Effect of Process Conditions on Olefin Selectivity during Conventional

a diaphragm metering pump (American Lewa, Inc.;. Model FCMK-1). The amount of propane produced .... drogen, was calculated from experimental values of...
1 downloads 19 Views 387KB Size
2580

Ind. Eng. Chem. Res. 1997, 36, 2580-2587

Effect of Process Conditions on Olefin Selectivity during Conventional and Supercritical Fischer-Tropsch Synthesis Dragomir B. Bukur,* Xiaosu Lang, Aydin Akgerman, and Zhentao Feng Department of Chemical Engineering, Kinetics, Catalysis and Reaction Engineering Laboratory, Texas A&M University, College Station, Texas 77843-3122

A precipitated iron catalyst (100 Fe/5 Cu/4.2 K/25 SiO2 on mass basis) was tested in a fixed-bed reactor under a variety of process conditions during conventional Fischer-Tropsch synthesis (FTS) and supercritical Fischer-Tropsch synthesis (SFTS). In both modes of operation it was found that: total olefin content decreases whereas 2-olefin content increases with either increase in conversion or H2/CO molar feed ratio. Total olefin and 2-olefin selectivities were essentially independent of reaction temperature. The effect of conversion was more pronounced during conventional FTS. Comparison of olefin selectivities in the two modes of operation reveals that total olefin content is greater while the 2-olefin content is smaller during SFTS. Also, both the decrease in total olefin content and the increase in 2-olefin content with increase in carbon number (i.e. molecular weight of hydrocarbon products) was significantly less pronounced during SFTS in comparison to the conventional FTS. The obtained results suggest that 1-olefins, and to a smaller extent n-paraffins, are the primary products of FTS. Secondary reactions (isomerization, hydrogenation, and readsorption) of high molecular weight R-olefins occur to a smaller extent during SFTS, due to higher diffusivities and desorption rates of R-olefins in the supercritical propane than in the liquid-filled catalyst pores (conventional FTS). Introduction R-Olefins are used as chemical intermediates for a number of important industrial and consumer products. The even-numbered carbon R-olefins (C4, C6, and C8) are used as comonomers for ethylene and propylene polymerization, whereas the higher molecular weight olefins are used in plasticizers, household detergents, and sanitizers (Lappin et al., 1996). Linear C10 olefins and others provide premium value synthetic lubricants. R-Olefins are produced in significant quantities during conventional Fischer-Tropsch synthesis (FTS) in fixedbed and fluid-bed reactors at Sasol in South Africa, and recently Sasol has built a large scale commercial plant for production of 1-pentene and 1-hexene utilizing raw streams from fluid bed FTS reactors (Waddacor, 1994). The purification process employed by Sasol entails a series of distillation steps to separate desired R-olefins from other products. Significant economic benefits can be achieved by increasing the alpha olefin selectivity of the raw FTS products and thus reducing the cost of product separation. Fischer-Tropsch synthesis in supercritical fluids provides means to accomplish this objective. In recent studies of FTS, on silica-supported cobalt-lanthanum and/or alumina-supported ruthenium catalysts, in a supercritical n-hexane Fujimoto and co-workers (Yokota and Fujimoto, 1991; Fan et al., 1992) have demonstrated certain advantages of this operation, including higher olefin selectivity, relative to gas phase and liquid phase (trickle bed) operation. In our Laboratory we studied FTS on a precipitated iron catalyst (Ruhrchemie LP 33/ 81 with nominal composition 100 Fe/5 Cu/4.2 K/25 SiO2 on mass basis) in a fixed-bed reactor at 250 °C using syngas with H2/CO molar feed ratio of 0.67 (coal-derived synthesis gas) and gas space velocity of 2 L (NTP)/(g of catalyst)‚h, under both supercritical (total pressure of 7 MPa; propane’s partial pressure is 5.5 MPa) and * To whom correspondence should be addressed. FAX: (409) 845-6446. E-mail: [email protected]. S0888-5885(96)00507-6 CCC: $14.00

conventional (P ) 1.5 MPa) operating conditions (Lang et al., 1995). It was found that supercritical FTS results in enhanced selectivity to 1-olefins (R-olefins) relative to conventional FTS, but it does not have significant effect on catalyst activity and hydrocarbon product distribution. The enhanced selectivity of high molecular weight R-olefins during SFTS is believed to be due to their higher diffusivities and desorption rates relative to conventional operation (Yokota and Fujimoto, 1991; Fan et al., 1992; Lang et al., 1995). In the absence of intraparticle diffusional limitations, the primary reaction products (R-olefins) can leave the catalyst pores without undergoing secondary reactions (isomerization, hydrogenation, and/or readsorption). Here, we report results from a study with the Ruhrchemie LP 33/81 catalyst, which was used originally in Arge fixed-bed reactors at Sasol (Dry, 1981), over a wide range of process conditions. Experimental data were obtained at different temperatures (T ) 235, 250, and 265 °C) and synthesis gas feed compositions (H2/ CO ) 0.67, 0.93, or 2.03) in both modes of operation (conventional and supercritical FTS) under steady state conditions. At a given reaction temperature and feed composition, gas space velocity was varied to achieve different levels of syngas conversion. Variations in residence time allow us to distinguish primary and secondary reaction steps that control olefin selectivity. Experimental Section Experimental equipment and procedures have been described previously (Lang et al., 1995). Experiments were conducted in a conventional downflow fixed-bed reactor (1.3 cm inside diameter, 40 cm3 effective bed volume for supercritical FTS, and 1 cm inside diameter, 27 cm3 effective bed volume for conventional FTS) embedded in an aluminum block with a two-zone heater. Carbon monoxide, hydrogen, carbon dioxide, and C5hydrocarbons were analyzed by on-line gas chromatography. Condensed C5+ hydrocarbons, collected for 6-8 h after reaching steady-state at a given set of reaction © 1997 American Chemical Society

