3.1 Biomass Conversion versus Crude Oil Conversion: Diverging Approaches

Crude oil feedstock has low functionality which makes it directly suitable for use as a fuel after prior processing (for example cracking and isomerisation). Functional groups, such as C=O and OH, are added to crude oil derived feedstock to produce bulk and specialty chemicals. Here, special care is taken to ensure selective addition of the functional group without over functionalisation of the substrates (Figure 3). Contrastingly, biomass derived feedstocks, such as cellulose and hemicellulose, contain far too much functionality to use directly as fuels or bulk chemicals, and therefore require selective defunctionalisation strategies (5, 12, 13).

3.2 Biomass Conversion Technologies

The three common conversion technologies for carbohydrates and cellulose are shown in Figure 4 (2, 14–16). Although production of first generation biofuels (bio-ethanol and bio-butanol) is well established, this process relies on starch and sugar feeds which compete with the food chain (16). Hydrolysis of cellulosic feedstock to fermentable sugars has been achieved and a number of commercial scale cellulosic ethanol plants have been and are being developed in several countries (for example the USA, Italy, Brazil and Canada) as alternative routes to bio-alcohols as fuels in order to avoid food sources as feedstock (17–20). Slow reaction rate, high cost and sensitivity of enzymes and energy intensive subsequent distillation and drying steps remain challenges to achieving cost-effectiveness in these processes. A cost-effective process that transforms biomass (such as wood, sugarcane bagasse or corn stover) to aromatic compounds has been developed by Anellotech, USA. This process uses zeolite-based catalysts, co-developed with Johnson Matthey, to produce gases which are then converted to benzene, xylene and toluene (bio-BTX) (21). Thermochemical processes such as gasification, pyrolysis, torrefaction and liquefaction require intense heating at elevated temperatures, therefore raising energy efficiency concerns. In addition, the selectivity in bio-oils produced from pyrolysis is extremely poor, therefore inevitably requiring expensive additional upgrading and separation steps (22–23). This has led to a number of investigations into novel pathways to reduce costs (24, 25).

6.1 Polysaccharides and Monosaccharides to 5-Hydroxymethylfurfural

The dehydration of hexoses affords HMF, some furfural, LA, formic acid and humins. On the other hand, dehydration of pentoses produces furfural (Scheme II) (2, 4).

HMF and furfural are important platform chemicals that can be upgraded to valuable commodity chemicals, fuels and fuel additives. HMF is formed as an intermediate during dry heating and roasting of foods containing high levels of carbohydrates (54). It has been found in foods and beverages such as dried fruits, baked foods, coffee, honey, citrus juice, grape juice, prune juice, wine and brandy amongst others. Daily HMF intake is estimated at 30 to 150 mg per person and is toxic to humans if ingested at high concentrations (55–57). HMF can be accessed by three dehydration steps of hexose sugars. The mechanistic pathway for acid catalysed dehydration of hexoses by expulsion of three water molecules is said to proceed via acyclic intermediates or cyclic intermediates (58–62). The cyclic mechanism (Scheme III) begins from the cyclic ketofuranose which undergoes dehydration at C₂ to form a carbenium cation, followed by consecutive β-dehydrations in the ring to afford HMF (61).

The acyclic pathway (Scheme IV) proceeds via formation of 1,2-enediol, which is the rate determining step. This is followed by two β-dehydrations and finally ring closure to yield HMF (61).

In both pathways, the selective conversion of hexoses to HMF is influenced by the isomerisation of glucose to fructose. Glucose isomerisation is catalysed by Lewis acids and Brønsted acids (63). Brønsted acids tend to degenerate monosaccharides to various byproducts (64–72) and afford low (<10%) yields of fructose. The use of organic bases such as piperazine, triethylamine, pyrrolidine, ethylenediamine, piperidine and morpholine has been reported. Fructose yields of up to 32% and 63% selectivity were achieved (66). These are the first organic base catalysed glucose to fructose isomerisation systems that gave yields and selectivities for fructose that are comparable to widely used Lewis acids (68–72).

The conversion of hexoses and cellulosic biomass to HMF has been dominated by Lewis acid and mineral acid catalysts and has recently been reviewed (70). Some noble metal-containing catalytic systems have demonstrated good activity for this transformation (Table I) (71–79).

In 2007, Zhang and coworkers reported that fructose dehydration proceeded in the presence of catalytic amounts of MCl₂ and M′Cl₃ (where M = Cr, Fe, Cu, Pd, Pt and M′ = Rh, Rh, Ru, Mo). The catalytic systems were selective to HMF and negligible amounts (0.08%) of LA were produced. Notably, only CrCl₂ gave high yields (70%) of HMF from glucose, whereas other Lewis acid catalysts afforded yields less than 10% (71). The role of Cr²⁺ is to convert the α-glucose to β-glucose and isomerise the β-glucose to fructose, which improves the yield of HMF. A mixture of CrCl₂ and CuCl₂ in 1-ethyl-3-methylimidazolium chloride ([EMIM]Cl) was used to transform cellulose into HMF (in 59% yield) in a single step. This catalytic system achieved this by prior hydrolysis of cellulose, followed by dehydration of glucose units into HMF. The authors suggest that CrCl₂ activates CuCl₂ to hydrolyse cellulose into glucose. Separate experiments with a CuCl₂ catalyst only and with a paired CuCl₂/CrCl₂ catalyst (where the relative proportion of CuCl₂ in the pair was higher than CrCl₂), show that the paired metal chlorides yield more than three times the amount of product (i.e. glucose from cellulose) than the CuCl₂ catalyst on its own. On the other hand, the CuCl₂/CrCl₂ pair with a relatively higher proportion of CrCl₂ resulted in a high HMF yield, consistent with CrCl₂ being a highly selective catalyst for the conversion of glucose to HMF. The catalytic system could be reused up to three times with consistent activity and selectivity (71).

