Heterogeneously Catalyzed Hydrothermal Processing of C5−C6 Sugars

Biomass has been long exploited as an anthropogenic energy source; however, the 21st century challenges of energy security and climate change are driving resurgence in its utilization both as a renewable alternative to fossil fuels and as a sustainable carbon feedstock for chemicals production. Deconstruction of cellulose and hemicellulose carbohydrate polymers into their constituent C5 and C6 sugars, and subsequent heterogeneously catalyzed transformations, offer the promise of unlocking diverse oxygenates such as furfural, 5-hydroxymethylfurfural, xylitol, sorbitol, mannitol, and gluconic acid as biorefinery platform chemicals. Here, we review recent advances in the design and development of catalysts and processes for C5−C6 sugar reforming into chemical intermediates and products, and highlight the challenges of aqueous phase operation and catalyst evaluation, in addition to process considerations such as solvent and reactor selection.


Introduction
The quest for sustainable resources to meet the energy, food, and materials nexus of needs for a rising global population, set against the backdrop of climate change and dwindling ecodiversity, necessitates new chemical technologies. Global energy demand is expected to rise 2% per annum, with 2050 energy consumption predicted to be twice that of 2001, with associated CO 2 emissions increasing from 6.6 to 11.0 GtC yr −1 . 1 The Copenhagen Accord stated that greenhouse gas concentrations in the atmosphere should be stabilized at a level that would prevent dangerous anthropogenic interference with the climate system, with the ensuing scientific consensus that the global temperature rise by the end of the 21st century should be <2°C relative to preindustrial levels. 2 In 2015, the Paris COP21 UN conference on climate change culminated in 195 countries adopting the first universal commitment to cut greenhouse gases, affirming the prior aspirational goal to limit warming <2°C and indeed strive to keep global temperature rises <1.5°C, heralding what some considered the dawn of the postfossil fuel era. 3 Realizing these ambitious targets will require a transition from traditional fossil fuel resources to renewables, notably nuclear, solar, wind, and wave power, and biomass. 4 Biomass is a generic term describing biogenic organic matter, typically formed through photosynthesis, with the potential to replace fossil feedstocks for the production of biobased transportation fuels, heat, power, and biomaterials, in addition to underpinning new chemical platforms ( Figure  1). 5−8 Recent decades have witnessed an inauspicious beginning to global biomass utilization, with large-scale diversion of edible plant crops toward bioethanol and first-generation biodiesel manufacture impacting upon food supply and prices. The ensuing "food versus fuel" debate highlighted the importance of identifying inedible and waste biomass feedstocks, in tandem with the biorefinery concept for the coproduction of high volume/low value fuels and low volume/high value chemicals. In 2004, The U.S. Department of Energy (DOE) identified a range of such sugar-derived platform chemicals accessible through the chemical or biochemical transformation of lignocellulosic biomass (Figure 2) that would be potential target products from biorefineries, several of which are featured in this Review.
Lignocellulose is the most abundant biomass source inedible to humans, prevalent in plant walls, and comprises a mixture of cellulose and hemicellulose carbohydrate polymers embedded in a lignin matrix. 9,10 Lignin is a three-dimensional network of polyaromatic alcohols, whose catalytic upgrading is now viewed as an important source of renewable value added chemicals. 11 Hemicellulose comprises pentose (xylose predominantly) and hexose units connected by different glycosidic bonds. Cellulose is a water insoluble, linear polysaccharide formed from glucose units linked via β-1,4-glycosidic bonds ( Figure 3). 12 Lignocellulose from different sources, regions, and seasons exhibits significant variations in the proportions of the aforementioned components; however, the combined hemicellulose and cellulose content generally exceeds 60 wt %. 13 Popular approaches to lignocellulosic biomass conversion to fuels and chemicals, via noncatalytic thermochemical and enzymatic biochemical platforms, including gasification 14−19 to produce syngas (CO/H 2 ) or H 2 , fast pyrolysis 20−24 or liquefaction 25−27 to produce bio-oils, or sugar fermentation to bioethanol, 28−32 butanol, 33−35 or other high value chemicals, 36−38 have been comprehensively reviewed. Thermochemical methods such as gasification are rather crude, unselective routes to the production of platform chemicals from highly functional biomass components, and syngas production from high-quality biomass as a route to subsequent hydrocarbon (e.g., Fischer−Tropsch) synthesis cannot therefore be rationalized unless the biomass feedstock is particularly recalcitrant and unsuitable for processing via more economic methods. Biomass fast pyrolysis offers a potential route to drop-in liquid hydrocarbon fuels, but is limited by the poor quality of the resulting bio-oils (high oxygen content, low heating value of typically 16−19 MJ kg −1 , i.e., one-half that of petroleum-   derived fuels, corrosive nature, high viscosity, immiscibility with conventional fuels, and poor chemical stability), which hence require costly and energy intensive upgrading treatments prior to use as a biofuel. Fast pyrolysis also requires predried biomass feeds, compromising the potential economic advantage of important wet feedstocks such as algae. 39,40 Hydrothermal liquification is a moderate temperature/high pressure process to depolymerize solid biomass and yields higher energy density (5−20 wt % oxygen content) bio-oils and chemicals, but is also unselective and requires high capital investment in robust reactor construction materials and energy intensive operation. 25 Biocatalytic processes generally offer superior activity and selectivity than chemical catalysts; however, while widely investigated in the context of biorefineries 41 for bioalcohol production, 42 processes become less viable when the target products are involatile and/or in low concentration, making their isolation economically prohibitive. 43 C 5 −C 6 sugar monomers derived from nonfood (hemi)cellulose have risen in prominence as key building blocks of biofuels and platform chemicals for either drop-in or new applications. However, obtaining such sugars from lignocellulosic biomass for chemical (or biochemical) processes is particularly challenging due to the difficulty of separating (hemi)cellulose from lignin, and hemicellulose from cellulose in turn. Hence, while enzymatic transformation of sugars to chemicals and fuels is an attractive prospect, this requires extensive processing of raw biomass. 31,44 Fractionation to isolate cellulose, lignin, and hemicellulose is typically performed via acid or base hydrolysis, steam explosion, 45 or organosolv treatments 46 to separate the polysaccharide from lignin. 47 Biomass pretreatment is an area of intensive research wherein greener methods are sought, and is expected to play an increasingly important role in biorefinery viability. 48 Such greener strategies have been the subject of reviews spanning supercritical fluids 49 (with CO 2 showing promise as both solvent and acid to hydrolyze biomass in subcritical pressurized water 50 ) to ionic liquids (ILs) 51−56 or deep eutectic solvents (DESs) whose properties can be tuned to dissolve lignin or cellulose selectively. 57−60 Once separated, cellulose fractions are typically hydrolyzed to fermentable sugars for further processing into fuels and/or chemicals via enzymatic, 61 chemical (acid or base), 62,63 supercritical water, 64 or more recently IL-based treatments. 65 ILs and DESs have similar properties, and are based on salts with poorly coordinated ions present in the liquid phase at normal operating temperatures (typically <150°C), but differ in the properties of their ionic components. ILs comprise a single discrete anion and cation pair, often based upon tetraalkylammonium, 1,3-dialkylimidazolium, tetraalkylphosphonium, and alkylpyridinium cations with a range of anions. ILs have essentially no vapor pressure at chemically relevant temperatures, and are thermally stable as liquids at temperatures exceeding 300°C. In contrast, DESs are mixtures of various anionic and/or cationic species formed when Lewis or Brønsted acids and bases are mixed. 59 The combination of ILs and CO 2 in the presence of a solid acid catalyst has been reported for phenolics extraction from lignocellulose. 66 Despite the advantages of ILs as reaction media for biomass conversion, their industrial uptake has been limited for several reasons: (i) the high cost of some ILs; (ii) energy intensive product separation and purification from ILs; (iii) lack of data regarding their toxicity and biodegradability; and (iv) IL contamination by oligomeric species and humins produced as byproducts during biomass processing, 67 which render these processes economically unfavorable on a commercial scale. Low-cost ILs have been developed (≤$2.50 kg −1 ), which enable competition with conventional solvents. 55 DESs are promising alternatives to ILs, which allow the use of benign Lewis acids and/or bases of general formula Cat + X − zY, where Cat + is cationic ammonium, phosphonium, or sulfonium species and X − is a Lewis base (typically a halide anion). DESs have been reviewed in detail, 59,60 and in the context of biomass fractionation DESs based on choline chloride with urea, malonic acid, lactic, malic, or oxalic acid 68 offer process advantages similar to those of ILs, while mitigating toxicity of the cationic (e.g., imidazolium, pyridinium, or pyrrolidinium) components. 69−71 Despite progress in this area of "green solvents", much research is required for their large-scale application to biomass fractionation, specifically to minimize solvent use, improve energy/atom economies, and minimize wastewater production during acid/ base neutralization. 72,73 Water recovery and reuse remains a challenge in biorefineries, and the cost of removing trace organic contaminants may be prohibitive, 74 with product separation hindered by physical and chemical barriers such as polarity and azeotropy. Imidazolium-based ILs have been advanced as a means to separate target components from biorefinery streams, 75 with examples spanning the extractive separation of ethanol−water mixtures or the removal of ethanol, butanol, or lactic acid from fermentation broths. Tuning ILs polarity and thermal properties by altering the cation−anion combination 76 makes them attractive alternatives to volatile organic solvents for separation/extraction processes. 77−79 Imidazolium-based ILs can aid the separation and dehydration of ethanol from azeotropic water mixtures by enhancing the relative volatility of ethanol; 80 4 ]. 81 ILs such as trihexyltetradecylphosphonium dicyanamide (P 666,14 [N(CN) 2 ]) are also promising for aromatics extraction from aqueous sugar solutions obtained from biomass pyrolysis, permitting their subsequent fermentation or catalytic upgrading. ILs have also shown particular promise as a medium for biomass conversion, with the catalytic transformation of glucose and fructose to 5-HMF receiving particular attention. 82,83 Soluble Lewis acid catalysts such as SnCl 4 are particularly effective for glucose conversion into 5-HMF in 1-ethyl-3-methylimidazolium tetrafluoroborate ([EMIM]BF 4 ).
[EMIM]BF 4 /SnCl 4 alone is also effective for the conversion of fructose, sucrose, inulin, cellobiose, and even starch to 5-HMF, demonstrating the generality of ILs as reactive solvents. 84 The IL catalytic performance can also be fine-tuned by varying the alkyl chain length of the cation, as observed in the direct conversion of glucose to 5-HMF in dialkylimidazolium chlorides with germanium(IV) chloride, where 5-HMF yields decreased from 38.4% to 23.3% when the alkyl group changed from butyl to decyl. 85 Mechanocatalysis offers an intriguing opportunity for solvent-free biomass processing. 86,87 For example, H 2 SO 4 impregnated lignocellulosic substrates offer a clean and efficient route to one-pot biomass depolymerization, in which mechanical forces and mineral acids convert dry lignocellulose into water-soluble products comprising oligosaccharides and lignin fragments. 86,88 Heterogeneous analogues, employing clays and simple metal oxides, have also been exploited in mechanocatalysis of cellulose to glucose, in addition to levoglucosan, levoglucosenone, and 5-hydroxymethyfurfural (5-HMF). 87 Catalysis underpins the development of sustainable technologies for the green chemical transformation of sugars and their reforming, 6,9,89 with biocatalysis, homogeneous catalysis, and heterogeneous catalysis each presenting theoretical, technical, and economic challenges for biomass utilization. Biocatalysis has delivered several successful industrial processes for bioderived chemicals production, generally in the food sector. Notable examples include xylitol (an artificial sweetener) from xylose using microorganisms, 90 and glucose isomerization to fructose using glucose isomerase. 91 It is also advantageous in applications where a societal desire for sustainable consumer products permits less economical routes to the manufacture of renewable chemicals for biopolymers based upon, for example, 1,3-propanediol, polylactic acid, or polyhydroxyalkanoate. 92 Bioprocesses to produce chemicals such as succinic acid, 1,4-butanediol, and isoprene are also being commercialized; 92 however, the market for these drop-in products remains volatile because they are competing against existing petrochemical routes whose cost is linked to (currently low) oil prices. 93 Challenges facing large-scale biocatalytic biomass processing include 94−96 (i) purification of low-cost waste biomass-derived feed streams, which may otherwise compromise enzyme activity; (ii) the necessity for buffered solutions to maintain an optimal environment, for example, pH 7−8; (iii) low operating temperatures to protect enzyme lifetime and maximize product selectivity, which afford lower activity; and (iv) high operational costs from frequent enzyme replacement due to irreversible deactivation.
Heterogeneous catalysts offer opportunities to selectively transform a wide range of biomass-derived sugar-containing feed streams into chemical and fuels over a wide operational regime (typically 25−300°C and 1−80 bar) and with a good tolerance to impurities and consequent long lifetimes. With respect to aqueous phase processing of waste-derived C 5 −C 6 sugars from cellulosic sources, 97 heterogeneous catalysts may complement biocatalysts. 98 Accessing key platform chemicals from xylose and glucose (the primary C 5 −C 6 sugars obtained from lignocellulose) requires a range of transformations, 99 including isomerization to xylulose/lyxose or fructose/ mannose, hydrogenation to xylitol or sorbitol, dehydration to furanic species such as furfural and 5-HMF, and selective oxidation to xylonic acid and gluconic acid, all of which may be heterogeneously catalyzed. 94,100,101 Catalytic approaches have also been adopted to upgrade the preceding sugar-derived chemicals through, for example, hydrodeoxygenation (HDO) or deoxydehydration. 23 Although a number of the preceding reactions may be catalyzed by soluble acids/bases, 102−105 such approaches suffer the same drawbacks common to homogeneous catalysis, difficult product separation and catalyst recovery/reuse and accompanying high volumes of waste, problematic storage/handling of hazardous materials, and process issues related to scale-up and continuous operation. Solid catalysts circumvent many of the preceding limitations, facilitating (i) low-cost recovery and recycling through centrifugation, filtration, or fixed-bed operation (rendering them suited to large-scale continuous industrial process units); (ii) robust performance under challenging reaction environments; (iii) safe storage/handling/operation within reactors constructed from conventional materials, extending the lifetime of operation units; and (iv) tunable physical and chemical properties allowing tailoring to individual reactions. 106−108

Scope of the Current Review
Despite a plethora of original research on heterogeneous catalysts for the transformation of (hemi)cellulose-derived C 5 − C 6 sugars into oxygenates, aliphatic, and cyclic compounds over the past five years, there have been no comprehensive reviews encompassing the range of associated acid, base, and metal catalysis reported. This Review focuses primarily on the aqueous phase reforming of such sugars, highlighting promising heterogeneously catalyzed isomerization, dehydration, hydrogenation/hydrogenolysis, and selective oxidation pathways, with a view to identifying structure−function relationships, and exploring the role of active phase and (where applicable) catalyst support in controlling overall activity, selectivity, and reusability. Throughout, we will try to unify diverse performance indicators (e.g., conversion, selectivity, yield, turnover frequency, and initial reaction rate) to permit quantitative benchmarking of disparate systems, and to assess the reliability of literature data in the light of analytical methods employed and the extent to which, for example, mass balances and reuse tests have been utilized. The impact of process engineering (e.g., reactor design, solvent effects, and cost issues) on catalysis and scale-up is also discussed, and emergent areas for aqueous phase sugar reforming identified. Indicative cost estimates for earth scarce catalyst components (e.g., platinum group metals) provide additional metrics for readers to assess the utility of competing catalyst technologies.
We hope that this Review will provide a useful resource for guiding academic and industrial researchers toward informed judgments regarding the selection of the most appropriate technology for platform chemicals production from biomass, and in turn encourage greater awareness of biobased trans-formations and communication between catalytic scientists and biotechnologists. The overall framework of this Review is summarized in Scheme 1.

