- Email: [email protected]
The addition of HZSM-5 to the Fischer–Tropsch process for improved gasoline production
Applied Catalysis A: General 267 (2004) 217–225
The addition of HZSM-5 to the Fischer–Tropsch process for improved gasoline production F.G. Botes a,∗ , W. Böhringer b b
a Sasol Technology, R&D Division, P.O. Box 1, Sasolburg 1947, South Africa Catalysis Research Unit, Department of Chemical Engineering, University of Cape Town, Rondebosch, 7701, South Africa
Received in revised form 5 March 2004; accepted 5 March 2004 Available online 22 April 2004
Abstract The combination of an alkali-promoted iron-based Fischer–Tropsch catalyst and an acidic co-catalyst (HZSM-5) for syngas conversion to hydrocarbons was studied in a Berty microreactor. It was found that a physical mixture of the two catalysts resulted in severe alkali migration from the iron catalyst to the zeolite, so that this mode of the bifunctional process does not seem viable. However, by separating the catalytic layers by means of a wire mesh inside the microreactor, it was confirmed that the addition of HZSM-5 to the Fischer–Tropsch process improved both the selectivity and the quality of the gasoline product fraction. The deactivation behaviour of the acidic co-catalyst is also reported. © 2004 Elsevier B.V. All rights reserved. Keywords: Fischer–Tropsch; Iron catalyst; HZSM-5; Bifunctional process; Gasoline (high octane); Aromatics
1. Introduction Presently, there is a lot of world-wide interest in the Fischer–Tropsch (FT) process as a method to produce synthetic liquid fuels from natural gas or coal [1]. Most view the following two directions as the most viable for the further commercialisation of the FT synthesis [2]: • The production of heavy wax with an FT process, followed by hydrocracking to mainly produce fuel in the middle distillate range (see also [1,3]). • The conversion of the FT product spectrum to high octane gasoline by using a zeolitic co-catalyst (see also [4]). This paper will focus on the latter option, namely the conversion of syngas to high quality gasoline. The Fischer–Tropsch process has two important disadvantages with regard to the production of gasoline. Firstly, the carbon number distribution of the product spectrum follows the statistical Schulz–Flory function. This means that the gasoline selectivity (C5 –C11 fraction) is limited to a theoretical maximum value of about 48% [4,5]. Secondly, the Fischer–Tropsch process produces mainly linear olefins ∗
Corresponding author. Tel.: +27-16-960-2914; fax: 27-11-522-3306. E-mail address: [email protected] (F.G. Botes).
0926-860X/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2004.03.006
and paraffins [5]. Because of the low octane value of these compounds, the gasoline requires extensive work-up to improve the octane number. The above two limitations of the Fischer–Tropsch process can at least in part be addressed by the addition of an acidic co-catalyst to the system [4,6]. The cracking of longer chain hydrocarbons and the oligomerisation of light olefins would increase the yield of gasoline range products. Skeletal isomerisation and aromatisation are also facile reactions over acidic zeolites. The branched and aromatics hydrocarbons formed by these reactions have a high octane value. Consequently, the use of a zeolite in combination with an FT catalyst may increase both the selectivity and the quality (octane value) of the gasoline. Various types and combinations of FT catalysts and zeolites have been studied. The range of FT catalysts include those based on iron [6–9], iron/manganese [10,11], cobalt [12,13], cobalt/manganese [14,15] and even iron/cobalt [16]. The use of different zeolites (mordenite, erionite, ZSM-11, ZSM-12, L, omega and beta) in combination with a Fischer–Tropsch catalyst has been investigated [17], while other researchers tested HZSM-5 [6,16], gallium-substituted HZSM-5 [14,15] and HY [10]. The two catalytic functions have also been combined in a variety of ways, ranging from a single reactor containing both catalytic functions (taking advantage of a possible synergistic effect between the two catalysts) to a dual reactor arrangement with the two cat-
218
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
alytic functions in subsequent reactors (so that the operating parameters of the reactors can be optimised individually). More detailed reviews of the literature are presented elsewhere [4,18].
2. Choice of catalysts for current study The aim of the study was to investigate the viability of a single-step bifunctional process, i.e. the two catalytic functions were to be contained in the same reactor. In order to obtain the desired hydrocarbon reactions over zeolite catalysts, reaction temperatures above 300 ◦ C are generally required [11,19]. Since cobalt and ruthenium as Fischer–Tropsch catalysts have a very high methane selectivity at such high temperatures, they are not expected to be effective in the bifunctional process [6]. Methane formation at high temperature is less excessive for iron-based FT [1,5]. Furthermore, a highly olefinic product spectrum can be obtained with an iron-based catalyst. These qualities make iron-based FT catalysts ideal for application in a single-step bifunctional process. However, iron catalysts need alkali promotion to attain the desired activity and selectivity [1]. Migration of the alkali and subsequent poisoning of the acid sites on the zeolite is thus a great concern for the bifunctional process [20]. Because of the following reasons, HZSM-5 seems to be the preferred choice of acidic co-catalyst for the bifunctional process [4,21]: its shape-selective properties limits the formation of hydrocarbons boiling above the gasoline range; its highly acidic nature gives it a high activity for the acid catalysed reactions (oligomerisation, cracking, isomerisation, aromatisation); its medium pore size makes it highly resistant to coking; it is stable under hydrothermal conditions. Consequently, the combination of an alkali-promoted iron catalyst and an acidic HZSM-5 co-catalyst was used for the current study.
of reactor can essentially be viewed as a short packed bed with an extremely high recycle ratio, and thus behaves approximately like a continuous stirred tank reactor (CSTR). The flow rates of the four feed streams were controlled by Brooks mass flow controllers. These feed streams included a commercial syngas stream (which contained a substantial amount of methane), as well as pure hydrogen, carbon dioxide and argon fed from bottles. The oil and reaction water products were condensed in a knock-out pot, while all uncondensed effluent passed through a back-pressure regulator into the vent system. 3.3. Catalyst loading A baseline FT run was performed with the iron catalyst (no HZSM-5 added). Two configurations of the bifunctional process were tested, namely the physical admixture of the two catalysts and a type of dual bed arrangement. Only the “high acidity” HZSM-5 was tested in the former configuration, while both zeolites were tested in the latter. For the dual bed arrangement, contact between the catalytic functions was avoided by a wire mesh between the catalyst layers. It should be noted that this configuration is distinctly different from a dual layer arrangement in a fixed bed reactor where the syngas is first converted over the FT catalyst, after which the hydrocarbons are isomerised and aromatised by an acid catalyst. Since a Berty reactor approximates CSTR behaviour, the two catalysts are exposed to more or less the same gas composition. 3.4. Catalyst activation and synthesis The synthesis conditions were selected based on published data for a fused iron catalyst operated in a fluidised bed reactor for which reasonable methane selectivities and ethylene/ethane ratios were reported [5]. The details of reduction and synthesis are presented in Table 1.
3. Experimental 3.1. Catalysts A fused iron catalyst with a low alkali level was obtained from Sasol. The catalyst was made according to the published preparation method for the fused iron catalyst used in Sasol’s High-temperature Fischer–Tropsch synthesis [5]. The two samples of ZSM-5 powder, obtained from Zeolyst International, will be referred to as the “high acidity” zeolite (silica/alumina molar ratio of 30) and the “low acidity” zeolite (silica/alumina molar ratio of 280). These were received in the ammonium form (NH4 ZSM-5) and were therefore calcined at 500 ◦ C for 16 h to convert them to the acidic form (HZSM-5). 3.2. Reactor system The synthesis experiments were performed in a Berty microreactor, containing a very thin layer of catalyst. This type
3.5. Gas sampling and product analysis Feed and product samples were taken in glass ampoules for later analysis according to the method described elsewhere [22]. The samples were analysed on two gas chromatographs (GCs). The Gowmac Series 600 GC–TCD, with two columns in series (a Porapaq Q column of 1.5 m followed by a 13× molsieve column of 2 m) and a thermal conductivity detector (TCD), was used to quantify the amounts of the permanent gases (hydrogen, carbon monoxide, carbon dioxide and methane) relative to the internal standard (argon). The hydrocarbon product spectrum was analysed with a Perkin Elmer XL GC–FID fitted with a Petrocol DH column (150 m) and a flame ionisation detector (FID). Since methane was analysed by both GCs, the methane analysis of the GC–TCD could be used to quantify the amounts of all the hydrocarbon components identified by the GC–FID.