Ind. Eng. Chem. Res., Vol. 36, No. 7, 1997 2581

condition, were analyzed using gas chromatography (Bukur et al., 1989; 1990a; Lang et al., 1995). Premixed synthesis gas (Iweco, Inc. >99.7% purity), containing approximately 5% of argon as an internal standard, was used as the feed. Propane (Phillips 66 Co. >99% purity) was pumped from a liquid propane dip tube tank using a diaphragm metering pump (American Lewa, Inc.; Model FCMK-1). The amount of propane produced during FTS could not be determined directly, due to the large amount of excess propane in the feed. Propane production rate was estimated from the measured amount of propene, by assuming that the propene/ propane ratio during SFTS is the same as that during conventional FTS. Olefin selectivities reported here are based on the analysis of gas phase products (C2-C5 hydrocarbons) and liquid products (C6-C15 hydrocarbons) collected in the ice trap at ambient pressure. A precipitated iron catalyst synthesized by Ruhrchemie AG (Oberhausen-Holten, Germany) was used in this test. The nominal catalyst composition is 100 Fe/5 Cu/4.2 K/25 SiO2 (on mass basis), and the preparation procedure is described elsewhere (Frohning et al., 1977; Dry, 1981). Catalyst was calcined in air at 300 °C for 5 h, and then crushed and sieved to 32/60 mesh size (0.48 mm in diameter). About 3.5 g of catalyst was diluted 1:6 by volume with glass beads of the same size prior to loading into the reactor. The catalyst was reduced with hydrogen at 220 °C, ambient pressure and a flow rate of 5100 cm3/min (linear superficial velocity of 150 cm/s) for 1 h. Following reduction, the catalyst was tested initially at baseline process conditions (1.5 MPa, 250 °C, 2 L (NTP)/(g of catalyst)‚h, H2/CO ) 0.67). After 67 h of conventional FTS under the baseline conditions the total pressure was increased to 5.5 MPa using propane as a balance gas (run FA-1724), while keeping the partial pressure and the flow rate of syngas at the baseline conditions. Since the reaction pressure and temperature (5.5 MPa and 250 °C, respectively) are well above the critical pressure and temperature of the propane (4.19 MPa and 96.7 °C, respectively), this is referred to as supercritical Fischer-Tropsch synthesis. Between 70 and 700 h on stream the catalyst was tested under different sets of process conditions. Following any change in process conditions, the reactor was allowed to operate at a given set of process conditions for at least 16 h, before the mass balance was performed (i.e. before the liquid products were collected in steady state traps at ambient and system pressure). On the basis of our previous studies with the Ruhrchemie catalyst (Bukur et al., 1990b) 16 h of operation under a given set of conditions is sufficient to achieve a quasi-steady state. In another test (run FB-1644) the Ruhrchemie catalyst was evaluated under process conditions similar to those used in run FA-1724, but without supercritical propane (conventional FTS). Only selected results and selectivity trends from run FB-1644 test are described here, for comparison with SFTS under similar conversions. Results and Discussion Catalyst Activity. Variations in syngas conversion during the first 140 h of testing are shown in Figure 1. After a 25 h conditioning period (results not shown) and 5 h of testing at the baseline conditions (30 h on stream) the syngas conversion was about 51%, which is in agreement with results from run FA-0844 (conversion of ∼50%) with the same catalyst (Lang et al., 1995). At 30 h on stream, the reactor exit line was clogged with wax product, and gas flow was resumed at 40 h and

Figure 1. Variation in syngas conversion during conventional FTS (P ) 1.5 MPa) and supercritical FTS (P ) 5.5 MPa) in run FA-1724.

stabilized at about 47 h on stream. Following this operational upset, the syngas conversion was lower (4143%) during the next 20 h of testing (period 1). This drop in conversion (catalyst activity) is believed to be due to formation of carbonaceous deposits on the catalyst, resulting in blockage of active sites. Upon introduction of propane (total pressure of 5.5 MPa; syngas partial pressure of 1.5 MPa) the syngas conversion increased, reaching 50% at 140 h (period 2). The latter value is about 8% less than that obtained in run FA-0844 (54% conversion at total pressure of 7 MPa, syngas partial pressure of 1.5 MPa), indicating that exposure of the catalyst to supercritical propane resulted in removal of some of the carbonaceous deposits and partial recovery of the activity. Between 140 and 627 h on stream, the catalyst was tested under different process conditions (variations in gas space velocity, reaction temperature, and H2/CO feed ratio at a total pressure of 5.5 MPa) and selected results are given in Table 1 (periods 3-17). An apparent activation energy of 105 kJ/mol was estimated from values of the apparent reaction rate constant at different reaction temperatures. The apparent rate constant, for a first-order reaction in hydrogen, was calculated from experimental values of conversion, usage ratio (at the reactor exit), and the average hydrogen to carbon ratio of hydrocarbon product, using the model of Zimmerman et al. (1989). The estimated value of the apparent activation energy is consistent with previously reported literature values on iron FTS catalysts and is about 20% higher than the value (86 kJ/mol) obtained from experiments with the same catalyst (after CO reduction) in a stirred tank slurry reactor (Zimmerman and Bukur, 1990). Values of (H2/CO) usage ratio provide information on a relative rate of water gas shift (WGS) reaction (lower usage ratio corresponds to higher WGS activity). As shown in Table 1, the usage ratio decreases with increase in syngas conversion at a constant temperature, and with increase in reaction temperature at constant syngas conversion. Catalyst Deactivation and Its Effect on Olefin Selectivities. Catalyst deactivation was checked by repeating the baseline SFTS conditions at 332-356 h (period 9), and 628-696 h (period 18) on stream. The syngas conversions were approximately 47% (340-356 h) and 35-40% (650-696 h), indicating moderate catalyst deactivation between 70 and 356 h (periods 2-9), and more rapid deactivation between 356 and 696 h (periods 10-18). Catalyst deactivation is attributed to partial poisoning of active sites by sulfur in the propane feed (>99.0% purity, less than 1 ppm of sulfur)

