Download PDF
Synlett
DOI: 10.1055/a-2603-8157
letter
Special Section 14th EuCheMS Organic Division Young Investigator Workshop

Famous but Unknown: Structural Characterization of the Ruthenium-MACHO Catalyst

Mike S. B. Jørgensen
,
Kasper Enemark-Rasmussen
,
Mariusz Kubus
,
René Wugt Larsen
,
Martin Nielsen  ∗

The authors thank Villum Fonden (19049) and the Independent Research Fund, Denmark (1127-00172B) for generous funding.
› Further Information
 
 PDF Download Permissions and Reprints All articles of this category


Abstract

In this work, we present the determination of the crystal structure of the widely used transition-metal complex known as ruthenium-MACHO. The crystal structure of the selectively formed product of the reaction between this complex and hydrogen chloride dissolved in chloroform was also evaluated. The structures determined in this work were compared and evaluated by using various spectroscopic techniques.


#

Key words

ruthenium catalyst - ruthenium-MACHO - pincer complexes - coordination chemistry - structure determination - crystal structure

It is difficult to engage in homogeneously catalyzed (de)hydrogenation chemistry without familiarizing oneself with the commercially available transition-metal complex known as ruthenium-MACHO (Ru-MACHO). This complex was first reported in 2011 by the Takasago chemist Wataru Kuriyama and his co-workers as a highly active catalyst for the homogeneous reduction of esters in methanol under mild conditions.[1] The catalytic versatility of this compound has been wildly explored during the last decade, and many interesting transformations have emerged, such as a mild reduction of CO2 to methanol [2] an acceptorless dehydrogenation of aqueous-phase methanol,[3] a dehydrogenative coupling of ethanol,[4] a cross-coupling of secondary alcohols to α-substituted ketones[5] or nitriles,[6] an upgrading of ethanol to higher secondary alcohols,[7] and a chemoselective hydrogenation.[8] These are just a small representative sample of the wide range of catalytic reactions promoted by the Ru-MACHO complex. It is interesting to note that, despite the vast number of examples reported in the literature where Ru-MACHO has been used as the primary catalyst, satisfactory crystallographic data for the complex have not previously been provided.[9]

Chemists who have worked with Ru-MACHO recognize its distinct physical appearance as an off-white, sticky, microcrystalline powder. This description is in sharp contrast to its very well-characterized analogues, the structures of which differ only in the aliphatic substituents on the phosphorus- or nitrogen-ligators;[10] these complexes all appear as nonsticky, clearly crystalline products. Furthermore, it has been shown that Ru-MACHO itself is not stable under the basic conditions in which the vast majority of its reactions are conducted, and it has been shown to decompose into a variety of different compounds, some of which are also catalytically active in hydrogenation reactions.[11] Given the small amount of irrefutable analytical data on this particular transition-metal complex, it is questionable whether this complex can be regarded as the ‘true’ precatalyst, a detail that is of paramount importance for the correct acquisition of data on catalytic systems.

Table 1 Elemental Analyses of Ru-MACHO (C29H30ClNOP2Ru) Samples as Received from the Vendors

Vendor

C

H

N

Sigma

57.20

5.00

2.30

Strem

58.15

5.19

2.24

Theoretical

57.38

4.98

2.31

In this study, by using various techniques, we investigated samples of Ru-MACHO acquired from two different vendors (Strem and Sigma) in an attempt to shed light on the molecular integrity of the complex. CHN elemental analyses showed a slight disagreement between the theoretical and the observed compositions (Table [1]). Even though all the elemental analyses appeared to give results close to the theoretical values, they were sufficiently off-value to give rise to concern regarding the integrity of the product mixtures.

Zoom Image
Figure 1 (a) 31P-{1H} CPMAS solid-state NMR spectrum. The insert shows the region with the isotropic chemical shifts, as well as the simulated spectrum (in red). (b) 13C-{1H} CPMAS solid-state NMR spectrum. The asterisk denotes a spinning sideband not properly suppressed by the TOSS sequence.

The two batches of Ru-MACHO exhibited almost identical 1H and 31P NMR spectra, with only minor differences in the observed 1H, 13C, and 31P chemical shifts. Because of the high degree of similarity between the two samples, only the spectra for the Sigma sample will be discussed below in greater detail.

