Você está na página 1de 20

View Article Online / Journal Homepage / Table of Contents for this issue

CRITICAL REVIEW

www.rsc.org/greenchem | Green Chemistry

Glycerol dehydration to acrolein in the context of new uses of glycerol


Benjamin Katryniok,a ,b S ebastien Paul,a ,b,c Virginie Belli` ere-Baca,d Patrick Reye and Franck Dumeignil*a ,b
Received 9th July 2010, Accepted 5th October 2010 DOI: 10.1039/c0gc00307g Catalytic dehydration of glycerol to acrolein has the potential to valorise the glut of crude glycerol issuing from biodiesel production. This reaction requires catalysts with appropriate acidity, and intensive research activities have been focused on the application of families of catalysts including zeolites, heteropolyacids, mixed metal oxides and (oxo)-pyrophosphates, as their acidic properties are well-known. Nevertheless, their deactivation by coking remains the main obstacle in the way of large-scale industrial applications. Considering this important issue, various technologies have been proposed for regenerating the catalysts. This review shows that a well-balanced combination of an appropriate catalytic system together with an adapted regeneration process could put large-scale industrial applications within reach.

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

Introduction

The nite reserves of fossil-fuel feedstocks have encouraged intensive research activities for developing substitutes/additives for fuels such as biodiesel, bioethanol or biokerosene. Due to their origin from biomass, they have the strong advantage of a lower carbon footprint than fuels derived from fossil-fuel resources. However, from an economic point of view, it is not yet possible to produce any of these biofuels competitively. Nevertheless, political decisions have pushed the production of fuels from bioresources in order to be able to fulll the CO2 reduction objectives xed by the Kyoto climate protocol.1 For example, the European Union has planned a progressive increase in the mandatory proportion of bioethanol blended in gasoline, and of biodiesel blended in diesel fuel. The blended amounts must reach 10% and 7% by 2015, respectively.2 The alternative gasoline production route using vegetable oils and fats has attracted the attention of many academic and industrial researchers.3 The raw materials for biodiesel production are vegetable oils and fats from canola, soy, corn, etc. and a mono-alcohol (usually methanol), which is used to cleave the fatty acids from their glycerol backbone to yield the desired fatty acid esters (Scheme 1). These esters can be directly used as biodiesels, but they are commonly blended with diesels derived from fossil-fuels to meet the regulations.

The capacity of biodiesel production is expanding all over the world. For example, in 2008, the USA produced 2.3 million tonnes of biodiesel, while the EU produced 7.8 million tonnes (Fig. 1), and these values are expected these to double by 2012.4,5 This growth is accompanied by a signicant increase in glycerol production, as this a signicant by-product (~10 wt%) of the biodiesel process (Scheme 1). Simple projections enable one to forecast that 1.54 million tonnes of glycerol will be generated worldwide in 2015,6 all of which will have to be efciently processed in order to achieve a sustainable business.

Fig. 1 Comparative evolution of the quantities of biodiesel produced in the EU and the US.5,7

Scheme 1 Transesterication reaction of vegetable oils to yield biodiesel.

a Univ. Lille Nord de France, F-59000, Lille, France. E-mail: Franck.Dumeignil@univ-lille1.fr; Fax: +33 (0)3.20.43.65.61; Tel: +33 (0)3.20.43.45.38 b CNRS UMR8181, Unit e de Catalyse et Chimie du Solide, UCCS, F-59655, Villeneuve dAscq, France c ECLille, F-59655, Villeneuve dAscq, France d RHODIA France, 52 Rue de la Haie Coq, 93308, Aubervilliers, France e ADISSEO France SAS, Antony Parc 2-10, 92160, Antony, France

Depending on the process used for the cleavage of the fatty acids, the purity of the crude glycerol can vary greatly. The crude glycerol obtained from most of the conventional biodiesel processes contains ca. 80 wt% of glycerol, but it also contains water, methanol, traces of fatty acids as well as various inorganic and organic compounds (called MONG, denoting Matter Organic Non-Glycerol) (Table 1). As a consequence, in most of the cases, crude glycerol must be puried by an expensive distillation step prior to further use, to meet the requirements of the downstream processes. The proportion of glycerol that is rened is actually steadily decreasing, due to the high cost of the distillation step, together with a rapid growth of the quantity of crude glycerol produced, and also (primarily) because of the absence of any market able to absorb the massive overproduction (Fig. 2). In fact, an increase in the price of glycerol together with a sustainable and signicant demand would automatically result in a decrease in Green Chem., 2010, 12, 20792098 | 2079

This journal is The Royal Society of Chemistry 2010

View Article Online

Table 1 Examples of qualities of glycerol derived from some biodiesel production processes.8 Compositions are given in wt% Company production site Robbe/Diester Saipol/Diester Diester/Bioenergy Compi` egne (France) Rouen (France) Marl (Germany) Glycerol Water MONGa Salts Methanol 65 31 <1 3 (Na2 SO4 ) 0.3 93 4 <1 2.5 (NaCl) 0.2 85 10 <0.5 4.5 (NaCl) <0.01 Fig. 3 Distribution of the glycerol consumption by sector/ application.10

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

MONG = Matter Organic Non-Glycerol.

Fig. 2 Global crude glycerol production compared to the amount distilled. Delta is the amount of glycerol that is not upgraded, and usually burned.6,7

the production costs of the transesterication process, which up to now had to be entirely compensated for by the sales of biodiesel.9 New economical ways of using glycerol must therefore be developed in order to substantially increase the demand and the price of crude glycerol, and therefore to ensure the sustainability of the biodiesel sector.10 More than 1500 direct applications of glycerol are already known, especially in cosmetics, pharmaceuticals and food industries.11 The large versatility of glycerol use is based on both its chemical and physical properties (Fig. 3). Due to the presence of three hydroxyl groups, glycerol is completely soluble in water and alcohols, whereas it is completely insoluble in hydrocarbons. It is a very hydrophilic species, and is employed as

such when the amount of water has to be controlled, namely in glue or other adhesives. Further, the presence of hydroxyl groups leads to the formation of intra- and inter-molecular hydrogen networks, which explains its high boiling point (290 C at atmospheric pressure) and high viscosity. This latter rheological property is at the origin of the use of glycerol as a softener in resins and plastics but also as a lubricant, for instance in pharmaceutical applications. In addition, glycerol is non-toxic and has a sweet taste. It can thus be incorporated in food, medicines and cosmetics. However, as mentioned earlier, the crude glycerol from the biodiesel processes contains impurities, and is therefore not suitable for such applications without a preliminary purication stage. In addition, the size of the existing market is not sufcient to absorb the huge amount of glycerol currently produced, and the gap between the absorption capacity of the market and the amount of glycerol produced will increase in the near future if no new applications are found. Today, the crude glycerol that is not rened is generally burned (Fig. 2), which must be considered as a tragic waste of a potentially very useful organic raw material. Glycerol is a molecule with a large functionalisation potential that offers numerous opportunities for chemical or biochemical conversions for producing value-added chemicals. A non-exhaustive selection of these possibilities is shown in Scheme 2, and is further discussed below. For the moment, only a few applications utilise glycerol on a large scale. One of them is the halogenation to epichlorhydrin,

Scheme 2 Selected glycerol valorisation target molecules.

2080 | Green Chem., 2010, 12, 20792098

This journal is The Royal Society of Chemistry 2010

View Article Online

which is an important intermediate for epoxy resins. The process uses hydrochloric acid in the presence of organic acids (caprylic acid Solvay/acetic acid Dow)12 as catalysts working in the gaseous phase at 180220 C under a pressure of 15 bar.13 This technology was commercialised in 2007 by Solvay, which operates an existing production facility in France that was formerly used to produce glycerol from epichlorhydrin. Another commercialised process for large-scale consumption of glycerol is the reforming over PtRh catalysts to yield syngas. This latter can either be used in the FischerTropsch process for the synthesis of alkanes (Biomass to Liquid, BtL) or for the synthesis of methanol as performed by BioMCN in Delfzijl (Netherlands) with a capacity of 200 kt/year.1416 From this technology is derived the sustainable production of hydrogen an energy vector with an expected high potential in the near future through the water-gas shift process. As another option, the etherication to glycerol tert-butyl ether (GTBE) targets the classical petrol-based economy. Indeed, this compound is an environmentally friendly additive in gasoline, and can benecially substitute for the problematic methyl tert-butyl ether (MTBE). Nevertheless, the associated process is still far from reaching industrial applications.17 The selective reduction and esterication of glycerol are two processes with potential impacts on glycerol consumption at a medium scale. The former can be used to yield either propylene glycol (MPG) or 1,3-propanediol (PD), which are valuable intermediates in the polymer industry. The latter targets the esters of glycerol namely, monoacylglycerol (MAG) and diacylglycerol (DAG) that nd applications as emulsiers, e.g., in food (margarines and sauces) or in cosmetics. This process can either be catalysed by a conventional catalyst or by enzymes (lipase-type).18 Recently, the production of MPG has been commercialised by Synergy Chemicals with a production capacity of 30 kt/year.10,19 A small-scale application, but aiming at the synthesis of high-value ne chemicals, consists on the partial oxidation of glycerol to carboxylic acids (i.e., glyceric acid, tartronic acid and ketomalonic acid), aldehydes (i.e., glyceraldehyde) or ketones (i.e., dihydroxyacetone). The main challenge is to nd oxidation catalysts that are highly selective to the target molecule, among a large quantity of possible reaction products. One example is the selective oxidation to dihydroxyacetone over Bi/Pt catalysts, giving 37% yield at 70% conversion of glycerol.20 Oxidation can also be performed using genetically modied bacteria or electrochemical reactions.21,22 However, one of the most promising routes to glycerol valorisation lies in its catalytic dehydration to acrolein, which is an important industrial intermediate for the chemical and the agro-industries. The synthesis of acrolein is currently based on the selective oxidation of propene over complex multicomponent BiMoOx -based catalysts (Scheme 3). The selectivity reached in this process is close to 85% at 95% conversion.23 New approaches with propane as a starting material are currently being explored at the laboratory scale, but they suffer from insufcient yields, which are incompatible with further commercialisation.24 Con-

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

sidering the petrochemical feedstock depletion issues, sustainable resources will become more and more competitive, not to mention their positive effect in terms of impact on the climate. Within this context, a massive bioresourced and sustainable acrolein production is an important challenge, which has been accepted by academic and industrial researchers. An economics study has shown that competitive production of acrolein from glycerol could be possible if the price of glycerol became less than ca. 300 US$/t.25 In January 2010, rened glycerol still cost 500550 US$/t, but crude glycerol was only around 100 US$/t. This makes this latter a potentially very competitive raw material for acrolein production, even though some technical difculties in the application of crude glycerol have still to be solved, in particular avoiding catalyst poisoning by the impurities.26 Due to its toxicity, acrolein is usually directly converted to the desired high value-added derivatives. Most acrolein is used for the synthesis of acrylic acid, which is used, for example, as a starting material for synthesising sodium polyacrylate (Scheme 4). According to its physical properties, this polymer, classied as a superabsorbent, nds uses in hygiene products, such as diapers. The annual worldwide market for this superabsorbent polymer (SAP) is estimated at 1.9 million tonnes for 2010.27

Scheme 4 Industrial synthesis of sodium polyacrylate, a superabsorbent polymer.

The second largest consumer of acrolein is represented by the synthesis of DL-methionine via 3-methylthio-propionaldehyde as an intermediate (Scheme 5).28 DL-Methionine is an essential amino-acid, which cannot be synthesised by living organisms, and is widely used in meat production to accelerate animal growth. The annual worldwide production capacity of DLmethionine is about 500 000 tonnes.29 As natural methionine sources like plants and microorganisms provide concentrations that are too low, it has to be synthesised on the industrial scale. It is estimated that the global demand will increase by 37% in the near future.30

Scheme 5 Chemical pathways for the industrial manufacture of methionine.

DL-

Scheme 3

Catalytically-assisted selective oxidation of propene.

In the following sections, we will present the state-of-theart on the sustainable production of acrolein by gas-phase dehydration of glycerol. In fact, since our last review on the dehydration of glycerol,31 the number of publications in this Green Chem., 2010, 12, 20792098 | 2081

This journal is The Royal Society of Chemistry 2010

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

eld has greatly increased (Fig. 4). In particular, the fundamental scientic aspects of the reaction have been illuminated. Therefore, we will rst discuss the general properties needed to give high catalytic performances, and then we will describe in detail the three most common types of catalytic systems used in this reaction. The possibility of catalyst regeneration, such as the dehydration of glycerol in the presence of oxygen, will also be commented upon. After a short look at liquid-phase dehydration, the fundamentals of the reaction will be discussed in a section devoted to the insights in the understanding of the mechanism. Finally, the possibility of using crude glycerol will be discussed, before concluding the discussion and giving an outlook on the possible directions for further research efforts.

