Key words
manganese - reduction - homogeneous catalysis - hydrogenation - hydrosilylation -
hydroboration - transfer hydrogenation
1
Introduction
Manganese is the 12th most abundant element in the earth’s crust[1] and is currently (2020) produced on an 18.5 million metric ton scale per year.[2] Manganese has many applications, the most prominent being in the steel- and iron-making
industries.[3] A unique property is that it can be found in a large range of oxidation states (–3
to +7), allowing it to play essential roles in assisting chemical synthesis. Manganese
is found in many enzymes vital to sustaining life on earth; most notable is its role
in the core of oxygen-evolving complexes, which assist in splitting water to produce
oxygen in plants.[4] In the lab, manganese-catalyzed oxidation reactions,[5] radical cyclization reactions,[6] and C–H activation reactions,[7] to name a few, demonstrate the chemical potential of this group 7 transition metal.
With a growing interest in creating more sustainable chemistry in the future, both
catalysis and manganese have been placed under the limelight. Recently, manganese’s
role in homogenous catalysis has been increasingly discussed in reviews[8] as well as book chapters.[9]
The reduction of organic molecules is an important chemical tool in synthesizing novel
high-value chemicals as well as bulk precursor molecules. The latter, in particular,
has seen increasing attention due to the potential of green hydrogen to recycle carbon
dioxide in the building of a methanol-based economy.[10] There are many known methods to add hydrogen across unsaturated π-bonds. Herein,
we will focus exclusively on the homogeneous manganese catalysts able to perform four
important reduction reactions: hydrosilylation, hydroboration, hydrogenation, and
transfer hydrogenation.
Hydrosilylation and hydroboration reactions make use of silanes or boranes, which
themselves are weak hydride sources. In the presence of a catalyst, the activation
of these readily available precursors affords the silylated or boronated species under
mild reaction conditions. In the case of C–C π-bond reduction, these functionalized
products are highly useful precursors for cross-coupling reactions, predominantly
Suzuki[11] and Hiyama cross-couplings. When used to reduce a polar C–X π-bond (X = nitrogen,
oxygen), the reduced functionalized products can be readily hydrolyzed to produce
the unfunctionalized reduced organic compound. Hydrogenation uses hydrogen gas as
a reductant. With complete atom economy (in principle), hydrogenation is the most
efficient reduction method. Transfer hydrogenation adds two hydrogen atoms across
an unsaturated bond, given by a sacrificial organic source (commonly
i
PrOH). The major advantages of this reduction method are the availability of the hydrogen
source, which often also serves as the solvent, and the mild reaction conditions,
which are often low temperature and do not require specific pressure-safe equipment;
this is particularly advantageous for small scale, laboratory systems.
Several significant breakthroughs have occurred in recent times, which have unleashed
manganese’s potential in reducing organic substrates. Particularly, the synthesis
of low-valent Mn(I) and Mn(II) pincer complexes and the utilization of metal–ligand
cooperation (MLC)[12] in manganese complexes have allowed the activation of reductants and substrates
previously not possible. From a few initial examples, half a decade ago, the list
of reduction transformations is now significant (vide infra).[13]
The growth of manganese pincer complexes as a complete topic was first reviewed by
Beller in 2017[14] and then by Milstein in 2018.[15] Since then, several reviews have included hydrogenation of manganese complexes in
comparison to other base metals.[16] Reviews that focus primarily on manganese include an early (2017) review on manganese
(de)hydrogenation reactions,[17] a review on hydrosilylation from Wang et al. in 2018,[18] a book chapter on MLC in Mn(I) hydrogenation,[12b] and, more current, in-depth reviews on transfer hydrogenation from Bastin and Sortais[19] and hydrogenation from Liu.[20] So far, no concise overview of all the latest developments of manganese-catalyzed
reduction reactions across different major reducing platforms exists.
This review has three main objectives: (a) to provide a concise overview of the literature
in the four reduction methods mentioned above, (b) to show the potential of manganese
as a catalyst for future reduction use, and (c) to offer insight into the current
major catalytic design constraints and requirements for the development of novel catalytic
systems and complexes. Each reduction method is broken down into three subsections:
a short introduction of early examples, an overview of the known manganese catalysts
able to perform the broad transformation, and a concise summary of the substrates
that have been shown to be reduced using that method.
2
Hydrosilylation
Manganese-catalyzed hydrosilylation reactions are the oldest known reduction reactions
covered in this review. Back in 1983, Faltynek et al. showed that a silylpentacarbonylmanganese(I)
species could reduce terminal alkenes via thermal or photochemical activation to their
corresponding alkylsiloxane with heptamethylcyclotetrasiloxane (HMCTS).[21] While the photochemical reaction cleanly produced the desired product, the thermally
activated system produced a mixture of products unselectively. Furthermore, it was
shown that manganese itself was not the active catalytic component; instead, the Mn–Si
bond was homolytically cleaved to afford the active silyl radical. The group of Hilal
(1987) used decacarbonyldimanganese(0) in the presence of silanes to produce an active
Mn(I) hydride, to reduce hexene under much milder conditions (40 °C vs. 180 °C).[22] Almost a decade later, Cutler used a Mn(I) acetyl complex to hydrosilylate esters
to ethers via a deoxygenative reduction and, a year later, reduce ketones to secondary
alcohols.[23] Particularly these early examples from the group of Cutler showed the promise of
low-valent manganese catalysts with loadings of 2.4 mol% enabling the reduction of
acetophenone in less than 4 minutes.
2.1
Catalyst Overview
A large spread of oxidation states has been used as pre-catalysts for manganese-catalyzed
hydrosilylation reactions (notable examples summarized in Figure [1]). As expected, many of these catalysts are active through a variety of different
mechanisms and activation pathways. Both photochemical and thermal activation have
been used depending on the catalyst, the silane, and the substrate. The divergent
approaches to Mn(III) hydrosilylation are great examples of the many possible catalytic
cycles. The Mn(V) pre-catalyst 6 reported by Du et al. is reduced to form a supposed Mn(III) active catalyst.[24] In contrast, Lugan, Darcel, and Sortais suggested that their Mn(I) pre-catalyst
3 requires oxidative addition of the silane to give a Mn(III) intermediate before photochemical
CO dissociation affords the active catalyst.[25] Alternatively, Mn(III) pre-catalyst 5 was used as early as 2000 by Magnus and co-workers to reduce α,β-unsaturated ketones.[26] Mn(0) pre-catalysts have also been used for a range of hydrosilylation reactions.
