Kinetics of the Oxidative Dehydrogenation of Ethanol to Acetaldehyde

Mar 4, 2004 - ... Università di Napoli “Federico II”, Complesso di Monte Sant'Angelo, ... expression is similar to a Mars and van Krevelen kineti...
1 downloads 0 Views 154KB Size
Download PDF
Ind. Eng. Chem. Res. 2004, 43, 1623-1633

1623

Kinetics of the Oxidative Dehydrogenation of Ethanol to Acetaldehyde on V2O5/TiO2-SiO2 Catalysts Prepared by Grafting R. Tesser, V. Maradei, M. Di Serio, and E. Santacesaria* Dipartimento di Chimica, Universita` di Napoli “Federico II”, Complesso di Monte Sant’Angelo, Via Cinthia, 80126 Napoli, Italy

Catalysts that are very active and selective in the oxidative dehydrogenation of ethanol to acetaldehyde have been prepared by grafting vanadyl triisopropoxide onto a silica support whose surface was coated with TiO2. Together with acetaldehyde, small amounts of byproducts were obtained, including acetic acid, acetals, and CO2. The kinetic behaviors of the catalysts were studied in the temperature range 100-180 °C, by changing the ethanol residence time, the molar ratio between the reagents, and the vanadium load. A four-step mechanism was adopted, consisting of the following steps: (i) dissociative adsorption of ethanol onto vanadium, (ii) R-hydrogen withdrawal by the metal to form acetaldehyde and a hydride group, (iii) oxidation of the formed hydride, and (iv) dehydration of the vanadium site to restore the original active site. This mechanism can be simplified, and the resulting kinetic expression is similar to a Mars and van Krevelen kinetic expression. The reaction is well-described by this kinetic model. The activation energy (11 kcal/mol) obtained for the main reaction is in agreement with the values reported in the literature for the same type of catalyst. The reaction rates depend linearly on the vanadium load. Introduction Interest in ethanol production from renewable natural sources as a possible alternative energy vector is increasing strongly throughou the world. For example, ethanol can usefully be used as a feedstock for producing acetaldehyde, ethyl acetate, acetic acid, ETBE, etc. Acetaldehyde can be obtained from ethanol by dehydrogenation, oxidation, and oxidative dehydrogenation, and different industrial techniques have been developed for these processes in the past.1,2 Acetaldehyde production via the oxidative dehydrogenation (ODH) of ethanol could be a promising alternative to the Wacker process, occurring more simply in a single step and in tubular reactors, provided that high activities and selectivities can be achieved under mild conditions. The partial oxidation of ethanol over different catalytic systems has been studied by several authors.3-10 In particular, supported V2O5 has been found to be an active and selective catalyst in promoting the oxidative dehydrogenation of ethanol to acetaldehyde.8-11 The reaction occurs under very mild conditions of temperature (150-250 °C) and pressure (1 atm).9-11 Moreover, in contrast to other ODH reactions, such as the ODH of light hydrocarbons, that occur at higher temperature, the surface oxygen of the V2O5 lattice is not directly involved in the ODH process, because the oxygen exchange reaction is too slow at the low temperatures used. Therefore, classical Mars and van Krevelen redox mechanism12 is not operative in this reaction, and other redox mechanisms must be considered. Acetaldehyde, the main reaction product, is obtained in very high selectivities, even though this compound is more oxidable than ethanol. According to Oyama and Somorjai,8 the reaction performed on V2O5 catalysts supported on SiO2 is * To whom correspondence should be addressed. E-mail: [email protected].

structure-insensitive. The same conclusion was reached by Lakshmi et al.13 using V2O5 catalysts prepared by impregnation on mixed oxides. Inamaru et al.14 studied ethanol dehydrogenation on V2O5 catalysts prepared by both impregnation and by chemical vapor deposition. They found that, in the latter case, the catalysts were more dispersed and more selective in the reaction. The effect of V2O5 dispersion, therefore, seems to be somewhat controversial, although in a previous work,10 we showed that a high dispersion of V2O5 on a TiO2-SiO2 support is important for obtaining high selectivities, a result that is also in agreement with the findings of Quaranta et al.9 In fact, catalysts prepared by grafting on an opportune support are more selective than catalysts prepared by impregnation, probably as a consequence of greater dispersion.10 It seems opportune, therefore, to distinguish between the catalytic action of vanadium oxide crystallites and that of dispersed monomeric or polymeric VO4 species anchored on a surface. In the latter case, three different types of bonds must be distinguished:15 VdO terminal bonds, bridging V-O-V bonds, and V-O-support bonds. It has recently been shown15 that VdO bonds are not responsible for the ODH of methanol to formaldehyde. Bridging V-O-V bonds are also poorly involved,15 whereas a dramatic effect on activities and selectivities can be observed as a consequence of changing the support.15,16 It is universally recognized that the support plays a fundamental role in this and other ODH reactions by favoring or inhibiting the aggregation of vanadium oxide on the surface and by allowing the V-O-support bonds to be more or less involved in the reaction. TiO2, in particular, strongly interacts with vanadium oxide, thus favoring the molecular dispersion of the active phase.9,15,17 However, TiO2 alone is not convenient as a support, because of its low specific surface area and low resistance to sintering. To eliminate the drawbacks of this support, it is a common

10.1021/ie034182s CCC: $27.50 © 2004 American Chemical Society Published on Web 03/04/2004