They reproduced the experiments over the lithium phosphate catalyst and compared its performance to those of a set of acid catalysts with well dened Hammett acidities (HAs) ranging from +3 to -8.2. They claimed that aqueous solutions from 10 to 40 wt% of glycerol could be converted at 300 C over aluminasupported phosphorous acid, to yield acrolein with a selectivity up to 75% at full glycerol conversion. The main by-product was acetol (hydroxyacetone), with a selectivity close to 10%. Neher et al. also pursued research on alumina-supported phosphorous acid, but with a view to producing 1,2- and 1,3-propanediol.37 The yield of acrolein formed as an intermediate was, however, a little lower (70.5%).

3 Acidity of the catalysts


Clearly, the acidity of the active phase is a crucial parameter that inuences the catalytic performance and stability. The rst quantitative studies on this effect were published by Dubois et al. in 2006.38,39 They carried out several tests with zeolites, Naon R , heteropolyacids (HPAs) and also with different types of acid-impregnated metal oxides. All these catalysts have a well dened HA ranging from -9 to -18. The selectivity for acrolein depended on the catalyst: For Naon R catalysts (HA = -12) and for tungsten oxide on zirconium oxide (HA = -14.5), the selectivity for acrolein was close to 70% at full conversion, while the selectivity over zeolitic catalysts (HA < +2) did not exceed 60%. The authors concluded that catalysts with an HA of -10 to -16 were the best candidates for the selective dehydration of glycerol to acrolein. Inspired by these rst results, a second research group focused its activities on the inuence of catalyst acidity. In 2007, Chai et al. published two papers on the sustainable production of acrolein from glycerol. Whereas the rst paper focused on niobium oxide as a catalyst,40 the second compared the catalytic performances of different types of acid catalysts.41 As the acidity of niobium oxide is inversely related to the calcination temperature of the hydrous niobium oxide precursor (as previously described by Chen et al.42 and Lizuka et al.43 ), this compound is particularly suitable as a model catalyst for studying the inuence of acidity on a given reaction. The best results were obtained for catalysts calcined at low temperature (250300 C). In addition to their high acidity, these catalysts are also advantageous as they have higher specic surface area than the catalysts calcined at higher temperatures. Nevertheless, the selectivity for acrolein barely exceeded 50% at 90% of glycerol conversion. In addition, the catalysts calcined at low temperature showed high carbon deposition, and were therefore subject to quick deactivation. From these observations, it was already possible to conclude that the strength of the acid catalyst must be carefully kept in a narrow range, because acids that are too weak give low selectivity, whereas acids that are too strong result in accelerated deactivation. Therefore, Chais team screened numerous catalysts, which were classied into four different groups according to their acidity.41 Group 1 contained basic catalysts with an HA higher than +7, like magnesium oxide. These catalysts showed no selectivity for acrolein at all. Group 2 contained catalysts with an HA between -3 and +7, like zirconium oxide. According to the work of Neher,35 these catalysts should give high acrolein This journal is The Royal Society of Chemistry 2010

Fig. 4 Evolution of the number of publications dealing with glycerol dehydration to acrolein (plotted using data collected using ScopusTM by performing a search with glycerol, dehydration, acrolein as keywords).

Historical context

It has been well known for a long time that heating glycerol induces its decomposition to acrolein and water, with a certain amount of by-products. However, an acid catalyst is needed to provide good control of the reaction and to obtain a signicant yield of the unsaturated aldehyde at a moderate temperature. The rst patent on this subject was published by Schering in 1930 in France.32 The reaction was carried out in the gaseous phase over supported lithium phosphate as a catalyst, with the yield of acrolein being close to 75%. Later, in 1934, Groll and Hearne from the Shell company applied for a patent for the dehydration of an aqueous glycerol solution in the presence of sulfuric acid at 190 C (supposedly the boiling point of the solution).33 The product was recovered in the gas phase with a yield of nearly 50%. Dehydration in the liquid phase was further studied and patented by Hoyt and Manninen, this time using a heterogeneous catalyst based on phosphorous acid supported on clay.34 It is interesting to note that these authors even at that time preferred petroleum oil as a reaction medium, due to its high boiling point of more than 300 C. Thereby, they could use higher reaction temperatures than Groll and Hearne,33 who were limited to 190 C, as the reaction was performed in the aqueous phase. This combination of a heterogeneous catalyst and a high boiling point reaction medium allowed them to obtain a yield of 72.3%. However, these early works in both the gaseous and the liquid phases remained undeveloped until the end of the 20th century, when cheaper glycerol from the biodiesel production process became available. In 1994 and 1995, Neher et al. published two patents35,36 as a follow-up of the pioneering work of Schering-Kahlbaum.32
2082 | Green Chem., 2010, 12, 20792098

View Article Online

selectivities, but, in practice, the selectivity did not exceed 30%, while the performances remained quite stable for 10 h on stream. Group 3, which was more promising, contained catalysts with an HA between -8 and -3. In this group, we nd alumina-supported phosphorous acid next to alumina-supported HPAs, niobium oxide (calcined at 400500 C), as well as HZSM zeolite and pure alumina. In good agreement with previous results,38,39 the selectivity was generally higher than that observed for Group 2 catalysts, with the exception of pure alumina and niobium oxide calcined at 500 C. Interesting results were also observed over the alumina-supported phosphotungstic HPA and over a mixed phase of tungsten oxide/zirconium oxide, with approximately 70% selectivity at 70% conversion in both cases after 2 h onstream. Unfortunately, these catalysts showed poor stability, and their performance signicantly decreased during time onstream (for both catalysts, the glycerol conversion dropped from 6869% after 1 h under reactant ow to 2325% at 10 h, the selectivity being more-or-less unaffected at roughly 6670%). The other catalysts were on the whole less efcient, and gave selectivities of 3555% at conversions of 55100% after 10 h onstream. In Group 4 were included catalysts with a HA less than -8, like Hb-zeolite, niobium oxide calcined at 350 C, and also alumina-silicate as well as sulfonated zirconium oxide. These catalysts were less selective to acrolein than those of Group 3, but their performances were less altered with time on-stream. As an example, the glycerol conversion over an aluminosilicate dropped only from 94% after 1 h under reactant ow to 75% after 10 h, the acrolein selectivity being more-or-less unchanged at roughly 4346%. Nevertheless, like niobium oxide, Group 4 catalysts were also subject to detrimental coke formation, with a carbon deposit of 100400 mg per gram of catalyst after ten hours on-stream. In addition to the acidic strength of the catalyst, the type of the acid sites present at their surfaces has an important inuence on their catalytic performance. When Chai et al. classied the catalysts according to their strength,41 they mixed up Brnsted and Lewis acid catalysts without differentiation. Whereas Brnsted acids donate a proton (as for the aforementioned silicotungstic acid, protonated zeolites or phosphorous acid), Lewis acids are acceptors for electron pairs (as for niobium oxide or pure alumina). From the catalytic results obtained by Chai et al. over these two different types of acids, one can conclude that they do not obey the same reaction pathway, as the authors systematically obtained rather low selectivities for acrolein over Lewis acids.41 A more detailed study on the different catalytic behaviors of Brnsted and Lewis acids in this reaction was proposed by Alhanash et al.,44 who compared a pure Brnsted acid catalyst (acidic caesium salts of phosphotungstic acid) to a pure Lewis acid catalyst (tin-chromium mixed oxide). They showed that: Lewis acid catalysts need a higher reaction temperature to be activated, due to a higher activation energy compared to Brnsted acid catalysts; Lewis acid catalysts give a larger selectivity for acetol, which is the major by-product observed during glycerol dehydration. These results were conrmed by Kim et al., who compared HZSM-5 aluminosilicates with different ratios of alumina and silica.45 This journal is The Royal Society of Chemistry 2010

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

According to Alhanash et al.44 the reaction over Brnstedtype acid catalysts starts with the protonation of the secondary hydroxyl group of glycerol by a proton from a Brnsted site, as postulated by Buhler for homogenous reaction media (Scheme 6a).87 The resulting intermediate is transformed by release of a hydronium ion (H3 O+ ) followed by a ketoenol rearrangement to give 3-hydroxypropionaldehyde and after a second dehydration step acrolein. The reacted Brnsted acid site can be regenerated (re-protonated) by the aforementioned released hydronium ion. The reaction over Lewis acid sites proceeds in a completely different way (Scheme 6b), as the glycerol interacts via concerted transfer of a terminal hydroxyl group on one of the metal centers and migration of the secondary proton to the other metal center. Consequently, 2,3-dihydroxypropene and pseudo-Brnsted sites are formed. The enol will tautomerise to acetol, whereas the pseudo-Brnsted site can either catalyse the dehydration reaction, as in the case of the aforementioned Brnsted sites, or can regenerate the initial Lewis site by thermal dehydration. Thereby, one can explain why Lewis acid sites generally show higher selectivity for acetol than Brnsted acid sites.

Scheme 6 Reaction mechanism over basic catalysts, as proposed by Kinage et al.44

As catalysts usually contain both acidic and basic sites, the inuence of the basicity must also be considered. Kinage et al. investigated the selective dehydration of glycerol to acetol and proposed a reaction mechanism over basic catalysts.46 In contrast to the mechanisms proposed over acid catalysts, the reaction over basic catalysts does not begin with a dehydration step but with a dehydrogenation step (Scheme 7). The resulting 2,3-dihydroxypropanal can either further react via consecutive dehydration and hydrogenation to yield acetol, or can form via a retro-aldol reaction formaldehyde and hydroxyacetaldehyde, which can be subsequently hydrogenated to ethylene glycol. With this reaction mechanism, one can explain the high selectivity for acetol observed by Chai et al. over basic catalysts.41 In this section, we have seen that the strength and the type of the catalyst acidity are of utmost importance for obtaining high catalytic performances. In the following sections, we will further discuss in detail the three most widely used catalytic systems for Green Chem., 2010, 12, 20792098 | 2083

View Article Online

Scheme 7 Reaction mechanism over protonated zeolite (H-MFI) catalyst as proposed by Yoda et al.46

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

the dehydration of glycerol: supported HPAs, zeolites, and the family of metal oxides, phosphates and pyrophosphates.

Supported heteropolyacids

The use of supported inorganic acids such as phosphorous acid or HPAs is one possibility to reach high performances in the dehydration of glycerol to acrolein, as shown by Neher and Chai.35,36,41 However, when using supported catalysts, new parameters in addition to acidity must be taken into account, namely the inuence of the support on the active-phase properties, the dispersion extent of the active phase on the support, and the distribution of the pore size of the support. In 2008 and 2009, Chai et al. published two papers dealing with zirconia- and silica-supported HPA catalysts.47,48 These studies showed that the nature of the catalyst support has a signicant effect on the thermal stability and on the dispersion of the Keggin-type active phase. The use of zirconia as a support led to better results, as far as the deactivation of the catalyst is concerned. The Keggin-anion density at the surface of the support was identied as being key for tuning the activity and the selectivity of the HPA for acrolein production, which is in good agreement with the conclusions of Ning et al., who used activated carbon-supported silicotungstic acid catalysts.49 The latter claimed that a 10 wt% silicotungstic acid-supported catalyst gives the best acrolein space-time yield ever reported in the literature (namely 68.5 mmol acrolein h-1 g-1 ), but this value is misleading, as the calculation is based on the mass of active phase and not on the mass of catalyst. This promising performance is attributed to the good dispersion of the HPA on the support surface and also to the relative quantities of strong acid sites. Our team also recently used silica-supported silicotungstic acid as a catalyst, but on supports that were
Table 2 Catalytic results obtained over various supported heteropolyacidsa Catalyst 15 wt% PW/SiO2 15 wt% PW/ZrO2 10 wt% SiW/AC 20 wt% PW/Al2 O3 20 wt% SiW/Al2 O3 30 wt% PMo/Q6 30 wt% PW/Q6 30 wt% SiW/Q6 30 wt% SiW/Q10 CsPW T react ( C) 315 315 330 275 275 325 325 325 275 275 mcat (g) 0.3 0.63 0.38 0.3 0.3 0.3 0.3 0.3 0.3 0.3 TOS (h) 10 10 5 5 5 5 5 5 5 1 C (%) 18 76 93 99 98 98 100 100 98 100

grafted with zirconia before impregnation of the active phase.50 We showed that the strong electronic interaction between the zirconia and the heteropolyanion results in a decreased strength of the Brnsted acid sites, whereby the long-term catalytic performances were increased in comparison to bare silica (69% acrolein vs. 24% respectively, after 24 h on-stream) due to less coking. Nevertheless, it is worth mentioning that the quantity of grafted zirconia and active phase necessitates careful balancing in order to avoid the introduction of undesired Lewis-acid character into the catalyst, which results in decreased selectivity for acrolein as shown by Alhanash et al.44 The inuence of the pore size distribution of the support was discussed for the rst time in 2007 by Tsukuda and coworkers,51 who worked with silica-supported acids. They focused their study on the inuence of the textural properties of the porous support on the selectivity, and used three commercial silicas with pore sizes of 3, 6 and 10 nm, namely CARiACT Q3, Q6 and Q10. The active phase was either a HPA or a conventional inorganic acid like phosphorous or boric acid. Good results were obtained using Keggin-type HPAs such as phosphotungstic acid (H3 PW12 O40 ) or silicotungstic acid (H4 SiW12 O40 ) deposited on the 6 nm silica support. At full conversion, the selectivity for acrolein was 65% and 74%, respectively, whereas the molybdenum-based homologous HPA (i.e., H3 PMo12 O40 ) did not yield high selectivity (34%) (Table 2). When a catalyst based on silicotungstic acid over silica with 10 nm pore diameter was used, the selectivity could be further increased to 86%, whereas for a support with smaller pores (namely 3 nm) the selectivity slightly decreased to 67%. Furthermore, it was observed that the small pore size of 3 nm resulted in a decreased catalytic activity (55% conversion), which most probably originates from enhanced coking resulting from steric limitations within the catalyst framework. The inuence of temperature was also