Notes: Conventional FTS during period 1; SFTS during periods 2-18 with Ptot. ) 5.5 MPA. Syngas partial pressure PH2+CO ) 0.7 MPa in periods 5 and 11, otherwise PH2+CO ) 1.5 MP. a

33.8 35.1 0.73 0.63 75.3 69.9 0.79 1.35 26.0 27.7 1.06 0.89 26.3 27.4 1.01 0.90 70.4 64.9 0.78 1.31 77.3 51.7 1.03 5.45 48.3 30.4 0.90 3.08 92.5 60.3 0.98 15.1 34.2 22.1 0.95 2.59 43.8 46.7 0.78 0.58 77.0 76.6 0.66 0.69 29.2 32.5 0.85 0.57 26.8 29.3 0.82 0.61 62.1 65.0 0.74 0.54 74.7 75.5 0.69 0.61 32.0 34.4 0.79 0.61 48.5 50.2 0.73 0.61 48.5 43.1 0.82 0.60 CO conv, % H2 + CO conv, % H2/CO usage ratio H2/CO exit ratio

18 17

624 235 0.5 162 0.93 598 235 2.2 772 0.93

16 15

571 250 4.0 1413 0.93 547 250 1.4 487 0.93

14 13

475 235 1.5 529 2.0

12

428 235 3.0 1047 2.0

11

406 250 2.0 1234 2.0 373 250 5.0 1750 2.0

10 9

352 250 2.0 680 0.67 331 265 1.4 507 0.67

8 7

259 265 7.1 2480 0.67 239 235 2.2 752 0.67

6 5

215 235 0.5 288 0.67 191 250 1.0 330 0.67

4 3

164 250 3.8 1340 0.67 140 250 2.0 680 0.67

2 1

67 250 2.0 175 0.67 time on stream, h T, °C SV, L(NTP)/(g cat)‚h GHSV, h-1 H2/CO feed ratio

period no. Table 1. Process Conditions and Catalyst Activity Results in Test FA-1724a

691 250 2.0 681 0.67

2582 Ind. Eng. Chem. Res., Vol. 36, No. 7, 1997

Figure 2. Effect of gas space velocity on (a) olefin content and (b) 2-olefin content during SFTS at 250 °C, H2/CO ) 0.67, and 5.5 MPa (run FA-1724).

and other factors which are believed to be responsible for deactivation of iron catalysts used for FTS (e.g. formation of inactive carbonaceous deposits, sintering, and/or formation of less active iron oxide phases). At the end of the test the catalyst (Carusorb 200, manufactured by Carus Chemical, containing 4 wt % potassium permanganate on activated alumina support) used for sulfur removal in the propane feed line was found to be ineffective. Data from the latter portion of the test (400-700 h) must be treated with caution due to potential effects of catalyst deactivation on product selectivity. Olefin selectivities (both total and 2-olefin) obtained during period 18 (691 h) were similar to those obtained in period 3 (similar syngas conversions), indicating that the catalyst deactivation and the operating procedures employed did not have a significant effect on olefin selectivities. Catalyst deactivation during conventional FTS (run FB-1644) was also checked by repeating baseline conditions at 295-316 h and at 571-598 h on stream. Under the baseline process conditions (1.5 MPa, 250 °C, H2/ CO ) 0.67, 2 L (NTP)/(g of catalyst)‚h) the syngas conversions were 62% at 76 h on stream, 63% at 310 h, and 55% at 582 h. Olefin selectivities at 310 h on stream were the same (within experimental errors) as those obtained at 76 h. This indicates that operating procedures used in this run did not have an effect on olefin selectivities. Liquid products were not collected during the last period of the run (571-598 h). Olefin SelectivitiesEffect of Gas Residence Time. Effects of gas residence time (i.e. gas hourly space velocity (GHSV) defined as the total volumetric (NTP) feed flow rate of synthesis gas and propane per unit bed volume, ca. 40 cm3) and carbon number on total olefin content ((1-olefin + 2-olefin)/(1-olefin + 2-olefin + nparaffin)) and 2-olefin content (2-olefin/(1-olefin + 2-olefin)) during SFTS are illustrated in parts a and b of Figure 2, respectively. Data shown in Figure 2 were obtained at 250 °C with H2/CO ) 0.67 and GHSV )