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
219
Table 1 Reduction and synthesis conditions for all experimental runs
4. Results and discussion
Catalyst loading Iron catalyst HZSM-5 powder
4.1. Bifunctional process—physical admixture of catalysts 5 g (unreduced mass) 5g
Reduction conditions Reduction gas Gas space velocity Temperature Pressure Time
Hydrogen 200 ml/min/g unreduced iron catalyst 420 ◦ C 20 bar 16 h
Total feed gas composition during synthesis (vol.%) H2 58.0 CO 12.5 CO2 12.0 CH4 5.5 Ar (internal standard) 12.0 Synthesis conditions Syngas space velocity Temperature Pressure
211 ml/min/g unreduced iron catalyst 330 ◦ C 20 bar
3.6. Product selectivities The syngas conversion, as well as the overall yield of FT product, was calculated from the analysis of the permanent gases. Even though carbon monoxide can be converted to carbon dioxide via the reversible water-gas-shift reaction, this is not considered part of the FT (hydrocarbon synthesis) reaction. The product selectivities were therefore expressed in terms of the hydrocarbon product spectrum only and reported on the basis of carbon atom percentage. All samples were submitted for the quick GC–TCD analysis; consequently, the conversion and methane selectivity could be calculated for every sample. However, due to the time consuming nature of the GC–FID analysis, a comprehensive hydrocarbon product distribution was only obtained for selected samples.
The methane selectivity of the bifunctional process run in the physical admixed configuration is compared to that of the baseline FT run in Fig. 1. The scatter in the data is attributed to the large amount of methane contained in the commercial syngas stream fed to the microreactor. In fact, the flow rate of methane into the reactor was normally around four times the amount that was produced by the FT reaction. Errors (for example caused by GC inaccuracies) in the methane inlet and outlet flow rates will be vastly amplified when these two large values are subtracted from each other in order to obtain the much lesser value of the methane production rate. Disregarding the data scatter, it is evident that the bifunctional process not only had a substantially higher methane selectivity than the FT run, but the increase over the first 200 h of synthesis also seems to be more pronounced. HZSM-5 cannot readily form methane from hydrocarbons. It thus appears as though alkali migration from the iron to the HZSM-5 during reduction and synthesis of the bifunctional process run caused a severe shift in the selectivity of the FT catalyst towards methane. In Fig. 2, the carbon monoxide conversion of the bifunctional process run in the physical admixed arrangement is compared to that of the baseline FT run. The conversions of the two experimental runs were very similar at the onset of synthesis. However, for the case of the bifunctional process run, there was a much more significant decline in the FT reaction rate over time-on-line than for the baseline FT run. From data presented elsewhere for Sasol’s high-temperature FT process [5], it is clear that a decrease in the alkali content of the iron catalyst not only affects the product selectivities, but also causes a slight decrease in conversion. This is precisely what has been observed for the bifunctional process run, namely an extensive shift in the product selectivity and
50 45
Baseline FT run
Methane selectivity [carbon atom %]
40 35
Physical mixture
30 25
Dual layers: "Low acidity" HZSM-5
20 15
Dual layers: "High acidity" HZSM-5
10 5 0 0
50
100
150
200
250
300
350
400
450
Time on line [hours] Fig. 1. Methane selectivities of baseline FT run and all bifunctional process runs.
220
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225 100
CO Conversion [%]
90
Baseline FT run
80
Physical mixture
70
Dual layers: "Low acidity" HZSM-5 60
Dual layers: "High acidity" HZSM-5
50
40 0
100
200
300
400
500
Time on line [hours] Fig. 2. CO conversions of baseline FT run and all bifunctional process runs.
a less substantial decrease in syngas conversion. Therefore, the decline in the FT reaction rate for the bifunctional process run is also ascribed to the lowering of the alkali content of the iron catalyst due to alkali migration from the FT catalyst to the zeolite. It was attempted to confirm the alleged alkali migration from the FT catalyst to the HZSM-5 by analysing the spent catalyst mixture. Since it was not possible to quantitatively separate the intimate mixture of very fine particles, chemical analysis of the spent zeolite (free from any iron) proved impossible. Therefore, the spent mixture was submitted for SEM-EDX analysis with the hope that alkali-contaminated HZSM-5 particles could be identified. However, it was not even possible to identify discreet zeolite particles under the SEM. Furthermore, those areas that yielded high intensity EDX peaks for silicon and aluminium also showed the presence of iron. Possibly the break-up of FT catalyst particles due to carbon formation [5] resulted in a fine dispersion of iron over the spent catalyst mixture; it could thus not be
proven conclusively by means of analysis that alkali was indeed present on the zeolite. Considering Fig. 3, it is evident that the bifunctional process in the physical admixed configuration has a low selectivity towards the C5 –C11 fraction (desirable product). This is mainly ascribed to the progressive shift in the product spectrum of the FT catalyst towards lighter and more paraffinic compounds. Because of the excessive production of light paraffins, it does not seem as if the bifunctional process would be commercially viable if there is an intimate contact between the catalytic functions. For this reason, no further experiments were performed with the catalysts in physical admixture. 4.2. Bifunctional process—catalysts in separate layers 4.2.1. Time-on-line behaviour of FT catalyst function The methane selectivities of the bifunctional process runs in dual layers are compared to that of the baseline FT run
Gasoline (C5 to C11) selectivity [carbon atom %]
60
50
Baseline FT run 40
Physical mixture 30
Dual layers: "Low acidity" HZSM-5
20
Dual layers: "High acidity" HZSM-5
10
0 0
50
100
150
200
250
300
350
400
450
Time on line [hours] Fig. 3. Gasoline selectivities of baseline FT run and all bifunctional process runs.
