`E.I. du Pont de Nemours & Co. and
`Acher-Daniels-Midland Co. v. Furanix Technologies BV
`IPR2015-01838
`
`
`
`cr-rnrrczu. ENGINEERING nzszucr-r AND DESIGN 8 7 (2 o o 9) 1318-1327
`
`1319
`
`food regulations. Nevertheless all the potential technologies
`(whether approved for food or non-food production) need to be
`able to overcome the pH and temperature instability and lim-
`ited solubility ir1 organic solvents. It is because of the nature
`of glucose therefore that one obvious starting point is to use
`enzymatic catalysis (water based and under mild conditions).
`In this paper we will review the alternative technologies and
`routes from glucose to FDA, and discuss some of the limita-
`tions and drallenges.
`
`Biomass as a raw material for
`2.
`biorefmeries
`
`Nature is producing vast amounts of biomass driven by sun-
`light via photosynthesis:
`
`YICO2 + nH2O —> (CI-120),. + 7102
`
`However, utilization of biomass for producing chemicals and
`fuels is sfill in its infancy with only 3.5% being used for food
`or non-food purposes. Plant biomass consists mainly of car-
`bohydrates, lignin, protein and fats. Out of an estimated 170
`billion metric tons of biomass produced every year roughly
`75% are in the form of carbohydrates which makes biomass
`carbohydrates the most abundant renewable resource (Roper,
`2002). Tbgether with their amenability towards enzymatic pro-
`cesses this makes carbohydrates the center of attention when
`looking for new and greener feedstocks to replace petroleum
`for producing commodity chemicals as well as fuels. In plant
`biomass most of the carbohydrates are stored as sugar poly-
`mers such as starch, cellulose or hernicellulose.
`Starch is the second largest biomass produced on earth and
`commonly found in vegetables, such as corn, wheat, rice, pota-
`toes and beans. The total world production in 2004 was 60
`million tons of whidl more than 70% came from corn. Starch
`
`consists of chains of glucose molecules, which are linked
`together by u-1,4 and u—1,6 glycosidic bonds. The two major
`parts of starch are amylose (20—30%), essentially linear a—1,4
`glucan chains and arnylopectin (70—80%), a branched molecule
`containing 4—5% u-1,6 linkages.
`Starch is industrially hydrolyzed to glucose by the three
`enzymes: a-amylase, glucoarnylase, and pullulanase (Schafer
`et al., 2007). Bacterial a-amylases (EC 3.2.1.1) catalyze the
`hydrolysis of internal a-1,4 glycosidic bonds. This reduces
`the viscosity, which is necessary for further processing. Glu-
`coamylase (EC 3.2.1.3) is an exo-amylase that is added to the
`partly hydrolyzed stardr after liquefaction. Glucose units are
`removed in a stepwise manner from the non—reducing end
`of the molecule. The third enzyme is pullulanase (EC 3.2.1.41).
`Industrially used pullulanases are heat stable enzymes, which
`act simultaneously with glucoarnaylase during saccharifica-
`tion. Pullulanases catalyze the hydrolysis of the a-1,6 linkages
`in amylopectin, and especially in partially hydrolysed amy-
`lopectin. Typical process conditions for production of glucose
`from starch are given in Table 1.
`Cellulose is a glucose polymer consisting of linear chains
`of glucopyranose units linked together via B-1,4 glucosidic
`
`bonds. Unlike starch, cellulose is a crystalline material where
`inter- and intra-molecular hydrogen bonding gives rise to the
`very stable cellulose fiber. Hernicellulose is a polysaccharide
`consisting of short highly branched chains of different car-
`bohydrate units, including five- as well as six-carbon units
`(e.g. xyloses, galactose, glucose, mannose and arabinose).
`Hernicelluloses are much easier to hydrolyze than cellu-
`lose. The structured portion of biomass, such as straw, com
`stover, grasses and wood, is made of lignocellulose com-
`posed mainly of cellulose (30—60%), hernicellulose (20—40%)
`and lignin (10—30%). Both cellulose and hernicellulose con-
`sist of carbohydrate components whereas lignin is a highly
`branched aromatic polymer.
