Hydrogenation in the Vitamins and Fine Chemicals Industry – An Overview

In the pharmaceutical and partly also in the fine chemicals industry many chemical conversions require stoichiometric amounts of reagents, and thus generate large amounts of waste [1, 2]. This is in contrast to the production of bulk chemicals which mostly relies on catalysis. This difference can be explained by the higher complexity of pharmaceuticals and fine chemicals which makes catalysis more demanding and process development more expensive.


Introduction
In the pharmaceutical and partly also in the fine chemicals industry many chemical conversions require stoichiometric amounts of reagents, and thus generate large amounts of waste [1,2]. This is in contrast to the production of bulk chemicals which mostly relies on catalysis. This difference can be explained by the higher complexity of pharmaceuticals and fine chemicals which makes catalysis more demanding and process development more expensive.
According to Sheldon's classification [3], most vitamins are typical fine chemicals with production volumes of about 100 to 10'000 tons per year. Some vitamins can be placed in the class of bulk chemicals. Typically these compounds have been produced industrially for decades in multi-step syntheses with high overall yields. The application of catalytic methods in the highly competitive field of vitamins has increased significantly in recent years because of price pressure on these products. Research and development is thus driven by the necessity to reduce waste, use less toxic reagents and solvents, improve energy efficiency, recycle catalysts and reagents, and combine unit operations to reduce costs and achieve more sustainable processes. These goals are mostly in accordance with the twelve principles of "green chemistry" [4,5,6].
Catalytic hydrogenation is certainly the most widely applicable method for the reduction of organic compounds and belongs to the most important transformations in chemical industry. Catalytic hydrogenations in the fine chemicals industry are usually carried out with heterogeneous catalysts. Homogeneous catalysts are typically applied for highly selective transformations, particularly enantioselective reductions. In the case of full recycling of heterogeneous or homogeneous catalysts, hydrogenation with molecular hydrogen is an atom economic transformation and undoubtedly the cleanest possible method for reducing a compound. Alternatively, hydrogen donors such as isopropanol or formic acid can be applied in transfer hydrogenations. In the field of catalytic hydrogenation reactions several significant inventions have been reported in the last 150 years. Sabatier and co-worker investigated the application of highly dispersed metals, e.g. nickel, in the hydrogenation of organic compounds [7,8]. The elective semihydrogenation of C≡C-bonds in presence of lead-doped palladium on calcium carbonate catalysts found by Lindlar was a further milestone in the field of catalytic hydrogenation reactions [9][10][11]. During the last decades asymmetric hydrogenations, pioneered by W.S. Knowles and R. Noyori, were a further highlight in the field of hydrogenation applied in organic synthesis [12,13].
Catalytic hydrogenations can be carried out in a variety of ways, either in the liquid or gas phase and in batch-wise or continuous mode. In continuous processes usually fixed-bed reactors or fluidized-bed reactors are used. The suitable choice of a reactor system depends on various factors such as, in particular, the choice of catalyst, the reaction conditions, heat formation, space-time-yield, residence time, hydrogen pressure, mass-transport phenomena, temperature, solvent and economic reasons.
In this contribution we will focus on industrially important catalytic hydrogenation reactions which are of interest for DSM Nutritional Products concerning the manufacture of vitamins, carotenoids and nutraceuticals. The various hydrogenations are organized by reaction types, rather than by the products prepared: hydrogenations of C=C double bonds, selective semi-hydrogenations of C≡C triple bonds, hydrogenations of C=X/C≡X multiple bonds (X = oxygen or nitrogen), and stereoselective hydrogenations.

Hydrogenation of C=C double bonds
Probably the most common hydrogenation reaction performed in industry is the hydrogenation of carbon-carbon double bonds. A wide variety of catalysts are available from commercial suppliers and this transformation is considered a robust and atomeconomical reaction. Even so, careful optimisation of the reaction conditions can be required to obtain full conversion and reduce or eliminate by-products. This is also important in the synthesis of vitamins and fine chemicals.