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
in Fig. 1. Within the experimental scatter of the data, no differences are visible between the various runs. Since the methane selectivity of the FT synthesis is very sensitive to the alkali level of the iron catalyst [5], it seems as if the wire mesh between the two layers was effective in avoiding significant alkali migration from the iron catalyst to the zeolite. In Fig. 2, the bifunctional process runs in dual layers are compared to the baseline FT run with respect to the carbon monoxide conversion over time-on-line. For each of the involved runs, the conversion was fairly constant during the period of synthesis. This means that the iron catalyst had a very stable activity during the course of each run. However, it seems as if the addition of HZSM-5 to the process resulted in a lowering of the FT reaction rate. Furthermore, it may be argued that the extent of the decrease in FT activity is dependent on the number of acid sites present in the reactor, since the decrease in conversion was more severe for the “high acidity” HZSM-5 than for the “low acidity” HZSM-5. It is highly unlikely that the FT catalyst was poisoned by some volatile compound (e.g. ammonia) contained in HZSM-5. Any compound in the zeolite that survived the 16 h of calcination at 500 ◦ C would surely have been stable at the lower reduction and synthesis temperatures as well. If the iron catalyst was affected by a poison contained in the feed gas, one would expect a continuous decrease in conversion over the synthesis period rather than the observed stable activity. The effect of HZSM-5 addition on the FT activity is clearly not a result of alkali migration either, since a loss of alkali from the iron catalyst would have reflected more notably in the methane selectivity than in the activity. Consideration was also given to the possibility that the HZSM-5 may have affected the FT catalyst activity by changing the concentrations of components in the gas phase. If the zeolites lowered the concentration of a reactive component by means of adsorption, one would expect a temporary effect on the conversion, since the small amount of zeolite would only have a limited capacity for such a substance. If a reactive compound or inhibitor is consumed or produced by a chemical reaction catalysed by the HZSM-5, one would expect a continuous change in the FT activity over time as the zeolite deactivated. A more feasible explanation is that the wire mesh, which was installed to separate the two catalysts, altered the hydrodynamics inside the Berty reactor. This may have affected the gas flow through the iron catalyst layer. The hypothesis can be tested by performing a baseline FT run with the second wire mesh in place in order to simulate the hydrodynamics of the bifunctional process experiments. The hydrodynamic explanation seems to be supported by the results of the bifunctional process run with the two catalysts in physical admixture. For this run, the second wire mesh was not installed. Even though the “high acidity” HZSM-5 was present in the reactor, the carbon monoxide conversion was initially comparable to that of the baseline FT run. The continuous decrease in the conversion during the course of the run was ascribed to the extensive alkali migration from the
221
iron catalyst to the zeolite. This drop in conversion noted for the physical admixed configuration is thus apparently unrelated to the lower FT reaction rate observed for the case of the dual layer arrangement. In view of the foregoing discussion, it does not seem as if the presence of an acidic zeolite had a direct effect on the syngas conversion rate in the bifunctional process, but this should be verified. 4.2.2. Deactivation of the acid catalyst function Over HZSM-5, additional propane is mainly produced as the hydrogen-rich byproduct of the aromatisation of olefins, but can also form via the cracking of longer chain aliphatics. Although additional propylene can be produced by the zeolite through cracking, propylene from the FT reaction is mostly consumed by the acid catalyst during the aromatisation and oligomerisation reactions. A high propane and low propylene selectivity is therefore indicative of a highly active zeolite. The propane and propylene selectivities are presented in Fig. 4a and b, respectively. From the data presented for the baseline FT run, it is clear that the C3 fraction produced by the iron catalyst was highly olefinic. The product of the bifunctional process generally contained much more propane and much less propylene than the standard FT product. At the beginning of the run, the “high acidity” HZSM-5 clearly increased the propane selectivity enormously, while the propylene was consumed almost completely. This is indicative of a very high initial activity for this zeolite. However, there was a dramatic change in the selectivities of the C3 components with synthesis time, indicating a rapid decrease in the initial high activity. From the graphs, it is also clear that the “low acidity” HZSM-5 had a substantially lower initial activity, but a much more stable performance over the course of the run. The linearity of the C6 paraffin fraction is presented in Fig. 5 in order to indicate the variations in the isomerisation activity of the acidic co-catalysts. The graph clearly shows that the product of the baseline FT run is highly linear, as would be expected [5]. On the other hand, the product from the bifunctional process is generally much more branched due to the isomerisation ability of the acid catalyst function. For the case of the “high acidity” HZSM-5, the C6 paraffin fraction initially contained mainly branched molecules, but the linearity increased steeply with time-on-line as a result of the rapid deactivation of this type of zeolite. The “low acidity” HZSM-5 had a lower initial isomerisation activity, but the deactivation was much less severe. The observations concerning the zeolite activity are confirmed if the aromatic content of the C8 carbon number fraction is considered (Fig. 6). Initially, the “high acidity” HZSM-5 had a high selectivity towards ethylbenzene and the xylenes, but this decreased rapidly as the zeolite deactivated. The “low acidity” zeolite had a lower but more constant C8 aromatic selectivity. In fact, considering Figs. 4–6, it seems as if the deactivation of the “high acidity” HZSM-5 was so severe that its activity dropped below that of the “low acidity” HZSM-5 after about 150 h on line. It also appears
222
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225 16
Baseline FT run "Low acidity" HZSM-5
Propane selectivity [carbon atom %]
14
"High acidity" HZSM-5
12 10 8 6 4 2 0 0
50
100
(a)
150
200
250
300
350
400
450
Time on line [hours] 16
Propylene selectivity [carbon atom %]
14 12 10 8 6 4
Baseline FT run "Low acidity" HZSM-5
2
"High acidity" HZSM-5
0 0
50
100
(b)
150
200
250
300
350
400
450
Time on line [hours]
Fig. 4. (a) Propane selectivities of baseline FT run and dual layer bifunctional process runs. (b) Propylene selectivities of baseline FT run and dual layer bifunctional process runs.
80
Amount of linear molecules in C6 paraffin fraction [%]
70 60 50 40 30 20
Baseline FT run "Low acidity" HZSM-5
10
"High acidity" HZSM-5
0 0
50
100
150
200
250
300
350
400
Time on line [hours] Fig. 5. Linearity of C6 paraffins—baseline FT run and dual layer bifunctional process runs.
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225 100
Baseline FT run "Low acidity" HZSM-5 "High acidity" HZSM-5
90
Aromatic content of C8 carbon number fraction [%]
223
80 70 60 50 40 30 20 10 0 0
50
100
150
200
250
300
350
400
Time on line [hours] Fig. 6. Aromatic content of C8 carbon number fraction—baseline FT run and dual layer bifunctional process runs.
as if the aromatisation ability of the “high acidity” zeolite was more notably affected by the deactivation than its isomerisation ability. The reason for this is that aromatisation is the slowest of the acid catalysed reactions and is therefore very sensitive to the activity of the catalyst.
the product spectrum contains similar amounts of olefins and paraffins, while the condensable fraction contains both naphthenes and aromatics. The overall carbon number distributions obtained for the two bifunctional process runs at the beginning of synthesis are compared to the FT carbon number distribution in Fig. 9. The notable lowering in the selectivities of the C2 and C3 carbon number fractions indicates that both ethylene and propylene were converted over the acidic co-catalyst. However, due to the higher stability of secondary carbenium ions compared to primary carbenium ions, propylene has a higher reactivity than ethylene; hence the more substantial lowering in the selectivity of the C3 fraction than in the C2 fraction. The overall carbon number distribution of the bifunctional process product is mainly distinguished from the Schulz–Flory distribution of the traditional FT product by the occurrence of two clear humps. The reason for the first hump at around C4 is that the longer aliphatic molecules are not stable in the presence of an active acid catalyst and are
4.2.3. The effect of HZSM-5 addition on the FT product spectrum Typical product distributions of the bifunctional process runs are presented in Figs. 7 and 8 for each of the two HZSM-5 zeolites at the beginning of synthesis. It is evident that a highly active “high acidity” zeolite produces a condensable product fraction consisting mainly of aromatics (Fig. 8). Since olefins are consumed and paraffins produced during the formation of aromatics, the light end of the product spectrum is almost completely saturated. For the “low acidity” HZSM-5 with its lower initial activity, the conversion of olefins to a mixture of aromatics and light paraffins is not as extensive (Fig. 7). Consequently, the light end of
Selectivity [carbon atom %] .
16
Paraffins Olefins Naphthenes Aromatics
14 12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
9
10
11
12
Carbon number Fig. 7. Product distribution of dual layer bifunctional process run with “low acidity” HZSM-5 as co-catalyst.
224
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
Selectivity [carbon atom%] .