The 31P-{1H} and 13C-{1H} solid-state CP/MAS spectra of Sigma sample of Ru-MACHO are shown in Figure [1]. Multiple overlapping peaks from various phosphorus sites are observed in the 31P spectrum. All the chemical shifts appear in the range δ = 48–63 ppm, revealing the presence of several phosphorus sites in the sample. The total spinning sideband envelope covers a chemical-shift range of more than +250 ppm, suggesting that each site exhibits a significant chemical-shift anisotropy (CSA). A full-peak simulation of the spectrum permits an estimation of the relative amounts of each phosphorus site, as well as the associated principal CSA components [see the Supporting Information (SI): Table S4].[12] The simulation is challenging due to the rather broad peak widths and similar chemical shifts; therefore, these values are only rough estimates, and some sites may be missing due to overlaps. However, on neglecting P12, P13, and P14, the remaining sites show only slight differences in their principal components, indicating highly similar chemical environments. Although cross-polarization does not permit a direct comparison of peak areas due to differences in cross-polarization dynamics, the large variation in the estimated relative proportions indicates that the phosphorus sites do not belong to the same crystal phase, but are more likely to originate from different structural isomorphs. The asymmetric unit cell contains half a molecule and, as such, we expected to observe only one phosphorus site for each Ru-MACHO-like structure. Thus, the solid-state 31P spectrum suggests that at least six or seven similar structures were present in the sample. This is also corroborated by the 13C-{1H} CP/MAS spectrum, in which peaks are observed at about δ = 30, 50, and 130 ppm, assigned to P– C H2, N– C H2, and the phenyl rings, respectively. Multiple peaks with different intensities are observed at each chemical-shift region, confirming the presence of multiple structurally similar compounds. The CO site is only observed as a minor broad peak at δ = 206 ppm, due to inefficient polarization transfer.

To further study the structure and the isomers of the Ru-MACHO complex, the samples were dissolved in THF-d 8 and analyzed by multiple NMR experiments. A total of five signals were observed in the 31P spectrum (Figure [2], bottom), whereas the 1H spectrum (Figure [2], top) revealed the presence of two hydride signals (δ = –15.06 ppm and δ = –14.14 ppm), as well as several NH signals. In addition to the 31P peak from the expected Ru-MACHO complex (δ = 52.9 ppm), four other signals were observed with 31P chemical shifts of δ = 62.3, 56.0, 48.2, and 29.6 ppm, respectively. Except for the signal at δ = 29.6 ppm, these results matched the chemical shifts observed in the solid state. To investigate the general structure, we recorded a two-dimensional (2D) 1H-31P HMBC spectrum, which showed long-range correlations from both aromatic and aliphatic hydrogens to all the observed 31P peaks. Furthermore, the chemical shifts were highly deshielded, indicating coordination to a metal center, which confirmed that all the compounds are structurally similar to the Ru-MACHO compound. We therefore speculate that the observation of multiple peaks in the 31P spectrum is due to the presence of several Ru-MACHO isomers. The relative orientation and position of the ligands in the various structures affects the degree of complexation of the phosphorus atoms, which results in a large variation in the 31P chemical shifts. This was further corroborated by the 1D selective NOESY spectra (SI; Figure S2). Here, the hydride signal at δ = –14.1 ppm exhibited a NOE correlation to the NH, whereas the hydride signal at δ = –15.2 ppm showed a NOE correlation to H1b, reflecting different internuclear distances for the two isomorphs. The most intense peak at δ = 52.9 ppm also showed a long-range HMBC correlation from the hydride at δ = –15.06 ppm.

Zoom Image
Figure 2 1H and 31P solution-state NMR spectra for the two sample batches dissolved in THF-d 8. 1H decoupling was applied during the interscan delay and the acquisition time for the 31P measurement.

By using routine 2D NMR techniques, such as 1H-13C HSQC, 1H-13C HMBC, and 1H-1H COSY, it was possible to make a full peak assignment of the main Ru-MACHO compound (SI; Table S5). Interestingly, the two phenyl rings on the phosphorus atoms are seen to be nonequivalent, primarily showing different 1H and 13C chemical shifts closer to the phosphorus (i.e., at C5/C9, C6/C11, and H6/H10), probably due to the position of the ring relative to the H, CO, and Cl ligands. In this respect, H4 showed a long-range HMBC correlation (3 J) to C5 only, and not to C9. Such 3 J couplings are highly dependent on the dihedral angle (following the Karplus equation[13]) and could potentially provide crucial information to elucidate the full spatial arrangement of the ruthenium ligands.