Sortais and Darcel used UV light to activate 1 for the reduction of both esters and acids to their respective aldehydes.[27] In both cases, no activity was observed with thermal activation (even at 100 °C),
in agreement with reports by Fuchikami for ester reduction.[28] In contrast, it has been suggested in multiple early reports that the dimanganese
carbonyl complex 1 can form the known Mn(I) reducing species HMn(CO)5 upon reaction with a tertiary silane at high temperature.[29] The group of Fuchikami used Mn2(CO)10 at 100 °C in the presence of diethylamine for amide reduction and, recently, the
Mn(I) hydride was also proposed as the active species in the hydrosilylation of alkenes
presented by Xie and co-workers.[30] Whether the photochemical hydrosilylation using decacarbonyldimanganese(0) also
has the same activation to a Mn(I) hydride is, as yet, unknown. Contrary to examples
presented so far, other catalysts do not exhibit oxidation or reduction of the metal
center in the activation step or throughout the catalytic cycle. Mechanistic studies
from the group of Trovitch show that their redox non-innocent Mn(II) complex 4 was able to insert into the Si–H σ-bond without a change in oxidation state.[31] In addition, the group of Royo studied their Mn(I) hydrosilylation reaction mixture
by 55Mn NMR spectroscopy and identified three active manganese species with a suggested
oxidation state of +1 (vide infra).[32] Likewise, Werlé et al. suggested no change from the Mn(I) oxidation state and an
η2 Si–H manganese coordination as the most likely activation pathway of phenylsilane
by MnBr(CO)5, instead of oxidative addition.[33]
Figure 1 Examples of manganese-based hydrosilylation pre-catalysts in various oxidation states
2.2
Development of Manganese-Catalyzed Hydrosilylation Reactions
In 2000, Magnus continued the work of Hilal and Faltynek on the hydrosilylation of
olefin bonds using manganese.[26] By using Mn(dpm)3 (i.e., catalyst 5, dpm = dipivaloylmethanato, Figure [1]) in an isopropanol solvent and with phenylsilane as reductant, α,β-unsaturated ketones
could be reduced to saturated unfunctionalized ketones. In 2016, Shenvi and co-workers
clarified the reaction by showing that a highly active reducing agent, isopropoxy(phenyl)silane,
was formed in situ.[34] The group also showed that other non-conjugated olefins could also be reduced. In
2017, Thomas showed that an EtBIPMnBr2 catalyst (7, BIP = bis(imino)pyridine), similarly to equivalent iron and cobalt complexes, could
be used to hydrosilylate and hydroborate terminal alkenes when in the presence of
the highly important co-catalyst NaO
t
Bu (Figure [2]).[35] In 2018, the group published a separate hydrosilylation study using NaO
t
Bu and complex 7 with a different Ar group.[36] A fast and selective reduction of alkenes to their linear organosilanes was presented:
25 °C and 4 hours sufficed for high yields. Trovitch followed this discovery with
complex 8, which could perform the same transformation.[37] Wang and Yang, on the other hand, used a commercially available Mn(I) compound,
MnBr(CO)5 (2), to also reduce aliphatic alkenes to linear organosilanes.[38] Perhaps surprisingly, without the use of specialized ligands, Wang could perform
the reduction under only slightly more reactive conditions, 60 °C and 4 hours. Later,
they showed that alkynes could be reduced to vinylsilanes. Three different systems
have been developed: firstly, E-vinylsilanes were observed in good selectivity (40:1) using 2 and AsPh3, as a ligand, at 150 °C;[39]
Z-vinylsilanes were formed when using Mn2(CO)10 in the presence of dilauroyl peroxide as an additive;[39] and finally, even higher selectivity towards the Z-vinylsilane was detected by the group of Zhang when using Mn2(CO)10 in the presence of blue light (Figure [2]B).[40] Recently, Xie et al. used the same Mn2(CO)10 pre-catalyst in the presence of JackiePhos (9) to give linear organosilanes from alkenes.[30]
Figure 2 C−C π-bond reduction using silanes and a manganese catalyst: A. Mn(II) and Mn(I)
complexes; B. Ligands/additives used in addition to the Mn(0) pre-catalyst Mn2(CO)10
Carbonyl hydrosilylation using a manganese catalyst, initiated by Cutler, was continued
by Chung and co-workers in 2000.[41] The first arene-containing cationic Mn(I) carbonyl complex, (arene)Mn(CO)3
+, was synthesized to reduce several ketones at room temperature using dimethylphenylsilane.
Particularly, non-bulky, electron-rich ketones gave high yields in short reaction
times. Ring slippage (η6 → η4) of the naphthalene coordinated to the metal was attributed to the unique ability
of the catalyst. Further developments in Mn(I) catalysts for the hydrosilylation of
carbonyl groups were presented by a collaboration between the group of Lugan and Sortais.[25a] The reduction of a range of aldehydes and ketones to alcohols gave high yields in
an average of 1–4 hours at room temperature in the presence of 250 nm light and catalyst
3 (Figure [1]). Mechanistic experiments suggest that the photochemical dissociation of the final
carbonyl ligand is most likely only possible from a Mn(III) intermediate, following
oxidative addition into the Si–H σ-bond.[25b] With the majority of carbonyl systems focusing on low-valent Mn(I), Du et al. used
a high oxidation state Mn(V) salen complex 6 (Figure [1]) to reduce ketones and aldehydes in high yields and at low catalyst loading (0.5
mol%).[24] Mechanistic experiments, including a Hammett plot, suggest the reduction of the
d2 manganese pre-catalyst to a Mn(III) hydride. This may also account for the short
induction period and significantly longer reaction times required at lower temperatures.