(all-rac)-α-Tocopherol (3) is the economically most important member of the group of vitamin E compounds, due to its biological and antioxidant properties. This fat-soluble vitamin is produced on a scale of >30'000 tonnes per year (mostly in form of its acetate derivative 4) for applications in human and animal nutrition. One of the key building blocks for the chemical production of synthetic vitamin E is trimethylhydroquinone (TMHQ, 1), which is converted into (all-rac)-α-tocopherol (3) by condensation with (all-rac)-isophytol (2, Scheme 1) [14][15][16]. The synthesis of isophytol can be carried out starting from acetone (5) and building up the isoprenoic side chain by a sequence of C2 and C3 elongations (via 8 and 10, Scheme 2). Another approach starts from citral (9) followed by reaction to linalool (11) and elongation to geranylacetone (13) [14,15]. Routes to 9 based on prenyl chloride and myrcene (6) are described in literature but are not competitive. The use of cheap isobutene (7) is preferred which represents also an access to other isoprenoic building blocks, e.g. 14. The C3-elongation can be carried out by Saucy-Marbet or Carroll reactions and the C2 elongation by ethynylation or vinylation reactions (vinyl Grignard addition) [1]. In such reaction sequences several hydrogenations of C=C bonds are necessary. A key intermediate in the synthesis of isophytol (2) is hexahydropseudoionone (12), which can be produced from pseudoionone (10) or geranylacetone (13). In the past the hydrogenation reactions were carried out batch-wise in presence of a Pd/C catalyst below 80 °C and <10 bar pressure [17]. This hydrogenation was also investigated using Pd-or Rh-containing polymers on Al2O3 [18]. The liquid phase hydrogenation of pseudoionone (10), geranylacetone (13) or dihydrogeranylacetone in presence of a suspended catalyst in a special reactor which allows an easy hold-up of the catalyst has been described in [19]. The synthesis of these saturated ketones in a continuous fixed-bed mode applying a Pd catalyst supported on SiO2 results in excellent yield under nearly full conversion [20].  The hydrogenation is usually carried out in a continuous mode at medium to low pressure and elevated temperature to prevent crystallization of the product from the reaction mixture. Yields are generally quantitative.
Alternatively, a range of different Pd-catalyzed hydrogenation conditions can be found in the literature, e.g. using Pd/C in solvents like carboxylic esters [21] or acetone [22], or palladium on acidic [23,24] or basic oxides [25] in lower alcohols. Also a Pt-catalyzed hydrogenation has been described using platinum nitrate and aluminium oxide in i-butanol [25].
An alternative access to TMHQ (1) starts from less methylated 1,4-benzoquinones, e.g. 2,6dimethylbenzoquinone (2,6-DMQ, 16), which is first hydrogenated to 2,6dimethylhydroquinone (17) and methylated later, e.g. by aminomethylation via 18 followed by hydrogenolysis to TMHQ (1) (Scheme 4) [26]. The K-vitamins are a group of substituted 2-methyl-1,4-naphthoquinones with (or without) a prenyl chain of different length in C-3 position ( Figure 1). The core unit of all K-vitamins is menadione (vitamin K3, 21) [27,28]. The standard synthesis of vitamins K1 and K2 (19,20) is the coupling of the aromatic unit with the (poly)prenyl side chain, similar to vitamin E. The direct coupling of the side chain to menadione (21) is not possible; therefore a key step is the hydrogenation of menadione to menadiol (22) so that alkylation can take place. The hydrogenation is usually carried out batch-wise using a palladium on carbon catalyst.