Ind. Eng. Chem. Res., Vol. 36, No. 7, 1997 2583

330-1340 h-1 and correspond to syngas conversions of 75-34% (periods 2-4 in Table 1). Results obtained at gas hourly space velocities of 680 and 1340 h-1 were nearly identical, whereas the residence time effect was clearly observed at the gas space velocity of 330 h-1. Ethylene selectivity was significantly higher at the two higher gas space velocities, whereas the increase of total olefin content at higher carbon numbers was much smaller (Figure 2a). Selectivity of 2-olefins decreased (i.e. R-olefin content increased) with an increase in GHSV. The same trends were observed in experiments with different syngas feed compositions (H2/CO ) 0.93 and 2.03). From these observations it is concluded that R-olefins, and to a smaller extent n-paraffins, are the primary products of FTS. This is consistent with results from previous studies with iron FT catalysts (Dry 1981; Schulz and Gokcebay, 1984; Dictor and Bell, 1986). At the present time there is no consensus whether some of n-paraffins and 2-olefins are formed by secondary hydrogenation and isomerization reactions, respectively, on sites where chain growth cannot take place or as primary products following secondary readsorption of R-olefins on FTS sites (Schulz et al., 1982; Madon et al, 1991, 1993). Upon readsorption, R-olefin becomes a reaction intermediate which can either continue to grow and terminate as a longer chain R-olefin, n-paraffin or 2-olefin or be terminated to n-paraffin or a 2-olefin of the same carbon number. Olefin selectivities (total olefin content and 2-olefin content) obtained in our previous study with supercritical propane (Lang et al., 1995) at 250 °C, 7 MPa total pressure (syngas partial pressure of ∼1.5 MPa), H2/CO ) 0.67, and GHSV ) 860 h-1 (syngas space velocity of 2 L (NTP)/(g of catalyst)‚h) were nearly the same as the ones obtained in the present study at GHSV ) 680 h-1 (baseline process conditions). In both cases the nominal residence time at reaction conditions was approximately 2.5 min, and syngas conversions were similar as stated above. Nominal residence time is calculated as the ratio of the effective bed volume (∼40 cm3) and the arithmetic average of inlet and outlet gas volumetric flow rate at reaction conditions. In run FA-0844 the ratio of molar flow rates of propane and syngas was about 4, whereas in run FA1724 it was about 3. These data show that olefin selectivities are independent of propane to syngas ratio (i.e. total system pressure), which is to be expected if the operating conditions are above the critical point of reaction mixture. From the economic point of view, it is advantageous to operate at as low a total pressure as possible, while ensuring operation in the supercritical region. These results also show that the operational upset at the beginning of the test did not have an effect on olefin selectivities during the early periods of SFTS in run FA-1724. Similar olefin selectivities were also obtained in run FA-1075 (total pressure of 5.5 MPa; syngas partial pressure of ∼1.5 MPa) conducted in our laboratory (Akgerman and Bukur, 1995), under the same baseline conditions as employed in run FA-1724. This represents an additional evidence that selectivities in periods 2-4 in run FA-1724 were not affected by the operational upset described above. Shapes of curves in Figure 2 reflect the carbon number (molecular weight) effect on secondary reactions. Ethylene is more reactive than other low molecular weight R-olefins, and thus its selectivity is low. The decrease in olefin content with increase in carbon number has been attributed to their greater adsorptivity (Anderson, 1956; Schulz et al., 1982), higher solubility in the liquid phase resulting in higher R-olefin concen-

trations (Dictor and Bell, 1986; Schulz et al., 1988), and/ or diffusion enhanced R-olefin readsorption (Madon et al, 1991; Iglesia et al., 1991). Madon et al. (1991, 1993) proposed that the decrease in olefin content with increasing molecular weight is due to decreasing diffusivities of longer chain molecules and their longer residence times in the catalyst pores. Due to diffusional restrictions, larger R-olefins spend longer times in a pore, which increases the probability of their readsorption on FTS active sites before exiting the pore. The increase in 2-olefin content with an increase in carbon number (Figure 2b) or with an increase in bed residence time (lower gas space velocity) is due to the same factors mentioned above. The longer residence time of high molecular weight R-olefins either in the catalyst pores or in the reactor itself increases the probability for secondary R-olefin readsorption followed by termination as 2-olefin on FTS and/or different type of sites. The issue of importance of the solubility effect is a controversial one. Madon and Iglesia (1993) argued that higher relative concentrations (solubility) of larger R-olefins in hydrocarbon liquids cannot increase the rate of secondary reactions under vapor-liquid equilibrium conditions (i.e. in the absence of transport restrictions). Recently, Kuipers et al. (1995) proposed a model for decrease in olefin content with increase in carbon number, which accounts for solubility and diffusional limitations corrected for preferential physisorption at the fluid-to-solid interface. They stated that the solubility effect on olefin selectivity is expected to be stronger than the effect of intraparticle diffusion at high reactant conversions. Their model is applicable only when all the products leave the reactor in the gas phase, whereas when the products leave the reactor as a liquid the solubility effect should not be taken into account. In our experiments (both conventional and supercritical mode of operation), the liquid products were leaving the reactor continuously, and thus the solubility effect was not contributing to the decline in total and 1-olefin contents with increasing carbon number. Effects of gas space velocity and carbon number on olefin selectivities during conventional FTS (run FB1644) at 1.5 MPa, 250 °C, H2/CO ) 0.67 are shown in Figure 3. Qualitative trends are the same as those observed during SFTS (Figure 2); i.e., the total olefin content increased (Figure 3a), whereas the 2-olefin content decreased (Figure 3b), with increase in gas space velocity (decrease in bed residence time). However, the bed residence time effect on selectivity was markedly higher during the conventional FTS, although conversions and nominal gas residence times were similar in both sets of experiments (0.9-4.8 min in run FB-1644 vs 1.3-5.1 min in run FA-1724). Carbon number effect on total olefin and 2-olefin selectivity was also more evident in the case of conventional FTS. For example, at GHSV ) 560 h-1 (nominal residence time, θ, of 0.9 min) the total olefin content decreased from 78% (C4 hydrocarbons) to 53% (C15 hydrocarbons), whereas 2-olefin content increased from 3.6% (C4 olefins) to 30% (C15 olefins). The corresponding changes during SFTS at GHSV ) 1340 h-1 (θ ) 1.3 min) and syngas conversion of 34%, were as follows: total olefin content decreased from 77% (C4 hydrocarbons) to 65% (C15 hydrocarbons), and 2-olefin content increased from 1.2% (C4 olefins) to 22% (C15 olefins). At higher syngas conversions, i.e. longer bed residence times, changes in total olefin content and 2-olefin content with carbon number during conventional FTS were significantly larger than those during SFTS, which is illustrated in Figure 4 for syngas conversions of 76% (FA-1724) and

