Asymmetric hydrogenation of ketones with planar chiral manganese(I) complexes

Abstract

Tridentate P,N,N ferrocenyl ligands 4a-b with planar chirality only have been prepared and evaluated in the manganese-catalysed hydrogenation of aromatic ketones. The Mn^I complexes 5a-b, prepared in situ from ligands 4a-b and [MnBr(CO)₅], showed a complex mixture of stereoisomers and cationic tricarbonyl/neutral dicarbonyl species. The catalytic activity of both complexes is excellent, with a total conversion of acetophenone in 4 h in the presence of 0.1 mol% catalyst and the mild K₂CO₃ base, under 30 bar of H₂ at 50 °C. Ees of up to 69% were obtained with bulky pivalophenone in the presence of 1 mol% catalyst 5b. The results of a computational study, carried out on the Mn-catalysed hydrogenation mechanism, agree with the greater activity and lower enantioselectivity of 4a with respect to previously reported P,N,N ferrocenyl ligands bearing both planar and central chirality, highlighting the useful role of each chiral source.

---

Introduction

Hydrogenation of unsaturated substrates is one of the most prominent processes in synthetic chemistry. Although many strategies exist, the use of dihydrogen is one of the cleanest and most selective. It becomes particularly important when dealing with chiral compounds, which constitute ca. 50% of the drugs available on the market.¹ However, activation of dihydrogen requires the use of catalysts, mainly metal catalysts and, until recently, noble metals were mostly used.² Thanks to a paradigm change that has put sustainability and environmental issues at the centre of chemists’ concerns, the last ten years have witnessed a surge of catalytic systems based on inexpensive, more abundant first-row transition metals.^2–11 Yet, finding active and selective systems, expanding the range of substrates and understanding mechanisms are outstanding challenges.

In the case of asymmetric catalysis, it is also necessary to couple good activity with high enantioselectivities, using relatively easy-to-access chiral ligands whenever possible. Our group has a long experience in developing planar chiral polytopic ferrocenyl ligands for asymmetric catalysis.^12–17 In particular, ferrocenyl thioether-phosphines^18,19 have proven very efficient in the iridium-catalysed hydrogenation of ketones^20,21 and alkenes.²² Our intention to move towards a more sustainable approach of catalysis prompted us to explore the chemistry of manganese. Indeed, since the first report of Beller et al. in 2016,²³ soon followed by the groups of Kempe²⁴ and Milstein²⁵ (Fig. 1), manganese(I) complexes have emerged as catalysts of choice for the hydrogenation of carbonyl groups. A literature survey shows that most successful systems are based on pincer tridentate ligands, although several catalysts bearing bidentate ligands also proved efficient.^8,9,26 Since the first use of a chiral manganese(I) complex in 2017 by Zirakzadeh and Kirchner in asymmetric transfer hydrogenation (TH),²⁷ a limited number of chiral manganese-based catalysts have been described for the asymmetric hydrogenation of carbonyl groups (Fig. 2).^26,28–39

Fig. 1 First Mn-based pincer-type complexes for hydrogenation reactions.

Fig. 2 First chiral Mn^I complexes for the transfer hydrogenation (top) and hydrogenation (bottom) of carbonyl groups.

The first example came with the group of Clarke et al., who disclosed in 2017 very active manganese(I) complexes bearing tridentate P,N,N ferrocenyl ligands with excellent enantioselectivities with aromatic ketones.²⁹ Soon after, Beller et al. reported manganese(I) complexes bearing chiral P,N,P ligands which gave high enantioselectivities with aliphatic ketones, but surprisingly much lower with aromatic ketones.²⁸

Previous work has shown that the presence of a coordinating NH group, such as in Clarke's P,N,N ligand, is often crucial for catalytic activity.^{31,33,38,40–42} After their first report in 2017, Clarke et al. explored the scope of their ferrocenyl ligands in several manganese-catalysed hydrogenation reactions (Fig. 3). In this way, a variety of ketones^29–31 as well as esters^29,30,43 and in situ-generated imines⁴⁴ could be hydrogenated.

Fig. 3 Mn(I) complexes with chiral P,N,N ferrocenyl ligands for the hydrogenation of ketones in EtOH.

Their ligands, derived from Ugi's amine, display both central and planar chirality and allowed reaching very high enantioselectivities in the hydrogenation of a range of aryl ketones. However, Ugi's amine and its derivative (S,R)-PPFA⁴⁵ are quite expensive and only one diastereoisomer of the derived ligands has been investigated in those reports, precluding the evaluation of a possible “match/mismatch effect”.

Given our expertise in the development of ferrocenyl ligands bearing planar chirality only, we wished to examine the influence of each type of chirality present in these ligands: what is the effect of the central chirality on activity and enantioselectivity? This prompted us to design and synthesize new versions of Clarke's P,N,N complex I, i.e.5a and 5b, with no asymmetric carbon between the ferrocene moiety and the NH group. We also carried out theoretical calculations to compare Clarke's system and the non-methylated counterpart.

Results and discussion

Synthesis of ligands and complexes

The synthesis of planar chiral P,N,N ligands 4 starts from enantiopure ferrocenyl alcohol 1 (Scheme 1), whose synthesis has been developed in our group.^14,18 Both enantiomers are available and both were used for the synthesis of enantiopure 4 and evaluation in catalysis. Thus, HBF₄·Et₂O was added to the ferrocenyl phosphinoalcohol 1, followed by pyridine to form the intermediate pyridinium salt 2. After workup and recovery of 2, 2-picolylamine was added and the reaction mixture in MeCN was refluxed for 24 h to give product 3a in 68% yield from 1. Finally, the phosphine deprotection was carried out by refluxing 3a in toluene in the presence of P(NMe₂)₃; purification by column chromatography on silicagel gave ligand 4a as a crystalline solid in 80% yield. A derivative bearing a methoxy substituent on the pyridyl group was also prepared: the introduction of an electron-donating substituent in ortho to the coordinating nitrogen is expected to play a role in catalysis. Thus, the 6-methoxy-2-pyridinemethanamine was introduced on the ferrocene backbone by reaction with the pyridinium salt 2 to produce 3b in 54% yield. Subsequent phosphine deprotection gave the expected ligand 4b in 71% yield. With these two ligands in hand, we studied their coordination chemistry on [MnBr(CO)₅].

Scheme 1 Synthesis of P,NH,N ligands 4a-b from ferrocenylphosphino alcohol 1 ((S)-enantiomer represented).