In 2014, a significant breakthrough in manganese-catalyzed hydrosilylation was published
by the group of Trovitch.[31] Complex 4 (Figure [3]) is formally neutral bearing the redox-active PDI (pyridinediimine) dianion ligand
with a low-valent Mn(II) center. The catalyst exhibited a TOF (turnover frequency)
of 76,800 h–1 and high performance at catalyst loadings as low as 0.01 mol% for the reduction of
ketones. Ester hydrosilylation was also possible for small, less sterically hindered
esters; extremely long reaction times of 10 days were required with increased steric
bulk. Developments published later (2017), showed that aldehydes were reduced at an
even faster rate (TOF = 4900 min–1) than ketones and that catalyst 4 could perform the formate dihydrosilylation with a TOF of 330 min–1.[42] Catalyst 4 is currently available to buy from many commercial chemical vendors. Mechanistic
studies suggest a modified Ojima mechanistic cycle where the redox-active ligand assistance
maintains a Mn(II) metal center throughout the reaction. Interestingly, an alternative
redox-innocent pentacoordinate complex 10, also synthesized by the group, gave TOF values of 2475 min–1 for benzaldehyde hydrosilylation, only half that of complex 4 and a very competitive rate in comparison to other metal catalysts.[43] The group also synthesized Mn(I) dimer 12 and Mn(III) hydride 11 supporting the same ligand backbone. The Mn(I) dimer showed a similar TOF for benzaldehyde
reduction in comparison to complex 4, but the TOF for formate dihydrosilylation was lower on a per manganese basis.[44] The Mn(III) hydride catalyst 11 showed poor performance in the hydrosilylation of carbonyls; however, it matched
the performance of the leading complex 4 in the reduction of formates and showed even more outstanding performance in the
reduction of ethyl acetate.[42] KIE (kinetic isotope effect) differences between complexes 4 and 11 indicate that the hydrosilylation reactions go through competing mechanisms. In 2017,
Huang and co-workers reported a similar, but chiral tridentate ligand coordinated
to Mn(II) (13) to reduce ketones to secondary alcohols in high enantiomeric excess.[45] Catalyst activation using NaBEt3H (2 mol%) was required. More recently, the group of Royo published a Mn(I) complex
with a bis-NHC ligand (15) (see Scheme [1]A), which was effective in the hydrosilylation of ketones.[32] Though more promising yields were obtained with phenylsilane, the readily available
silane polymethylhydrosiloxane (PMHS) could also afford alcohols in moderate to good
yields. Further examples of Mn(II) hydrosilylation of aldehydes and ketones were presented
by the groups of Madrahimov[46] and Sunada.[47] The latter managed to utilize TMDS (tetramethyldisiloxane) as the silane source.
Figure 3 Developments on Mn(II) complexes initiated by Trovitch in 2014
Apart from the examples of Cutler and Trovitch, not many examples of ester hydrosilylation
were published in the literature until recently. Werlé and Leitner used their Mn(I)
triazole catalyst 16 (Scheme [1]) to reduce esters to alcohols as the major product instead of ethers like the Cutler
group; however, small amounts of ether (10–20%) could also be detected.[48] High selectivity for the alcohol product was obtained by using Mn(I) complexes 15 (Royo) and 14 (Bagh), reported in 2020.[49] The group of Sortais formed silyl acetals, in contrast, by using the readily available
Mn2(CO)10 complex under UV light conditions (LED 265 nm) (Scheme [1]).[27b] Aldehydes were isolated in high yields following acidic hydrolysis. Sydora, Stradiotto,
and Turculet used their catalyst 17 to reduce a methyl benzoate derivative and methyl heptanoate to their respective
alcohols in high yield, although no other examples were given. Instead, the group
focused on the, also challenging, hydrosilylation of amides.[50] With catalyst 17 (2 mol%), a range of tertiary amides were deoxygenatively reduced overnight at room
temperature. Only two other minor examples have shown the potential of manganese to
reduce amides via a deoxygenative hydrosilylation: work from Fuchikami, using Mn2(CO)10 with a diethylamine additive in thermally activated conditions,[51] and Pannell, using CpMn(CO)3 to reduce DMF and DEF within a mechanistic study.[52] Nitrile reduction using silanes and a manganese catalyst was also most recently
performed by the group of Kundu.[53]
Scheme 1 A. Recent manganese catalysts for more challenging carbonyl reductions using hydrosilanes;
B. Hydrosilylation of esters into their different possible products (selected examples
are used as conditions)
An important field of manganese hydrosilylation research has been the pursuit of a
catalyst that can reduce a range of carboxylic acids since no currently known homogenous
manganese catalyst exists for the transfer hydrogenation or hydrogenation of carboxylic
acids. The group of Sortais and Darcel used Mn2(CO)10 under photochemical conditions to reduce carboxylic acids to their respective aldehydes
after acidic hydrolysis (Scheme [1]B).[27a] During their silane screening, the alcohol product was also formed as the major
product when secondary silanes were used; impressively, this was even the case with
cheap and industrially applicable TMDS. However, the strengths or limitations of the
system for use in the synthesis of alcohols were not studied in this publication.
Though a large scope of aliphatic carboxylic acids was reduced to their respective
aldehydes in high yield, benzylic acids could not be reduced with greater than 30%
conversion. In 2019, Werlé and Leitner contributed to this field by reducing a small
number of benzylic and aliphatic carboxylic acids to alcohols (using catalyst 16, see Scheme [1]).[48] More recently (2021), Werlé and co-workers developed this work to show that the
commercially available manganese complex MnBr(CO)5 (2) could reduce a large scope of acids to alcohols within 4 hours for most aliphatic
acids and 2 hours for aromatic acids.[33] Separate experiments showed that as low as 0.5 mol% catalyst loading could be used
at the expense of time to generate equally high yields. A gram-scale reaction performed
using only 1.5 equiv. of PhSiH3 was conveniently isolated after a Kugelrohr distillation in 93% yield. Mechanistic
and gas phase analysis suggested the loss of 1 CO molecule with η2 Si–H manganese coordination, where Mn(I) remains the active oxidation state of the
metal throughout the reaction.
Another challenging hydrogenation transformation using a manganese catalysts is the
conversion of CO2 to MeOH. Gonsalvi and Kirchner showed that their (PNP)Mn(CO)2H catalyst 18 (Scheme [1]) could dramatically catalyze the single reduction of carbon dioxide to formate,
which requires 24 hours without catalyst, and the triple reduction to methanol, which
gave almost quantitative yields at 80 °C.[54] In 2021, Garcia et al. reported using MnBr(CO)5 with Et3SiH at mild CO2 pressures (<4 bar) to selectively obtain either the (silyl)formate or bis(silyl)acetal,
depending on solvent composition.[55]
3
Hydroboration
Manganese-catalyzed hydroboration reactions are not as common historically in comparison
to hydrosilylation reactions. In 2016, Zhang and Zheng reported the use of manganese
catalyst 19 for the chemoselective reduction of ketones over alkenes and the selective Markovnikov
hydroboration of styrene derivatives (Scheme [2]).[56] The group utilized a pincer Mn(II) complex supported by a terpyridine ligand. The
hydroboration of aliphatic alkenes gave, with much lower selectivity, the anti-Markovnikov
product. Since then, the rapid development of manganese-catalyzed hydroboration has
taken place.