`Currently, there is intensive research on the use of lig-
`nocellulosic raw material as a biomass source for producing
`chemicals and fuels (as exemplified by many of the other
`articles in this special edition). However this research still
`faces considerable challenges due to lignocellulose being
`remarkably resistant towards hydrolysis and enzymatic attack
`(Peters, 2007). Energy demanding thermal pre—treatment oflig-
`nocellulose is necessary in order to break up the extremely
`stable cellu1ose—hemicellulose—lignin composites prior to
`adding cellulose-hydrolyzing enzymes and the current sit-
`uation does not allow the efficient use of lignocellulosic
`materials. Nevertheless, there is little doubt given the great
`abundance of lignocellulose that in the future this will become
`an attractive option. It is therefore important to continue to
`develop processes that can economically convert lignocellu-
`lose into chemicals. Moreover, glucose is one of the most
`abundant monosaccharides in biomass, accessible by enzy-
`matic or chemical hydrolysis from starch, sugar or cellulose.
`Furthermore, a range of chemical products can be obtained
`from glucose which gives it a key position as a basic raw mate-
`rial/building block.
`
`3.
`
`Glucose — a biorefinery building block
`
`Fermentation of polymer building blocks is already under
`commercial introduction. For example, Cargill produces lac-
`tic acid by fermentation and products based on polylactic
`acid are being introduced to the market. Several companies
`focus on succinic acid as a polymer building block, but also as
`a potential raw material for chemicals (e.g. butanediol). 1,3-
`propandiol is marketed by DuPont Tate & Lyle BioProducts
`for Soronam polytrirnethylene terephthalate (P'I'I‘) polyester.
`Likewise Cargill is working on developing 3—hydroxypropionic
`acid (3—HP). 3-HP is a potential raw material for existing
`chemicals such as propanediol and acrylic acid. Polyhydrox—
`yalkanoate (PHA) is marketed by Telles, a JN between ADM
`and Metabolix. Roquette, the French starch producer, has com-
`mercialized isosorbide, a derivative of sorbitol. Isosorbide is
`used as a co-monomer for high temperature polyethylene
`terephthalate. However, even if commercialization of polymer
`building blocks made by fermentation is commercially under-
`way, the technology has certain drawbacks such as loss of
`carbon as C02, low yields and difficult recovery of the products
`
`Table 1 - Process conditions for production of glucose from starch.
`
`Process
`
`Temperature (°C)
`
`Dry substance content (%)
`
`Jet cooking/dextrinization
`Saccharification
`
`105/95
`60
`
`30-35
`30-35
`
`pH
`
`5.2-5.6
`43-4.5
`
`Process time (11)
`
`0.1/1-2
`25-50
`
`
`
`1320
`
`chemical engineering research and design 8 7 ( 2 0 0 9 ) 1318–1327
`
`from the fermentation broth. The technology presented here
`(combined chemical and enzymatic catalysis from glucose)
`has the potential to overcome these problems and represents
`a promising next generation technology.
`One chemical transformation (besides fermentations) of
`carbohydrate monomers for the degradation of functional-
`ity is the dehydration reaction. This facilitates the removal of
`some of the functional groups in carbohydrates and allows the
`formation of defined building blocks. Triple dehydration of glu-
`cose yields HMF—a building block molecule that subsequently
`can be transformed into a multitude of bio-based chemicals.
`By a subsequent hydration reaction or an oxidation, HMF can
`be converted into levulininc acid or FDA, respectively. Both
`of these molecules are on the list of the 12 bio-based plat-
`form chemicals identified as being of highest potential to be
`converted into new families of useful molecules (Werpy and
`Petersen, 2004). In the following we will focus on the dehydra-
`tion of glucose to HMF as an example of the need to efficiently
`combine enzymatic aqueous processes with inorganic het-
`erogeneous catalytic processes that have so far mainly been
`developed for running reactions within the petrochemical
`industry.
`HMF is in itself a rather unstable molecule. It can be found
`in natural products such as honey and a variety of heat pro-
`cessed food products formed in the thermal decomposition
`of carbohydrates. Interestingly, HMF can be chemically con-
`verted into a range of other valuable chemicals. The oxidation
`of HMF is of particular interest. Here, the ultimate objective
`is to obtain FDA as suggested by Schiwek et al. (1991). The
`diacid can be used as a replacement for terephthalic acid in the
`production of polyethylene terephthalate and polybutylene
`terephthalate (Gandini and Belgacem, 1997; Kunz, 1993) which
`was recently reviewed by Moreau et al. (2004). The partially oxi-
`dized compounds can also be used as polymer building blocks
`
`although these are more difficult to produce selectively. FDA is
`a chemically very stable compound. Its only current uses are
`in small amounts in fire foams and in medicine where it can
`be used to remove kidney stones.