2584 Ind. Eng. Chem. Res., Vol. 36, No. 7, 1997

Figure 5. Comparison of R-olefin content during conventional FTS (run FB-1644) and SFTS (run FA-1724) at 250 °C (H2/CO ) 0.67; syngas conversions 82% (FB-1644) and 76% (FA-1724)).

Figure 3. Effect of gas space velocity on (a) olefin content and (b) 2-olefin content during conventional FTS at 250 °C, H2/CO ) 0.67, and 1.5 MPa (run FB-1644).

Figure 4. Comparison of (a) olefin content; and (b) 2-olefin content during conventional FTS (run FB-1644) and SFTS (run FA-1724) at 250 °C (H2/CO ) 0.67; syngas conversions 82% (FB-1644) and 76% (FA-1724)).

82% (FB-1644). Changes in R-olefin content (1-olefin/ (1-olefin + 2-olefin + n-paraffin)) with carbon number, for both modes of operation at syngas conversion of about 80%, are shown in Figure 5. It can be seen that selectivity of C2 and C7+ R-olefins is significantly higher during supercritical FTS, and this is of potential commercial importance. For all other cases reported in this study, R-olefin content can be calculated from the given

2-olefin and total olefin content values as: R-olefin content (wt %) ) (100% - 2-olefin content) × (olefin content)/100. Results in Figures 2 and 3 show that the gas space velocity had a marked effect on the higher carbon number olefin selectivity during conventional FTS and a relatively small effect during SFTS. This behavior is unexpected, since nominal gas residence times were even shorter during conventional FTS under similar syngas conversions (due to a smaller bed volume in run FB-1644). During conventional FTS the reaction mixture inside the reactor is distributed among two phases: gas and liquid. High molecular weight hydrocarbons (C8+) are leaving the reactor preferentially in the liquid phase, the flow rate of which increases along the reactor length. In a fixed-bed reactor the residence time of the liquid phase is much longer than that of the gas phase. This increases probability for readsorption of high molecular weight R-olefins and leads to increased formation of n-paraffins and 2-olefins via secondary reactions (Schulz et al., 1982, 1988). During SFTS operation, there is only one phase in the reactor, and the residence times of all products, regardless of their molecular weight, are the same. As can be seen from Figures 2-4, selectivities of low molecular weight hydrocarbons in both modes of operation are similar, with the exception of ethylene selectivity. These products are primarily in the gas phase or distributed between the gas phase and the liquid phase (conventional FTS), and their residence times are similar in both modes of operation. Therefore, similar selectivities were obtained in both modes of operation. From results shown in Figures 2-4 (C3 and C4 hydrocarbons), it is concluded that primary R-olefin selectivity (content) is 70-80% and that of n-paraffins is 20%. Also, the 2-olefin content of C4-C6 hydrocarbons is 2-4% at short bed residence times, and this either is the result of double bond shift (isomerization of 1-olefins) or represents the primary selectivity of 2-olefins. As mentioned earlier, several explanations were proposed in the literature for the carbon number effect on olefin selectivity (increased adsorptivity and/or solubility, or decrease in diffusivity with increase in molecular weight of R-olefins). Our results from conventional and supercritical FTS experiments may be explained in terms of the diffusion enhanced R-olefin readsorption model of Iglesia et al. (1991, 1993), and enhanced desorption of adsorbed primary products in a supercritical fluid (Yokota and Fujimoto, 1991; Fan et al., 1992, 1995). Larger R-olefins spend longer time in the catalyst pores than smaller ones, due to their lower diffusivities, and this increases the probability for