Sacrolein (%) 57 76 67 52 64 65 33 74 86 98

Sacetol (%) 7 12 8 8 8 7 5 7 7 nc

Y acrolein (%) 10 58 62 51 63 64 33 74 84 98

STY (mmol g-1 h-1 ) 0.6 1.6 11.1 5.1 6.3 3.8 2.0 4.4 5.1 49.0

Ref. 47 47 49 52 52 51 51 51 51 44

a T react = reaction temperature; mcat = mass of catalyst; TOS = time on stream; C = conversion of glycerol; Sacrolein = selectivity for acrolein; Sacetol = selectivity for acetol; Y acrolein = yield of acrolein; STY = space-time yield; nc = not communicated; PW = H3 PW12 O40 ; SiW = H4 SiW12 O40 ; CsPW = Cs2.5 H0.5 PW12 O40 ; AC = active carbon; Q6 = CARiACT-Q6; Q10 = CARiACT-Q10.

2084 | Green Chem., 2010, 12, 20792098

This journal is The Royal Society of Chemistry 2010

View Article Online

studied, 275 C being optimal in terms of acrolein yield, which is consistent with earlier studies on Brnsted acid catalysts. Similar results were obtained by Atia et al., who investigated the catalytic performance of the silicotungstic acid supported on alumina with 5 nm and 12 nm pore sizes.52 The reaction conditions for the catalytic tests were chosen such as to match those used by Tsukuda et al.,51 for the sake of easier comparison. The selectivity for acrolein increased from 65% for the alumina support with a 5 nm pore size to over 85% for the alumina support with a 12 nm pore size, both at full conversion and for catalysts loaded with 20 wt% of active phase (Table 2). Furthermore, Atia et al. found the same optimum reaction temperature as Tsukuda, since the alumina-based catalysts also showed complete conversion for a temperature higher than 275 C. These results show that the size of the pores has a direct inuence on the selectivity of the catalysts. If steric limitations like pores that are too narrow hinder the rapid desorption and diffusion of the reactants and of the products in the porous network of the catalyst, then condensation is more likely to occur, which results in lower selectivity for acrolein and deactivation of the catalyst due to carbon deposition (coke). From the aforementioned studies, one can identify two important parameters that govern the catalytic performances of supported inorganic acids: (i) the dispersion of the active phase and (ii) the textural properties of the support. These two parameters were further investigated by Chai et al. and Sato et al., who focused on supported HPAs and on their catalytic performances with regard to the textural properties of the porous support.5355 They carried out a huge number of performance tests, by varying both the loading of active phases and the nature of the supports. Here also, phosphomolybdic acid was unselective compared to phosphotungstic acid and silicotungstic acid.51 With 2030 wt% silicotungstic acid deposited on two porous silicas with pore sizes of 6 and 10 nm, respectively, they obtained a selectivity for acrolein up to ca. 87% at 275 C in both cases. Smaller pore diameters (3 nm) or a higher reaction temperature (300 C) always led to a decrease in the catalytic performance. It is worth mentioning that the inuence of the BET specic surface area of the support was also studied, but this proved to have only a small effect. Whereas Chai and Tsukuda51,53 showed that silicotungstic acid and phosphotungstic acid gave the best performances, due to their higher acidities, the group of Bruckner et al.56 recently decided to focus their investigations on supported phosphomolybdic (H3 PMo12 O40 ) and the vanado-phosphomolybdic (H4 PMo11 VO40 ) acids. Using electron paramagnetic resonance spectroscopy (EPR), they showed that the Keggin structure of the molybdenum compounds was partially destroyed upon impregnation on silica-alumina supports and that further calcination at 350 C led to total decomposition to simple metal oxides. These catalysts were then tested in the glycerol dehydration reaction, and the oxidation state of the metal atoms was followed by Operando EPR. The catalytic performance was close to that already observed for these compounds by Chai et al.,53 that is to say quite poor (the selectivity for acrolein did not exceed 25%). On the other hand, the Operando analysis of the catalyst revealed that the oxidation state of the metals was 3+ for vanadium and 4+ for molybdenum, which indicates This journal is The Royal Society of Chemistry 2010

that they were reduced during the reaction. When oxygen was co-fed, the reduction degree of vanadium was limited to 4+ and that of molybdenum to 5+. Furthermore, the coking was slowed down, and the authors claimed that the carbon was preferentially deposited on the vanadium centers. The carbon species constituting the coke layer was also analysed. According to the EPR signal, the carbon was mainly in the sp3 hybridisation state, which corresponds to non-double-bonded CC chains, like from the radical polymerisation of the C C bond in acrolein. Furthermore, if oxygen was present in the feed, the formation of C O carbonyl groups was observed. Another idea for using heteropolycompounds was investigated by Dubois et al., who used salts of HPAs obtained by complete or partial substitution of the protons of HPAs by metal cations.57 They used silicotungstic or phosphotungstic acid as bases, and prepared their respective salts using elements such as caesium, rubidium, calcium or bismuth, or transition metals like zirconium, lanthanum, iron or hafnium. The partially neutralised caesium-containing HPAs (Cs2.5 H0.5 PW12 O40 and Cs2.5 H1.5 SiW12 O40 ) and rubidiumcontaining HPAs (Rb2.5 H0.5 PW12 O40 ) required signicantly lower reaction temperatures (260280 C vs. 300350 C) than the totally proton-free compounds, and further showed higher selectivity for acrolein in oxydehydration (90% vs. 5070%). The best result was reported for Cs2.5 H1.5 SiW12 O40 , with a 93.1% yield of acrolein. Nevertheless, when this catalyst was used without co-feeding of oxygen the catalytic performance was much lower, with an acrolein yield of no more than 40%. Comparable results with caesium salts of phosphotungstic acid were reported by Alhanash et al., who also obtained acrolein yields of up to 98% during the rst hour of reaction, which is the highest performance ever reported, with a productivity up to 49 mmol g-1 h-1 (Table 2).44 Nevertheless, these authors admit that the catalyst deactivates rapidly, and that excellent performances claimed were observed only for the rst few hours of reaction. In conclusion, all these results suggest that the use of Keggin-type heteropolycompounds is a highly promising means of obtaining high catalytic performances. They offer many possibilities for tuning their acidity due to the possible variations in their composition, which can be adjusted by modifying the central atom, the addenda atoms and the counter-ions. When these compounds are deposited on a support, it is also possible to adjust an additional parameter, as one can even control the pore size of the catalyst, and thus reduce diffusion limitations.

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

5 Zeolites
The use of supported heteropolycompounds is not the only way of obtaining good catalytic performances. Zeolitic structures are promising, and have been well studied since the work of Dubois et al.39 In addition to ZSM-5 and b-zeolite (already patented by Dubois et al.39 ), Li and Zhuang et al. studied other zeolite structures such as MCM-49, MCM-22, MCM-56 and ZSM11.58,59 The obtained acrolein yields were in the range of 7085% for all the catalysts, with nearly no decrease in the catalytic performance during 400 h of reaction. In 2007, Okuno and coworkers patented metallosilicate catalysts with an MFI structure, a typical structure of zeolites Green Chem., 2010, 12, 20792098 | 2085

View Article Online

Table 3 Catalytic performances observed over various zeolitesa Catalyst b-Zeolite HZSM-5 MFI+Ga MFI (28)b MFI (200)b ZSM-5 (30)b ZSM-5 (150)b ZSM-5 (500)b HZSM-5 without Na T react ( C) 300 300 360 360 360 315 315 315 320 mcat (g) 4.2 6.4 9 9 9 0.3 0.3 0.3 1.2 TOS (h) 1.5 1 nc nc nc 2 2 2 10 C (%) 100 79 96 92 100 52 76 39 100 Sacrolein (%) 57 49 62 69 61 42 64 44 60 Sacetol (%) 10 6 nc nc nc 7 5 4 5 Y acrolein (%) 57 39 59 63 61 22 49 17 60 STY (mmol g-1 h-1 ) 5.8 2.6 4.6 4.8 4.6 17.0 37.9 13.4 0.3 Ref. 39 39 60 61 61 45 45 45 63

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

a T react = reaction temperature; mcat = mass of catalyst; TOS = time on stream; C = conversion of glycerol; Sacrolein = selectivity for acrolein; Sacetol = selectivity for acetol; Y acrolein = yield of acrolein; STY = space-time yield; nc = not communicated. b Si/Al ratio.

with a 3D porous network.60 In addition to aluminosilicates, they also prepared gallosilicates and ferrosilicates. Nevertheless, good catalytic performances were only obtained over the aluminosilicates and gallosilicates, whereas ferrosilicates were less selective due to their higher redox-character. The aluminosilicates offered the most stable performances without any ion exchange, with a selectivity for acrolein around 65%. In this patent, the results claimed over zeolitic catalysts were quite similar to those reported in Dubois et al.s work (i.e., 70% selectivity for acrolein at full glycerol conversion).39 In further publications, the same Japanese team reported the tuning of the acidity by modifying the Si/Al ratio of the zeolite. The best results were achieved using a ratio of 28, which resulted in yield of acrolein of 63% (Table 3).61,62 Kim et al. studied the inuence of the silica/alumina ratio in zeolites of the HZSM-5 type.45 They chose commercial HZSM5 catalysts with Si/Al atomic ratios ranging from 23 to 1000. A detailed characterisation of the acid properties using ammonia temperature-programmed desorption revealed an inverse relationship between the total acidity and the silica/alumina ratio. In fact, the catalyst with a silica/alumina ratio of 30 had ca. 9 times more acid sites (66 mmol NH3 g-1 ) than the catalyst with a ratio of 500 (7 mmol NH3 g-1 ). According to the observed desorption temperature, Kim et al. also showed that low silica/alumina ratios result in a higher number of strong acid sites. Nevertheless, these very strong acid catalysts showed only poor selectivity for acrolein (no more than 41.6% for the catalyst with a silica/alumina ratio of 30, for instance). Therefore, Kim et al. investigated the type of the acid sites and found that these low silica/alumina ratios lead to Lewis acid catalysts, whereas higher ratios give Brnsted acid catalysts, as already postulated by Alhanash et al.44 They concluded that a Brnsted acid catalyst with a high number of sites was more effective for obtaining a high acrolein yield. These two favorable conditions were gathered in a HZSM-5 with a silica/alumina ratio of 150, which gave 63.8% selectivity at 75.8% conversion at 315 C (Table 3). Furthermore, the inuence of the reaction temperature on the last mentioned catalyst was studied, showing that a reaction temperature below 315 C gave only low conversion (no more than 50.3%), whereas reaction temperatures above 315 C resulted in poor acrolein selectivity (no more than 57.4%). The idea of tuning the number of Brnsted acid sites by adjusting the silica/alumina ratio is one possible means of optimisation. An alternative was proposed by Schuths research
2086 | Green Chem., 2010, 12, 20792098

group, consisting of the replacement of the Brnsted protons at the catalyst surface by sodium cations through an ionexchange step.63 They used a commercial HZSM-5 catalyst with a silica/alumina ratio of 65 and varied the number of surface protons between 0.5 and 1. Unsurprisingly, the sodium-free catalyst exhibited the best performance in glycerol dehydration, with a selectivity for acrolein of 60% at full conversion (Table 3), which was explained by a higher number of acid sites, which were not replaced by sodium. Further, Zhou and co-workers proposed a synthesis of microand mesoporous ZSM-5 composites using dual templates, subsequently used as catalysts in the dehydration of glycerol to acrolein.64 The best result achieved with these catalysts was 73.6% acrolein selectivity at 98.3% glycerol conversion, which is quite a good performance. The inuence of the textural parameters is not clear for the zeolite compounds used, but the authors claim that mesopores are more favorable than micropores, as the latter are likely to induce diffusional limitations. A more detailed analysis of the impact of the pore size of ZSM-5 zeolites on the selectivity was performed by Pathak et al.65 They showed that the selectivity for acrolein increases with the pore size, whereas the selectivities to acetaldehyde, formaldehyde and acetol decrease in the same time. Nevertheless, an optimum pore size could not be reliably determined, as only four different pore sizes, namely 0.54, 0.74, 3.15 and 11.2 nm, were investigated. The use of zeolites as catalysts for the dehydration of glycerol offers another attractive means of accomplishing this reaction, linked with the possibility of tailoring the acid properties by tuning the bulk composition (silica/alumina ratio), by incorporation of transition-metal oxides (like gallium oxide or iron oxide) or by modication of the surface (ion-exchange). Furthermore, the structure and the mean pore diameter of the porous network of zeolites can be tuned, whereby as in the case of supported inorganic acids the diffusion limitations can be avoided.