`Several extensive reviews describing the chemistry of HMF
`and its derivatives have been reported (see Fig. 1). The most
`recent review focuses on chemical transformation of biomass
`to a variety of chemicals with particular emphasis on the
`dehydration of monosaccharides giving either furfural (from
`pentoses) or HMF from hexoses, respectively (Corma et al.,
`2007). Moreau et al. (2004) described the recent catalytic
`advances in substituted furans from biomass and focused
`especially on the ensuing polymers and their properties. A
`review by Lewkowski (2001) on the chemistry of HMF and its
`derivatives also appeared recently. Two other relevant reviews
`are from Cottier and Descotes (1991) and Kuster (1990).
`The mechanism for the dehydration of fructose to HMF has
`been interpreted to proceed via two different routes; either
`via acyclic compounds or cyclic compounds (Haworth and
`Jones, 1944; Kuster, 1990; Van Dam et al., 1986; Antal et al.,
`1990). Besides HMF, the acid-catalyzed dehydration can lead to
`several other by-products such as insoluble polymers, called
`humins or humic acids. In an industrial process it is very
`important to find the right process conditions that avoid the
`formation of humins as these, besides lowering the selectivity
`of the reaction, potentially can clog up your reactor or deacti-
`vate the heterogeneous catalysts.
`In spite of all the research carried out within this area an
`efficient way of producing HMF or its corresponding dicar-
`boxylic acid, FDA, still remains to be found. Traditionally,
`chemists have been struggling with finding an inexpensive
`way of producing pure HMF. Given the immense field of its
`application, it is interesting that relatively few of the listed
`reviews have described the challenges that might be faced in
`
`Fig. 1 – HMF as a precursor for a range of commercial chemicals.
`
`
`
`cr-rnnczu. ENGINERING RESEARCH AND DESIGN 8 7 (2 oo 9) 1318-1327
`
`1321
`
`Table 2 — Typical reaction conditions for immobilized glucose isomerase.
`
`Process
`Isornerization
`
`Temperature (°C)
`50-60
`
`Dry substance content (9%)
`40-50
`
`pH
`7-8
`
`Process time (11)
`0.3-3
`
`a biorefinery manufacturing HMF or its derivatives. The most
`likely bioreflnery scenario will not be restricted to one product
`but make a series of high and low value products (including
`fuel). This allows the biorefinery to shift focus from one prod-
`uct to another if the market changes. In the case of I-IMF or
`FDA production this means that producing purely HMF or FDA
`is not the ultimate target and side-strearns producing other
`valuable products besides HMF or FDA can actually be of ben-
`efit One potential by-product of value is levulinic acid. This is
`formed via a rehydration of HMF to give levulininc acid along
`with formic acid. Both of these molecules are valuable prod-
`ucts that are potentially worth isolating as side streams. In
`this respect the goal of completely selective dehydration may
`in the future be misplaced.
`The synthesis of HMF is based on the acid-catalyzed
`triple dehydration of C6-sugar monomers, mainly glucose and
`fructose. However, various polysaccharides have also been
`reported as I-IMF sources (Rapp, 1987). The most convenient
`method for the preparation of HMF is by dehydration of fruc-
`tose. When starting from ketohexoses (such as fructose) the
`dehydration reaction proceeds more efficiently and selec-
`tively. This can be explained by aldohexoses (such as glucose)
`only being able to enolyze to a low degree which is consid-
`ered the limiting step in the production of HMI-' from glucose.
`However, glucose is the favored source of HMF due to the
`lower cost of glucose compared to fructose. Fructose may be
`obtained by enzyme or acid-catalyzed hydrolysis of sucrose
`and inulin orby the isomerization of glucose to fructose. Inulin
`is a linear B-2,1 linked fructose polymer which is temiinated
`by a single glucose unit. It is found as a food reserve in a
`number of plants including Jemsalem artichoke and chicory.
