Catal Lett (2009) 131:122128 DOI 10.

1007/s10562-009-9923-0

Conversion of Glycerol to Acrolein in the Presence of WO3/ZrO2 Catalysts

Arda Ulgen Wolfgang Hoelderich

Received: 30 January 2009 / Accepted: 26 February 2009 / Published online: 26 March 2009 Springer Science+Business Media, LLC 2009

Abstract Dehydration of glycerol to acrolein was performed on WO3/ZrO2 solid heterogeneous catalysts in a continuous ow xed bed reactor. A maximum of 75% acrolein selectivity was achieved at 100% conversion of glycerol. The acrolein selectivity correlates with the concentration of weak acidic sites of the WO3/ZrO2 catalysts that were free of basic sites. Keywords Glycerol (glycerin) Acrolein Alcohol dehydration Solid acid WO3/ZrO2 Sustainable technology

1 Introduction Worldwide glycerol production is increasing due to the growing biodiesel production based on triglycerides. On the other hand the current demand of glycerol, which is formed as an inevitable by-product thereof, is not increasing at the same rate. Consequently, the use of glycerol as a starting material becomes economically and environmentally interesting [1, 2]. The motivation of the present investigations is the valorisation of glycerol by its dehydration to acrolein. Acrolein nds direct application in medicine, water treatment, and petroleum industry as biocide. Rened acrolein is used for the synthesis of methionine, fragrances as well as dyes. The oxidation of crude acrolein to acrylic acid, however, is its major eld of use. The polymer of acrylic acid has super absorbent properties and is the main
A. Ulgen W. Hoelderich (&) RWTH Aachen University, Worringerweg 1, 52074 Aachen, Germany e-mail: hoelderich@rwth-aachen.de

component of hygienic pads and diapers. In addition polyacrylates are applied for wall paper paintings. Worldwide 85% of the acrylic acid is produced by captive oxidation of acrolein [3, 4]. Acrolein is currently produced by the oxidation of propylene using multi component mixed oxide catalysts [35]. The source of propylene is crude oil. Shifting the source of carbon for the production of acrolein from propylene to glycerol would not only encourage biodiesel producers by enabling commercial pathways for their by-product, but also it would avoid using crude oil in the production of acrolein. By that, acrolein would be a biomass based intermediate and a CO2-neutral production would be provided. Already in 1918 Sabatier et al. [6]reported the decomposition of glycerol to different products, including acrolein in the presence of alumina catalysts. The company Schering-Kahlbaum [7] patented a process in 1930, in which 80% acrolein yield was claimed by the dehydration of glycerol using metal phosphates impregnated on pumice stone. 18 years later Hoyt et al. [8] patented a heterogeneously catalyzed continuous ow xed bed process for the production of acrolein from glycerol. In that patent the consistence of the catalyst material was reported as ortho phosphoric acid supported on diatomaceous earth, which was mixed with a petroleum oil fraction having a boiling point between 300 and 400 C. An acrolein yield of 72.8% was reported. In 1993 Neher et al. [9] claimed the production of 1,2and 1,3-propandiol from glycerol. The rst step described in this patent is the dehydration of glycerol using a-alumina supported phosphoric acid for the production of acrolein. The acrolein yield was reported as 70.5%. In the current decade this reaction has gained immense academic and industrial interest. In collaboration with