C 5 −C 6 SUGAR TRANSFORMATIONS
Prior to a detailed account of catalytic systems developed for the aqueous reforming of C 5 −C 6 sugars, we introduced the most common reaction classes (isomerization, dehydration, hydrogenation/hydrogenolysis, and selective oxidation) and their interrelations.

Isomerization
Xylose, mainly derived from xylan hemicelluloses by acid/base hydrolysis, is a five-carbon sugar with lower calorific value than glucose. It can exist as a pyranose ring, a furanose ring, or an open-chain structure in water. 109 Glucose also exhibits three conformers in water: a pyranose ring, a furanose ring, and an open-chain (aldohexose) structure. 96 Although most popularly known for their use as sweeteners, xylose/xylulose and glucose/ fructose are important chemical intermediates; the higher reactivity of xylulose and fructose as compared to xylose and glucose rendering the former more easily valorized, and hence targets for the heterogeneously catalyzed isomerization of the latter. The overall reaction network is more complicated than first appears due to side reactions such as the epimerization of xylose to lyxose and of glucose to mannose, the retroaldolization of fructose to glyceraldehyde and dihydroxyacetone, and glucose conversion to glycolaldehyde and erythrose, as illustrated in Figure 4. 96,109,110 Furthermore, all four sugar reactants (and their intermediates) can irreversibly react to form solid humins or other degradation products, which are intractable for valorization. The maximum reported xylulose yield from xylose isomerization is therefore 27%, with a corresponding lyxose yield of 11%, at a xylose conversion of 60% (albeit after only 7 min at 110°C). 109 For glucose isomerization, kinetic studies reveal that the energy barrier from glucose to fructose is 17% lower than that to mannose, and hence fructose is the dominant isomeric product under appropriate kinetic control. 96

Dehydration
Fructose dehydration yields 5-HMF, which features in the top 10 list of valuable biobased chemicals according to the U.S. DOE, 111 and is a hot topic of both academic and commercial research 112 due to its potential as a versatile platform chemical from which to access 12 important furan derivatives and nonfuranic compounds. 82,113,114 However, 5-HMF production direct from glucose is more challenging than from fructose due to the additional (slow) isomerization that must precede fructose dehydration, and this multistep process often results in poor 5-HMF yields ( Figure 5). 115 The analogous dehydration product from xylose is furfural, itself an important chemical feedstock for furfuryl alcohol, THFA, and pentanediols production. Unfortunately, the typical temperatures required for xylose and fructose dehydration of 120−200°C are significantly higher than those for xylose/glucose isomerization of 100−120°C, 115,116 and consequently xylulose and lyxose formation compete with furfural ( Figure 6). 117 Furfural can subsequently fragment into organic acids, or oligomerize into furilic species (not shown). Complete dehydration of organic acids yields carbonaceous deposits, while xylose and its isomers readily form dark colored solid humins, and such side reactions result in low selectivity toward furfural and significant deficits in carbon mass balance. Humins are common byproducts of higher temperature catalytic sugar transformations, formed notably alongside furfural, formaldehyde, formic acid, and levulinic acid synthesis, and whose undesired production is always favored at higher reaction temperatures, driving the quest for more active catalysts.

Hydrogenation/Hydrogenolysis
Xylitol is a five-carbon artificial sweetener with unusual anticarcinogenic properties whose sweetening capacity exceeds that of sucrose, yet offers lower insulin requirements, and hence is especially suited to diabetics. Demand for xylitol has thus risen within the food, cosmetic, and pharmaceutical sectors. 118−120 Its six-carbon analogue, sorbitol, finds direct application as a sweetener, laxative, and cosmetic thickener, and can be valorized through dehydration to sorbitan and isosorbide, or by hydrogenolysis to other polyols. Sorbitan and isosorbide are used in medical therapies, while their esters are important emulsifiers and food stabilizers; moreover, isosorbide can be further deoxygenated to yield light (<C 6 ) alkanes for the production of second-generation biofuels suitable for the gasoline pool. 121,122 Catalytic routes to both xylitol and sorbitol from bioderived feedstocks are thus of growing interest, with xylose and glucose hydrogenation, respectively, considered promising pathways, although the associated reaction temperatures of 100−140°C promote competing isomerization and dehydration side reactions ( Figure 7). 123,124 Hydrogenolysis offers an alternative route from biomass-derived carbohydrates to a range of sugar alcohols, 101 but occurs through more complex reaction networks than hydrogenation. Glucose hydrogenolysis, for example, features competing hydrogenation to sorbitol, dehydration to 5-HMF, and carbon chain fragmentation and subsequent cascade reactions (Figure 8), and hence necessitates multifunctional catalysts or combinations of different catalyst classes to achieve the desired product; 125,126 the development of selective hydrogenolysis catalyst systems remains a challenge. 127

Selective Oxidation
In the aerobic selective oxidation (selox) of C 5 −C 6 sugars, most reports focus on glucose oxidation to gluconic acid, with other monosaccharides largely neglected. Gluconic acid finds widespread use in the food, detergent, and pharmaceutical industries, and it is produced commercially by fermentation (enzymatic oxidation) of glucose, presenting problematic enzyme separation, wastewater removal, and a narrow window of reaction conditions. 129 There is hence an opportunity for heterogeneously catalyzed routes.
Gluconic and glucaric acid are expected major products of glucose selox; however, glucose possesses both aldehyde and primary and secondary alcohol functions, all of which are oxidizable, resulting in a wide product distribution ( Figure 9). Moreover, as an isomer of glucose, fructose can also be oxidized to 2-keto-D-gluconic acid and D-threo-hexo-2,5-diulose ("5-ketofructose"), although for simplicity this reaction network is omitted. 130−132 Selox of the fructose dehydration product, 5-HMF, also produces 2,5-furandicarboxylic acid, 133 a valuable monomer proposed as a potential renewable replacement for terephtalic acid. Another problem in glucose oxidation is the common requirement for an alkaline medium (pH = 9−10), 134,202 which poses an additional constraint on the catalyst stability. Nevertheless, the oxidation of glucose has the advantage of operating under mild temperatures (40−80°C), and hence   suppressing isomerization and undesired solid humin formation reactions, as compared to dehydration and hydrogenation/ hydrogenolysis pathways.

HETEROGENEOUS CATALYSTS FOR SUGAR
TRANSFORMATIONS The preceding overview of xylose and glucose transformations highlights the complexity of associated reaction networks and hence requirement for both active and selective heterogeneous catalysts to direct reforming of C 5 −C 6 sugars from biomass. Such catalysts are now discussed according to their classification and associated applications.