6 Mixed metal oxides, phosphates and pyrophosphates


As a nal group of catalysts used for the dehydration of glycerol, we summarise in this section the results of various studies concerning metal oxides, phosphates and pyrophosphates. In addition to the aforementioned niobium oxide,40 the group of Chai studied binary mixtures of zinc oxide, tin oxide, zirconia, This journal is The Royal Society of Chemistry 2010

View Article Online

Table 4 Catalyst

Catalytic performances observed over metal oxides, phosphates and oxophosphates, with and without co-feedinga T react ( C) 315 315 315 315 280 280 300 300 300 280 300 280 mcat (g) 0.56 0.57 0.61 0.6 0.2 0.2 0.2 0.2 0.2 0.8 17 5 Co-feed O2 O2 O2 TOS (h) 10 10 10 2 1 1 10 10 10 5 5 nc C (%) 75 88 91 86 66 59 81 100 97 100 100 83 Sacrolein (%) 47 51 35 43 62 72 95 80 58 92 74 69 Sacetol (%) 10 12 14 17 10 7 9 6 15 0 11 10 Y acrolein (%) 35 45 32 37 41 42 77 80 56 92 74 57 STY (mmol g-1 h-1 ) 1.3 1.6 1.2 1.1 0.9 0.9 6.9 7.1 5.0 4.1 1.9 9.3 Ref. 40 40 40 66 67 67 68 68 68 72 38 79

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

Nb2 O5 -350 Nb2 O5 -400 Nb2 O5 -500 TiAl-600 SAPO-11 SAPO-34 VPO-700 VPO-800 VPO-900 Fex (PO4 )y 9 wt% WO3 /ZrO2 19 wt% WO3 /ZrO2

a T react = reaction temperature; mcat = mass of catalyst; co-feed = nature of the co-feeding gas; TOS = time on stream; C = conversion of glycerol; Sacrolein = selectivity for acrolein; Sacetol = selectivity for acetol; Y acrolein = yield of acrolein; STY = space-time yield; nc = not communicated; Mx Oy -*** indicates the calcination temperature (in C); TiAl = TiO2 /Al2 O3 binary mixture; SAPO = silica-alumina phosphate; VPO = vanadium oxophosphate.

titania, alumina and silica.66 Whereas the catalytic performances of these mixed oxides remained rather low, with acrolein yields not exceeding 36%, the study of the acidity and of the textural parameters clearly showed that the less acid catalysts have increased selectivity for acetol, and that small pore diameters induce a decrease in the selectivity for acrolein. Suprun et al. published a paper on phosphate-modied titania, alumina and silica/alumina (SAPO),67 and found a strong inuence of the acidic and textural parameters on the catalytic performances. As in the case of supported inorganic acids and of zeolites, a relationship between the total acidity and the selectivity for acrolein was observed. In fact, the most acidic catalyst, a silica-alumina phosphate, showed a selectivity for acrolein up to 72% (Table 4). Furthermore, Suprun et al. pointed ) are less active due out that microporous materials (56 A to internal diffusion limitations, and tend to induce increased carbon deposition, as observed on SAPO-34. A similar catalytic system, namely vanadium oxophosphates and oxo-pyrophosphates, was studied by Wang et al.68,69 As in the case of niobium oxide, the acidity of these compounds can be tuned by modifying their calcination temperature, but in an inverse relationship, as high calcination temperatures result in a larger number of acid sites. The catalysts were tested in the glycerol dehydration reaction, with co-feeding of oxygen. The best result was obtained using a vanadium pyrophosphate calcined at 800 C, which gave a selectivity for acrolein of 66% at full conversion (Table 4). Nevertheless, this result was not obtained over the catalyst with the largest number of acid sites. Actually, as the specic surface decreased for high calcination temperatures due to sintering of the solid, the selectivity consequently dropped drastically in this case. Dubois extended the study of pyrophosphates using boron and aluminium as cations, and observed that the boron-containing compounds give the highest acrolein yields, with a maximum of 77.8%.70 Other metal pyrophosphates were studied by Liu and coworkers, who used rare earth metals such as lanthanum, neodymium and cerium.71 All the considered pyrophosphates exhibited quite similar catalytic performances, with the best results being achieved over Nd4 (P2 O7 )3 . As in the case of the aforementioned vanadium pyrophosphates, the acidity of the neodymium comThis journal is The Royal Society of Chemistry 2010

pounds increases with the calcination temperature, whereby the optimum in the catalytic performance (80% yield of acrolein) was found for a calcination temperature of 500 C. The use of phosphates instead of pyrophosphates was studied by Deleplanque et al.72 They prepared iron phosphates using different preparation techniques, such as hydrothermal reaction or precipitation. The highest acrolein yield without oxygen cofeeding was 92.1% (Table 4). In the so-called oxy-dehydration reaction, i.e., with co-feeding of molecular oxygen, the selectivity for acrolein decreased to 62.5% at full conversion, due to the total oxidation to carbon oxide species (COx ) as an important side reaction. Therefore, Dubois investigated the addition of group 1 and group 2 metal cations to these catalytic formulations, such as caesium, strontium and potassium.73 Doping with caesium increased the acrolein yield to 72.2%, which was thus signicantly higher than that observed over the undoped iron phosphate (62.5%), but still lower than the yield obtained without oxygen co-feeding (92.1%). The idea of doping the phosphate species was also investigated by Matsunami et al., who used phosphatemodied silica with alkali salts. Nevertheless, in this latter case, the yield of acrolein did not exceed 67%.74,75 In 2006, Dubois et al. proposed the use of tungsten oxide on zirconia as a catalyst for glycerol dehydration. Total conversion of glycerol was achieved at 300 C with a catalyst containing 9 wt% of tungsten oxide, with a yield of acrolein of 74%.38 Redlingshofer et al. further studied this binary metal oxide system.76,77 The reaction was carried out at a lower temperature (260 C). Good performances were observed with an acrolein yield between 77 and 79%. Nevertheless, the catalysts tended to deactivate on-stream, with the acrolein yield decreasing at a rate of 5% every 10 h. Additionally, these authors described the possibility of regenerating the catalysts by oxygen posttreatment at 350 C for 5 h. Bythis treatment, they claimed that the initial catalytic performances could be recovered.78 This well-known combination of tungsten oxide on zirconia was further studied by Ulgen et al.79 They showed that the acidity of the solid increases with the amount of tungsten oxide. For the catalyst containing 19 wt% of tungsten oxide, an acrolein yield of 57% was observed (Table 4). Furthermore, they studied the inuence of the calcination temperature on the physical properties of the catalysts. Whereas Green Chem., 2010, 12, 20792098 | 2087

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

the temperature treatment had only a low impact on the acidic properties, the number of basic sites signicantly decreased when the calcination temperature increased. Moreover, the textural parameters were quite dramatically modied, as the sintering led to larger pores but reduced specic surface areas. For these two reasons, the thermal treatment had in any case a positive impact on the catalytic performance, as the decrease in basicity led to the suppression of acetol formation, and the larger pore size reduced the deactivation extent due to coking, as internal diffusion limitations were lower. The above-described family of mixed metal oxides, phosphates and pyrophosphates is rather heterogeneous. Some compounds, like niobium oxide, tungsten oxide and pyrophosphates, offer the possibility of controlling their acid strength via the calcination step. This thermal treatment generally also has an inuence on the specic surface area and the pore size due to sintering. In comparison to the aforementioned zeolites and supported HPAs, the control of the physical properties of the mixed metal oxides and phosphates seems nevertheless less straightforward.

Regeneration of spent catalyst

From the studies reported above, it is clear that very efcient catalysts for the dehydration of glycerol to acrolein can be prepared. Unfortunately, these catalysts quickly deactivate, most probably because of the formation of coke, which makes it difcult to straightforwardly use them in industrial plants at the current time. Up to now, only a few studies have been devoted to the characterisation of carbon deposits and to the sites responsible in these catalytic systems. Ere et al. used FTIR spectroscopy to characterise the spent catalyst and observed C O and C C signals.56 They attributed these signals to cyclic anhydride species and claimed that the Brnsted acid sites were the centers of coke formation. Suprun and coworkers conrmed that the acid sites were responsible for the formation of carbon deposits on phosphate catalysts.67 They further found that the quantity of coke increased with higher reaction temperatures and also with decreasing the pore diameter of the catalyst (Table 5). For example, the coke loading on SAPO-11 increased from 4.6 wt% at 280 C to 9.5 wt% at 320 C. From the ammonia desorption experiments, one can also see that the acid strength increases in the order Al2 O3 -PO4 TiO2 -PO4 < SAPO-11 SAPO-34, which correlates well with the amount of carbon deposit found over these catalysts. Indeed, the quantity of deposited carbon increased for all the reaction temperatures in the same order.

Furthermore, the hydrogen-to-carbon (H/C) ratio for the carbon species was determined by elementary analysis. All the catalysts showed a decrease in the H/C ratio at higher reaction temperatures, meaning that the carbon deposit became more decient in hydrogen (Table 5). For example, the H/C ratio over SAPO-34 decreased from 0.62 at 280 C to 0.47 at 320 C. This can be ascribed to the formation of highly condensed, unsaturated carbonaceous aromatic species. Solutions still have to be found to avoid or to limit the coke deposit on the catalyst surface or at least to optimise the regeneration of the catalysts. Although a fundamental approach is still needed for studying efcient means for limiting the coke deposition during the reaction, at least three kinds of solutions have been proposed so far for continuously regenerating the catalysts: (i) co-injection (co-feeding) of oxygen or hydrogen with the gas feed to yield in situ regeneration;39 (ii) cyclic regeneration of the used catalyst by injection of a ow or of pulses of air or oxygen;83 and (iii) circulation of the catalyst in a moving-bed reactor with regeneration in a separate parallel vessel (as for the FCC process).25 Whereas the rst option is associated with the risk of generating explosive conditions and/or to oxidise the various reaction products, the second alternative is accompanied by the disadvantage of a loss in productivity. The third option does not suffer from these drawbacks, but the construction and the operation of a circulating bed reactor implies serious technological difculties that can only be overtaken by skilled persons through ne process control. The idea of working with air in the reactant mixture was rst proposed by Dubois for catalysts of the zeolite type.39 To avoid the explosivity range, the oxygen fraction must never overpass 7 vol%. Therefore, the composition of the reaction feed was adjusted to 6% of oxygen for 4.5% of glycerol, the remaining 89.5% being steam. Thereby, the authors claim to reduce catalyst deactivation and even to inhibit the formation of by-products such as acetol in these conditions. The same process was further used for vanadium pyrophosphates,69 boron phosphates,70 iron phosphates,72 tungsten oxide on zirconia79 and caesium salts of phosphotungstic acid.57 As a general trend, the selectivity for by-products such as acetol, acetaldehyde and formaldehyde decreased, whereas the formation of organic acids, like acrylic acid, acetic acid and formic acid, increased, due to subsequent oxidation of the aldehydes formed. For some catalysts, like zeolites, vanadium oxophosphates, and tungsten oxide on zirconia, the selectivity for acrolein was not or only slightly affected. In the case of the caesium salt of phosphotungstic acid, an increase in the selectivity for acrolein by a factor of two was even observed (93%

Table 5 Correlation between acidity and the carbon deposit for phosphate catalystsa T react = 280 C Catalyst Al2 O3 -PO4 TiO2 -PO4 SAPO-11 SAPO-34
a

T react = 300 C H/C ratio 0.54 0.55 0.68 0.62 Coke loading (wt%) 3.9 6.8 7.4 12.7 H/C ratio 0.52 0.50 0.55 0.54

T react = 320 C Coke loading (wt%) 5.8 8.9 9.5 16.2 H/C ratio 0.49 0.48 0.40 0.47

Ac (mmol NH3 g-1 ) 295 258 1330 498

Dp (A) 111 101 6 5

SBET (m2 g-1 ) 118 38 172 49

Coke loading (wt%) 2.4 3.1 4.6 9.6

Ac = total acidity; Dp = pore diameter; SBET = specic surface area; T react = reaction temperature; SAPO = silica-alumina-phosphate.

