ¹³C NMR resonances of the carbonyl and formate carbon atoms of the minor 6b(cis) isomer could not be observed. In the IR spectrum of the 6b(cis,trans) mixture a band at 1612 cm⁻¹ was attributed to the ν_as(CO₂) vibration in accord with earlier assignments of η¹-O-formato tungsten complexes.²⁸ Additional IR bands at 1863 and 1578 cm⁻¹ of the 6b(cis,trans) mixture were assigned to the ν(CO) and ν(NO) ligand vibrations of both compounds showing that both compounds cannot be distinguished in the 1500 to 1900 cm⁻¹ region of the IR spectra.

Despite the fact that 6b(cis) was the minor constituent of the isomeric mixture in solution, tiny crystals of this minor component could precipitated from solution by slow diffusion of pentane into a concentrated THF solution. The X-ray diffraction study revealed a pseudo-octahedral framework with cis nitrosyl and formate ligands (Fig. 7). The asymmetric unit of this minor isomer 6b(cis) contained two crystallographically independent tungsten species displaying intermolecular hydrogen bonding³¹ between the protic H_N atom of one molecule and the O_formato of an adjacent molecule forming linear chains in the three dimensional lattice. It should be noted here that the H_N atom and the formate group were found to be disposed anti in the solid state structure, which was somewhat surprising when cis addition of H₂ to 3a,b is taken into consideration, followed by CO₂ insertion into the M–H bond occurring with retention of the configuration at the metal centers (Scheme 2). One possible reason for the thermodynamically favored re-orientation of the H_N atom by a prototopic rearrangement could be the adoption of an energetically preferred intermolecular hydrogen bonding network requiring anti disposition of the H_N atom and the formate ligand in the crystal lattice.

Fig. 7 Molecular structure of 6b(cis). Thermal ellipsoids are drawn at the 50% probability level. All hydrogen atoms are omitted for clarity except H1. Selected bond distances [Å] and bond angles [°]: W1–N1 2.287(5), W1–C1 1.930(8), W1–N2 1.792(6), W1–O3 2.185(5), C1–O1 1.172(8), C2–O3 1.271(7), C2–O4 1.213(8), N1–W1–C1 91.6(3), N2–W1–O3 101.9(2), N1–W1–N2 178.3 (3) P1–W1–P2 158.32(7).

In addition to these findings DFT calculations support the observation that the major trans NO/formate isomer 6b(trans) is thermodynamically more stable than the trans CO/formate isomer 6b(cis) by ΔE = −4.4 kcal mol⁻¹ (see Scheme 2 and ESI†), which seems to match well reality. The fact that in the crystallization experiment of the 6b(cis,trans) mixture 6b(cis) was found to be less soluble than 6b(trans) can now be attributed to a secondary crystallization effect based on the mentioned hydrogen bonding system in the solid state causing lower solubility of the minor constituent in solution. In addition it should be noted that the proton at N1 and the formate ligand of the structure of 6b(cis) are in anti position, which somewhat contradicts the fact that in 5a,b(trans) and 5a,b(cis) they are expected to be syn due to the Z stereochemistry of the H₂ addition and the expected retention of the configuration at the metal center associated with the CO₂ insertion process. Therefore the proton at N1 is assumed to have undergone an inversion process when forming the crystals of 6b(cis), which seems likely on the basis of our DFT calculations, which revealed only minor energetic differences (approx. 2 kcal mol⁻¹) between the syn and anti structures in favour of the anti arrangements.

Stoichiometric hydrogenation of carbon dioxide

The 6a,b(cis,trans) mixtures can be deprotonated at the H_N atom employing 1 equiv. of Na[N(SiMe₃)₂] at room temperature with regeneration of 3a,b. Sodium formate was formed in addition in accord with eqn (1) closing thus a stoichiometric cycle of consecutive reaction steps of H₂ and CO₂ additions in a “pseudo-catalytic” reaction course. The permanent presence of a base seemed not to interfere with 6b(cis,trans) and to initiate side reactions, but the preferential addition of CO₂ over the H₂ addition to 3a,b seemed in the presence of a H₂–CO₂ mixture a great obstacle to establish a genuine catalytic reaction course. Therefore, if catalysis should be accomplishable, it could be envisaged at higher temperatures making the CO₂ addition to some extent reversible enabling thus under catalytic conditions H₂ before CO₂ addition to 3a,b (Scheme 3).

Scheme 3 Schematic cycle for the CO₂ hydrogenation using 3a,b.

Being aware of this situation, we attempted catalysis and carried out at 140 °C 3 types of hydrogenation experiments of CO₂ applying the simultaneous presence of H₂ (70 bar) and CO₂ (10 bar) and of catalytic amounts of 3a,b or of the 5b(cis,trans) mixture (5 mol% loading) in the presence of Na[N(SiMe₃)₂] as stoichiometric agent. After 15 h all the experiments revealed formation of HCOONa according to eqn (1), but the yields were low, only 4%, 2% and 3%, respectively, as examined by ¹H NMR (DMF as internal standard) (Table 1). To improve the yield of the formate salt, we varied the type of the base using DBU, KOtBu or NEt₃ in the presence of 3a as a catalyst with otherwise the same conditions as above. However, the yield of the formate salts [HCOO][DBUH] (3.5%), HCOOK (2.5%) or HCOOH–NEt₃ (0.5%), did not increase (Table 1). Furthermore, the addition of B(C₆F₅)₃ or the Et₃SiH–B(C₆F₅)₃ mixture as co-catalysts led in the presence of 3a and Na[N(SiMe₃)₂] even to a decrease in yields.