A remarkable stereoselective hydrogenation of a trisubstituted olefinic C=C double bond was used in several total synthesis routes to the water-soluble vitamin (+)-biotin (26, Scheme 6) [32]. This product is produced on a scale of about 100 tonnes per year, and only the (3aS,4S,6aR)-stereoisomer exhibits full biological activity. Stereocenter C-4 of the thiophane ring can be introduced by catalytic hydrogenation of the exocyclic olefin 24 with undefined double-bond stereochemistry on Pd/C or other heterogeneous catalysts, yielding 25 with the desired all-cis relative configuration at centers C-4, C-3a, and C-6a. The N-benzyl protecting groups are stable under those conditions. This strategy, originally developed by Goldberg and Sternbach [33,34], was later on used by other groups in various syntheses of racemic and optically active intermediates [32].  (3aS,4S,6aR)-stereoisomer Scheme 6. Stereoselective heterogeneous hydrogenation in routes towards (+)-biotin.

Semi-hydrogenation of C≡C triple bonds (Lindlar type)
The semi-hydrogenation of carbon-carbon triple bonds to alkenes is one of the most useful hydrogenations for the production of vitamins, however careful choice of catalyst and reaction conditions are required to obtain high selectivity. In general, hydrogenation of acetylenes with a metal catalyst results in the formation of the fully saturated alkane product, since the second hydrogenation (alkene to alkane) is generally faster than the first (alkyne to alkene). However, as long as some of the starting alkyne remains in the reaction mixture, selectivity can be high since the alkynes bind more strongly to the metal surface. The selectivity can be enhanced by the use of suitable catalyst poisons which modify the activity of the metal catalyst.
One of the most widely used and most selective catalysts is the one originally developed by Lindlar [9,11]. With this catalyst, the palladium supported on calcium carbonate is doped with a lead acetate solution during manufacture. This catalyst can then be used either directly in the hydrogenation or modified further by an organic compound such as an amine. Since the hydrogen is delivered from the metal surface to the alkyne, usually high selectivity is obtained for the Z-(cis)-alkene product. "Lindlar catalysts" generally have 5% palladium loading and 2-5% lead loading, depending on the application.
One of the earliest uses of the catalyst developed by Lindlar was the semi-hydrogenation of a vitamin A key intermediate (27,Scheme 7) to give tetraene 28. Whilst this could be achieved with poisoned palladium on charcoal or palladium on calcium carbonate [35], selectivities were significantly higher with the lead-doped catalyst and the reaction could easily be stopped after the uptake of just one equivalent of hydrogen gas. From MBE, the chain is extended in a sequential manner to obtain dehydroisophytol (31, Scheme 9). This is then reduced in another semi-hydrogenation to give isophytol (2, cf. Scheme 2) [15]. Isophytol can then be coupled with TMHQ (1), as described previously, to form α-tocopherol (cf. Scheme 1). As with MBY-MBE, the hydrogenation is carried out in a batch-wise process at 2-5 bar hydrogen pressure.  As an extension to the work above, another fragrance compound, dimethyloctenol (DMOE, 36), can be prepared by the combination of a hydrogenation of a C=C double bond and a selective semi-hydrogenation of a C≡C triple bond (Scheme 11). Methylheptenone (34) can be hydrogenated with a range of Pd/C or Pd/Al2O3 catalysts to give methylheptanone (35). The reaction was run without solvent at a range of temperatures (30-80 °C) and pressures (1-10 bar hydrogen). The latter compound was ethynylated to give the alkyne 14, which can undergo semi-hydrogenation with high selectivity (up to 95%) to give the desired tertiary allyl alcohol 36. Lindlar catalysts with varying amounts of lead-doping were successful at moderate temperatures (20-40 °C) and pressures (1-10 bar hydrogen) [36].
New approaches for the application of Lindlar-type catalysts are the use of supported palladium nanoparticles. By carefully controlling their preparation, a narrow range of diameters can be obtained and deposited on a carbon support [37,38]. These catalysts allow the hydrogenation of C≡C bonds with low metal loadings and in several cases with a high selectivity, e.g. hex-3-yne can be hydrogenated to hex-3-ene in high selectivity at full conversion, however to the best of our knowledge, these have not yet been applied on an industrial scale for the production of vitamins and fine chemicals. Further trends in the research on Lindlar hydrogenations are focusing on the addition of FeCl2 and tetramethylammonium chloride to the catalyst, and the use of palladium on metal sintered fibers. This allows the hydrogenation of triple bonds in a continuous reaction mode, also with low Pd loadings, in which Pb doping is not necessary [39][40][41][42].