In another report a solid Ag-containing HPA (Ag₃PW₁₂O₄₀ HPA) was used as a solid catalyst to dehydrate fructose and glucose to HMF. HMF yields of 78% and 76% were obtained from fructose and glucose respectively. The Ag₃PW₁₂O₄₀ cluster exhibited higher catalytic activity and selectivity as compared to when Brønsted acids (HCl and phosphotungstic acid (H₃PW₁₂O₄₀)) and Lewis acid (silver nitrate (AgNO₃)) were used separately in glucose dehydration to HMF. Therefore, the higher catalytic activity of Ag₃PW₁₂O₄₀ can be attributed to the synergistic effect of Lewis acid and Brønsted acid sites in the Ag₃PW₁₂O₄₀ cluster. Infrared (IR) experiments revealed that the substrate accumulated on the Ag₃PW₁₂O₄₀ catalysts and could be the reason for the high activity displayed by this catalyst. Fourier transform (FT)-IR spectra of Ag₃PW₁₂O₄₀ and its adsorbed fructose and glucose confirmed the absorption of substrate on part of the catalyst. Notably, the Ag₃PW₁₂O₄₀ cluster as a catalyst could be recycled and reused up to six times (73).

In a further report, CrC_l2 and RuCl₃ in a 4:1 molar ratio catalysed the conversion of cellulose to HMF in [EMIM]Cl IL in a 60% yield. Gram scale production of HMF was achieved with this CrCl₂/RuCl₃ system. Furthermore, lignocellulosic biomass from reed was also converted to HMF (41% yield) and furfural (26%) using the same catalytic system and conditions (74). CuCl₂ in 1-ethyl-3-methylimidazolium acetate ([EMIM [Ac]) has been used to produce HMF in ≈70% yield (75).

Good catalytic performance has been recorded for the dehydration of fructose to HMF with iridium(III) chloride (IrCl₃) and AuCl₃•HCl catalysts in HCl and 1-butyl-3-methylimidazolium chloride ([BMIM Cl). A HMF yield of 89% was obtained with IrCl₃ catalyst in 30 min at 120°C while AuCl₃•HCl gave 44% HMF yield under the same conditions (77).

6.2 Polysaccharides and Monosaccharides to Furfural

The synthesis of furfural has been achieved by using Brønsted and Lewis acid catalysts (80–82). In a report by Zhang et al. CrCl₃, CrCl₃/lithium chloride (LiCl), FeCl₃•6H₂O, CuCl₂•H₂O, copper(I) chloride (CuCl), LiCl and AlCl₃ were screened for catalytic activity and selectivity in xylose and xylan conversion to furfural, under microwave irradiation. AlCl₃ gave the best performance for the production of furfural from xylose (82.2%) and xylan (84.8%) (80).

Furfuryl alcohol (FA) has also been obtained directly from xylose using a combination of Pt/SiO₂ and ZrO₂-SO₄ catalysts. The SiO₂ was reported to promote xylose to xylulose isomerisation (with xylose conversion of 65%), while the ZrO₂-SO₄ catalysed the dehydration of xylulose to furfural. Finally, furfural was selectively transformed to FA by Pt/SiO₂ (81).

6.3 5-Hydroxymethylfurfural and Furfural as Sources of Biofuels

Sections 6.1 and 6.2 provide avenues to HMF and furfural, versatile renewable platform chemicals that can provide access to a variety of biofuels and chemicals (Schemes V and VI). Since the 1920s, the Quaker Oats Process has been producing furfural and FA from agricultural residues (83). AVA Biochem, Switzerland, produces 300 tonnes of 90% purity aqueous HMF and 20 tonnes of highly pure crystalline HMF per annum, (84) and there are examples of patented pilot scale processes that have produced HMF using organic catalysts, mineral acid catalysts and without a catalyst (85–88). Although Lewis and Brønsted acid catalysts have dominated efforts to produce these platform chemicals, noble metal containing catalytic systems (including monometallic and bimetallic species) have shown remarkable efficacy in converting HMF and furfural to the chemicals illustrated in Schemes V (89) and VI (90). In Section 7 we will discuss some of the noble metal promoted catalytic transformations illustrated in Schemes V and VI.

7.1 Polysaccharides and Monosaccharides to 2,5-Dimethylfuran, 2-Methylfuran, 2,5-Dimethyltetrahydrofuran, 2-Methyltetrahydrofuran and 5-Ethoxymethylfurfural

In comparison to the leading biofuel, bioethanol, DMF has a higher energy density (30 MJ l^–1), a higher octane number of 119 and a lower volatility (boiling point = 92–94°C) than bioethanol (Table II). Furthermore, it is immiscible with water therefore, in this regard, is similar to gasoline. These properties make DMF a better biomass derived liquid fuel. Policy limitations and/or cost-effective production of DMF at an industrial scale are likely reasons why DMF has not yet been taken up as a fuel additive of choice, given its many technical advantages over bioethanol. DMF can be obtained from the HDO of HMF in the presence of a catalyst. Metal catalysts have been widely used to promote this reaction (88). For example, Dumesic et al. used a CuRu/C catalyst system that produced 71% DMF in 10 h when 1-butanol was used as the solvent of the reaction at 220°C and 6.8 bar H₂ (91). This catalyst system could also be utilised to promote the conversion of HMF to DMF, where the HMF was initially prepared by mineral acid catalysed dehydration of corn stover (93).