Table 1 Hydrogenation of CO₂

Entry^a	Cat	Base	Time (h)	Yield^b (%)
a Unless and otherwise stated 10 bar CO2, 70 bar H2, THF as solvent, 140 °C temperature and 5 mol% catalyst with respect to base were used as reaction conditions. b Yield on the basis of ^1H NMR integration using DMF as internal standard. c No catalyst was used.
1	3a	Na[N(SiMe3)2]	15	4
2	3b	Na[N(SiMe3)2]	15	2
3	5b(cis,trans)	Na[N(SiMe3)2]	15	3
4	3a	DBU	15	3.5
5	3a	KOtBu	15	2.5
6	3a	Et3N	15	0.50
7^c	—	Na[N(SiMe3)2]	17	0.14
8	3a	TMP	15	—
9	3a/BCF	Na[N(SiMe3)2]	15	0.70
10	3a/Et3SiH/BCF	Na[N(SiMe3)2]	22	0.30

---

At this stage it can be stated that CO₂ hydrogenation is in principle possible, but for effective such catalyses further tuning efforts have to be taken into consideration, presumably by attempts to make the CO₂ adducts of type 4 more labile. It should be mentioned at this point that since strong bases are known to react with CO₂ particularly at high temperatures and pressures, a control experiment was also carried out in the absence of the metal catalyst applying only Na[N(SiMe₃)₂] as the catalyst and keeping all the other conditions the same. The yield of sodium formate was in this case only 0.14%, which is significantly lower than the yields obtained in the presence of 3a,b.

It deserves mentioning that after the CO₂ hydrogenation experiments (applying especially the NaN(SiMe₃)₂ base), the reaction mixtures were found to contain a considerable amount of white precipitate in the THF reaction mixture. When THF was evaporated in vacuo and the residue was dissolved in D₂O to quantify the yield of the formate salt by ¹H NMR spectroscopy (DMF internal standard), the solution was still found to contain white particles even after addition of excess D₂O. Therefore, formation of another species is anticipated to be produced in the hydrogenation of CO₂ (note that the product sodium formate is expected to be soluble in D₂O). The nature of this species could however not be unravelled, but in principle it should be added to the overall yield of these catalytic reactions.

Conclusions

In conclusion we have reported a new synthetic route for the preparation of isomeric diamagnetic M(NO)Cl₃(PN^HP) (M = Mo, 1a(syn,anti); W, 1b(syn,anti)) complexes and their reduction to rare paramagnetic species M(NO)Cl₂(PN^HP) {M = Mo, 2a(syn,anti); W, 2b(syn,anti)}. Furthermore, we have shown that the excessively reduced pentacoordinate amides Mo,W(0)(PNP)(NO)(CO) (3a,b; PNP = (iPr₂PCH₂CH₂)₂N)) can activate CO₂ in a metal–ligand assisted way via 2e + 2e additions across the M=N bonds forming the pseudo-octahedral 4a,b(trans) compounds. The CO₂ approach occurs thermodynamically controlled from the CO ligand side. Kinetically there is practically no distinction between the sides of attack. The amine hydride species M(NO)(CO)H(PN^HP) (M = Mo, 5a(cis,trans) W, 5b(cis,trans)) formed by H₂ addition to 3a,b reacted with CO₂ at room temperature to produce η¹-formato complexes M(NO)(CO)(PN^HP)(η¹-OCHO) {M = Mo, 6a(cis,trans); W, 6b(cis,trans)}. Stoichiometric hydrogenations of CO₂ to formate salts could thus be accomplished via elimination of the formate ligands of type 6 complexes induced through the presence of a strong base regenerating the amides 3a,b. Catalytic hydrogenation of CO₂ was then also approached at somewhat elevated temperatures and pressures. However, the apparently too high stability of the CO₂ addition products of type 4 are anticipated to largely block the catalytic reaction course and cause low yields. This conclusion rendered the idea that in these Mo and W systems catalysis could eventually be achieved by tuning the 4 type complexes for reversibility in CO₂ additions. It is finally worth mentioning that this study demonstrated for the first time that reductions of CO₂ are within reach utilizing middle, non-noble transition metal compounds as catalysts.

Experimental section

General procedures

All experiments were carried out under an atmosphere of nitrogen using either dry glove box or Schlenk techniques. Reagent grade solvents benzene, tetrahydrofuran, pentane, toluene, diethyl ether were dried with sodium benzophenone and distilled prior to use under N₂ atmosphere. CH₂Cl₂ and Et₃N were dried over calcium hydride and distilled. Deuterated solvents were dried with sodium benzophenone ketyl (THF-d₈, toluene-d₈ and C₆D₆) and calcium hydride (CD₂Cl₂) and distilled via freeze–pump–thaw cycle prior to use. The M(NO)Cl₃(NCMe)₂,²⁹ M(NO)(CO)(PNP),^23a {M = Mo, W; PNP = N(CH₂CH₂iPr₂)₂} complexes and HN(CH₂CH₂ⁱPr₂)₂ ³⁰ were prepared according to literature procedures. KOtBu, Na[N(SiMe₃)₂] and DBU were purchased from commercially available sources and used without further purification. The NMR spectra were measured with a Varian Mercury 200 spectrometer (at 200.1 MHz for ¹H, at 81.0 MHz for ³¹P), with Varian Gemini-300 instrument (¹H at 300.1 MHz, ¹³C at 75.4 MHz), with Bruker-DRX 500 spectrometer (500.2 MHz for ¹H, 202.5 MHz for ³¹P, 125.8 MHz for ¹³C) and Bruker-DRX 400 spectrometer (400.1 MHz for ¹H, 162.0 MHz for ³¹P, 100.6 MHz for ¹³C). All chemical shifts for ¹H and ¹³C{¹H} are expressed in ppm relative to tetramethylsilane (TMS) and for ³¹P{¹H} relative to 85% H₃PO₄ as an external standard reference. Signal patterns are as followed: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; v, virtual triplet. IR spectra were obtained either ATR or KBr methods using Bio-rad FTS-45 instrument. Elemental analyses were carried out at Anorganisch-Chemisches Institut of the University of Zurich.