123

Conversion of Glycerol to Acrolein

123

Arkema, we found the high activity of WO3/ZrO2 systems for the dehydration of glycerol in the gas phase for the rst time in 2005. The experiments with outstanding 7380% acrolein yields at total conversion of glycerol are described extensively in our corresponding patents [1013]. In 2007 Matsunami et al. [14] patented a process in which glycerol was dehydrated to acrolein using different phosphates. The acrolein yield was reported as 61.2% using a Si, P, Cs composite catalyst. In the same year Li [15] patented the same reaction using acidic zeolites. The highest acrolein yield was reported as 82.1% using ZSM11 as catalyst at 320 C. Also in 2007 S. Sato reported the high activity of silicotungstic acid supported on mesoporous silica resulting in 86.2% selectivity of acrolein and a glycerol conversion of 98.3% at 275 C [16]. The screening of several heterogeneous catalysts for the dehydration of glycerol was comprehensively reported by Xu et al. [17, 18] in 2007, too. 65% yield on acrolein was achieved by using 15 wt% WO3/ZrO2 catalysts. Recently, Ding et al. [19] published that silicotungstic acids supported on activated carbon leads to 75% acrolein selectivity at 93% glycerol conversion under He atmosphere at 330 C. As mentioned before, we reported in our patents that WO3/ZrO2 systems give suitable catalysts for the dehydration of glycerol to acrolein [1013]. In the present case study, we investigate such WO3/ZrO2 systems in more detail. Thereby, we study the inuence of tungsten content, the acidity/basicity of the zirconia carrier, the modication by calcination temperature and the physical properties on the catalytic performance.

water, the remaining slurry was placed in a ceramic bowl and dried 6 h at 110 C followed by calcination at 600 C for 6 h. Both drying and calcination steps were conducted in a box shaped programmable oven (Nabertherm N 7 with a Logotherm S 19 Program Controller, Tmax = 1,000 C) under ambient pressure and atmosphere without addition of any gases. The obtained powder was formed to pellets under 10 tons for 20 min, which was then crushed and sieved. A fraction of 0.51.0 mm particle size was used for the characterisation and screening experiments. 2.2 Catalyst Characterization BET surface areas, pore volumes and average pore diameters were obtained at 77 K on a Micrometrics ASAP 2010 Gas Sorption and Porosimetry System. After activation of the samples at 300 C for 3 h under vacuum, the adsorptiondesorption was conducted by passing nitrogen over the samples. BET surface areas were determined over a relative pressure range from 0.05 to 0.20 bars. Measurement of catalytic acidity and basicity were conducted with a temperature programmed desorption (TPD) unit; TPD-Pro 1100 from CE instruments equipped with a TCD detector. After heating the sample at 500 C for drying purposes, NH3 or CO2 was physisorbed at room temperature. Then the sample was heated up to 600 C at 1/min and the amount of desorbed gas was recorded by the TCD. Bulk elemental analyses were carried out with inductive couple plasma atomic emission spectroscopy on a Spectroame D (Spectro Analytic Insturment). Powder X-Ray diffraction (XRD) data were recorded on a Siemens Diffractometer (D 5000) operated at 45 kV and 40 mA, using nickel ltrated Cu Ka radiation with 1.5406 A between 1.5 and 40 (2h), at a scanning speed of 0.02/ min. The measured patterns were in good agreement with the literature, showing that WO3 was present on the ZrO2 carrier [20]. 2.3 Catalytic Reaction The gas-phase dehydration of glycerol under atmospheric pressure was conducted in a 100 cm continuous plug ow xed bed reactor in form of a coil, made of stainless steel tube with a 0.6 cm internal diameter. The solid catalyst was lled into the reactor which was placed afterwards in a temperature controlled, box shaped, GC-like, oven. The temperature of the oven was controlled with a thermocouple, connected to a Eurotherm controller. The temperature homogeneity within the oven was fullled with an installed radial fan. The observed temperature uctuation within the oven and over the time was not over 0.5 C.

2 Experimental 2.1 Catalysts ZrO2 powder and WO3/ZrO2 pellets with 19 wt% WO3 were kindly provided by St. Gobain Norpro, Ohio, USA. WO3/ZrO2 powders with ve different WO3 contents (between 2.11 and 15.43 wt% WO3) were obtained from Daiichi KKK, Japan, via Arkema CRRA, France. In these cases there is nothing disclosed about the method of preparation. These Daichii KKK catalysts were used without prior modication. Several homemade catalysts have been prepared according to the following impregnation method recipe, which was conducted under ambient atmosphere. A certain amount of ammonium (para) tungstate (Sigma Aldrich, Germany) and 200 mL of water were heated to 80 C and stirred 2 h, yielding a clear solution. Into this solution, ZrO2 (Provided by St. Gobain/Norpro, Ohio, USA) was added and stirred for another 4 h. After evaporation of the