16
Paraffins Olefins Naphthenes Aromatics
14 12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
9
10
11
12
Carbon number Fig. 8. Product distribution of dual layer bifunctional process run with “high acidity” HZSM-5 as co-catalyst. 20
Baseline FT run "Low acidity" HZSM-5 "High acidity" HZSM-5
Selectivity [carbon atom %]
18 16 14 12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Carbon number Fig. 9. Overall product carbon number distributions—baseline FT run and dual layer bifunctional process runs.
readily cracked down to lighter olefins and paraffins. The C4 aliphatics seems to be the preferred products of these cracking reactions due to the high stability of the tertiary carbenium ion. The second hump at about C8 corresponds to the favoured distribution of aromatics. There is also a fairly sharp cut-off in the product spectrum at around C10 –C11 . The reason for this is that long chain aliphatics are not stable at these temperatures in the presence of an acid catalyst, while aromatics higher than about C11 are too large to be formed readily inside the pores of HZSM-5. For the baseline FT run, the gasoline selectivity,1 decreased steadily over the course of the experiment as there was a slight shift in the product spectrum towards the lighter end (see Fig. 3). The decline in gasoline selectivity towards the end of the bifunctional process runs in the dual layer configuration can thus not solely be ascribed to the deactiva-
1
The gasoline fraction is defined as the C5 –C11 carbon number range.
tion of the acid catalyst function, since the amount of compounds that are reactive over a zeolite would have decreased as the selectivity of the FT catalyst shifted towards lighter and more paraffinic hydrocarbons. Fig. 3 also indicates that the addition of the “low acidity” HZSM-5 to the FT process resulted in a substantial improvement in the selectivity of the C5 –C11 fraction throughout the course of the experiment, probably mainly due to the oligomerisation and cyclisation of light olefins. To the contrary, the “high acidity” HZSM-5 did not significantly improve the gasoline selectivity of the FT process during the early stages of the run. The proposed reason is the vast amount of aromatics produced by such a highly active zeolite, which is accompanied by the formation of light paraffins that fall outside the petrol range. As this zeolite deactivated, oligomerisation and cyclisation presumably became the predominant reactions, resulting in an increase in the gasoline selectivity. However, towards the end of the run it seemed as if the gasoline selectivity for the case of the “high acidity” HZSM-5 dropped more sharply
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
than for the other zeolite, possibly because of its very rapid deactivation rate. It is also evident that the gasoline selectivity of the bifunctional process is in general substantially higher than that of the traditional FT process. For the case of the “low acidity” HZSM-5, the increase in C5 –C11 selectivity over the baseline FT run was between 25 and 35% throughout the run.
225
tion negatively impacted on the gasoline selectivity during the latter stages of the experiment. It was thus found that, in order to optimise the gasoline production, a zeolite with a lower aluminium content should rather be employed. It was also seen that the liquid fraction from the bifunctional process was rich in aromatics, naphthenes and branched aliphatics. This product will therefore have a much higher octane number than the predominantly linear aliphatic compounds produced by the traditional FT process.
5. Conclusions It was found that the physical mixture of an alkalipromoted iron catalyst and HZSM-5 resulted in extensive alkali migration from the FT catalyst to the zeolite. This caused a severe shift in the selectivity of the FT catalyst towards light paraffins that cannot be converted to higher value products by the zeolite. The selectivity of these low value paraffins was so high that the commercial viability of a bifunctional process in this mode is doubtful. However, by separating the two catalyst layers with a wire mesh in the Berty reactor, it was shown that carbon monoxide hydrogenation with subsequent work-up of the hydrocarbons can be performed in the same reactor with great success provided the migration of alkali can be avoided. In a fluidised bed reactor, alkali transfer between separate iron catalyst and zeolite particles is also expected to occur during collisions between the particles. The extent of this migration will be dependent on the overall time of contact between the particles and is a topic for further experimental investigation. As expected, the deactivation rate of the zeolitic co-catalyst was related to its aluminium content, with the “high acidity” HZSM-5 deactivating more rapidly than the “low acidity” HZSM-5. Due to the extensive deactivation, the product spectrum of the bifunctional process with the “high acidity” HZSM-5 changed dramatically over the first week of synthesis. In fact, the deactivation was so severe that the performance of the “high acidity” HZSM-5 dropped below that of the “low acidity” HZSM-5 after about 150 h of synthesis. The more stable activity throughout the run and the enhanced performance during the latter stages of the run probably makes the “low acidity” HZSM-5 a more preferred choice as the acid catalyst function of the bifunctional process. The experimental results have shown that the use of a “highly acidic” zeolite in the bifunctional process does not significantly improve the gasoline selectivity of the FT process. At the beginning of the run, the extensive production of aromatics was accompanied by the formation of light paraffins that fall outside the petrol range, while rapid deactiva-
References [1] H. Schulz, Appl. Catal. A 186 (1999) 3. [2] L. Xu, S. Bao, R. O’Brien, A. Raje, B.H. Davis, Chemtech 28 (1) (1998) 47. [3] R.L. Espinoza, A.P. Steynberg, B. Jager, A.C. Vosloo, Appl. Catal. A 186 (1999) 13. [4] V. Udaya, S. Rao, R.J. Gormley, Catal. Today 6 (3) (1990) 207. [5] M.E. Dry, in: J.R. Anderson, M. Boudart (Eds.), Catalysis Science and Technology, vol. I, Springer, Berlin, 1981, Chapter 4. [6] H. Schulz, H.L. Niederberger, M. Kneip, F. Weil, Stud. Surf. Sci. Catal. 61 (1991) 313. [7] P.D. Caesar, J.A. Brennan, W.E. Garwood, J. Ciric, J. Catal. 56 (1979) 274. [8] C.D. Chang, W.H. Lang, A.J. Silvestri, US Patent 4 086 262 (1978), to Mobil Oil Corporation. [9] A. Shamsi, V.U.S. Rao, R.J. Gormley, R.T. Obermeyer, R.R. Schehl, J.M. Stencel, Appl. Catal. 27 (1986) 55. [10] N.O. Egiebor, W.C. Cooper, B.W. Wojciechowski, Appl. Catal. 55 (1989) 47. [11] K. Müller, H.D. Deckwer, M. Ralek, Stud. Surf. Sci. Catal. 12 (1982) 267. [12] A. Shamsi, V. Udaya, S. Rao, R.J. Gormley, R.T. Obermyer, R. Schehl, J.M. Stencel, Ind. Eng. Chem. Prod. Res. Dev. 23 (1984) 513. [13] J.M. Stencel, V.U.S. Rao, J.R. Diehl, K.H. Rhee, A.G. Dhere, R.J. DeAngelis, J. Catal. 84 (1983) 109. [14] G. Bäurle, K. Guse, M. Lohrengel, H. Papp, Stud. Surf. Sci. Catal. 75 (1993) 2789. [15] M. Baerns, N. Guan, E. Körting, U. Lindner, M. Lohrengel, H. Papp, Int. J. Energy Res. 18 (1994) 197. [16] V.U.S. Rao, R.J. Gormley, Hydrocarbon Process. 59 (11) (1980) 139. [17] H. Hammer, M. Joisten, S Lüngen, D Winkler, Int. J. Energy Res. 18 (1994) 223. [18] F.G. Botes, The addition of HZSM-5 to the Fischer–Tropsch process for improved gasoline production, M.Sc. Thesis (Engineering), University of Cape Town, 2002. [19] N. Guan, Y. Liu, M. Zhang, Catal. Today 30 (1996) 207. [20] S.A. Butter, A.W. Chester, A.B. Schwartz, US Patent 4 298 695 (1981), to Mobil Oil Corporation. [21] K. Jothimurugesan, S.K. Gangwal, Ind. Eng. Chem. Res. 37 (1998) 1181. [22] H.L. Niederberger, Zeitliche Änderung der Aktivität und der Selektivität bei der Umsetzung CO-reicher Synthesegase an einem Eisen/H-ZSM-5-Kombinationskatalysator, Dr.-Ing. Disseration, University of Karlsruhe, 1988.