Preparation of isomeric mixture of Mo(NO)Cl₃(PN^HP), 1a(syn,anti)

To a solution of Mo(NO)Cl₃(NCMe)₂ (0.200 g, 0.636 mmol) in 10 mL of THF, 0.16 g (196 mg, 0.64 mmol) (iPr₂PCH₂CH₂)₂NH ligand in 5 mL THF was added. The resulting mixture was kept stirring overnight to obtain a light green precipitate and the solution becomes dark green. Then the solvent was separated from the precipitate and the precipitate was washed with minimum amount of THF and extracted with dichloromethane and dried in vacuo to obtain the powdery isomeric mixture of 1a(syn,anti) in 55% yield.

¹H NMR (400 MHz, CD₂Cl₂): δ = 4.4 (broad singlet, NH, 1a(syn)), 3.63–3.5 (m, NCH₂), 3.23–3.21 (m, NCH₂), 3.1–3.02 (m, CH), 2.3–2.2 (m, –PCH₂), 2.05 (m, PCH₂), 1.52–1.39 (m, CH₃, 1a(syn)), 1.34–1.30 (m, CH₃, 1a(anti)). ¹³C{¹H} NMR (100.6 MHz, CD₂Cl₂): δ = 55.9 (t, ^vJ_C–P = 3.6 Hz, NCH₂, 1a(syn)), 29 (m, CH), 28 (m, PCH₂), 21 (m, CH₃), 20.4 (m, CH₃). ³¹P{¹H} NMR (400 MHz, CD₂Cl₂): 69.3 (s, ⁱPr₂P, 1a(syn)), 71 (s, 1a(anti)). IR (cm⁻¹, ATR): 1653 (ν_NO), expected m/z: 538, observed m/z: 538.2. Anal. Calcd for C₁₆H₃₇Cl₃MoN₂OP₂: C, 35.74; H, 6.94; N, 5.21. Found: C, 36.03; H, 6.39; N, 5.67.

Preparation of isomeric mixture of W(NO)Cl₃(PN^HP), 1b(syn,anti)

WCl₃(NO)(NCMe)₂ (200 mg, 0.5 mmol) was dissolved in 10 mL of THF. 0.160 g (0.52 mmol) of aminodiphosphine (iPr₂PCH₂CH₂)₂NH was added via a syringe. After a few minutes stirring a light red precipitate was observed and the reaction was allowed to stir overnight. Then the solution was removed partially under vacuum and the reddish residue was filtered off and washed with THF. The obtained residue was extracted with dichloromethane and the solvent was removed in vacuo to obtained the isomeric mixture of 1b(syn,anti) in 62% yield. ¹H NMR (400 MHz, CD₂Cl₂): δ = 4.47 (broad singlet, NH, 1b(syn)), 5.5 (broad singlet, NH, 1b(anti)) 3.71–3.57 (m, NCH₂, syn,anti), 3.19–3.05 (m, CH), 2.30–2.23 (m, –PCH₂), 2.09–2.06 (m, PCH₂), 1.47–1.19 (m, CH₃). ¹³C{¹H} NMR (100.6 MHz, CD₂Cl₂): δ = 55.9 (t, ^vJ_C–P = 5.9 Hz, NCH₂, 1b(anti)), 55 (t, ^vJ_C–P = 4.8 Hz, NCH₂, 1b(anti)), 27.8 (m, CH), 27.3 (m, –PCH₂), 20 (m, CH₃). ³¹P{¹H} NMR (400 MHz, CD₂Cl₂): 56.8 [s, ¹J_P–W (d, satellite) = 184.8 Hz, 1b(syn)], 58.3 [s, ¹J_P–W (d, satellite) = 189.3 Hz, 1b(anti)]. IR (cm⁻¹, ATR): 1596 (ν_NO), Anal. Calcd for C₁₆H₃₇Cl₃N₂OP₂W: C, 30.72; H, 5.96; N, 4.48. Found: C, 30.72; H, 5.88; N, 4.55.

Preparation of Mo(NO)Cl₂(PN^HP), 2a(syn,anti)

The isomeric mixture of 1a(syn,anti) (0.030 g, 0.056 mmol) was added to a stirred suspension of approximately 1% sodium amalgam (0.0013 g, 0.057 mmol) in 15 mL THF. Stirring was continued overnight at room temperature to ensure the completion of the reaction. The final supernatant green solution was filtered off from the mercury containing residue and evaporated to dryness. The resulting residue was washed with pentane and extracted with THF. Finally the isomeric mixture of 2a(syn,anti) was obtained as a green powder after removal of THF in vacuo. Yield 74%. IR (cm⁻¹, ATR): 1595 (ν_NO), Anal. Calcd for C₁₆H₃₇Cl₂MoN₂OP₂: C, 38.26; H, 7.42; N, 5.58. Found: C, 37.88; H, 7.12; N, 5.29.