Ind. Eng. Chem. Res., Vol. 36, No. 7, 1997 2585

secondary R-olefin readsorption, double bond isomerization and hydrogenation reactions. Ethylene, which has relatively large diffusivity due to its small molecular size, is significantly more reactive than other R-olefins (Schulz et al., 1982; Iglesia et al., 1993). Its Thiele modulus for readsorption reaction can be large and similar to that of higher molecular weight R-olefins. Thus, ethylene has a much greater probability for readsorption, followed by chain growth initiation or termination to ethane, than other low molecular olefins, and its selectivity is low in comparison to C3-C6 olefins. Ethylene selectivity during conventional FTS was smaller than during SFTS at comparable bed residence times, i.e. syngas conversions (Figures 2-4). However, the pore residence time of ethylene is greater during conventional FTS, because ethylene diffusivity is smaller in the liquid-filled pores (conventional FTS), than in the supercritical propane (SFTS). As stated earlier, at a given gas space velocity, the carbon number effect on either the total olefin or 2-olefin content was much more evident during the conventional FTS. This is due to the dominant effect of the intraparticle diffusion on the residence time of R-olefins inside the pores and to slow removal (desorption) of adsorbed R-olefins from the catalyst surface. In both modes of operation, diffusivity decreases with an increase in carbon number (molecular weight), but since diffusivities are significantly smaller in hydrocarbon wax than in the supercritical propane (see below), the intraparticle diffusional resistance during conventional FTS is larger and the carbon number effect on olefin selectivity is stronger. Also, as proposed by Fujimoto and co-workers (Yokota and Fujimoto, 1991; Fan et al., 1992, 1995), the adsorption/ desorption equilibrium of high molecular weight R-olefins is shifted to the right in supercritical fluids due to in situ extraction of these products from the catalyst surface. During conventional FTS, high molecular weight R-olefins are strongly adsorbed on the catalyst surface and may react to form secondary products before leaving the surface. In the above discussion, we have provided a rational explanation for the observed gas space velocity and carbon number effects on olefin selectivity trends during the conventional and supercritical FTS. Here we present results from related studies in our laboratory, which suggest that the intraparticle diffusional restrictions are indeed significant under the experimental conditions employed in the present study. In a study with fused iron catalyst, Zimmerman et al. (1989) showed that intraparticle diffusion restrictions were significant in experiments with 30/60 mesh size catalyst particles. For example, at 250 °C and 1.48 MPa, the experimental value of the catalyst effectiveness factor was 0.36, and the corresponding value of the reaction rate constant (for the first order reaction with respect to hydrogen) was 36 mmol/(g of catalyst)‚h‚MPa. Calculated values of the effectiveness factor, based on molecular diffusivity of hydrogen in hydrocarbon wax of 1.6 × 10-8 (m2/s), were 0.23-0.31, which is in reasonable agreement with the experimental value. In the present study, the reaction rate constant for the rate of synthesis gas consumption during conventional FTS under baseline process conditions (250 °C, 1.5 MPa, 2 L(NTP)/(g of catalyst)‚h, H2/CO ) 0.67) was 111 mmol/(g of catalyst)‚h‚MPa in run FB-1644, and 90 (mmol/(g of catalyst)‚h‚MPa) in run FA-1724. The ruhrchemie catalyst used in the present study is more active than the fused catalyst used in the study by Zimmerman et al., and the intraparticle diffusional restrictions during conventional FTS are expected to be

larger (assuming that there are no significant differences in the porous structure of the two catalysts) in the present study. We expect that intraparticle diffusional effects for reactants were significant (effectiveness factor less than 1) during SFTS (run FA-1724) but to a smaller extent than during conventional FTS. During run FA-0844 (Lang et al., 1995), the FTS was conducted under both conventional and supercritical conditions, and a slight increase in syngas conversion was observed during the SFTS compared to conventional FTS (54% vs 50%) which is consistent with the expected increase in diffusivities of reactants in supercritical propane. Fan et al. (1992, 1995) observed an increase in CO conversion on alumina supported ruthenium catalyst during SFTS in n-hexane, compared to liquid phase FTS (trickle bed operation with co-current flow of syngas and n-hexadecane). This was attributed to higher diffusivities of reactants in supercritical n-hexane relative to liquid n-hexadecane. Higher diffusivities of reactants result in higher reactant concentrations in the pellet, which increases the rate of syngas consumption. This increase in partial pressures coupled with change in H2/CO molar ratio may also have an effect on hydrocarbon selectivities and their functionalities (Iglesia et al., 1993). We did not attempt to estimate a value of the reaction rate constant for the R-olefin readsorption reaction and hence cannot quantify the intraparticle diffusional resistance for this reaction. However, we have determined experimentally molecular diffusivities of C8 hydrocarbons in both Fischer-Tropsch hydrocarbon wax (Rodden, 1988; Erkey et al., 1990) and supercritical propane (Akgerman and Bukur, 1996) utilizing the Taylor dispersion method. Molecular diffusivity of n-octane (C8H18) in hydrocarbon wax at 250 °C and 1.48 MPa (the same conditions as in run FB-1644) was 5.6 × 10-9 m2/s, whereas in supercritical propane at 250 °C and 6.2 MPa (similar to reaction conditions used in run FA-1724) the molecular diffusivity of 1-octene was 8.7 × 10-8 m2/s. These data show that (1) diffusivities of n-octane in wax are smaller than that of hydrogen and (2) diffusivities in supercritical propane are significantly higher (more than 10-fold) than in the liquid hydrocarbon wax (conventional FTS operation). The implications of these results are that (a) intraparticle diffusional restrictions are likely to be significant for high molecular weight R-olefins (gC8) during both conventional FTS and SFTS and (b) carbon number effects on olefin selectivity are expected to be less pronounced during supercritical FTS. The latter was observed in the present study. Olefin SelectivitiessEffect of Reaction Temperature. The effect of reaction temperature on olefin selectivities during SFTS at a nearly constant syngas conversion (∼30%) is shown in Figure 6 (data from periods 3, 6, and 7 in Table 1). Total olefin content of C3+ hydrocarbons (Figure 6a) and 2-olefin content (Figure 6b) are essentially independent of reaction temperature in the range 235-265 °C. Ethylene selectivity was the highest at the lowest reaction temperature. The same trends were observed at a higher syngas conversion (∼70%), and in experiments with syngas molar feed H2/CO ratios of 0.93 and 2.03, as well as during conventional FTS (run FB-1644). Results from previous studies with iron FT catalysts showed three different types of behavior; i.e. in some cases the olefin selectivity increased with increase in temperature, but on some catalysts either no effect or the opposite trends were observed (Schulz et al., 1982; Dictor and Bell, 1986; Donnelly and Satterfield, 1989).