Although the signals detected in the dissolved state match those detected in the solid state, there is a discrepancy between the relative signal intensities. To determine if this is caused by the choice of solvent, we decided to measure the 1H and 31P spectra in three additional solvents: acetone, acetonitrile (ACN) and dichloromethane (DCM). The results are shown in Figure [3]. Peaks with almost identical 31P chemical shifts are observed, but with a rather large variation in their relative proportions. The same is seen for the 1H spectra, where variations in chemical shifts and relative intensities are observed for the hydride signals. DCM is a particularly interesting solvent, with the presence of two hydride signals in high abundance. We applied 1D selective NOESY experiments, as well as 1H-13C HMBC, to investigate the structural differences for these two compounds. The NOESY spectra (SI; Figure S2) showed that the hydride at δ = –15.2 ppm is close in space to an aromatic hydrogen (corresponding to position H10) as well as the NH. Although the hydride at δ = –14.5 ppm also exhibits a correlation to an aromatic hydrogen, it does not have any correlations to an amine hydrogen. Conversely, in the HMBC spectrum (SI; Figure S3), the hydride at δ = –14.5 ppm has a strong long-range correlation to the CO ligand, whereas for the hydride at δ = –15.2 ppm, only a trace correlation peak can be seen in the chemical-shift region of CO, which reflects a difference in the dihedral angle between the hydride and CO in the two observed compounds. These findings therefore strongly indicate that most of the multiple phosphorus and hydride sites observed in both the solid and liquid state originate from Ru-MACHO complexes with various orientations of the H, NH, Cl, and CO ligands. Furthermore, it is established that the choice of solvent influences the stability of the different isomorphs.

The ATR-FTIR spectrum in the insert in Figure [4] clearly shows the presence of two strongly IR-active transitions for two different carbonyl ligands at ν(CO) = 1916 and 1907 cm–1, as well as three different, weakly IR-active, well-defined Ru–H stretching frequencies at 1941, 1961, and 1974 cm–1, respectively. In addition, two different broad (indicative of extensive hydrogen bonding) N–H stretching frequencies at 3139 and 3076 cm–1, respectively, are observed. The far-infrared (FIR) data (SI; Figure S6) show two distinct absorbances for the RuCl2-MACHO congener at 318 and 330 cm–1, respectively, which can be attributed to two different RuII–Cl stretching modes. It is peculiar to note that the same pattern is not immediately observed in the FIR data for Ru-MACHO; instead, only one very weak band at 315 cm–1 is observed.

Zoom Image
Figure 3 31P and 1H solution-state NMR spectra for the Sigma batch in acetone, MeCN, DCM, and THF. The 1H spectrum shows only the hydride chemical-shift region.

Density functional theory (DFT) simulations (gas phase) of the vibrational transitions under the harmonic oscillator approximation of the syn/anti-periplanar conformations of the PNP backbone are consistent with the experimental finding of two different carbonyl absorbances separated by 11 cm–1 as compared with 13 cm–1 observed in silico for this separation. A larger discrepancy is found when comparing the computationally predicted Ru–H and N–H stretching modes with experiment. The DFT simulations show that the syn-periplanar conformer is ~143 cm–1 bathochromically shifted relative to the anti-periplanar conformer. This frequency decrease is much more significant than the experimentally observed value of 88 cm–1. It is therefore difficult to conclude whether or not the observed red-shift between the two experimentally observed N–H stretching frequencies is due to the presence of syn/anti isomerism.

Zoom Image
Figure 4 ATR-FTIR spectra of commercially available Ru-MACHO from Strem (black trace) and Sigma (red trace). Both spectra show the same two peculiar, well-separated, different carbonyl stretching modes, as well as three different Ru–H stretching modes, as shown in the insert. See the main text for details.
Zoom Image
Figure 5 ATR-FTIR spectrum of RuCl2-MACHO obtained from the reaction between Ru-MACHO and hydrochloric acid in chloroform. The spectrum shows the characteristic normal stretching mode of the CO ligand at 1945 cm–1.

Interestingly, we have observed that the reaction of Ru–MACHO with hydrochloric acid under ambient conditions results in protonation of the hydrido ligand and a slow effervescence, presumably due to the elimination of dihydrogen. The FTIR spectrum of the resulting species (Figure [5]) revealed only a small contamination by Ru–H modes, and the two ν(CO) modes related to Ru–MACHO are merged into one less bathochromically shifted absorbance at 1945 cm–1, indicative of a weaker Ru–CO bond. The fact that only one ν(CO) stretching mode is observed for the reaction product is a strong indication that the two bands observed for Ru-MACHO are indeed a result of the presence of two different conformers, as the respective reaction product is symmetric along the axial bond vector and, as a result, syn/anti isomerism is spectroscopically irrelevant.

To elucidate the identity of the product species obtained from the reaction between hydrochloric acid and Ru–MACHO, yellow orthorhombic crystals of sufficient quality for X-ray diffraction analysis were grown by allowing diethyl ether to diffuse into the yellow chloroform solution under ambient conditions.