Although most catalytic reactions were carried out by pre-mixing ligands 4a-b and [MnBr(CO)₅] in the chosen solvent before hydrogenation, the manganese(I) complexes were also synthesized, isolated and characterized so that their performances could be compared to those of the in situ-generated catalysts. To this aim, the ligand 4a and [MnBr(CO)₅] were heated under argon for 24 h in a cyclohexane/toluene mixture to achieve complete conversion. Three species were observed by ³¹P{¹H} NMR in CD₂Cl₂: a major signal at 89.85 ppm and two smaller signals at 56.51 and 45.11 ppm in an approximate 1 : 0.46 : 0.34 ratio. In the ¹H NMR spectrum, three signals appeared above 8.5 ppm, of which one situated at 9.2 ppm, typical of protons in ortho to a coordinated pyridine nitrogen. We initially envisaged that the expected fac- or mer-[Mn(CO)₃(4a)]⁺Br⁻ cationic complex is the major species, while the other species may either be the other geometrical isomer, or a neutral complex still possessing the bromide ligand but bearing only two CO ligands (Scheme 2). A similar outcome was obtained when preparing complex 5b: three signals were observed by ³¹P NMR in CD₂Cl₂ at 90.1, 60.9 and 45.4 ppm (ca. 1 : 0.2 : 1.4 ratio), along with a very small signal at −23.5 ppm, typical of a free phosphine.

Scheme 2 Synthesis of the manganese complexes 5a-b, with three envisaged structures shown.

When the counteranion of 5b was changed from Br⁻ to BARF⁻ and the ³¹P NMR recorded in acetone-d₆, the same three signals were observed with a different distribution, the species at 90 ppm becoming the major one (2.5 : 0.7 : 1). The signal at −23.5 ppm disappeared, replaced by a small signal at ca. 74 ppm, probably corresponding to the complex bearing a non-coordinated pyridine moiety.²⁹ In order to further probe the influence of the solvent, solid-state NMR analyses were performed on both 5a and 5b: the ³¹P spectra show only one species at ca. 90 ppm for 5a, whereas two species were observed in the case of 5b, the major one at ca. 90 ppm with a second, minor species at ca. 45 ppm. IR analysis of the complexes by ATR confirmed the solid NMR analysis, as 5a showed two intense CO bands at 1924 and 1830 cm⁻¹ and an additional medium band at 1850 cm⁻¹, whereas the spectrum of 5b (ATR) shows the presence of at least four bands: this may indicate the coexistence of two complexes: a cationic complex fac-5b, bearing three carbonyl ligands (medium bands at 2034 and 1957 cm⁻¹ of equal intensity, the third band being probably hidden under the 1918 cm⁻¹ band), and a neutral complex with a bromide and only two CO ligands (two strong bands at 1918 and 1832 cm⁻¹). IR spectra of 5a-b were also recorded in DCM solutions so as to evaluate the influence of the solvent on the ligand organisation around the metal: apart from a small blue shift of most bands, the number and distribution was the same as in the solid state for complex 5b. However, the spectrum becomes simpler for complex 5a, as only two bands are now visible, corresponding to the two major bands present in the ATR spectrum. Finally, mass spectrometry analysis of 5a confirmed the presence of the [M]⁺ peak, albeit very weak, whereas the main observed species is the cation with only two CO ligands. On the other hand, the mass spectrum of 5b shows [M]⁺ as the major peak, i.e. the cationic species bearing three CO ligands, along with peaks of species resulting from the loss of one, two or three CO.

In their 2019 paper, Clarke et al. discussed the structure of complex I (Fig. 3) bearing either a Br⁻ anion or a non-coordinating BARF⁻ anion and observed the reversible interconversion of two species at ca. 90 ppm and 45 ppm by simply changing the polarity of the NMR solvent: while the 90 ppm species is overwhelmingly present in acetone-d₆, the 45 ppm species is largely predominant in an alcoholic solvent (Table 1).³⁰ They attributed the two species as two geometrical fac and mer isomers of the cationic species bearing three carbonyl ligands. The infrared analysis (ATR) of solid I confirmed their hypothesis, with the presence of three bands at 2037, 1962 and 1931 cm⁻¹, whereas mass spectrometry showed the presence of three carbonyl ligands. This is somewhat logical as BARF⁻ is a non-coordinating anion, with a low propensity to interact with the Mn centre. However, analysis of I with a bromide anion showed a different outcome: whereas ³¹P NMR and mass spectrometry analyses seemed to confirm the presence of a tricarbonyl complex, the infrared spectrum (ATR) only revealed two bands at lower wavenumbers (1921 and 1842 cm⁻¹), similar to that of our complex. These values are typical of neutral manganese(I) complexes bearing two carbonyl ligands: in addition to the complex described by Milstein et al.⁴⁶ and previously mentioned by Clarke et al., all examples of neutral Mn^I complexes of this type and bearing tridentate P,N,N ligands display only two bands in the carbonyl region, usually at 1915–1925 cm⁻¹ and 1830–1846 cm⁻¹.^38,47–49 Furthermore, Morris et al. described in 2018 two Mn^I complexes bearing P,N,N ligands, with a small structural variation: whereas the neutral complex, bearing one bromide and two carbonyl ligands, displayed two IR bands at 1915 and 1828 cm⁻¹, the analogous cationic complex bearing three carbonyl ligands and a bromide counteranion, showed three bands at 2032, 1955 and 1914 cm⁻¹ (ATR).⁴⁷ Finally, manganese(I) complexes described by Liu et al. in 2021 and possessing fac-arranged tridentate P,N,N ligands according to an XRD study, exhibit only one species in acetone-d₆ with a ³¹P NMR resonance at ca. 45 ppm, whereas the HR-MS data indicate a species possessing only two carbonyl ligands. Unfortunately, no IR data was reported for these complexes.⁴⁹

Table 1 Comparison of IR and ³¹P NMR data of reported P,N,N-manganese(I) complexes