Solid Acids/Bases for Isomerization
Since the pioneering work of Lobry de Bruyn and Alberda van Ekenstein (LBAE) in 1895 on hexose isomerization in alkaline medium, 135 extensive research has been undertaken to improve the isomerization efficiency in terms of conversion, selectivity, and yield through heterogeneous catalysis. Solid Lewis acids and solid bases have been explored in most detail for the isomerization of C 5 −C 6 sugars (Table 1), exemplified by zeolitic Lewis acids, 136−138 hydrotalcite solid bases, 139 and metallosilicate solid bases. 140 3.1.1. Zeolitic Solid Lewis Acids. Zeolites are well-known crystalline aluminosilicates, with rigid microporous structures and well-defined channels of molecular dimensions. The nanoporous nature of zeolites offers high surface areas, shapeselective reactions, and efficient mass/heat transfer. 141 4 ] − tetrahedra necessitating extraframework cations (e.g., Na + and K + ) to provide electroneutrality. These cations are ion-exchangeable, permitting the introduction of new charge-balancing cations (e.g., H + , Ag + , Ca 2+ , and Ce 3+ ) with attendant modification of the zeolite acid/ electronic properties; for instance, charge compensation with protons imparts strong Brønsted acidity. The zeolite framework may also be modified to incorporate heteroatoms (e.g., Fe, Mn, and Sn) and thereby create new porous structures and impart Lewis acidity. 109,142,143 The heteroatomic Sn-beta zeolite, first synthesized by Corma and co-workers, has successfully been employed as a catalyst for the Meerwein−Ponndorf−Verley (MPV) reduction of carbonyl compounds, 144 and has also found application to xylose isomerization, giving a 27% xylulose yield at 100°C at 60% conversion (entry 1, Table 1). Sn-beta (possessing 0.8 nm pores, entry 2, Table 1) has also been investigated for glucose isomerization in comparison with a related Ti-beta (0.8 nm pores), and mesoporous Sn-MCM-41 and Ti-MCM-41. 145 Snbeta and Ti-beta offered the highest fructose yields of 32% and 14%, respectively, from a 10 wt % glucose solution at 110°C, with the former maintaining good activity even when a 45 wt % glucose solution was employed. The catalytic performance was partly attributed to the unique environment within large pore β-zeolites, with no reaction observed over a TS-1 zeolite (0.5− 0.6 nm pores). Isomerization activity was reduced considerably for Sn-MCM-41, possibly reflecting a lower degree of Sn incorporation into the MCM framework. The active sites were identified as Sn atoms coordinated within the zeolite framework because neither SnCl 4 ·5H 2 O nor SnO 2 alone catalyzed isomerization. Key factors influencing glucose isomerization  are (i) framework or extra-framework active sites; 146 (ii) the solvent (e.g., water or methanol); 147 and (iii) the nature of exchange cations. 136 Davis and co-workers demonstrated that framework tin sites in Sn-beta catalyzed the isomerization of glucose to fructose via a Lewis acid-mediated intramolecular hydride shift in water (Figure 10a), 133 whereas the former epimerization into mannose occurs by a Lewis acid-mediated intramolecular carbon shift. Extra-framework Sn located within the hydrophobic zeolite channels isomerized glucose to fructose in water and methanol through a base-catalyzed H + transfer mechanism. Isotope labeling ( 2 H and 13 C) of glucose highlights that its isomerization products are formed through either hydrogen or carbon shifts, varying with the catalyst and solvent as summarized in Figure 10b. Cation-exchanged zeolites X, Y, and A are also active for glucose isomerization. Ca-and Ba-exchanged A, X, and Y zeolites exhibited low selectivity to fructose, whereas moderate basicity NaX and KX achieved 86% fructose selectivity (entry 4, Table 1), albeit only for glucose conversions <25% and potentially accompanied by cation leaching.
3.1.2. Hydrotalcite Solid Bases. Hydrotalcites (HTs), a subset of layered double hydroxides and found in nature as anionic clays, are important solid base catalysts that exhibit high catalytic activity and robustness in water. 149,150 With a general formula of [M(II) 1−x M(III) x (OH) 2 ] x+ (A x/n n− )·mH 2 O, hydrotalcites comprise brucite-like hydroxide sheets containing octahedrally coordinated M 2+ (e.g., Mg 2+ ) and M 3+ (e.g., Al 3+ ) cations, separated by interlayer A n− charge-balancing anions, where x is the fraction of M 3+ (typically 0.17 < x < 0.34 for aluminum) and m denotes the water of crystallization. HTs exhibit an unusual so-called "memory effect", in which their calcination results in the formation of high area mixed metal oxides, whose subsequent rehydration can restore the original layered double hydroxide structure. HTs can function as strong Lewis bases, with basicity dependent upon the M(II):M(III) ratio within the hydrotalcite sheets. The chemical composition, and textural properties, of HTs can hence be tuned to modify their catalytic performances.
A variety of commercial Mg−Al hydrotalcites, prepared as the hydroxide, carbonate, and mixed carbonate−hydroxide anionic forms, have been explored for glucose isomerization. 151 Hydroxide-containing structures were both more active and somewhat more selective than the carbonate form [entry 5, Table 1]; however, all of these HTs deactivated after only 15% glucose conversion, concomitant with a rapid fall in their initial high (90%) selectivity to fructose. The initial activity and selectivity were claimed to return upon recycling, presumably due to the removal of reversibly adsorbed products during regeneration.
The hydrophobic/hydrophilic nature of commercial Mg−Al HTs was also investigated in batch and continuous aqueous phase glucose isomerization. 110 Hydrophobic materials exhibited lower conversion but superior selectivity toward fructose, exceeding 92% at 30% glucose conversion in 24 h. The main byproducts were from fructose retroaldolization to glyceraldehyde and dihydroxyacetone (and glucose conversion to glycolaldehyde and erythrose) (Figure 4), whose strong adsorption was ascribed to catalyst deactivation. Acidic degradation products such as lactic acid were also responsible for neutralizing the hydrotalcites and magnesium leaching; however, hot filtration tests on these materials suggested that isomerization occurred exclusively via heterogeneous catalysis. Structure−performance relations of Mg−Al hydrotalcites were examined recently for glucose isomerization, employing different Mg:Al molar ratios, textural properties, and morphologies (shown in Figure 11). 152 Catalysts were prepared by coprecipitation, employing a range of pH, Mg:Al ratios, aging temperatures, and solvents to tune their basicity and structure. Isomerization was catalyzed by weak base sites, specifically interlayer anions accessible at the edges of primary particles and defects on the basal planes of the hydrotalcite, with glucose conversion scaling with base site density. Dimensions of coherent crystallographic domains and primary particles, and their agglomeration, are both strong influences on catalytic activity, and may explain the higher activity and selectivity of the porous Mg 0.75 Al 0.25 (OH) 2 (CO 3 ) 0.125 ·0.71H 2 O (HT1-10-RT in Figure 11) [entry 7, Table 1]. These HTs were recyclable without loss of activity or selectivity, although some magnesium leaching due to the presence of lactic acid byproduct was again reported. These catalysts were precipitated by a NaOH/Na 2 CO 3 solution, and hence the possibility of homogeneous catalysis by leached/residual Na + cannot be discounted in this study.
3.1.3. Other Solid Base Catalysts. In comparison with zeolitic Lewis acids and hydrotalcites, the solid bases of metallosilicates and zirconium compounds are less studied for sugar isomerization. Lima and co-workers investigated glucose isomerization in water at 100°C catalyzed by a variety of metallosilicates such as Na 9 Si 12 Ti 5 O 38 (OH), finding fructose yields of between 20% and 40% after only 2 h (entry 9, Table  1). 140 Catalytic performances were comparable or superior to those achievable with a commercial NaX zeolite or NaOH under these reaction conditions, although deactivation occurred after two recycles due to a combination of sodium/potassium leaching, loss of crystallinity, and surface passivation. Zirconium salts such as zirconium carbonate and zirconium phosphate are also active for glucose isomerization, with the carbonate best performing with a maximum conversion of 45% and 76% selectivity to fructose at 120°C, and exhibiting excellent stability over five recycles (entry 10, Table 1). Sulfated zirconia, synthesized in both bulk form 154 and as ultrathin conformal monolayers over mesoporous SBA-15, 155 exhibits modest glucose conversions up to 20%, but high fructose selectivity >85% (with 5-HMF as a significant byproduct of fructose dehydration), with the latter thin film system exhibiting excellent hydrothermal stability.
3.1.4. Process Considerations. Glucose isomerization to fructose is equilibrium-limited; thus innovative reactor designs are required to facilitate fructose removal and thereby drive the forward reaction and also minimize side reactions of fructose, which are problematic in water. 138 Equilibrium-limited reactions, or reactions wherein the catalyst deactivates and requires periodic reactivation, are well-suited to a moving bed reactor design such as those employed in immobilized enzymatic glucose−fructose isomerization accompanied by quasi-continuous chromatographic separation, 156 and could be readily extended to improve heterogeneously catalyzed isomerization. Figure 12a illustrates this reactor concept, in which six reactive chromatographic fixed-bed reactors are interconnected to form a closed-loop. Three functional zones have different roles in the process: zone I between the desorbent and the extract line is for regeneration of the solid, zone II is for reaction/separation, and zone III between the feed and the eluent line is for reaction/separation and recycling of solvent.
Three different lines (feed, desorbent, and extract) are fed by three different pumps, as shown in Figure 12b. An additional fourth pump closes the loop and recycles flow from the last column to the first. At any time, only one valve associated with a given line is open. Port switching is achieved by simultaneously closing the open valve (e.g., De1) and opening the next valve in the direction of the flow (e.g., De2). In this design, the integration of reaction and separation in one unit operation is advantageous for glucose isomerization, where it enables product removal from the reaction zone and displacement of the equilibrium. 157 Enhanced yields of fructose from glucose have also been achieved through the use of aqueous alcohol cosolvents over Si/Al zeolite catalysts. Reactively formed fructose in methanol immediately reacts to form methyl fructoside. In a subsequent step, water is added to hydrolyze the methyl fructoside back to fructose. 158 Lewis acidic Sn-beta also shows enhanced glucose conversion and fructose yield with a methanol solvent; 136 Density Functional Theory (DFT) calculations suggest this results from a differing solvation behavior of water versus methanol in the hydrophobic pores of Sn-beta. 159 3.1.5. Summary of Solid Base Isomerization Catalysts. A relatively narrow range of catalysts have been assessed for xylose/glucose isomerization, typically operating between 90 and 110°C and offering modest sugar conversion albeit with notably high fructose selectivity. Few studies quantify catalyst performance in terms of either initial rates of sugar conversion, per site turnover frequency (TOF), or isomer productivity, hampering accurate benchmarking. Catalyst characterization is also rather restricted to either acid/base site densities and/or classification as Brønsted/Lewis, with acid/base strength distributions rarely reported. The apparent activation energy for glucose isomerization over NaX zeolites of 104 kJ mol −1138 is around 22 kJ mol −1 lower than that over the hydroxyl form of hydrotalcites, 151 indicative of a different reaction mechanism, although there is little fundamental work in this regard; over  NaA zeolites, isomerization is reported as first order in glucose for concentrations <1 wt %, but zero order >5 wt %; in contrast, over NaX zeolites the transition from first to zero order only occurs at glucose concentrations >10 wt %. 138 The work of Davis and co-workers is notable, revealing that Sn-beta zeolite maintains activity and selectivity under acidic conditions. This is significant because it unlocks the possibility for one-pot transformations compatible with acid pretreatments employed to isolate carbohydrate components of biomass. 145 Catalyst deactivation during isomerization is generally attributed to a loss of active sites through leaching, or site-blocking by strongly adsorbed products, although the origin is rarely investigated, with Sn-beta 145 and Sn−Al-beta 160 exhibiting the best stability over repeated reuse.
Despite efforts to develop new catalysts, isomerization efficiency is constrained by virtue of its slight endothermicity (ΔH = 3 kJ mol −1 ) and reversibility (K eq ≈ 1 at 25°C), restricting the maximum attainable fructose yield (according to Moliner et al., the equilibrium conversion of glucose at 90°C is 55% 145 ). However, Paniagua and co-workers have designed a two-step process utilizing different commercial zeolites to enhance xylose isomerization, through its initial conversion to xylulose and subsequent reaction with methanol to form methyl xyluloside (step 1), which could in turn be hydrolyzed back to xylulose (step 2), affording a 47% xylulose yield over H-USY(6) at 100°C. The same process translated to glucose, resulting in a remarkable 55% fructose yield at 120°C, higher than for any single catalytic step in Table 1. Scale-up of this two-step reaction may prove a challenge for commercial development. Hot-compressed (hydrothermal and supercritical) water, which is known to fragment or hydrolyze cellulose within woody biomass, 163 has also been harnessed to initiate ionic and radical isomerization. However, the resulting reaction networks are complicated by competing dehydration, condensation, and degradation of fructose. Future chemical technologies would be better focused at lower temperature and neutral reaction media to reduce side reactions.