Preparation of W(NO)Cl₂(PN^HP), 2b(syn,anti)

0.075 g (0.12 mmol) of the isomeric mixture of 1b(syn,anti) was added to a stirred suspension of 1% sodium amalgam (0.003 g, 0.13 mmol) in 15 mL THF. Stirring was continued overnight at room temperature to ensure the completion of the reaction. The final supernatant solution was filtered off from the mercury containing residue and evaporated to dryness. The residue was washed with pentane and extracted with THF and later with toluene. Recrystallisation from cold toluene–pentane solution afforded green crystals of 2b(syn,anti). Yield 78%. IR (cm⁻¹, ATR): 1547 (ν_NO). Anal. Calcd for W(NO)Cl₂(PNP): C, 32.56; H, 6.32; N, 4.75. Found: C, 32.71; H, 6.22; N, 4.78.

General procedure for the preparation of M(NO)(CO)(PNP)(OCO), (M = Mo, 4a(trans); W, 4b(trans))

M(NO)(CO)(PNP) (0.065 mmol, M = Mo, 3a; W, 3b) was dissolved in 0.5 mL THF-d₈ in a J. Young NMR tube. The tube was taken out from the glove box and frozen with liquid nitrogen. The nitrogen atmosphere was removed via a freeze–pump–thaw cycle and the tube was filled with 2 bar of CO₂ and sealed. Formation of 4a(trans) or 4b(trans) took place immediately within five minutes in quantitative yield.

4a(trans). ¹H NMR (400 MHz, THF-d₈): δ = 3.47–3.35 (m, 2H, –NCH₂), 2.85–2.78 (m, 2H, –NCH₂), 2.42–2.36 (m, 2H, CH), 2.25–2.17(m, 2H, CH), 2.08–2.03 (m, 2H, CH₂), 1.81–1.72 (m, 2H, –PCH₂), 1.35–1.28 (m, 12H, CH₃), 1.23–1.15 (m, 12H, CH₃). ¹³C{¹H} NMR (100.62 MHz THF-d₈): δ = 247 (m, CO), 156.4 (s, CO₂₎, 53.2 (t, ^vJ_CP = 3.6 Hz, NCH₂), 23.79 (s, CH), 19.83 (t, ^vJ_C–P = 6 Hz, CH₂), 16.59 (s, CH₃), 16.21 (s, CH₃), 15.74 (s, CH₃). ³¹P{¹H} NMR (161.97 MHz, THF-d₈): δ = 61.9 (s). IR (cm⁻¹, KBr): 1593 (ν_NO), 1732 (ν_CO), 1902 (ν_CO). Anal. Calcd for C₁₈H₃₆MoN₂O₄P₂: C, 43.03; H, 7.22; N, 5.58. Found: C, 42.67; H, 7.28; N, 5.34.

4b(trans). ¹H NMR (400 MHz, THF-d₈): δ = 3.58–3.49 (m, 2H, –NCH₂), 2.92–2.85 (m, 2H, –NCH₂), 2.52–2.46 (m, 2H, CH), 2.28–2.20(m, 2H, CH), 2.11–2.06 (m, 2H, –PCH₂), 1.81–1.76 (m, 2H, –PCH₂), 1.37–1.29 (m, 12H, –CH₃), 1.25–1.11 (m, 12H, –CH₃). ¹³C{¹H} NMR (75.45 MHz, THF-d₈): δ = 250.6 (m, CO), 159.3 (s, CO₂₎, 57.7 (t, ^vJ_C–P = 3.6 Hz, NCH₂), 27.53 (t, ^vJ_CP = 10.9 Hz, CH), 26.87 (t, ^vJ_CP = 12.2 Hz, CH), 23.62 (t, ^vJ_CP = 7.7 Hz, PCH₂), 19.66 (t, ^vJ_CP = 5.9 Hz, CH₃), 19.05(s, CH₃), 18.61 (s, CH₃). ³¹P{¹H} NMR (161.97 MHz, THF-d₈): 56.4 [s, ¹J_PW (d, satellite) = 308.2 Hz]. IR (cm⁻¹, KBr): 1554 (ν_NO), 1741 (ν_CO2), 1879 (ν_CO). Anal. Calcd for C₁₈H₃₆N₂O₄P₂W: C, 36.63; H, 6.15; N, 4.75. Found: C, 36.29; H, 6.27; N, 4.72.

Reaction of 3a and 3b with CO₂ at 213 K. M(NO)(CO)(PNP) (0.03 mmol) (M = Mo, 3a; W; 3b) was dissolved in 0.5 mL THF-d₈ in a J. Young NMR tube. The tube was taken out of the glove box and frozen with liquid nitrogen. The nitrogen atmosphere was removed via freeze–pump–thaw cycle. The tube was filled with 2 bar of CO₂ under frozen condition and sealed. Then the tube was immediately placed in a 300 MHz NMR spectrometer preset at −60 °C and ³¹P{¹H} NMR spectra was recorded.

³¹P{¹H} NMR data for the reaction of 3a with CO₂ at 213 K: ³¹P{¹H} NMR (121.48 MHz, 213 K, THF-d₈): δ = 63.1 (s, formation of 4a(trans)), 62.7 (s, formation of 4a(cis)).