Mn^I complex	^31P NMR (ppm, solvent)	IR, ν(CO) (cm^−1)
	5a (X = Br): 89.85 (M), 56.51, 45.11 (1 : 0.46 : 0.34) (CD2Cl2)	5a: 1924 (s), 1850 (m), 1830 (s) (ATR)
	5b (X = Br): 90.09, 60.88, 45.43 (M) (1 : 0.2 : 1.4) (CD2Cl2)	5b: 2034 (m), 1957 (m), 1918 (s), 1832 (s) (ATR)
	5b (X = BARF): 90.07 (M), 61.88, 43.91 (acetone-d6)
	I (X = BARF):^29,30 90.17 (acetone-d6)	I (X = BARF):^29,30 2037, 1962, 1931 (ATR)
	ca. 45 (alcohol, not specified)	I (X = Br):^29,30 1921, 1842 (ATR)
		II (X = Br):^30 2025, 1942, 1905, 1831 (ATR)
	I (X = Br):^29,30 90.14 (acetone-d6)
	ca. 45 (EtOD)
	II (X = Br):^30 89.13, 43.64 (CD2Cl2)
	89.20 (M), 43.86 (acetone-d6)
	86.49, 62.93, 42.17 (M) (MeOD)
	89.2 (CD2Cl2)	1917, 1838 (NaCl)
	63.9 (DMSO-d6)	2032, 1955, 1914 (ATR)
	83.1 (DMSO-d6)	1915, 1828 (+impurity at 2023) (ATR)
	R = H, X = Br: 46.87 (acetone-d6)	Not reported
	R = H, X = BARF: 44.24 (acetone-d6)
	R = Me, X = BARF: 44.53 (acetone-d6)

---

Slow crystallisation of 5a from a CH₂Cl₂/hexane mixture allowed us to confirm the solid-state structure of 5a by X-ray diffraction as fac-octahedral, with the tridentate ligand occupying three mutually cis positions around the metal centre, three carbonyl ligands and a non-coordinated Br⁻ anion (Fig. 4).

Fig. 4 Molecular views of 5a (top) and 5b (bottom) showing 50% probability displacement ellipsoids and the atom numbering scheme. The H atoms are omitted for clarity.

Whereas most structural features are similar to those of analogous manganese(I) complexes described in the literature, the absence of the methyl substituent in α to the ferrocenyl unit led to small changes compared to Clarke's reference complex I (Table 2): the Mn(1)–P(1) and Mn(1)–N(2) bonds in 5a are slightly shorter, and the N(1)–Mn(1)–P(1) angle is significantly more open than in Clarke's complex (95.00(10)° for 5a against 90.59(19)° for I), showing a high deviation from the ideal 90°. Additionally, the N(1)–Mn(1)–C(1) angle, measured at 169.42(17)°, is far from the ideal 180° value: this may be due to the presence of the bulky phosphine moiety, pushing the carbonyl ligand. The carbonyl ligand trans to the phosphorus donor also presents an important deviation from linearity with a O(2)–C(2)–Mn(1) angle of 170.8(4)°. The crystals were further analysed by NMR in CDCl₃, giving two main ³¹P{¹H} NMR signals at 90.6 and 56.4 ppm in a 2.75 : 1 ratio, along with a very small signal at 45.9 ppm. However, the single crystals were so small that it was difficult to pick them up cleanly for NMR analysis. Therefore, it is difficult to conclude whether this ratio is due to rapid scrambling of the species in solution or to the presence of non-crystalline material in the NMR solution.

Table 2 Bond lengths and angles of complexes 5a-b compared to literature complex I

Bond lengths/angles (Å/°)	Complex 5a (Br^−)		Complex 5b (MnBr4^2−)		Complex I (BArF^−)^29
a This value is probably erroneously reported in ref. 29.
trans to Py	C(1)–Mn(1)	1.803(4)	C(3)–Mn(1)	1.797(4)	1.783(10)
trans to P	C(2)–Mn(1)	1.847(5)	C(2)–Mn(1)	1.888(5)	1.824(11)
trans to NH	C(3)–Mn(1)	1.810(4)	C(1)–Mn(1)	1.810(4)	1.787(10)
Mn–Py	N(1)–Mn(1)	2.068(3)	N(1)–Mn(1)	2.114(3)	2.053(7)
Mn–NH	N(2)–Mn(1)	2.099(3)	N(2)–Mn(1)	2.098(3)	2.122(7)
Mn–P	Mn(1)–P(1)	2.3224(11)	Mn(1)–P(1)	2.3363(10)	2.353(2)
	N(1)–Mn(1)–N(2)	79.97(13)	N(1)–Mn(1)–N(2)	77.82(12)	79.4(3)
	N(1)–Mn(1)–P(1)	95.00(10)	N(1)–Mn(1)–P(1)	92.76(9)	90.59(19)
	N(2)–Mn(1)–P(1)	88.86(10)	N(2)–Mn(1)–P(1)	88.59(9)	88.32(18)
	C(1)–Mn(1)–N(1)	169.42(17)	C(3)–Mn(1)–N(1)	171.83(14)	90.6(5)^a
	C(3)–Mn(1)–N(2)	174.20(16)	C(1)–Mn(1)–N(2)	175.77(16)	175.6(4)
	C(2)–Mn(1)–P(1)	175.85(14)	C(2)–Mn(1)–P(1)	178.33(13)	174.8(3)
trans to Py	O(1)–C(1)–Mn(1)	175.5(4)	O(3)–C(3)–Mn(1)	176.8(3)	175.9(8)
trans to P	O(2)–C(2)–Mn(1)	170.8(4)	O(2)–C(2)–Mn(1)	174.6(4)	173.0(8)
trans to NH	O(3)–C(3)–Mn(1)	179.8(5)	O(1)–C(1)–Mn(1)	176.8(4)	179.4(9)

---

Similarly, the fac-octahedral structure of 5b was confirmed by X-ray analysis of crystals obtained by slow crystallization from a CH₂Cl₂/hexane mixture. It shows again a fac-tridentate coordination of ligand 4b, but surprisingly the anion turned out to be the manganese(II) complex [MnBr₄]²⁻ shared between two units of the cationic manganese(I) complex (see SI for details). This can be rationalized by an adventitious partial oxidation during the crystallization process.

The Mn(1)–P(1) and Mn(1)–N(2) bonds are in the same range as those of fac-5a, whereas the C(2)–Mn(1) bond (trans to the phosphine) is significantly longer in this case (1.847(5) Å for fac-5a against 1.888(5) Å for fac-5b). Again, the N(1)–Mn(1)–P(1) angle is much wider than the other heteroatom–manganese–heteroatom angles with a value of 92.76(9)°, and the C(3)–Mn(1)–N(1) angle is much narrower than the ideal 180° value, with 171.83(14)°.