Scheme 2 Hydroboration of terminal alkenes using Mn(II) pincer catalysts (7, Ar = 2,6-diisopropylbenzene)
Scheme 3 Simple schematic representation of the two ways manganese hydrides can catalyze a
hydroboration reaction (for the purposes of this simplification, a ketone substrate
is shown)
3.1
Catalyst Overview
The development of both Mn(I) and Mn(II) pincer complexes appears to have unleashed
the potential of manganese-catalyzed hydroboration. To date, mechanistic investigations
have not been able to differentiate a clear difference in mechanism between Mn(I)
and Mn(II) complexes. However, studies of intermediate manganese complexes by experimental
or computational experiments have shown, more often than not, the involvement of a
manganese hydride intermediate.
Du’s Mn(V) catalyst 6 (see Figure [1]) is thought to be reduced to form an active hydride species, a possible Mn(III)
hydride, in a similar fashion to the hydrosilylation reactions.[57] By adding two different boranes, one of which was deuterated, in the presence of
the manganese catalyst under standard reaction conditions, no scrambling was observed.
On the basis of these findings, it was believed that the manganese hydride acted as
a weak Lewis acid to activate HBpin rather than as a direct hydride transferring catalyst.
Gade et al. proposed the same Lewis acid activated pathway for their asymmetric Mn(II)
catalyst.[58] In contrast, the proposed mechanistic cycles for Mn(I) catalysts typically follow
a direct manganese hydride transfer, as seen in the examples by Maji[59] and Leitner.[60] The two different manganese hydride pathways are shown in generic form in Scheme
[3]. Also, when Trovitch and Baik performed computational calculations of their Mn(II)
catalyzed dihydroboration of nitriles, direct insertion of the substrate into the
Mn–H bond was in plausible agreement with experimental observations. The manganese
hydride could be re-formed via a σ-bond metathesis.[61] Leitner showed that a well-designed ligand could assist B−H activation and, therefore,
the regeneration of the manganese hydride.[60]
3.2
Development of Manganese-Catalyzed Hydroboration Reactions
Following the initial work by Zhang, the groups of Thomas (2018)[36] and de Ruiter (2021)[62] used their Mn(II) catalysts 7 and 20, respectively, to form the linear organoboronate compounds from both aromatic and
aliphatic alkenes. In 2020, the group of Rueping used a bis(imino)pyridine (BIP) ligand
– also used by Thomas (catalyst 7, Scheme [2]) – to synthesize a (BIP)MnCl2 catalyst for the single reduction of propargylic alcohols and amines as well as internal
alkynes.[63] This afforded highly valuable, functionalized alkenes, with a possible application
as substrates for cross-coupling reactions. Even more recently, the group of Kirchner
first used their Mn(I) catalyst 21 (Scheme [2]) with HBPin to also reduce terminal alkenes into the respective linear organoborates
at low catalyst loadings (0.25 mol%).[64] Interestingly, terminal alkynes could be reduced with unique selectivity, the trans-1,2-diborated product isolated as the major product (Scheme [4]). Although the highest observed selectivity was only 55% and products could only
be isolated up to 44% yield, the reaction is unprecedented in transition-metal catalysis
and affords a valuable building block in organic chemistry. Further experiments and
DFT calculations propose an acceptorless pathway with the release of dihydrogen within
the mechanism.[64]
Scheme 4 Reduction of terminal alkynes using Mn(I) catalyst 21 (fully drawn in Scheme [2])
Hydroboration of carbonyl compounds also affords chemically valuable products, often
in the form of alcohols, after silica-activated hydrolysis. By using specifically
designed pincer ligands, the groups of Zhang and Zheng,[56] Du,[57] Gonsalvi and Kirchner,[65] Maji,[59] and Leitner[60] could all form alcohols from different substrates using pinacolborane as the reduction
source (Figure [4]). Most examples showed high catalytic activity at low catalyst loadings (<1 mol%)
and low temperatures, with exceptions being the work of Leitner, which required elevated
temperatures (90–115 °C) for the reduction of substrates more challenging to reduce.
Trovitch and Baik synthesized diborylamines, useful for cross-coupling chemistry,
from nitriles via double hydroboration.[61] This reduction also needed higher temperatures of around 80 °C. The potential for
asymmetric hydroboration reductions using chiral pincer ligands was realized by Gade
et al.[58] Their catalyst 24 (2.5 mol%) reduced ketones to their respective secondary alcohols in high enantiomeric
excesses.
Scheme 5 Simple Mn(II) catalysts for the hydroboration of C−X π-bonds (‘B’ = BPin)
Figure 4 Hydroboration of polarized C–X π-bonds using pinacolborane and specifically designed
pincer catalysts; substrates reduced are written in red (8, Ar = 2,6-diisopropylbenzene)[61]
More recently, when the de Ruiter group investigated their manganese catalyst 20 (see Scheme [2]), they discovered that the simple manganese triflate catalyst Mn(OTf)2(CH3CN)2 (25; Scheme [5]) gave comparative activity for the reduction of esters, carbonates, and nitriles.[66] Jacobi von Wangelin et al. also showed that their simple Mn(hmds)2 catalyst (26; hmds = hexamethyldisilazane) reduces an extensive array of organic compounds (esters,
amides, pyridines, carbonates, polycarbonates, diimides, and carbon dioxide) via hydroboration
(Scheme [5]).[67] Also, in this case, a manganese hydride was detected: upon the reaction of the pre-catalyst
with pinacolborane, a cluster of MnH(hmds)6 was formed, which could stoichiometrically reduce benzaldehyde via direct hydride
transfer from the metal.