Resveratrol (40, Scheme 12) is a polyphenol that can be isolated from natural sources such as red grapes or giant knotweed. It has attracted significant attention due to its possible health benefits in areas such as anti-cancer, anti-aging, anti-inflammatory and cardiovascular protection. The synthesis of resveratrol has been reported a number of times and can be industrially performed using Mizoroki-Heck coupling of 3,5-diacetoxy-styrene and 4acetoxybromobenzene as key step [43,44].
An alternative approach to resveratrol uses the selective hydrogenation of tolan derivatives 37 in presence of a Lindlar-type catalyst to form intermediate 38 [45]. The precursor 37 can be synthesized by Sonogashira coupling (Scheme 12). In general this procedure can be applied to the synthesis of electron-rich stilbene derivatives as in the preparation of natural compounds like combretastatin (39) [46].

Z-Butene-1,4-diol (42) is an intermediate in the synthesis of vitamin B6 (43), and is synthesized by selective hydrogenation of butynediol (41) in presence of a Pd/CaCO3
catalyst (Scheme 13) [47,48]. The hydrogenation also proceeds well in water. The application of Pd/C or Pd/Al2O3 has been described under similar conditions [49]. The hydrogenation can also be carried out in a monolith bubble column reactor. Full conversion and high selectivity can be achieved in presence of a Pd-catalyst [50].

Hydrogenation of C=O and C≡N functional groups
As well as the hydrogenation of C-C multiple bonds, the hydrogenation of C-X multiple bonds is used extensively in the production of vitamins and fine chemicals.
L-Ascorbic acid (vitamin C, 46) is the vitamin that is produced on the largest volume worldwide (approximately 100,000 tonnes per year). The most important method for its industrial manufacture is the Reichstein process [53], which transforms D-glucose (44) into L-ascorbic acid via a mixed series of high-yielding chemical and microbiological steps. In the first step of the synthesis sequence D-glucose is hydrogenated to D-sorbitol (45) in presence of a Ni-alloy catalyst (Scheme 14). The reaction is usually carried out at high pressure and elevated temperature in a batch-mode or continuous process. Under these conditions the product is obtained in high selectivity and almost quantitative yield with only minor amounts of D-mannitol and L-iditol as by-products.
Alternatively to the catalytic hydrogenation, microbiological [54] and electrochemical [55] methods also are known for the reduction of D-glucose to D-sorbitol. Vitamin B1 (thiamine chloride, 49, Scheme 15) contains two heterocyclic rings linked by a methylene unit. In all industrial syntheses, the key intermediate is Grewe diamine (48), onto which the thiazole ring is constructed. In general, the diamine is prepared by hydrogenation of a pyrimidino nitrile (47) using a nickel-alloy catalyst. This was first reported in 1944 by Huber [56] who found that the use of palladium and platinum supported catalysts gave significant amounts of secondary amine 50 as by-product. The use of nickel catalysts also gave significant quantities of secondary amines, but these could be reduced to less than 5% by the addition of ammonia to the reaction mixture.
Modern industrial syntheses still use the same basic process, although optimisation has improved the reaction significantly, further reducing the unwanted by-products. Reactions proceed in a batch-wise process with a solvent saturated with ammonia at moderate (>10 bar) hydrogen pressure to ensure high activity and throughput. The catalyst can be recycled multiple times and usually remains in the reaction after batch. The choice of nickel catalyst is also important and recently nickel-alloy catalysts (Centoprime ® ) have been developed that reduce amine by-products further [57][58][59][60]. Resveratrol (40) can be produced industrially by an alternative route to that described earlier (cf. Scheme 12). A key intermediate in this production is the benzylic alcohol 52 [43]. This is prepared by the hydrogenation of acetophenone derivative 51 using a transition metal catalyst [61]. A number of catalysts can successfully perform this transformation, but the suppression of by-products is essential. Use of palladium or platinum on carbon resulted in yields of up to 90% (depending on the solvent); however levels of critical by-products were too high to be applied on large scale. The problems were solved by the use of nickel-alloy catalysts. Using these catalysts in ethyl acetate (2-10 bar hydrogen) allowed the production of the required alcohol 52 in greater than 95% yield. In addition the catalyst can remain in the reactor and be recycled multiple times resulting in a cost effective process.

