1310
Ind. Eng. Chem. Res. 1999, 38, 1310-1315
Catalytic Wet Air Oxidation on a Ru/TiO2 Catalyst in a Trickle-Bed Reactor Jean-Christophe Be´ ziat,† Miche` le Besson,*,† Pierre Gallezot,† and Sylvain Dure´ cu‡ Institut de Recherches sur la Catalyse-CNRS, 69626 Villeurbanne, France, and De´ partement Recherche, TREDI, 54505 Vandœuvre-le` s-Nancy, France
The catalytic wet air oxidation of oxygenated pollutants has been investigated in a co-current downflow trickle-bed reactor, over a temperature range 423-473 K at a total pressure of 5 MPa, using a 2.8% Ru/TiO2 catalyst. When the catalyst was employed under integral conditions (high catalyst mass/liquid flow ratio, approximate residence time of 5.1 h gRu L-1) it was possible to convert 5 g L-1 aqueous solutions of cyclohexanol, succinic acid, and acetic acid into CO2 and H2O. Experimental kinetic data were measured, in the reactor operating differentially at 463 K, at different conversions of succinic acid obtained at each step of the successive recycling of the reaction medium. The intermediate compounds detected were identified as acrylic and acetic acid. For the six main reactions identified, the reaction rates were described by LangmuirHinshelwood type rate equations, assuming competitive adsorption of liquid-phase components. The comparison of the experimental data with kinetic modeling showed an excellent agreement. The model indicates a very low adsorption of acetic acid (KACE e 0.005 L mmol-1) compared to succinic and acrylic acid (KSUC ) 0.13 and KACR ) 0.16 L mmol-1), even though the kinetic constants were of the same order of magnitude. Long-term experiments with this catalyst demonstrate its stability under the reaction conditions used. Introduction Wet air oxidation (WAO) is an emergent technology that depollutes organic effluents in industrial waste waters by oxidizing with air or oxygen the effluents dissolved or suspended in water. The complete mineralization of organic pollutants into CO2 requires high temperature and pressure conditions, which affect the economy of the process. This is due to expensive operating conditions and the large costs in reactor construction needed to cope with the high pressures and reactor corrosion. Therefore, catalysts that accelerate the rate of the WAO reaction and thus diminue the severity of reaction conditions are being sought to improve the competitivity of this technology with respect to conventional water treatment processes. Heterogeneous catalysts which can be easily separated from the reaction medium in batch or continuous processes are better suited than soluble catalysts which have to be recovered by an additional separation process.1-3 Different transition metal oxides have been used, but the main drawback is their dissolution in the corrosive reaction mixture.4-7 Previous studies in this laboratory8,9 have shown that aqueous solutions of carboxylic acids, including acetic acid which is particularly difficult to oxidize, can be completely converted into CO2 at 190200 °C under 50 bar of air in the presence of supported ruthenium catalysts. Active carbon and graphite supports were resistant to leaching in acidic medium, but they were slowly oxidized at temperatures higher than 150 °C and therefore could not be used for continuous operation over a long period of time. Titanium dioxide was selected as the support for ruthenium in view of its stability in acidic and oxidizing medium. Aqueous * To whom correspondence should be addressed. E-mail:
[email protected]. † Institut de Recherches sur la Catalyse-CNRS. ‡ TREDI.
