Volume 29, Issue 28 e202300193
Research Article
Open Access

A Non Expected Alternative Ni(0) Species in the Ni-Catalytic Aldehyde and Alcohol Arylation Reactions Facilitated by a 1,5-Diaza-3,7-diphosphacyclooctane Ligand

Juliane Heitkämper

Juliane Heitkämper

Universität Stuttgart, Institut für Theoretische Chemie, Pfaffenwaldring 55, 70569 Stuttgart, Germany

Institut de Química Computacional i Catàlisi and Departament de Química, Universitat de Girona, Maria Aurèlia Capmany 69, 17003 Girona, Catalonia, Spain

Contribution: Data curation (lead), Formal analysis (lead), Writing - original draft (equal)

Search for more papers by this author
Dr. Sergio Posada-Pérez

Corresponding Author

Dr. Sergio Posada-Pérez

Institut de Química Computacional i Catàlisi and Departament de Química, Universitat de Girona, Maria Aurèlia Capmany 69, 17003 Girona, Catalonia, Spain

Contribution: Data curation (lead), Formal analysis (lead), Supervision (lead), Visualization (lead), Writing - review & editing (equal)

Search for more papers by this author
Sílvia Escayola

Sílvia Escayola

Institut de Química Computacional i Catàlisi and Departament de Química, Universitat de Girona, Maria Aurèlia Capmany 69, 17003 Girona, Catalonia, Spain

Donostia International Physics Center (DIPC), Donostia, Euskadi, Spain

Contribution: Data curation (lead), Formal analysis (lead), ​Investigation (lead), Visualization (lead), Writing - original draft (equal)

Search for more papers by this author
Prof. Dr. Miquel Solà

Prof. Dr. Miquel Solà

Institut de Química Computacional i Catàlisi and Departament de Química, Universitat de Girona, Maria Aurèlia Capmany 69, 17003 Girona, Catalonia, Spain

Contribution: Data curation (equal), Formal analysis (equal), Funding acquisition (equal), Validation (lead), Writing - review & editing (supporting)

Search for more papers by this author
Prof. Dr. Johannes Kästner

Prof. Dr. Johannes Kästner

Universität Stuttgart, Institut für Theoretische Chemie, Pfaffenwaldring 55, 70569 Stuttgart, Germany

Contribution: Conceptualization (equal), Supervision (lead), Validation (lead), Writing - review & editing (supporting)

Search for more papers by this author
Dr. Albert Poater

Corresponding Author

Dr. Albert Poater

Institut de Química Computacional i Catàlisi and Departament de Química, Universitat de Girona, Maria Aurèlia Capmany 69, 17003 Girona, Catalonia, Spain

Contribution: Conceptualization (lead), Data curation (equal), Formal analysis (equal), Funding acquisition (lead), ​Investigation (equal), Supervision (lead), Visualization (equal), Writing - original draft (equal), Writing - review & editing (lead)

Search for more papers by this author
First published: 23 January 2023
Citations: 2

Graphical Abstract

As in cross-coupling reactions, in the alcohol arylation reaction catalysed by a system combining nickel and a 1,5-diaza-3,7-diphosphacyclooctane ligand, the presence of a Ni(0) intermediate is under debate by Density Functional Theory calculations.

Abstract

For decades there were many attempts to dispense with stoichiometric amounts of metal reagents for the synthesis of secondary alcohols. In 2021, the synthetic results of Newman and collaborators pioneered a synthesis still with metals, but not as reactants. Instead, they serverd as catalytic engines. Here we present a description by means of Density Functional Theory calculations of how this process can occur, and an attempt is made to shed light on the mechanism that facilitates the attainment of secondary alcohols, emphasizing the eternal cross-coupling debate of whether the catalytically active species is Ni(0) or they are really taking shortcuts following the course of Ni(II). Effective Orbital analyses give a clear picture. Furthermore, this paper provides insight not only into the nature of the ligands of the metal catalyst but also the role of the base.