The crystallographic analysis confirmed our assumption that, upon reaction with HCl, the Ru–MACHO compound extrudes H2 to yield the trans-dichlorido species (RuCl2-MACHO) presented in Figure [6] (left). This species has previously been reported by Zhang et al.[14] However, these authors observed co-crystallization of the corresponding cis-dichlorido species, probably as a consequence of their chosen synthetic protocol, and it was not further investigated. Interestingly, the PNP backbone is slightly skewed relative to the PNP coordination plane, formally reducing the molecular symmetry from Cs to C1 , probably as a result of a packing effect. The structure resembles a distorted octahedron with a Cl–Ru–Cl angle of 174° and a PNP bite angle of 163°. The packing exhibits two hydrogen bonds between two neighboring molecules (Figure [6]; bottom left) with symmetry-related distances of Cl⋯H(N) ≈ 2.261 Å [Cl⋯N(H) = 3.149 Å].

Zoom Image
Figure 6 Top left: Crystal structure of the RuCl2-MACHO complex isolated from the reaction mixture of Ru-MACHO and chloroform (see the main text). Bottom left: Two hydrogen bonds are observed between the nearest-neighboring RuCl2-MACHO molecules in the solid-state structure. Top right: Ru-MACHO molecular structure showing that the amine proton and the hydrido ligand are oriented in a syn-periplanar manner. Bottom right: Depiction of a small part of the chain-like structure of hydrogen bonds between neighboring molecules in the solid-state structure. Thermal ellipsoids are plotted at 50% probability. All hydrogen atoms on carbon are omitted for clarity.

Because the Ru-MACHO product appeared to us to have a very low solubility in many common laboratory solvents such as THF, DCM, toluene, benzene, DMF, or DMSO, it was difficult to settle on an appropriate crystallization system without over-rapid precipitation of the product. This circumstance might contribute to an explanation of why there has been no previous elucidation of the crystal structure of Ru-MACHO.

High-quality crystals of Ru-MACHO were, however, obtained by slow diffusion of tert-butyl methyl ether into an almost-saturated solution of Ru-MACHO in DMSO under ambient conditions for seven days, after which the vial was slowly cooled to 4 ℃ for another 14 days. This yielded small, high-quality, rod-shaped crystals that could be harvested at the interface between the solvent and the precipitant. Ru-MACHO crystallizes in the Pnma space group and contains three co-crystallized molecules of DMSO per formula unit. The solvent molecules are disordered over two positions (SI; Figure S1). This model converged to R 1 = 4.24% and wR 2 = 11.03% [for reflections where I ≥ 2σ(I)].[15] This unsatisfactory agreement factor is attributed to the presence of disordered solvent of crystallization. As a check, a model omitting reflections arising from the solvent of crystallization was realized by using Platon Squeeze [16] (Figure [6]; right). After omitting the solvent-related contributions from the model, a new convergence to R 1 = 3.23% and wR 2 = 7.12% was reached. This is a clear indication that the Ru-MACHO in the present model is well defined. The structure resembles a distorted octahedral geometry with a PNP bite angle of 163°. The Ru–Cl distance of 2.541 Å is 0.143 Å longer than the mean Ru–Cl distance of 2.398 Å observed for the RuCl2-MACHO complex, reflecting the larger trans-influence of the hydrido ligand as compared with the chlorido ligand. A Hirschfelt atom refinement performed by using the program NoSpherA2 [17] in conjunction with Olex2 [18] gave a Ru–H distance of 1.46(3) Å (solvent-free model) and converged to R 1 = 2.97% and wR 2 = 6.40%. The hydrido ligand and the amine proton of the PNP ligand are oriented in a syn-periplanar manner, which is not typical for this type of monohydrido transition-metal pincer complex. In fact, only three such examples have previously been reported in the Cambridge Structure Database: as can be seen in Figure [7], they are all examples of PNP pincers with more-or-less the same auxiliary ligand sphere.[19]

Zoom Image
Figure 7 Upper three structures: Examples of monohydrido pincer complexes showing a syn configuration with crystal structures published in the Cambridge Structure Database. Lower structure in box: This work.