1624 Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004

practice to use a silica support coated with TiO2.9,10,17,18 This TiO2-SiO2 support is thermostable and retains the high specific surface area characteristic of SiO2, but it has the chemical characteristics of TiO2 on the surface. Different methods have been adopted for preparing this support by using both impregnation and grafting techniques, and this support has been extensively studied and characterized.9,10,18 Consequently, TiO2-SiO2 is a very good and well-characterized support for dispersed V2O5 catalysts. For these reasons, we used it in the present work. Despite the significant attention paid by various authors to the ODH of ethanol and to the action of vanadium-based catalysts, only one kinetic study has been published on this reaction, namely, the work by Gomez et al. using a VMgO as catalyst.11 Other kinetic studies are available, but they are related to the use of other less active catalysts.4,5 In the present work, therefore, we have studied the kinetics of the ODH of ethanol on a V2O5/TiO2-SiO2 catalyst prepared by grafting vanadyl triisopropoxide onto a support of silica coated with TiO2 using a multistep grafting procedure. We tested the kinetic behavior of the catalyst by varying the reagent concentrations; the residence time; the temperature; the vanadium load; the acid and basic characteristics of the catalyst; and the presence in the feed of reaction products such as water or acetaldehyde. A reaction scheme consisting of five reactions was developed for considering the evolution with time of all the reaction products, namely, acetaldehyde, acetic acid, carbon dioxide, acetals, and diethyl ether. A kinetic law for interpreting both the main reaction from ethanol to acetaldehyde and all other oxidations occurring in the reaction scheme was derived by assuming a redox mechanism occurring in the following four steps: (i) dissociative adsorption of ethanol on vanadium giving place to an ethoxy group, (ii) R-hydrogen withdrawal by the metal to form acetaldehyde and a hydride group, (iii) oxidation of the formed hydride, and (iv) dehydration of the vanadium site to restore the original active site. The same mechanism has also been employed by other authors for the same reaction on different catalysts3 and for the similar reaction of the ODH of methanol to formaldehyde on vanadium-based catalyst.15 The kinetic law derived was found to be identical, in mathematical form, to the one that can be obtained from the classical Mars and van Krevelen mechanism.12 A discussion of the kinetic behavior of the reaction, the possible alternatives to the postulated reaction mechanism, and the values obtained for the kinetic parameters concludes the paper. Experimental Section Support and Catalyst Preparation. The support of silica coated with TiO2 was prepared in a three-step grafting procedure, using titanium isopropoxide dissolved in toluene and a commercial silica (Grace S432), calcined at 500 °C for 8 h. The quantity of titanium isopropoxide dissolved in toluene and used in each grafting step was about 50% more than the amount corresponding to a monolayer, assuming a conventional stoichiometry of one hydroxyl per alkoxide molecule. After calcination, the silica was put into contact with the titanium alkoxide solution by refluxing at the boiling point of the solvent for 6 h under constant stirring. The solid obtained was then filtered, washed with toluene,

dried at 100 °C, steamed at 190 °C to eliminate the residual alkoxide groups from the surface by hydrolysis, and finally calcined at 500° for 2 h. The procedure described above was then repeated two additional times to obtain the TiO2-SiO2 support, that is, a silica support coated with more than a monolayer of TiO2. More details about the preparation method and the properties of this support are reported elsewhere.10,18-20 The catalysts were prepared by putting the described support into contact with a solution of vanadyl triisopropoxide normally dissolved in n-hexane at room temperature for 24 h under a He atmosphere. Some catalyst samples were also been prepared by dissolving the vanadyl triisopropoxide in other solvents, such as dioxane and tetrahydrofurane. In the latter case, titanium and vanadyl alkoxides were mixed with each other in a molar ratio of 12:1 and then treated with an equimolecular amount of water containing traces of HCl to promote partial hydrolysis and alkoxide condensation. The obtained precursor was then anchored directly onto the silica surface in the usual way. This procedure, described in detail in a previous work,10 largely simplifies the preparation of well-dispersed vanadium catalyst. After the reaction, the samples were filtered, washed with the used solvent, dried at 105 °C, steamed at 190 °C for 2 h, and finally calcined at 500 °C for 2 h. Different catalyst samples were prepared with different vanadium loads, as reported in Table 1, where the used supports and prepared catalysts are listed together with some of their relevant properties and the operating conditions employed. All of the reagents used were furnished by Fluka and were of the highest level of purity available. Catalyst and Support Characterization Techniques. The titanium loads were determined using a colorimetric method,21 after dissolution of the titanium in a concentrated sulfuric acid solution, dilution, and then treatment with H2O2. The vanadium load in the prepared catalysts was determined by atomic adsorption spectroscopy, after dissolution in concentrated sulfuric acid. The support and prepared catalysts were subjected to XRD analysis on an X3000 Seifert diffractometer equipped with a lithium fluoride monochromator on the diffracted beam. The scans were collected within the range of 4-44° (2θ) using Mo KR radiation. Diffuse reflectance spectra were obtained on a Shimatzu AV2101 spectrophotometer. FTIR and DRIFT spectra were collected using a Nicolet AVATAR 360 instrument. Textural analyses were carried out on a Thermoquest Sorptomatic 1990 instrument (Fisons Instruments) by determining the nitrogen adsorption/desorption isotherms at 77 K. The samples were thermally pretreated under vacuum overnight to 473 K (heating rate ) 1 K/min). Specific surface areas and pore size distributions were determined using the BET22 and DollimoreHeal23 methods, respectively. The catalyst mainly used in the kinetic runs, previously treated with a stream of pyridine, was also subjected to temperature-programmed desorption in a He flow stream (20 cm3/min) using a TPDRO 1100 ThermoFinnigan instrument, to evaluate the number of acid sites on the catalytic surface. After treatment of the sample at 170°C for 1 h to eliminate the physically adsorbed pyridine, the temperature was gradually increased from 170 to 500 °C at a scanning rate of

Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004 1625 Table 1. Operating Conditions and Reagents Used in Catalyst Preparation and Some Related Catalyst Properties Silica Support silica type

thermal treatment temperature

specific surface area (m2/g)

porosity (cm3/g)

hydroxyl density (mmol/g)