Introduction

The apparent simplicity of the OH functional group that defines the alcohols, and specially its natural origin for the primary alcohols,1-4 hides somewhat the complexity of the synthesis of secondary and tertiary alcohols. Their importance in drug synthesis5-8 motivates continued efforts towards improving their synthesis. Among the spectrum of possibilities, the reaction of aldehydes to secondary alcohols with organometallic reagents is of vital importance not only academically, but industrially. On the other hand, the formation of C−C bonds, using organomagnesium reagents, or also Grignard reagents, is achieved but only in a stoichiometric way and with a fairly limited variety of functional groups,9-12 apart from the necessity of low temperatures that are required because of the exothermic nature of the reaction. In order to overcome the limitations of Grignard reagents, an alternative is to replace the organohalides and magnesium metal agents by catalysts. In detail, the Nozaki–Hiyama–Kishi reaction allows the reductive coupling of organohalides and aldehydes by chromium,13-15 in combination with a nickel catalyst.16, 17 Mechanistically, nickel allows the activation of the carbon-halogen bond and then transmetalates it with chromium, thus generating in situ the nucleophilic species of organochrome. However, the solution is far from definitive since chromium does not act catalytically, although it has been possible to reduce its use.18-22 Since then, a series of studies made it possible to make a clear evolution to remove chromium. Without these works, our current understanding of this reaction would not be possible. Cheng and Majumdar first discovered that the Ni/dppe pair could catalyze the reaction, although they needed a stoichiometric amount of Zn,23 which was required as well by Weix and coworkers, showing that catalysts with N-based ligands such as bipyridine or PyBox, combined with nickel as metal, led to complex secondary alcohols24 using aryl bromides and aldehydes as reagents. Krische contemporaneously achieved the reduction with sodium formate, therefore in a milder form the combination of reductive coupling and transfer hydrogenation,25 using a rhodium(I) catalyst and double the amount of aryl iodide to make feasible the arylation of both alkyl like aryl substituted aldehydes, but at high temperature. Continuing with the aim of the synthesis of secondary alcohols, the α-arylation of the primary homologues is in the process of expansion. MacMillan and coworkers, combining nickel and iridium, photochemically achieved this with aliphatic alcohol and aryl bromides.26 However, the reaction required stoichiometric amounts of Zn.27-29 This bottleneck, together with the fact that an excess of alcohol was needed, generates undesired ketones as by-products. At that time, nothing had been studied in the absence of electron-rich heterocycles or benzyl alcohols paving the way to new challenges.

Zhang, Xu, Findlater and collaborators reported an electrochemical arylation capable of interacting with aldehydes and alcohols, but in a very limited way.30 And more recently, the attempts of Newman and co-workers have followed the same path, with the aim of avoiding the stoichiometric use of any metallic element in the reactions, placing a clear emphasis on the use of Ni(0) as a catalytically active species. Thus, cross-coupling reactions with aldehydes and alcohols as reagents could lead to the synthesis of ketones from aryl triflates with the catalytic framework Ni(0)/triphos.31 It was proposed that the process took place thanks to the Heck reaction,32 with β-hydride elimination of an intermediate with the nickel alkoxide ligand that would subsequently give the ketone. Then the goal became more ambitious, and although failed, it happened to reconvert this Ni-alkoxide intermediate to obtain alcohols, in line with other studies that lead to O-arylation,33 or simply unwanted secondary reactions.34-36 Another synthetic effort by Newman's group made it possible to overcome these obstacles in reductive arylation and alcohol α-arylation to reach secondary alcohols from the coupling of organohalides without the need for any stoichiometric amount of metal compound. A nickel catalyst with the 1,5-diaza-3,7-diphosphacyclooctane ligand (P2N2) was used as a solution,37, 38 which allows not only the cross-coupling of aryl iodides with primary alcohols or aldehydes to obtain secondary benzyl alcohols,39 but also in an efficient way, to avoid the secondary reactions of Grignard reagents. It was proposed that the mechanism of the reaction under investigation proceeds via a Ni(0) species, despite its relatively expected high instability,40 extendable especially to palladium, too.40, 41 In the present work, the aldehyde arylation reaction catalyzed by Ni catalyst armed with the P2N2 ligand, shown in Figure 1, is studied in terms of Density Functional Theory (DFT) calculations, aiming to reveal the underlying reaction mechanism and conditions that influence it at the molecular level.