123

124

A. Ulgen, W. Hoelderich

At the adjusted reaction temperature, a 20 wt% aqueous glycerol solution was pumped into the reactor at a rate of 1030 g/h. If required, a regulated ow of oxygen was introduced to the reactor as well. The reaction products were collected in a double jacket ask. The temperature of the collector ask was kept at -2 C using a cryostat all the time. Figure 1 illustrates the set-up used. Two balances were used for instant observation of the amount of the glycerol solution as well as the collected amount of the product mixture. The reactions were usually conducted for 8 h. A sample for analysis was taken each hour from the collected product mixture and analysed on a Fisons GC 8000 gas chromatograph equipped with a ame ionization detector and a Chromosorb W-AW column (2 mm i.d., 2.5 m long), set to on-column injection. Conversion of glycerol and product selectivity were calculated according to the following formulas: Glycerol conversion (% moles of reacted glycerol moles of glycerol fed in

gures reached steady values. The given conversion and selectivity gures are calculated by the values of at least three measurements.

3 Results and Discussion 3.1 Effect of Temperature An initial series of experiments were conducted with unmodied 19 wt% WO3/ZrO2 (St. Gobain/Norpro, Ohio, USA) for the determination of optimal reaction temperature. For each experiment 9.0 mL of the catalyst, corresponding to 10.0 g, were placed into the reactor. A 20 wt% aqueous glycerol solution was used as reactant, which was fed into the reactor at a rate of 30 g/h together with 2.5 mln/min oxygen. The results are presented in Fig. 2. At temperatures higher than 240 C, glycerol is extensively converted. The acrolein selectivity, however, shows a maximum at 280 C. At lower temperatures the intermolecular dehydration, yielding oligomers of glycerol, is thermodynamically favoured over the desired intramolecular dehydration forming acrolein. At temperatures higher than 280 C, the formation of CO and CO2 is possible. These two reasons are responsible for the selectivity decrease of acrolein. That is also supported by the fact that the untrapped product mass increased with rising temperatures, from 0.1 wt% at 240 C to 2.0 wt% at 320 C.

Product selectivity (mol% moles of carbon in a defined product moles of carbon in reacted glycerol The mass balance was usually above 98% and the calculated carbon balance of the detected compounds was usually above 90% except for the initial phase of one hour. After the incubation phase, the selectivity and conversion
Fig. 1 Drawing of the used set-up for the continuous dehydration reaction of glycerol

Oxygen flow regulator

Temperature controller Pump

Bubbler

Condenser By-pass Mass flow controller

D

Glycerol solution Reactant scale Oxygen bottle

Reactor

Cryostat Product scale Product container

Oven Sample for GC

123

Conversion of Glycerol to Acrolein

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 220

125

nature of oxygen is generating the amelioration of the results and not solely the existence of an inert gas. 3.2 Effect of WO3 Content
X Glycerol S Acrolein

240

260

280

300

320

340

Reaction temperature [ C]

Fig. 2 Effect of reaction temperature on glycerol conversion and acrolein selectivity using 19 wt% WO3/ZrO2 catalyst

Due to these results, 280 C was selected as standard reaction temperature for further investigations. Additionally, the catalyst bed was reduced to 4.5 mL so that by intention only an incomplete conversion is achieved. This fact enables a better comparison of the catalytic performance in terms of conversion, which is not possible in case of complete conversion. The results of two experiments (with and without oxygen ow) in which partial conversion of glycerol was provoked by using 4.5 mL of catalyst at a reactant feed rate of 23 g/h, are given in the Table 1. The positive effect of oxygen gas has been reported previously [10]. With the addition of oxygen to the reactant ow, the conversion dropped slightly. Another possible reason for this effect is the blockage of the active sites by molecular oxygen in form of dative bonds. The important effect, however, is the increase in acrolein selectivity and decrease in the main side product selectivities. Similar reasons mentioned above for the conversion change could be responsible for this desirable effect. Especially the blockage of neighbouring active sites with molecular oxygen could suppress the stability of such transition states, which would else lead to undesired side products. It is possible that the positive effect caused by oxygen co-ow could be due to a slight decrease in residence time. This possibility was assessed by an experiment in which nitrogen gas was used instead of oxygen. The results were very similar to the ones obtained from the experiment where no gas was added. This shows that the chemical