solutions of succinic acid were completely oxidized with air at 190 °C on a 2.8 wt % Ru/TiO2 catalyst in stirred tank reactor with an initial rate of 43 molC h-1 molRu-1.10 The treated water contained no ruthenium nor titanium within the detection level of 0.5 ppm and 0.1 ppm, respectively, indicating that metal and support were not leached and the catalyst underwent no loss of activity in successive recycling operations after the second run. The WAO treatment of industrial waters, e.g., from the chemical and paper pulp industries, is potentially a very powerful technology if the waste waters could be depolluted as close as possible to their point of emission and the treated water immediately recycled in the production line. Indeed, closed loop operation minimalizes the risk of polluting the hydrosphere. This scheme implies that WAO should be conducted on line, preferentially in continuous reactors. The aim of the present work was to study the feasibility of WAO in a trickle-bed reactor. Aqueous solutions of succinic acid, acetic acid and cyclohexanol, were oxidized with air in a trickle-bed reactor packed with a fixed bed of Ru/TiO2 catalyst. Stress was laid on activity measurements over long periods of time because very stable heterogeneous catalysts are imperatively required to develop reliable and economically competitive WAO processes. Conversion and product distribution were measured as a function of residence time, and the kinetic modeling of succinic acid oxidation was achieved to extract the rate constants of the various steps in a scheme accounting for the product distribution. Experimental Section Catalysts. Most of the experiments were carried out with the 2.8% Ru/TiO2 catalyst (Engelhard ref Q500069, 50 m2 g-1) which was used for batch studies in stirred tank reactor.10 This catalyst in powder form (dp
10.1021/ie980539u CCC: $18.00 © 1999 American Chemical Society Published on Web 02/27/1999
Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 1311
Figure 1. Trickle-bed reactor setup. Key: R ) reactor; M ) manometer; TLC ) temperature indicator controller; FIC ) flow indicator controller; PCV ) pressure control valve; PSV ) pressure security valve.
) 10-5 m) could not be used as such in the fixed-bed reactor, which requires catalysts in grain form. The powder without lubricant was compressed in a manual pilling machine to obtain cylindrical pellets (3 mm large, 1 mm thick). The pellets were coarsely crushed and then screened to obtain granules in the size range 0.8-2 mm. Before use they were reduced at 300 °C under flowing hydrogen (15 L h-1), cooled under hydrogen, and passivated by flowing a mixture of 1% O2 in N2 at room temperature, for a controlled and limited surface oxidation.11 Reactor and Reaction Procedure. Experiments were carried out in a micropilot trickle-bed reactor schematically described in Figure 1. Air from a gas cylinder was depressurized to 7 MPa and its flow rate adjusted with a mass flow controller (Brook 5850 TR) to give a constant flow of 5-50 L h-1. Aqueous solutions of succinic acid were fed with a high-pressure pump (Shimadzu LC6A)(6-600 mL h-1). Air and liquid flows were mixed at the inlet of the reactor which consisted of an Hastelloy C22 tube equipped with an axially located thermocouple well and heated with an electrical furnace (length, 150 mm; internal diameter, 10 mm; internal volume, 12 cm3). The reactor was packed with the catalyst placed between two layers of metal-free, supporting material. Gas and liquid were flown in cocurrent downflow mode through the reactor. At the reactor outlet the liquid and gas flows were cooled and recovered in the gas-liquid separator. The gas flow was depressurized to atmospheric pressure by a backpressure regulator (Brook 5835P). To maintain a con-
stant level in the separator, the liquid was continuously withdrawn through an automatic valve (Kammer) which was controlled via a regulator (Pyromat 320S), using a manometer (Elsag-Bailey) measuring the differential pressure in the gas-liquid separator. The reactor loaded with 1-12 g of 2.8% Ru/TiO2 catalyst was operated under the following conditions. It was pressurized with 5 MPa of air flowing at 5 NL h-1. The aqueous solution of succinic acid at 5 g L-1 was pumped at high flow rate (300 mL h-1) to saturate the catalyst with the liquid and to fill partially the gasliquid separator, and then the liquid flow rate was reduced to standard operating conditions (60 mL h-1) and the regulation of liquid level in the separator was activated. The reactor temperature was increased by increments of 50 K every 30 min until the standard reaction temperature (463 K) was attained. Analysis. Samples were periodically withdrawn from the reactor to measure the pH, the total organic carbon (TOC) and the product distribution. Analysis of organic products was performed by HPLC with UV and RID detectors in series using an ion-exchange column (Sarasep Car-H) with dilute H2SO4 solutions as eluent (0.01 N, 0.5 mL min-1). During oxidation of succinic acid, acrylic and acetic acids were identified. Maleic, fumaric, and oxalic acids might be intermediates, but they are easily transformed. The overall abatement of organic effluents was monitored with a TOC analyser equipped with an IR detector (Shimadzu 5050A). The TOC was determined by subtracting the inorganic carbon (CO2 evolved after treatment of the sample in
1312 Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 Table 1
a
entry
time on stream (h)
effluent (5 g L-1)
T (K)
conversion (%)
acetic acid (%)
TOC abatement (%)
1 2 3 4 5 6 7 8 9 10 11 12
0-100 100-500 350-400 420-500 500-625 625-675 675-775 775-860 860-940 940-1000 1100-1200 1200-1300
succinic acid succinic acid succinic acid succinic acid cyclohexanol cyclohexanol cyclohexanol acetic acid acetic acid acetic acid succinic acid succinic acid
463-433 423-463 453 463 453 463 473 453 463 473 463 453
100-99.5 70-100 100 100 98.8 100 100 65.0 86.0 95.0 100 100
0-4.6 6.7-0.4 1.2 0.4 12.0 5.8 1.2 35.0 14.0 5.0 1.2 3.3
100-95 65-99.8 99.1 99.8 76.3 94.0 98.0 65.0 86.0 95 99.2 96.8
a
Reaction conditions: 5 MPa of air at 5 NL h-1; liquid flow rate, 60 mL h-1; corresponding residence time, 5.1 gRu h L-1.