Alternative manganese-based systems have also recently been developed in contrast
to small monomeric molecular catalysts. Zheng and Zhang used a terpyridine-based Mn(II)
coordination polymer as a pre-catalyst for the reduction of a range of carbonyls as
well as styrene derivatives via both hydrosilylation and hydroboration.[68] Another example by Copéret and co-workers showed that manganese could be used with
surface organometallic chemistry (SOMC) without an organic ligand; aldehydes and ketones
could be reduced with comparative performance.[69]
4
Hydrogenation
The earliest examples of manganese-catalyzed hydrogenation reactions were published
in 2016, starting with the work of Beller and co-workers.[70] They used their Mn(I) catalyst 27 (Figure [5]) to reduce nitriles, ketones, and aldehydes under a H2 pressure atmosphere. Soon thereafter, this group added to this discovery by using
the slightly different PNP catalyst 28 to reduce esters.[71] Interestingly, catalyst 28 exhibits facial coordination, which appears to be the kinetic product, instead of
the more commonly found meridional coordination. By monitoring the reaction in comparison
to mer-28, it was hypothesized that the coordination of the pre-catalyst had no impact on the
catalytic activity and that a common active species would be present in the mechanistic
cycle. That same year, Kempe[72] and Milstein[73] used their own Mn(I) complexes 29 and 30, respectively, to reduce ketones and esters, respectively (summarized in Figure [5]). The rapid development in catalytic design and substrate reduction that followed
is outlined below.
Figure 5 First examples of manganese-catalyzed hydrogenation were published by the groups
of Beller, Kempe, and Milstein, all in 2016
4.1
Catalyst Overview
From all the possible oxidation states of manganese observed to date, only manganese(I)
species have been used as hydrogen transfer catalysts in hydrogenation reactions.
The next most important catalytic design feature allowing catalytic hydrogenation
to take place has been the ligand framework. MLC has been of utmost importance for
activating hydrogen, with multiple different systems developed (Figure [6]). ‘Type A’ catalysts can undergo deprotonation on the ligand backbone; this generates
an important change in the first coordination sphere of the metal, as the N-heteroaromatic
ring loses its aromaticity and allows the coordinating nitrogen atom to act as an
amide donor. ‘Types B–D’ catalysts all contain an M/N–H system in the complex, an
important feature shown to improve the activity of noble-metal complexes that are
commonly associated with the work of the Noyori group. Also, here, deprotonation/reprotonation
within the catalytic cycle is possible; however, it should be noted that, through
studies with the original Noyori catalysts, several proposed mechanisms of N–H involvement
exist, discussed elsewhere.[74] For the purpose of this concise review, we will simply refer to these systems as
MLC systems, as, with all mechanisms, the ligand and metal play integral roles in
the activation of substrates.[12a] These four catalytic systems have been the most common in the manganese hydrogenation
literature to date. The complexes commonly form a classical pincer coordination geometry;
however, facial coordination has also been observed in several active catalysts (e.g.,
complex 28, Figure [5]). ‘Type E’ catalysts have been used by the groups of Garcia and Kirchner for manganese-based
hydrogenation reactions; it shows an alternative system where the coordinating solvent
or substrate (symbolized by X in Figure [6]) assists the metal in the activation of hydrogen (vide infra). There are a couple
of noteworthy exceptions to these five systems: Beller et al.’s use of MnBr(CO)5 for the hydrogenation of quinolines,[75] Hu et al.’s biomimetic manganese catalyst 36 (Scheme [6]A; see below) for aldehyde, ketone, and imine hydrogenation,[76] and Khusnutdinova and Nervi’s use of hydroxy deprotonation on their MLC system 46 (Scheme [8]C; see below) for the single reduction of CO2.[77]
Figure 6 A breakdown of the basic ligand structures used for manganese-catalyzed hydrogenation
reactions
Figure 7 Overview of two important features of ligand design that affect the hydrogenation
of manganese-catalyzed reactions: A. Pidko group bidentate 31 and tridentate 32 complexes, which show different levels of thermal stability; B. Comparison between
PNP (‘type B’) and NNP (‘type D’) complexes for the hydrogenation of organic substrates
(majority of the comparison is summarized from Liu et al.[78])
Understanding the mechanistic effect of the ligand system has become an important
focus of research for the development of novel, highly active hydrogenation catalysts.
In a recent publication from the group of Filonenko and Pidko, the high-performance
(TOF = 41,000 h–1, TON = 200,000) and thermally stable (at 120 °C) Mn(I) catalyst 32 (Figure [7]A) was used for a range of carbonyl hydrogenation reactions.[79] It highlighted the important effect of the ‘third arm’, in this case the phosphine,
for stabilizing the tridentate NHC–NP complex 32, since the equivalent bidentate complex 31 (Figure [7]A), also by Pidko, underwent rapid thermal decomposition at >75 °C.[80]
The effect of changing the electronic or steric environment of the complex by changing
the substituents on similar PNP (‘type B’) and NNP (‘type D’) complexes was studied
by Lan and Liu.[78] They observed clear differences between the two types, with higher activity for
more electron-rich and less bulky NNP complexes compared to similar PNP derivatives
(Figure [7]B). This discovery is in line with the development made by Pidko and co-workers,
whose highly active complex 32 contained an electron-donating and planar NHC group. With this rationale, Lan and
Liu used complex 33, previously used by Beller for the hydrogenation of amides,[81] for the first hydrogenation of quinolines by a manganese catalyst. The catalyst
outperformed similar less electron-donating and bulkier PNP complexes as expected.
Unexpected are the catalytic activity differences between complexes 33 with 27, which have previously been reported in screening reactions by the groups of Beller[82] and Leitner.[83] In the one-pot reduction of carbon monoxide assisted by alcohols from the Leitner
group, complex 33 afford only minor amounts of formate and undetectable amounts of MeOH. Under the
same conditions, catalyst 27 produced almost 3 mmol of MeOH with >99% selectivity. When the group of Beller tested
the formamide reduction (for amine-assisted CO hydrogenation), low amounts of MeOH
and amine were detected with catalyst 33 in comparison to 27. Likewise, for ethylene carbonate reduction, catalyst 33 afforded similarly low yields of product in comparison to other PNP ‘type A’ catalysts,
whereas the PNP ‘type B’ catalyst 27 gave between 8–12 times higher TON for MeOH and ethylene glycol (Figure [7]B). The computational calculations performed by the group of Lan and Liu[78] suggest that the less electron-donating and bulkier PNP complexes favor rapid activation
of hydrogen to form the manganese hydride via a lower activation barrier and more
stable manganese hydride. In contrast, the NNP complexes have a lower activation barrier
for the hydride transfer from the metal to the carbonyl (in silico calculations used
an amide as the model carbonyl). The ability of the ligand to facilitate different
steps in the mechanism gives an increased insight into the challenges in predicting
improved novel catalysts for hydrogenation transformations.