³¹P{¹H} NMR data for the reaction of 3b with CO₂ at 213 K: ³¹P{¹H} NMR (121.48 MHz, 213 K, THF-d₈): δ = 58.1 (s, formation of 4b(trans)), 57.9 (s, formation of 4b(cis)).

Preparation of Mo(NO)(CO)(PN^HP)(η¹-OCHO), {mixture of 6a(cis) and 6a(trans)}. 3a (0.015 g, 0.033 mmol) in 0.5 mL THF was frozen in a J. Young NMR tube. The nitrogen atmosphere was removed via a freeze–pump–thaw cycle. Then the tube was filled with 2 bar of H₂ and sealed and kept for three days to allow formation of 5a(cis) and 5a(trans). After formation of the equilibrium mixture of hydride complexes 5a(cis) and 5a(trans) along with 3a (1 : 1 : 1), the hydrogen pressure was slowly released and CO₂ (2 bar) was purged into that mixture instantaneously. Immediate formation of 6a(cis) and 6a(trans) was observed along with 4a(trans) in a ratio of 1 : 1 : 1 as revealed by the ³¹P{¹H} NMR.

Selected spectroscopic data: ¹H NMR (400.1 MHz, THF-d₈) data: δ = 8.5 (s, 1H, OCH, 6a(cis)), 8.1 (s, 1H, OCHO, 6a(trans)). ¹³C{¹H} NMR (100.6 MHz, THF-d₈): δ 248 (s, CO, 6a(trans)), 246.5 (s, CO, 6a(cis)), 170 (s, OCHO, 6a(cis)), 168.8 (s, OCHO, 6a(trans)), 156 (s, CO₂, 4a(trans)). ³¹P{¹H} NMR (161.97 MHz, THF-d₈): 60.9 (s, 6a(trans)), 60.0 (s, 6b(cis)). Since the resulting product contains mixtures of 6a(cis), 6a(trans) and 4a(trans) which could not be separated, elemental analyses could not be provided.

W(NO)(CO)(PN^HP)(η¹-OCHO), {mixture of 6b(cis) and 6b(trans)}. A solution of the mixture of 5b(cis) and 5b(trans) (0.020 g, 0.036 mmol) in 0.5 mL THF was frozen in a J. Young NMR tube. The nitrogen atmosphere was removed via a freeze–pump–thaw cycle. Then the tube was filled with 2 bar of CO₂ and sealed. Formation of 6b(cis) and 6b(trans) took place immediately upon shaking the Young NMR tube in quantitative yield in a 1 : 9 ratio. Suitable orange crystals were obtained for diffraction upon layering pentane to a concentrated THF solution of the product. Yield (by ³¹P{¹H} NMR): 10% (6b(cis)), 90% (6b(trans)). ¹H NMR (400.1 MHz, THF-d₈): δ = 8.52 (s, 1H, OCH, 6b(trans)), 8.0 (s, 1H, OCH, 6b(cis)), 2.93 (m, 4H, –NCH₂), 2.41–2.26 (m, 4H, –CH), 1.66–1.62 (m, 4H, CH₂P), 1.35–1.25 (m, 24H, CH₃). ¹³C{¹H} NMR (100.6 MHz, THF-d₈): δ 244.7 (s, CO), 164.3 (s, CO₂H₎, 57.7 (s, NCH₂), 24.5 (t, ^vJ_CP = 13.1 Hz, CH), 24.01 (t, ^vJ_CP = 9.5 Hz, CH), 19.08 (t, ^VJ_CP = 8.3 Hz, CH₂), 17.2 (s, CH₃), 14.80 (s, CH₃). ³¹P{¹H} NMR (161.97 MHz, THF-d₈): 51.3 (s, 6b(trans)), 52.6 (s, 6b(cis)). IR (cm⁻¹, KBr): 1578 (ν_NO), 1612 (ν_OCHO), 1863 (ν_CO). Anal. Calcd for C₁₈H₃₈N₂O₄P₂W: C, 36.50; H, 6.47; N, 4.73. Found: C, 36.25; H, 6.41; N, 4.53.

General procedure for the reaction of the formate complexes (mixture of 6a(cis) and 6a(trans)) and (mixture of 6b(cis) and 6b(trans)) with Na[N(SiMe₃)₂]

The mixture of the formate complexes (0.03 mmol) of molybdenum (mixture of 6a(cis) and 6a(trans) along with 4b(trans)) and tungsten (mixture of 6b(cis) and 6b(trans)) was dissolved in THF in a J. Young NMR tube. Then 1 equiv. of Na[N(SiMe₃)₂] was added. The mixture was shaken vigorously. The ³¹P{¹H} NMR spectra revealed immediate regeneration of 3a or 3b. Then the reaction mixture was dried in vacuo and extracted with pentane to remove the 3a or 3b. The remaining solid insoluble in pentane was dissolved in D₂O. The ¹H NMR spectra in D₂O revealed the formation of HCOONa.

General procedures for the CO₂ hydrogenations. A steel autoclave was charged under nitrogen with the catalysts (0.1 mmol, 5 mol% with respect to base) dissolved in 0.5 mL of THF and 20 equiv. of Na[N(SiMe₃)₂] (1 M in THF). The tube was pressurized with 40 bar H₂ and allowed to stir at room temperature for half an hour. After this period of time 10 bar of CO₂ was charged. Then the total pressure was kept at 80 bar filling with H₂ pressure. The reaction mixture was heated to 140 °C (Caution!!: pressure rises to 100 bar). After certain reaction times, the autoclave was cooled to room temperature and the pressure was released. The THF solvent was evaporated to dryness and D₂O was added to dissolve the solid residue. Dimethylformamide (2 μL) was added to the reaction mixtures as an internal standard for the determination of the yield by ¹H NMR spectroscopy. ¹H NMR (300 MHz, D₂O): δ = 8.3 (s, 1H, HCOONa).