GraceS432

500 °C

282

1.02

0.92

TiO2/ SiO2 Support, TiO2 Load after Each Grafting Step

sample

step

TiO2 anchored (mmol/g)

TiO2-SiO2 TiO2-SiO2 TiO2-SiO2

I II III

0.88 1.70 2.23

TiO2 (wt %)

specific surface area (m2/g)

porosity (cm3/g)

7.0 13.7 17.8

274

0.96

Catalyst Preparation and Properties

precursor/solvent/support

acronym

VO(i-Pr)3 used (mmol/g)

VO(Oi-Pr)3/n-hexane/TiO2-SiO2 VO(Oi-Pr)3/ n-hexane/TiO2-SiO2

V/Ti-Si(1) V/Ti-Si(4)

0.07 0.40

TiO2-SiO2 (g)

hexane (mL)

T (°C)

V2O5 content (wt %)

specific surface area (m2/g)

porosity (cm3/g)

6 6

75 75

25 25

0.65 3.56

243 258

0.30 0.28

Catalyst Preparation and Properties

precursor/solvent/support

acronym

VO(Oi-Pr)3 used (mmol/g)

[VO(Oi-Pr)3+ Ti(Oi-Pr)4]h/THF/SiO2 [VO(Oi-Pr)3+ Ti(Oi-Pr)4]h/dioxane/SiO2

(V-Ti)h/Si(T) (V-Ti)h/Si(D)

0.55 0.58

10 °C/min. The temperature was then kept constant at 500 °C for about 2 h. The same catalyst, after pretreatment at 400 °C for 1 h in air, was reduced with hydrogen (diluted to 6% in nitrogen), by heating the sample from 100 to 500 °C at a scanning rate of 5 °C/min in the same TPDRO 1100 ThermoFinnigan instrument. Methods, Techniques, and Operating Conditions Used in the Catalytic Runs. Kinetic runs were performed in a stainless steel tubular reactor with an internal diameter of 1 cm. The reactor was externally jacketed and kept isothermal with a fluidized bed of sand. Liquid ethanol was fed, by a syringe pump, into a vaporizer chamber kept at 170 °C and was then sent, after the addition of a stream of oxygen and helium, into a stainless steel coil kept at the same temperature as the reactor. The composition of the vapors at the outlet of the reactor was analyzed gas-chromatographically by withdrawing small samples through an on-line sampling valve kept at 150 °C. The gas chromatograph used was an HP 5890 instrument, with a Restek Rt-Q-Plot 30 m × 0.32 mm column. Helium was used as the carrier gas. The conditions used for the analyses were as follows: temperature held at 40 °C for 2 min, increased at a rate of 20 °C/min to 160 °C, and then maintained at this temperature for 20 min. A thermal conductivity detector (TCD) kept at 210 °C was used for detection. All TCD response factors were determined independently by analysis of different binary systems consisting of ethanol in a mixture with each individual reaction product. Samples of powdered catalyst, generally 0.3 g, were placed inside the reactor on a bed of glass wool. Two thermocouples located immediately upon and under the catalytic bed allowed the validity of the isothermal conditions to be controlled. Three different sets of kinetics runs were performed. Kinetic runs were mainly performed on a single catalyst containing 3.56 wt % of V2O5. A feed rate of inert diluent (helium) of 22 cm3/min was kept constant during all runs. The first set of runs was performed in the

Ti(Oi-Pr)4 used (mmol/g)

SiO2 (g)

solvent (ml)

T (°C)

V2O5 content (wt %)

TiO2 content (wt %)

6.91 6.74

6 6

100 100

25 25

2.23 1.91

7.17 6.75

temperature range 100-180 °C by changing the ethanol residence time from 9 to 70 gcat‚h/mol but keeping the molar ratio between ethanol and oxygen constant; the second set was carried out in the temperature range 140-160 °C by signifcantly varying both the ethanol residence time (from 10 to 200 gcat‚h/mol) and the molar ratio between ethanol and oxygen. The third set of runs is related to the effect of the vanadium load on the activity and was conducted on all available catalysts containing different amounts of of V2O5, as can be seen in Table 1. In these latter experiments, 0.5 g of catalyst was used in each run performed in a temperature range of 100-190 °C with a constant ethanol residence time of 26.6 gcat‚h/mol. Results are reported in terms of the ethanol conversion and product yields. The ethanol conversion is defined as

C)

(number of moles of ethanol reacted) (number of moles of ethanol fed)

while the yields of the ith product is defined as

Yi )

(number of moles of product i formed) (number of moles of ethanol fed)

Results and Discussion Catalyst and Support Characterization. In Table 1, the operating conditions used in the preparation of the supports and catalysts, together with some the properties of the resulting materials, are reported. In the same table is also reported the particular amount of TiO2 loaded on the SiO2 after each grafting step. As can be seen from these data, the TiO2 loading progressively increases with each grafting step, but to a lower and lower extent. Examination of the X-ray diffraction pattern of the support obtained after three grafting steps reveals very small crystallites of anatase and a predominant amorphous deposit of TiO2. XPS analyses

1626 Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004 Table 2. Operating Conditions and Experimental Results for Runs Performed with a Constant Ethanol/Oxygen Molar Ratio feed composition ethanol oxygen helium ethanol produst yielda (%) W/F catalyst (liquid) (gas) (gas) temperature conversiona (gcat/h‚molEtOH) (g) (mL/h) (mL/min) (mL/min) (°C) (%) acetaldehyde acetic acid acetals CO2 ethylic ether 17.7 26.6 64.8 9.7 17.7 26.6 64.8 9.7 17.7 26.6 64.8 9.7 17.7 26.6 64.8 9.7 26.6 64.8 9.7 26.6 64.8 a