2088 | Green Chem., 2010, 12, 20792098

This journal is The Royal Society of Chemistry 2010

View Article Online

Table 6 Catalyst

Comparison of catalytic performance observed in glycerol dehydration, with and without co-feedinga T react ( C) 300 300 300 300 300 300 280 280 280 280 280 280 260 nc 275 275 275 275 mcat (g) 4.2 4.2 6.4 6.4 0.2 0.2 0.3 0.3 1.3 0.8 5 5 23 23 0.3 0.3 9.2 9.2 Co-feed O2 O2 O2 O2 O2 O2 O2 H2 SO2 TOS (h) 3 1.5 1 1 nc nc nc nc 5 5 nc nc nc 1 5 1 24 24 C (%) 100 100 88 79 100 95 100 96 100 100 73 83 100 83 79 100 87 69 Sacrolein (%) 57 57 46 49 65 60 65 81 63 92 74 69 93 47 96 98 69 71 Sacetol (%) 0 10 2 6 6 12 8 6 0 0 8 10 nc 3 1 nc 7 9 Y acrolein (%) 57 57 40 39 65 57 65 78 63 92 54 57 93 39 76 98 60 49 STY (mmol g-1 h-1 ) 5.8 5.8 2.7 2.6 5.8 5.1 0.8 0.9 1.7 4.1 8.9 9.3 3.0 0.7 37.9 49.0 1.2 1.0 Ref. 39 39 39 39 68 68 70 70 72 72 79 79 57 57 44 44 81 81

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

b-Zeolite b-Zeolite HZSM-5 HZSM-5 VPO-800 VPO-800 BPO BPO Fex (PO4 )y Fex (PO4 )y 19 wt% WO3 /ZrO3 19 wt% WO3 /ZrO2 CsPW CsPW 0.5% Pd-CsPW CsPW WO3 /ZrO2 WO3 /ZrO2

a T react = reaction temperature; mcat = mass of catalyst; Co-feed = nature of the eventual co-feeding gas; TOS = time on stream; C = conversion of glycerol; Sacrolein = selectivity for acrolein; Sacetol = selectivity for acetol; Y acrolein = yield of acrolein; STY = space-time-yield; nc = not communicated; VPO = vanadium oxophosphate calcined at 800 C; BPO = boron phosphate; CsPW = Cs2.5 H0.5 PW12 O40.

vs. 47%; Table 6). On the other hand, for boron phosphates and iron phosphates, the co-feeding of oxygen led to a signicant decrease in the selectivity for acrolein, which was ascribed to the over-oxidation to carbon oxides (CO/CO2 ). To facilitate the activation of molecular oxygen, doping with metal cations like caesium, potassium, strontium, silver and platinum was probed by Dubois, which led to an acrolein selectivity of around 70%.73 A comparable approach was followed by Kasuga et al., who used protonated MFIs as catalysts.80 When they injected air in the gas reactant feed, the selectivity for acrolein on unmodied catalysts was rather low and did not exceed 45% after 24 h under reactant ow. To optimise the regeneration, Kasuga et al. thus also proposed to modify the MFI protonated-zeolite with a small amount of metal (Pt, Pd, Ru, Cu, Ir or Au). By this mean of doping, they claimed that it is possible to accelerate the splitting of the dioxygen used during regeneration.80 The best results were obtained with 0.1 wt% Pt and 1 wt% Au (80.7 and 79.7% of acrolein yield, respectively, at full glycerol conversion after 150 min under reaction conditions). The same concept to reduce the carbon deposit, but this time by co-feeding hydrogen, was investigated by Alhanash et al., who used caesium salts of HPAs.44 Following a similar principle as that proposed by Kasuga et al.,80 they probed the doping with small quantities of noble metals such as ruthenium, palladium and platinum. Thereby, they could increase the catalyst activity by a factor of two after 5 h of reaction (conversion of 79% vs. 41%; Table 6) without any signicant impact on the selectivity for acrolein (96% vs. 98%). The best results were obtained with palladium on the caesium salt of phosphotungstic acid, over which a productivity of 37.9 mmol g-1 h-1 was achieved. Dubois proposed a third kind of feed additive for enhancing the long-lasting performances of tungsten oxide on zirconia. Instead of oxygen or hydrogen, he injected 250 ppm of sulfur dioxide into the feed in order to slow down the deactivation.81 In fact, the addition of SO2 even induced a slight positive impact on the selectivity for acrolein (73% vs. 72% without SO2 ) and This journal is The Royal Society of Chemistry 2010

signicantly inhibited the formation of acetol (selectivity of 0.2% vs. 2.4% without SO2 ). After 24 h on-stream, the catalytic performance observed when co-feeding SO2 were signicantly improved, still with an excellent conversion of 87% (initially 100%), whereas it was 69% in the absence of SO2 (Table 6). The results obtained using co-feeding of oxygen are quite different and strongly depend on the studied catalytic system. As mentioned earlier, the oxygen concentration must be kept below 7 mol% in order to stay out of the explosion limits. Wang et al. studied the inuence of the oxygen concentration on the catalytic performance of vanadium phosphate catalysts. They observed that too little (as well as too much) oxygen has a negative impact on the selectivity for acrolein.69 In fact, low quantities of oxygen led to larger amounts of acetol, whereas high concentrations induced an increase in the selectivity for acetaldehyde, acetic acid, acrylic acid and of course to carbon oxides, which is explained by the successive over-oxidation of the products. The exceptions where the co-feeding process is unfavorable to the selectivity for acrolein as over iron phosphate and boron phosphate can be easily explained by the redox character of the catalysts, which promotes oxidation processes. In these cases, co-feeding with hydrogen might be a good way to overcome this problem. The process using co-feeding of sulfur dioxide might also be an interesting approach, even if one should keep in mind its toxicity, which is certainly the most important drawback of this method. Furthermore, this reagent may be incompatible with some catalytic systems, due to some possible poisoning issues. The second process for regenerating the spent catalyst namely, periodic regeneration was studied by Arita et al., who regenerated a used H-ZSM5 catalyst under an air ow. They evidenced the possibility of recovering the initial performances.82 However, the exothermicity of the regeneration led to the formation of hot-spots. Consequently, the temperature during regeneration exceeded the reaction temperature by more than 100 C, at which the catalyst risks thermal decomposition. Green Chem., 2010, 12, 20792098 | 2089

View Article Online

Therefore, this technique is not applicable for temperaturesensitive catalytic systems. A more academic approach was followed by Atia et al. Their preliminary work involved long-term runs (up to 300 h), and they found that over alumina-supported heteropolyacid catalysts the selectivity for acrolein remained quite stable, whereas glycerol conversion more-or-less linearly decreased. This effect is typically observed for deactivation by deposition of coke on the catalyst surface, and is linked with the progressive decrease in the number of accessible active sites. To verify this hypothesis, spent catalysts were regenerated under a ow of 1% of oxygen in nitrogen at 325 C for 24 h. After this treatment, the regenerated catalyst showed performances identical to those of the fresh catalyst. To quantify the coking effect, the spent catalyst was also analysed by thermogravimetry, which straightforwardly conrmed the hypothesis of carbon deposition.52 Meanwhile, Corma et al. adopted the idea of a circulating bed reactor,25,83 which was previously disclosed in a Dubois patent.38,39 These authors studied the opportunity of injecting crude glycerol directly into FCC plants. An advantage would be the use of existing facilities and therefore avoiding having to invest in building specic infrastructures. Note that this idea was already proposed, but under another form, by Dubois, who pointed out the possibility of injecting glycerol into propylene oxidation plants to concomitantly yield acrolein from both sources.84 In the FCC plants, the heat recovered by the burning of coke could be used to provide the energy necessary for the evaporation of glycerol. Corma et al. concluded that an autothermal process is possible in that case. Nevertheless, the authors did not limit their view to the dehydration reaction, but also investigated the possibilities for reforming glycerol at 500 600 C to produce ethylene and propylene. In this approach, the glycerol feedstock may be an alternative or a complement to naphtha cracking. An original approach was followed by Kasuga et al., who focused their investigations on various possibilities for pretreating catalysts (protonated MFI zeolite or Alox alumina).85 Three types of pretreatments were tested: (i) a ow of acetol, water and nitrogen; (ii) a ow of acrolein, water and nitrogen; and (iii) a ow of acetol, water and air. Irrespective of the pretreatment method, the Alox catalyst always led to very poor acrolein yields, while over the MFI catalyst all these

pretreatments induced higher acrolein selectivity in the rst few minutes after switching to the reactant ow. However, after 150 min on-stream, the catalytic performances returned to values similar to those obtained without any preliminary pretreatment, which means that this effect is only effective in the early stages of the reaction. Despite this, the authors pointed out the possibility of recovering the initial performance by regenerating the catalysts at 500 C under a ow of air.

Dehydration of glycerol in the liquid phase

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

As mentioned in the introduction, the rst experimental results on dehydration of glycerol in the liquid phase were reported by Groll in 1936,33 and by Hoyt and Manninen in 1951,34 in the presence of either homogenous or heterogeneous catalysts. Since these studies, less than a dozen articles and patents have been published on liquid-phase dehydration, while there has been a boom in the number of publications dealing with the gas-phase reaction. In the following section, we give a brief overview of the dehydration reaction in the liquid phase, and point out the major drawbacks of this process. Ramayya et al. studied the dehydration of glycerol in batch experiments using an aqueous solution of sulfuric acid, and conditions close to the critical point of water (namely P = 22.1 MPa, and T = 374 C).86 (Near) supercritical water offers the advantage that its physical properties, like the dielectric constant or the ion product, can be adjusted by varying temperature and pressure. The mixture was heated in a batch reactor at 350 C, pressurised to 34.5 MPa and acidied with 5 mM sulfuric acid. In these conditions, and at a residence time of 25 s, the conversion of glycerol and the selectivity for acrolein were 55% and 86%, respectively (Table 7). Considering the fact that sulfuric acid is soluble in supercritical water, this process can be classied as homogeneous catalysis with, the consequent catalystreaction mixture separation issues. Furthermore, a batch process is not applicable for large-scale production of acrolein. Nevertheless, the idea of using supercritical water as a reaction medium was a new aspect, and Buhler et al. decided to study it in more detail in 2001.87 Instead of working in a batch reactor, they built a ow-type reactor and used supercritical water both as a solvent and a catalytic system. They observed various decomposition products such as acetaldehyde, formaldehyde,

Table 7 Catalytic performances observed for glycerol dehydration in the liquid phasea Catalyst H2 SO4 H3 PO3 /clay H2 SO4 No catalyst ZnSO4 ZnSO4 H2 SO4 KHSO4 MgSO4 5 wt% H3 PO4 /Al2 O3 FePO4 CuPO4
a

T react ( C) 190 300 350 356 360 360 400 280 280 280 320 280

P (MPa) 0.1 0.1 34.5 45 25 34 34.5 0.1 0.1 0.1 0.1 0.1

Solvent Water Parafn Water Water Water Water Water Parafn Parafn Parafn Xylene Sulfolane

Reactor type Batch Batch Batch Continuous Continuous Continuous Continuous Continuous Continuous Continuous Continuous Continuous

C (%) nc nc 55 71 50 62 92 97 92 89 87 98

S (%) nc nc 86 38 75 59 81 82 56 57 78 85

Y (%) 49 72 47 27 38 37 74 80 52 51 68 84

Ref. 33 34 86 87 88 89 90 91 91 91 93 93

T react = reaction temperature; P = pressure; C = conversion of glycerol; S = selectivity for acrolein; Y = yield of acrolein; nc = not communicated.