Details are in the caption following the image

Reaction under investigation in the current study.

Results and Discussion

To reduce the computational cost, we simplified the catalyst by considering methyl groups instead of aryl or cyclohexyl groups on the nitrogen or phosphorus atoms, respectively. However, experiments reported the large influence in the reaction yield of the substituents on the ligand.39 To systematically investigate this effect, we identified the TOF determining intermediate (TDI) and TOF determining transition state (TDTS) and the energetic span, according to the energetic span model,42 using the simplified model system. Then, this energetic span has been recomputed with different substituents on the ligand to explore their influence, by means of the comparison of the energy barrier.

After a careful search of isomers, the most stable conformer of the simplified catalyst is shown in Figure 2a. The simplification by methyl groups becomes a way to reduce the complexity of further isomers with the larger ligands and thus, determine precisely the whole reaction pathway finally locating the right TDI and TDTS. Herein, a hydrogen atom of the methyl group coordinates with the metal center to form an agostic interaction. Apparently, coordination with the nickel center out of the square planar coordination plane has a stabilizing effect. Moreover, when substituting methyl with a bulkier phenyl group, the latter conformation is preferred since the phenyl ring interacts with the metal center. This stabilizing effect was found previously,43 and is particularly highlighted as it has an influence on the stability of intermediates and transition states, as it will be discussed later. The interaction between the metal and the nearest methyl group is clearly illustrated in Figure 2a, with an activation of the C−H bond up to 1.125 Å, compared to the others of 1.109 Å.

Details are in the caption following the image

(a) The most stable conformer of the simplified catalyst 0 and (b) the formed complex TMP⋅⋅⋅HBr (selected distances in Å).

Figure 3 provides an overview of the studied reaction pathways with the corresponding labeling, while Figure 4 shows the analogous energy diagram for the simplified system as well as the structures of the transition states. The reaction mechanism starts with the catalytic active species in the catalytic cycle, I, which is generated through an endothermic process from 0, in which I (with X1=Br) is 10.2 kcal mol−1 higher in energy than 0. To form I, the TMP (2,2,6,6-tetramethylpiperidine) and HBr (or HI) are essential. Both acids are very strong and favor the release of protons. The TMP acts as a base, resulting in a very stable TMP⋅⋅⋅HBr complex shown in Figure 2b. The bond distance of the accepted proton to the nitrogen is of 1.158 Å, very similar to that of the other proton with a bond distance of 1.034 Å. Moreover, the H−Br bond is lengthened to 1.885 Å compared to 1.438 Å in a single HBr molecule. To screen the role of TMP as a base, a discussion of the use of different bases other than TMP is presented later.

Details are in the caption following the image

Reaction mechanism of the aldehyde arylation reaction catalyzed by a nickel center with a 1,5-diaza-3,7-diphosphacyclooctane ligand. X1=Br for the first catalytic cycle, but it can be Br or I for the following cycles.

Details are in the caption following the image

Energy diagram of the investigated reaction mechanism (above) and structures of the transition states (below). Relative Gibbs energies of the stationary points with respect to I+Reactants, in kcal mol−1; and selected distances in red are given in Å.

From I, a nickel hydride (II) is formed via a β-elimination of the alcohol, resulting in the formation of a ketone. To overcome the transition state, TSI, 19.6 kcal mol−1 are required. In past studies, it was proposed that subsequently the nickel hydride undergoes reductive elimination to form a Ni(0) species III.39 However, it is located thermodynamically 33.1 kcal mol−1 above I (still disregarding transition states to get there). Thus, it is found to be too high in energy to be formed even at 75 °C. In addition, a possible lower triplet state was ruled out since the triplet spin state is found 19.1 kcal mol−1 higher in energy than the singlet ground state. Later we discuss different possibilities for stabilizing intermediate III, but first, the path to go directly from II to IV ruling out the need of a Ni(0) intermediate in between is discussed.