`Industrially fructose is produced from glucose by the enzyme
`glucose isomerase (EC 5.3.1.5). The equilibrium conversion
`under industrial conditions is 50% making chromatographic
`separation necessary in order to obtain the industrial product
`of 55% fructose, which has sweetness similar to sucrose. Glu-
`cose isomerase is used industrially as an immobilized enzyme
`with typical reaction conditions as shown in Table 2.
`Commercial immobilized glucose isomerase preparations
`used in a packed column have half-lives between 100 and
`200 days. Most columns therefore last for more than 1 year
`and productivities are typically around 15 tons of syrup dry
`substancdkg immobilized enzyme.
`
`4.
`
`Case studies
`
`4.1.
`
`Case 1: conversion of glucose/fructose to HMI-'
`
`To date most of the work regarding the add-catalyzed con-
`version of fructose, and to a less extent glucose, into HMI-'
`has been carried out in aqueous reaction media. Obviously
`water being very abundant and non-hazardous is the pre-
`ferred solvent of choice when exploring green and sustainable
`chemistry. Furthermore water is a good solvent for dissolving
`the monosacdraride substrates (fructose and glucose) as well
`as the product, HMF. However the dehydration of fructose to
`yield HMF in aqueous media is hampered by a competitive
`rehydration process resulting in the by—products levulinic acid
`
`and formic acid. In addition soluble and insoluble polyrner—
`ization products (humins), that are thought to arise from the
`self- and cross-polymerization of HMF, fructose and other by-
`products seem to be more pronounced in an aqueous reaction
`medium than an organic one (Van Dam et al., 1986). Neverthe-
`less, several interesting papers have been published on the
`dehydration of fructose into HMF. The conversion of glucose
`into HMF is more difficult and as a result there are only a few
`publications on this process.
`
`4.1.1. Aqueous media
`Several mineral acids such as HCI, H2504 and H3P04 have
`been employed in the homogeneous catalyzed dehydration of
`fructose to yield HMF (Newth, 1951; Medniclr, 1962; Roman-
`Leshkov et al., 2006). So far, however, the yield and selectivity
`of reactions carried out in aqueous reaction media are not
`comparable to those observed in aprotic high—boiling organic
`solvents such as DMSO where the solvent also serves as
`
`the catalyst (Musau and Munavu, 1987). Despite high yields
`and selectivity, the cost of removing high—boiling solvents
`makes these solvents unsuitable for indusnial and large-scale
`processes. Heterogeneous catalysts have, due to separation
`and recycling considerations, drawn more attention than
`homogenous catalysts. The use of various acidic heteroge-
`neous catalysts such as niobic acid (Nb205-nH20) and niobium
`phosphate (Nb0PO4) have been reported to have an intermedi-
`ate selectivity of about 30% for the production of HMF at about
`80% conversion offructose (Camiti et al., 2006). Zirconium and
`titanium phosphates/pyrophosphates have been shown to
`have a very high selectivity ofup to 100% at 100 °C in a period of
`18 min for the formation of HMI-' in water. However as the reac-
`
`tion time increases, the selectivity drops fast which is thought
`to be due to the formation of polymeric by-products. Addi-
`tionally, titanium oxides (T102), zirconium oxides (Zr02) and
`H-forrn zeolites catalyze the dehydration reaction (Moreau et
`al., 1996). Especially interesting is the direct conversion of glu-
`cose to HMF which can be enhanced up to 5-fold compared
`to the hydrothermal dehydration, by employing an a—TiO2
`at 200°C (Watanabe et al., 2005a,b). The main disadvantage
`with these catalysts seems to be the high temperature needed
`in order for the reaction to proceed without limited selec-
`tivity and conversion rates. Highly acidic cation-exchange
`resins such as those derivatized with sulfonic acid groups are
`also effective catalysts, providing the acidity of mineral acids
`together with the advantages of the heterogeneous catalysts
`(Rigal et al., 1981). These, often polystyrene based resins, can
`only tolerate temperatures up to around 130 “C, which reduces
`the range oftheir application. However this temperature range
`seems to be sufficient to overcome the activation energ bar-
`rier, when simultaneously applying the effect of microwave
`heating (Qi et al., 2008).
`
`4.1.2. Modified aqueous media and two-phase systems
`Phase modifiers have within the last couple of years proved
`very effective in promoting the conversion of fructose to HMI-'.