An even better catalyst system was reported by Wang et al. with a selective bimetallic catalyst (PtCo nanoparticles supported on hollow carbon nanospheres, PtCo@HCS) for the conversion of HMF to DMF at 180^oC with a 98% yield in 1-butanol. The PtCo@HCS could be recycled and reused up to three times, although by the third cycle, a drop in DMF yield (from 98 to 72%) occurred (94). The role of the catalyst support in this reaction appears to be critical. For example, when Pt nanoparticles supported on activated carbon (Pt/AC) and graphitic carbon (Pt/GC) were employed in this reaction, only 9% and 56% DMF yields respectively were achieved. However, PtCo nanoparticles on activated carbon (PtCo/AC) and graphitic carbon (PtCo/GC) gave conversions of 100% HMF and 98% DMF, effectively proving that the two metals work cooperatively to transform HMF into DMF. Selective HDO of HMF over Ru/Co₃O₄ and PdAu/C has also been reported. These reactions were conducted in tetrahydrofuran (THF) using H₂ at 130°C for 24 h and at 60°C for 6 h respectively; giving excellent DMF yields of 100% and 93.4% respectively (95, 96). It is interesting to note that Ru and Pd alone without either Co or Au are less effective in catalysing HMF to DMF. X-Ray photoelectron spectroscopy (XPS) spectra of PdAu/C catalyst reveal that Pd atoms in the PdAu/C may have electronic poor states (after charge transfer from Pd to Au atoms occurs) which strongly contributed to the enhancement of HDO activity thus improving DMF yield.

Notably, Ru/C and Pd/C with and without ZnCl₂ in converting HMF to DMF were reported by Zu et al. (96); Ru/C/ZnCl₂ produced 41% DMF and 52% 2,5-dihydroxymethyl furan (DHMF) and Ru/C alone gave a minimal amount of DMF from HMF. The higher activity displayed by the bimetallic catalyst (Ru/C/ZnCl₂) as compared to Ru/C suggests enhanced HDO in the presence of Zn²⁺, provided by the synergy between the two metals. It also explains why Pd/C/ZnCl₂) gave better yield than Ru/C/ZnCl₂) in this study (97, 98). In the presence of a Brønsted acid Amberlyst^®-15, instead of ZnCl₂, a mixture of products was obtained from HMF, namely DMTHF (13%), DMF (6%), tetrahydrofurfuryl alcohol (THFA) (3%), 2,5-bis(hydroxymethyl)tetrahydrofuran (DHM‐THF) (23%), 2-methyltetrahydrofuran-5-aldehyde (4.5%) and methyltetrahydrofuryl alcohol (6%) and 2-hexanone (1.6%) (Schemes V and VI). This suggests that the acidity of ZnCl₂ is not the only reason for the enhanced activity of the catalyst, resulting in high DMF yield. Since the authors detect leached zinc in the solution, it is likely that the role of Zn²⁺ which leaches into solution during the reaction is as described by Abu-Omar and co-workers, where Zn²⁺ ions bind to the substrate and activates its cleavage upon encountering Ru–H or Pd–H sites on the catalyst surface (98).

Similarly, in the presence of a RhCl₃/HX catalytic system (X = I and Cl), a mixture of furan ring saturated and ring-opening products were also obtained from xylose, glucose, fructose, sucrose, inulin, cellulose and corn stover. The reaction is conducted in aqueous media and the main products obtained were DMTHF (up to 85% from fructose and 80% from cellulose) and MTHF (up to 80% from xylose and 56% from corn stover) (99, 100). These are gas chromatography (GC) yields and have not been isolated. DMHTHF has been hydrogenated to 1,6-hexandiol, which was converted into caprolactone and caprolactam (101).

Generally, molecular hydrogen has been the source of hydrogen in HDO reactions; but the use of hydrogen transfer reagents is safer than using molecular hydrogen in HDO reactions. Here, formic acid as a hydrogen source in the conversion of HMF, fructose and lignocellulose (with Ru/C catalyst) to chemicals provide useful examples of non-molecular hydrogen in HDO reactions. For example, when HMF was used as a substrate, 37% DMF, 43% 5-formyloxymethyl furfural (FMF) and 3% LA were obtained. Other examples include the initial direct conversion of lignocellulose and carbohydrate substrates in a Brønsted acid IL to catalyse the conversion of these substrates to HMF, and then subsequent use of formic acid as hydrogen source and Ru/C as catalyst to convert HMF to FMF, which was further converted to DMF. Interestingly, the feedstock in this reaction appears to be crucial in determining the yield of DMF. The yields of DMF from cellulose is 16% whilst the yield is 10% from sugars compared to when pure HMF was used as the substrate (37%) (102). Isopropyl alcohol can also be used as a hydrogen source in place of formic acid in the Ru/C catalysed conversion of HMF to DMF. The yield of DMF obtained was 81% and the reaction was said to proceed by the formation of 2,5-bis(hydroxymethyl)furfural as an intermediate (103).