The addition of HZSM-5 to the Fischer–Tropsch process for improved gasoline production F.G. Botes a,∗ , W. Böhringer b b
a Sasol Technology, R&D Division, P.O. Box 1, Sasolburg 1947, South Africa Catalysis Research Unit, Department of Chemical Engineering, University of Cape Town, Rondebosch, 7701, South Africa
Received in revised form 5 March 2004; accepted 5 March 2004 Available online 22 April 2004
Abstract The combination of an alkali-promoted iron-based Fischer–Tropsch catalyst and an acidic co-catalyst (HZSM-5) for syngas conversion to hydrocarbons was studied in a Berty microreactor. It was found that a physical mixture of the two catalysts resulted in severe alkali migration from the iron catalyst to the zeolite, so that this mode of the bifunctional process does not seem viable. However, by separating the catalytic layers by means of a wire mesh inside the microreactor, it was confirmed that the addition of HZSM-5 to the Fischer–Tropsch process improved both the selectivity and the quality of the gasoline product fraction. The deactivation behaviour of the acidic co-catalyst is also reported. © 2004 Elsevier B.V. All rights reserved. Keywords: Fischer–Tropsch; Iron catalyst; HZSM-5; Bifunctional process; Gasoline (high octane); Aromatics
1. Introduction Presently, there is a lot of world-wide interest in the Fischer–Tropsch (FT) process as a method to produce synthetic liquid fuels from natural gas or coal [1]. Most view the following two directions as the most viable for the further commercialisation of the FT synthesis [2]: • The production of heavy wax with an FT process, followed by hydrocracking to mainly produce fuel in the middle distillate range (see also [1,3]). • The conversion of the FT product spectrum to high octane gasoline by using a zeolitic co-catalyst (see also [4]). This paper will focus on the latter option, namely the conversion of syngas to high quality gasoline. The Fischer–Tropsch process has two important disadvantages with regard to the production of gasoline. Firstly, the carbon number distribution of the product spectrum follows the statistical Schulz–Flory function. This means that the gasoline selectivity (C5 –C11 fraction) is limited to a theoretical maximum value of about 48% [4,5]. Secondly, the Fischer–Tropsch process produces mainly linear olefins ∗
Corresponding author. Tel.: +27-16-960-2914; fax: 27-11-522-3306. E-mail address: [email protected] (F.G. Botes).
0926-860X/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2004.03.006
and paraffins [5]. Because of the low octane value of these compounds, the gasoline requires extensive work-up to improve the octane number. The above two limitations of the Fischer–Tropsch process can at least in part be addressed by the addition of an acidic co-catalyst to the system [4,6]. The cracking of longer chain hydrocarbons and the oligomerisation of light olefins would increase the yield of gasoline range products. Skeletal isomerisation and aromatisation are also facile reactions over acidic zeolites. The branched and aromatics hydrocarbons formed by these reactions have a high octane value. Consequently, the use of a zeolite in combination with an FT catalyst may increase both the selectivity and the quality (octane value) of the gasoline. Various types and combinations of FT catalysts and zeolites have been studied. The range of FT catalysts include those based on iron [6–9], iron/manganese [10,11], cobalt [12,13], cobalt/manganese [14,15] and even iron/cobalt [16]. The use of different zeolites (mordenite, erionite, ZSM-11, ZSM-12, L, omega and beta) in combination with a Fischer–Tropsch catalyst has been investigated [17], while other researchers tested HZSM-5 [6,16], gallium-substituted HZSM-5 [14,15] and HY [10]. The two catalytic functions have also been combined in a variety of ways, ranging from a single reactor containing both catalytic functions (taking advantage of a possible synergistic effect between the two catalysts) to a dual reactor arrangement with the two cat-
218
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
alytic functions in subsequent reactors (so that the operating parameters of the reactors can be optimised individually). More detailed reviews of the literature are presented elsewhere [4,18].
2. Choice of catalysts for current study The aim of the study was to investigate the viability of a single-step bifunctional process, i.e. the two catalytic functions were to be contained in the same reactor. In order to obtain the desired hydrocarbon reactions over zeolite catalysts, reaction temperatures above 300 ◦ C are generally required [11,19]. Since cobalt and ruthenium as Fischer–Tropsch catalysts have a very high methane selectivity at such high temperatures, they are not expected to be effective in the bifunctional process [6]. Methane formation at high temperature is less excessive for iron-based FT [1,5]. Furthermore, a highly olefinic product spectrum can be obtained with an iron-based catalyst. These qualities make iron-based FT catalysts ideal for application in a single-step bifunctional process. However, iron catalysts need alkali promotion to attain the desired activity and selectivity [1]. Migration of the alkali and subsequent poisoning of the acid sites on the zeolite is thus a great concern for the bifunctional process [20]. Because of the following reasons, HZSM-5 seems to be the preferred choice of acidic co-catalyst for the bifunctional process [4,21]: its shape-selective properties limits the formation of hydrocarbons boiling above the gasoline range; its highly acidic nature gives it a high activity for the acid catalysed reactions (oligomerisation, cracking, isomerisation, aromatisation); its medium pore size makes it highly resistant to coking; it is stable under hydrothermal conditions. Consequently, the combination of an alkali-promoted iron catalyst and an acidic HZSM-5 co-catalyst was used for the current study.
of reactor can essentially be viewed as a short packed bed with an extremely high recycle ratio, and thus behaves approximately like a continuous stirred tank reactor (CSTR). The flow rates of the four feed streams were controlled by Brooks mass flow controllers. These feed streams included a commercial syngas stream (which contained a substantial amount of methane), as well as pure hydrogen, carbon dioxide and argon fed from bottles. The oil and reaction water products were condensed in a knock-out pot, while all uncondensed effluent passed through a back-pressure regulator into the vent system. 3.3. Catalyst loading A baseline FT run was performed with the iron catalyst (no HZSM-5 added). Two configurations of the bifunctional process were tested, namely the physical admixture of the two catalysts and a type of dual bed arrangement. Only the “high acidity” HZSM-5 was tested in the former configuration, while both zeolites were tested in the latter. For the dual bed arrangement, contact between the catalytic functions was avoided by a wire mesh between the catalyst layers. It should be noted that this configuration is distinctly different from a dual layer arrangement in a fixed bed reactor where the syngas is first converted over the FT catalyst, after which the hydrocarbons are isomerised and aromatised by an acid catalyst. Since a Berty reactor approximates CSTR behaviour, the two catalysts are exposed to more or less the same gas composition. 3.4. Catalyst activation and synthesis The synthesis conditions were selected based on published data for a fused iron catalyst operated in a fluidised bed reactor for which reasonable methane selectivities and ethylene/ethane ratios were reported [5]. The details of reduction and synthesis are presented in Table 1.
3. Experimental 3.1. Catalysts A fused iron catalyst with a low alkali level was obtained from Sasol. The catalyst was made according to the published preparation method for the fused iron catalyst used in Sasol’s High-temperature Fischer–Tropsch synthesis [5]. The two samples of ZSM-5 powder, obtained from Zeolyst International, will be referred to as the “high acidity” zeolite (silica/alumina molar ratio of 30) and the “low acidity” zeolite (silica/alumina molar ratio of 280). These were received in the ammonium form (NH4 ZSM-5) and were therefore calcined at 500 ◦ C for 16 h to convert them to the acidic form (HZSM-5). 3.2. Reactor system The synthesis experiments were performed in a Berty microreactor, containing a very thin layer of catalyst. This type
3.5. Gas sampling and product analysis Feed and product samples were taken in glass ampoules for later analysis according to the method described elsewhere [22]. The samples were analysed on two gas chromatographs (GCs). The Gowmac Series 600 GC–TCD, with two columns in series (a Porapaq Q column of 1.5 m followed by a 13× molsieve column of 2 m) and a thermal conductivity detector (TCD), was used to quantify the amounts of the permanent gases (hydrogen, carbon monoxide, carbon dioxide and methane) relative to the internal standard (argon). The hydrocarbon product spectrum was analysed with a Perkin Elmer XL GC–FID fitted with a Petrocol DH column (150 m) and a flame ionisation detector (FID). Since methane was analysed by both GCs, the methane analysis of the GC–TCD could be used to quantify the amounts of all the hydrocarbon components identified by the GC–FID.