The data collected in the literature and by ourselves is contrasting and makes it difficult to conclude about the organisation of the coordination sphere around manganese. In an attempt to solve this issue, we conducted a frequency analysis by DFT calculations to simulate the IR spectra for the most plausible species. Preliminary investigations were done on a simplified ligand model (see SI, Fig. S34), and calculations on the full molecules were done only on the four lower-energy structures: fac-[Mn(CO)₃(4a)]⁺Br⁻ (two diastereomers), mer-[Mn(CO)₃(4a)]⁺Br⁻, fac-[MnBr(CO)₂(4a)] with Br trans to P and mer-[MnBr(CO)₂(4a)] with Br in the axial position (Fig. S35). The entire molecules were calculated at the full quantum-mechanical level. The optimized geometries and IR spectra in the CO stretching region are shown in Fig. S35, with relative Gibbs energies and calculated frequencies in cm⁻¹. There are in fact two possible fac tricarbonyl structures, the first corresponding to that experimentally observed by X-ray diffraction and the second one with the opposite central chirality at the metal centre. The second one has a higher Gibbs energy by 3.5 kcal mol⁻¹, suggesting that the ligand chirality fully controls the metal chirality. The mer-tricarbonyl isomer has an even higher Gibbs energy. Thus, these two alternative tricarbonyl isomers should only be present in insignificant amounts in solution at equilibrium to be observed (0.27% and 0.14%, respectively, relative to the major fac isomer). The calculated pattern for the more stable fac isomer matches quite well that of the minor observed species (Fig. 5), except for a shift of all bands to higher wavenumbers, which is a common phenomenon for DFT-calculated IR spectra. Concerning the neutral complexes, their calculated Gibbs energies are higher than those of the ionic tricarbonyl species, but the significance of this comparison is limited, because they correspond to the standard state (e.g. all 1 M concentrations including for CO in solution). At the quite low [CO] that is likely present in solution, the dicarbonyl species will obviously be stabilized, rendering their generation possible, especially when flushing the solution with Ar to remove the dissociated CO ligand. The relative Gibbs energy of the two dicarbonyl isomer may however be directly compared and this is largely in favour of the fac isomer. Again, the two calculated bands for this species, besides a slight shift to higher wavenumbers, match those experimentally observed for the major species (Fig. 5). Their placement at much lower wavenumbers than those of the cationic complex is confirmed. Thus, according to the calculations, the observed second isomer at equilibrium in solution could only be the dicarbonyl fac-species, assuming the CO molar concentration is low enough. Indeed, if pumping off CO completely, the fac-dicarbonyl complex should remain the only residual complex. This could partly explain the discrepancies in the results reported in the literature and obtained with the different characterization techniques, as those may have been obtained following repeated solvent changes or vacuum-pumping of the solvents. Although the data on our compounds 5a and 5b were obtained with the highest possible care, we cannot assign with confidence the structure of the third minor species observed for 5a.

Fig. 5 Experimental (left) and calculated (right, fac isomers) IR spectra of 5a. For the latter, an artificial Lorentzian line broadening of 12 cm⁻¹ (width at half-height) was applied.

Catalysis: hydrogenation of aromatic ketones

Acetophenone, a common standard for a first evaluation of catalytic systems, was first investigated in the inexpensive and benign ethanol (Table 3). A sub-stoichiometric amount of base was added, as it was previously proved essential.²³ K₂CO₃ was initially chosen, both for its low cost and its inertness towards sensitive functional groups. At 1 mol% of a 4a/[MnBr(CO)₅] mixture, using the racemic ligand, the conversion into 1-phenylethanol was complete after 16 h at 50 °C under 40 bar of H₂ (entry 1). Pleasingly, the conversion was still complete under milder conditions (6 h at 50 °C and also at 30 °C under 30 bar of H₂, entries 2 and 3). Only after 2 h at 30 °C was the conversion incomplete (41%, entry 4).

Table 3 Mn-catalysed hydrogenation of acetophenone in EtOH with racemic ligand 4a^a

Entry	Base	H2 pressure (bar)	T (°C)	t (h)	Conv.^b (%)
a Typical reaction conditions: 1.09 mmol acetophenone, 1 mol% rac-4a, 1 mol% [MnBr(CO)5], 10 mol% base, 1.09 mmol internal standard (dodecane), EtOH (1.6 mL, 0.55 M). b Determined by GC analysis. c 0.85 mmol acetophenone, 1 mol% rac-5a complex, 10 mol% base, 0.85 mmol internal standard (1-Me-1-naphthalene), EtOH (4.0 mL, 0.2 M), conversion determined by ^1H NMR.
1	K2CO3	40	50	16	>99
2	K2CO3	30	50	6	>99
3	K2CO3	30	30	6	>99
4	K2CO3	30	30	2	41
5	tBuOK	30	30	2	81
6	NaOMe	30	30	2	57
7^c	K2CO3	30	30	4	>99

---

When stronger bases were used, a higher conversion was obtained, with a promising 81% in the presence of tBuOK (entry 5). Routine blank tests showed that no conversion occurred when the manganese complex, the base or H₂ were missing, and a similar outcome was observed with [MnBr(CO)₅] without ligand. The presence of the N–H functional group proved essential, since 1% conversion of acetophenone was observed when using a similar ligand (4a-Me) bearing a N–Me group instead,⁵⁰ under otherwise identical conditions (see Table 4, entry 4 and SI for synthetic details). Finally, using the isolated manganese complex 5a instead of the in situ-generated complex gave essentially the same results (entry 7), hence the remainder of the investigation involved premixing 4a and [MnBr(CO)₅] just before catalysis.

Table 4 Mn-catalysed hydrogenation of acetophenone: effect of solvent, base and substrate concentration^a

Entry	[Ketone] (mol L^−1)	Base	Solvent	H2 pressure (bar)	t (h)	Conv.^b (%)	ee^b (%)
a Typical reaction conditions: 0.55 M: 1.09 mmol acetophenone, 0.011 mmol rac-4a or (S)-4a, 0.011 mmol [MnBr(CO)5], 0.109 mmol base, 1.09 mmol internal standard (dodecane), solvent (1.6 mL), 30 °C. 0.2 M: 0.34 mmol acetophenone, 0.0034 mmol 4a-rac or 4a-(S) or 4a-(R), 0.0034 mmol [MnBr(CO)5], 0.034 mmol base, 0.85 mmol internal standard (dodecane), solvent (4 mL), 30 °C. b Determined by GC analysis, known absolute configuration in parentheses. c Reaction carried out with ligand 4a-Me.
1	0.55	tBuOK	iPrOH	30	2	99	—
2	0.55	tBuOK	iPrOH	30	1	94	—
3	0.2	tBuOK	iPrOH	30	2	>99	41 (S)
4^c	0.2	tBuOK	iPrOH	30	2	1	—
5	0.2	K2CO3	iPrOH	30	2	99	—
6	0.2	tBuOK	iPrOH	0	2	46	55 (S)
7	0.2	tBuOK	EtOH	30	4	>99	58 (S)
8	0.55	tBuOK	EtOH	0	2	1	—
9	0.2	K2CO3	EtOH	0	4	0	—
10	0.2	tBuOK	MeOH	30	4	58	54 (S)
11	0.2	tBuOK	MeOH	0	16	0	—
12	0.2	tBuOK	tBuOH	30	4	0	—
13	0.2	tBuOK	tBuOH	30	8	25	17 (S)
14	0.2	tBuOK	tBuOH	30	22	91	17 (S)
15	0.2	tBuOK	t-amOH	30	4	0	—
16	0.2	tBuOK	Toluene	30	4	0	—