MF has a high density (913 kg cm^–3), low flash point (–11°C) and low solubility in water, unlike ethanol. The energy density of MF is higher than that of ethanol and similar to DMF (Table II) (91). MF is a water soluble liquid, it is flammable and has a chocolate odour. It is also used as feedstock, as a solvent and has been widely used in pesticides and the perfume industry (104, 105). MF production proceeds by dehydration of xylose followed by HDO of furfural. The best catalyst for this reaction is a polymer supported Pd(II) complex which exhibited 100% yield conversion of furfural into MF with H₂ pressure 106).

EMF is another compound that can be accessed by etherification of HMF (Scheme V). It is a potential biofuel and has an energy density of 30.3 MJ l^–1, which is higher than that of ethanol and comparable to gasoline. It has, thus, been proposed as an additive and/or substitute for diesel as a fuel (89–91). Lin and coworkers have used Ag_x H_3-x PW catalysts, prepared through a H⁺ exchange for Ag⁺, for esterification of HMF to EMF. In using this series of catalysts, AgH₂PW displayed higher catalytic performance than Ag₂HPW and Ag₃PW in the etherification of HMF; producing a 69.5% yield of EMF when fructose was the starting material. The catalyst could be recycled and showed no marked decrease in activity (107).

In addition to EMF, 2,5-bisalkoxymethylfurans have also been identified as potential renewable liquid fuels. Bell and coworkers have reported a one-pot process involving the sequential dehydration and reductive etherification of fructose to 2,5-bisalkoxymethylfurans. The dehydration and etherification of fructose was performed using Amberlyst^®-15 and Pt/Sn/alumina at 110°C to form 5-(alkoxymethyl)furfurals and 2-(2,2-alkoxyethyl)-5-(alkoxymethyl)furans. At 60°C and under H₂ pressure, 5-bisalkoxymethylfurans and 2,5-bisalkoxymethyl furans could still be obtained in good yields (108).

7.2 Polysaccharides and Monosaccharides to Furfuryl Alcohol and Pentanediols

FA is prepared industrially by the catalytic hydrogenation of furfural using Ni and Cu/CrO catalysts. If the catalyst used is not selective, hydrogenation of furfural yields not only the desired products but also a variety of byproducts including furan, THF, THFA and ring-opening products, such as pentanol and pentanediols (Scheme VI) (109–114). Pd/SiO₂ as catalyst for the hydrogenation of furfural using 1 bar of H₂ at 230°C has been reported by Sitthisa et al. to produce FA in a yield of 65% (109). Similarly, furfural transformation to FA has been performed in the vapour phase using silica supported Pt nanoparticles. With small Pt nanoparticles on SiO₂, decarboxylation of furfural to furan was observed while larger sized Pt nanoparticles produced furan and FA (Scheme V) (110).

HDO of furfural can also produce 1,2-pentanediol (1,2-PDO), 1,4-pentanediol (1,4-PDO) and 1,5-pentanediol (1,5-PDO) which can be used as monomers for the synthesis of polyesters (Scheme VI) (111). In an investigation of the effect of adding noble metals (including Ru, Rh, Pt, Pd) to Ir-ReO_x /SiO₂ in the one-pot conversion of furfural to 1,5-PDO, Pd 0.66 wt%-Ir-ReO_x /SiO₂ was found to be superior. The Pd 0.66 wt%-Ir-ReO_x /SiO₂ gave 71.4% 1,5-PDO while 28.6% of the products consisted of 1,4-PDO, 1,2-PDO, 1-pentanol, 2-pentanol, THFA and 2-MTHF (112).

Catalyst characterisation revealed that Pd–Ir–ReO_x /SiO₂ catalyst consisted of Pd metal particles with ReO_x species (Pd–ReO_x ) and Ir metal particles with ReO_x species (Ir–ReO_x ) (Figure 5). The Pd–ReO_x species catalyse hydrogenation of furfural into THFA and the Ir–ReO_x species catalyse the HDO of THFA to 1,5-PDO. The ReO_x species modified Pd sites (i.e. decreased Pd metal average particle size) and increased the dispersion of Pd metal. This effect of ReO_x is said to be linked to the high performance of Pd–Ir–ReO_x /SiO₂ in the hydrogenation of furfural into THFA. Supported Ir was found to be in the metallic state and the Re species form low valent oxide clusters (ReO_x ) that partially cover the Ir metal surface, therefore the Ir–ReO_x species can catalyse the hydrogenolysis of THFA to 1,5-PDO (112).

More recently, Huber et al. reported a new process for the production of 1,5-PDO (84% yield) from furfural. This route to 1,5-PDO proceeds via hydrogenation → dehydration/hydration → ring-opening tautomerisation and → hydrogenation reactions, using Ni, γ-Al₂O₃ and Ru catalysts respectively. While this process has more reaction steps than conventional hydrogenation of furfural to THFA followed by HDO of THFA to 1,5-PDO, techno-economic assessments demonstrate that the process is economically viable (113).

Hronec and co-workers described a novel, highly selective transformation of furfural to CPO (Scheme V). Selective rearrangement of furfural to CPO by 5%Pt/C is achieved in water, at reaction temperatures of 140°C to 190°C and hydrogen pressures of 30 to 80 bar. Notably, when 5% Ru/C and 5% Pd/C were used under the same conditions, a mixture of typical furfural hydrogenation products (FA, THFA, MF and MTHF) were obtained (114).