ZrO2 can have acidic and basic sites, i.e., is an amphoteric material [21]. We studied the modication of acidic/basic sites of the ZrO2 support by the addition of various amounts of WO3, and the corresponding effects on the catalytic performance for the dehydration of glycerol. A series of WO3/ZrO2 catalysts were used to study the effect of WO3 content (Daiichi KKK, Japan). The acidity and basicity of the catalysts were measured with NH3 and CO2 TPD. The increasing temperature axis implies increasing strength of the acidic or basic sites. On the other hand the increasing signal correlates to the amount of corresponding sites. For both acidic and basic sites, two different site strengths, weak, and strong, are observed. While the amount of weak acidic sites increases with increasing WO3 amount, an opposite behaviour was observed for basic sites, as shown in Figs. 3 and 4. NH3 measurement of the sample with 9.17 wt% loading shows a sharp strong acidity peak, which is

Fig. 3 NH3 TPD of WO3/ZrO2 catalysts with different WO3 amounts

Table 1 Glycerol conversion and product selectivity observed for the dehydration of glycerol at 280 C in the presence WO3/ZrO2 catalysts with 19 wt% WO3 Co-owed gas Co-ow (mln/min) Conversion (%) Selectivity (%) Acrolein O2 N2
a

Propion aldehyde 0.8 0.3 0.7

Hydroxy propanone 10.2 8.0 10.8

Othersa 20.4 17.4 21.3

11.33 11.33

82.9 72.7 80.2

68.6 74.3 67.2

Others are CO, CO2, acetaldehyde, acetone, acetic acid, propionic acid, acrylic acid, and unidentied products

123

126

A. Ulgen, W. Hoelderich

Fig. 4 CO2 TPD of WO3/ZrO2 catalysts with different WO3 amounts

believed to be a measurement artefact. Even though the amount of basic sites decreases with increasing loading, a strict monotonous trend cannot be observed. The reason for this could be the differences within the available surface areas of the measured catalysts, since the physical occurrences during a TPD measurement take place on the surface. While ZrO2 was cited as a quite neutral material, it was found that basic sites can exist on its surface, depending on the preparation method. Especially if it is obtained from Zr(OH)2, it is well possible that basic sites are available [22]. Intermittent strong basic sites on the mentioned catalysts should be due to deviations in their preparation method, particularly the calcination parameters. The impact of calcination is discussed briey in the next chapter (Figs. 3, 4).

The effects of acidity/basicity of these WO3/ZrO2 systems on the catalytic performance for the dehydration of glycerol were investigated in corresponding screening experiments. The experiments were conducted at 280 C, using 4.5 mL of catalyst (5.0 g) in the absence of oxygen ow for studying solely the effect of catalyst composition. The 20 wt% aqueous glycerol solution was fed into the reactor at a rate of 23 g/h. Table 2 summarizes the results of the experiments obtained using mentioned commercial and also homemade catalysts. The increasing WO3 amount led to an increasing acrolein selectivity starting from 11.3% over a 2.11 wt% WO3/ ZrO2 system to 55.3% with a 15.43 wt% WO3/ZrO2. Combined consideration of Table 1 with Figs. 3 and 4 shows that the existence of weak acidic sites as well as the absence of basic sites is preferred for high acrolein selectivity. On the other hand, abundance of weak basic sites enables higher selectivity of undesirable hydroxy propanone. At a possible commercialization of such an acrolein production by the dehydration of glycerol, a purication of acrolein by distillation has to be carried out. All the identied by-products except for propionaldehyde have signicantly different boiling points than acrolein, which enables their separation. Even though propionaldehyde, which is listed in Table 2, is produced in small amounts during the dehydration of glycerol under given reaction parameters, its existence is undesired for further purication processes. However, the tendency for the selectivity of propionaldehyde is signicantly coupled to acrolein selectivity. The selectivity of other identied by-products was not affected by the variation of WO3 amount in the catalytic system. The comparison of commercial and homemade catalysts reveals that the method of preparation has a big impact on