2586 Ind. Eng. Chem. Res., Vol. 36, No. 7, 1997

Figure 6. Effect of reaction temperature on (a) olefin content and (b) 2-olefin content during SFTS at 5.5 MPa and H2/CO ) 0.67 (run FA-1724).

Figure 8. Effect of gas feed composition on (a) olefin content and (b) 2-olefin content during conventional FTS at 1.5 MPa and 250 °C (run FB-1644).

Olefin SelectivitiesEffect of Reactant Composition. Figure 7 illustrates the effect of gas feed composition on olefin selectivity during SFTS at 235 °C and syngas conversion of about 30% (data from periods 6, 11, and 15 in Table 1). Total olefin content was lower (Figure 7a), and 2-olefin content higher (Figure 7b) when the synthesis gas with H2/CO ) 2.03 was used (representative of syngas obtained from steam reforming or partial oxidation of natural gas). Olefin selectivities were similar in experiments with H2/CO ) 0.67 (period 6) and H2/CO ) 0.93 (period 15). Results obtained at 250 °C, 1.5 MPa and syngas conversions of ∼30% during conventional FTS (run FB1644) indicate that total olefin decreased and 2-olefin content increased, as the H2/CO ratio in the feed increased (Figure 8). Concentration of surface hydrogen determines chain termination probabilities and olefin content, and it increases with increase in H2/CO molar feed ratio. It has been suggested that high surface concentrations of hydrogen favor termination reactions (i.e. increased selectivity of low molecular weight products) and termination to paraffins rather than olefins, as well as secondary R-olefin isomerization reactions (Dictor and Bell, 1986; Madon and Iglesia, 1993). Figure 7. Effect of gas feed composition on (a) olefin content and (b) 2-olefin content during SFTS at 5.5 MPa and 235 °C (run FA1724).

The rates of both primary and secondary reactions increase with increasing temperature, and the effect of temperature on olefin selectivities is determined by values of activation energies of the primary and secondary reactions (higher temperature favors a reaction with higher activation energy). From our results it appears that activation energies of primary and secondary reactions are similar.

Summary Effects of process conditions (reaction temperature, gas space velocity, and feed composition) on olefin selectivity were studied in a fixed-bed reactor during conventional FTS (reaction pressure of 1.5 MPa) and FTS in supercritical propane (total pressure of 5.5 MPa). It was found that total olefin content decreased and 2-olefin content increased with either decrease in gas space velocity or increase in the H2/CO molar feed ratio, whereas olefin selectivities were essentially independent of reaction temperature.

Ind. Eng. Chem. Res., Vol. 36, No. 7, 1997 2587

Results from bed residence time effect studies in both modes of operation indicate that R-olefins are the dominant primary products of FTS. Normal paraffins are primary products also, and their fractional content on the Ruhrchemie catalyst was approximately 20%. Selectivity of n-paraffins and 2-olefins increases, whereas R-olefin selectivity decreases with increase in carbon number, due to secondary reactions of R-olefins. Bed residence time and carbon number (molecular weight) effects were more pronounced during conventional FTS than during supercritical FTS. During conventional FTS high molecular weight products (C8+ hydrocarbons) leave the reactor preferentially in the liquid state, and their residence time is longer than that of either the gas phase products or the products formed during supercritical FTS. Diffusivities of high molecular weight R-olefins in the liquid hydrocarbon wax are significantly lower than the corresponding diffusivities in the supercritical propane, and hence, the pore residence time of these products is longer during conventional FTS. A longer residence time in the reactor and/or catalyst pores increases probability for secondary readsorption of R-olefins, and results in secondary formation of n-paraffins and 2-olefins. Also, desorption rates of heavy R-olefins are higher in supercritical propane than in liquid-filled pores during conventional FTS, which results in higher selectivity of primary products. At high syngas conversions (∼80%), selectivities of high molecular weight R-olefins during SFTS were significantly higher than those obtained during conventional operation. These results indicate that SFTS is a potentially attractive route for synthesis of high molecular weight R-olefins from the synthesis gas. Acknowledgment This work was supported by the U.S. Department of Energy (University Coal Research Program) under Grant DE-FG22-92PC92545 and the Texas Engineering Experiment Station. Literature Cited Akgerman, A; Bukur, D. B. Fischer - Tropsch Synthesis in Supercritical Fluids. Quarterly Technical Progress Report, April 1-June 30, 1995, for DOE Grant DE-FG22-92PC92545; Texas Engineering Experiment Station: College Station, TX. Akgerman, A; Bukur, D. B. Fischer - Tropsch Synthesis in Supercritical Fluids. Quarterly Technical Progress Report, April 1-June 30, 1996, for DOE Grant DE-FG22-92PC92545; Texas Engineering Experiment Station: College Station, TX. Anderson, R. B. Catalysts for the Fischer-Tropsch synthesis. In Catalysis; Emmett, P. H., Ed.; Van Nostrand-Reinhold: New York, 1956; Vol. IV, pp 29-255. Bukur, D. B.; Lang, X.; Rossin, J. A.; Zimmerman, W. H.; Rosynek, M. P.; Yeh, E. B.; Li, C. Activation studies with a promoted precipitated iron Fischer-Tropsch catalyst. Ind. Eng. Chem. Res. 1989, 28, 1130-1140. Bukur, D. B.; Mukesh, D.; Patel, S. A. Promoter effects on precipitated iron catalysts for Fischer-Tropsch synthesis. Ind. Eng. Chem. Res. 1990a, 29, 194-204. Bukur, D. B.; Lang, X.; Mukesh, D.; Zimmerman, W. H.; Rosynek, M. P.; Li, C. Binder/support effects on the activity and selectivity of iron catalyst in the Fischer-Tropsch synthesis. Ind. Eng. Chem. Res. 1990b, 29, 1588-1599. Dictor, R.; Bell, A. T. Fischer-Tropsch synthesis over reduced and unreduced iron oxide catalysts. J. Catal. 1986, 97, 121-136.