`Polar organic solvents that are miscible with water are added
`in order to increase the rate of the reaction to HMF and reduce
`
`the rate of the rehydration process forming by—products (Van
`
`
`
`1322
`
`chemical engineering research and design 8 7 ( 2 0 0 9 ) 1318–1327
`
`Dam et al., 1986). Commonly employed aqueous phase mod-
`ifiers are acetone, DMSO and polyethylene glycol (PEG) (Qi et
`al., 2008; Chheda et al., 2007; Van Dam et al., 1986). A further
`modification of the aqueous phase system is the introduction
`of a second immiscible phase to create a two-phase reaction
`system. An organic phase extracts the HMF from the aqueous
`phase as it is produced and consequently reduces the forma-
`tion of rehydration and polymeric by-products. Even with an
`initial concentration of fructose as high as 50 wt%, remark-
`able results with selectivity of 77% and a conversion of 90% at
`◦
`180
`C with HCl as the catalyst have been reported. In compar-
`ison similar conditions in water resulted only in a selectivity
`of 28% and a conversion of 51% (Román-Leshkov et al., 2006).
`
`4.1.3. Non-aqueous organic solvents
`Until now, the best results for the dehydration of fructose to
`HMF have been made in high-boiling organic solvents. The low
`concentration of water prevents the rehydration of HMF to lev-
`ulinic acid and formic acid. Iodine catalyzes the dehydration
`◦
`of the fructose part of sucrose in anhydrous DMF at 100
`C.
`Glucose is unaffected under the same conditions (Bonner et
`al., 1960). High selectivity has also been obtained when using
`PEG-600 as a solvent together with catalytic HCl. With the
`acid present a 1:1 solution of fructose and PEG-600 can be
`◦
`obtained at 85
`C (Kuster and Laurens, 1977). The first really
`high yields were reported by Nakamura and Morikawa (1980)
`using a strongly acidic ion-exchange resin as the catalyst in
`◦
`DMSO at 80
`C. These conditions gave a yield of 90% after
`8 h. The rate of the reaction was strongly affected by the type
`of resin used (Nakamura and Morikawa, 1980). Quantitative
`yields, without the use of a catalyst, were reported soon after
`◦
`in DMSO at 100
`C for 16 h (Brown et al., 1982). Good results
`were also obtained during an investigation of the optimum
`fructose concentration in DMSO. With 8.5 molar equivalents
`of DMSO with respect to fructose, a yield of 92% was obtained
`◦
`at 150
`C without any catalyst after 2 h (Musau and Munavu,
`1987).
`None of the above examples are suitable for production
`on a large-scale. High-boiling aprotic solvents such as DMSO,
`DMF and NMP are all miscible with water as well as many
`other common organic solvents. This makes separation of the
`desired products very difficult. Furthermore, both DMF and
`NMP are considered to be teratogenic.
`
`Supercritical/subcritical solvents
`4.1.4.
`Since the best results for the dehydration of hexoses to HMF
`have been in high-boiling organic solvents, the use of low-
`boiling solvents in their sub- or supercritical state would be
`an interesting alternative. Subcritical water has emerged in
`recent years as a feasible alternative to organic solvents at
`larger scale. Its unique intrinsic acidic and basic properties,
`makes it particularly interesting as a reaction medium for the
`dehydration of carbohydrates. When glucose is dehydrated in
`pure subcritical water, HMF is formed with greater selectivity
`than when using sulfuric acid or sodium hydroxide as cata-
`lysts under the same pressures and temperatures (Simkovic
`et al., 1987). Watanabe et al. (2005a) explored the use of differ-
`ent TiO2 and ZrO2 catalysts in highly compressed water. The
`anatase-TiO2 catalyst showed both basic and acidic properties
`and catalyzed the conversion of glucose to HMF. Yields were
`only about 20%, but the selectivity was more than 90%. The
`basic properties of the catalyst were thought to catalyze the
`isomerization of glucose to fructose, whereas the acidic prop-
`erties were thought to catalyze the dehydration (Watanabe et
`
`al., 2005b). Yields of up to 50% were obtained when using fruc-
`tose as the starting sugar and different zirconium phosphates
`as catalysts in subcritical water. No rehydration products
`were observed, yet the highest selectivity was not more than
`61%. By-products were humins and furaldehyde (Asghari and
`Yoshida, 2006). Interesting results have recently been reported
`on the catalytic effect of H3PO4, H2SO4 and HCl in the direct
`conversion of glucose to HMF in water at 523 K. It was con-
`cluded that the weakest acid, H3PO4, was the best catalyst for
`the conversion of glucose into HMF and the strongest acid,
`HCl, was the best catalyst for the conversion of HMF to lev-
`ulinic acid. The best yield for HMF was 40% (Takeuchi et al.,
`2008). More extensive studies on the kinetics of the dehydra-
`tion of d-glucose and d-fructose in sub- and supercritical water
`have been made as well as the behavior of HMF under similar
`conditions (Kabyemela et al., 1999; Asghari and Yoshida, 2007;
`Chuntanapum et al., 2008).