0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33

1.10 0.73 0.30 2.00 1.10 0.73 0.30 2.00 1.10 0.73 0.30 2.00 1.10 0.73 0.30 2.00 0.73 0.30 2.00 0.73 0.30

7.70 5.13 2.10 13.96 7.70 5.13 2.10 13.96 7.70 5.13 2.10 13.96 7.70 5.13 2.10 13.96 5.13 2.10 13.96 5.13 2.10

22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22

100 100 100 120 120 120 120 140 140 140 140 160 160 160 160 170 170 170 180 180 180

3.24 3.67 10.19 3.62 6.54 7.01 13.91 9.74 11.35 23.23 28.54 23.67 30.96 37.16 58.79 31.12 54.25 68.88 48.28 76.51 86.43

3.07 3.39 9.29 3.31 6.14 6.55 12.95 9.03 10.28 21.22 25.92 20.16 26.78 31.79 48.46 28.13 45.82 57.93 40.51 63.15 69.21

0.02 0.09 0.07 0.15 0.04 0.09 0.13 0.10 -

0.11 0.18 0.69 0.03 0.22 0.25 0.57 0.39 0.69 1.31 1.56 1.65 2.03 2.67 5.59 1.09 4.90 6.77 3.97 7.78 11.08

0.88 0.90 1.12 2.46 0.38 1.41 0.84 1.71 2.48 2.83

0.06 0.10 0.21 0.29 0.17 0.20 0.39 0.32 0.37 0.72 1.05 0.89 1.18 1.43 2.26 1.48 2.03 1.56 1.95 2.99 3.30

Conversion and yields are averaged values.

Figure 1. FTIR spectra of silica at different titanium loadings in comparison with the spectrum of pure silica.

of similar samples of TiO2-SiO29,10 have shown that an atomically homogeneous dispersion is achieved for lower TiO2 coverages corresponding to less than a monolayer (from 0.6 to 1.0), after which small aggregates and crystallites can be formed. The original SiO2 support shows an intense FTIR absorption band at 3747 cm-1, corresponding to isolated silanol groups (Si-OH).24,25 These silanol groups are strongly reduced as a consequence of titanium isopropoxide grafting, as can be seen in Figure 1. This demonstrates the efficacy of the alkoxide grafting reaction in dispersing titanium on silica surface. Included in Table 1 is a list of the prepared catalysts and their main related properties. As can be seen, the specific surface area is negatively affected by the increase of the vanadium loading in the range considered. XRD analysis indicates the absence of crystalline V2O5. This observation was also confirmed by examining the DRIFT spectra, which showed a moderate increase of the signal corresponding to V-O-V bonds with increasing vanadium load, that is, the occurrence of

vanadium aggregation, probably giving rise to polymeric linear species anchored on TiO2 surface. From a TPD run on adsorbed pyridine, it was possible to evaluate an overall number of acid sites corresponding to 123.6 µmol/gcat. These acid sites are strong and relatively homogeneous, considering that pyridine desorption mainly occurs at about 500 °C. TPR with hydrogen gave a sharp peak starting at 325 °C and reaching a maximum at about 500 °C. The amount of hydrogen consumed corresponded to a reduction of V5+ to V3+. Catalytic Results. In Tables 2-4 the kinetic results obtained in three different sets of catalytic runs, together with the operating conditions employed in each case, are reported. In particular in Table 2, the results of the runs performed by changing temperature (100180 °C) and ethanol residence time (10-65 gcat‚h/mol) but keeping the molar ratio between ethanol and oxygen constant are reported. In contrast, inTable 3 are reported the results obtained in the runs performed by changing the temperature (140-160 °C), the ethanol residence time (10-200 gcat‚h/mol), and the molar ratio between the reagents. Finally, in Table 4 are reported the results of the kinetic runs performed at different temperatures (130-190 °C) on different catalysts characterized by a different vanadium contents. In these latter runs, the other operating conditions, such as the ethanol residence time and reagent molar ratio, were kept constant. In correspondence with each experimental flow rate and temperature combination, different samples of the gaseous outlet mixture were withdrawn and sent to the GC for analysis, to evaluate the ethanol conversion and product yields. Data reported in Tables 2-4 are averaged values for both the conversion and yields, evaluated under steady-state conditions, although all available data were used for the mathematical regression analysis. Additional kinetic runs were then performed to evaluate the effect on the reaction rate of the presence of water in the feed stream. To this end, a mixture of water and ethanol in a molar ratio 1:1 was used. The results

Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004 1627 Table 3. Operating Conditions and Experimental Results for Runs Performed with a Variable Ethanol/Oxygen Molar Ratio feed composition ethanol oxygen helium ethanol produst yielda (%) W/F catalyst (liquid) (gas) (gas) temperature conversiona (gcat/h‚molEtOH) (g) (mL/h) (mL/min) (mL/min) (°C) (%) acetaldehyde acetic acid acetals CO2 ethylic ether 9.7 26.6 48.6 97.3 194.6 9.7 11.4 14.9 19.5 26.6 48.6 97.3 149.7 194.6 a

0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33 0.33

2.00 0.73 0.40 0.20 0.10 2.00 1.70 1.30 1.00 0.73 0.40 0.20 0.13 0.10

7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7

22 22 22 22 22 22 22 22 22 22 22 22 22 22

140 140 140 140 140 160 160 160 160 160 160 160 160 160

11.96 17.83 28.35 58.28 81.43 27.24 28.20 31.85 41.25 58.64 88.97 87.26 100.0 99.59

11.06 16.30 25.84 52.50 70.04 23.42 24.16 27.39 35.40 47.97 69.39 70.00 78.14 87.93

0.07 0.09 0.04 -

0.52 0.86 1.41 4.73 5.04 1.70 2.04 2.31 3.06 6.22 9.39 7.82 10.09 10.37

0.85 0.78 0.82 1.10 2.15 6.94 6.40 9.19 -

0.38 0.67 1.09 1.05 7.33 1.20 1.14 1.29 1.63 2.30 3.24 3.03 2.57 1.29

Conversion and yields are averaged values.