Solid Acids for Dehydration
Catalytic dehydration of C 5 −C 6 sugars is conventionally catalyzed by liquid acids, such as H 2 SO 4 , HCl, H 3 PO 4 , oxalic, and maleic acid, in aqueous solution. 164,165 Lewis acid metal halides (e.g., CrCl 3 ) have also been investigated in conjunction with a liquid Brønsted acid (e.g., HCl). 166,167 However, liquid acid processes are associated with corrosion, toxicity, and separation and product recovery issues (and associated high  168 In xylose dehydration, H-form ferrierite, ZSM-5, mordenite, beta, and Y zeolites possessing similar acid densities, and external and specific surface areas, were compared to determine the impact of zeolite structure (entry 2, Table 2). 169 Pore channel size influenced both conversion and selectivity to furfural, with conversion linearly proportional to channel diameter as shown in Figure 13, while H-ZSM-5 exhibited the highest furfural selectivity due to the similarity of its channel size to the product dimensions (size selectivity). However, mass balances were not reported, an important omission for such microporous materials wherein slow in-pore diffusion may favor side reactions leading to humins, although in this instance furfural selectivity appeared time-independent. Dealumination and desilication also improved the catalytic activity of H-ZSM-5 due to the resulting change in solid acidity, with dealuminated H-ZSM-5 exhibiting higher furfural selectivity than the parent H-ZSM-5 due to an increase in Brønsted acid sites (entry 2, Table 2). 169 The effect of Si:Al ratio has also been explored in H-MCM-22-catalyzed xylose dehydration. 170 Decreasing the Si:Al ratio enhanced dehydration activity without adversely affecting furfural selectivity. The delaminated analogue ITQ-2 has a significantly higher external surface area than H-MCM-22, but a similar acid loading, and hence both exhibit comparable performances (for Si:Al = 24). ITQ-2 was however easier to regenerate, suggesting differences in the nature of carbonaceous deposits formed during reaction suppressed deactivation by coking. Acid loading and strength were both shown to influence the catalytic performance of hierarchical USY, beta, and ZSM-5 zeolites (entry 5, Table 2) 171 prepared by alkaline treatment, and whose acid strength, density, and distribution were quantified ( Figure 14). When applied to fructose dehydration, these modified zeolites offered higher 5-HMF selectivity than their parent materials ( Figure 15). As mentioned in section 3.1.1, framework-substituted Sn-beta zeolite is an efficient catalyst for xylose and glucose isomerization, 109,136 and has been explored in conjunction with HCl as a Brønsted acid catalyst for one-pot xylose conversion to furfural in water at low temperature; catalyst stability in such an acidic environment remains difficult, and the use of corrosive mineral acids undesirable.
Microporous zeolites exhibit poor hydrothermal stability at high reaction temperatures (>150°C) 112 and weaker mechanical strengths as compared to dense metal oxides. Moreover, dissolved zeolitic species formed in water can catalyze homogeneous reactions of fructose and 5-HMF. 172 174−179 Early studies show that the sulfonic acid resin Amberlyst-15 catalyzed fructose dehydration at 80°C in an IL and dimethyl sulfoxide (DMSO) solvent mixture. 180 Mesoporous SO 3 H-MCM-41 silica and SO 3 H-hybrid-organic-silica also showed activity for xylose dehydration in DMSO or water, biphasic   toluene/water, or MIBK/water solvents. 181 Poorly ordered microporous hybrid materials prepared via co-condensing 3mercaptopropyl-trimethoxysilane with bis-trimethoxysilylethylbenzene exhibited lower furfural selectivity as compared to the mesoporous MCM-41 silica functionalized postsynthesis (entry 8, Table 2), with selectivity increasing strongly with reaction temperature.
More recently, silica nanoparticles with a mesoporous silica shell and dense silica core (MSHS) have been developed for the catalytic dehydration of xylose to furfural in water. These were modified by sulfonic acids as Brønsted acids (SO 3 H-MSHS), or by aluminum as Lewis acids (Al-MSHS) ( Figures  16 and 17). 116 SO 3 H-MSHS outperformed Al-MSHS in respect of selectivity to furfural due to the latter driving xylose isomerization to lyxose. The hydrothermal stability of SO 3 H-MSHS was superior to other mesoporous silica materials such as MCM-41 (entry 6, Table 2), although the forcing reaction conditions (170−190°C) and associated lack of reported humin formation are surprising. Nevertheless, these MSHS materials suffer from relatively low surface areas and consequent low acid site densities.
In an effort to address the issue of low active site loadings, "hairy" solid acid catalysts have been designed by outward growth of poly(sodium 4-styrenesulfonate) brushes from the surface of silica core particles ( Figure 18). Subsequent acidification resulted in poly(4-styrenesulfonic acid) (PSSH) brushes (entry 10, Table 2). 182 Such PSSH catalysts displayed improved 5-HMF yields from fructose dehydration in water than the free homopolymer acid, attributed to a unique solvation microenvironment formed by the densely grafted PSSH chains in the brush structures. The sulfonic acid hybrid catalysts could be reused several times with a minimal decrease in the 5-HMF yield (entry 8, Table 2).
Sulfonated graphene and graphene oxides have also been tested for xylose dehydration; 183 sulfonated graphene oxide proved an active and water-tolerant solid acid catalyst even at very low catalyst loadings down to 0.5 wt % (relative to xylose), and able to sustain its initial activity for 12 reuses at 200°C, with an average furfural yield of 61% versus 44% for the uncatalyzed system (entry 7, Table 2).
Glucose dehydration to 5-HMF is economically more attractive than from fructose because the former can be obtained directly from cellulose hydrolysis. Nevertheless, glucose conversion is more challenging because it requires a tandem catalytic process involving a Lewis acid-or basecatalyzed isomerization and subsequent Brønsted acid catalyzed dehydration. Glucose is also susceptible to side reactions such as condensation and mutarotation, 184,185 and the mechanism of its isomerization to fructose and associated active sites remains debated. 146,186,187 The design of efficient catalysts for glucose dehydration to 5-HMF remains problematic for low temperature aqueous operation. 188 3.2.2.2. Sulfated Metal Oxides. Bifunctional sulfated zirconia (SZ) has been investigated in detail for one-pot 5-HMF production from glucose (entry 9, Table 2). 154 A comparison of acidic properties and associated reactivity toward glucose and fructose revealed that submonolayer SO 4 coverages provide the optimal balance between base sites on the exposed zirconia responsible for glucose isomerization to fructose, and polydentate Brønsted acid sulfoxy species coordinated to the underlying tetragonal zirconia support efficient for fructose dehydration to 5-HMF ( Figure 19). TOFs for fructose dehydration evidenced that this transformation was catalyzed solely by Brønsted acidic sulfoxy groups. However, these materials possessed low surface areas, and hence sulfation of high surface area ZrO 2 /SBA-15 prepared via a conformal ZrO 2 coating method was developed to create high area analogues over a mesoporous SBA-15 template. 155 Resulting bilayer SZ/ SBA-15 exhibited excellent hydrothermal stability, and a 3-fold enhancement in 5-HMF productivity from both glucose and fructose as compared to nonporous SZ counterparts. SZ/SBA-   15 is also a promising catalyst for the one-pot cascade synthesis of alkyl levulinates from glucose. 189 The hydrothermal stability of this material is particularly interesting as this is one of the key desirable properties for catalysts in aqueous phase biomass processing.
3.2.3. Metal Phosphates. Metal phosphates have been widely investigated in academia and industry as heterogeneous catalysts. 190,191 Early studies reveal niobium phosphate as active for the aqueous phase dehydration of fructose to 5-HMF, 190 although little correlation was reported between acidic properties and performance with conversions of between 30% and 60% and high initial 5-HMF selectivity, although in all cases selectivity declined sharply over the course of reaction to unidentified (presumably polymeric) secondary products.
Doped and oxide supported vanadyl phosphates (VOPO 4 · 2H 2 O) with Brønsted and Lewis acid sites have also been studied for fructose dehydration. Doped vanadyl phosphates were prepared by isomorphic substitution of VO 3+ groups with trivalent metals Fe 3+ , Cr 3+ , Ga 3+ , Mn 3+ , and Al 3+ (entry 14, Table 2). 192 In the case of supported vanadyl phosphates, the support acidity was also important in promoting dehydration. For the doped samples, Fe-VOPO 4 ·2H 2 O offered the best activity and selectivity, and was able to convert concentrated aqueous fructose solutions with a 5-HMF productivity of 376 mmol g cat −1 h −1 at 80°C, with minimal insoluble polymers or 5-HMF rehydration reported. Fructose dehydration over copper phosphates was dependent on their surface acidity (Brønsted/Lewis character) and morphology ( Figure 20) (entry 16, Table 2). 193 Heat treatment of CuHPO 4 ·H 2 O nanoneedles transformed it into α-Cu 2 P 2 O 7 nanocrystals at 600°C and rod-like nanostructures at 900°C. The thermally processed phosphates exhibited weak acidity (+3.3 ≤ H 0 ≤ +4.8) but enhanced productivity as compared to H 3 PO 4 , with the α-Cu 2 P 2 O 7 -900 giving up to a 36% 5-HMF yield.
Calcium and α-strontium phosphates have been investigated for glucose, fructose, and cellulose conversion in hot compressed water (entry 15, Table 2). 194 These catalysts were prepared by a modified coprecipitation method, resulting in worm-like SEM morphologies as shown in Figure 21. Although they are nonporous with very low (0.5 g m −2 ) surface areas, high activity was observed for glucose and fructose dehydration, and the hydrolysis and dehydration of cellulose ( Figure 22), with CaP 2 O 6 and α-Sr(PO 3 ) 2 phosphate favoring 5-HMF formation from glucose and fructose.
In comparison with the preceding niobium, vanadium, copper, and alkaline earth phosphates, zirconium phosphates (ZrPO) are more extensively investigated. Crystalline ZrPO, precipitated from ZrCl 2 , was explored for fructose dehydration in subcritical water at 240°C, achieving around 80% fructose conversion after 120 s with 61% selectivity to 5-HMF. 195 However, these aggressive conditions resulted in a high background rate of dehydration to 5-HMF. Metal phosphates of aluminum (AlPO), titanium (TiPO), zirconium (ZrPO), and niobium phosphates (NbPO) have also been compared for glucose dehydration, 196 with a view to examine the effect of different phosphate species and associated acidity. Acid strength and activity both increased with decreasing electronegativity of the metal cation, with NbPO > ZrPO > TiPO > AlPO. The Brønsted:Lewis acid site distribution also influenced 5-HMF selectivity, with excess Lewis acidity driving unselective glucose   transformation into humins (see entry 13, Table 2). The modification of NbPO and ZrPO through silylation to eliminate unselective Lewis acid sites afforded a drastic increase in 5-HMF selectivity, which reached 60% (albeit at 135°C).
Porous metal phosphates have also attracted interest with a view to enhancing their catalytic performance. Cheng et al. reported mesoporous zirconium phosphates ( Figure 23) obtained by hydrothermal synthesis using organic amine (dodecylamine and hexadecylamine) templates, which exhibited high conversion (up to 96%) for xylose dehydration in water and corresponding high furfural yields reaching 52% (entry 12, Table 2). 197 This was attributed to their open internal structures and abundant Brønsted/Lewis acid sites; the mesoporous ZrPO catalyst showed good stability and was easily regenerated by calcination.
While (transition) metal phosphates show promise for C 5 − C 6 sugar dehydration, control over their textural properties is limited due to restructuring under high temperature calcination to remove structure-directing agents, 198,199 while mild calcination or solvent extraction (e.g., acid and ethanol) is insufficient to completely remove organic templates. Synthesis of porous metal phosphates is also restricted currently to a limited range of metals: zirconium, titanium, and niobium. 200 Continued effort should be devoted to preparing porous and ordered metal phosphates with improved active site accessibility.
3.2.4. Composite Metal and Nonmetal Oxides. Metal oxides are extensively used in catalysis as active components or supports, and have been investigated in the aqueous phase reforming of C 5 −C 6 sugars. A comparison of TiO 2 and ZrO 2 with homogeneous mineral acids (H 2 SO 4 and HCl) highlighted their advantages for both glucose and fructose dehydration during microwave irradiation (and sand bath heating to 200°C ). 201 ZrO 2 promoted glucose isomerization to fructose, whereas TiO 2 drove both glucose isomerization and subsequent dehydration of reactively formed fructose to 5-HMF. TiO 2 and ZrO 2 were both efficient for 5-HMF production from fructose (entry 19, Table 2), suppressing levulinic and formic acid products of 5-HMF rehydration.
Supported metal oxides have also been utilized; for example, TiO 2 (8−9 nm anatase) carbon nanocomposites were prepared by a microwave-assisted method for the catalytic dehydration of xylose into furfural ( Figure 24). 202 TiO 2 /RGO (reduced graphene oxide) and TiO 2 /CB (carbon black) exhibited high furfural yields (67−69%) under identical conditions, and showed excellent stability with negligible Ti leaching over 3 cycles. This synthetic approach could be generalized to a variety of metal oxides and composites.
Mixed metal oxides offer interesting opportunities to tune Brønsted and Lewis acidity relative to their oxide constituents. CeO 2 −Nb 2 O 5 mixed oxides prepared by coprecipitation were tested for fructose dehydration; 203 although no crystalline mixed phases were observed, the CeO 2 :Nb 2 O 5 ratio influenced the strong acid site density and associated conversion and selectivity ( Figure 25). Higher Nb 2 O 5 loadings favored fructose dehydration and 5-HMF selectivity; however, there was no clear evidence of any synergy, with pure niobia superior to any of the ceria composites.
Nonmetal oxides are also sporadically reported for the catalytic dehydration of sugars. For instance, graphene oxides have been examined in xylose conversion to furfural, in comparison with graphene, sulfonated graphene, and sulfonated graphene oxides (Figure 26). 183 Although sulfated graphene oxides showed the best 5-HMF selectivity and yield, this study does show the potential of nonmetal oxides for the C 5 −C 6 sugar dehydration.
3.2.5. Other Solid Acid Catalysts. Heteropolyacids are an important class of catalysts due to their superacidity and strong redox properties for which they have found application in selox. Heteropolyacid salts have also been investigated in catalytic sugar transformations, with Fan et al. reporting that Ag 3 PW 12 O 40 catalyzed 5-HMF production from fructose and   glucose; 204 it was claimed that fructose could be dehydrated to 5-HMF with 94% selectivity and 78% yield within only 1 h at 120°C (entry 18, Table 2). Silver ion-exchanged silicotungstic acid Ag 4 [Si(W 3 O 10 ) 4 ]·nH 2 O has also been proven an effective catalyst for fructose and sucrose dehydration to 5-HMF in superheated water, 205 achieving 98% fructose conversion and an 86% 5-HMF yield. However, the cost of such high loading (10 wt %) Ag-exchanged catalysts would likely be prohibitive for scale-up as compared to, for example, acidified carbons. 117 Table 2), 206 with a yield of 86% from fructose in DMSO (versus 56% in water, this difference is significant because DMSO is itself an effective Brønsted acid catalyst for this dehydration 206 ). Somewhat lower yields of 48% 5-HMF were obtained from glucose ( Figure 27). The authors claimed that the "low polar" 5-HMF product may be stabilized in the hydrophobic channels of the ionic crystals.
3.2.6. Process Considerations. A continuous process for fructose dehydration to 5-HMF has been developed around water tolerant niobia catalysts using a fixed-bed tubular flow reactor, 209 in which both the preheater and the reactor are held isothermal via a hot air circulating oven ( Figure 28). The reactor design enabled efficient temperature control between     90 and 110°C. Pressure control between 2 and 6 bar was achieved using a micrometric valve at the end of the reaction line to avoid water evaporation and the formation of gas bubbles in the catalyst bed. 5-HMF selectivity increased with fructose conversion under continuous flow conditions, contrary to observations in batch, wherein 5-HMF selectivity normally decreases with conversion. Short residence times accessible in continuous flow operation allow rapid product removal and suppression of undesired side reactions.
A major limitation to the development of economic C 5 −C 6 sugar dehydration in pure water is the accompanying poor selectivity to 5-HMF or furfural due to product degradation. 210 Use of a biphasic continuous reactor is recognized as one of the most promising methods to improve the yield of 5-HMF, whereby a water-immiscible organic solvent is used for continuous 5-HMF extraction from the aqueous phase, thereby facilitating efficient product separation and limiting degradation reactions. 211−215 Xylose and fructose conversion to furfural or 5-HMF, respectively, is reported to proceed with improved selectivity (up to 90%) when conducted in such biphasic reactors, which also permit continuous product extraction. 59,214 A continuous flow, biphasic fixed-bed reactor for furfural production from xylan and xylose has been reported 216 utilizing a mixed catalytic bed of a Lewis acid gallium-containing USY zeolite for xylose isomerization and a Brønsted acid ionexchanged resin (Amberlyst-36) for hemicellulose hydrolysis and xylulose dehydration (Figure 29). High product selectivity was achieved via efficient extraction of furfural to the organic phase, with furfural yields of 72% from xylose and 69% from xylan obtained.
Addition of solvent modifiers including DMSO, 217,218 1methyl-2-pyrrolidinone (NMP), poly(1-vinyl-2-pyrrolidinone) (PVP), and alcohols 218−220 is reported to inhibit aqueous phase side reactions and increase 5-HMF productivity. While the use of aprotic solvents such as DMSO can improve 5-HMF yields, separation from the high-boiling point solvent is difficult due to attendant product decomposition; hence there is growing interest in molecular simulation to obtain fundamental insight into the design of such mixed solvent systems. Desirable solvent features include (i) nonmiscibility of the cosolvents; (ii) high partition coefficients for extracting the desired compound from water without extracting the catalyst; and (iii) higher boiling points than the product to aid distillation and final product separation. MD simulations to investigate the solubility of 5-HMF in 2-butanol/MIBK biphasic systems 221 demonstrate the potential of in silico screening of organic solvent mixtures to assist process design. 2-sec-Butylphenol (SBP), propyl guaiacol (PG), and propyl syringol (PS), which can be derived from lignin, have also proven viable renewable cosolvents 222 for furfural, 5-HMF, levulinic acid, and γ-valerolactone production from C 5 −C 6 sugars. Process optimization also necessitates the development of heterogeneous catalysts, which selectively partition within the aqueous phase. 223,224 As discussed in section 1.1, ILs, DES, and CO 2 are potential cosolvents for use in aqueous phase catalysis to enhance the dissolution of nonpolar substrates and/or provide a second phase to extract products into. DESs are becoming a popular green solvent for catalytic reactions; 222 however, there are few reports on catalytic processing of sugars in biphasic DES/H 2 O or IL/H 2 O systems, with most studies concentrating on enzymatic transformations in which DESs enhance substrate solubility and increase reaction rates and in some cases regio-or enantioselectivity. 66 One promising example of a chemocatalytic transformation concerns the mixed solvent systems of betaine hydrochloride (BHC) and choline chloride (ChCl) in water for fructose and inulin dehydration to 5-HMF. 225 MIBK addition enabled the continuous extraction of 70% 5-HMF from the ChCl/BHC/H 2 O system, which could be recycled seven times. Alternatively, addition of an AlCl 3 Lewis acid catalyst to ChCl/BHC/H 2 O promoted glucose isomerization to fructose, with a 40% 5-HMF yield. ZrO 2 -catalyzed 5-HMF production from glucose proceeded more effectively in IL−water mixtures (10−50 wt % water in 1,3-dialkylimidazolium chloride) than pure water or IL; a 50 wt % 1-hexyl-3methyl imidazolium chloride mixture offered 53% 5-HMF in only 10 min at 200°C. 226 3.2.7. Summary of Solid Acid Dehydration Catalysts. The catalytic performances of the preceding solid acids are summarized in Table 2, and represent a more diverse collection than employed for xylose and glucose isomerization (which reflected principally zeolites and hydrotalcites). Reaction temperatures for dehydration are generally higher (120−200°C ) than employed for isomerization (90−120°C), with the attendant possibility of side reactions of reactants, intermediates, and products. Wide variations are observed in sugar conversions; however, product yields were relatively low, with the highest yield observed over the sulfonated graphene oxide and Ag 3 PW 12 O 40 (entries 7 and 18, Table 2). Although few studies provide kinetic data or mass balances, the latter is extremely important because the xylose, glucose, or fructose dehydration may be accompanied by a range of side reactions and require detailed HPLC analysis. 154 Catalyst reusability has only been explored for select systems, notably the sulfated graphene oxide 183 and heteropolyacids 206 highlighted above. A high density of accessible and medium-strong acid sites is required for low temperature (<150°C) sugar dehydration, with Brønsted acidity generally considered a key factor for high activity and selectivity to furfural (Figure 30) 208 and 5-HMF, with the latter also favored by porous structures. Water is both the reaction media of choice in biomass conversion, and byproduct of dehydration; hence competing rehydration and hydrolysis pathways are inevitable, limiting selectivity and presenting a challenge to most solid acids; metal oxide−water  interfaces drive a range of chemistries, including acid−base, ligand exchange, and/or redox processes, 227,228 and require further study in the context of sugar dehydration. Water is also a common poison of solid acids, either through leaching of active sites or strong adsorption and site-blocking, and watertolerant solid acids would thus represent a key development for commercialization. 229