Table 2 Glycerol conversion and product selectivity observed for the dehydration of glycerol at 280 C in the presence of WO3/ZrO2 catalysts with different WO3 amounts Origin of catalyst WO3 content (wt%) Conversion (%) Selectivity (%) Acrolein Daiichi KKK Daiichi KKK Daiichi KKK Daiichi KKK Daiichi KKK Norpro Homemade Homemade Homemade
a

Propion aldehyde 0.9 1.0 1.0 1.1 1.5 0.2 0.8 2.2 1.3

Hydroxy propanone 40.2 41.5 29.1 29.7 25.1 2.1 29.7 14.8 9.0

Othersa 47.6 29.7 44.0 34.0 18.1 25.1 48.9 30.8 33.8

2.11 5.31 7.27 9.17 15.43 19 0.94 4.28 8.21

51.1 37.7 45.4 38.2 57.7 60.5 37.2 62.0 83.0

11.3 27.8 25.9 35.2 55.3 72.6 20.6 52.2 55.9

Others are CO, CO2, acetaldehyde, acetone, acetic acid, propionic acid, acrylic acid, and unidentied products

123

Conversion of Glycerol to Acrolein

127

catalytic performance and that there is still potential for improvement. 3.3 Effect of Calcination Three catalysts obtained from Daiichi KKK were calcined at 600 C for 6 h under atmospheric conditions with the already mentioned calcination method. The BET and TPD measurements show the effects of the calcination on the catalytic properties (Table 3). For all three catalysts, the BET surface areas decreased and the pore size increased as expected. Apparently, during the calcination process the heat induced the combination of small pores to form bigger pores and this occurrence led to smaller BET areas. The calcination process had also immense effects on the acidic/basic properties of the catalytic system (Table 4). The comparison of Figs. 5 and 6 with Figs. 3 and 4 reveal the changes of the acidic/basic properties of the catalysts after the calcination process. The amount of weak and strong acidic sites decreased considerably. More signicant is the extensive disappearance of both weak and strong basic sites. Probably, the calcination process has ameliorated the dispersion of WO3 on the ZrO2, so that the basic sites of ZrO2 are extensively covered. Even more important is the elimination of remaining Zr(OH)2 units by their dehydration to ZrO2. Additionally, the drop of the
Table 3 Effect of calcination on the pore system of WO3/ZrO2 catalysts WO3 content (wt%) 5.31 5.31 9.17 9.17 15.43 15.43 Calcination Before After Before After Before After BET surface (m2/g) 106 48 144 57 140 59 Mean pore diameter (A) 63 107 56 87 62 77

Fig. 5 NH3 TPD of WO3/ZrO2 catalysts with different WO3 amounts after calcination

Fig. 6 CO2 TPD of WO3/ZrO2 catalysts with different WO3 amounts after calcination

BET area led to an overall decrease on all sites per gram of catalyst. These catalysts were screened under the same conditions and with the same parameters as in the previous chapter. The glycerol conversion and product selectivity obtained with these catalysts differ radically from the ones obtained using their uncalcined predecessors. The rst observation is the considerable increase of glycerol conversion. Apparently,

Table 4 Dependence of glycerol conversion and product selectivity observed for the dehydration of glycerol at 280 C on the calcination of WO3/ZrO2 catalysts with different WO3 amounts WO3 amount (wt%) Calcination procedure Conversion (%) Selectivity (%) Acrolein 5.31 5.31 9.17 9.17 15.43 15.43
a

Propion aldehyde 1.0 1.2 1.1 1.4 1.5 0.8

Hydroxy propanone 41.5 15.9 29.7 9.5 25.1 10.3

Othersa 29.7 18.1 34.0 26.8 18.1 16.8

Before After Before After Before After

37.7 47.5 38.2 95.6 57.7 88.7

27.8 64.8 35.2 62.3 55.3 72.1

Others are CO, CO2, acetaldehyde, acetone, acetic acid, propionic acid, acrylic acid, and unidentied products

123

128

A. Ulgen, W. Hoelderich

increased pore size of the catalyst had a positive effect on transport phenomena. The acrolein selectivity reached to values between 64.8 and 72.1. The absence of basic sites and strong acidic sites on the catalyst accounts for the increase of acrolein selectivity. Additionally, the absence of basic sites decreased the formation of hydroxy propanone, too. No signicant changes were recorded for the formation of propionaldehyde in relation to the calcination process.