On the one hand, there is a direct reaction pathway in which the Ar−I bond cleaves and both halves coordinate with the nickel center. Simultaneously the H and Br ligands join to form HBr. This step is concerted and possesses a transition state TSII,D with a relative Gibbs energy of 28.6 kcal mol−1, thus the associated energy barrier is 29.0 kcal mol−1 with respect to II. On the other hand, an alternative pathway was observed, in which the ligand is actively involved. From the hydride species II, TSII,S1 involves the H transfer to the nitrogen atom of the ligand to form III.1 complex overcoming an energy barrier of 29.5 kcal mol−1. This hydrogen transfer to the ligand was observed in a catalytic hydrogen oxidation mechanism using the same type of ligand.44 This reaction step is sensitive depending on the electronic environment of the nitrogen and thus on the substituents at it (see below). This hydrogen transfer step was observed when a R1-I reactant already coordinates to the Ni center to form a tetrahedrally coordinated complex involving the iodide. Without the R1-I coordination, structure III.1 would be 7.6 kcal mol−1 higher in energy, so clearly the iodide is coordinated to the nickel center to form a tetrahedrally coordinated nickel complex, which is confirmed by a Mayer Bond Order (MBO)45 of 0.65 between the iodide and nickel, in accordance with previous results greater than 0.5.46 In a subsequent step, an oxidative addition process, the R1-I bond splits to coordinate with the nickel center to form III.2, via the transition state TSII,S2 with an energy barrier of 15.5 kcal mol−1. In a final step, the hydrogen is transferred to the bromide ion to release HBr and form IV in a fast reaction via TSII,S3 with an energy barrier of 11.6 kcal mol−1.

Apart from the proposed catalytic cycle in Figure 3, alternative steps/pathways were also considered. For instance, from structure II, an undesired side reaction of the reactant R1-I to form the de-halogenated compound R1-H and VII via TSII,SR was also explored. This side reaction implies an energy barrier of 29.5 kcal mol−1 and therefore, it is a competitive pathway. The ratio of rate constants of this side reaction and the main reaction, which is sensitive depending on the substituents on the nitrogen atom, determines the yield of the desired product or the extent of the formation of undesired by-products.

The predicted C−C bond formation between the aldehyde and the R1 group occurs after the aldehyde coordinates to IV to form V. The energy barrier to obtain the product coordinated to the catalyst (TSV) is overall 26.6 kcal mol−1. This step is greatly stabilized by coordinating agents below and above the square plane of the active center. It increases by 4.4 kcal mol−1 if the methyl group is not coordinated by a Ni⋅⋅⋅H interaction,47 and further decreases by 6.8 kcal mol−1 if an ArCF3 ligand is on the nitrogen atom of the ligand instead of a methyl group.

The mechanism is finally completed by the exchange of the product by a 1-PhEtOH molecule to initiate a new catalytic cycle in an endothermic step. However, the whole catalytic pathway is exergonic, since completing the catalytic cycle, the system ends up at I again releasing 25.7 kcal mol−1.

According to the energetic span model,42 intermediate VI and TSII describe the TDI and TDTS of the catalytic cycle and create an energetic span of 32.2 kcal mol−1 (see Figure 5a). Even though this energy barrier is remarkable, even at 75 °C, there is the key factor that, after the first catalytic cycle, one iodide is coordinated to the nickel center instead of Br. Therefore, X1 could be iodide instead of bromide in Figure 3, from the second catalytic pathway. Figure 5b shows that this fact has a particular influence on TSII,SR, decreasing the energetic span to 28.5 kcal mol−1. Thus, iodide is likely to be part of the active catalyst rather than bromide.

Details are in the caption following the image

Energetic span for the simplified catalyst with X=Br, I (relative Gibbs energies in kcal mol−1 with respect to I+Reactants) for (a) the transition state TSII,D and (b) TSII,S1.