Figure 2. Oxidation of succinic acid over 2.8% Ru/TiO2 catalyst. Influence of temperature on TOC abatement; residence time δ ) 5.1 h gRu L-1.
concentrated phosphoric acid) from the total carbon (CO2 evolved after catalytic combustion on platinum at 700 °C). The experimental TOC values were compared with the TOC values calculated from the HPLC results; there was good agreement between both techniques which allowed us to determine the conversion to carbon dioxide (error less than 3%). The initial reaction rates r0(SUC) and r0(TOC) were obtained at low conversion from the curves giving the succinic acid conversion or the TOC abatement (100 (TOC0-TOC)/TOC0) as a function of residence time expressed in gRu h-1 L-1. They were expressed either in molSUC h-1 molRu-1 or in molC h-1 molRu-1. Results and Discussion Studies at High Conversion of Different Oxygenated Compounds. It was first verified that TiO2 was not active for cyclohexanol and succinic acid oxidation under the standard conditions. The reactor was then loaded with 11 g of 2.8% Ru/TiO2 catalyst to obtain high conversion. Table 1 summarizes the successive WAO reactions runs carried out with different effluents at 5 g L-1 concentration and different temperatures (measured in the middle of the catalyst bed). At the start of the oxidation experiment at 463 K (entry 1), TOC abatement was close to 100%. Therefore, the temperature was gradually decreased to attain a TOC abatement level of 70% at 423 K. The TOC abatement for succinic acid oxidation as a function of reaction temperature and time on stream during the time interval 100-500 h (entries 2, 3, and 4) is given in Figure 2. Succinic acid was completely converted at 453-463 K with a TOC abatement greater than 99%. The remainder of carbon was found in the form of small amounts of acetic acid (1.2 and 0.4% at 453 and 463 K,
respectively). The oxidation of this short chain acid is often the slower step to achieve complete conversion of oxygenated molecules to CO2 such as phenolic compounds. It should be noted that the catalyst deactivated when the reaction was conducted at 423 K and, to a lesser extent, at 433 K. This was not a permanent deactivation since the catalyst reactivated when the reaction was conducted at higher temperatures. Indeed, after successive WAO of succinic acid, cyclohexanol, and acetic acid, the WAO of succinic acid was carried out again after 1100 h on stream and only minor deactivation was observed (compare entries 3 and 4 with 12 and 11 at 453 and 463 K, respectively). Indeed, the conversion of succinic acid was still complete, but slightly increased amounts of acetic acid were observed (3.3 and 1.1% instead of 1.2 and 0.4% at 453 and 463 K, respectively) and, as a consequence, lower TOC abatements. The reversible deactivation when the reactions were run at 423-433 K can be attributed to an increasing coverage of the ruthenium surface by oxygen. Cyclohexanol was slightly more difficult to convert into CO2 since the conversion was 98.8% at 453 K with a TOC abatement of 76.3%. As expected from previous studies8-10,12-14 acetic acid was the most difficult to oxidize (Table 1, entries 8-10). The analysis of the treated water by ICP-AES did not show any trace of ruthenium or titanium within the detection limit of 0.5 ppm for ruthenium and 0.1 ppm for titanium. Studies at Low Conversion of Succinic Acid. The reactor was stopped after 1300 h on stream and 1.5 g of the used catalyst was collected to reload the reactor. The oxidation of succinic acid solution was performed at constant temperature (463 K) and liquid flow rate (60 mL h-1) and under standard air pressure (5 MPa) and air flow rate (5 NL h-1). We verified that the temperature was constant along the catalyst bed, so that the reactor operated in an isothermal mode. The reaction was run for about 100 h, and then the treated solution was collected, analyzed, and submitted to another WAO run by recycling it in the liquid supply reservoir. To obtain higher residence times this procedure was repeated nine times albeit with shorter periods of time on stream (typically from 100 to 50 h). This procedure was used to increase the residence time while maintaining the liquid and the gas flow rates, thus the hydrodynamic effects and the wetting of the catalyst bed were kept constant. Figure 3a gives the conversion, TOC abatement, and product distribution as a function of residence time, and Figure 3b shows the concentration profiles of reaction products plotted vs TOC abatement for this experiment. The oxidation of succinic acid
Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 1313
Figure 4. Reaction scheme for the oxidation of succinic acid. SUC ) succinic acid, ACR ) acrylic acid, and ACE ) acetic acid.