Table 4. Operating Conditions and Experimental Results for Runs Performed with Catalysts of Various Vanadium Contents and Prepared with Different Grafting Solventsa feed composition catalyst acronym V/Ti-Si(1)

V/Ti-Si(4)

(V-Ti)h/Si(D)

(V-Ti)h/Si(T)

a

ethanol oxygen ethanol produst yield (%) catalyst (liquid) (gas) temperature conversion (g) (mL/h) (mL/min) solvent (°C) (%) acetaldehyde acetic acid acetals CO2 0.5 0.5 0.5 0.5 0.5 0.5 0.3 0.3 0.3 0.3 0.3 0.3 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

1.1 1.1 1.1 1.1 1.1 1.1 0.74 0.74 0.74 0.74 0.74 0.74 1.1 1.1 1.1 1.1 1.1 1.1 1.1 1.1

7.7 7.7 7.7 7.7 7.7 7.7 5.13 5.13 5.13 5.13 5.13 5.13 7.7 7.7 7.7 7.7 7.7 7.7 7.7 7.7

hexane hexane hexane hexane hexane hexane hexane hexane hexane hexane hexane hexane dioxane dioxane dioxane dioxane THFb THFb THFb THFb

100 130 150 160 170 190 100 120 140 160 170 180 130 150 170 190 130 150 170 190

0.50 1.30 3.12 5.52 7.77 29.50 3.24 7.01 23.24 37.16 54.25 76.51 7.01 10.76 23.67 56.54 10.21 22.31 53.20 95.16

0.50 1.30 2.71 4.63 6.53 26.50 3.07 6.55 21.22 31.79 45.82 63.15 6.48 9.73 19.67 40.16 9.63 19.25 45.15 67.33

0.15 0.09 0.10 0.06 0.13 0.07 0.26 0.09

0.41 0.82 1.25 2.48 0.11 0.25 1.31 2.67 4.90 7.78 0.40 0.80 2.22 2.23 0.34 1.52 3.93 6.41

1.12 1.41 2.48 1.00 6.17 0.83 1.81 15.85

ethylic ether 0.51 0.06 0.20 0.72 1.43 2.03 2.99 0.13 0.23 0.64 7.85 0.25 0.63 2.05 5.68

W/F ) 26.6 gcat/h‚molEtOH, helium feed ) 22 mL/min. b THF ) tetrahydrofurane.

Table 5. Operating Conditions and Experimental Results for Runs Performed with a 1:1 Ethanol/Water Molar Ratio in the Feed feed composition ethanol oxygen helium ethanol produst yield (%) W/F catalyst (liquid) (gas) (gas) temperature conversion (gcat/h‚molEtOH) (g) (mL/h) (mL/min) (mL/min) (°C) (%) acetaldehyde acetic acid acetals CO2 ethylic ether 9.7 26.6 48.6 97.3

0.33 0.33 0.33 0.33

2.00 0.73 0.40 0.20

7.7 7.7 7.7 7.7

22 22 22 22

160 160 160 160

obtained are collected in Table 5. As can be seen by comparing the conversions and yields of this table with those reported in Table 3, obtained under the same conditions, the ethanol reaction rate is negatively affected by the presence of water in the feed stream. Another interesting observation is that the selectivity is strongly increased, changing from 85 to 95%, because of the absence of some byproducts such as acetals and CO2. A kinetic experiment was then performed by directly feeding acetaldehyde, instead of ethanol, onto the

20.30 55.11 63.91 81.01

19.22 53.19 61.67 78.54

-

0.07 -

-

0.74 1.91 2.24 2.47

catalyst under the same conditions as used for the ethanol-water mixture. The results of this experiment confirmed the strong resistance of acetaldehyde to oxidation on vanadium catalysts, giving only small amounts of CO2 and acetic acid. In addition, two other kinetic experiments were conducted to evaluate the influence of the acid-base properties of the catalysts. In one case, ethanol containing a small amount of pyridine was fed into the reactor, kept at 160 °C, which contained 0.31 g of a catalyst with a V2O5 content of 3.56%, giving a residence time of 26.6

1628 Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004

dFi dW

Figure 2. Ethanol conversion vs amount of pyridine (micromoles) fed to the reactor.

gcat‚h/mol. In Figure 2, the ethanol conversions obtained after the addition of a stoichiometric amount of pyridine (123.6 µmol/g) and two and three times this quantity are reported. As can be seen, the activity of the catalyst initially decreased slightly and then remained constant. The selectivities remained roughly unchanged. The conclusion is that the acid character of the catalyst does not influence this reaction and the slight decrease of activity observed initially can reasonably be attributed to the steric hindrance of the adsorbed pyridine. The second experiment consisted of saturating the catalyst with CO2, to observe the eventual effect of basic sites on the reaction rate. The conversion remained unchanged, in agreement with the TPD observations, according to which no basic sites of reasonable strength are present on the surface of this catalyst. As mentioned before, the catalyst was also reduced with hydrogen. With the reduced catalyst, kinetic runs were performed at 160 °C with a residence time of 26.6 gcat‚h/mol. The reduced catalyst exhibited a slight decrease in ethanol conversion, from 0.58 to 0.48, with an unchanged selectivity. As V5+ is reduced by hydrogen to V3+, which is unstable in the presence of the oxygen of the reaction, the valence of vanadium is probably intermediate between 3 and 5. However, this seems to have a small effect on the reaction rate and no effect on the selectivity. Reaction Kinetics and Mechanism. On the basis of the obtained products, the following reaction scheme can be assumed