4.2
Development of Manganese-Catalyzed Hydrogenation Reactions
Between the early developments of Beller and Kempe in 2016 and the most recent low
catalyst loading publication from Filonenko and Pidko (2021), there have been several
noteworthy additions to the hydrogenation of carbonyl groups literature. Following
the use of a tridentate ‘type B’ catalyst by Sortais and co-workers,[84] the same group developed a series of bidentate manganese complexes in 2018–2019
that only required 50–60 °C to reduce a range of ketones in high yields (Scheme [6]).[85] The most recent of the publications used a phosphonium ylide intermediate to activate
hydrogen via a non-classical MLC pathway.[86] The group of Kirchner performed highly selective hydrogenation of aldehydes; catalyst
18 reduced aldehydes, while ketones, esters, alkynes, nitriles, heteroaromatics, α,β-unsaturated
bonds, and alkenes were all not hydrogenated (Scheme [6]).[87] In 2019, Hu and co-workers used a completely different style of catalyst (36) based on a biomimetic model to reduce a range of carbonyls (Scheme [6]).[76] This example, however, did not exhibit any preferable activity in carbonyl reduction,
still requiring pressures of 50 bar (as in the work of Sortais). In 2021, the group
of Kirchner also performed the hydrogenation of ketones and α,β-unsaturated ketones
using their (dippe)MnnPr(CO)3 catalyst [21, dippe = 1,2-bis(diisopropylphosphino)ethane] under relatively mild conditions: 10
bar H2 and 60 °C.[88] A well-known migratory insertion reaction followed by hydrogenolysis of the acyl
intermediate was used to activate the manganese alkyl complex and form the active
manganese hydride (Scheme [6]B). The unusual ‘type E’ ligand framework (Figure [6]), without the possibility to perform MLC through the permanent ligand framework,
was able to activate dihydrogen with an inner-sphere substrate assistance (Scheme
[6]C).
Scheme 6 Manganese catalysts for carbonyl hydrogenation: A. Catalyst examples from the literature;
B. Activation of Kirchner’s hydrogenation catalyst 21; C. Suggested mechanistic cycle of the Mn(I)dippe (‘type E’, Figure [6]) catalyst used by Kirchner and Garcia
After original developments from Beller and Milstein in ester hydrogenation, the group
of Pidko formed a bidentate manganese complex that could also reduce esters.[89] The catalyst could reduce a range of esters in high yields at similar temperatures
to the previous publications; however, slightly elevated hydrogen pressures and significantly
higher base loading were required. The same group later contributed computational
insights into the reaction.[90] Clarke and co-workers also made an important contribution to the field with their
complex 37 (seen in Figure [8]); this catalyst could reduce esters at particularly low catalyst loadings and reduce
ketone asymmetrically to give secondary alcohols in high yield with enantiomeric excesses
up to 97%.[91] Further developments from the group also include using the racemic version of 37 to reduce enantioenriched α-chiral esters without loss of stereochemistry.[92] The groups of Beller,[93] Zhong,[94] Wen and Zhang,[95] Morris,[96] and Wang, Han and Ding[97] have all also added to the field of asymmetric ketone hydrogenation using enantiomerically
enriched chiral manganese complexes 38–42 (Figure [8]).
Figure 8 Manganese-based catalysts for the asymmetric hydrogenation of ketones (37, Ar = Ph;[91]
[98]
41, variations of Ar were used[97])
Nitrile hydrogenation has only been replicated, since Beller’s original 2016 publication
(with catalyst 27, Figure [5]), using diphosphine bidentate ligands on a Mn(I) center (‘type E’).[99] The best conditions for the reduction of a range of benzonitriles were obtained
by Garcia et al., using low pressure (6.9 bar) and <100 °C for 15 minutes.[99a] The (dippe)MnOTf(CO)3 catalyst, expectedly, follows a similar mechanistic cycle as the other ‘type E’ catalysts
(Scheme [6]C). Imine reduction, traditionally one of the more facile functional group reductions,
was performed by the group of Sortais using their simple bidentate catalyst 34, also used to reduce ketones (Scheme [6]A). Subsequently, the group deployed this finding in the reductive amination of aldehydes
and amines, using H2 as the reduction source: the reaction was performed in a stepwise process with the
crude imine hydrogenated to prevent the faster hydrogenation of aldehyde over the
condensation reaction.[100] The reduction of imines was later (2019) also performed by Kempe using a variation
on their triazine-based catalyst 29 (see Figure [5]).[101] As an extension to the work started in 2016 on ketone reduction, thorough mechanistic
insights into the role of the doubly deprotonated manganese hydride [Mn–H]K2 species and the imine hydrogenation mechanism was presented. In 2020, Mao and Wang
reported the use of the ligand- and base-free hydrogenation system with MnBr(CO)5,[102] developed earlier that year by Beller et al., for hydrogenation of imines and quinolines.
The group of Beller showed that at 15 bar H2 and at 45 °C, the presence of only MnBr(CO)5 was sufficient to hydrogenate quinolines in high yield (Scheme [7]).[75] Thorough mechanistic investigations led to the understanding that several Mn(II)
species were generated during the reaction via disproportionation; however, neither
of the major isolated Mn(II) species was involved in the hydrogen transfer. Instead,
the species were shown to act as Lewis acids in the activation of the quinoline with
the Mn(I) intermediate MnH(CO)5 identified as an active hydride transfer compound. MLC manganese complexes have also
been used to hydrogenate quinolines and N-heteroaromatic molecules: the Liu group’s
33 (vide supra, Figure [7]),[78] then the group of Rueping with a ‘type B’ catalyst,[103] and finally asymmetric quinoline hydrogenation was performed by Lan and Liu.[104] Despite the use of a base and a specialized ligand system, none of the three systems
had any significant improvements in temperature, H2 pressure, or time.
Scheme 7 Hydrogenation of quinolines using the commercially available MnBr(CO)5, presented by Beller et al.