---

The influence of the solvent on the activity and enantioselectivity was evaluated using 4a in the presence of tBuOK (Table 4). In iPrOH, the conversion was total after 2 h and very high after 1 h (entries 1 and 2). Lowering the ketone concentration to 0.2 M did not have any notable influence on the conversion under H₂, and 41% ee was obtained (entry 3). As mentioned earlier, the manganese complex formed with N-methylated ligand 4a-Me instead of 4a proved unreactive (entry 4). Ligand 4a showed a high activity in the presence of K₂CO₃ even with a lower concentration (entry 5). However, iPrOH is known to efficiently promote TH and, indeed, a test carried out without H₂ gave 46% conversion and 55% ee after 2 h (entry 6).⁵¹ When a reaction was carried out under H₂ in EtOH, a poorer reagent for TH, a higher 58% ee was found (entry 7) and blank tests in EtOH without H₂ (entries 8 and 9) confirmed the absence of significant conversion. As hydrogenation and TH may proceed via different mechanisms, the ee value obtained under H₂ in iPrOH (41%) may result from the interplay of two mechanisms. In order to rule out this possibility, it was decided to avoid iPrOH, even if it gives faster reactions. In the more challenging MeOH, a lower conversion was observed under 30 bar H₂ with a similar ee as in EtOH (54%, entry 10), whereas no conversion occurred under TH conditions (entry 11).

Higher alcohols were also tested but all led to lower conversions and ees (entries 12–15). In the case of tBuOH, we hypothesized that this was due to its high viscosity, but the use of lower melting t-amyl alcohol proved futile (entry 15). As similar catalytic systems showed very high activities in a tBuOK/toluene medium,^26,28 these conditions were also evaluated but no conversion was observed in our case (entry 16).

As promising results were obtained in terms of activity with K₂CO₃ in EtOH under mild conditions (Table 3, entry 7), we checked whether the nature of the base could have an influence on enantioselectivity. To our delight, similar ees were reached with tBuOK and K₂CO₃ under the same conditions (Table 5, entries 1 and 2). When the catalytic charge was lowered to 0.1 mol%, 16 h were necessary to reach a modest 27% conversion (entry 3). The ketone concentration was thus increased to 0.6 M, and a better 85% conversion was obtained, albeit with a lower 21% ee (entry 4). Raising the temperature to 50 °C allowed reaching complete conversion and a higher ee (50%, entry 5). Under identical conditions, the Clarke system I gave a 70% ee with a complete conversion,³⁰ showing the importance of the central chirality on the ferrocenyl ligand. However, our system proved much more active, since a complete conversion was still reached when the reaction time was shortened to 4 h (entry 7), against a 37% conversion for Clarke's.³⁰ Even after 2 h, a very satisfying 81% conversion was achieved, with 56% ee (entry 8). Finally, the role of H₂ pressure and temperature were further analysed: whereas the pressure could be lowered to 30 bar without any loss in activity (and even a slight gain in enantioselectivity, entry 10), setting the temperature at 30 °C resulted in an important drop in the conversion (35%, entry 9).

Table 5 Mn-catalysed hydrogenation of acetophenone in EtOH: effect of substrate concentration, catalyst loading, Base/Mn ratio, H₂ pressure and temperature^a

Entry	[Ketone] (mol L^−1)	Cat. loading (mol%)	Base	Amount base (mol%)	H2 pressure (bar)	T (°C)	t (h)	Conv.^b (%)	ee^b (%)
a Typical reaction conditions: (1) [ketone] = 0.2 M: 0.85 mmol acetophenone, 0.0085 mmol (S)-4a, 0.0085 mmol [MnBr(CO)5], 0.085 mmol base, 0.86 mmol internal standard (dodecane), EtOH (4 mL); (2) [ketone] = 0.6 M: 2.4 mmol acetophenone, 0.0025 mmol (S)-4a, 0.0024 mmol [MnBr(CO)5], 0.12 mmol base, 0.84 mmol internal standard (dodecane), EtOH (4 mL); (3) [ketone] = 0.7 M: 2.8 mmol acetophenone, 0.0028 mmol ligand, 0.0028 mmol [MnBr(CO)5], 0.14 mmol base, 0.84 mmol internal standard (dodecane), EtOH (4 mL). b Determined by GC analysis, known absolute configuration in parentheses.
1	0.2	1	tBuOK	10	30	30	4	>99	58 (S)
2	0.2	1	K2CO3	10	30	30	4	>99	57 (S)
3	0.2	0.1	K2CO3	5	30	30	16	27	n.d.
4	0.6	0.1	K2CO3	5	30	30	16	85	21 (S)
5	0.6	0.1	K2CO3	5	30	50	16	>99	50 (S)
6	0.7	0.1	K2CO3	5	50	50	16	>99	51 (S)
7	0.7	0.1	K2CO3	5	50	50	4	>99	49 (S)
8	0.7	0.1	K2CO3	5	50	50	2	81	56 (S)
9	0.7	0.1	K2CO3	5	50	30	4	35	59 (S)
10	0.7	0.1	K2CO3	5	30	50	4	>99	59 (S)

---

The influence of the reaction time on conversion and enantioselectivity was further investigated (Table 6). For this purpose, the temperature was kept at 30 °C to work under mild conditions, thus the catalytic charge was fixed at 1 mol% to keep a reasonable activity. A 36% conversion was measured after 2 h, which compares well with the 41% conversion obtained in a more concentrated medium (Table 3, entry 4), with 60% ee. The conversion was complete after 4 h and a slight drop in ee was observed at longer times, with 55% ee after 16 h (entry 4).

Table 6 Mn-catalysed hydrogenation of acetophenone in EtOH: influence of the reaction time on conversion and ee at 30 °C.^a

Entry	t (h)	Conv.^b (%)	ee^b (%)
a Typical reaction conditions: 0.85 mmol acetophenone, 0.0080 mmol (S)-4a, 0.0080 mmol [MnBr(CO)5], 0.084 mmol K2CO3, 0.84 mmol internal standard (dodecane), EtOH (4 mL, 0.2 M), 30 bar H2, 30 °C. b Determined by GC analysis, known absolute configuration in parentheses.
1	2	36	60 (S)
2	3	93	57 (S)
3	4	>99	57 (S)
4	16	>99	55 (S)

---

With the optimal conditions in hand, we then evaluated ligand 4b (Table 7). A slower reaction was observed, with a 25% conversion after 2 h (entry 2) and only 38% conversion after 4 h (entry 4), vs. a total conversion after 4 h with 4a (entry 3).