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
219
Table 1 Reduction and synthesis conditions for all experimental runs
4. Results and discussion
Catalyst loading Iron catalyst HZSM-5 powder
4.1. Bifunctional process—physical admixture of catalysts 5 g (unreduced mass) 5g
Reduction conditions Reduction gas Gas space velocity Temperature Pressure Time
Hydrogen 200 ml/min/g unreduced iron catalyst 420 ◦ C 20 bar 16 h
Total feed gas composition during synthesis (vol.%) H2 58.0 CO 12.5 CO2 12.0 CH4 5.5 Ar (internal standard) 12.0 Synthesis conditions Syngas space velocity Temperature Pressure
211 ml/min/g unreduced iron catalyst 330 ◦ C 20 bar
3.6. Product selectivities The syngas conversion, as well as the overall yield of FT product, was calculated from the analysis of the permanent gases. Even though carbon monoxide can be converted to carbon dioxide via the reversible water-gas-shift reaction, this is not considered part of the FT (hydrocarbon synthesis) reaction. The product selectivities were therefore expressed in terms of the hydrocarbon product spectrum only and reported on the basis of carbon atom percentage. All samples were submitted for the quick GC–TCD analysis; consequently, the conversion and methane selectivity could be calculated for every sample. However, due to the time consuming nature of the GC–FID analysis, a comprehensive hydrocarbon product distribution was only obtained for selected samples.
The methane selectivity of the bifunctional process run in the physical admixed configuration is compared to that of the baseline FT run in Fig. 1. The scatter in the data is attributed to the large amount of methane contained in the commercial syngas stream fed to the microreactor. In fact, the flow rate of methane into the reactor was normally around four times the amount that was produced by the FT reaction. Errors (for example caused by GC inaccuracies) in the methane inlet and outlet flow rates will be vastly amplified when these two large values are subtracted from each other in order to obtain the much lesser value of the methane production rate. Disregarding the data scatter, it is evident that the bifunctional process not only had a substantially higher methane selectivity than the FT run, but the increase over the first 200 h of synthesis also seems to be more pronounced. HZSM-5 cannot readily form methane from hydrocarbons. It thus appears as though alkali migration from the iron to the HZSM-5 during reduction and synthesis of the bifunctional process run caused a severe shift in the selectivity of the FT catalyst towards methane. In Fig. 2, the carbon monoxide conversion of the bifunctional process run in the physical admixed arrangement is compared to that of the baseline FT run. The conversions of the two experimental runs were very similar at the onset of synthesis. However, for the case of the bifunctional process run, there was a much more significant decline in the FT reaction rate over time-on-line than for the baseline FT run. From data presented elsewhere for Sasol’s high-temperature FT process [5], it is clear that a decrease in the alkali content of the iron catalyst not only affects the product selectivities, but also causes a slight decrease in conversion. This is precisely what has been observed for the bifunctional process run, namely an extensive shift in the product selectivity and
50 45
Baseline FT run
Methane selectivity [carbon atom %]
40 35
Physical mixture
30 25
Dual layers: "Low acidity" HZSM-5
20 15
Dual layers: "High acidity" HZSM-5
10 5 0 0
50
100
150
200
250
300
350
400
450
Time on line [hours] Fig. 1. Methane selectivities of baseline FT run and all bifunctional process runs.
220
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225 100
CO Conversion [%]
90
Baseline FT run
80
Physical mixture
70
Dual layers: "Low acidity" HZSM-5 60
Dual layers: "High acidity" HZSM-5
50
40 0
100
200
300
400
500
Time on line [hours] Fig. 2. CO conversions of baseline FT run and all bifunctional process runs.
a less substantial decrease in syngas conversion. Therefore, the decline in the FT reaction rate for the bifunctional process run is also ascribed to the lowering of the alkali content of the iron catalyst due to alkali migration from the FT catalyst to the zeolite. It was attempted to confirm the alleged alkali migration from the FT catalyst to the HZSM-5 by analysing the spent catalyst mixture. Since it was not possible to quantitatively separate the intimate mixture of very fine particles, chemical analysis of the spent zeolite (free from any iron) proved impossible. Therefore, the spent mixture was submitted for SEM-EDX analysis with the hope that alkali-contaminated HZSM-5 particles could be identified. However, it was not even possible to identify discreet zeolite particles under the SEM. Furthermore, those areas that yielded high intensity EDX peaks for silicon and aluminium also showed the presence of iron. Possibly the break-up of FT catalyst particles due to carbon formation [5] resulted in a fine dispersion of iron over the spent catalyst mixture; it could thus not be
proven conclusively by means of analysis that alkali was indeed present on the zeolite. Considering Fig. 3, it is evident that the bifunctional process in the physical admixed configuration has a low selectivity towards the C5 –C11 fraction (desirable product). This is mainly ascribed to the progressive shift in the product spectrum of the FT catalyst towards lighter and more paraffinic compounds. Because of the excessive production of light paraffins, it does not seem as if the bifunctional process would be commercially viable if there is an intimate contact between the catalytic functions. For this reason, no further experiments were performed with the catalysts in physical admixture. 4.2. Bifunctional process—catalysts in separate layers 4.2.1. Time-on-line behaviour of FT catalyst function The methane selectivities of the bifunctional process runs in dual layers are compared to that of the baseline FT run
Gasoline (C5 to C11) selectivity [carbon atom %]
60
50
Baseline FT run 40
Physical mixture 30
Dual layers: "Low acidity" HZSM-5
20
Dual layers: "High acidity" HZSM-5
10
0 0
50
100
150
200
250
300
350
400
450
Time on line [hours] Fig. 3. Gasoline selectivities of baseline FT run and all bifunctional process runs.