In 2017, Beller and co-workers reported the use of complex 33 to reduce amides to amines and alcohols (Scheme [8]).[81] A year later, the group of Milstein reported the reduction of amides via a deoxygenative
hydrogenation (43, Scheme [8]).[105] In this case, a Lewis acid was required to allow the reaction to occur, most probably
due to its function in assisting dehydration. Following this work, three groups (Milstein,
Rueping, Leitner), in the space of two months, all independently, published reports
of carbonate and cyclic carbonate reduction, to afford a pair of alcohols, or a diol,
and methanol.[83a]
[106] Summarized in Scheme [8]B, the groups of Leitner (27) and Milstein (30) used lower hydrogen pressures, while the group of Rueping (44) used 1 mol% of catalyst to produce diols in more than 90% yields consistently. The
even more challenging carbamates and urea functional groups were hydrogenated by Milstein
and co-workers using a similar, but a little more electron-donating and less sterically
bulky, pincer backbone (45).[107] The same catalyst also performed best in the screening of catalysts performed by
their group for the hydrogenation of cyclic imides (published 2020).[108]
Scheme 8 Manganese catalysts for the hydrogenation of more challenging carbonyl groups: A.
Different conditions and catalysts used for the reduction of amides to give either
the deoxygenative product or a mixture of amines and alcohols; B. Conditions and catalysts
for the reduction of challenging carbonyl-containing functional groups; C. Manganese
catalysts for the hydrogenation of carbon dioxide
In 2017, Kirchner, Gonsalvi, and co-workers, published the single reduction of carbon
dioxide to formate by using manganese catalyst 18 (Scheme [8]C).[109] By using stoichiometric amounts of DBU, CO2 was reduced to the DBUH+HCOO– product by a ‘type A’ manganese catalyst (20 ppm) in the presence of LiOTf (10 mol%)
as Lewis acid. The formate species was isolated in 63% yield (TON = 31,600). Khusnutdinova
and Nervi went on to reduce carbon dioxide to formate using a similar Mn(I) catalyst/DBU
system.[77] Interestingly, an OH group on their catalyst 46, in contrast to the majority of NH group systems, was deprotonated in a bio-inspired
fashion to allow for MLC to occur (Scheme [8]C). In addition, CO2 activation by different MLC manganese catalysts was studied by the Milstein group,[110] and, more recently, by Kirchner and Gonsalvi, who used catalyst 21 to synthesize the formate–DBU aggregate from CO2, albeit in lower TON than their previous example.[111] The group of Prakash performed the first reduction of carbon dioxide to methanol
(TON = 36); a major limitation of the reaction was that it needed to be done in a
stepwise process.[112] More recently (2021), the group of Leitner made an important contribution to the
field with a single-step hydrogenation of carbon dioxide to methanol. In their work,
an isopropoxide metal salt was required to prevent the buildup of the resting state,
a formate–manganese complex. In the presence of the PNP (‘type B’) catalyst 47* and Ti(O
i
Pr)4, carbon dioxide could be reduced to methanol with an increased TON of 160 (Scheme
[8]C).[113] CO2 hydrogenation using a manganese catalyst has also been performed in silico to aid
catalyst and reaction environment (i.e., solvent) design.[114]
Carbon monoxide hydrogenation to methanol was reported in 2019 by the group of Beller
and Checinski.[82] They used an amine-based promoter to capture the carbon monoxide, which generated
a formamide that could then be reduced to methanol, re-forming the amine-based promotor.
An important development came two years later from the Leitner group, in which the
amine-based promotor was substituted with an alcohol.[83b] Using the same catalyst 27, in the presence of ethanol, an improved TON of 4023 was reached. The developments
from the group of Leitner on alcohol-assisted direct reduction to methanol (from CO2 and CO) lead to the possibility of a catalytic ‘breeding’ of methanol, since the
product, mediator, and, potentially, solvent can be identical. This was demonstrated
by the group[83b]
[113] and is summarized in Scheme [9].
Scheme 9 Direct hydrogenation to methanol via a catalytic ‘breeding’ system presented by the
group of Leitner
Kirchner, Rueping, and Beller have also separately contributed to the hydrogenation
of C–C double or triple bonds with Mn(I) catalysts (Scheme [10]). Kirchner et al. hydrogenated a range of alkenes with their MnnPr(CO)3dippe catalyst 21 at low temperature (25–60 °C) to afford the saturated compounds in high yields.[115] Rueping used Mn(I) catalyst 48 with an SNP pincer ligand to perform single hydrogenation of alkynes to afford Z-alkenes with high selectivity.[116] Likewise, the group of Beller obtained Z-alkenes – from alkynes – almost exclusively via a concerted outer-sphere mechanism,
using catalyst 27.[117] The mild conditions (30–60 °C) seemed critical for the selective single reduction,
since the Z-alkene products could only be reduced to the corresponding alkanes at 140 °C. The
base also played a pivotal role in the reaction, recycling the thermodynamically stable
and nonreactive off-cycle trans-hydride species.
Scheme 10 C−C π-bond hydrogenation using Mn(I) catalysts; reaction equation above shows reduction
of alkynes to Z-alkenes under generic conditions
5
Transfer Hydrogenation
With the explosion of MLC manganese chemistry in 2016, the first transfer hydrogenation
of ketones using a well-defined manganese catalyst was reported. The group of Beller
tested multiple Mn(I) pincer complexes for the transfer hydrogenation of acetophenone
using
i
PrOH as the hydrogen source.[118] The best performing catalyst, the phosphine-free N,N,N-pincer complex 49 (Figure [9]), could reduce a large range of ketones under mild conditions via a convenient laboratory
procedure that precludes the need for high pressure or more active and expensive silanes
or boranes.
Figure 9 Manganese catalysts for the transfer hydrogenation of ketones to secondary alcohols
5.1
Catalyst Overview
Similarly to hydrogenation reactions, Mn(I) metal centers and MLC appear to be crucial
components in the design of highly active catalysts. A common theme in almost all
the catalysts was a Noyori-type M–NH bond in the complex. The most significant differences
in catalytic design between transfer hydrogenation catalysts and hydrogenation catalysts
result from the, on average, lower temperature requirements for manganese-catalyzed
transfer hydrogenation reactions. This allows for more readily available, easier to
synthesize, and cheaper ligands. The added thermal stability from a tridentate ligand
adds no detectable benefit in the room temperature to 60 °C transfer hydrogenation
reactions. Sortais and co-workers explicitly explored this trait by reacting a range
of 28 diamines with MnBr(CO)5.[119] The in situ generation of a bidentate ligated complex allowed good yields (>65%)
to be observed with eight of the diamines. This included using a chiral diamine 55 (Figure [9]) to give 30–90% ee for the reduction of a range of benzylic ketones.[119]
5.2
Development of Manganese-Catalyzed Transfer Hydrogenation Reactions
Following the work of Beller in 2016, the groups of Sortais,[120] Leitner,[121] and Kundu[122] added to this transfer hydrogenation work with similar bidentate nitrogen-based
Mn(I) catalysts 50–52 (Figure [9]). Later, Chen used the biomimetic-inspired bidentate catalyst 53,[123] similar to catalyst 46 used by the Khusnutdinova group for the hydrogenation of carbon dioxide (see Scheme
[8C]). Hydroxy deprotonation allowed for the transfer hydrogenation of ketones and aldehydes.