Table 7 Mn-catalysed hydrogenation of acetophenone in EtOH: comparison of ligands 4a and 4b^a

Entry	Cat.	t (h)	Conv.^b (%)	ee^b (%)
a Typical reaction conditions: 0.85 mmol acetophenone, 1 mol% (S)-4a-b, 1 mol% [MnBr(CO)5], 10 mol% K2CO3, 0.84 mmol internal standard (dodecane), 30 bar H2, 30 °C, EtOH (4 mL, 0.2 M). b Determined by GC analysis, known absolute configuration in parentheses.
1	4a	2	36	60 (S)
2	4b	2	25	n.d.
3	4a	4	>99	57 (S)
4	4b	4	38	69 (S)
5	4b	8	91	68 (S)
6	4a	16	>99	55 (S)
7	4b	16	>99	68 (S)

---

This suggests that the increased steric bulk in proximity of the metal centre is the dominant cause of the observed trend. However, we noted an appreciable increase in enantioselectivity from 57 to 68% for the hydrogenation of acetophenone. Finally, the substrate scope was studied with various aromatic and heteroaromatic ketones (Table 8). The reaction time was set at 16 h in order to maximize the conversion for all substrates, as the enantioselectivity was found reasonably stable in time. The introduction of a Cl substituent in ortho to the keto group (entries 3 and 4) did not alter the conversion under these reaction conditions, but the enantioselectivity dropped by 10% with both ligands. When the same substituent was positioned in para to the keto group (entries 5 and 6), the ee was identical to that of acetophenone, which denotes a negative steric effect of the chlorine atom, but no electronic effect. In contrast, the introduction of an electron-withdrawing CF₃ group in place of the methyl group of the ketone led to a slight decrease in activity for 4a but a severe drop in ees for both ligands (entries 7 and 8). Interestingly, bulkier substrates still gave complete conversions (entries 10–12), with up to 69% ee for pivalophenone with 4b (entry 12). By comparison, Clarke's system I with both planar and central chirality gave a 85% ee with pivalophenone, although under different reaction conditions.²⁹ Finally, heterocyclic substrates are tolerated but do not furnish the product in high ee (47% and 64% respectively, entries 13 and 14). From these results it can be concluded that the absence of central chirality on the ferrocenyl P,N,N ligands 4 translates most of the time into a decrease in enantioselectivity, with the notable exception of 4-chloroacetophenone, for which our catalytic system gave 56% ee against 27% with Clarke's.²⁹

Table 8 Substrate scope with 4a and 4b^a

Entry	Substrate	Cat.	Conv.^b (%)	ee^b (%)
a Typical reaction conditions: 0.85 mmol ketone, 1 mol% (S)-4a-b, 1 mol% [MnBr(CO)5], 10 mol% K2CO3 30 bar H2, 30 °C, EtOH (4 mL, 0.2 M), 16 h. b Determined by GC analysis using dodecane (0.84 mmol) as internal standard for conversion. c Determined by ^1H NMR in CDCl3 using trichloroethylene (1.81 mmol) as I.S. d Reaction time: 24 h.
1		4a	>99	55 (S)
2		4b	>99	68 (S)
3		4a	>99^c	47 (S)
4		4b	>99^c	55 (S)
5		4a	>99^c	56 (S)
6		4b	>99^c	68 (S)
7		4a	85^c^d	13 (S)
8		4b	>99^c^d	12 (S)
9		4a	>99^c	32 (S)
10		4b	>99^c	58 (S)
11		4a	>99^c	65 (S)
12		4b	>99^c	69 (S)
13		4a	>99^c	47 (S)
14		4b	>99^c	64 (S)

---

Computational investigation

In order to rationalise and support our results, a computational study was performed on both catalytic systems. In a recent contribution,³¹ Clarke et al. have computationally explored the hydrogenation of several prochiral ketones with one of their catalysts (III, Fig. 3) using DFT, pointing to a standard bifunctional (outer-sphere) mechanism and finding excellent agreement between the predicted and observed enantioselectivities for a few different substrates. Therefore, we restricted our investigation to this mechanism, using our catalyst 5a and the analogous Clarke's catalyst I (Fig. 3) for the hydrogenation of acetophenone, though we opted to use a different computational level (see details in the Experimental section). The results are shown in Fig. 6. The pathway is essentially identical to that described by Clarke et al., although certain interpretations required slight revision (vide infra). All optimized geometries can be viewed in the SI. The cycle starts with the 16-electron [(P,N^−H,N)Mn(CO)₂] complex (A), which is obtained from the pre-catalyst (either 5a or I) by deprotonation of the central NH donor of the P,N,N ligand and CO dissociation. The first step is H₂ activation, which is assisted by one ethanol molecule as proton shuttle. Thus, EtOH addition to A yields the H-bonded adduct B, followed by coordination of H₂ to the open coordination site to produce C, which is a standard dihydrogen complex of Mn^I rather than a dihydride complex of Mn^III. The Gibbs energy of this intermediate is quite high, because of the absence of enthalpic stabilization by the Mn–(H₂) interaction (the association is in fact endothermic with ΔH = +2.7 and +2.9 for systems 4 and I, respectively) and the entropic penalty. The H₂ ligand is then heterolytically cleaved via transition state TS_CD to yield the hydride complex D, where a proton is restored on the central N donor of the P,N,N ligand. Removal of EtOH from D yields the 18-electron hydride complex E, which is the lowest-energy species around the cycle (rate-determining intermediate). Subsequently, the ketone docks by H-bonding onto the NH function of the P,N,N ligand (F), followed by the hydride transfer from Mn to the ketone C atom viaTS_FG, which is the enantiodiscriminating step. Differently from the literature investigation,³¹ which reported direct (concerted) formation of the alcohol by an asynchronous H⁺/H⁻ transfer, the computational method utilized in the present investigation revealed a stepwise process, with the presence of a local minimum for the [(P,N,N)Mn(CO)₂]⁺(PhMeCHO⁻) ion pair resulting from hydride transfer (G). This is an 18-electron σ-complex of formally cationic Mn^I with the anionic alkoxide, which is held in the coordination sphere through a 3c-2e interaction with the incipient C–H bond, while the O atom maintains an H-bond with the central NH function of the neutral P,N,N ligand. Proton transfer then follows with a very small activation barrier viaTS_GH to generate the alcohol, which remains H-bonded to the deprotonated central N atom of the anionic P,N^−H,N ligand (H). For both systems 5a and I, hydrogenation of the re face, leading to the S central chirality of the phenylethanol product, is preferred presumably thanks to a π-stacking interaction between the substrate phenyl ring and the pyridine ring of the P,N,N ligand. The final step is alcohol removal to regenerate the initial complex A. Note that the H-bond formed between A and the S-phenylethanol product (to generate S-H) is stronger than that formed by the ethanol solvent (to generate B). The concentration difference, on the other hand, plays in favour of B. Thus, establishing the Gibbs energy of the rate-determining intermediate (RDI) and comparing the Gibbs energy span for the two catalysts is not clearcut. In addition, as also pointed out in the previous investigation,³¹ additional alcohol molecules may participate in the stabilization of the RDI. Nevertheless, based on the assumption that S-H is the cycle resting state, the calculated energy Gibbs span (from S-H to TS_CD) is smaller for catalyst 5a, in agreement with its observed greater activity. Concerning the enantioselectivity, the enantiodiscriminating step for both catalysts is the hydride transfer (TS_FG), thus the enantioselectivity is linked to [G(TS_FG-si) − G(TS_FG-re)] (ΔΔG^‡). According to the results in Fig. 6, these values (1.6 kcal mol⁻¹ for 5a and 1.1 kcal mol⁻¹ for I) lead to the prediction of a greater enantioselectivity for 5a (88%) than for I (72%), which is against the experimental evidence. Thus, the computational method adopted in the current investigation (see Experimental section) does not appear entirely adequate. Since the previous investigation of the similar catalyst III,³¹ which was conducted at a different level of theory, could satisfactorily reproduce the experimentally observed ee for a number of substrates, the same computational method has also been applied to reoptimize the enantiodiscriminating transition states. The new results are in better agreement with the experimental evidence, since the new ΔΔG^‡ values are 0.65 kcal mol⁻¹ (ee = 50%) for 5a and 0.74 kcal mol⁻¹ (ee = 55%) for I.