As indicated in Figure 3, from the energy well of the reaction, i. e. intermediate VI, it is also conceivable that the coordinated product alcohol undergoes a β-elimination via TSVI, which is energetically competitive with a β-elimination of 1-PhEtOH (see Figure 4) via TSI. However, a β-elimination of the product is not observed experimentally. A possible explanation is that 1-PhEtOH is present at high concentrations, favoring the reaction rate of the formation of II.

Since intermediate VI and the following TSII turned out to dominate the energetic span, these species were recomputed with different ligands to understand the performance of different catalysts. Experimentally it was found that, for example, the use of an PCyNArCF3 catalyst forms the desired product, while PCyNArOMe yields mainly the quinoline side product (VII). Therefore, the methyl group interacting with the nickel center was substituted by ArOMe and ArCF3 and the energetic span recomputed. As can be seen in Figure 6, the performance of the catalyst seems to depend on a sensitive interplay between the energy barriers for the side reaction and the desired transition state TSII. In case of ArOMe (Figure 6b) the side reaction has the same energy barrier as the main reaction, yielding a formation of the side product VII in large amount, lowering the yield of the desired product. In case of the ArCF3 the side reaction is slightly higher in energy than the transition state for the main product formation (0.5 kcal mol−1). As known from enantioselective reactions, just 1–2 kcal mol−1 are already enough to fully favor one enantiomer over the other and therefore, it is possible that this small energy difference decides over the formation of the main or side product.

Details are in the caption following the image

Energetic span using (a) the PCyNArCF3 catalyst and (b) the PCyNArOMe catalyst for X=I (relative Gibbs energies in kcal mol−1 with respect to VI).

As already discussed, the Ni(0) species III is very high in energy. However, there are a few possibilities to stabilize this complex. On the one hand, the coordination of a benzene ring to the metal center is found in the literature to have a stabilizing effect.48 As can be seen in Figure 7, coordination of a toluene molecule indeed stabilizes intermediate III by 23.1 kcal mol−1. Note that in this comparison we added phenyl rings to the nitrogen as they might influence this effect a lot and also not species I, but species 0, was taken as a reference. Coordination of a second ligand in place of toluene further stabilizes III (see Figure 7). However, this complex is apparently catalytically inactive since there is no space around the nickel center to coordinate reactants to undergo the catalytic reaction. Separating the second ligand to form III in order to make space for catalysis would end in a too high energy barrier to be overcome. A discussion on the Ni(0) species III.1 is provided later.

Details are in the caption following the image

Stabilization of intermediate III by toluene, a second ligand or intramolecular deprotonation of the Ni-hydride to make anionic Ni(0)-Br and a protonated amine (relative energy values in kcal mol−1).

Experimentally it was found that the choice of the base has a great influence on the reaction. The use of TMP as a base was critical and with other common bases or additives like pyridine, triethylamine (NEt3) or K3PO4 no reaction was observed.39 As shown in Figure 8, DFT calculations show that pyridine or phosphoric acid coordinate stronger to the nickel centre than PhEtOH, therefore inhibiting the reaction by preventing the formation of I. For complexes with sterically hindered bases like NEt3 or TMP (with similar pka values of 10.8 and 11.1 respectively) no analogous square planar coordination was found. As shown in Figure 9, for the example of TMP, the bromide and TMP coordinate above and below the plane, resulting in a less stable species than I. Therefore, they do not inhibit the reaction. However, this argument does not explain why TMP outperforms NEt3. It is possible that NEt3, as the bulkiest base, is sterically very demanding, hindering the trimolecular reaction from 0 to I in which the catalyst, alcohol and base needs to be in close proximity. Unfortunately, it is barely possible to study such processes from a computational point of view. Nevertheless, DFT calculations allow to lead to this conclusion and a potential precipitation of a salt when adding TMP in the crude is a plausible explanation.

Details are in the caption following the image

Relative Gibbs energies for the coordination of different bases to the metal centre with respect to 0, in kcal mol−1.

Details are in the caption following the image

Coordination of TMP to the catalyst (selected distances indicated in Å).