Also, notably, Pidko used Mn–NHC complex 31 (Figure [9]) at very low catalyst loading (75 ppm) to reduce ketones almost quantitatively.[80]
Scheme 11 A. Transfer hydrogenation catalysts for the reduction of aldehydes, imines, esters,
and N-heterocycles (37, Ar = 3,5-dimethyl-4-methoxybenzene[130]); B. Reductive cross-coupling of N-heterocycles with an asymmetric Mn(I) base-activated
transfer hydrogenation catalyst; C. Switchable primary and secondary amine products
from the transfer hydrogenation of nitriles using SNN-based manganese catalysts
Multiple groups have also devised chiral ligand backbones to allow for the asymmetric
reduction of ketones to afford enantiomerically enriched secondary alcohols (Figure
[9]). The group of Kirchner and Zirakzadeh used a ferrocenyl-containing PNP′ pincer
ligand in catalyst 54 to reduce acetophenone in up to 85% ee and full conversion.[124] The group of Sortais used a cheaper, chiral diamine ligand 55 to generate the active species in situ, which afforded alcohols in high yield and
ee.[119] Morris and co-workers and, later, the Beller group also made chiral ligands for
asymmetric transfer hydrogenation;[125] however, they did not obtain the same ee levels. Mezzetti performed an in-depth
analysis on the transfer hydrogenation of acetophenone derivatives using their manganese
catalyst 57, despite particularly low ee values being observed for the alcohol products.[126] In 2020, Sortais screened a range of chiral PN bidentate ligands, with ligand 59 performing the best.[127] Only mild improvements in ee values in comparison with their chiral NN bidentate
system 55 were observed. In addition to minor mechanistic investigations reported in communications,
Pidko et al. performed a thorough investigation, including computational analysis,
of the asymmetric transfer hydrogenation of acetophenone using a simple chiral diamine
ligand, inspired by the work of Sortais.[128] Recently, another significant contribution to the field has been made by the group
of Sun;[129] by using complex 60, formed from an amino acid derived NN-bidentate ligand and MnBr(CO)5, a range of benzyl ketones could be reduced in high yields (up to 90%) and high ee
values (up to 93%). Computational calculations showed that π–π stacking between the
ligand and the substrate plays an important role in the enantiocontrol.
In 2020, Clarke et al. reported that catalyst 37 could be used for the double transfer hydrogenation of esters to give alcohols (Scheme
[11]).[130] The reduction of C–N π-bonds was demonstrated independently by the groups of Khusnutdinova
and Sortais in 2019.[131] Both aldehydes and imines could be reduced in high yield by catalysts 46 and 50. In addition, Khusnutdinova showed that N-heterocycles could also be reduced, albeit
in lower yield. In a specific study by Zhang and Ci (2020), the transfer hydrogenation
of N-heterocycles was performed within a sequential cross-coupling reaction catalyzed
by 61. Yields of up to 84% could be obtained (Scheme [11]B).[132]
The Adhikari/Maji and Garcia groups both, separately, using Mn(I) catalysts 63/64 and 62 (Scheme [11]), reduced nitriles using different hydrogen sources: the ammonia–borane complex
and butan-2-ol, respectively.[133] Both groups could isolate the primary amine in high yields; however, the group of
Maji could also isolate the secondary amine by modifying the polarity of the solvent
and optimizing with a modified catalyst (63). In addition, the nitrile could be coupled with a primary amine to create more diverse
secondary amines (Scheme [11]C). The group of Morrill used catalyst 65 (Scheme [11]A), an outer-sphere variation of complex 18 (used by Kirchner and Gonsalvi; see Scheme [1]), to reduce nitroarene compounds to their corresponding N-methylarylamines by using methanol as both the hydrogen donor and methylation source.[134]
The ammonia–borane complex has also been used as a hydrogen source in multiple single
reductions of alkynes to their corresponding alkenes. The Driess[135] and El-Sepelgy/ Azofra[136] groups could isolate E-alkenes and Z-alkenes, respectively (Figure [10]). While both used Mn(II) pre-catalysts, Driess used an N-heterocyclic silylene backbone
in 66, whereas El-Sepelgy and Azofra used a more traditional PNP backbone in 67 (Figure [10]), typical on Mn(I) centers for both hydrogenation and transfer hydrogenation. Interestingly,
the Mn(II) pre-catalyst was assumed to be reduced to a Mn(I) hydride prior to the
catalytic cycle due to the presence of the ammonia–borane complex, a known reducing
species.[136] The group of Rueping continued the work on Z-alkene formation by using a Mn(I) catalyst (47; Figure [10]) and methanol as the only sacrificial hydrogen source.[137] The group of Lacy also managed to reduce C–C π-bonds on chalcones by using amino
acid derived bidentate ligand 68 (Figure [10]) and MnBr(CO)5.[138]
Figure 10 Manganese-based transfer hydrogenation catalysts for the reduction of C–C π-bonds
(66, R =
t
Bu[135])
Conclusion and Perspective
6
Conclusion and Perspective
Manganese has become a metal of great interest due to its chemical versatility and
its high natural abundance. By using important catalytic design features that were
initially developed on rare earth and noble metal catalysts to improve performance,
the development of manganese-catalyzed reduction reactions has skyrocketed. With
a future focus of chemistry being the 12 principles of green chemistry, the use of
abundant metals and catalytic processes dictates that manganese catalysis can play
an important role. However, continual development and increased understanding of the
limitations of the current systems are still absolute requirements in replacing and
ending the reliance on highly active rare earth and noble metals. We hope this report
can assist in this regard, by highlighting current limitations and unknowns in the
literature.