Fig. 6 Gibbs energy profile for the catalytic cycle of acetophenone hydrogenation with 5a and I. The (S) enantiomer of the chiral P,N,N ligand 4a was used in the calculations.

Conclusions

New ferrocenyl-based tridentate P,N,N ligands 4a and 4b, bearing planar chirality only, were prepared and evaluated in the manganese-catalyzed hydrogenation of aromatic ketones. The characterization of the manganese(I) complexes 5a and 5b proved complex, as they revealed to be very dependent on experimental conditions. On these grounds, with the help of DFT calculations, the pre-catalysts are likely mixtures of the fac-cationic tricarbonyl complex and of a neutral dicarbonyl complex. Our systems proved more active than Clarke's under very mild conditions, with complete conversion of acetophenone in 4 h at 50 °C and 30 bar of H₂, in the presence of only 0.1 mol% of 5a and the mild K₂CO₃ base. Even though complexes 5a-b allowed reaching reasonable ees, the absence of central chirality turned out to be detrimental for enantioselectivity. This behaviour was supported by calculations. The presence of the NH functionality also proved crucial for activity, but the effect of a substituent on the pyridine moiety in 5b was not notable in terms of ees. The evaluation of our systems in other catalytic reactions is currently under investigation.

Author contributions

The manuscript was written through the contributions of all authors. All authors have given approval to the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: experimental details, structural and computational details, NMR and IR spectra of new compounds. See DOI: https://doi.org/10.1039/d5nj03919c.

CCDC 2490601 and 2490602 contain the supplementary crystallographic data for this paper.^52a,b

Acknowledgements

The authors acknowledge the Ministère de la Recherche et de l’Enseignement Supérieur (PhD grant to R. Y.) and the Indo-French Centre for the Promotion of Advanced Research (IFCPAR/CEFIPRA) (PhD grant to U. B.). The technical assistance provided by Dr Yannick Coppel and Dr Christian Bijani for recording the solid and liquid NMR spectra, and that provided by N. Martins-Froment of the ICT-FR 2599 (Toulouse, France – ict.ups-tlse.fr) for recording the mass spectra, are gratefully acknowledged. We are also grateful to the CALMIP mesocentre of the University of Toulouse for the allocation of computational resources.