Metal Catalysts for Hydrogenation
Early studies revealed Raney nickel and ruthenium, platinum, cobalt, copper, and rhodium as promising catalysts for selective hydrogenation of xylose, glucose, and fructose. 230−232 Although the hydrogenation of keto or aldose carbonyls should be routine, selectivity is hampered by the tendency for sugars to undergo competitive isomerization, polymerization, degradation, and dehydration ( Figure 6). This challenge has largely been addressed through the use of promoters and cocatalysts, or modification of support structure or electronic properties.  Table 3), 233 enhancing activity 7-fold ( Figure 31) for Fepromoted Raney Ni. The promoter type, distribution, and the oxidation state also influenced performances, with low valent species acting as Lewis sites binding glucose through the oxygen of the CO group, polarizing the bond to favor hydrogenation via nucleophilic attack by surface hydrogen. Promoter stability remains to be addressed.
3.3.1.2. Ni−P Amorphous Alloys. Inspired by Raney Ni, a skeletal Ni−P amorphous alloy catalyst (Raney Ni−P) was prepared by alkali leaching of a Ni−Al−P precursor, 234 which exhibited higher TOFs (per surface Ni) than Raney Ni catalyst in glucose hydrogenation to sorbitol, attributed to P promotion of active Ni sites (entry 2, Table 3). The Raney Ni−P catalyst outperformed conventional Ni−P alloys due to a higher surface area and density of surface Ni atoms.
3.3.1.3. Ni−B Amorphous Alloys. Silica supported amorphous Ni−B was also investigated in glucose hydrogenation. 235 The as-prepared Ni−B/SiO 2 catalyst was more active than other Ni catalysts, such as crystalline Ni−B/SiO 2 , Ni/SiO 2 , and commercial Raney Ni (entry 3, Table 3), attributed to the high dispersion of Ni and electronic perturbation of metallic Ni by B, which was proposed to induce stronger CO adsorption to Ni atoms. The addition of Cr, Mo, and W promoters further improved activity (Figure 31). High promoter concentrations resulted in site-blocking of active Ni sites. The question of selectivity was largely neglected in the preceding Ni alloy studies, with sorbitol implicit as the major product even though Lewis acid promoters are also expected to drive glucose isomerization. 236 3  Table 3) exhibited higher activity than crystalline Co and Ni analogues, with Cr, Mo, and W further increasing conversion, attributed to enhanced dispersion of the Co active species (and hence carbonyl adsorption). Ultrafine Ru−B particles in amorphous Ru−B alloys likewise conferred higher activity in glucose hydrogenation than crystalline Ru−B and pure Ru powders. In another study, 240 addition of trace Cr in the form of Cr 2 O 3 similarly promoted glucose hydrogenation by increasing Ru−B alloy dispersion.
3.3.2. Ni. Raney Ni was employed historically for the hydrogenation of sugar compounds due to its low cost; however, supported nickel catalysts often offer higher activity. Kusserow et al. investigated Ni over SiO 2 , TiO 2 , Al 2 O 3 , and carbon supports in comparison with commercial Raney Ni for the aqueous phase hydrogenation of glucose in batch and flow, 124 prepared by impregnation, sol−gel, precipitation, and templated syntheses ( Figure 32). Impregnated alumina proved the best (entry 6, Table 3), with a conversion comparable to that of an industrial Ni68T catalyst (despite containing only 5 wt % Ni versus 66 wt %) and similar sorbitol selectivities of around 90%. Ni on TiO 2 and SiO 2 exhibited lower conversions, with carbon providing poor activity. Specific sorbitol productivities followed Al 2 O 3 > TiO 2 > SiO 2 > C. Prereduction temperature strongly influenced performance, with catalysts reduced at 500°C more active than those reduced at 300°C. Further investigations of catalyst pretreatment for Ni/SiO 2 in glucose hydrogenation 241 revealed that calcination prior to reduction gave higher conversion and selectivity than direct reduction due to improved decomposition of the nickel precursor.
Recently, Ni/Cu/Al hydrotalcite precursors and Ni/Cu/Al/ Fe hydrotalcite-like catalysts were developed for the hydrogenation of glucose and fructose, respectively. Preparation and activation treatments of Ni/Cu/Al hydrotalcite precursors had a significant impact on glucose hydrogenation; 242 high temperature reduction increased activity and selectivity toward sorbitol over a Ni 1.85 Cu 1 Al 1.15 catalyst (entry 7, Table 3) 243 independent of pH, and also increased the activity of a Ni 4.63 Cu 1 Al 1.82 Fe 0.79 HT for fructose hydrogenation ( Figure  33).
3.3.3. Ru. Ruthenium catalysts are more extensively investigated than Ni analogues for the hydrogenation of sugars, being more efficient for aqueous phase carbonyl hydrogenation. 244,245 Ru supported on activated charcoal via cation or anion exchange was tested in a continuous trickle-bed reactor for glucose hydrogenation at 100°C and 80 bar H 2 as a function of residence time, 241 with activity inversely proportional to Ru loading, and sorbitol selectivity inversely proportional to residence time, longer reaction leading to sorbitol epimerization to mannitol. Continuous-flow operation hence offered improved selectivity over batch. This catalyst was stable over several weeks' operation without Ru leaching. Overall Ru/C catalysts were comparable to, or more active, selective, and stable than, Raney Ni catalysts, which suffered from leaching of both Ni and promoters. 246 Ru nanoparticles have been dispersed on mesoporous carbon (Ru/C-NFs) microfibers, using Al 2 O 3 microfibers as templates by a chemical vapor deposition route. 247 Ru nanoparticles appear embedded within the mesoporous carbon matrix ( Figure 34). Such Ru catalysts exhibited significantly enhanced activity and stability for glucose hydrogenation (entry 10, Table  3) due to the open mesoporous structure of the support, the unique microfiber morphology, and hydrogen spillover at the interface between the embedded Ru nanoparticles and carbon support. N-doping of the carbon also enhanced catalytic performance, attributed to stronger hydrogen adsorption, and better Ru wettability and electronic properties.
In fructose hydrogenation, Ru/C was compared to Pt/C and Pd/C catalysts. Heinen et al. found the furanose form of fructose reacted over Ru/C, whereas the pyranose form was inert under the same conditions (1 bar H 2 , 72°C), 248 despite these anomers adsorbing with comparable strength. It is interesting to note that the selectivity to mannitol for Pd/C and Pt/C could be improved by tin promotion. Structure−reactivity  251 Hydrogenation was first order with respect to hydrogen, and also first order with respect to D-glucose for concentrations <0.3 M (but zero order at high concentration). Modeling implicated Langmuir−Hinshelwood−Hougen−Watson (LHHW) kinetics wherein the surface reaction was ratedetermining, but was unable to distinguish between (i) noncompetitive (molecular or dissociative) adsorption of H 2 and D-glucose; (ii) competitive adsorption of molecular H 2 and D-glucose; or (iii) competitive adsorption of dissociatively chemisorbed hydrogen and D-glucose. In contrast to carbon supported Ru, glucose hydrogenation over silica analogues revealed growth/aggregation of Ru nanoparticles. In situ X-ray absorption spectroscopy and electron microscopy showed that SiO 2 supported Ru was in an oxidized state when air-exposed, undergoing reduction and concomitant sintering upon exposure to 40 bar H 2 at 100°C in aqueous solution. However, the authors proposed that glucose acts to stabilize Ru against sintering (Figure 35), 252 preventing migration of Ru species during hydrolysis of the SiO 2 surface. This Ru/SiO 2 catalyst was entirely selective to sorbitol.     Ru/HYZ (H-formed zeolite Y) has also been investigated in the selective hydrogenations of xylose to xylitol and glucose to sorbitol, 123,256 as a function of reaction temperature, reaction time, catalyst loading, Si:Al ratio, and Ru concentration. Ru/ HYZ exhibited exceptional TOFs in glucose hydrogenation, exceeding Ru/TiO 2 and Ru/NiO-TiO 2 (entry 13, Table 3), with >97% sorbitol selectivity. Reuse tests also found that Ru/ HYZ had good stability, although some Ru nanoparticle aggregation was observed after four runs (Figure 36). 256 The influence of preparation method used to introduce Ru into ZSM-5 was investigated for glucose hydrogenation. 257 Ru-/ZSM-5 was obtained either by adding a Ru precursor to the parent ZSM-5 synthetic mixture (Ru/ZSM-5-TF), or through loading Ru nanoparticles on a commercial microporous ZSM-5 (Ru/ZSM-5-MS) or an alkali-treated mesoporous ZSM-5 (Ru/ZSM-5-AT) by incipient wetness impregnation. All of the Ru/ZSM-5 catalysts outperformed Ru/C (entry 14, Table 3), notably Ru/ZSM-5-TF, which offered the highest conversion and sorbitol selectivity. Ru incorporated in the Ru/ZSM-5-TF catalyst exhibited a uniform size and homogeneous distribution of nanoparticles throughout ZSM-5 ( Figure 37).
Ru has also been explored over mesoporous MCM-41 silica (3.8 wt % Ru/MCM-41), prepared by impregnation and reduction with formaldehyde. This exhibited higher activity and sorbitol selectivity than a range of other transition metal catalysts for glucose hydrogenation (Figure 38). 254 Ru/MCM-41 has twice the metal surface area of Ru/C, suggesting the excellent performance is simply due to greater active site dispersion. Although the Ru/MCM-41 catalyst could be reused, the metal surface area decreased from 4.5 to 0.78 m 2 g −1 , and hence stability is an issue.
Metal oxides are also popular supported for Ru, with Mishra et al. reporting Ru/NiO−TiO 2 for xylose and glucose hydrogenation. 253,258 The TiO 2 support was first modified by nickel chloride impregnation and subsequent oxidation.
Catalytic performance in xylose hydrogenation to xylitol was comparable to Raney Ni, Ru/C, and Ru/TiO 2 under identical reaction conditions; however, NiO addition enhanced conversion, yield, and selectivity toward xylitol. Considering glucose hydrogenation, the NiO-modified TiO 2 catalyst improved sorbitol selectivity by lowering fructose production and its subsequent hydrogenation to mannitol (entry 15, Table  3). Ru on poly(styrene-co-divinylbenzene) amine-functionalized polymers (Ru/PSN) was studied recently for xylose hydrogenation to xylitol. 263 In addition to facile Ru/PSN recovery and reuse, its selectivity to xylitol exceeded 90%. In a related study, Ru nanoparticles embedded within a mesoporous hyper-cross-linked polystyrene were investigated for glucose hydrogenation (entry 12, Table 3). 255 Two reaction mechanisms were considered: the interaction of glucose with hydrogen spilled-over from Ru to the support; and the direct reaction of adsorbed glucose with molecular hydrogen from the reaction medium. Catalyst deactivation was attributed to the complete consumption of "stored hydrogen" from spillover of the as-prepared catalyst following the initial stage of reaction.
Mechanistic aspects of Ru-catalyzed hydrogenation of bioderived carbonyl compounds (albeit not the C 5 −C 6 parent sugars) offer some insight into the active species. 244 Jae et al. studied 5-HMF hydrogenation to 2,5-DMF by XPS, XAS, and HRTEM, noting that fresh Ru nanoparticles were dominated by surface RuO x , which underwent in situ reduction to metallic Ru resulting in deactivation. 264 They proposed that RuO x behaved akin to a Lewis acid, promoting an MPV pathway via interhydride transfer from 2-propanol to 5-HMF to form 2,5-bis(hydroxymethyl)furan. Prereduced Ru/C exhibited only moderate hydrogenolysis activity toward 5-HMF, with ∼30% selectivity to 2,5-DMF. A synergy between physically mixed (partially) oxidized and metal phases increased 2,5-DMF selectivity to 70%. DFT calculations and experimental studies of Ru-catalyzed levulinic acid, 245,265 furfural, 266 L-arabinose, 267 and 2-butanone 268 hydrogenation have highlighted a crucial role for surface H 2 O in regulating substrate activity, Ru surface oxidation state, and intermediate reactivity, although the nature of the participating water species has yet to be fully elucidated. Cooperative interactions between CO functions in reactive intermediates adsorbed over Ru with coadsorbed, neighboring water molecules are proposed to decrease the barriers toward levulinic acid hydrogenation to γ-valerolactone; higher experimental rates were observed as the hydrogen-bond donor capability of the solvent increased. 265 However, computational studies suggest that water undergoes dissociative     271−273 However, in the context of xylose/glucose hydrogenation to xylitol/sorbitol, Pt has proven less popular than Ni or Ru catalysts.
Pt on activated carbon cloth (ACC), see Figure 39a−c, was investigated for glucose hydrogenation. 259 As compared to traditional powder or granular carbon supports, ACC displayed a high surface area and narrow micropore size distribution, which favored efficient liquid phase mass transport. Pt was introduced by cation exchange after oxidation of the ACC support, with TEM ( Figure 39d) confirming a homogeneous dispersion of Pt in the carbon fibers (2−3 nm diameter nanoparticles). Pt/ACC catalysts were highly active for glucose hydrogenation, giving a sorbitol yield >99.5%, superior to conventional Pt/C catalysts. ACC was inexpensive and resistant to acidic and basic conditions. Activated charcoal is also a good support for Pt-catalyzed glucose hydrogenation, 274 and fructose hydrogenation to mannitol. 275 However, in common with many studies utilizing carbon supports, the chemical composition and distribution of surface functions on the carbon were poorly characterized, limiting their reproducibility. Kanie et al. studied glucose and fructose hydrogenation at 130−270°C in subcritical water under 50 bar H 2 over Pt nanoparticles protected by polyethylenimine (Pt-PEI). 276 Unfortunately, these catalysts were unselective, converting glucose to 1,2-propanediol, 1,2-hexanediol, and ethylene glycol, and fructose to 1,2-propanediol, 1,2-hexanediol, and glycerol.
Bifunctional Pt catalysts employing acidic γ-Al 2 O 3 and basic hydrotalcite supports have been examined for xylose and glucose hydrogenation (entry 18, Table 3 for glucose hydrogenation). 260 A combined sorbitol and mannitol yield of 68% was obtained from glucose over Pt/γ-Al 2 O 3 in combination with hydrotalcite, with an alkaline medium increasing the concentration of open-form glucose, which is more readily hydrogenated. The same combination achieved 99% xylose conversion and 82% xylitol yield. However, these catalysts lost significant activity after only two runs at 90°C.
3.3.5. Other Hydrogenation Catalysts. Copper catalysts popular in redox and photocatalysis, 277−281 and Cu/SiO 2 and Cu/ZnO/Al 2 O 3 , have also been studied in fructose hydrogenation to mannitol. 261 Ultrasonication enhanced the activity of Cu/SiO 2 (and a Raney Ni comparator), but did not improve selectivity (Figure 40), whereas it deactivated a Cu/ZnO/Al 2 O 3 catalyst. A commercial methanol synthesis, CuO−ZnO (61 wt % CuO and 39 wt % ZnO) catalyst was also active for fructose hydrogenation. 262 In both cases, mannitol was the major hydrogenation product with some sorbitol formed through fructose isomerization to glucose. The best mannitol selectivity was around 70% for any Cu catalysts.
3.3.6. Process Considerations. Catalytic hydrogenation of aqueous solutions of glucose to sorbitol has been explored in a high pressure, trickle-bed reactor over supported Ni 282 and Ru catalysts. 283 Selectivity toward sorbitol was improved over charcoal supported Ru catalysts at short residence times; 283 long residence times favored sorbitol epimerization to mannitol. Ru/C catalysts with 1−2 wt % metal prepared by anionic or cationic adsorption afforded 100% glucose and selectivities of 99% to sorbitol under 80 bar H 2 and 100°C. However, scale-up of trickle-bed reactors is hindered by multiple hydrodynamic states, nonuniform temperature distributions, large pressure drops, and liquid holdup. 284,285 The wetting efficiency of solid catalysts, and gas−liquid−solid mass transfer limitations, also present barriers to continuous high pressure heterogeneous hydrogenations. 286,287 In stirred tank reactors, hydrogen is often sparged and mixed using hollow impellers to improve gas solubility; however, hydrogen dissolution in aqueous sugar solutions and uniform radial and axial transport is a challenge in trickle-bed reactors, even in counter-current flow operation. 288 Airlift loop reactors enhance the contact area between gas and liquid phases, and provide more favorable flow patterns for catalytic hydrogenation. Wen and co-workers reported the batch-wise catalytic hydrogenation of glucose to sorbitol in an airlift reactor. 289 Here, gas in the  Chem. Rev. XXXX, XXX, XXX−XXX storage tank passed through a buffer tank prior to injection at pressure into the reactor in which 1.5 m 3 of 50 wt % glucose solution and 37.5 kg of Ni catalyst were added and mixed by continuous hydrogen addition. The sorbitol yield in the airlift loop reactor (98.6%) was higher than that in a stirred tank reactor under the same reaction conditions of 140°C and 70 bar. In a separate study, Ru/γ-Al 2 O 3 monoliths were assessed using an external airlift loop tubular reactor with independent recycle of gas and liquid, and compared to powder analogues using an integrated stirred tank reactor. 290 External mass transfer limitations were observed for the powder catalyst, whereas the monolithic catalyst exhibited internal mass transfer resistance. Catalyst performance was sensitive to reaction temperature and configuration ( Figure 41).
Membrane trickle-bed reactors, which may offer integrated reaction and product separation, have also been studied for glucose hydrogenation to sorbitol, 291 and are compared against a conventional trickle-bed reactor in Figure 42a. In the membrane reactor, gas is uniformly distributed along the reactor bed through the membrane, with liquid reactant flowed over the catalyst bed (Figure 42b) delivering superior conversion due to the distributed addition of H 2 . While membrane reactors are finding growing application, 292 their use in sugar transformations remains limited due to the lack of available membranes stable under reaction conditions, and the added challenge of incorporating on-stream product separation.
Solvent selection may also enhance catalytic hydrogenation of sugars through improved gas solubility. Use of CO 2 as a cosolvent for the catalytic conversion of biomass was recently reviewed, 49 wherein the aqueous phase catalytic hydrogenation of 5-HMF to DMF over Pd/C was promoted in CO 2 /H 2 O mixtures, 293 with poor DMF selectivity obtained in pure water or supercritical CO 2 . Multiphase systems based on water, ioctane, and IL mixtures have also been reported for the Ru/Ccatalyzed hydrogenation of levulinic acid to γ-valerolactone, in which substrate and product partition in the aqueous phase, with the metal catalyst residing in the IL, and i-octane aids phase separation for product recovery and catalyst reuse. 60 Transfer hydrogenation has also been recently explored for the one-pot hydrogenation of cellulose and hemicellulose (xylan) to sugar alcohols, as a means to obviate the requirement for renewable H 2 and associated high pressures and expensive reactors. In both cases, Ru/C catalysts are efficacious in conjunction with i-propanol as a hydrogen source, with Ru/C-Q10 delivering 37% sorbitol and 9% mannitol from milled cellulose at 190°C, 294 and 5 wt % Ru/C affording an 80% xylitol yield from xylan at 140°C (albeit only with trace H 2 SO 4 to initiate hemicellulose hydrolysis to xylose) 295 and a similar performance from sugar cane bagasse. In the former case, highly dispersed cationic Ru is proposed as the active site responsible for transfer hydrogenation, with metallic Ru nanoparticles on alumina inert. Recycling of residual i-propanol, and its regeneration from the reactively formed acetone byproduct, will be essential for large-scale processing.
3.3.7. Summary of Hydrogenation Catalysts. Ni, Ru, and Pt catalysts have been most extensively explored for xylose and glucose hydrogenation (Table 3). Quantitative catalyst comparison is, however, hampered by a lack of kinetic analyses in most studies regarding external and internal mass transfer limitations, which are problematic in such three-phase hydrogenations, and reporting of carbon mass balances. A wide range of hydrogen pressures (30−160 bar) have been employed to date, without adequate precautions to ensure that reactions are first order in catalyst and that reactions are not controlled by slow hydrogen transport at either the gas−liquid or the liquid−  Chem. Rev. XXXX, XXX, XXX−XXX V solid interfaces. Typical reaction temperatures span 80−120°C, comparable to those under which glucose and xylose isomerization occur; however, most hydrogenation studies report very high selectivity to the direct sugar alcohol hydrogenation product. This surprising observation may reflect the difficulty in distinguishing sorbitol and mannitol by HPLC, which coelute over most columns, and hence questions remain as to whether high reported xylitol and sorbitol selectivity are reliable. Ru/ZSM-5 and Pt/activated carbon appear the best catalysts for glucose and xylose hydrogenation; however, there are wide variations in the sugar concentrations that hamper direct comparisons, and mass balances are again rarely reported, which is concerning giving the potential for competing isomerization and dehydration pathways (particularly where acidic, basic, or carbon supports with ill-defined surface properties are employed), that may promote humin or coke formation. Few efforts have been made to determine the hydrogenation mechanism. About one-third of studies suggest that glucose hydrogenation is zero order in glucose for sugar concentrations >10 wt %, switching from first order at lower glucose concentration due to saturation of the catalyst surface by the sugar. Surprisingly, hydrogenation is also reported first order in H 2 pressure (rather than one-half order as for typical olefin hydrogenation). Deactivation is principally attributed to leaching of metals (especially Ni), and poisoning or siteblocking of active sites. Catalyst development should target more efficient desorption of reactive intermediates, and improved mass transfer through the use of high area and/or porous supports. Loss of precious metal by leaching may require encapsulated nanoparticles, 296 such as those developed by Joo et al. in which Pt nanoparticles were enclosed within porous silica to form a Pt core/porous silica-shell structure. 297 Such core−shell motifs have shown high thermal stability in CO oxidation and ethene hydrogenation, preventing metal sintering or loss, and are amenable to metal-catalyzed C 5 −C 6 sugar hydrogenation.  121 Early studies of Pt catalysts suggested they deactivated rapidly, and were strongly dependent on alkaline reaction media. 298 Pd has subsequently attracted significant interest due to its ability to selectively oxidize primary and secondary alcohols. 299,300 Pd/C, 301 Pd/Al 2 O 3 , 302 Pd−Bi/C, 303 and Pd−Bi/SiO 2 304 have been studied for glucose selox to gluconic acid, with alkaline conditions and Bi promoters originally believed to enhance activity and selectivity and/or prevent overoxidation of Pd surfaces and hence their deactivation (the latter hypothesis has been since discredited 305−308 ). Promoter leaching in situ and on-stream deactivation have hampered industrial take-up of such catalyst systems.
Efforts to improve the performance of Pd catalysts have used new supports such as polymers and zeolites, and promoters. 309 For instance, Witonśka et al. studied tellurium-modified Pd (5% Pd−x%Te/support, in which x = 0.5−1 wt % and the support = SiO 2 or Al 2 O 3 ) in comparison with Bi-promoted analogues. 310 Supported Pd−Te catalysts were more active, selective [entry 14, Table 4], and stable, with no Te leaching for loading x ≤ 1 wt %. This Te promotion was ascribed to the formation of interfacial Pd−Te compounds.