Nr

)

Ri,j(rj) ∑ j)1

with i ) 1-8 representing the involved components as numbered above and j ) 1-5 representing the reactions in the assumed scheme; Ri,j represents the corresponding stoichiometric coefficients for component i in reaction j. The integration is possible only after the definition of the kinetic laws for each of the five mentioned reactions. To define the kinetic laws for the oxidation reactions (eqs 1-3), we first considered the four-step reaction mechanism depicted in Chart 1, which has been proposed by different authors for the ODH of ethanol on different oxide catalysts3 and for the ODH of methanol on vanadium-based catalysts.15 An alternative mechanism could be suggested in which the first step is the chemisorption of oxygen to form a vanadium peroxy radical that can withdraw the R-hydrogen from ethanol, thereby favoring the formation of acetaldehyde. Reduction of the peroxide by decomposition would restore the original catalytic site. A similar mechanism was suggested to explain the oxidation of butane to maleic anhydride by Agaskar et al.26 However, this mechanism can probably also be simplified to a Mars and van Krevelen-like mechanism for interpreting the kinetic data. By considering the previously reported mechanism and assuming the second and third steps to be the slowest, we can write it in a compact way in which only two reactions appear, the first being reduction and the second oxidation (see Chart 2). The two reactions can also be written as

Siteox + CH3CH2OH f Sitered + CH3CHO Sitered + 1/2O2 f Siteox + H2O This mechanism is formally identical to that of Mars and van Krevelen12 and gives rise, therefore, to the following kinetic laws for the oxidation reactions

r1 ) 1+

1

(1) CH3CH2OH + /2O2 f CH3CHO + H2O (2) CH3CHO + 1/2O2 f CH3COOH

r2 ) 1+

(3) CH3COOH + 2O2 f 2CO2 + 2H2O (4) 2CH3CH2OH + CH3CHO T CH3CH(OCH2CH3)2 + H2O (5) 2CH3CH2OH f C2H5OC2H5 + H2O To simplify the kinetic approach, we assigned a number to each of the components as follows: 1, ethanol; 2, acetaldehyde; 3, acetic acid; 4, acetal; 5, carbon dioxide; 6, oxygen; 7, water; and 8, ethyl ether. Because the reactor is isothermal, it is possible to evaluate the concentration profiles of both the reagents and the products inside the reactor by integrating the differential equation

(1)

r3 ) 1+

k1P1 k1P1

(2)

koxPO21/2 k2P2 k2P2

(3)

koxPO21/2 k3P3 k3P3

(4)

koxPO21/2

where koxPO21/2 is the rate of the active site reoxidation, which is the same in all cases. Simpler kinetic laws were adopted for the reactions 4 and 5, considering that the contributions of these reactions are very low

r4 ) k4P2

(5)

r5 ) k5P12

(6)

Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004 1629 Chart 1

Chart 2

To the temperature dependence into account, all of the kinetic parameters appearing in eqs 2-6 were expressed in terms of two parameters according to the Arrhenius law

kj ) Aj exp(-Ej/RT)

(7)

Then, the system of differential eqs 1 was numerically integrated using a fourth-order Runge-Kutta method. All kinetic data related to Tables 2 and 3 were subjected to mathematical regression analysis involving mini-

mization of the objective function N

Φ(β) )

2 [Xexp - Xcal ∑ i i (β)] i)1

(8)

where X represents the conversion or a yield, N is the number of experimental runs, and β is the vector of the kinetic parameters. The best fitting parameters obtained are reported in Table 6. In Figure 3 the conversions of ethanol obtained at different temperatures for the runs of Table 2 are

1630 Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004 Table 6. Arrhenius Parameters for the Considered Reactions ln K ) ln A - Ea/RT reaction

ln A

Ea (kcal/mol)

correlation coefficient

ethanol f acetaldehyde catalytic site reoxidation acetaldehyde f acetic acid acetic acid f CO2 ethanol + acetaldehyde f acetals ethanol f ethylic ether

12.61 ( 1.75 10.47 ( 1.46 27.04 ( 25.95 44.56 ( 33.32 5.46 ( 5.08 48.64 ( 22.01

10.9 ( 1.4 11.7 ( 1.2 35.2 ( 21.3 47.1 ( 27.4 7.5 ( 4.2 47.0 ( 18.1

0.9347 0.9593 0.6365 0.6517 0.6675 0.7925

Figure 3. Ethanol conversion as a function of residence time for a feed with constant ethanol/oxygen molar ratio at different temperatures.

Figure 4. Acetaldehyde yield as a function of residence time for a feed with constant ethanol/oxygen molar ratio at different temperatures.

reported as a function of the ethanol residence time. Symbols are experimental data, whereas lines are calculated. The deviation observed for the run performed at the high temperature of 180 °C is probably due to the difficulty of controlling the bed temperature, considering the strong exothermicity of the reaction (-41.35 kcal/mol). The agreement obtained in the other cases is satisfactory. In Figure 4, the simulation results obtained for the same runs but related to the yields of the main reaction product acetaldehyde are reported. In Figure 5, the conversions of ethanol obtained at the two different temperatures of 140 and 160 °C are reported as a function of the ethanol residence time. These kinetic runs correspond to the second set of runs

Figure 5. Ethanol conversion as a function of residence time for a feed with variable ethanol/oxygen molar ratio at different temperatures.