`Nevertheless, the overall results from sub- and supercrit-
`ical water have so far been unsatisfactory in terms of yields.
`Bicker et al. (2003) explored other low-boiling solvents such
`as acetone, methanol and acetic acid. An acetone/water mix-
`◦
`ture at 180
`C and 20 MPa gave 99% conversion of fructose
`and a selectivity of 77% to HMF. This excellent result was
`explained by the structural similarities between acetone and
`DMSO, which would promote the furanoid form of fructose
`and hence favor the formation of HMF. The authors also pro-
`pose a continuous process for the reaction (Bicker et al., 2003,
`2005).
`
`Ionic liquids
`4.1.5.
`Another attractive alternative to high-boiling organic solvents
`is the use of ionic liquids. Their unique physical properties
`such as negligible vapor pressure and non-flammability make
`them particularly suitable as solvents for large-scale produc-
`tion. There is a possibility to design and functionalize the
`ions of the ionic liquid, giving them ability to work both as
`solvent and reagent for certain reactions. There are several
`examples of ionic liquids that have the ability to solubilize nat-
`ural polymers such as cellulose, starch and chitin. This opens
`an excellent opportunity to convert crude biomass into fine
`chemicals (Liu et al., 2005; El Seoud et al., 2007).
`The first dehydrations of fructose and glucose with the help
`of ionic liquids date back 25 years. Fructose was dehydrated
`in the presence of pyridinium chloride to HMF in high purity
`with 70% yield. The corresponding result for glucose was only
`5% (Fayet and Gelas, 1983). In 1-butyl-3-methylimidazolium
`tetrafluoroborate and 1-butyl-3-methylimidazolium hexaflu-
`orophosphate, yields up to 80% from fructose were obtained
`using DMSO as a co-solvent and Amberlyst-15 resin as the
`catalyst. The DMSO helped to solubilize the starting fructose
`and the reaction was faster than in DMSO alone. Performing
`the reaction in 1-butyl-3-methylimidazolium tetrafluorobo-
`rate alone gave a yield of 50% within 3 h (Lansalot-Matras
`and Moreau, 2003). The best results so far from fructose were
`made by using the acidic 1-H-3-methylimidazolium chloride
`as reaction medium. This acted both as solvent and cata-
`◦
`lyst giving a yield of 92% after 15–45 min at 90
`C. There was
`no sign of HMF decomposition and glucose remained com-
`pletely unreacted (Moreau et al., 2006). Recently remarkably
`good results were found using the ionic liquid 1-ethyl-3-
`methylimidazolium chloride together with CrCl2, giving a total
`yield of 70% HMF directly from glucose and virtually no lev-
`ulinic acid. The authors propose that the actual catalytic
`−
`specie is the CrCl3
`ion formed together with the solvent
`
`
`
`chemical engineering research and design 8 7 ( 2 0 0 9 ) 1318–1327
`
`1323
`
`based on fructose (Halliday et al., 2003). Carlini et al. (2005)
`reported that HMF, as a starting reagent or produced one pot
`from fructose, was oxidized to the corresponding dialdehyde
`in water with methylisobutylketone (MIBK), as well as pure
`organic solvents, with vanadyl phosphate (VPO) based cata-
`lysts (Zr, Nb, Cr, Fe modified) as such or using a TiO2 support
`◦
`at 75–200
`C and 1 MPa. However, the reported yields were
`low (H20:MIBK = 0:30–5:30, HMF conversion 3–10%, selectivity
`to DFF 100–60%, respectively). Considering the oxidation as a
`stand-alone reaction and changing the solvents to less polar
`ones (benzene, toluene) better conversion rates and selectiv-
`ity were obtained, and using MIBK as a solvent lead to 98%
`conversion with 50% selectivity. However, in DMF the results
`◦
`are even better (at 150
`C) giving 84% conversion and 97%
`selectivity.