Nanoporous/Colloidal Au.
Gold catalysis is now highly topical in large part due to the work of Haruta 311 and Hutchings, 312 and has been heavily investigated for selox, 313−315 notably the aerobic oxidation of alcohols to aldehydes and aldehydes to carboxylic acids. 300,316,317 Selox of sugars has also been examined using Au catalysts in the aqueous phase, 318−321 particularly over titania supports, which interestingly can themselves perform photocatalytic glucose oxidation. 322 The effect of catalyst preparation on glucose selox over Au bimetallic and trimetallic nanoparticles has highlighted the continued efforts to improve gold selox catalysts. 121 Here, we focus attention on recyclable catalyst systems.
Porous Au represents a new class of heterogeneous catalyst suitable for surface functionalization, 323 and unsupported nanoporous Au appears as an effective catalyst for glucose selox to gluconic acid with >99% selectivity under mild conditions. 134,318 Yin et al. prepared nanoporous Au by dealloying Ag/Au alloys (Figure 37), noting that activity was dependent on reaction temperature, pH of the reaction solution, and particle size (Figures 43 and 44). Active sites for glucose oxidation were low-coordinate surface Au atoms at corners and step edges (entry 2, Table 4). Such nanoporous Au catalysts were less active than Au nanoparticles due to their lower density of the under-coordinated active sites. Nonetheless, such porous Au catalysts are amenable to modification to form nanocomposites or direct incorporation into microreactors.
Colloidal Au nanoparticles have also been studied in selox and compared to monometallic Pt, Pd, and Rh colloids and bimetallic Au−Pt, Au−Pt, and Au−Rh colloids for glucose oxidation. 324 Under acidic conditions, activity was low for Au and Pt (TOF = 51−60 h −1 ) and worse for Rh and Pd (TOF < 2 h −1 ), whereas bimetallic colloids were far superior (Au−Pt TOF = 295 h −1 and Au−Pd TOF = 92 h −1 ), highlighting strong synergies. Mirescu et al. compared Au/TiO 2 to Au colloids, 325 showing that both catalysts exhibited >99% selectivity toward gluconic acid (entry 3, Table 4). Direct use of unsupported colloids is problematic due to the difficulty in their recovery from the reaction media, although stabilizing polymers, such as naturally occurring chitosan, can overcome this. Supported Au catalysts are however preferable if commercial glucose oxidation is to be realized.
3.4.3. Au. Au/C catalysts have been investigated under pH control (7−9.5) and uncontrolled pH in aqueous glucose selox with O 2 under mild conditions (50−100°C, pO 2 = 1−3 bar). 326 Au/C offered total conversion and high selectivity at all pH values, and was more active at low pH than supported Bipromoted Pd and Pt catalysts. Au was also less sensitive to poisoning than the group VIII metals, but still underwent leaching and gradual deactivation over four runs.
The impact of mass-transport, 327 employing high glucose concentrations and O 2 partial pressures, 328 and the preparation method have been examined for Au/Al 2 O 3 , 329 and its performance has been compared to Au/TiO 2 and Au/C to determine their long-term stability, 330 and the role of particle size effects relative to Au/ZrO 2 , Au/TiO 2 , and Au/CeO 2 for glucose oxidation. 321, 331 Au on silica and mesoporous SBA-15 ( Figure 45) 332 were also compared to Au/TiO 2 and Au/Al 2 O 3 . Highest productivity (48 mol g Au −1 h −1 ) was obtained for Au supported on Al 2 O 3 -modified SBA-15 (entry 11, Table 4), attributed to gold stabilization by surface aminopropyl groups. All catalysts were selective to gluconic acid. Au nanoparticle  Chem. Rev. XXXX, XXX, XXX−XXX X synthesis by direct reduction was more effective than deposition of colloidal Au sols. Au/resins have also been tested for glucose selox 333 using ion-exchanged resins to obtain acid or base sites, and modified by, for example, amines to serve as stabilizers and reductants to create well-dispersed Au (Figure 46a). Catalytic performances decreased with decreasing basicity of anion-exchanged resins, while resin hydrophilicity enhanced activity (entry 6, Table 4). Operation under neutral conditions offers a more environmentally benign process, but catalyst stability was problematic. The same researchers investigated Au/cellulose for glucose selox (entry 5, Table 4), 319 with the biopolymer support stabilizing 2 nm Au nanoparticles (Figure 46b), through direct deposition by solid grinding with a volatile dimethyl Au(III) acetylacetonate precursor and subsequent H 2 reduction. These catalysts exhibited a high TOF of 11 s −1 in the production of sodium gluconate at 60°C and pH 9.5 (gluconic acid exists as the gluconate ion in aqueous solution at neutral pH). This activity is comparable to Au/C for similar size Au nanoparticles, which are easier to obtain with a cellulose support due to surface OH groups; however, the Au loading over cellulose was limited to 0.23 wt % due to the low cellulose surface area. As for carbon, the performance of cellulose supported Au nanoparticles was improved under basic conditions.