2090 | Green Chem., 2010, 12, 20792098

This journal is The Royal Society of Chemistry 2010

View Article Online

allylic alcohol and propionaldehyde, as well as acrolein. The acrolein yield varied between 10 and 27% depending on the reaction conditions, with the best result (27%) obtained at a reaction temperature of 350 C under 45 MPa pressure (Table 7). As mentioned earlier, these tests were carried out in the absence of any catalyst or acid component, which explains the poor yields in acrolein. Recently, a combination of the ideas of Hoyt and Manninen34 and of Ramaya et al.,86 namely using a supported catalyst in supercritical water as a reaction medium, was proposed by Ott and coworkers.88 They investigated the dehydration of glycerol in a batch reactor under supercritical conditions using zinc sulfate as a catalyst. The choice of zinc sulfate as a catalyst is quite surprising, as this compound is not known as a strong acid catalyst, which is typically needed for this reaction. However, they justied their choice considering material stress issues: water in its supercritical form is very corrosive, and therefore requires the use of special and expensive steel grades for reactors; this corrosive power would become even stronger if acidic compounds were subsequently added to the medium, leading to unacceptable reactor material stress. In these conditions, the authors showed that supercritical conditions are not necessarily the optimal ones. Actually, the best catalytic performance, namely 75% selectivity at 50% conversion, was obtained in the near-critical region of water (310 C at 25 MPa) (Table 7). This might be explained by the degradation of acrolein when it is subjected to temperatures that are too high. In 2007, Lehr et al. also published results on the dehydration of biomass-derived polyols in sub- and super-critical water.89 As far as glycerol is concerned, they reported 59% glycerol conversion with almost 60% selectivity for acrolein over zinc sulfate as a catalyst under subcritical water conditions, like Ott et al. (Table 7).88 To determine if the dehydration of glycerol in the liquid phase might offer a sustainable source of acrolein at an industrial scale, Watanabe et al. attempted to optimise the acrolein yield.90 They combined the supercritical conditions in a ow-type reactor with sulfuric acid as a catalyst. Several blank tests performed in the absence of a catalyst gave low glycerol conversions, but high acrolein selectivities with a yield up to 16%. The best results were obtained at 400 C under 34.5 MPa with a sulfuric acid concentration of 5 mM. In these conditions, the yield of acrolein was 74% at a glycerol conversion around 92% (Table 7). One of the main issues in this work was to increase the formation rate of acrolein without modifying its decomposition rate, which was achieved by adding acid in supercritical conditions. In 2007, Suzuki et al. and Yoshimi et al. revamped the idea of Hoyt and Manninen,34 using a liquid reaction medium with high boiling point at atmospheric pressure.91,92 Potassium sulfate and bisulfate were used as catalysts in a batch reactor, where the glycerol solution was added dropwise in a parafn solution maintained at 280 C. The produced acrolein was continuously evaporated from the reaction mixture and was recovered in an aqueous solution, and yields close to 80% were achieved. This technique was further adapted by Takanori et al., who used iron or copper phosphate as catalysts in high-boiling organic solvents (e.g. n-octane, m-xylene or sulfolane) as reaction media.93,94 They obtained (to our knowledge) the highest ever reported acrolein yield in the liquid phase, over copper sulfate in sulfolane (84%; Table 7). This journal is The Royal Society of Chemistry 2010

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

Thus, we have seen that the dehydration of glycerol in the liquid phase (at atmospheric pressure or in near-critical or supercritical conditions) leads to the formation of acrolein. The application of supercritical conditions is an especially interesting concept the change in the physical properties of water inducing dehydration of glycerol even in the absence of a catalyst but, unfortunately, the yield of acrolein remains rather low in this case. On the other hand, the combination of acids and near-critical or supercritical conditions results in much better performances, but also in the generation of an extremely corrosive medium. Therefore, the reaction vessel, which must resist high temperature and pressure, must be specically designed, implying important investment/maintenance costs, which is a major drawback for industrial applications. Recent results show that the initial idea of using high-boiling liquids as reaction media at atmospheric pressure seems more preferable. Furthermore, the proposed process in this latter case, where the reactant is constantly fed into the reaction mixture and the acrolein is separated by evaporation upon formation, can be classied as a continuous process. Nevertheless, the accumulation of heavier by-products in the reaction mixture seems quite probable in this case. Moreover, the adaptation of this kind of process for industrial application is a real challenge, as the results from laboratory scale (300 mL) cannot easily be transferred to plant size. In conclusion, one can state that a more practical process seems to still be a topical issue of research before commercial applications of liquid-phase dehydration of glycerol to acrolein can be reliably envisaged.

Reaction mechanism

Whereas the research for an efcient catalyst can follow (at least in an initial screening step) a more-or-less applied approach, a more fundamental aspect lies in the understanding of the reaction mechanism, which further enables back-optimisation of the catalytic formulations. This requires the identication of the intermediate steps and the explanation of the formation of the by-products. A rst proposal was made by Buhler et al. for the glycerol activation in the liquid phase.87 Two pathways were claimed, either via an ionic or a radical mechanism (Scheme 8). The ionic reaction (Scheme 8a) begins with the protonation of glycerol either on one primary hydroxyl group or on the secondary hydroxyl group. Consequently, the elimination of a water molecule leads to the formation of a carbocation. In the case of the secondary carbocation, the only possible product after release of H3 O+ is acrolein, whereas either acrolein or acetaldehyde + formaldehyde can be formed starting from the primary carbocation. The radical pathway (Scheme 8b) starts with the abstraction of a hydrogen atom from a primary carbon using an OH radical. The resulting radical species further loses an hydroxyl radical (OH ), which leads to the formation of acrolein. Tsukuda et al. who performed the dehydration in the gaseous phase developed a more formal reaction framework (Scheme 9).51 The rst dehydration step leads to the formation of two enols, which are in tautomeric equilibrium with the corresponding ketone (acetol) and aldehyde (3-hydroxypropionaldehyde). The latter may subsequently react in a second dehydration reaction to yield acrolein, or undergoes a retroaldol reaction to Green Chem., 2010, 12, 20792098 | 2091

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

Scheme 8 Reaction pathways to the formation of acrolein considering (a) an ionic pathway and (b) a radical pathway.87

liberate formaldehyde + acetaldehyde, which in the presence of oxygen can easily be oxidised to acetic acid. In agreement with the previous proposal of Buhler et al.,87 Tsukuda et al. conrmed that the key to obtain high acrolein selectivity lies in the control of the rst dehydration step, whereby the formation of 3hydroxypropionaldehyde must be favored over the formation of acetol, which is identied as the main by-product of the process. The same mechanism has been further amended by Chai et al. (Scheme 9),41 who added two possible hydrogenation steps, one starting from acrolein and the other from acetol to yield allylic alcohol and 1,2-propanediol, respectively. In the catalytic system studied, these by-products can indeed be observed. Furthermore, the hydrogenation of formaldehyde to methanol or its decomposition to CO + H2 are mentioned.

Scheme 10 Reaction network proposed by Corma et al.25

Scheme 9 Mechanism proposed by Tsukuda et al. and further completed by Chai et al.41,51

These two rst proposals were the rst approaches for explaining the observed by-products from glycerol dehydration. Successively, Corma et al.25 postulated a more complex interwoven reaction framework (Scheme 10), where the main reaction steps are the same as those previously described by Tsukuda and Chai. In order to achieve a deeper understanding and specically investigate the secondary reactions, the authors used acetol as
2092 | Green Chem., 2010, 12, 20792098

a reactant at 350 C. The products were classied into three groups: acids (9.4%), other aldehydes (52.0%) and coke (27.7%). Furthermore, the formation of low amounts of acetone (4.0%), acetaldehyde (1.4%) and carbon monoxide (1.2%) was observed. Nevertheless, the conversion of acetol did not exceed 25%, which means that this compound is quite stable and does not easily undergo consecutive reactions. This explains why it is usually identied as the major by-product in the reaction of dehydration of glycerol to acrolein. Even though only small amounts of acetone were observed, Corma et al. investigated the consecutive reactions starting from this compound by repeating their method. The resulting product distribution was dominated by unsaturated hydrocarbons like butene (27.9%), propylene (1.7%), C5 + C68 aromatics (6.6%) and as a product of these compounds coke (31.9%). Other identied products were various acids (20.2%) and aldehydes (7.3%), but, as in the case of hydroxylacetone, the conversion of acetone remained low and did not exceed 14%. Nevertheless, with their ndings, they enlarged the reaction scheme proposed by Chai and Tsukuda by adding further side-reactions emanating from acetol. Thereby, they explained the formation of oligomers and coke by consecutive reactions starting from acetol via acetone and acetaldehyde. The idea of investigating the consecutive reactions starting from a by-product was picked up by Suprun et al. for 3hydroxypropionaldehyde and acetol.67 They could thus conrm the retroaldol reaction previously postulated by Tsukuda,51 which leads to the formation of formaldehyde and acetaldehyde issuing from 3-hydroxypropionaldehyde. Additionally, signicant amounts of coke were found, which was ascribed to the formation of cyclic C6 compounds identied by GCMS analysis (Scheme 11a). The authors further identied numerous furan derivatives as products from acetol (Scheme 11b). This journal is The Royal Society of Chemistry 2010

View Article Online

Scheme 11 Products derived from (a) 3-hydroxypropionaldehyde and (b) acetol.67

In the previous section, we have seen that the co-feeding of oxygen has a positive impact on the long-term performances of the catalysts. Nevertheless, the presence of oxygen also

results in a new type of product distribution. Indeed, new by-products, such as organic acids, are observed, and a new reaction scheme has thus been very recently proposed by Deleplanque et al. (Scheme 12).72 In addition to dehydration, which yields the already known products like acetol and acrolein, or their reductions to acetone and allylic alcohol, the authors postulate some new oxidation reactions. Starting from formaldehyde, originating from the aforementioned retroaldol reaction, oxidation leads to the formation of formic acid, which is easily over-oxidised to CO or CO2 . When acrolein is oxidised, acrylic acid is observed, which may further react to give acetic acid by decarbonylation. Alternatively, the latter can also be formed by oxidation of acetaldehyde. On the other hand, the formation of propanoic acid is explained by the reduction of allylic alcohol in a rst step and subsequent oxidation of the formed 1-propanol. In addition to these oxidation reactions, the authors also describe the formation of acetals from condensation of glycerol with formaldehyde. This may be possible when the glycerol concentration is rather high compared to the oxygen concentration, whereby the oxidation of formaldehyde to formic acid becomes limited. Nevertheless, one can conclude that the presence of oxygen results in a much higher number of by-products, which explains the lower selectivity for acrolein observed in the case of the dehydration reaction conducted in the presence of oxygen. All these reaction mechanisms are mostly formal and actually do not explain the real interactions with the surface of the solid catalysts. Therefore, Yoda et al. developed a mechanism for the dehydration over H-MFI zeolites inspired by the results of in situ infrared observations.95 The mechanism is similar to the aforementioned ones, whereby glycerol interacts with the bridging hydroxyl-group of the zeolite to give alkoxy species (Scheme 13). When glycerol is bound by a terminal OH-group, acetol is formed by the known ketoenol tautomerism. On the other hand, the interaction with the secondary OH-group leads to

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

Scheme 12 Reaction scheme proposed for dehydration of glycerol in the presence of oxygen.72

This journal is The Royal Society of Chemistry 2010

Green Chem., 2010, 12, 20792098 | 2093

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

Scheme 13 Reaction mechanisms over (a) Brnsted and (b) Lewis acid sites as proposed by Alhanash et al.95

the formation of an intermediate of 3-hydroxypropionaldehyde, and, after a second dehydration step, to the formation of acrolein. Based on these proposed reaction mechanisms, two research groups calculated the energy states for the different reaction pathways. Nimlos et al. concentrated on the calculations of the neutral and the protonated reaction pathways using the Gaussian 03 algorithm with B3LYP/6-311G sets for the molecular orbitals.96 For the so-called neutral glycerol, they found a very high activation barrier of 70.9 kcal mol-1 for the dehydration

of the secondary hydroxyl-group (Fig. 5). In contrast, the second dehydration step starting from the intermediate 3hydroxy-propionaldehyde required only 29.7 kcal mol-1 . The dehydration of the primary OH-group, leading to the formation of acetol, was found to be energetically even more unfavorable, with an activation energy of 73.2 kcal mol-1 . Nevertheless, the formation of acetol is exothermic and liberates 4.0 kcal mol-1 , whereas acrolein requires +9.3 kcal mol-1 . The calculations for the protonated glycerol as proposed by Buhler et al.87 show that the activation energies are much lower