Figure 3. Oxidation of succinic acid at 463 K: (a) concentration profiles of reaction products and TOC abatement as a function of the residence time; (b) concentration profiles vs TOC abatement. Key: O ) succinic acid; b acrylic acid; × ) acetic acid; + ) TOC. Table 2. Comparison of the Initial Activities in Batch and Trickle-Bed Reactors reactor
r0(TOC) (molC h-1 molRu-1)
r0(SUC) (molSUC h-1 molRu-1)
autoclave trickle bed
43 5
16 2
yielded small amounts of acrylic acid and up to 12.5% of acetic acid. The former disappeared rapidly whereas the concentration of the latter decreased only after succinic acid was completely converted. These results are quite similar to those obtained in batch studies conducted with the same solutions of succinic acid, under the same temperature and pressure conditions, and on the same 2.8% Ru/TiO2 catalyst, albeit in powder form.10 However, the initial rate of reaction was much higher in the autoclave as shown in Table 2 which gives the initial rate of TOC abatement and succinic acid conversion. The effect of catalyst mass was studied by performing the continuous reaction with 1 g of catalyst instead of 1.5 g. The fact that the reaction rate per mole of ruthenium was found to be the same may indicate that the reaction kinetics was not perturbed by external diffusion limitations. The difference in initial reaction rate could be due to internal mass transfer limitations in the pores of the compressed powder of 2.8% Ru/TiO2 used for the trickle-bed experiments and/or to a nonhomogeneous wetting of the catalyst by the trickling solution. Moreover the activity data for batch experi-
Figure 5. Oxidation of succinic acid at 453 K. Comparison of model predictions with experimental product concentrations. Rate parameters are as in Table 3.
ments was obtained from the fresh catalysts, whereas here the initial activity was measured after 1300 h on stream. Indeed, previous measurements in an autoclave10 have shown that the activity of the catalyst decreased with the oxygen coverage of the ruthenium particles down to a steady-state activity or to an inadequate coverage of catalyst particles by liquid reactant. The influence of the residence time on conversion and product distribution was carried out by operating at different liquid flow rates instead of recycling the solutions. The reaction data were very similar to those shown in Figure 3 except that the rates of succinic and acetic acid conversion at high residence time were slightly lower. This could be due to the fact that, when the reactor was operated at low liquid flow rates with constant air flow rate of 5 NL h-1, the liquid was dragged down the gas stream so that the real contact time of the liquid with the catalyst could be smaller. The total time on stream corresponding to the experiment conducted with the reactor loaded with 1.5 g of catalyst, taking into account the first series of experiment conducted with 11 g of catalyst (vide supra), was 1000 h. During this period of time, the initial rate of succinic acid conversion decreased from 2 to 1.7 molSUC h-1 molRu-1, i.e., experienced a 15% loss of activity.
1314 Ind. Eng. Chem. Res., Vol. 38, No. 4, 1999 Table 3. Calculated Values of the Kinetic Model Parameters rate const (mmol h-1 gRu-1)
adsorp const (L mmol-1)
k1
k2
k3
k4
k5
k6
KSUC
KACR
KACE
3.5
9.5
15
4.6
>55
32.5
0.13
0.16