Chemical Reviews
Base-free glucose selox has been a recent focus of attention by Hutchings and co-workers, with 1 wt % Au nanoparticles supported over P25 titania by sol-immobilization or depositionprecipitation offering good conversion and excellent selectivity toward the aerobic selox of glucose to gluconic acid at 160°C and 3 bar oxygen. 334 Gluconic acid yields of 67% were obtained following mild calcination of the sol-immobilized Au/TiO 2 , attributed to removal of the PVA stabilizer ligands improving access to active sites without compromising particle size (optimum ∼2.5 nm), with negligible glucaric or glycolic acid observed. However, the role (if any) of the titania support remains unclear, and no kinetics were reported; hence the high selectivity to the single carboxylic acid product remains unclear.
3.4.4. Supported Alloys. A bimetallic Pd−Pt/charcoal catalyst, prepared from Pd−Pt colloids using a tetra-noctylammonium chloride (NOct 4 Cl) stabilizer, has been reported for glucose selox. 335 A synergy was found for Pd 88 Pt 12 /C impregnated with Bi, although it was also claimed that the surfactant stabilizer acted to modify catalytic reactivity, rendering it difficult to distinguish the roles of metal and organic components. Ag−Au bimetallic nanoparticles with a Au shell, 76 Ag−Au bimetallics formed from a physical mixture of monometallic nanoparticles, 121 Au−Pt bimetallics with a Pt-rich shell, 243 and Au/Pt/Ag trimetallic nanoparticles have all been explored for glucose selox. 336,337 Trimetallic systems were an order of magnitude more active than pure Au counterparts of the same particle size, and exhibited excellent stability. Comotti et al. also compared Au−Pt, Au−Pd, and Au−Rh free and supported colloids 324 Enhanced TOFs were obtained for the bimetallic catalysts, although similar performance was observed for mono-and bimetallic catalysts under alkaline conditions. Au−Pt bimetal nanoparticles with a Pt-rich shell were reported recently for glucose aerobic selox, 243 delivering 10-fold rate enhancements over similar size pure gold nanoparticles. Bimetallic Au−Pt catalysts prepared by the rapid injection of excess NaBH 4 into AuCl 4 − /PtCl 6 2− /PVP outperformed pure Au nanoparticles. High activity was attributed to the small particle diameter (∼1.5 nm) and the presence of negatively charged Au and Pt atoms due to electron donation from the    PVP, as shown in Figure 47a. In contrast, Au−Pt prepared by dropwise addition of NaBH 4 or alcohol reduction exhibited poor activity. Zhang et al. also reported that PVP-stabilized Ag core /Au shell bimetallic particles (1.4 nm) exhibited higher activity in glucose selox than Au−Ag particles prepared by dropwise addition of NaBH 4 (which had a larger mean size, entry 19, Table 4). 76,338 They proposed that charge transfer from Ag to Au neighbors was responsible for the improved selox (Figure 47b), although there was no spectroscopic evidence (e.g., XPS or valence band measurements) to support their hypothesis. Benkóand co-workers also studied Au−Ag alloys, in this case supported on SiO 2 , for glucose selox, 339 finding similar results, that is, that activity was dependent on particle size and the Au:Ag ratio, with Au−Ag alloys more active than pure Au nanoparticles (Ag nanoparticles being inactive). The alloy enhancement was however assigned to Ag active sites and associated increased O 2 activation, although excess Ag blocked surface Au sites, suggesting both elements are required to form the requisite surface active ensemble. Basefree glucose aerobic selox under extremely mild conditions (1 bar O 2 and 60°C) has also been demonstrated over bimetallic Au−Pd nanoparticles supported by sol-immobilization on calcined MgO. Pd strongly promotes gold, affording 62% conversion and 100% selectivity to gluconic acid following mild alloying treatment, although the composition of the alloy and surface termination were not reported. 320 Au/Pt/Ag trimetallic nanoparticles (TNPs) prepared by dropwise NaBH 4 addition (D) 337 were compared for aerobic glucose selox against other Au mono-and bimetallic catalysts (entry 20, Table 4). Bimetallic Au/Pt(7/3)(D) with a TOF of 4290 h −1 was twice as active as pure Au(D) (TOF = 2170 h −1 ), while Au 70 Pt 20 Ag 10 (D) (T-1 in Figure 48(left)) gave the highest TOF; activity was dependent on particle size and more importantly alloy composition (Figure 48(right)) with XPS and quantum-chemical DFT calculations implicating negatively charged Au; the latter differs from results obtained for Ag core / Au shell bimetallics 76 in which particle size effects were believed to be the dominant factor.
3.4.5. Process Considerations. The continuous selective oxidation of glucose to gluconic acid has been reported over 0.25 wt % Au/Al 2 O 3 using a continuous stirred tank reactor (CSTR) reactor shown in Figure 49 for 70 days operation at 40°C , pH 9, and 1 bar O 2 to study long-term catalyst stability, 340 with unreacted glucose recirculated into the reactor. After 5 days operation, the reaction attained a steady-state specific activity of 150−200 mmol min −1 g Au −1 . Reversible catalyst deactivation occurred using high glucose feed concentrations, with activity recovered upon lowering the glucose concentration, indicating the excellent stability and robust nature of this catalyst for industrial application.
Pelleted or formed catalysts with large particle sizes are necessary for continuous reactors to facilitate their separation and/or entrainment, for example, within a basket. 341 In such cases, support textural and mechanical properties are critical to maximize intraparticle diffusion and minimize catalyst friability. 342 Pelleted catalysts often employ eggshell-type catalysts to improve active site accessibility. Particle morphology was explored using slurry, pellet, and rotating foam catalysts for Pt− Bi/Al 2 O 3 -catalyzed oxidation of glucose to gluconic acid. 343 The rotating foam block exhibited superior gas−liquid and liquid−solid mass transfer coefficients, which increased the surface oxygen concentration and afforded higher glucose conversion at lower catalyst loadings than the slurry or pellet   catalysts. Further enhancements were obtained using washcoated versus sol gel prepared foam blocks; the former comprised sintered micrometer sized particles exhibiting both macro-and mesoporosity, allowing rapid oxygen diffusion and a high dispersion of the bimetal active phase.
3.4.6. Summary of Glucose Oxidation Catalysts. Glucose selective oxidation to gluconic acid (or gluconate) is generally conducted under ambient oxygen, which hinders the extent of metal nanoparticle surface oxidation. While the latter is of benefit for gold catalysts, wherein anionic gold is believed the active site in selox, the converse applies to Pt and Pd systems in which electron-deficient (PdO 308,344 and PtO 2 345 ) are strongly implicated by ex situ and operando X-ray spectroscopy. Unlike sugar isomerization, dehydration, and hydrogenation, selox occurs at lower temperature (35−60°C) and hence affords high selectivity to the desired gluconic acid in most catalyst systems, although almost all studies show that glucose selox is very pH-sensitive. For instances, Au on activated carbon is far more active under basic conditions, with TOFs reaching 1.5 × 10 5 h −1 (42 s −1 ) per surface Au atom for glucose selox at 50°C and pH 9.5. 346 This need for a strong alkaline medium makes reactor design and handling of the reaction media more hazardous. Solid base metal oxide supports such as MgO or hydrotalcites may thus be advantageous over, for example, carbons (see entry 12, Table  4), potentially obviating the need for additional pH control, although support stability is contentious with Mg 2+ leaching into the reaction media, opening routes to competing homogeneous sugar catalysis. 320,347 The importance of employing alkali-free routes to hydrotalcites was recently highlighted in Au-catalyzed 5-HMF selox over Mg−Al supports. 348 Au/cellulose, Au/resin, and Au/ZrO 2 also show remarkable TOFs (entries 5, 6, and 7, Table 4), as do Au-derived bimetallic Au/Ag or trimetallic Au/Pt/Ag and Au/Pt/Rh catalysts (entries 19, 20, and 21, Table 4). TOFs vary hugely across supported and colloidal catalysts, from <1 to 1.5 × 10 5 h −1 , although little information is provided on either mass balance or product selectivity in some cases (entries 18, 19, 20, and 21, Table 4), and the assumption is that only gluconic acid is formed. Kinetic studies are also limited, with an order of 1.5 in glucose concentration for Au/cellulose for concentrations between 2.5 and 10 wt %, whereas Au/Al 2 O 3 exhibited a reaction order of only 0.4 with respect to glucose concentration. Detailed kinetic and mechanistic investigations are still required to understand the respective roles of metal, support, and solution pH, and identify the optimum reaction conditions in terms of temperature and oxygen pressure. Recycle tests reveal that 0.45 wt % Au/TiO 2 (entry 4, Table 4) was the most stable glucose selox catalyst, with complete conversion and high selectivity observed for 17 consecutive runs. In contrast, Au/ MgO lost 70% activity after only three runs (entry 12, Table 4), and there is a notable dearth of reusability studies for colloidal catalysts.
3.4.7. Nonglucose Monosaccharide Oxidations. The selective oxidation of less common C 5 (arabinose, ribose, xylose, lyxose) and C 6 (mannose, rhamnose, and galactose) monosaccharides is also of interest for applications in the food, pharmaceutical, and cosmetic industries. L-Arabinose and Dgalactose are readily obtained from water-soluble arabinogalactans within woody biomass. 349 Selective aerobic oxidation of these monosaccharides catalyzed by supported Pd, Au, Pt, and bimetallic Pd−Au catalysts to the valuable arabinonic acid and galactonic acid products was comprehensively reviewed by Kusema and Murzin. 350 In brief, the relative oxidation reactivity of sugar isomers depends on the accessibility of the oxidizable carbonyl group. Saccharides can be classified as unprotected aldoses or (1-O)-protected aldoses, such as glucose or (1-O)glucose methyl-glucose, respectively, or ketoses such as fructose, sorbose, and sucrose.
Pentose and hexose oxidation to their aldonic acids is reported over Pd/Al 2 O 3 , Pt/Al 2 O 3 , and Au/TiO 2 , revealing striking (and unexplained) differences between the metals. 351 Au proved most active and selective for both pentose (arabinose > xylose > ribose > lyxose) and hexose (D-glucose, D-galactose, and D-mannose) oxidation, achieving >99.5% selectivity to the aldonic acids in every case. In general, Pd was less selective to the oxidation of C 5 than C 6 sugars, with the exception of xylose oxidation to xylose acid, which occurred with 99% selectivity (as was also observed over Pd/C 352 ), whereas Pt exhibited the reverse behavior.
Selective oxidation of the disaccharide D-lactose (comprising glucose and galactose units joined by a β-1,4-glycosidic linker) to lactobionic acid, an important antioxidant in food and medicine, has also been reviewed, highlighting the importance of tuning support porosity and acidity, and metal loading in Pdcatalyzed oxidation. 100 Au/TiO 2 353 and bimetallic Cu−Au/ TiO 2 354 are reported excellent catalysts for cellobiose oxidation to gluconic acid with 100% conversion and 89% selectivity at 145°C and 10 bar O 2 . Au/C is also active for cellobiose oxidation to gluconic acid under base-free conditions at 145°C and 5 bar O 2 , with surface phenolic functions on the carbon support providing acidity to hydrolyze the glycosidic bond while metallic Au catalyzing oxidation of the resulting glucose. 355 Excellent gluconic acid selectivity (approaching 80%) was observed, with the support pore diameter influencing activity likely through modifying the adsorption mode of cellobiose, glycosidic bonds being more readily activated in larger pores. Bifunctional strategies also employ Au supported on acidic multiwall carbon nanotubes, 356 polyoxometalates, 357,358 and Pt on sulfonated carbon. 359

FUTURE PERSPECTIVES
Catalytic routes for biomass conversion to renewable fuels and chemicals will underpin the valorization of bioderived feedstocks this century. The development of new heterogeneous catalysts tailored for biorefinery applications represents a grand challenge that requires collaboration across the bio-chemoengineering interfaces. The complex nature of biomass means that no single conversion technology based on thermocatalytic or biocatalytic processing alone will likely prove successful, with a shift toward toolkits based around (tandem) biochemo catalytic routes arising from collaborations between biochemists, catalytic and materials chemists, chemical engineers, and experts in molecular and process simulation to exploit innovative reactor designs and integrated process optimization and intensification.
Mechanistic studies will continue to be of growing importance in elucidating new reaction pathways and developing associated kinetic models to guide the design process. Insight into catalyst deactivation will derive from a broader application of operando spectroscopies of liquid-phase transformations, underpinned by in-silico insight into molecular adsorption and surface dynamics. Practical implementation of new catalytic processes will be supported by advanced online monitoring and control systems for faster reaction optimization to maintain catalyst performance. Many C 5 −C 6 sugar trans-Chemical Reviews Review DOI: 10.1021/acs.chemrev.6b00311 Chem. Rev. XXXX, XXX, XXX−XXX formations involve multiphase systems (e.g., gas−solid−liquid with cosolvents and pH regulation); hence multiscale modeling will become increasingly important. Process intensification 157 will also have a significant role in improving the energy efficiency of conversion and separation unit operations, for example, through the use of spinning-disc, 360 oscillatorybaffled, 361 or microfluidic 362 reactors. Catalytic hydrogenation and oxidation of sugars is still effected by noble metals, and hence strategies to reduce precious metal usage (e.g., by substitution with base metals or the use of single atom alloys 363 ) will likely come to the fore. However, early/mid transition metals are often more susceptible to deactivation than noble metals, and will require efficient reactivation/ recycling protocols to minimize waste. Improved and more user-friendly decision support systems for risk management of biorefineries, coupled with more extensive life cycle analysis tools, will also prove invaluable to ensure process economic viability and sustainability.