X-ray diffraction analyses

The data collection and structure refinement data for 2a(anti), 2b(syn), 4b(trans) and 6b(cis) are presented in (Table 2). Single-crystal X-ray diffraction data were collected at 183(2) K on a Xcalibur diffractometer (Agilent Technologies, Ruby CCD detector) for all compounds using a single wavelength Enhance X-ray source with MoKα radiation (λ = 0.71073 Å).^32a The selected suitable single crystals were mounted using polybutene oil on the top of a glass fiber fixed on a goniometer head and immediately transferred to the diffractometer. Pre-experiment, data collection, data reduction and analytical absorption corrections^32b were performed with the program suite CrysAlis^Pro.^32a The crystal structures were solved with SHELXS97^32c using direct methods. The structure refinements were performed by full-matrix least-squares on F² with SHELXL97.^32c All programs used during the crystal structure determination process are included in the WINGX software.^32d PLATON^32e was used to check the result of the X-ray analyses and DIAMOND^32f was used for the molecular graphics. CCDC 888516 (for 4b(trans)), 888517 (for 6b(cis)), 960495 (for 2a(anti)) and 960496 (for 2b(syn)) contain the supplementary crystallographic data (excluding structure factors) for this paper.

Table 2 Crystallographic data for compounds 2a(anti), 2b(syn), 4b(trans) and 6b(cis)^a

	2a(anti)	2b(syn)	4b(trans)	6b(cis)
a The unweighted R-factor is R1 = ∑(Fo − Fc)/∑Fo; I > 2σ(I) and the weighted R-factor is wR2 = {∑w(Fo^2 − Fc^2)^2/∑w(Fo^2)^2}^1/2.
CCDC	960495	960496	888516	888517
Empirical formula	C16H37Cl2MoN2OP2	C16H37Cl2N2OP2W	C18H36N2O4P2W	8(C18H38N2O4P2W), C4H8O
Formula weight (g mol^−1)	502.26	590.16	590.28	4810.46
Temperature (K)	183(2)	183(2)	183(2)	183(2)
Wavelength (Å)	0.71073	0.71073	0.71073	0.71073
Crystal system, space group	Orthorhombic, P212121	Orthorhombic, P212121	Monoclinic, P21/c	Tetragonal, I41/a
a (Å)	7.3971(1)	7.4129(1)	7.8847(1)	36.3651(6)
b (Å)	13.2575(2)	13.2003(2)	26.3644(4)	36.3651(6)
c (Å)	23.8535(4)	23.8800(5)	11.5872(2)	14.8550(3)
α (°)	90	90	90	90
β (°)	90	90	106.979(2)	90
γ (°)	90	90	90	90
Volume (Å^3)	2339.24(6)	2336.72(7)	2303.70(7)	19644.6(8)
Z, density (calcd) (Mg m^−3)	4, 1.426	4, 1.678	4, 1.702	4, 1.626
Abs coefficient (mm^−1)	0.933	5.316	5.178	4.859
F(000)	1044	1172	1176	9632
Crystal size (mm^3)	0.30 × 0.11 × 0.06	0.12 × 0.10 × 0.06	0.33 × 0.27 × 0.20	0.10 × 0.08 × 0.03
θ range (°)	2.88 to 32.58	2.88 to 28.28	2.70 to 28.28	2.44 to 25.68
Reflections collected	55 639	16 728	37 089	27 807
Reflections unique	8495/[Rint = 0.0485]	5804/[Rint = 0.0372]	5720/[Rint = 0.0263]	9319/[Rint = 0.0800]
Completeness to θ (%)	99.9	99.9	99.9	99.9
Absorption correction	Analytical	Analytical	Analytical	Analytical
Max/min transmission	0.951 and 0.799	0.765 and 0.644	0.432 and 0.275	0.995 and 0.987
Data/restraints/parameters	7529/0/225	5804/61/257	5574/0/252	5971/0/510
Goodness-of-fit on F^2	1.030	0.895	1.294	0.987
Final R1 and wR2 indices [I > 2σ(I)]	0.0351, 0.0852	0.0329, 0.0488	0.0203, 0.0400	0.0488, 0.0619
R 1 and wR2 indices (all data)	0.0420, 0.0872	0.0473, 0.0505	0.0212, 0.0403	0.0972, 0.0744
Absolute structure parameter	−0.02(3)	−0.007(7)
Largest diff. peak and hole (e Å^−3)	1.107 and −0.613	2.044 and −1.158	0.657 and −1.231	1.241 and −0.845

---

Acknowledgements

Financial support from the Swiss National Science Foundation and the University of Zürich are gratefully acknowledged.

