Ind. Eng. Chem. Res. 2005, 44, 645-650
645
APPLIED CHEMISTRY Gas-Phase Epoxidation of Propylene and Ethylene Torsten Berndt* and Olaf Bo1 ge Leibniz-Institut fu¨ r Tropospha¨ renforschung e.V., Permoserstrasse 15, 04318 Leipzig, Germany
New results from the catalyst-free epoxidation route for small alkenes in the homogeneous gas phase using O3 as the oxidizing agent (Berndt, T.; Bo¨ge, O.; Heintzenberg, J.; Claus, P. Ind. Eng. Chem. Res. 2003, 42, 2870.) are presented. The formation of propylene oxide was studied in a wide range of experimental conditions such as total pressure, temperature, feed composition, residence time, and reactor dimension. In the absence of dilution gas, for a propylene conversion of 91.8%, a selectivity to propylene oxide of 93.3% was found. An optimal O3 utilization of ∼95% (reacted propylene/initial O3 ) 0.95) was achieved by limiting the propylene conversion to 2030%. The epoxidation of ethylene yielded a selectivity to ethylene oxide of 90% and better in a wide range of conditions. The maximum formation rates of propylene oxide and ethylene oxide measured were 3640 and 2870 g per hour and per liter of reaction volume, respectively. Introduction Small epoxides of interest on an industrial scale are propylene oxide (PO) and ethylene oxide (EO) with annual worldwide production capacities of ∼5 × 106 tons each.1-3 In the future, an increasing market demand of these epoxides is expected. They are precursors for the production of polymers (polyesters, polyurethanes) or solvents (glycols). For EO, there exists a well-established direct formation route oxidizing ethylene by oxygen on a silver catalyst.1,3 For PO, however, this route does not work. Presently, PO is produced in complex processes either via the antiquated chlorohydrin route coupled with a large effluent load or via indirect oxidation routes (use of peracids or hydroperoxides) coupled with the formation of large amounts of coproducts.2,3 For the latter, there exists an enhanced fire and explosion risk. Therefore, a more efficient and environmentally acceptable route for PO is desired comparable to that of EO.4 An attractive route seems to be the oxidation of propylene with H2O2 producing PO with high selectivity (> 90%) and water in the presence of titanium silicalite.5,6 Disadvantages here are the expensive H2O2 production and the explosive potential associated with concentrated peroxide solutions. At present, however, the H2O2 route is favored for the commercial process of the future.7 A further interesting route is the oxidation of propylene with O2/H2 mixtures over supported nanosized gold particles producing PO with high selectivity (>99%) but at very low propylene conversion (1-2%).8 Very recently, results of an improved approach were reported describing a propylene conversion close to 10% with a selectivity to PO better than 90%.9 Remaining hurdles are the inefficient utilization of H2 and the strong deactivation of the catalyst with time on stream. In a first communication from this laboratory, results of a novel route operating in the homogeneous gas phase * To whom correspondence should be addressed. E-mail:
[email protected]. Fax: +49 341 235 2325.
without a catalyst were published.10 The reaction was conducted in a flow tube under low-pressure conditions with a residence time of less than 10 ms. In this process, the “real” oxidizing agent is a nitrogen(V) oxide (NO3, N2O5, N2O6) arising from the reaction of O3 with NO2. In the overall process, NO2 is not consumed. In the absence of NO2, however, the epoxidation step does not work and a simple ozonolysis takes place. Using He as dilution gas, a selectivity to PO of up to 98% was achieved for a propylene conversion of 30%. The process is not afflicted with the production of coproducts, and no noticeable environmental impact is expected. The subject of the present work is a more comprehensive characterization of this reaction process depending on total pressure, temperature, feed composition, residence time, and reactor dimension. Key parameters are the selectivity to PO, the propylene conversion, and the formation rate of PO per volume. Because O3 represents an expensive reactant (such as propylene), special attention was also paid to an efficient utilization of oxidizing agent O3. In addition, the general approach was also applied for the epoxidation of ethylene. Here, a comparison of the reaction parameters is possible taking as the reference the data of the established, Agcatalyzed direct oxidation route of ethylene. Experimental Section The experimental setup was roughly the same as that in the previous work.10 In addition to the flow tube of the former study, here reactor I (length, 100 mm; inner diameter, 5 mm; quartz glass), in a few experiments a second tube was used, called reactor II (length, 150 mm; inner diameter, 17 mm; quartz glass). The resulting surface/volume ratios are 8 (reactor I) and 2.35 cm-1 (reactor II). O3 was produced in an ozone generator (Wedeco, 16 HC; O2 feed) and the concentration was monitored by UV absorption measurements at 255 nm (PerkinElmer, Lambda 2). The resulting O3/O2 mixture (5-8 vol % O3 in O2) was brought together with NO2 in