`
`4.2.2. Oxidation of HMF to FDA
`The above-described DFF may either be used as a valuable by-
`product or as an intermediate for obtaining FDA. On the other
`hand, catalytic reactions leading to the formation of FDA are
`also reported.
`Partenheimer and Grushin (2000) obtained DFF from HMF
`using metal bromide catalysts (Co/Mn/Zr/Br). The reactions
`were carried out in acetic acid at atmospheric pressure and
`also at 70 bar; the yields were 57% and 63% with the conver-
`sion of HMF 98% and 92%, respectively. Cobalt as a catalyst
`was also used by Ribeiro and Schuchardt (2003). Using cobalt
`acetylacetonate as a bi-functional acidic and redox catalyst
`◦
`encapsulated in silica in an autoclave at 160
`C, they obtained
`FDA, from fructose via HMF formation, with 99% selectivity to
`FDA at 72% conversion of fructose. By in situ oxidation of HMF
`to FDA starting from fructose, Kröger et al. (2000) described a
`way of producing FDA via acid-catalyzed formation and sub-
`sequent oxidation of HMF in a MIBK/water mixture using solid
`acids for fructose transformation and PtBi-catalyst encap-
`sulated in silicone and swollen in MIBK. The reaction was
`carried out in a reactor divided with a PTFE-membrane in
`order to prevent the oxidation of fructose. However, though
`in principle the integration process has been described, the
`yields remain quite low. The resulting yield of FDA was 25%
`based on fructose. In the oxidation of HMF to FDA the use
`of noble metals was first studied by Vinke et al. (1991). Here,
`mainly Pd, Pt, Ru supported on different carriers were used as
`the aerobic oxidation catalysts. Although all the noble met-
`als revealed catalytic activities, only Pt supported on Al2O3
`remained stable and active and gave quantitative yields of
`FDA. The reactions were carried out in water at pH 9 using
`◦
`a reaction temperature of 60
`C and a partial oxygen pressure
`of 0.2.
`
`4.2.3. Oxidation of HMF to FDA derivatives
`A new approach to the oxidation of HMF has been reported
`recently by Taarning et al. (2008) using methanol as both
`solvent and reagent. They performed a reaction with a gold
`◦
`nanoparticle catalyst in an autoclave at 130
`C and 4 bars of
`dioxygen, and obtaining FDA with 98% yield (according to GC
`analysis) and 60% isolated yield after sublimation.
`
`5.
`
`Process technology
`
`Table 3 indicates some of the key features of possible routes for
`the conversion of fructose to HMF. A number of observations
`can be made:
`
`Fig. 2 – Oxidation of HMF to DFF and FDA.
`
`and that it catalyzes the isomerization of -glucopyranose
`to fructofuranose, which is subsequently dehydrated to HMF
`(Zhao et al., 2007). Bao et al. (2008) concluded that ionic liquids
`with a Lewis acid moiety were more efficient than those with a
`Brønsted acid counterpart when dehydrating fructose. These
`ionic liquids were also successfully immobilized on silica, giv-
`ing a yield of up to 70% from fructose to HMF and completely
`retained their catalytic activity after five reaction cycles (Bao
`et al., 2008).
`
`Case 2: HMF oxidation to 2,5-diformylfuran and
`
`4.2.
`FDA
`
`FDA has been identified by the U.S. Department of Energy
`(DOE) biomass program as one of the 12 chemicals that in the
`future can be used as a feedstock from biomass in biorefiner-
`ies (Werpy and Petersen, 2004). Due to the presence of the
`two carboxylic acid groups, FDA is considered to be a biore-
`newable building block to form polymers from biomass and
`therefore become an alternative to terephthalic, isophthalic
`and adipic acids, which are all produced from fossil fuels. Sug-
`ars in the form of mono- and disaccharides are easily available
`from biomass. The hexose type monosaccharides such as glu-
`cose and fructose can be catalytically dehydrated into HMF
`(Corma et al., 2007; Gallezot, 2007; Moreau et al., 2004). HMF
`can then be oxidized into FDA using a variety of routes and
`reaction types with stochiometric amount of oxidants. Most of
`them are described in a review by Lewkowski (2001), including
`electrochemical oxidation, use of barium and potassium per-
`manganates, nitric acid and chromium trioxide. In this section
`we will focus on the recently reported catalytic routes for the
`oxidation of HMF into FDA.