A striking difference in the packing between the RuCl2-MACHO and the Ru-MACHO crystal structure is seen in the extended chain structure of the latter, in which intermolecular hydrogen bonds between the chlorido ligand and the amine proton cause each Ru-MACHO molecule to interact directly with two neighboring molecules. This observation is corroborated by the observation that the N–H absorbance in the IR spectra is significantly red-shifted by ν < 3200 cm–1, which is in line with previously observed intermolecular hydrogen bonding in RuII amine complexes.[20] It cannot be ruled out that this extended network of hydrogen bonding is an important factor in the stabilization of the lattice and, therefore, makes a significant contribution to the unfriendly solubility profile of this product. In addition, the closest distance between aromatic planes of neighboring molecules in the Ru-MACHO crystal structure is only 4.33 Å, which falls within the range for π–π interaction, albeit on the long side.[21] It is tempting to assume that this Van der Waals interaction between neighboring molecules contributes significantly to the stabilization of the lattice, together with the aforementioned network of hydrogen bonds.

In conclusion, the solid-phase structure of the widely used Ru–MACHO precatalyst has been elucidated, together with that of its dichlorido derivative, obtained by the reaction of the parent complex with HCl in chloroform with liberation of H2. The peculiar spectral features, such as multiple Ru–H and CO vibrational absorbances, have been attributed to the presence of both the syn- and anti-periplanar conformations of (H)N–Ru–(H), a confirmation that has been shown to have a significant impact on the CO vibrational frequency. This conclusion has been corroborated by DFT calculations. Furthermore, the fact that the RuCl2­-MACHO complex shows only one CO stretching frequency reveals that the presence of a hydride in the Ru-MACHO structure imposes significant changes in the electron-density distribution around the Ru center, whereby a nonnegligible difference between the syn- and anti-conformers makes these spectroscopically distinguishable. Unfortunately, it has not been possible, in our hands, to isolate other species that would explain the peculiar NMR results and thereby disclose why so many apparently different species are present in the product.


#

Conflict of Interest

The authors declare no conflict of interest.

Acknowledgment

We are grateful for the help and support supplied by the NMR Center at The Technical University of Denmark.

Supporting Information

    Supporting information for this article is available online at https://doi-org.accesdistant.sorbonne-universite.fr/10.1055/a-1337-2603-8157.
  • Supporting Information

Corresponding Author

Martin Nielsen
Technical University of Denmark, DTU Chemistry
Kemitorvet 207, 2800 Kgs. Lyngby
Denmark   

Publication History

Received: 28 March 2025

Accepted after revision: 08 May 2025

Accepted Manuscript online:
08 May 2025

Article published online:
18 June 2025

© 2025. Thieme. All rights reserved

Georg Thieme Verlag KG
Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany


   Permissions and Reprints All articles of this category
Zoom Image
Figure 1 (a) 31P-{1H} CPMAS solid-state NMR spectrum. The insert shows the region with the isotropic chemical shifts, as well as the simulated spectrum (in red). (b) 13C-{1H} CPMAS solid-state NMR spectrum. The asterisk denotes a spinning sideband not properly suppressed by the TOSS sequence.
Zoom Image
Figure 2 1H and 31P solution-state NMR spectra for the two sample batches dissolved in THF-d 8. 1H decoupling was applied during the interscan delay and the acquisition time for the 31P measurement.
Zoom Image
Figure 3 31P and 1H solution-state NMR spectra for the Sigma batch in acetone, MeCN, DCM, and THF. The 1H spectrum shows only the hydride chemical-shift region.
Zoom Image
Figure 4 ATR-FTIR spectra of commercially available Ru-MACHO from Strem (black trace) and Sigma (red trace). Both spectra show the same two peculiar, well-separated, different carbonyl stretching modes, as well as three different Ru–H stretching modes, as shown in the insert. See the main text for details.
Zoom Image
Figure 5 ATR-FTIR spectrum of RuCl2-MACHO obtained from the reaction between Ru-MACHO and hydrochloric acid in chloroform. The spectrum shows the characteristic normal stretching mode of the CO ligand at 1945 cm–1.
Zoom Image
Figure 6 Top left: Crystal structure of the RuCl2-MACHO complex isolated from the reaction mixture of Ru-MACHO and chloroform (see the main text). Bottom left: Two hydrogen bonds are observed between the nearest-neighboring RuCl2-MACHO molecules in the solid-state structure. Top right: Ru-MACHO molecular structure showing that the amine proton and the hydrido ligand are oriented in a syn-periplanar manner. Bottom right: Depiction of a small part of the chain-like structure of hydrogen bonds between neighboring molecules in the solid-state structure. Thermal ellipsoids are plotted at 50% probability. All hydrogen atoms on carbon are omitted for clarity.
Zoom Image
Figure 7 Upper three structures: Examples of monohydrido pincer complexes showing a syn configuration with crystal structures published in the Cambridge Structure Database. Lower structure in box: This work.