10.1021/ie049464m CCC: $30.25 © 2005 American Chemical Society Published on Web 01/19/2005
646
Ind. Eng. Chem. Res., Vol. 44, No. 4, 2005
a gas mixer held at room temperature, NO2/O3 ) 3-5.5 on a molar basis. Before entering the heated reaction zone, the stream of propylene (ethylene) was added. If needed, the O3/O2/NO2 mixture as well as the olefin stream was diluted with N2. Downstream the heated reaction zone, the reactor pressure was measured by a capacitive manometer (Baratron). For taking the reactor temperature, a thermocouple was located at half the length of the heated reaction zone at the outside of the reactor wall. The measured temperature stability was (1-2 °C (Sto¨rk Tronic, TRS 4). All flow rates were set by calibrated mass flow controllers (MKS, 1259/1179). For gas-phase analysis, the reaction gas mixture was pumped continuously through a cell equipped with a White mirror system for on-line FT-IR spectroscopy (Nicolet, Magna 750). The olefins, the oxides, acetaldehyde, and CO2 were analyzed simultaneously from the FT-IR measurements using calibrated spectra of each species. The uncertainty of these determinations was less than 3-5%. From FT-IR measurements, there was no clear indication for the formation of CO. Caused by larger uncertainties in the determination of the amount of reacted ethylene, in this reaction system the selectivity to the products (EO or CO2) is given with respect to the initial O3 concentration. In each case, the total consumption of O3 during the reaction was confirmed by FT-IR measurements. For the reaction with propylene, the stated selectivity to the products (PO, acetaldehyde, CO2) is based on the reacted propylene. Furthermore, for on-line GC-MS analysis (HP 6890 with HP-MSD 5973), a small gas stream was pumped continuously through a GC loop for identification of the oxidation products including the trace compounds. The chemicals and gases used had a stated purity as follows: ethylene, 99.9% (Linde); propylene, 99.5% (Messer); ethylene oxide, 99.8% (Fluka); propylene oxide, >99% (Aldrich); acetaldehyde, >99.5% (Fluka); NO2, 98.5% (Gerling Holz); O2, 99.9996% (Linde); N2, 99.999% (Linde). Results and Discussion Epoxidation of Propylene. (1) Dependence on Total Pressure. The results from the first communication10 indicated that the proposed route is bound to lowpressure conditions clearly below atmospheric pressure. That is an unusual reaction condition for an industrial process. Atmospheric pressure or higher seems to be more convenient, whereas from the point of view of chemical engineering such a low-pressure process is possible. We had to find out whether the low-pressure conditions were urgently needed, and if so, what was the preferred pressure range. Experiments were carried out at a total pressure of 25, 50, 100, and 200 mbar using nearly identical feed composition: 17-20 vol % propylene, 2.9-3.3 vol % O3, and 18 vol % N2; residence time 2.4 ms; usage of reactor I. At a temperature of 300 °C the selectivity to PO decreased clearly with increasing total pressure: 89.1 (25 mbar); 84.9 (50 mbar); 70.5 (100 mbar), and 56.6% (200 mbar). The selectivity to acetaldehyde showed an inverse behavior: 11.4 (25 mbar); 15.1 (50 mbar); 23.3 (100 mbar), and 25.6% (200 mbar). The same trend was observed for CO2. It is to be noted that CO2 was generally formed in low amounts close to the detection limit of the FT-IR analysis, here: 0.28 (25 mbar); 0.91 (50 mbar); 1.2 (100 mbar), and 1.4% (200 mbar). The amount of converted propylene regarding the initial O3 was close to unity in the whole
Figure 1. Selectivity to the products (propylene oxide, acetaldehyde, and CO2), propylene conversion, and ratio of reacted propylene/initial O3 depending on the initial ratio propylene/O3 for an O3 content of 4.3-5.2 vol % (temperature, 300 °C; pressure, 25 mbar; residence time, 2.4 ms; reactor I).