References
1. Gunstone FD, Heming MPD (2004) Lipid Technol 16:177 2. Paglario M, Ciriminna R, Kimura H, Rossi M, Pina CD (2007) Angew Chem 229:4516 3. Etzkorn WG, Pedersen SE, Snead TE (2002) Kirk-Othmer Encycl Chem Technol 1:265 4. Mourey S (1999) Info Chim Mag 412:90 5. Weissermel K, Arpe H-J (1994) Industrielle organische chemie, vol 310. Wiley-VCH, Weinheim 6. Sabatier P, Gaudion G (1918) Comptab Rend 166:1033 7. French Patent 695,931 (1930) to Schering-Kahlbaum AG 8. Hoyt HE, Manninen TH (1948) US Patent 2,558,520 to US Industrial Chemicals Inc 9. Neher A, Haas T, Arntz D, Klenk H, Girke W (1995) US Patent 5,387,720 to Degussa AG 10. Dubois JL, Duquenne C, Hoelderich WH (2005) F Patent 2,882,052 to Arkema SA 11. Dubois JL, Duquenne C, Hoelderich WH, Kervennal J (2005) F Patent 2,882,053 to Arkema SA 12. Dubois JL, Duquenne C, Hoelderich WH (2005) F Patent 2,884,817 to Arkema SA 13. Dubois JL, Duquenne C, Hoelderich WH (2006) F Patent 2,884,818 to Arkema SA 14. Matsunami E, Takahashi T, Kasuga H, Arita Y (2007) WO Patent 119,528 to Nippon Shokubai Co. Ltd 15. Li XZ (2007) CN Patent 101,070,276 A to Shanghai Huayi Acrylic Acid Co 16. Tsukuda E, Sato S, Tajahashi R, Sodesewa T (2007) Cat Comm 8:1349 17. Chai S-H, Wang H-P, Liang Y, Xu B-Q (2007) Green Chem 9:1130 18. Chai S-H, Wang H-P, Liang Y, Xu B-Q (2007) J Catal 250:342 19. Ning L, Ding Y, Chen W, Gong L, Lin R, Lu Y, Xin Q (2008) Chin J Catal 29:212 20. Martinez A, Prieto G, Arribas MA, Concepcion P, Sanchez-Royo JF (2007) J Catal 248:288 21. Tanabe K, Yamaguchi T (1994) Cat Today 20:185 22. Tanabe K (1989) New solid acids and bases: their catalytic properties, vol 55. Elsevier

4 Conclusion The production of acrolein from glycerol by its dehydration in a continuous manner is shown to be possible and tungstated zirconia systems are very suitable catalysts for that. The optimal reaction temperature is 280 C. Below and above this temperature unwanted side reactions occur and thus decrease the acrolein selectivity. The addition of oxygen to the reaction system reduces the formation of side products; in particular the very undesirable hydroxy propanone formation is inhibited. The acrolein selectivity correlates on one hand positively with the existence of the weak acidic sites on the catalyst and on the other hand negatively with the basic sites. Thus, the catalytic properties can be improved by increasing amount of tungsten oxide in the catalyst resulting in populating acidic sites. Another important conclusion is the absence of any relationship between the strong acidic sites and acrolein selectivity. Calcination is apparently a very suitable tool for the elimination of basic sites and for the enhancement of weak acidic sites, thus resulting in dramatically better catalytic performance.
Acknowledgments We would like to thank Arkema S. A. for their nancial support and St. Gobain/Norpro for providing the ZrO2 and WO3/ZrO2 materials.

123