Stereoselective hydrogenation of C=C double bonds
Stereoselective processes, in particular the asymmetric hydrogenation of C=C double bonds, play an increasingly important role in the total synthesis of isomerically pure biologically active products. The naturally occurring fat soluble antioxidant (2R,4'R,8'R)-α-tocopherol (53, Scheme 17) is not available in sufficient amounts from natural source starting material.
In the course of the considerable efforts towards an economic total synthesis of 53 during the last decades [16], exceptionally efficient new asymmetric hydrogenation processes were developed for the introduction of chirality to the aliphatic tocopherol side chain. The homogeneous asymmetric hydrogenation of allylic alcohols catalyzed by ruthenium complexes could be performed on pilot scale with substrate-to-catalyst ratios of up to 150'000 (Scheme 17). The C10-building block (E)-54 was transformed into (R)-57 with >99% selectivity by using (S)-MeOBIPHEP (56, Ar = p-Tol, X = OCH3) as ligand. Under similar conditions, hydrogenation of (E)-55 gave (R,R)-58 (>98% RR) with the catalyst derived from (S)-p-Tol-BIPHEMP (56, Ar = p-Tol, X = CH3) [62].
Even two chiral centers can be introduced by the one-pot reduction of unfunctionalized trialkyl substituted olefins in the presence of Ir-BArF complexes containing chiral P,Nligands (Scheme 18). By applying this novel retrosynthetic concept, an (all-R)-tocopherol derivative could be obtained for the first time from the corresponding (all-E)-tocotrienol derivative in a collaboration of the Pfaltz group with DSM Nutritional Products. Asymmetric hydrogenation of γ-tocotrienyl acetate (R,E,E)-59 with pyridyl phosphinite 60 yielded acetate 61 with complete conversion and excellent stereoselectivity (>98% R,R,R) [63,64]. A Rh(I)-catalysed highly diastereoselective hydrogenation is the basis for a very short synthesis of (+)-biotin (26) developed by Lonza and applied on a technical scale (Scheme 19). The hydrogenative key transformation was the result of a cooperation with the catalysis group of former Ciba-Geigy [65]. After ligand screening and optimisation, the conversion of substrate 63, easily accessible from diketene via cheap tetronic acid (62), could be improved to high selectivity (65a:65b >99:1) with ferrocene derived josiphos2 (64) as ligand. Although the synthesis via thiolactone 66 and subsequent stereoselective heterogeneous hydrogenation of olefin 67 (cf. Chapt. 2, Scheme 6, 23→24) was operated on tonne scale, production had to be terminated due to high production costs. The final debenzylation step to yield 26 led to destruction of the chirality in the expensive (R)-methylbenzylamine auxiliary used for introduction of one of the two nitrogen functionalities.  Scheme 23. Asymmetric transfer hydrogenation of (R)-levodione in the synthesis of (3R,3'R)zeaxanthin.

Conclusions
Various types of hydrogenation reactions are indispensable parts of economically and ecologically beneficial manufacturing processes towards valuable products in the vitamins and fine chemicals industry. The increasing importance of environmentally benign production methods is addressed by developing concepts for improving the efficiency of transformations, continuous processing and recycling, and achieving high chemo-and stereoselectivities, thus avoiding laborious separation protocols and waste formation. The examples presented and discussed in detail show many achievements in this field during recent decades, but also the necessity to further search for alternative solutions.