Homogeneous catalysts such as [Pd(OAc)₂]/D^t BPF (D^t BPF = 1,1′-bis(di-tert butylphosphino) ferrocene), [Rh(CO)₂(acac)]/H₂WO₄, {[Cp*Ru (CO)₂]₂(μ-H)}⁺(OTf^–)/HOTf (Cp* = C₅Me₅ and OTf = CF₃SO₃^–) and cis -[Ru(6,6′-Cl₂bpy)₂(OH₂)₂](OTf)₂ (6,6′-Cl₂bpy = 6,6′-dichloro-2,2′-bipyridine) catalyse HDO of C–O bonds (115–118). For example, {[Cp*Ru(CO)₂]₂(μ-H)}⁺(OTf^–) is an efficient catalyst for the dehydration of 1,2-PDO followed by hydrogenation to form n -propanol under acidic conditions. This catalyst produced 54% n -propanol after 30 h at 92% 1,2-PDO conversion (119, 120). These homogeneously catalysed reactions are typically carried out at temperatures ranging from 80°C to 120°C, meaning the active species is likely a molecular catalyst or a ‘cocktail’ of catalysts, comprising molecular catalyst and nanoparticles (121).

Direct conversion of furfural to γ-valerolactone (GVL) by Brønsted and Lewis acid zeolite catalysts has been reported. Up to 80% GVL yield was achieved through Meerwein-Ponndorf-Verley (MPV) reduction of furfural to FA, followed by hydrolysis of FA to LA and then selective MPV hydrogenation of LA to GVL (122, 123).

7.3 Polysaccharides and Monosaccharides to Levulinic Acid, Ethyl Levulinate and γ-Valerolactone

Acid catalysed dehydration of hexoses leads to the formation of HMF, which can undergo rehydration in acidic medium to give LA and formic acid (Scheme V). LA is a particularly attractive platform molecule, because it can be converted into valuable chemicals and advanced biofuels (124). For example, valeric biofuels which are compatible with current transportation fuels are accessible from LA (125, 126). If dehydration of hexoses is carried out in ethanol, then esterification of LA with ethanol leads to ethyl levulinate (EL) (Scheme V). EL is an acceptable diluent for biodiesel fuels with high saturated fatty acid content (127).

GFBiochemicals’ Caserta plant in Milan, Italy, currently produces LA and its coproduct FA from non-edible biomass feedstock (wood, grass, wheat straw and cellulose) via an acid catalysed process (ATLAS Technology^TM). This company also has plans to expand its production portfolio to LA derivatives, such as LA-esters and LA-ketals, GVL, MTHF and methylbutanediol (MeBDO) (128). Hydrogenation of LA in the presence of a metal catalyst with an external H₂ source or formic acid as the source of hydrogen gives GVL. GVL is a promising bio-derived molecule and it is attractive for its physical and chemical properties. It has low toxicity, an acceptable and definitive smell which makes detection of leaks and spills easy, it is safe to store and move globally in large quantities, has high flash (96°C) and boiling (207°C) points as well as a low melting point (–31°C). It is also used as an additive in the food industry and as a green solvent (129–133). Bond et al. have proposed a pathway to renewable butene, using GVL, which can be oligomerised to long chain alkenes for liquid fuels and commodity chemicals production. Using formic acid as the source of hydrogen in the hydrogenation of LA is more economical because dehydration of hexoses produces formic acid in equimolar amounts to LA. In the presence of a catalyst, formic acid decomposes to carbon dioxide and H₂ and hence can be used as an internal source of hydrogen (Scheme VII) (131).

Quaker Oats, USA, produces GVL on a commercial scale from LA using chromium(III) oxide (Cr₂O₃) and copper(II) oxide (CuO) at 200°C (134). Metal catalysts, such as Pd, Ir, Ni, Cu, Pt, Re, Rh, Au and Ru have been used in the presence of H₂ to produce GVL from LA (135–138). Manzer studied this reaction using Ir, Rh, Ru, Pd, Rh, Pt and Re supported on carbon, and found that 5% Ir, Pd and Rh were active in the conversion of LA to GVL but their selectivity to GVL was rather low. Ni had the lowest activity and selectivity under the same conditions and Ru had the best activity and selectivity (Figure 6) (135).

While heterogeneous catalysts have dominated LA to GVL reactions, (135–141) homogenous catalysts have also been studied. Particularly, water soluble homogeneous catalysts have been attractive because GVL does not form an azeotropic mixture with water, therefore the catalyst can be recovered by distillation and reused (129–131, 142). Hydrogenation of LA to produce GVL has been examined in aqueous solution using various water soluble phosphine ligands in combination with the metal precursors Ru(acac)₃ and RuCl₃•x H₂O (Table III) (143–147). GVL was obtained in 97% yield within 5 h, using a Ru(acac)₃/TPPTS (TPPTS = p (C₆H₄-m -SO₃Na)₃) catalytic system (144, 147).

Formic acid was used as a hydrogen source in [(η⁶-C₆Me₆)Ru(bpy)(H₂O)]SO₄ catalysed transformation of LA, in water, to give in GVL and 1,4-PDO. In this same study, [Ru(acac)₃]/PBu₃/NH₄PF₆ afforded 100% GVL from LA (142). Mika et al. demonstrated that LA could be reduced effectively to GVL in the presence of an in situ catalyst generated from Ru(acac)₃ and different sulfonated phosphines with the general formula R_n P(C₆H₄-m -SO₃Na)_3–n (n = 1, 2; R = Me, ⁿ Pr, ⁱ Pr, ⁿ Bu, Cp) without the need for a base. The ⁿ BuP(C₆H₄-m- SO₃Na)₂ showed the highest activity at 99% GVL. The catalyst was successfully recycled for six consecutive runs without loss of activity (146).