Notes and references

1. (a) Carbon Dioxide as Chemical Feedstock, ed. M. Aresta, Wiley-VCH, Weinheim, 2010; (b) M. Cokoja, C. Bruckmeier, B. Rieger, W. A. Herrmann and F. E. Kühn, Angew. Chem., Int. Ed., 2011, 50, 8510–8537.
2. (a) T. Sakakura, J.-C. Choi and H. Yasuda, Chem. Rev., 2007, 107, 2365–2387; (b) X. Yin and J. R. Moss, Coord. Chem. Rev., 1999, 181, 27–59.
3. (a) P. G. Jessop, T. Ikariya and R. Noyori, Chem. Rev., 1995, 95, 259–272; (b) P. G. Jessop, F. Joó and T. Chih-C, Coord. Chem. Rev., 2004, 248, 2425–2442; (c) W. Leitner, Angew. Chem., Int. Ed. Engl., 1995, 34, 2207–2221.
4. (a) C. Gunanathan and D. Milstein, Acc. Chem. Res., 2011, 44, 588–602; (b) M. D. Fryzuk and P. A. MacNeil, Organometallics, 1983, 2, 682–684; (c) A. Friedrich, M. Drees, M. Käss, E. Herdtweck and S. Schneider, Inorg. Chem., 2010, 49, 5482–5494; (d) C. M. Fafard, D. Adhikari, B. M. Foxman, D. J. Mindiola and O. V. Ozerov, J. Am. Chem. Soc., 2007, 129, 10318–10319; (e) E. Khaskin, M. A. Iron, L. J. W. Shimon, J. Zhang and D. Milstein, J. Am. Chem. Soc., 2010, 132, 8542–8543; (f) C. Gunanathan, B. Gnanaprakasam, M. A. Iron, L. J. W. Shimon and D. Milstein, J. Am. Chem. Soc., 2010, 132, 14763–14765; (g) L. Schwartsburd, M. A. Iron, L. Konstantinovski, E. Ben Ari and D. Milstein, Organometallics, 2011, 30, 2721–2729; (h) S. Kohl, L. Weiner, L. Schwartsburd, L. Konstantinovski, L. J. W. Shimon, Y. Ben-David, M. A. Iron and D. Milstein, Science, 2009, 324, 74–77; (i) E. Ben-Ari, G. Leitus, L. J. W. Shimon and D. Milstein, J. Am. Chem. Soc., 2006, 128, 15390–15391.
5. (a) R. Noyori, M. Koizumi, D. Ishii and T. Ohkuma, Pure Appl. Chem., 2001, 73, 227–232; (b) S. Clapham, A. Hadzovic and R. Morris, Coord. Chem. Rev., 2004, 248, 2201–2237; (c) T. Ikariya and A. J. Blacker, Acc. Chem. Res., 2007, 40, 1300–1308; (d) C. Gunanathan and D. Milstein, Acc. Chem. Res., 2011, 44, 588–602; (e) H. Grützmacher, Angew. Chem., Int. Ed., 2008, 47, 1814–1818; (f) Y. Blum, D. Czarkie, Y. Rahamim and Y. Shvo, Organometallics, 1985, 4, 1459–1461.
6. C. A. Huff, J. W. Kampf and M. S. Sanford, Organometallics, 2012, 31, 4643–4645.
7. (a) M. Vogt, M. Gargir, M. A. Iron, Y. Diskin-Posner, Y. Ben-David and D. Milstein, Chem. – Eur. J., 2012, 18, 9194–9197; (b) M. Vogt, A. Nerush, Y. Diskin-Posner, Y. Ben-David and D. Milstein, Chem. Sci., 2014, 5, 2043–2051.
8. (a) C. M. Mömming, E. Otten, G. Kehr, R. Frçhlich, S. Grimme, D. W. Stephan and G. Erker, Angew. Chem., Int. Ed., 2009, 48, 6643–6646; (b) A. E. Ashley, A. L. Thompson and D. O'Hare, Angew. Chem., Int. Ed., 2009, 48, 9839–9843; (c) G. Ménard and D. W. Stephan, J. Am. Chem. Soc., 2010, 132, 1796–1797; (d) J. Boudreau, M. A. Courtemanche and F. G. Fontaine, Chem. Commun., 2011, 47, 11131–11133; (e) L. J. Hounjet, C. B. Caputo and D. W. Stephan, Angew. Chem., Int. Ed., 2012, 51, 4714; (f) I. Peuser, R. C. Neu, X. Zhao, M. Ulrich, B. Schirmer, J. A. Tannert, G. Kehr, R. Fröhlich, S. Grimme, G. Erker and D. W. Stephan, Chem. – Eur. J., 2011, 17, 9640.
9. Y. Jiang, O. Blacque, T. Fox and H. Berke, J. Am. Chem. Soc., 2013, 135, 7751–7760.
10. J. F. Hull1, Y. Himeda, W.-H. Wang, B. Hashiguchi, R. Periana, D. J. Szalda, J. T. Muckerman and E. Fujita, Nat. Chem., 2012, 4, 383–388.
11. (a) C. Federsel, R. Jackstell and M. Beller, Angew. Chem., Int. Ed., 2010, 49, 6254–6257; (b) A. H. Chelsea and M. S. Sanford, J. Am. Chem. Soc., 2011, 133, 18122–18125.
12. (a) S. Wesselbaum, T. Vom Stein, J. Klankermayer and W. Leitner, Angew. Chem., Int. Ed., 2012, 51, 7499–7502; (b) M. J. Sgro and D. W. Stephan, Angew. Chem., Int. Ed., 2012, 51, 11343–11345.
13. P. G. Jessop, T. Ikariya and R. Noyori, Nature, 1994, 368, 231–233.
14. P. G. Jessop, Y. Hsiao, T. Ikariya and R. Noyori, J. Am. Chem. Soc., 1996, 118, 344–355.