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
in Fig. 1. Within the experimental scatter of the data, no differences are visible between the various runs. Since the methane selectivity of the FT synthesis is very sensitive to the alkali level of the iron catalyst [5], it seems as if the wire mesh between the two layers was effective in avoiding significant alkali migration from the iron catalyst to the zeolite. In Fig. 2, the bifunctional process runs in dual layers are compared to the baseline FT run with respect to the carbon monoxide conversion over time-on-line. For each of the involved runs, the conversion was fairly constant during the period of synthesis. This means that the iron catalyst had a very stable activity during the course of each run. However, it seems as if the addition of HZSM-5 to the process resulted in a lowering of the FT reaction rate. Furthermore, it may be argued that the extent of the decrease in FT activity is dependent on the number of acid sites present in the reactor, since the decrease in conversion was more severe for the “high acidity” HZSM-5 than for the “low acidity” HZSM-5. It is highly unlikely that the FT catalyst was poisoned by some volatile compound (e.g. ammonia) contained in HZSM-5. Any compound in the zeolite that survived the 16 h of calcination at 500 ◦ C would surely have been stable at the lower reduction and synthesis temperatures as well. If the iron catalyst was affected by a poison contained in the feed gas, one would expect a continuous decrease in conversion over the synthesis period rather than the observed stable activity. The effect of HZSM-5 addition on the FT activity is clearly not a result of alkali migration either, since a loss of alkali from the iron catalyst would have reflected more notably in the methane selectivity than in the activity. Consideration was also given to the possibility that the HZSM-5 may have affected the FT catalyst activity by changing the concentrations of components in the gas phase. If the zeolites lowered the concentration of a reactive component by means of adsorption, one would expect a temporary effect on the conversion, since the small amount of zeolite would only have a limited capacity for such a substance. If a reactive compound or inhibitor is consumed or produced by a chemical reaction catalysed by the HZSM-5, one would expect a continuous change in the FT activity over time as the zeolite deactivated. A more feasible explanation is that the wire mesh, which was installed to separate the two catalysts, altered the hydrodynamics inside the Berty reactor. This may have affected the gas flow through the iron catalyst layer. The hypothesis can be tested by performing a baseline FT run with the second wire mesh in place in order to simulate the hydrodynamics of the bifunctional process experiments. The hydrodynamic explanation seems to be supported by the results of the bifunctional process run with the two catalysts in physical admixture. For this run, the second wire mesh was not installed. Even though the “high acidity” HZSM-5 was present in the reactor, the carbon monoxide conversion was initially comparable to that of the baseline FT run. The continuous decrease in the conversion during the course of the run was ascribed to the extensive alkali migration from the
221
iron catalyst to the zeolite. This drop in conversion noted for the physical admixed configuration is thus apparently unrelated to the lower FT reaction rate observed for the case of the dual layer arrangement. In view of the foregoing discussion, it does not seem as if the presence of an acidic zeolite had a direct effect on the syngas conversion rate in the bifunctional process, but this should be verified. 4.2.2. Deactivation of the acid catalyst function Over HZSM-5, additional propane is mainly produced as the hydrogen-rich byproduct of the aromatisation of olefins, but can also form via the cracking of longer chain aliphatics. Although additional propylene can be produced by the zeolite through cracking, propylene from the FT reaction is mostly consumed by the acid catalyst during the aromatisation and oligomerisation reactions. A high propane and low propylene selectivity is therefore indicative of a highly active zeolite. The propane and propylene selectivities are presented in Fig. 4a and b, respectively. From the data presented for the baseline FT run, it is clear that the C3 fraction produced by the iron catalyst was highly olefinic. The product of the bifunctional process generally contained much more propane and much less propylene than the standard FT product. At the beginning of the run, the “high acidity” HZSM-5 clearly increased the propane selectivity enormously, while the propylene was consumed almost completely. This is indicative of a very high initial activity for this zeolite. However, there was a dramatic change in the selectivities of the C3 components with synthesis time, indicating a rapid decrease in the initial high activity. From the graphs, it is also clear that the “low acidity” HZSM-5 had a substantially lower initial activity, but a much more stable performance over the course of the run. The linearity of the C6 paraffin fraction is presented in Fig. 5 in order to indicate the variations in the isomerisation activity of the acidic co-catalysts. The graph clearly shows that the product of the baseline FT run is highly linear, as would be expected [5]. On the other hand, the product from the bifunctional process is generally much more branched due to the isomerisation ability of the acid catalyst function. For the case of the “high acidity” HZSM-5, the C6 paraffin fraction initially contained mainly branched molecules, but the linearity increased steeply with time-on-line as a result of the rapid deactivation of this type of zeolite. The “low acidity” HZSM-5 had a lower initial isomerisation activity, but the deactivation was much less severe. The observations concerning the zeolite activity are confirmed if the aromatic content of the C8 carbon number fraction is considered (Fig. 6). Initially, the “high acidity” HZSM-5 had a high selectivity towards ethylbenzene and the xylenes, but this decreased rapidly as the zeolite deactivated. The “low acidity” zeolite had a lower but more constant C8 aromatic selectivity. In fact, considering Figs. 4–6, it seems as if the deactivation of the “high acidity” HZSM-5 was so severe that its activity dropped below that of the “low acidity” HZSM-5 after about 150 h on line. It also appears
222
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225 16
Baseline FT run "Low acidity" HZSM-5
Propane selectivity [carbon atom %]
14
"High acidity" HZSM-5
12 10 8 6 4 2 0 0
50
100
(a)
150
200
250
300
350
400
450
Time on line [hours] 16
Propylene selectivity [carbon atom %]
14 12 10 8 6 4
Baseline FT run "Low acidity" HZSM-5
2
"High acidity" HZSM-5
0 0
50
100
(b)
150
200
250
300
350
400
450
Time on line [hours]
Fig. 4. (a) Propane selectivities of baseline FT run and dual layer bifunctional process runs. (b) Propylene selectivities of baseline FT run and dual layer bifunctional process runs.
80
Amount of linear molecules in C6 paraffin fraction [%]
70 60 50 40 30 20
Baseline FT run "Low acidity" HZSM-5
10
"High acidity" HZSM-5
0 0
50
100
150
200
250
300
350
400
Time on line [hours] Fig. 5. Linearity of C6 paraffins—baseline FT run and dual layer bifunctional process runs.
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225 100
Baseline FT run "Low acidity" HZSM-5 "High acidity" HZSM-5
90
Aromatic content of C8 carbon number fraction [%]
223
80 70 60 50 40 30 20 10 0 0
50
100
150
200
250
300
350
400
Time on line [hours] Fig. 6. Aromatic content of C8 carbon number fraction—baseline FT run and dual layer bifunctional process runs.
as if the aromatisation ability of the “high acidity” zeolite was more notably affected by the deactivation than its isomerisation ability. The reason for this is that aromatisation is the slowest of the acid catalysed reactions and is therefore very sensitive to the activity of the catalyst.
the product spectrum contains similar amounts of olefins and paraffins, while the condensable fraction contains both naphthenes and aromatics. The overall carbon number distributions obtained for the two bifunctional process runs at the beginning of synthesis are compared to the FT carbon number distribution in Fig. 9. The notable lowering in the selectivities of the C2 and C3 carbon number fractions indicates that both ethylene and propylene were converted over the acidic co-catalyst. However, due to the higher stability of secondary carbenium ions compared to primary carbenium ions, propylene has a higher reactivity than ethylene; hence the more substantial lowering in the selectivity of the C3 fraction than in the C2 fraction. The overall carbon number distribution of the bifunctional process product is mainly distinguished from the Schulz–Flory distribution of the traditional FT product by the occurrence of two clear humps. The reason for the first hump at around C4 is that the longer aliphatic molecules are not stable in the presence of an active acid catalyst and are
4.2.3. The effect of HZSM-5 addition on the FT product spectrum Typical product distributions of the bifunctional process runs are presented in Figs. 7 and 8 for each of the two HZSM-5 zeolites at the beginning of synthesis. It is evident that a highly active “high acidity” zeolite produces a condensable product fraction consisting mainly of aromatics (Fig. 8). Since olefins are consumed and paraffins produced during the formation of aromatics, the light end of the product spectrum is almost completely saturated. For the “low acidity” HZSM-5 with its lower initial activity, the conversion of olefins to a mixture of aromatics and light paraffins is not as extensive (Fig. 7). Consequently, the light end of
Selectivity [carbon atom %] .
16
Paraffins Olefins Naphthenes Aromatics
14 12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
9
10
11
12
Carbon number Fig. 7. Product distribution of dual layer bifunctional process run with “low acidity” HZSM-5 as co-catalyst.
224
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
Selectivity [carbon atom%] .