Ir trihydride catalysts bearing a PNP pincer ligand showed extremely high activity in LA to GVL conversion giving up to 98% GVL yield. Recovery of the catalyst was demonstrated. This was done by removal of the product by vacuum and the remaining catalyst was redissolved in LA and exhibited similar activity to before distillation (149). Conversion of LA to GVL was studied using p -cymene Ru(II) N -heterocyclic carbene (NHC) complexes as (pre)catalyst in water. The complex (Table III) was reduced in situ to form highly catalytically active Ru nanoparticles which deactivated in subsequent catalytic runs due to aggregation (150).

Horváth and coworkers showed that LA and a small excess of formic acid can be converted to GVL in the presence of Shvo’s catalysts. At 100°C, the catalyst rearranged to form a Ru hydrido catalytically active species (Table III) and yields higher than 99% after 8 h were recorded. The formation of 1,4-PDO and MTHF, the typical side products of the reduction of GVL with molecular hydrogen, was not observed (131).

Solvent free conversion of LA to GVL has been achieved with pyrazolyl-phosphite and pyrazolyl-phosphinite Ru(II) complexes as (pre)catalysts, using both molecular hydrogen and formic acid as hydrogen sources. With H₂ (15 bar) 100% GVL yield obtained at 110°C, while with formic acid at 100°C, 100% GVL was obtained. In order to understand whether formic acid acts as a hydrogen transfer reagent or a source of H₂ through prior decomposition, in situ NMR studies, where the reaction was monitored in a J Young nuclear magnetic resonance (NMR) tube containing the (pre)catalyst, KOH and formic acid at 120°C, were carried out. Within 30 min of the reaction, hydrogen gas was observed by proton (¹H)-NMR spectroscopy with a H₂ signal at 4.56 ppm. The active species was found to be a cocktail of nanoparticles and molecular catalysts as evidenced by mercury poisoning tests. The catalyst could be recycled and reused up to three times (151).

8.1 Polysaccharides and Monosaccharides to Sugar Alcohols and Diols

Hydrogenation of cellulose, oligosaccharides, disaccharides and monosaccharides results in the formation of sugar alcohols or polyols such as xylitol, sorbitol and mannitol which are important chemicals widely used as monomers in the plastics industry. Sugar alcohols also have applications in the food industry in the substitution of sugars such as glucose, sucrose and fructose. Xylitol is most frequently used in place of sugar, as it has similar sweetness and lower energy content than sucrose. From sugar alcohols it is also possible to produce renewable hydrogen and hydrocarbons which will be discussed in Section 8.2 of this review (152–161).

The catalytic conversion of cellulose to sugar alcohols is a two-step process, which includes hydrolysis of cellulose to monosaccharides, followed by the hydrogenation of the sugars to sugar alcohols. Sels and coworkers used water soluble HPAs and a Ru/C catalyst to transform cellulose to sugar alcohols. At 190°C and (95 bar H₂), the conversion of cellulose was 100% with 85% yield (152). Palkovits et al. have also demonstrated that the transformation of cellulose to sugar alcohols was possible using a combination of HPAs and Ru/C. Sugar alcohols were obtained in a yield of 81% (153). Liu et al. have also demonstrated cellulose to sugar alcohols conversion with Ru/C at 245°C and 60 bar H₂. They obtained 40% sugar alcohols of which 30% was sorbitol and 10% mannitol (154). It is striking to see that in the presence of HPAs, which are acidic, catalytic conversion of cellulose to sugar alcohols on Ru/C proceeds to give higher yields at comparably lower temperatures than when no HPAs are added.

Hydrolytic hydrogenation of cellulose was shown by Fukuoka et al. The reaction gave 58% sorbitol yield at 190°C in water over a Ru/AC catalyst (155). Tsubaki et al. have presented a catalytic system that consists of Pt/RGO (RGO = reduced graphene oxide) for the hydrogenation of cellulose and cellobiose to sorbitol. Sorbitol yield was 91.5% from cellobiose and 58.9% from cellulose (156).

Ru and Os complexes have been used, as homogeneous catalysts, in conversion of fructose (to glycerol, propylene glycol (PG), ethylene glycol (EG) and sugar alcohols) (157, 158), mannose and inulin (to sugar alcohols) (159) and glucose (to sorbitol) (160). Recently, Wang and coworkers reported the first example of a highly selective homogenous Ru(II)/H₂SO₄ catalytic system which promoted the conversion of cellobiose and cellulose (at 100°C and 50 bar H₂ over 16 h) to sorbitol, 1,4-sorbitan, glucose and a negligible amount of HMF (0.1%). Starting from cellobiose, 91.4% sorbitol, 3.1% 1,4-sorbitan and 5.3% glucose were afforded at 100% conversion, whilst from cellulose 21.9% sorbitol, 7.6% 1,4-sorbitan and 8.4% glucose were produced at 40% conversion. Sorbitol can be further converted to sorbitan and isosorbide; thus suggesting that the sorbitan in the above study may have been produced from further reaction of the sorbitol. The authors further proposed a mechanism for the reaction (Scheme VIII) which involved in situ generation of the catalytically active Rh–H species 1 which inserted into the C=O bond of glucose (to form 2) of the initially hydrolysed cellulose. This is then followed by a metal-ligand assisted proton transfer to the glucose O¹ and C¹ 3 coupled with sorbitol expulsion to regenerate the (pre)catalyst (161).