pressure range. From a mechanistic point of view, that stands for an efficient O3 utilization for the propylene conversion independent of total pressure. Assuming a lower limit for the selectivity to PO of 80% for a viable, commercial process, the data from this set of experiments suggest an operating pressure of 50 mbar or below. To investigate the influence of inert additions in the carrier gas, at a total pressure of 100 mbar, experiments with 56 vol % N2 (15 vol % propylene; 1.4 vol % O3) have been performed yielding a selectivity to PO of 79.9%. This is in accordance with the finding of the 100 mbar experiments in ref 10, where He was used as the dilution gas. Obviously, in the presence of sufficient, inert additions (N2, He, Ar, etc.) a selectivity to PO of 80% and better is possible as well for a total pressure of 100 mbar. (2) Dependence on Feed Composition. For an economically priced process it is desirable to perform the chemical conversion only in the presence of the reactants needed without any dilution gases. Furthermore, a high reactant conversion per pass is desirable, keeping the effort for separation and recycling as low as possible. Therefore, in the following investigations, the feed consisted of the O3/O2/NO2 stream from the gas mixer and propylene only. By varying the initial reactant concentrations from an excess of propylene over O3 up to equal molar conditions, the influence of the extent of propylene conversion on the selectivity to PO was investigated. Figure 1 shows experimental results at 300 °C depending on the initial ratio propylene/O3 for an O3 content of 4.3-5.2 vol % (pressure, 25 mbar; residence time, 2.4 ms; usage of reactor I). The selectivity to PO is not affected by the initial ratio of the reactants. Even for equal molar conditions, the selectivity remains at the level of more than 85%. For acetaldehyde there is a slight decrease and for CO2 an increase with decreasing initial ratio propylene/O3. The amount of converted propylene regarding the initial O3 drops from 99.7 (initial propylene/O3 ) 5) to 69.7% for
Ind. Eng. Chem. Res., Vol. 44, No. 4, 2005 647 Table 2. Influence of Residence Time and Reactor Dimension on the Reaction Parameters for the Epoxidation of Propylene reactor pressure/mbar temperature/°C residence time/ms feed C3H6/vol % feed O3/vol % ∆[C3H6]/[O3]0/% propylene conversion/% selectivity to PO/% selectivity to acetaldehyde/% selectivity to CO2/%
Figure 2. Selectivity to propylene oxide, propylene conversion, and the ratios of reacted propylene/initial O3 as well as N2O5 (offgas)/initial O3 for different temperatures (pressure, 10 mbar; 2.0 vol % propylene; 5.4 vol % O3; reactor I). Table 1. Experimental Data for a (Nearly) Equal Molar Feed of Propylene and O3: Influence of Residence Time, Concentration Level, and Total Pressure pressure/mbar temperature/°C residence time/ms feed C3H6/vol % feed O3/vol % propylene conversion/% selectivity to PO/%
25 275 2.3 5.5 5.2 66.0 86.7
25 275 29 5.1 5.0 72.1 85.6
25 275 29 2.4 2.5 68.9 86.3
10 275 3.1 5.1 5.0 67.8 94.2
the equal molar conditions (initial propylene/O3 ) 1.06), lowering the propylene conversion for the latter from the theoretical total conversion to 65.8%. Investigating the possible reason for the decreasing O3 efficiency with decreasing initial propylene/O3 ratios, the residence time (increase by a factor of 12), the concentrations of both reactants (the half each), and the total pressure (from 25 to 10 mbar) were changed. Table 1 summarizes the findings for a temperature of 275 °C. There is no clear dependence of the propylene conversion on one of the three parameters varied. Merely the selectivity to PO is significant higher in the 10 mbar experiment. This is due to the general increase of this parameter with decreasing total pressure. The results from this set of experiments suggest limiting the propylene conversion to 20-30%, cf. Figure 1. Under these circumstances, the introduced O3 oxidizes propylene very efficiently. It is to be noted here that O3 represents a valuable reactant making an efficient utilization necessary. Furthermore, it was of interest if a propylene conversion close to total is possible and what the selectivity to PO is at this working point. Figure 2 shows the temperature-dependent findings using O3 in excess over propylene (pressure, 10 mbar; 2.0 vol % propylene; 5.4 vol % O3; residence time, 2.4 ms calculated for 300 °C; usage of reactor I). It is remarkable that even under conditions of an O3 excess and, therefore, in the presence of high amounts of nitrogen(V) oxides, a selectivity to PO of 90% and higher can be obtained (300 °C; propylene conversion, 91.8%; selectivity to PO, 93.3%; reacted propylene/initial O3, 33.9%). In this example, only onethird of the introduced O3 is consumed in the process
I
I
II
25 300 1.6 22.5 4.2 94.6 17.7 88.7 11.6 0.44
25 300 25 20.3 4.1 93.6 18.9 88.0 12.7 0.34
25 300 27 20.3 4.2 92.4 18.9 89.3 12.3 0.45
of propylene epoxidation. By means of FT-IR spectroscopy, the occurrence of N2O5 in the off-gas was observed. N2O5 is one member of the nitrogen(V) oxides. Because O3 and, therefore, the nitrogen(V) oxides were present in excess over propylene, it is not so surprising to measure such a species. The presence of N2O5 downstream from the reactor, however, does not prove that N2O5 is the “real” oxidizing agent. N2O5 can be formed from other nitrogen(V) oxides in the presence of NO2 in the colder parts of the flow system after the heated reaction zone.11 From the data depicted in Figure 2, it can be speculated that at lower temperatures (below 225 °C) the O3 balance can be roughly fulfilled, ∆[propylene]/[O3] + [N2O5]/[O3] ) 100%. (Note: O3 + 2NO2 f N2O5 + O2.) With increasing temperature, the concentration of N2O5 decreases, indicating a thermal decomposition of N2O5 itself or one of the other nitrogen(V) oxides in the heated reaction zone. Obviously, a complete consumption of O3 (nitrogen(V) oxides as intermediates) seems to be possible only in the presence of an excess of propylene (propylene conversion lower than 20-30%). Otherwise nitrogen(V) oxides can thermally decompose, or in the case of low temperatures, they can leave the reactor in the off-gas. Products of the thermal decomposition are NO2 and O2: e.g., N2O5 f 2NO2 + 0.5O2 or NO3 f NO2 + 0.5O2.11 This is in line with the NO2 balance measured by means of FT-IR spectroscopy (NO2 loss fraction of