To further explore the nature of the reaction intermediates, a summary of the effective oxidation state (EOS) results for intermediate II, III.1, III.2, and III is shown in Table 1 (see Table S1 in the Supporting Information for detailed information on other intermediates),49 to provide information about the oxidation state of Ni. From the EOS results, we can observe that intermediate III is a Ni(0) species (R >90 %), and the same oxidation state holds when changing the phosphine ligand, including coordination of an explicit solvent molecule or adding a second phosphine ligand unit. For the rest of the intermediates, the assignation of the oxidation state is less clear (R <55 %). In most of the cases (intermediates 0, I, III.2, IV, V, and VI), the last electron pair to be assigned is located on an effective fragment orbital (EFO) of the phosphine ligand, which gets a global oxidation state of 0, leaving an oxidized Ni center with an OS of +2.50, 51 There are some exceptions, intermediates II and III.1, which have an OS of 0 on the Ni atom. In the case of intermediate II, the last occupied EFO is located on the Ni and the first unoccupied EFO is located on the H atom (see Figure 10), which gets an oxidation state of +1. Nevertheless, one has to note that in this case, the populations of the frontier EFOs are very close, 0.556 for the last occupied and 0.511 for the first unoccupied EFO, so the actual OS of the Ni may lay in between 0 and +2. Then, for intermediate III.1, the assignation of an OS of 0 to the Ni center is more clear (the last occupied EFO located on the Ni fragment has a significant occupation of 0.673) and the low value of R is only due to our fragment selection since the last-assigned electron pair, disputed between R1 (OS of −1) and I (OS of 1), should be assigned to R1+I considered as a single fragment. Having revealed the OS of 0 for this species, we added it to the row of Ni(0) complexes in Figure 7.

Table 1. Occupation number of last occupied and first unoccupied EFOs and Effective Oxidation State (EOS) of each fragment within the system and global reliability index R (%) for intermediates II, III.1, III.2 and III.

Fragment

Last occ. EFO

First unocc. EFO

Ox. State

R [%]

II

M (Ni)

0.556

0.194

0

54.5

X1 (Br)

0.724

0.017

−1

H

0.000

0.511

+1

P-Ligand

0.632

0.104

0

III

M (Ni)

0.814

0.233

0

91.1

P-Ligand

0.644

0.146

0

III.1

M (Ni)

0.673

0.175

0

54.6

X (I)

0.881

0.458

+1

X (Br)

0.795

0.012

−1

R1

0.504

0.282

−1

P-Ligand-H

0.740

0.146

+1

III.2

M (Ni)

0.844

0.550

+2

52.7

X (I)

0.687

0.015

−1

X (Br)

0.802

0.013

−1

R1

0.577

0.053

−1

P-Ligand-H

0.640

0.152

+1

Details are in the caption following the image

EFOs with occupation numbers between 0.450 and 0.650 in intermediate a) II and b) III.2. The orbitals are visualized with an isocontour of 0.05 (with the exception of H orbital, where an isocontour of 0.1 was used).

Mechanistically, from the point of view of the oxidation state, we start the catalytic cycle with an Ni(II) species that gets partially reduced with the formation of II, and it could become clearly Ni(0) after the release of the ketone and the decordination of the Br (formation of III). However, as stated above, this is not affordable thermodynamically. From intermediate II a second reaction can take place when R1I is inserted on this intermediate leading to the formation of III.1 which is also a Ni(0) species, in agreement with past work of Borys and Hevia,52 Newman and coworkers,39, 53 and previously with palladium by Amatore and Jutand.54 Then, the metal center is oxidized back to Ni(II) with the formation of III.2. Both reaction paths converge to intermediate IV that latter evolves to V and VI, these last three intermediates are all Ni(II) species. These bidirectional changes from Ni(II) to Ni(0), with Ni(I) in between, have also been exposed very recently with great potential for cross-coupling reactions by Doyle and coworkers,55 but we have not characterized any Ni(I) species,56 and it is ruled out from the current study.