Donnelly, T. J.; Satterfield, C. N. Product distributions of the Fischer-Tropsch synthesis on precipitated iron catalysts. Appl. Catal. 1989, 52, 93-114. Dry, M. E. The Fischer-Tropsch synthesis. In Catalysis-Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer-Verlag: New York, 1981; Vol. 1, pp 160-255. Erkey, C.; Rodden, J. B.; Akgerman, A. Diffusivities of Synthesis Gas and n-Alkanes in Fischer-Tropsch Wax. Energy Fuels 1990, 4, 275-281. Fan, L.; Yokota, K.; Fujimoto, K. Supercritical phase FischerTropsch synthesis: catalyst pore size effect. AIChE J. 1992, 38, 1639-1648. Fan, L.; Yokota, K.; Fujimoto, K. Characterization of mass transfer in supercritical-phase Fischer-Tropsch synthesis reaction. Top. Catal. 1995, 2, 267-283. Frohning, C. D.; Rottig, W.; Schnur, F. The fixed bed synthesis. In Chemicals from Coal; Falbe, J., Ed.; Thieme: Stuttgart, Germany, 1977; pp 234-257. Iglesia, E.; Reyes, S. C.; Madon, R. J.; Soled, S. L. Selectivity control and catalyst design in the Fischer-Tropsch synthesis: Sites, pellets and reactors. J. Catal. 1991, 129, 238-256. Iglesia, E.; Reyes, S. C.; Madon, R. J. Transport-enhanced R-olefin readsorption pathways in Ru-catalyzed hydrocarbon synthesis. Adv. Catal. 1993, 39, 221-302. Kuipers, E. W.; Vinkenburg, I. H.; Oosterbeek, H. Chain length dependence of R-olefin readsorption in Fischer-Tropsch synthesis. J. Catal. 1995, 152, 137-146. Lang, X.; Akgerman, A.; Bukur, D. B. Steady-state FischerTropsch synthesis in supercritical propane. Ind. Eng. Chem. Res. 1995, 34, 72-77. Lappin, G. R.; Nemec, L. H.; Sauer, J. D.; Wagner, J. D. Higher olefins. In Kirk-Othmer Encyclopedia of Chemical Technology, 4th ed.; Wiley: New York, 1996; Vol. 17, pp 839-858. Madon, R. J.; Iglesia, E. The importance of olefin readsorption and H2/CO reactant ratio for hydrocarbon chain growth on ruthenium catalysts, J. Catal. 1993, 139, 576-590. Madon, R. J.; Reyes, S. C.; Iglesia, E. Primary and secondary reaction pathways in ruthenium-catalyzed hydrocarbon synthesis. J. Phys. Chem. 1991, 95, 7795-7804. Rodden, J. B. Diffusion coefficients for several dilute solutes in n-eicosane, n-octacosane, and Fischer - Tropsch wax at 373533 K. Ph.D. Thesis, Texas A&M University, College Station, TX, 1988. Schulz, H.; Rosch, S.; Gokcebay, H. Selectivity of the FischerTropsch CO-hydrogenation. In Coal: Phoenix of ‘80s, Proceedings of the 64th CIC Coal Symposium; Al Taweel, A. M., Ed.; Canadian Society for Chemical Engineering: Ottawa, Canada, Vol. 2, 1982; pp 486-493. Schulz, H.; Gokcebay, H. Fischer-Tropsch CO-hydrogenation as a means for linear olefins production. In Catalysis of organic reactions; Kosak, J. R., Ed.; Marcel Dekker: New York, 1984; pp 153-169. Schulz, H.; Beck, K.; Erich, E. Mechanism of the Fischer - Tropsch process. Stud. Surf. Sci. Catal. 1988, 36, 457-471. Waddacor, M. Sasol’s R300m R-olefins plant marks new milestone. Chem. Process. SA 1994, 1 (No. 10), 2-5. Yokota, K.; Fujimoto, K. Supercritical phase Fischer-Tropsch synthesis reaction. 2. The effective diffusion of reactant and products in the supercritical phase reaction. Ind. Eng. Chem. Res. 1991, 30, 95-100. Zimmerman, W. H.; Bukur, D. B. Reaction kinetics over iron catalyst used for Fischer-Tropsch synthesis. Can. J. Chem. Eng. 1990, 68, 292-301. Zimmerman, W. H.; Rossin, J. A.; Bukur, D. B. Effect of particle size on the activity of a fused iron Fischer-Tropsch catalyst. Ind. Eng. Chem. Res. 1989, 28, 406-413.

Received for review August 16, 1996 Revised manuscript received April 7, 1997 Accepted April 12, 1997X IE960507B X Abstract published in Advance ACS Abstracts, June 1, 1997.