16
Paraffins Olefins Naphthenes Aromatics
14 12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
9
10
11
12
Carbon number Fig. 8. Product distribution of dual layer bifunctional process run with “high acidity” HZSM-5 as co-catalyst. 20
Baseline FT run "Low acidity" HZSM-5 "High acidity" HZSM-5
Selectivity [carbon atom %]
18 16 14 12 10 8 6 4 2 0 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Carbon number Fig. 9. Overall product carbon number distributions—baseline FT run and dual layer bifunctional process runs.
readily cracked down to lighter olefins and paraffins. The C4 aliphatics seems to be the preferred products of these cracking reactions due to the high stability of the tertiary carbenium ion. The second hump at about C8 corresponds to the favoured distribution of aromatics. There is also a fairly sharp cut-off in the product spectrum at around C10 –C11 . The reason for this is that long chain aliphatics are not stable at these temperatures in the presence of an acid catalyst, while aromatics higher than about C11 are too large to be formed readily inside the pores of HZSM-5. For the baseline FT run, the gasoline selectivity,1 decreased steadily over the course of the experiment as there was a slight shift in the product spectrum towards the lighter end (see Fig. 3). The decline in gasoline selectivity towards the end of the bifunctional process runs in the dual layer configuration can thus not solely be ascribed to the deactiva-
1
The gasoline fraction is defined as the C5 –C11 carbon number range.
tion of the acid catalyst function, since the amount of compounds that are reactive over a zeolite would have decreased as the selectivity of the FT catalyst shifted towards lighter and more paraffinic hydrocarbons. Fig. 3 also indicates that the addition of the “low acidity” HZSM-5 to the FT process resulted in a substantial improvement in the selectivity of the C5 –C11 fraction throughout the course of the experiment, probably mainly due to the oligomerisation and cyclisation of light olefins. To the contrary, the “high acidity” HZSM-5 did not significantly improve the gasoline selectivity of the FT process during the early stages of the run. The proposed reason is the vast amount of aromatics produced by such a highly active zeolite, which is accompanied by the formation of light paraffins that fall outside the petrol range. As this zeolite deactivated, oligomerisation and cyclisation presumably became the predominant reactions, resulting in an increase in the gasoline selectivity. However, towards the end of the run it seemed as if the gasoline selectivity for the case of the “high acidity” HZSM-5 dropped more sharply
F.G. Botes, W. Böhringer / Applied Catalysis A: General 267 (2004) 217–225
than for the other zeolite, possibly because of its very rapid deactivation rate. It is also evident that the gasoline selectivity of the bifunctional process is in general substantially higher than that of the traditional FT process. For the case of the “low acidity” HZSM-5, the increase in C5 –C11 selectivity over the baseline FT run was between 25 and 35% throughout the run.
225
tion negatively impacted on the gasoline selectivity during the latter stages of the experiment. It was thus found that, in order to optimise the gasoline production, a zeolite with a lower aluminium content should rather be employed. It was also seen that the liquid fraction from the bifunctional process was rich in aromatics, naphthenes and branched aliphatics. This product will therefore have a much higher octane number than the predominantly linear aliphatic compounds produced by the traditional FT process.
5. Conclusions It was found that the physical mixture of an alkalipromoted iron catalyst and HZSM-5 resulted in extensive alkali migration from the FT catalyst to the zeolite. This caused a severe shift in the selectivity of the FT catalyst towards light paraffins that cannot be converted to higher value products by the zeolite. The selectivity of these low value paraffins was so high that the commercial viability of a bifunctional process in this mode is doubtful. However, by separating the two catalyst layers with a wire mesh in the Berty reactor, it was shown that carbon monoxide hydrogenation with subsequent work-up of the hydrocarbons can be performed in the same reactor with great success provided the migration of alkali can be avoided. In a fluidised bed reactor, alkali transfer between separate iron catalyst and zeolite particles is also expected to occur during collisions between the particles. The extent of this migration will be dependent on the overall time of contact between the particles and is a topic for further experimental investigation. As expected, the deactivation rate of the zeolitic co-catalyst was related to its aluminium content, with the “high acidity” HZSM-5 deactivating more rapidly than the “low acidity” HZSM-5. Due to the extensive deactivation, the product spectrum of the bifunctional process with the “high acidity” HZSM-5 changed dramatically over the first week of synthesis. In fact, the deactivation was so severe that the performance of the “high acidity” HZSM-5 dropped below that of the “low acidity” HZSM-5 after about 150 h of synthesis. The more stable activity throughout the run and the enhanced performance during the latter stages of the run probably makes the “low acidity” HZSM-5 a more preferred choice as the acid catalyst function of the bifunctional process. The experimental results have shown that the use of a “highly acidic” zeolite in the bifunctional process does not significantly improve the gasoline selectivity of the FT process. At the beginning of the run, the extensive production of aromatics was accompanied by the formation of light paraffins that fall outside the petrol range, while rapid deactiva-
References [1] H. Schulz, Appl. Catal. A 186 (1999) 3. [2] L. Xu, S. Bao, R. O’Brien, A. Raje, B.H. Davis, Chemtech 28 (1) (1998) 47. [3] R.L. Espinoza, A.P. Steynberg, B. Jager, A.C. Vosloo, Appl. Catal. A 186 (1999) 13. [4] V. Udaya, S. Rao, R.J. Gormley, Catal. Today 6 (3) (1990) 207. [5] M.E. Dry, in: J.R. Anderson, M. Boudart (Eds.), Catalysis Science and Technology, vol. I, Springer, Berlin, 1981, Chapter 4. [6] H. Schulz, H.L. Niederberger, M. Kneip, F. Weil, Stud. Surf. Sci. Catal. 61 (1991) 313. [7] P.D. Caesar, J.A. Brennan, W.E. Garwood, J. Ciric, J. Catal. 56 (1979) 274. [8] C.D. Chang, W.H. Lang, A.J. Silvestri, US Patent 4 086 262 (1978), to Mobil Oil Corporation. [9] A. Shamsi, V.U.S. Rao, R.J. Gormley, R.T. Obermeyer, R.R. Schehl, J.M. Stencel, Appl. Catal. 27 (1986) 55. [10] N.O. Egiebor, W.C. Cooper, B.W. Wojciechowski, Appl. Catal. 55 (1989) 47. [11] K. Müller, H.D. Deckwer, M. Ralek, Stud. Surf. Sci. Catal. 12 (1982) 267. [12] A. Shamsi, V. Udaya, S. Rao, R.J. Gormley, R.T. Obermyer, R. Schehl, J.M. Stencel, Ind. Eng. Chem. Prod. Res. Dev. 23 (1984) 513. [13] J.M. Stencel, V.U.S. Rao, J.R. Diehl, K.H. Rhee, A.G. Dhere, R.J. DeAngelis, J. Catal. 84 (1983) 109. [14] G. Bäurle, K. Guse, M. Lohrengel, H. Papp, Stud. Surf. Sci. Catal. 75 (1993) 2789. [15] M. Baerns, N. Guan, E. Körting, U. Lindner, M. Lohrengel, H. Papp, Int. J. Energy Res. 18 (1994) 197. [16] V.U.S. Rao, R.J. Gormley, Hydrocarbon Process. 59 (11) (1980) 139. [17] H. Hammer, M. Joisten, S Lüngen, D Winkler, Int. J. Energy Res. 18 (1994) 223. [18] F.G. Botes, The addition of HZSM-5 to the Fischer–Tropsch process for improved gasoline production, M.Sc. Thesis (Engineering), University of Cape Town, 2002. [19] N. Guan, Y. Liu, M. Zhang, Catal. Today 30 (1996) 207. [20] S.A. Butter, A.W. Chester, A.B. Schwartz, US Patent 4 298 695 (1981), to Mobil Oil Corporation. [21] K. Jothimurugesan, S.K. Gangwal, Ind. Eng. Chem. Res. 37 (1998) 1181. [22] H.L. Niederberger, Zeitliche Änderung der Aktivität und der Selektivität bei der Umsetzung CO-reicher Synthesegase an einem Eisen/H-ZSM-5-Kombinationskatalysator, Dr.-Ing. Disseration, University of Karlsruhe, 1988.