`
`4.2.1. Oxidation of HMF to DFF
`Though production of FDA from HMF has been of great interest
`recently, there are few papers on catalytic aerobic oxidation of
`HMF. In the catalytic route to form FDA the partially oxidized
`intermediate 2,5-diformylfuran (DFF) is often observed (Fig. 2).
`The dialdehyde is a useful product to form other deriva-
`tives, and a number of studies have reported on the selective
`formation of DFF. Thus, Halliday et al. (2003) reported oxida-
`tion of HMF to DFF using an in situ reaction protocol where
`HMF was directly generated from fructose and not isolated.
`Hence, using ion-exchange resins and, then, VOP-type cata-
`lysts the authors obtained DFF with a maximum yield of 45%
`
`
`
`1324
`
`CHEMICAL ENGINEERING RESEARCH AND DESIGN 8 7 (2 0 o 9) 1318-1327
`
`Table 3 — Key features of possible routes for the conversion of fructose to I-IMF.
`
`Mode of
`operation“
`
`Catalyst”
`
`Temp.
`
`Fructose
`concentration
`
`Solvent media‘
`
`Highest
`yield
`
`Reference
`
`13
`B
`B
`B
`B
`
`B
`
`B
`
`B
`B
`C
`B
`C
`
`Hetero.
`Homo.
`Homo.
`Hetero.
`Hetero.
`
`Homo.
`
`80°C
`170°C
`90°C
`165 "C
`80°C
`
`180°C
`
`Hetero.
`
`90°C
`
`Hetero.
`
`Hetero.
`Hetero.
`
`110°C
`100°C
`85 °G
`100°C
`165 °c
`
`3-4% (wlw)
`10% (wlw)
`3—50% (wlw)
`10% (wlw)
`6% (wlw)
`3% (wlw)
`30% (wlw)
`50% (wlw)
`10% (wlw)
`30% (wlw)
`6—10% (wlw)
`6—10% (wlw)
`10-20%. (wlw)
`6% (wlw)
`05-3.5% (wlw)
`
`Water, MIBK
`Water, DMSO, MIBK, 2-butanol, DCM
`I-IMIM*CI‘
`Water, MIBK
`Wafef
`
`Water, DMSO, PVP, MIBK, 2-butanol
`
`Water, DMSO, PVP, MIBK, 2-butanol
`
`Water
`water, MIBK
`Water
`Water
`Water, MIBK
`
`41%
`87%
`92%
`69%
`42%
`59%
`76%
`71%
`59%
`54%
`31%
`74%
`26%
`85%
`-
`
`Carlini et a1. (2005)
`Chheda et al. (2007)
`Moreau et al. (2006)
`Moreau et al. (1996)
`Cartini et al. (2004)
`
`Roman-Leshkov et al. (2006)
`
`Roman-Leshkov et a1. (2006)
`
`Carlini et al. (1999)
`
`Benvenuti et al. (2000)
`Rivalier et al. (1995)
`
`‘ Process is continuous (C) or batch (B).
`" Catalyst is homogenous (homo.) or heterogenous (hetero).
`C Solent media are: methylisobutylketone (MIBK), dimethyl sulfoxide (DMSO), poly(1-vinyl-2-pyrrolidinone) (PVP), dicholorrnethane (DCM), and
`1-H-3-methyl imidazolium chloride (HMIM*C1').
`
`o Catalysttype
`A variety of catalysts like mineral and organic acids,
`salts, and solid acid catalysts such as ion-exchange resins
`and zeolites have been used in the dehydration reaction.
`The homogeneous acid-catalyzed processes are frequently
`associated with low selectivity (30—50%) for HMI-‘ at a
`relatively high conversion (50—70%) (Carlini et al., 1999).
`Moreover, problems related to separation and recycling of
`the mineral acid as well as of plant corrosion are expected.
`Thus, recent research has been based on heterogeneous
`acid catalysts which have considerable potential for indus-
`trial application (Carlini et al., 1999).
`o Mode of operati