Notes and references

1. R. U. McVicker and N. M. O’Boyle, J. Med. Chem., 2024, 67, 2305–2320.
2. Y. Wang, M. Wang, Y. Li and Q. Liu, Chem, 2021, 7, 1180–1223.
3. M. Garbe, K. Junge and M. Beller, Eur. J. Org. Chem., 2017, 4344–4362.
4. G. A. Filonenko, R. van Putten, E. J. M. Hensen and E. A. Pidko, Chem. Soc. Rev., 2018, 47, 1459–1483.
5. T. Zell and R. Langer, ChemCatChem, 2018, 10, 1930–1940.
6. F. Kallmeier and R. Kempe, Angew. Chem., Int. Ed., 2018, 57, 46–60.
7. N. Gorgas and K. Kirchner, Acc. Chem. Res., 2018, 51, 1558–1569.
8. K. Azouzi, D. A. Valyaev, S. Bastin and J.-B. Sortais, Curr. Opin. Green Sustainable Chem., 2021, 31, 100511.
9. E. S. Gulyaeva, E. S. Osipova, R. Buhaibeh, Y. Canac, J.-B. Sortais and D. A. Valyaev, Coord. Chem. Rev., 2022, 458, 214421.
10. A. Torres-Calis and J. J. García, ACS Omega, 2022, 7, 37008–37038.
11. N. Ahmed, J. Organomet. Chem., 2024, 1009, 123071.
12. S. Mourgues, D. Serra, F. Lamy, S. Vincendeau, J.-C. Daran, E. Manoury and M. Gouygou, Eur. J. Inorg. Chem., 2003, 2820–2826.
13. J. Guadalupe Lopez Cortes, O. Ramon, S. Vincendeau, D. Serra, F. Lamy, J.-C. Daran, E. Manoury and M. Gouygou, Eur. J. Inorg. Chem., 2006, 5148–5157.
14. N. Mateus, L. Routaboul, J.-C. Daran and E. Manoury, J. Organomet. Chem., 2006, 691, 2297–2310.
15. A. Labande, J.-C. Daran, E. Manoury and R. Poli, Eur. J. Inorg. Chem., 2007, 1205–1209.
16. S. Bayda, A. Cassen, J.-C. Daran, C. Audin, R. Poli, E. Manoury and E. Deydier, J. Organomet. Chem., 2014, 772–773, 258–264.
17. R. Malacea, J.-C. Daran, R. Poli and E. Manoury, Tetrahedron: Asymmetry, 2013, 24, 612–620.
18. L. Routaboul, S. Vincendeau, J.-C. Daran and E. Manoury, Tetrahedron: Asymmetry, 2005, 16, 2685–2690.
19. L. Routaboul, S. Vincendeau, C.-O. Turrin, A.-M. Caminade, J.-P. Majoral, J.-C. Daran and E. Manoury, J. Organomet. Chem., 2007, 692, 1064–1073.
20. E. Le Roux, R. Malacea, E. Manoury, R. Poli, L. Gonsalvi and M. Peruzzini, Adv. Synth. Catal., 2007, 349, 309–313.
21. J. M. Hayes, E. Deydier, G. Ujaque, A. Lledós, R. Malacea-Kabbara, E. Manoury, S. Vincendeau and R. Poli, ACS Catal., 2015, 5, 4368–4376.
22. M. Biosca, M. Coll, F. Lagarde, E. Brémond, L. Routaboul, E. Manoury, O. Pàmies, R. Poli and M. Diéguez, Tetrahedron, 2016, 72, 2623–2631.
23. S. Elangovan, C. Topf, S. Fischer, H. Jiao, A. Spannenberg, W. Baumann, R. Ludwig, K. Junge and M. Beller, J. Am. Chem. Soc., 2016, 138, 8809–8814.
24. F. Kallmeier, T. Irrgang, T. Dietel and R. Kempe, Angew. Chem., Int. Ed., 2016, 55, 11806–11809.
25. N. A. Espinosa-Jalapa, A. Nerush, L. J. W. Shimon, G. Leitus, L. Avram, Y. Ben-David and D. Milstein, Chem. – Eur. J., 2017, 23, 5934–5938.
26. G.-W. Wang, M.-W. Chen, Y.-Q. Wu, Q.-X. Xie, Z. Liu, R.-Z. Liao and Y.-G. Zhou, ACS Catal., 2025, 15, 3418–3427.
27. A. Zirakzadeh, S. R. M. M. de Aguiar, B. Stöger, M. Widhalm and K. Kirchner, ChemCatChem, 2017, 9, 1744–1748.
28. M. Garbe, K. Junge, S. Walker, Z. Wei, H. Jiao, A. Spannenberg, S. Bachmann, M. Scalone and M. Beller, Angew. Chem., Int. Ed., 2017, 56, 11237–11241.
29. M. B. Widegren, G. J. Harkness, A. M. Z. Slawin, D. B. Cordes and M. L. Clarke, Angew. Chem., Int. Ed., 2017, 56, 5825–5828.
30. M. B. Widegren and M. L. Clarke, Catal. Sci. Technol., 2019, 9, 6047–6058.
31. C. L. Oates, A. S. Goodfellow, M. Buehl and M. L. Clarke, Angew. Chem., Int. Ed., 2023, 62, e202212479.
32. F. Ling, H. Hou, J. Chen, S. Nian, X. Yi, Z. Wang, D. Song and W. Zhong, Org. Lett., 2019, 21, 3937–3941.
33. L. Zhang, Y. Tang, Z. Han and K. Ding, Angew. Chem., Int. Ed., 2019, 58, 4973–4977.
34. L. Zhang, Z. Wang, Z. Han and K. Ding, Angew. Chem., Int. Ed., 2020, 59, 15565–15569.
35. H. Jayaprakash, P. Coburger, M. Wörle, A. Togni and H. Grützmacher, Chem. – Eur. J., 2022, 28, e202201522.
36. Y. Zhao, L. Zhang, Y. Tang, M. Pu and M. Lei, Phys. Chem. Chem. Phys., 2022, 24, 13365–13375.
37. M. Wang, S. Liu, H. Liu, Y. Wang, Y. Lan and Q. Liu, Nature, 2024, 631, 556–562.
38. J. Yang, L. Yao, Z. Wang, Z. Zuo, S. Liu, P. Gao, M. Han, Q. Liu, G. A. Solan and W.-H. Sun, J. Catal., 2023, 418, 40–50.
39. F. Ling, J. Chen, S. Nian, H. Hou, X. Yi, F. Wu, M. Xu and W. Zhong, Synlett, 2020, 285–289.
40. J. R. Khusnutdinova and D. Milstein, Angew. Chem., Int. Ed., 2015, 54, 12236–12273.
41. A. M. Krieger, P. Kuliaev, F. Q. Armstrong Hall, D. Sun and E. A. Pidko, J. Phys. Chem. C, 2020, 124, 26990–26998.
42. R. van Putten, E. A. Uslamin, M. Garbe, C. Liu, A. Gonzalez-de-Castro, M. Lutz, K. Junge, E. J. M. Hensen, M. Beller, L. Lefort and E. A. Pidko, Angew. Chem., Int. Ed., 2017, 56, 7531–7534.
43. M. B. Widegren and M. L. Clarke, Org. Lett., 2018, 20, 2654–2658.
44. C. L. Oates, A. S. Goodfellow, M. Bühl and M. L. Clarke, Green Chem., 2023, 25, 3864–3868.
45. T. Hayashi, K. Yamamoto and M. Kumada, Tetrahedron Lett., 1974, 15, 4405–4408.
46. A. Kumar, N. A. Espinosa-Jalapa, G. Leitus, Y. Diskin-Posner, L. Avram and D. Milstein, Angew. Chem., Int. Ed., 2017, 56, 14992–14996.
47. K. Z. Demmans, M. E. Olson and R. H. Morris, Organometallics, 2018, 37, 4608–4618.
48. Z. Császár, R. Kovács, M. Fonyó, J. Simon, A. Bényei, G. Lendvay, J. Bakos and G. Farkas, Mol. Catal., 2022, 529, 112531.
49. C. Liu, M. Wang, S. Liu, Y. Wang, Y. Peng, Y. Lan and Q. Liu, Angew. Chem., Int. Ed., 2021, 60, 5108–5113.
50. U. Bandyopadhyay, New chiral, tridentate, phosphine group containing ferrocenyl ligands for asymmetric catalysis with non-noble metals, Université Toulouse 3, Toulouse, 2021.
51. Other blank tests were done in iPrOH with no base, no manganese or no ligand, and no conversion was observed in all cases.
52. (a) CCDC 2490601: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2plnzy; (b) CCDC 2490602: Experimental Crystal Structure Determination, 2025 DOI:10.5517/ccdc.csd.cc2plp00.

---