15. (a) F. Joó, G. Laurenczy, L. Nádasdi and J. Elek, Chem. Commun., 1999, 971–972; (b) C. A. Huff and M. S. Sanford, ACS Catal., 2013, 3, 2412–2416.
16. E. Graf and W. Leitner, J. Chem. Soc., Chem. Commun., 1992, 623–624.
17. F. Gassner and W. Leitner, J. Chem. Soc., Chem. Commun., 1993, 1465–1466.
18. Y. Himeda, N. Onozawa-Komatsuzaki, H. Sugihara and K. Kasuga, Organometallics, 2007, 26, 702–712.
19. R. Tanaka, M. Yamashita and K. Nozaki, J. Am. Chem. Soc., 2009, 131, 14168–14169.
20. (a) S. Enthaler, K. Junge and M. Beller, Angew. Chem., Int. Ed., 2008, 47, 3317–3321; (b) S. Gaillard and J.-L. Renaud, ChemSusChem, 2008, 1, 505–509; (c) A. Correa, O. G. Mancheño and C. Bolm, Chem. Soc. Rev., 2008, 37, 1108–1117; (d) C. Federsel, A. Boddien, R. Jackstell, R. Jennerjahn, P. J. Dyson, R. Scopelliti, G. Laurenczy and M. Beller, Angew. Chem., Int. Ed., 2010, 49, 1–6; (e) R. Langer, Y. Diskin-Posner, G. Leitus, L. J. W. Shimon, Y. Ben-David and D. Milstein, Angew. Chem., Int. Ed., 2011, 50, 9948–9952.
21. R. M. Bullock, Chem. – Eur. J., 2004, 10, 2366–2374.
22. (a) S. Chakraborty, O. Blacque, T. Fox and H. Berke, ACS Catal., 2013, 3, 2208–2217; (b) E. Peterson, A. Y. Khalimon, R. Simionescu, L. G. Kuzmina, J. A. K. Howard and G. I. Nikonov, J. Am. Chem. Soc., 2009, 131, 908; (c) V. K. Dioumaev and R. M. Bullock, Nature, 2000, 424, 530–532.
23. (a) S. Chakraborty, O. Blacque, T. Fox and H. Berke, Chem. – Asian J., 2014, 9, 328–337; (b) A. Dybov, O. Blacque and H. Berke, Eur. J. Inorg. Chem., 2011, 652–659; (c) S. Chakraborty, O. Blacque, T. Fox and H. Berke, Chem. – Asian J., 2014, 9, 2896–2907.
24. S. Chakraborty, O. Blacque, T. Fox and H. Berke, Chem. – Eur. J., 2014, 20, 12641–12654.
25. S. Chakraborty and H. Berke, ACS Catal., 2014, 4, 2191–2194.
26. (a) P. J. Walsh, F. J. Hollander and R. G. Bergman, J. Am. Chem. Soc., 1988, 110, 8729–8731; (b) P. J. Walsh, F. J. Hollander and R. G. Bergman, Organometallics, 1993, 12, 3105–3123; (c) J. L. Bennett and P. T. Wolczanski, J. Am. Chem. Soc., 1994, 116, 2179–2180; (d) J. de With and A. D. Horton, Organometallics, 1993, 12, 1493–1496.
27. M. W. Bouwkamp, A. A. Batinas, P. T. Witte, T. Hubregtse, J. Dam, A. Meetsma, J. H. Teuben and B. Hessen, Organometallics, 2008, 27, 4071–4082.
28. (a) Z. Chen, H. W. Schmalle, T. Fox and H. Berke, Dalton Trans., 2005, 580–587; (b) J. Höck, H. Jacobsen, H. W. Schmalle, G. R. J. Artus, T. Fox, J. I. Amor, F. Bäth and H. Berke, Organometallics, 2001, 20, 1533–1544; (c) F. Furno, T. Fox, H. W. Schmalle and H. Berke, Organometallics, 2000, 19, 3620–3630; (d) F. Liang, H. Jacobsen, H. W. Schmalle, T. Fox and H. Berke, Organometallics, 2000, 19, 1950–1962; (e) Y. Zhao, H. W. Schmalle, T. Fox, O. Blacque and H. Berke, Dalton Trans., 2006, 73–85.
29. L. Bencze and J. Kohan, Inorg. Chim. Acta, 1982, 65, L17–L19.
30. A. A. Danopoulos, A. R. Wills and P. G. Edwards, Polyhedron, 1990, 9, 2413–2418.
31. CCDC 888516 (for 4b(trans)) and 888517 (for 6b(cis)), 960495 (for 2a(anti)) 960496 for 2b(syn)) contain the supplementary crystallographic data (excluding structure factors) for this paper.
32. (a) Agilent Technologies (formerly Oxford Diffraction), Yarnton, England, 2011; (b) R. C. Clark and J. S. Reid, Acta Crystallogr., Sect. A: Fundam. Crystallogr., 1995, 51, 887–897; (c) G. M. Sheldrick, Acta Crystallogr., Sect. A: Fundam. Crystallogr., 2008, 64, 112–122; (d) L. J. Farrugia, J. Appl. Crystallogr., 1999, 32, 837; (e) A. L. Spek, J. Appl. Crystallogr., 2003, 36, 7–13; (f) K. Brandenburg, DIAMOND, Crystal Impact GbR, Bonn, Germany, 2007.

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 888517, 888516, 960496 and 960495. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c5dt00278h

---