Process and Economic Considerations
While there are significant advances in catalytic technology for biomass conversion, widespread uptake and the development of next-generation biofuels and chemical feedstocks is ultimately hampered by uncertainty of industry over investment strategy. Progressive government policies and incentive schemes are essential to place biomass-derived chemicals on a comparable footing with heavily subsidized fossil fuel-derived resources. In the short-to medium-term, it is likely that so-called drop-in replacements of crude oil-derived chemicals, or methods to coprocess biomass in conventional oil refineries, will be most cost-effective to circumvent high capital costs of new processing facilities. Targeting processes that require drop-in chemical feedstocks to replace existing fossil-derived chemicals has the advantage of providing raw materials to mature markets wherein necessary infrastructure and technology exists, thereby maximizing the impact of value added chemicals. In parallel, bioderived chemicals such as lactic acid offer a new value chain of products that may offer enhanced performance and potential for business opportunities. Techno-economic analysis of biomass conversion processes will be essential to guide industry as to the most efficient processing conditions to employ. Biomass pretreatments employed to isolate sugars are among the most wasteful steps in a biorefinery, and hence an area where improvements in the atom and energy efficiency of lignocellulose processing are needed, such that preliminary acid hydrolysis/extraction of sugars from lignocellulose occurs under mild conditions with minimal waste generation. To overcome some of the infrastructure problems for biomass processing, economic models for biorefineries have proposed enhanced revenue that could be achieved by making more use of underutilized infrastructure in corn wet and dry mills, or pulp and paper operations.
To reduce dependency on fossil fuels, it is essential that the conversion processes used to produce fuels, chemicals, and materials from biomass are economically viable. The market for bioderived fuels and chemicals is linked to crude oil prices, which for biofuels only becomes competitive when oil prices are >$50−75 per barrel. 364 Techno-economic evaluation is critical when applying newly developed technologies, particularly when precious noble metal catalysts are involved. Insight into the challenges facing a heterogeneously catalyzed process utilizing bioderived sugars can be gained from related renewable technologies for producing methanol, dimethyl ether, or hydrogen, and fast pyrolysis bio-oil from biomass conversion. 24,369 These analogues highlight that biomass costs and capital costs are the two major factors that determine plant running costs for methanol synthesis (Figure 50a) and miscanthus pyrolysis (Figure 50b). 24 Plant running costs are also influenced by opportunities for the local sale of district heat to generate revenue, which is more significant for H 2 production due to the high levels of heat generated. 369 Oil pricing, and the availability of cheap H 2 from fracking, are a major barrier to biomass-derived routes to commodity chemicals such as methanol, DME, and H 2 , in the absence of green subsidies. Production costs of sugar-derived chemicals such as sorbitol and gluconic acid will be affected by similar factors, in addition to feedstock availability, number of operating units (e.g., catalysts, reactors, separators, and maintenance), energy consumption, waste treatment, staffing, and location proximate to infrastructure, but may be counterbalanced by the higher value of final product. Below, we estimate the cost of four metal catalysts used for the hydrogenation of glucose to sorbitol (Pt and Ru) or the oxidation of glucose to gluconic acid (Au and Pd). We adopt the method of Thielecke and co-workers, for continuous-flow glucose oxidation over 70 days, 340 to extrapolate the 70 day productivity employing 1 g of metal catalysts to the five examples and hence calculate the final cost of either sorbitol or gluconic acid (Table 5). For glucose hydrogenation to sorbitol, Ru catalysts (entry 2) are cheaper than Pt catalysts (entry 1), and there is no significant difference between batch (entry 2) versus continuous mode (entry 3) operation over Ru. For glucose oxidation, gold is more efficient than Pd in terms of both productivity and product cost, reflecting the current intensive focus on gold-catalyzed biomass transformations; continuous gluconate production over Au/Al 2 O 3 is much more cost-effective than batch production of the acid.
Reactor design and in particular process intensification will also enhance the commercial viability of biomass conversion technologies, 370 permitting more compact, cost-effective, and safer biorefineries. 371 Process designs wherein heat recovery and utilization are coupled will help to mitigate energy losses. 221 Because many of the reactions in this Review may be biphasic, reactor design will also need to account for the thermodynamics of phase separation and partitioning, which are significant technological challenges. Process design will require multiscale modeling for complex multiphase systems to include interactions at interfaces, solvent, and pH effects, along with new instrumental technologies and designs of new reactors. 361 Innovative separation processes will also be important for on-stream feedstock purification and product recovery, recycling of unreacted components, and the treatment of effluent streams in bio/chemical processes to meet product standards and environmental regulations. 372−374 Adsorption, ion-exchange, chromatography, solvent extraction or leaching, evaporation or distillation, crystallization or precipitation, and membrane separation are all expected to be featured in biorefinery plant designs. 292,375 The choice of separation technology will depend on intrinsic chemical properties and on economical flexibility. 376 Reactive distillation, combining chemical conversion and separation steps in a single unit, is a promising technology to reduce operating complexity and expense, 377 but necessitates a reaction media compatible with the temperature/pressure to effect distillation of the product (byproduct) of interest. Separation of thermally sensitive compounds, or macromolecules that have very low volatility, requires further advances in membrane separation and membrane reactors to become an attractive technology. 292,375 Hydrogen is critical to many of the steps in transforming C 5 −C 6 sugars into platform chemicals (and their products) and drop-in biofuels, and hence increasing demand is anticipated for renewable sources of hydrogen such as from water electrolysis (using renewable energy) or biomass gasification/reforming. 378 Use of molecular hydrogen is also predominantly accompanied by a requirement for noble metal catalysts and high pressure operation (>30 bar), together negatively impacting on the economics of hydrogenation. 379 Transfer hydrogenation offers an alternative, attractive route to in situ hydrogen generation, 380,381 permitting mild reaction conditions and obviating the need for pressurized reactor systems, such as in the cascade production of the renewable solvent γ-valerolactone via stepwise fructose dehydration to levulinic acid and subsequent transfer hydrogenation. 382,383 Formic acid is a particularly attractive hydrogen donor in this application 384,385 as a byproduct of levulinic acid production, and has been successfully applied in the one-pot cascade conversion of fructose to γ-valerolactone over Ru catalysts in the presence of mineral acids, 386 although the requirement for noble metal catalysts and corrosive nature of formic acid remains a challenge to catalyst stability/lifetime. Catalytic transfer hydrogenation employing an alcohol as both solvent and H-donor via the MPV mechanism is hence more favorable, being catalyzed by non-noble metal oxide Lewis acid/bases such as ZrO 2 . 387 Secondary alcohols are generally more efficient H-donors as compared to primary alcohols, but tertiary alcohols cannot serve as H donors due to the lack of an α-H. 380 Lewis acidic catalysts such as zeolites with tetravalent metal dopants, for example, Ti, Sn, Zr, and Hf, 379,388 and amorphous ZrO 2 or highly dispersed ZrO 2 on SBA-15 387,389−391 have shown remarkable activity in MPV reduction of levulinic acid. Moderate strength solid base catalysts are also reported as efficient catalysts for this reaction. 392 Direct γ-valerolactone production from C 5 −C 6 sugars would require coupling of a MPV process with an upstream Brønsted acid-catalyzed step for sugar dehydration. 393−395 It must be recognized that any transfer hydrogenation process requires either a low-cost, abundant bioderived source of sacrificial hydrogen donor, or routes to reuse any unreacted donor and regenerate the hydrogen donor from, for example, the reactively formed ketone byproduct.

Future Catalyst Development
Biomass pretreatments such as steam explosion 45 or enzymatic 61 and chemical (acid or base) 62,63 routes to fractionate and hydrolyze cellulose will ultimately produce aqueous sugar streams. From a practical and environmental perspective, the development of stable catalysts able to operate in the aqueous phase is hence critical. One of the most important challenges in the development of new catalytic materials is the design of catalysts with controllable active sites and strong stability that are robust under hydrothermal conditions. Support materials based upon niobia, titania, zirconia, or tungstates exhibit excellent stability under hydrothermal conditions, and catalysts based on such oxide supports will be of growing importance for high temperature sugar transformations. Templated porous carbons with controlled surface functionality will also likely come to the fore in aqueous phase sugar conversion due to their stability under acidic and alkaline environments. 396 A major influence on the selectivity of catalytic sugar transformations is the prevalence of simultaneous side reactions (e.g., isomerization accompanying dehydration or hydrogenation at high reaction temperatures) and attendant deactivation of active sites and accumulation of insoluble side products (e.g., solid humins and residual organic species). Several strategies may offer improved selectivity: (i) passivation of catalyst supports for hydrogenation and oxidation to minimize the contribution of supports in reactions; 258,397,398 (ii) development of materials exposing preferred crystal facets, which have different adsorptive or catalytic capacities; 398−400 (iii) development of catalysts with tunable hydrophobicity to minimize water inhibition and favor interaction with organic reactants in water; 401,402 and (iv) control over acid site distributions to tune Brønsted:Lewis acid ratio and strength. 403,404 Advanced nanomaterials with hierarchical macro-mesoporous or meso-microporous architectures offer a means to improve mass transport as compared to microporous materials through superior in-pore accessibility. 405 Besides improved diffusion, hierarchical porous materials are also effective in stabilizing highly dispersed active species. A recent study reports a new concept of catalyst design by spatially orthogonal chemical functionalization of the macroporous−mesoporous hierarchical pore network of silica, which offers new possibilities for cascade reactions. 406 Atomic layer deposition (ALD) is becoming a more mature technology for catalyst preparation that is underpinning breakthroughs in catalyst development. 407 ALD employs selflimiting chemical reactions between gaseous precursors and a solid surface to deposit thin films of single or bimetallic metals 408 or oxides. Precursors can infiltrate mesoporous materials, producing highly uniform, conformal thin films coatings on surfaces with atomic scale precision. 409,410 The development of a diverse range of coated support materials with desirable properties such as improved hydrothermal stability and/or acid:base properties for hydrothermal sugar conversion should be possible by the application of ALD to coating tailored porous architectures generated by dual templating methods.
To bridge the gap between homogeneous and heterogeneous catalysis and improve precious metal usage, single atom (site) catalysts offer exciting opportunities for application in sugar transformations. 411 While the concept of single site catalysts is common in framework solids or immobilized organometallic complexes, 412 single site catalysts based around atoms anchored to graphene are attracting significant attention. 413,414 The application of N-doped graphene encapsulated nanoparticles generated via pyrolysis of well-defined amine-ligated metal complexes 415 offers an interesting route to prepare nonprecious metal-based chemo-selective hydrogenation catalysts, with the selection of nitrogen ligands hoped to tuned catalyst performance. Such iron-based catalysts are active for transfer hydrogenation using formic acid, 416 and while not explored in sugar conversion would appear to be attractive materials to explore.
In catalytic C 5 −C 6 sugar reforming, many cascade reactions exist, for example, glucose isomerization to fructose over solid base or Lewis acid sites followed by fructose dehydration to 5-HMF over Brønsted acid sites, 154 and the stepwise hydrogenation of glucose to sorbitol over metal (Ni, Pt, or Ru) sites and subsequent hydrogenolysis to polyols over metal-acid bifunctional catalysts. 127 Isosorbide, an important intermediate for the synthesis of a wide range of pharmaceuticals, chemicals, and polymers, is another promising target for cascade synthesis directly from cellulose or sugars wherein metal-promoted solid acids show promise. 417−420 Such cascade reactions have great advantages with respect to atom economy, reducing time, labor and resource management, and waste generation. 421 One-pot catalytic synthetic processes are highly desirable in industry and necessitate the use of multifunctional catalysts. 422

CONCLUSIONS
Significant progress has been made in the development of heterogeneous catalysts and associated processes for the transformation of C 5 and C 6 sugars related to biorefinery applications. Waste-derived sugars are a promising feedstock for the production of renewable chemicals and advanced transportation fuels through low temperature, predominantly hydrothermal routes, which exploit advances in materials and surface chemistry to engineer tailored inorganic and hybrid inorganic−organic solid catalysts with tunable acid/base character and/or electronic/geometric properties. However, as with all areas of catalysis, efforts are required to standardize reaction conditions, reactor designs, and performance indicators to permit quantitative comparisons of disparate catalytic systems.
Aqueous phase sugar processing presents new challenges for heterogeneous catalysis related to the variable nature of the feed stream (in terms of component composition and concentration), solubility of sugars and reaction products, and the dissolution of reactive gases. 211,423,424 Biphasic media and cosolvents may serve to overcome some of these issues, and facilitate continuous processing and integrated product separation, although solvent selection must adhere to general green chemistry and sustainable technology principles, 425−428 being nontoxic, safe to handle, inflammable, and noncorrosive. Such aspects are particularly important when water itself is considered a finite resource, whose use in product isolation and purification must be minimized.

Notes
The authors declare no competing financial interest. Karen Wilson is Professor of Catalysis and Research Director of the European Bioenergy Research Institute at Aston University, where she holds a Royal Society Industry Fellowship. Her research interests lie in the design of heterogeneous catalysts for clean chemical synthesis, particularly the design of tunable porous materials for sustainable biofuels and chemicals production from renewable resources. She was educated at the Universities of Cambridge and Liverpool, and following postdoctoral research at Cambridge and the University of York, was appointed a Lecturer and subsequently Senior Lecturer at York, prior to appointment as a Reader in Physical Chemistry at Cardiff University.
Adam Lee is Professor of Sustainable Chemistry and an EPSRC Leadership Fellow in the European Bioenergy Research Institute, Aston University. He holds a B.A. (Natural Sciences) and Ph.D. from the University of Cambridge, and following postdoctoral research at Cambridge and Lecturer/Senior Lecturer roles at the Universities of Hull and York, respectively, held Chair appointments at Cardiff, Warwick, and Monash universities. His research addresses the rational design of nanoengineered materials for clean catalytic technologies, with particular focus on sustainable chemical processes and energy production, and the development of in situ methods to provide molecular insight into surface reactions, for which he was awarded the 2012 Beilby Medal and Prize by the Royal Society of Chemistry.