Figure 6. Acetaldehyde yields as a function of residence time for a feed with variable ethanol/oxygen molar ratio at different temperatures.

reported in Table 3 in which the ratio between the reagents was largely changed. Again, symbols represent experimental data, and lines are calculated. The obtained agreement is quite satisfactory. In Figure 6, the simulation results obtained for the same runs as Figure 5 but related to the yields of acetaldehyde are reported. In Figure 7, an example of the simulation of the evolution with time of the product distribution is reported for a fixed temperature of 160 °C. As can be seen, in all cases, a satisfactory agreement was obtained, but the validity of the adopted kinetic model can be better appreciated in Figures 8 and 9, where all of the experimental data for the ethanol conversions and acetaldehyde yields, respectively, are compared with the corresponding calculated values.

Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004 1631

Figure 7. Acetaldehyde yields and byproduct distribution as a function of residence time for a feed with variable ethanol/oxygen molar ratio at a temperature of 160 °C.

Figure 10. Arrhenius plots for the kinetic constants related to ethanol oxidation and to reoxidation of the catalytic sites.

Figure 8. Calculated vs experimental ethanol conversions. Figure 11. Ethanol conversion for catalysts at different vanadium loading. The curves are simulations obtained by correlating kinetic constants with vanadium content.

Figure 9. Calculated vs experimental acetaldehyde yields.

Finally, in Figure 10, the Arrhenius plots for the kinetic constants k1 and kox are reported. As can be seen, a linear trend was obtained in both cases. It is worth mentioning that the slopes in the two cases are similar, in agreement with the similar activation energies obtained in the regression analysis. This means that the ratio of k1/kox can be considered roughly constant for

different temperatures. It is then clear that the catalyst reoxidation is the lower step in the redox cycle; this is likely a peculiarity of the catalyst depending on both the type of vanadium sites prevalent on the surface and the chemical environment. In fact, the activation energy found by us for a V2O5/TiO2-SiO2 catalyst is in agreement with the values found by Quaranta et al.,9 who used well-dispersed vanadium-based catalysts on TiO2SiO2, by Oyama and Somorjai,8 who used the same support but poorly dispersed catalysts. The activation energy found by Gomez et al.11 was, on the contrary, higher (16.2 kcal/mol for the main oxidation reaction and 23.5 kcal/mol for the catalyst reoxidation) on a VMgO catalyst. This clearly shows the advantage of a TiO2 vanadium chemical environment. It is worth mentioning, finally, that ethanol oxidation to acetaldehyde is probably structure-insensitive, as suggested by Oyama and Somorjai,8 when crystalline aggregates of V2O5 are present in catalysts containing high vanadium amount. In this case, V-O-V bonds, which are largely prevalent, promote the reaction. When molecular dispersion is obtained, the V-O-support bonds become primarily responsible for the observed activities and selectivities. The selectivities, in the latter case, are higher than thoses observed for poorly dis-

1632 Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004

persed and crystalline V2O5 catalysts.9,10 The runs performed on well-dispersed catalysts, with different vanadium loadings, can be roughly interpreted assuming a linear dependence of the reaction rate with the vanadium catalyst concentration, as can be appreciated in Figure 11. Conclusions Very active and selective catalysts have been prepared by grafting vanadyl alkoxide onto a silica support coated with TiO2. This type of catalyst gives rise to high conversion of ethanol to acetaldehyde at very low temperature (130-180 °C), which represents a useful perspective in view of industrial applications. The high dispersion of the catalyst strongly improves the selectivity, and the activity is a linear function of vanadium load amount. Acetaldehyde is relatively stable on this catalyst, and this is the reason for the high selectivities observed. The acid and basic properties of the catalyst have a minimal influence on the catalyst performance in this reaction, and therefore, only the redox properties are responsible for the reaction. The mechanism of the reaction does not involve surface lattice oxygen, because, at the temperature used, the oxygen exchange reaction is too slow. We have suggested two different possible mechanisms that, when simplified, correspond in the kinetic analysis to a classical Mars and van Krevelen kinetic law. The kinetic parameters for all of the occurring reactions were evaluated, and the activation energy found for the main reaction to acetaldehyde is in agreement with the values reported in the literature on the same type of catalyst.8,9 The VMgO11 catalyst, in contrast, has exhibited higher activation energies for both ethanol oxidation to acetaldehyde and catalyst reoxidation. This means that the TiO2 vanadium chemical environment is favorable to this reaction, in agreement with the suggestion made by other authors that V-O-support bonds are determinant for the activity and selectivity of different reactions, because the effect of the support is often dramatic. Acknowledgment Thanks are due to MIUR (Italian Ministry of Research and Education) for the financial support. List of Symbols Aj ) preexponential factor for kinetic constant kj Ej ) activation energy for kinetic constant kj Fi ) molar flow rate of component i in the feed kj ) kinetic constant for reaction j kox ) kinetic constant for catalytic site reoxidation P1 ) partial pressure of ethanol in the gas mixture P2 ) partial pressure of acetaldehyde in the gas mixture P3 ) partial pressure of acetic acid in the gas mixture PO2 ) partial pressure of oxygen in the gas mixture rj ) rate of reaction j W ) weight of catalyst loaded in the reactor Xcalc ) calculated values of conversion or yields i Xexp ) experimental values of conversion or yields i Ri,j ) stoichiometric coefficient of component i in reaction j