Fig. 5

Calculation for neutral and protonated reaction pathway to acrolein and acetol.96

2094 | Green Chem., 2010, 12, 20792098

This journal is The Royal Society of Chemistry 2010

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

Fig. 6

Calculated energies for the glycerol dehydration over MFI-zeolite.97

than for the non-protonated one. Starting from the protonated secondary hydroxyl group of glycerol, the activation energy of the transition state of the rst dehydration is only 22.4 kcal mol-1 vs. 70.9 kcal mol-1 without proton. The second dehydration step, which is calculated to 24.6 kcal mol-1 in the presence of a proton, is also 5.1 kcal mol-1 lower than without (29.7 kcal mol-1 ). Furthermore, the reaction pathway for protonated glycerol leads to the exothermic formation of acrolein with an energy release of 18.0 kcal mol-1 . As in the case of the nonprotonated reaction, the formation of acetol requires a higher activation energy of 24.9 kcal mol-1 compared to that required for the formation of acrolein (22.4 kcal mol-1 ). Moreover, the differences in the nal energy states of the two products are quite low (acetol: -18.3 kcal mol-1 ; acrolein: -18.0 kcal mol-1 ). From these results, one can conclude that the formation of acetol is thermodynamically less favored than the formation of acrolein, due to the associated higher barrier of activation energy. Nevertheless, the difference is rather small, which suggests that it might be difcult to entirely suppress the formation of acetol. The calculations based on the results obtained for the reaction over ZSM-5 zeolite catalysts as proposed by Yoda et al.95 were performed by Kongpatpanich et al., who used Gaussion 03 with M06-2X/6-31G set for the zeolite 12T cluster model.97 For the rst dehydration step, they found an activation energy slightly higher than that obtained by Nimlos et al. for the protonated glycerol (41.4 kcal mol-1 vs. 22.4 kcal mol-1 (Fig. 6).96 The activation energy of the second dehydration was calculated at 38.5 kcal mol-1 and the nal energy state was +19.5 kcal mol-1 . It is difcult to explain these differing results and, unfortunately, Kongpatpanich has not yet calculated the reaction pathway leading to the formation of acetol. The mechanisms described give quite good explanations for the observed products in the glycerol dehydration reaction, such as acrolein, acetol, formaldehyde, acetaldehyde, 1,2-propanediol and coke. The recently published mechanisms over basic catalysts and over the different types of acid catalysts are the rst attempts to explain the different product selectivities observed over these types of catalysts. This journal is The Royal Society of Chemistry 2010

10 Direct use of crude glycerol as a reactant: an important economic issue


As mentioned in the introduction, the price of glycerol drastically depends on the required technical grade. Whereas rened glycerol cost between 500 and 600 US$ t-1 in January 2010, crude glycerol could be obtained at the same time for only 200 US$ t-1 .26 Nevertheless, the glycerol market has previously proven to be highly volatile, with costs up to 1800 US$ t-1 for rened glycerol and up to 500 US$ t-1 for crude glycerol in 20072008.26 It should also be mentioned that glycerol availability and price also depend on political decisions concerning biodiesel production and agricultural subsidies. Therefore, making possible the use of crude glycerol is a crucial issue for achieving a sustainable and economically viable process for the production of acrolein. However, as previously mentioned, crude glycerol is generally contaminated with by-products issued from biodiesel process (inorganic traces like salts and organic traces like water, esters, fatty acids and alcohols). The use of crude glycerol as a feedstock may therefore induce serious difculties, either by poisoning of the catalysts or by causing plugs due to deposition of highboiling organic materials or inorganic salts. To our knowledge, at this time no catalytic test using this type of feedstock has been reported in the literature all the disclosed studies were carried out using diluted aqueous solutions prepared from rened glycerol. Furthermore, it has been found that the yield of acrolein strongly depends on the glycerol concentration in the reactor feed, with a decrease in the yield with increasing concentration.35,36 However, the use of crude glycerol is of prime economical importance, and different ideas have already been proposed to avoid the costly distillation of crude glycerol and to overcome the obstacles raised by its use as a feedstock. Kijenski et al. proposed a modied evaporation system, where the crude glycerol is brought into contact with an inert liquid at high temperature.98 The glycerol is evaporated and fed to the reactor diluted in an inert carrier gas, whereas the impurities remain in the inert liquid. The authors describe an example of a process in which a solution of 75 wt% of glycerol with up to 3 wt% impurities is added to silicon-oil heated to 330 C. Green Chem., 2010, 12, 20792098 | 2095

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

To facilitate the use of crude glycerol, Dubois and Sereshiki et al. instead suggest a uidised inert solid heated to high temperature.99,100 This leads to the evaporation of glycerol, whereas the impurities remain inside the uidised bed, which can be continuously regenerated. The regeneration may consist in a ushing step with water to dissolve inorganic salts followed by a thermal treatment to burn off the organic deposits. As an example, the authors describe a process whereby a solution of 18 wt% glycerol with 2 wt% sodium chloride is put into contact with a uidised bed of silica particles. The uidisation is performed using nitrogen at 310 C. The author claims a separation/recovery of 99.9% of the introduced sodium chloride. Another process enabling the use of crude glycerol was also proposed by Dubois.101 As the evaporation of crude glycerol at high temperature induces reactions between glycerol and MONGs, this author proposed the possibility of forming acetals by reacting glycerol with the acrolein in a rst step (Scheme 14). These acetals have lower boiling points than glycerol (185 240 C vs. 290 C), and thus the aforementioned undesired sidereactions are said to be reduced. The formed acetals are then evaporated in the second step and subsequently dehydrated over an acid catalyst in the third step, to nally obtain acrolein. The acrolein formed is partially recycled for the acetalisation in the rst step.

by the use of heterogeneous catalytic systems are claimed to be lower than those from the conventional homogeneous process, as the leaching of the catalyst is negligible in this case. This process is planned or already in application in six units in Europe, and North & South America, with a total capacity of nearly 1 million tonnes per year. Nevertheless, for the near future, both homogenous and heterogeneous processes will still have to deal with the availability of different grades of crude glycerol on the global market. Therefore, the question concerning the economic sense of investing (or not) in a supplementary unit for the evaporation of crude glycerol is difcult to answer at this stage.

11

Conclusion

Scheme 14 Process based on an intermediate acetalisation step to purify crude glycerol.

In conclusion, one can see that the use of crude glycerol necessitates a supplementary upstream step at the level of the reactant evaporation. All the proposed concepts have in common the fact that inorganic impurities like salts are retained, whereby the poisoning of the catalyst is inhibited. Nevertheless, we have to mention that important progress has been made concerning the improvement of the quality of the glycerol issuing from biodiesel production processes.102 The early processes were based on homogenous catalytic cleavage of the vegetable oils and fats in the presence of a base for example, one of the new processes developed by the French company Axens uses a heterogeneous catalyst.103 This Esterp-H process provides exceptionally high glycerol purities close to 98%, with only a few traces of inorganic salts.104 Furthermore, the costs generated
2096 | Green Chem., 2010, 12, 20792098

The depletion of the petrol feedstock and the global warming caused by CO2 emission have led to a change in the energy policies. Biofuels have been proposed as an alternative to fossil fuels for reducing the impact of greenhouse gases. Due to political directives, the production of bioethanol for blending with gasoline and the production of biodiesel have drastically increased. As a consequence, the market has been ooded with crude glycerol co-produced in the biodiesel production process. As glycerol is a highly functionalisable molecule, many processes for its valorisation have been recently published. In particular, the catalytic dehydration to acrolein a precursor of sodium polyacrylate used as a superabsorbent polymer and of DLmethionine used for cattle feeding is of high economic interest. This reaction is catalysed over Brnsted and Lewis acid sites according to two different mechanisms, which are still a matter of debate and will undoubtedly be rened in the near future. Furthermore, the picture of the inuence of the acidity on the catalytic performances has become clearer, showing that optimal acidic strength is needed for high and constant performance. As far as gas-phase dehydration is concerned, a large variety of highly efcient catalysts has been proposed, which can be divided into three different categories: (i) supported HPAs; (ii) zeolites; and (iii) mixed-metal oxide type catalysts, including phosphates. The best catalytic performance has, up to now, been achieved with caesium salts of HPAs, yielding up to 98% acrolein in the very early stages of the reaction. On zeolitic and mixed oxide catalysts, the acrolein yield is generally slightly lower (around 70 to 80%) and the temperature needed to reach high conversions is higher (around 300 C vs. 275 C for HPAs). Nevertheless, the family of zeolites and supported HPAs offers many possibilities for tuning acidity and textural parameters, which makes possible ne-tuning to yield tailor-made catalysts. However, the main problem remains rapid catalyst deactivation due to carbon deposition. Up to now, no catalyst has had a half-life of more than a few days, which is of course not sufcient for commercial applications. Therefore, various approaches have been published to regenerate the coked catalysts. These are: the use of a uidised circulating catalyst, which is continuously regenerated in a parallel reactor; the introduction of oxygen, hydrogen or sulfur dioxide into the reaction feed for continuous regeneration of the catalyst inside the reactor; and discontinuous regeneration by alternating the glycerol feed and the oxygen feed. It was even proposed to dope the catalysts with noble metals for facilitating the splitting of the dioxygen used for regeneration. At This journal is The Royal Society of Chemistry 2010

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

this time, the co-feeding of oxygen seems to be the most practical way for extending catalyst life-time, as this technique requires relatively low investment and is easy compared to pulse-wise regeneration or to a uidised process. Nevertheless, this method is not applicable for all kinds of catalysts, as redox properties may lead to signicant decrease in the selectivity for acrolein. Liquid-phase dehydration was well studied with the rise of research into supercritical uids. Nevertheless, a dead-end has been reached, as dehydration just in supercritical water without acid catalyst results in low acrolein yields, while the combination of supercritical conditions with acids is problematic in terms of material stress on the reaction vessel, which would lead to signicant investment costs. The second and more promising route is the use of high-boiling solvents containing solid catalysts. With this technology, a quasi-continuous process can be achieved, even if scale-up to the industrial scale remains risky. As a consequence, the dehydration of glycerol in the liquid phase appears to be a rather uneconomic process. Finally, we have shown the need for the use of crude glycerol as a feedstock in regard to economic aspects. We have not come across any attempt to directly use crude glycerol over catalytic formulations, but two promising upstream technologies have been presented, which aim to eliminate the impurities of the crude glycerol during evaporation and thus avoid the costly crude glycerol distillation. As the number of publications concerning this kind of technology is still very limited, it is difcult at the moment to foresee which of them is the most promising. Furthermore, the quality of crude glycerol from biodiesel production is constantly increasing, reaching a purity up to 98% without any inorganic salts, which makes industrial applications of the proposed purication techniques rather improbable, at least in the long-term.

References
1 United Nations Framework Convention on Climate Change, http://unfccc.int. 2 Directive 2003/30/EC of the European Parliament and of the council of 05/08/2003 on the promotion of the use of biofuels or other renewable fuels for transport. 3 L. C. Meher, D. V. Sagar and S. N. Naik, Renew. Sustain. Energy Rev., 2006, 10, 248268. 4 Eurostat European Commission, http://epp.eurostat.ec.europa.eu. 5 U.S. National Biodiesel Board, http://www.biodiesel.org. 6 Global Industry Analysts, http://www.strategyr.com/Glycerin_ Market_Report.asp. 7 European Biodiesel Board, http://www.ebb-eu.org. 8 J.-L. Dubois and G. S. Patience (Arkema), WO2009044051, 2009. 9 M. J. Haas, A. J. Mc Aloon, W. C. Yee and T. A. Foglia, Biores. Tech., 2006, 97, 671678. 10 M. Pagliaro and M. Rossi, in The Future of Glycerol, 2nd edn (RSC Green Chemistry Book Series), 2010. 11 C. S. Callam, S. J. Singer, T. L. Lowary and C. M. Hadad, J. Am. Chem. Soc., 2001, 123, 11743. 12 D. Siano, E. Santacesaria, V. Fiandra, R. Tesser, G. Di Nuzzi, M. Di Serio and M. Nastasi (ASER), WO 2006111810, 2006. 13 D. Schreck, W. Kruper, F. Varjian, M. Jones, R. Campbell, K. Kearns, B. Hook, J. Briggs and J. Hippler (DOW Chemicals), WO 2006020234, 2006. 14 D. A. Simonetti, J. Ross-Hansen, E. L. Kunkes, R. R. Soares and J. A. Dumesic, Green Chem., 2007, 9, 1073. 15 R. R. Soares, D. A. Simonetti and J. A. Dumesic, Angew. Chem. Int. Ed., 2006, 45, 3982. 16 BioMNC, http://www.biomnc.eu.