EG, 1,2-PG, 1,3-PG and glycerol are the simplest sugar alcohols and are used as emulsifiers, dehumidifying agents, anti-freeze agents, lubricants, solvents, polymer monomers and as pharmaceutical intermediates. Diols are currently produced from fossil derived feedstock (petroleum) by hydration of propylene and ethylene oxide (162). In 2008, Chen and coworkers reported the direct conversion of cellulose to EG in a yield of 61% using a nickel-tungsten carbide catalyst (163). Since then, studies aimed at understanding the reaction mechanism of cellulose to EG and 1,2-PG have been conducted (164–167). The major mechanistic pathway involves hydrolysis of cellulose (to glucose and oligosaccharides) → retro-aldol condensation (of glucose to form glycol aldehyde and erythrose) → hydrogenation of glycol aldehyde to EG. If glucose isomerises to fructose, then 1,2-PG is formed (Scheme IX) (165, 166).

Catalysts such as Ru/AC/H₂WO₄ have been employed in cellulose conversion to EG (168). Liu and coworkers reported efficient conversion of cellulose into EG, 1,2-PG and sorbitol, as dominant products on Ru/C in the presence of tungsten trioxide (WO₃). It was found that WO₃ crystallites catalysed hydrolysis of cellulose as well as selective C–C bond cleavage, while Ru/C was responsible for hydrogenation (169).

EG and 1,2-PG production from sugar alcohols has also been reported. The most studied are sorbitol and xylitol while Ni- and Ru-based catalysts are most used. For example, Ru nanofibres, Ru/C, Ni-Re and Ni/Al₂O₃ in the presence of bases such as calcium hydroxide (Ca(OH)₂), calcium oxide (CaO), KOH and sodium methoxide (NaOMe) catalyse sugar alcohols to EG and 1,2-PG (170–173).

The Changchun Dacheng Group, China, has developed a process that produces 10,000 tonnes of 2,3-butandiol (2,3-BD) annually. This process integrates bio and catalytic conversion by firstly enzymatically (by glucoamylase) hydrolysing starch to glucose. Then a catalyst is employed in conversion of the glucose to sorbitol, which is then catalytically cracked into a mixture of C₂ to C₄ diols and polyols. After purification steps, 97% purity 1,2-PG, EG and 2,3-BD are obtained, in which the yield of 2,3-BD is 5% (167).

8.2 Renewable Hydrogen and Hydrocarbons from Polysaccharides, Disaccharides and Monosaccharides

Cellulose, oligosaccharide, disaccharide or monosaccharide and sugar alcohols have been transformed to hydrogen and hydrocarbons (alkanes). This work was pioneered by Dumesic (26, 174–177). Aqueous phase dehydration/hydrogenation (APD/H) produces alkanes and aqueous phase reforming (APR) leads to hydrogen and gaseous alkanes (primarily methane) (26, 174–177). Cortright and coworkers have demonstrated that hydrogen can be produced from sugars and sugar alcohols in an aqueous phase reforming process using a Pt-based catalyst. Glucose was converted to hydrogen and gaseous alkanes, with a hydrogen yield of 50% (174). Later, they used Pt/SiO₂–Al₂O₃, Pd/SiO₂–Al₂O₃ and Pt/Al₂O₃ to convert sorbitol to hexane. Sorbitol is transformed on these bifunctional catalysts by a pathway that involves the formation of hydrogen and CO₂ on the appropriate metal catalyst (Pt or Pd) and the dehydration of sorbitol on a solid acid catalyst (SiO₂ or Al₂O₃). This is followed by hydrogenation of the dehydrated intermediates (for example isosorbide or enolic species) by hydrogen produced in the water gas shift reaction on Pt or Pd, which leads to the overall conversion of sorbitol to alkanes plus CO₂ and water (175).

Beller has recently described a procedure for hydrogen generation from biomass (such as glucose, fructose, cellobiose, cellulose and ligncellulose) using Ir and Ru pincer-complexes as (pre)catalysts. Hydrogen was produced using ppm amounts of (pre)catalyst at a turnover frequency (TOF) of up to 10,269 h^–1 (from cellobiose in 1 h) and a turnover number (TON) of up to 6125 (from cellulose in 1 h) (176).

Dumesic proposed that C₇ to C₁₅ alkanes (which have a molecular weight appropriate for transportation fuel components) can be produced from carbohydrates by acid catalysed dehydration, followed by aldol condensation (over solid base catalysts) to form large organic compounds. These molecules may be converted into alkanes by HDO over bifunctional catalysts (Mg-Al oxide and Pd/Al₂O₃) (177). The total HDO of cellulosic biomass including biomass derived substrates such as glucose, xylose, sugar alcohols, LA, HMF and furfural to produce C₅ to C₁₅ alkanes has recently been reviewed (178).

9.1 Polysaccharides, Disaccharides and Monosaccharides to Levulinic Acid, Formic Acid, Acetic Acid and Glycolic Acid

Cellulose and glucose derived carbohydrates can be transformed into a range of organic acids (Scheme X) (89–91, 179).

The formation of LA from cellulose in the presence of mineral acids or Lewis acids has been extensively studied. This multistep process involves hydrolysis of cellulose to glucose ↔ fructose → dehydration of fructose to HMF → and rehydration of HMF to LA and formic acid (Section 7.3) (180–182). Acetic acid can be obtained by hydrothermal oxidation of glucose or cellulose in the presence of a base, such as NaOH and Ca(OH)₂ (183, 184).