β ) vector of adjustable parameters Φ ) objective function

Literature Cited (1) Acetaldehyde. In Ullmann’s Encyclopedia of Industrial Chemistry, 6th ed.; Wiley-VCH: Weinheim, Germany, 2002; Vol. A1, pp 31-44. (2) Chauvel, A.; Lefebre, G.; Castex, L. Procedes de petrochimies Caracteristiques techniques et economiques; Technip: Paris, 1986; Vol. 2, p 33. (3) Idriss, H.; Seebauer, E. G. Reactions of ethanol over metal oxides. J. Mol. Catal. A: Chem. 2000, 152, 201. (4) Tsuruya, S.; Tsukamoto, M.; Watanabe, M.; Masai, M. Ethanol oxidation over Y-type zeolite ion-exchanged with copper(II) and cobalt (II) ions. J. Catal. 1985, 93, 303. (5) Parlitz, B.; Hanke, W.; Fricke, R.; Richter M.; Roost, U.; Ohlmann, G. The catalytic oxidation of ethanol on Cr/SiO2 catalysts and some relations to the structure. J. Catal. 1985, 94, 24. (6) Ling, T. R.; Chen, Z. B.; Lee M. D. Catalytic behavior and electrical conductivity of LaNiO3 in ethanol oxidation. Appl. Catal. A: Gen. 1996, 136, 191. (7) Bagiev, V.; Dadashev, G.; Adzhamov, K.; Alkhazov, T. Hetero-homogenic conversion of ethanol on Sn-Mo oxide catalyst. Kinet. Katal. 1992, 33 (3), 460. (8) Ted Oyama, S.; Somorjai, G. A. Effect of Structure in Selective Oxide Catalysis: Oxidation Reactions of Ethanol and Ethane on Vanadium Oxide. J. Phys. Chem. 1990, 94, 5022. (9) Quaranta, N. E.; Soria, J.; Corte´s Corbera`n, V.; Fierro, J. L. G. Selective Oxidation of Ethanol to Acetaldehyde on V2O5/TiO2/ SiO2 catalysts. J. Catal. 1997, 171, 1. (10) Santacesaria, E.; Sorrentino, A.; Tesser, R.; Di Serio, M.; Ruggiero, A. Oxidative dehydrogenation of ethanol on V2O5/TiO2SiO2 catalysts obtained by grafting vanadium and titanium alkoxides on silica. J. Mol. Catal. A: Chem. 2003, 204-205, 617. (11) Gomez, M. F.; Arrua, L. A.; Abello, M. C. Kinetic Study of Partial Oxidation of Ethanol over VMgO Catalyst. Ind. Eng. Chem. Res. 1997, 36, 3468. (12) Mars, P.; van Krevelen, D. W. Oxidations carried out by means of vanadium oxide catalysts. Chem. Eng. Sci. 1954, 3, 41. (13) Lakshmi, J. L.; Ihasz, N. J.; Miller, J. M. Sinthesis, characterization and methanol partial oxidation studies of V2O5 catalysts supported on TiO2-SiO2 and TiO2-ZrO2 sol-gel mixed oxide. J. Mol. Catal. A: Chem. 2001, 165, 199. (14) Inamaru, K.; Misono, M.; Okuhara, T. Structure and catalysis of vanadium oxide overlayers on oxide supports. Appl. Catal. A: Gen. 1997, 149, 133. (15) Weckhuysen, B. M.; Keller, D. E. Chemistry, spectroscopy and the role of supported vanadium oxides in heterogeneous catalysis. Catal. Today 2003, 78, 25. (16) Wachs, I. E.; Briand, L. E.; Jih-Mirn, Jehng; Burcham, L.; Gao, X. Molecular structure and reactivity of the group V metal oxides. Catal. Today 2000, 57, 323. (17) Vogt, E. T. C.; Boot, A.; van Dillen, A. J.; Geus, J. W.; Janssen, F. J.; van der Kerkhof, F. M. G. Preparation and performamce of a silica-supported vanadia on titania catalyst for the selective reduction of nitric oxide with ammonia. J. Catal. 1988, 114 (2), 313. (18) Iengo, P.; Di Serio, M.; Gazzoli, D.; Santacesaria, E. Preparation and properties of new acid catalysts obtained by grafting alkoxides and derivatives on the most common support. Part III: Grafting titanium alkoxides and sulphate derivatives on silica. Appl. Catal. 1998, 170, 225. (19) Monaci, R.; Rombi, E.; Solinas, V.; Sorrentino, A.; Santacesaria, E.; Colon, G. Oxidative dehydrogenation of propane over V2O5/TiO2/SiO2 catalysts obtained by grafting titanium and vanadium alkoxides on silica. Appl. Catal. A.: Gen. 2001, 214, 203. (20) Iannazzo, V.; Neri, G.; Galvagno, S.; Di Serio, M.; Tesser, R.; Santacesaria, E. Oxidative dehydrogenation of isobutane over V2O5- based catalysts prepared by grafting vanadyl alkoxides on TiO2-SiO2 supports. Appl. Catal. A: Gen. 2003, 246, 49. (21) Snell, F. R. D.; Ettre, L. S. Encyclopedia of Industrial Chemical Analysis; Interscience: New York, 1974; Vol. 19, p 107. (22) Brunauer, S.; Emmet, P. H.; Teller, E. Adsorption of gases in multimolecular layers. J. Am. Chem. Soc. 1938, 60, 309.

Ind. Eng. Chem. Res., Vol. 43, No. 7, 2004 1633 (23) Dollimore, D.; Heal, G. R. An improved method for the calculation of pore-size distribution from adsorption data. J. Appl. Chem. 1964, 14, 109. (24) Astorino, E.; Peri, J. B.; Wiley, R. J.; Busca, G.; Spectroscopic characterization of silicalite-1 and titanium silicalite-1. J. Catal. 1995, 157, 482. (25) Odenbrand, C. U. I.; Andersson, S. L. T.; Andersson, L. A. H.; Brandin, J. G. M.; Busca, G. Characterization of silica-titania mixed oxides. J. Catal. 1990, 125 (2), 541.

(26) Agaskar P. A.; De Caul, L.; Grasselli, R. K. A molecular level mechanism of n-butane oxidation to maleic anhydride over vanadyl pyrophosphate. Catal. Lett. 1994, 23, 339-351.

Received for review October 14, 2003 Revised manuscript received January 19, 2004 Accepted January 27, 2004 IE034182S