17 M. Di Serio, L. Casale, R. Tesser and E. Santacesaria, Energy Fuels, 2010, DOI: 10.1021/ef901230r, in press. 18 N. O. V. Sonntag, J. Am. Oil. Chem. Soc., 1992, 75, 795. 19 Synergy Chemicals, http://www.senergychem.com. 20 H. Kimura, Appl. Catal. A: Gen., 1993, 105, 147158; H. Kimura, K. Tsuto, T. Wakisaka, Y. Kazumi and Y. Inaya, Appl. Catal. A: Gen., 1993, 96, 217228. 21 K. Nabe, N. Izuo, S. Yamada and I. Chibata, Appl. Environ. Microbiol., 1979, 38, 1056. 22 R. Ciriminna, G. Palmisano, C. Della Pina, M. Rossi and M. Pagliaro, Tetrahedron Lett., 2006, 47, 6993. 23 G. W. Keulks, L. D. Krenzke and T. M. Notermann, Adv. Catal., 1978, 27, 183. 24 M. M. Lin, Appl. Catal. A: Gen., 2001, 207, 1. 25 A. Corma, G. W. Huber, L. Sauvanaud and P. OConnor, J. Catal., 2008, 257, 163171. 26 ICIS Market-reporter, 6 January 2010, http://www.icis.com. 27 Global Industry Analysts, http://www.strategyr.com/Super_ Absorbent_Polymers_Market_Report.asp. 28 A. Yamamoto, Encyclopaedia of Chemical Technology, 3rd edn, 1978, vol. 2, p. 403. 29 M. P. Malveda, H. Janshekar, K. Yokose, CEH Marketing Research Report SRI Consulting, Major Amino Acids, June 2006. 30 M. P. Malveda, H. Janshekar, K. Yokose, Chemical Economics Handbook-SRI Consulting, June 2006, 502.5000 B. 31 B. Katryniok, S. Paul, M. Capron and F. Dumeignil, ChemSusChem., 2009, 2, 719730. 32 Schering-Kahlbaum FR695931, 1930. 33 H. Groll and G. Hearne (Shell), US 2042224, 1936. 34 H. Hoyt and T. Manninen (US Ind. Chemicals. Inc.), US 2558520, 1951. 35 A. Neher, T. Haas, A. Dietrich, H. Klenk and W. Girke (Degussa), DE 4238493, 1994. 36 A. Neher, T. Haas, A. Dietrich, H. Klenk and W. Girke (Degussa), US 5387720, 1995. 37 A. Neher and T. Haas (Degussa), US 5426249, 1995. 38 J.-L. Dubois, C. Duquenne, W. Hoelderich and J. Kervennal (Arkema), WO 2006087084, 2006. 39 J.-L. Dubois, C. Duquenne and W. Hoelderich (Arkema),WO 2006087083, 2006. 40 S.-H. Chai, H.-P. Wang, Y. Liang and B.-Q. Xu, J. Catal., 2007, 250, 342349. 41 S.-H. Chai, H.-P. Wang, Y. Liang and B.-Q. Xu, Green Chem., 2007, 9, 11301136. 42 Z. H. Chen, T. Lizuka and K. Tanabe, Chem. Lett., 1984, 13, 1085. 43 T. Lizuka, K. Ogasawara and K. Tanabe, Bull. Chem. Soc. Jpn., 1983, 56, 2927. 44 A. Alhanash, E. F. Kozhevnikova and I. V. Kozhevnikov, Appl. Catal. A: Gen., 2010, 378, 1118. 45 Y. T. Kim, K.-D. Jung and E. D. Park, Microporous Mesoporus Mater., 2010, 131, 2836. 46 A. K. Kinage, P. P. Upare, P. Kasinathan, Y. K. Hwang and J.-S. Chang, , Catal. Commun., 2010, 11, 620623. 47 S.-H. Chai, H.-P. Wang, Y. Liang and B.-Q. Xu, Green Chem., 2008, 10, 10871093. 48 S.-H. Chai, H.-P. Wang, Y. Liang and B.-Q. Xu, Appl. Catal. A: Gen., 2009, 353, 213222. and Q. Xin, 49 L. Ning, Y. Ding, W. Chen, L. Gong, R. Lin, Y. Lu Chin. J. Catal., 2008, 29(3), 212214. 50 B. Katryniok, S. Paul, M. Capron, C. Lancelot, V. Belli` ere-Baca, P. Rey and F. Dumeignil, Green Chem., 2010, 12, 19221925. 51 E. Tsukuda, S. Sato, R. Takahashi and T. Sodesawa, Catal. Commun., 2007, 8, 13491353. 52 H. Atia, U. Armbruster and A. Martin, J. Catal., 2008, 258, 71 82. 53 B.-Q. Xu, S.-H. Chai, T. Takahashi, M. Shima, S. Sato and R. Takahashi (Nippon Catalytic Chem. Ind.), WO 2007058221, 2007. 54 T. Sato and R. Takahashi (Nippon Catalytic Chem. Ind.), JP 2008088149, 2008. 55 H. Jo, S.-H. Chai, T. Takahashi and M. Shima (Nippon Catalytic Chem. Ind.), JP 2007137785, 2007. 56 S. Ere, U. Armbruster, U. Bentrup, A. Martin and A. Bruckner, Appl. Catal. A, 2010, DOI: 10.1016/j.apcata.2010.04.042, in press. 57 J.-L. Dubois, Y. Magatani and K. Okumura (Arkema), WO 2009127889 and WO 2009128555, 2009.

This journal is The Royal Society of Chemistry 2010

Green Chem., 2010, 12, 20792098 | 2097

View Article Online

Published on 12 November 2010 on http://pubs.rsc.org | doi:10.1039/C0GC00307G

58 X.-Z. Li (Shanghai Huayi Acrylic Acid Co), CN 101070276, 2007. 59 A. Zhuang, C. Zhang , S. Wen, X. Zhao and T. Wu (Shanghai Huayi Acrylic Acid Co), CN 101225039, 2008. 60 M. Okuno, E. Matsunami, T. Takahashi, H. Kasuga, M. Okada and M. Kirishik (Nippon Catalytic Chem. Ind.), WO 2007132926, 2007. 61 M. Okuno, E. Matsunami, T. Takahashi and H. Kasuga (Nippon Catalytic Chem. Ind.), JP 2007301505, 2007. 62 M. Okuno, E. Matsunami, T. Takahashi and H. Kasuga (Nippon Catalytic Chem. Ind.), JP 2007301506, 2007. 63 C.-J. Jia, Y. Liu, W. Schmidt, A.-H. Lu and F. Schuth, J. Catal., 2010, 269, 7179. 64 C.-J. Zhou, C.-J. Huang, W.-G. Zhang, H.-S. Zhai, H.-L. Wu and Z. S. Chao, Stud. Surf. Sci. Catal., 2007, 165, 527530. 65 K. Pathak, K. M. Reddy, N. N. Bakhshi and A. K. Dalai, Appl. Catal. A, 2010, 372, 224238. 66 L.-Z. Tao, S.-H. Chai, Y. Zuo, W.-T. Zheng, Y. Liang and B.-Q. Xu, Catal. Today, 2010, DOI: 10.1016/j.cattod.2010.03.073, in press. 67 W. Suprun, M. Lutecki, T. Haber and H. Papp, J. Mol. Catal. A: Chem., 2009, 309, 7178. 68 F. Wang, J.-L. Dubois and W. Ueda, J. Catal., 2009, 268, 260267. 69 F. Wang, J.-L Dubois and W. Ueda, Appl. Catal. A, 2010, 276, 2532. 70 J.-L. Dubois (Arkema), WO2010046227, 2010. 71 Q. Liu, Z. Zhang, Y. Du, J. Li and X. Yang, Catal. Lett., 2008, 127(3-4), 419428. 72 J. Deleplanque, J.-L. Dubois, J.-F Devaux and W. Ueda, Catal. Today, 2010, DOI: 10.1016/jcattod.2010.04.012, in press. 73 J.-L. Dubois (Arkema), WO2009044081, 2009. 74 E. Matsunami, T. Takahashi and H. Kasuga (Nippon Catalytic Chem. Ind.), JP 2007268363, 2007. 75 E. Matsunami, T. Takahashi and H. Kasuga (Nippon Catalytic Chem. Ind.), JP 2007268364, 2007. 76 H. Redlingshoefer, C. Weckbecker, K. Huthmacher and A. Doerein (Evonik Degussa), WO 2008092533, 2008. 77 H. Redlingshoefer, C. Weckbecker, K. Huthmacher and A. Doerein (Evonik Degussa), WO 2008092533, 2008. 78 H. Redlingshoefer, C. Weckbecker, K. Huthmacher and A. Doerein (Evonik Degussa), WO 2008092534, 2008. 79 A. Ulgen and W. Hoelderich, Catal. Lett., 2009, 131, 122128.

80 H. Kasuga and M. Okada (Nippon Catalytic Chem. Ind.), JP 2008137950, 2008. 81 J.-L. Dubois (Arkema), WO 2009156664, 2009. 82 Y. Arita, H. Kasuga and M. Kirishiki (Nippon Catalytic Chem. Ind.), JP 2008110298, 2008. 83 P. OConnor, C. Corma, G. Huber and L. Savanaud (Bioecon. Internat. Holding), WO 2008052993, 2008. 84 J.-L. Dubois (Arkema), FR 2897058, 2007. 85 H. Kasuga (Nippon Catalytic Chem. Ind.), JP 2008137952, 2008. 86 S. Ramayya, A. Brittain, C. De Almeida, W. Mok and M. J. Antal, Fuel, 1987, 66, 1364. 87 W. Buhler, E. Dinjus, H. J. Ederer, A. Kruse and C. Mas, J. Supercritical Fluids, 2001, 22, 3753. 88 L. Ott, M. Bicker and H. Vogel, Green Chem., 2006, 8(2), 214220. 89 V. Lehr, M. Sarlea, L. Ott and H. Vogel, Catal. Today, 2007, 121(12), 121129. 90 M. Watanabe, T. Iida, Y. Aizawa, T. M. Aida and H. Inomata, Biores. Technol., 2007, 98, 1285. 91 N. Suzuki and M. Takahashi (KAO Corp.), JP 2006290815, 2006. 92 Y. Yoshimi, Y. Masayuki, A. Torakichi and A. Takanori (Showa Denko), JP2009179569, 2009. 93 A. Takanori and Y. Masayuki (Showa Denko), JP2009292773, 2009. 94 A. Takanori and Y. Masayuki (Showa Denko), JP2009292774, 2009. 95 E. Yoda and Y. Ootawa, Appl. Catal. A, 2009, 360, 6670. 96 M. R. Nimlos, S. J. Blanksby, X. Qian, M. E. Himmel and D. K. Johnson, J. Phys. Chem. A., 2006, 110, 61456156. 97 K. Kongpatpanich, T. Nanok, B. Boekfa and J. Limtrekul, Prepr. Pap. Am. Chem. Soc. Div. Pet. Chem., 2010, 55(1), 115118. 98 J. Kijenski, A. Migdal, O. Osawaru and E. Smigiera (Inst. Chemii Przemyslowe), EP 1860090, 2007. 99 J.-L. Dubois (Arkema), WO 2008129208, 2008. 100 B. R. Sereshiki, S.-J. Balan, G. S. Patience and J.-L. Dubois, Ind. Eng. Chem. Res., 2010, 49, 10501056. 101 J.-L. Dubois (Arkema), WO2009081021, 2009. 102 C. Perego and D. Bianchi, Chem. Eng. J., 2010, DOI: 10.1016/j.cej.2010.01.036, in press. 103 R. Stern, G. Hillion, J-J. Rouxel and L. Serge (Institut Franc ais du Petrol), US 5908946, 1999. 104 L. Bournay, D. Casanave, B. Delfort, G. Hillion and J. A. Chodorge, Catal. Today, 2005, 106, 190192.

2098 | Green Chem., 2010, 12, 20792098

This journal is The Royal Society of Chemistry 2010

Você também pode gostar