Conclusion

We have reported DFT studies of a catalytic method to access secondary alcohols from aryl iodides. This reaction is catalyzed by a metal catalyst that combines nickel and a 1,5-diaza-3,7-diphosphacyclooctane ligand. Even though and in line with cross-coupling reactions, the presence of a Ni(0) intermediate is considered throughout the reaction mechanism, here it must be discarded apparently. Instead an alternative proposal is announced, with a catalytic pathway that alternatively changes the halide lowering the kinetic cost of the catalysis. Unexpectedly, this alternative mechanism also includes Ni(0) intermediates. The mechanistic study also describes that electron-donating groups on the aryl ligands bonded to the metal enhance the catalytic activity, as well as unveils the role of the base, indispensable for the reaction. Unfortunately, our computational study does not provide a complete understanding of the nature of the base. Thus, there are still mechanistic points at the root of the experiments that need to be further studied computationally,39 for example to explain why, with pyridine as a base, the alcohol is not obtained but quinoline and residually the ketone product.

Computational Details

DFT calculations have been performed with the Gaussian16 program package.57 Geometry optimizations and subsequent frequency calculations were performed with the BP86 functional, i. e. the gradient generalized approximation (GGA) functional of Becke and Perdew,58 adding D3(BJ) dispersion corrections,59 and using the def2-SVP basis set.60 All minima were verified to possess only real frequencies and transition structures only a single mode with imaginary frequency. The Gibbs energies were calculated at 348.15 K and are presented according to a liquid phase reference state. At converged geometries, the electronic energies were calculated using the B3LYP functional, i. e. the hybrid GGA functional of Becke, Lee, Yang, and Parr,61 and the def2-TZVP basis set. The solvent effects were accounted for with the Polarizable Continuum Model (PCM),62 using toluene as a solvent. Numerical integration was carried out on an ultrafine grid. All calculations were performed for singlet spin states, verifying that the triplet state was not accessible in any case. Reported Gibbs energies are electronic energies obtained at the B3LYP−D3(BJ)/def2-TZVP(Toluene)//BP86-D3(BJ)/def2SVP level of theory with added ZPEs, thermal corrections and entropy contributions to the Gibbs energy obtained at the BP86/def2-SVP level. Conformational searches have been carried out with the crest tool,63 to account for the conformational complexity of the system.

The characterization of the Ni oxidation state has been done by means of the EOS analysis with the APOST-3D program.64 The EOS and partial atomic charges have been computed at the BP86-D3BJ/def2-SVP level of theory (same as geometry optimization) using the topological fuzzy Voronoi cells (TFVC) 3D-space partitioning and a 40 x 146 atomic grid for numerical integration. Together with the EOS results we have analyzed the corresponding EFOs and their occupations in the cases were low values or the reliability index (R <55 %) were found. The R (%) index is obtained from the relative occupation of the frontier EFOs, and measures how close is the OS assignation to the actual electron distribution.

Acknowledgments

J.H. acknowledges the financial support received in the form of a Ph.D. scholarship from the Studienstiftung des Deutschen Volkes (German National Academic Foundation). S.P.P. thanks the Spanish Ministerio de Ciencia e Innovación for Juan de la Cierva Formación fellowship (FJC2019-039623-I) and Marie Curie fellowship (H2020-MSCA-IF-2020-101020330). S.E. thanks Universitat de Girona and DIPC for an IFUdG2019 PhD fellowship. A.P. is a Serra Húnter Fellow and ICREA Academia Prize 2019. M.S. and A. P. thank the Spanish Ministerio de Ciencia e Innovación for projects PID2020-13711GB−I00 and PID2021-127423NB-I00 and the Generalitat de Catalunya for project 2021SGR623. The authors acknowledge support by the state of Baden-Württemberg through bwHPC and the German Research Foundation (DFG) through grant no INST 40/575-1 FUGG (JUSTUS 2 cluster). We are extremely grateful to Eric S. Isbrandt and Prof. Dr. Stephen G. Newman for insightful discussions about the catalytic cycle.

    Conflict of interest

    There are no conflicts of interest to declare.

    Data Availability Statement

    The data that support the findings of this study are available in the supplementary material of this article.