Efficient, VO²⁺, Pb²⁺ and Ni²⁺ catalysed conversion of glucose, cellulose and lignocellulose into lactic acid, which is a high value chemical used for the production of fine chemicals and biodegradable plastics, has been achieved. Lactic acid yields ranging from 25% to 60% were obtained at high temperatures (185–187). This topic has been reviewed (188).

Another approach reported by Jin et al. is the conversion of glucose or cellulose to HMF and lactic acid, then oxidation of the lactic acid and HMF to give acetic acid (189). Glycolic acid production from cellulose has been less studied, nonetheless Keggin-type Mo polyoxometalates (POMs) have been successfully employed in the conversion of cellulose to glycolic acid (190).

Apart from being a coproduct of HMF hydration, formic acid can be produced from cellulose via oxidation. Formic acid is a starting material for the production of formate esters, which can be utilised in the production of a large variety of useful organic derivatives such as aldehydes, ketones, carboxylic acids and amides. It has been used in the rubber, agricultural and pharmaceutical industries. It has applications as a mordant, auxiliary agent in the dyeing industry, a disinfectant and its formate salt is a useful de-icing agent (191). Notably, formic acid has energy content at least five times higher than that of commercially available lithium ion batteries and therefore represents a convenient hydrogen carrier in fuel cells making it a highly exploitable chemical on the hydrogen energy storage front (192).

Currently formic acid is produced from fossil derived feedstock. The production of formic acid from renewable resources such as cellulose is desirable for the advancement of green chemistry. Formic acid has been produced from glucose and cellulose using alkali metal bases (193), vanadium substituted Keggin-type POM (H₅PV₂Mo₁₀O₄₀) (194) and ferric sulfate (Fe₂(SO₄)₃) (194).

9.2 Polysaccharides, Disaccharides and Monosaccharides to 2,5-Furandicarboxylic Acid, Gluconic Acid and Glucaric Acid

The oxidation of HMF can generate 2,5-furandicarboxylic acid (FDCA) (Scheme X), which is an important platform molecule, which has been identified as a potential alternative to petroleum-based terephthalic acid for the production of resins and polymers (195, 196). FDCA is produced by a two-step process from cellulose or sugars (197). HMF is first prepared from C₆ sugars or cellulose (59–62), followed by HMF oxidation to FDCA with oxidants (198), metal catalysts (199–204) or enzymes (205).

Recently, Avantium in The Netherlands and BASF, Germany, have announced that they are jointly preparing for the construction of the world’s first commercial plant for FDCA production from sugars. The Avantium process transforms fructose to HMF methyl ether, which is then oxidised (by [Co(OAc)] or [Mn(OAc)] and NaBr) to FDCA. The FDCA (replacement for petroleum-based terephthalic acid) will be combined with EG to produce polyethylenefuranoate (PEF), a next generation polyester. PEF is 100% bio-based and has a high market potential as a future packaging and film material (206).

Gluconic and glucaric acid are important intermediates and are used in the pharmaceutical, detergent and food industries (207). Currently, gluconic and glucaric acid are mainly produced by the enzymatic oxidation of glucose. Research efforts have been devoted to developing catalytic oxidation of glucose to gluconic acid (208–211). For instance Au/TiO₂ is an effective glucose oxidation catalyst and in the aqueous phase (using O₂ oxidant) at 40–60°C, 100% conversion of glucose and 100% selectivity of gluconic acid was recorded in 4 h (212).

Comotti et al. used mono and bimetallic catalysts (Rh, Pd, Pt and Au) supported on carbon to effect the transformation of glucose to gluconic acid in water. The combination of Au-Pd resulted in 100% selectivity (213). The direct conversion of cellulose to gluconic acid is more necessary, however few reports of the route are known. The conversion of cellobiose and cellulose to gluconic acid over POM-supported Au nanoparticles (Au/Cs_2.2H_0.8PW₁₂O₄₀) was reported. Cs_2.2H_0.8PW₁₂O₄₀ without Au exhibited a low activity for the conversion (25%) of cellobiose and the main product was glucose. Upon loading of Au particles, the conversion of cellobiose increased to 96% and gluconic acid was the major product. This meant that the acidic sites (Cs_2.2H_0.8PW₁₂O₄₀) catalysed the hydrolysis of cellobiose and cellulose, and Au nanoparticles are responsible for the oxidation of glucose to gluconic acid (214). Cellobiose conversion to gluconic acid has also been demonstrated using sulfonated carbon supported Pt (Pt/C–SO₃H) where 47% yield was obtained (215).

There are some studies on the oxidation of glucose to glucaric acid. A Pt/SiO₂ (4 wt%) catalyst afforded 66% yield of glucaric acid at 90°C under 5 bar of O₂, while at 119°C under 280 bar of O₂ over a Pt–Au/TiO₂ (4_Pt–4_Au wt%) catalyst a 71% yield of glucaric acid was recorded (216, 217).

Rennovia, USA, has developed a two-step catalytic process for producing adipic acid from glucose. The process starts by a selective catalytic oxidation of glucose to glucaric acid, followed by a selective catalytic HDO of glucaric acid to adipic acid (Scheme XI) (216–218). The Rennovia renewable adipic acid process offers potential for significant commercial and environmental advantages compared to the current petrochemical process for the production of adipic acid (218).