- Email: [email protected]
γ-Al2O3 catalysts in Fischer-Tropsch synthesis
Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
Contents lists available at SciVerse ScienceDirect
Journal of the Taiwan Institute of Chemical Engineers journal homepage: www.elsevier.com/locate/jtice
Effect of lanthanum doping on the lifetime of Co/g-Al2O3 catalysts in Fischer-Tropsch synthesis Mohammad Reza Hemmati a,b, Mohammad Kazemeini a,*, Jamshid Zarkesh b, Farhad Khorasheh a a b
Department of Chemical and Petroleum Engineering, Sharif University of Technology (SUT), Tehran, Iran Research Institute of Petroleum Industry (RIPI), National Iranian Oil Co (NIOC), Tehran, Iran
A R T I C L E I N F O
A B S T R A C T
Article history: Received 24 November 2011 Received in revised form 22 January 2012 Accepted 2 March 2012 Available online 10 April 2012
Cobalt-based catalysts were prepared on gamma alumina supports, and their behaviour for different Fischer-Tropsch synthesis (FTS) conditions was assessed. Although Co/g-Al2O3 is a well-known FTS catalyst, its durability ought to be improved to make the industrial process economically feasible. The effect of lanthanum doping on the catalyst lifetime was examined utilising reactor tests and catalyst characterization techniques including TPR, ICP and N2 porosimetry. Reactor test results revealed that an optimum amount of lanthanum improved catalyst activity and selectivity. Increasing amounts of lanthanum doping up to about 1.1 wt% seemed to modify the chemical composition of the support resulting in improved catalyst selectivity and lifetime. Further increase in lanthanum doping up to 2.7 wt% only marginally enhanced the catalyst selectivity and lifetime. TPR results revealed that the high temperature peak due to cobalt aluminate phase shifted to lower temperatures with increasing amount of doped lanthanum possibly due to the formation of lanthanum/aluminium mixed oxides. Crown Copyright ß 2012 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Cobalt catalyst Fischer-Tropsch synthesis Deactivation Lanthanum doping
1. Introduction Cobalt and iron are the most widely used active metals for the preparation of the Fischer-Tropsch synthesis (FTS) catalysts [1]. Iron catalysts are more selective toward the olefin and gasoline cuts, produce more CO2, have lower durability and a lower price. The research field is still wide open to investigate all aspects of iron-based FTS catalysts including the effects of feed impurities that are of considerable importance in commercial plants [2]. Both catalysts can be used in different reactors and recent investigations have reported the benefits and disadvantages of various reactors for each catalyst [3]. Cobalt-based catalysts result in a wide range of product distributions, produce water as a major product, and are more expensive than iron-based catalysts. Thus, cobalt-based catalysts need to have a longer lifetime to overcome their negative price trade-off and be economically competitive with iron-based catalysts. The most important deactivation agent for cobalt catalysts is water with both reversible and irreversible effects [4–7]. Another important deactivating agent for the system
* Corresponding author at: Department of Chemical and Petroleum Engineering, Sharif University of Technology, Azadi Avenue, Tehran 11365-9465, Iran. Tel.: +98 21 66165425; fax: +98 21 66164823. E-mail addresses: [email protected] (M.R. Hemmati), [email protected] (M. Kazemeini), [email protected] (J. Zarkesh), [email protected] (F. Khorasheh).
undertaken is carbon deposition for which mechanism of composition as well as origin and effects are considered elsewhere in detail [8,9]. Other factors such as reactor type and H2/CO ratio have a lesser effect on the deactivation of cobalt catalysts [10]. Catalyst preparation conditions, support properties, support modifications prior to impregnation, conditions and number of calcination steps, and the type of precursor are other important factors affecting the catalyst selectivity and lifetime. Due to repetition of calcination steps, for example, porosimetric properties and also particle size and particle size distribution deteriorated [11]. The type and properties of alumina and silica, which are most commonly used as the support for cobalt-based catalysts, alter the deactivation behaviour of the catalyst [12–15]. The reduction procedure can also affect the primary activity and the deactivation behaviour [16]. The cobalt dispersion and crystal size are important properties affecting the behaviour of the catalyst [17– 20]. There is an optimum range for cobalt crystal size as smaller crystals deactivate rapidly by oxidative effect of water and larger crystals lead to lower cobalt dispersion. It should be noted that cobalt crystal oxidation is a function of some interacting parameters such as the H2O/H2 (molar) ratio in the reaction media [9] as well as the type of support. Furthermore, crystal size also has effects on activity and selectivity of the catalysts. In fact, studies showed that smaller crystals led to more methane and less C5+ species. On the other hand, smaller crystals might have not provided the required active sites for a reaction to take place [21].
1876-1070/$ – see front matter . Crown Copyright ß 2012 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jtice.2012.03.002
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
Ultimately, smaller crystals might be more prone toward sintering and their reduction could be more troublesome. This investigation deals with possible modification of the support prior to cobalt impregnation. The properties of g-alumina support may be modified by doping with elements such as molybdenum [22] and lanthanum [23–26]. In addition to modifying the alumina support, lanthanum can also be utilised as activity and selectivity promoter for cobalt based FTS catalysts. The influence of La loading on Zr-Co/activated carbon (AC) catalysts was reported recently [24]. The experimental results showed that when a low La loading of 0.2 wt% was added to the ZrCo/AC catalyst, CO conversion and C5+ selectivity increased while methane selectivity decreased. However, high loadings of La resulted in a decrease in catalyst activity and C5+ selectivity and an increase in methane selectivity. The results also indicated that lanthanum loading had no significant effect on cobalt dispersion. The addition of a small amount of La increased the reducibility of the Zr-Co/AC while excess amount of La led to the decrease of the reducibility of Co catalyst thus resulted in higher methanation activity [24]. These types of catalysts were also used to investigate the effect of La on the production of a-alcohols in the C1–C18 range under mild processing conditions. The selectivity towards alcohols was improved by La doping. The reducibility of the catalyst decreased and the Co dispersion improved due to the strong interaction between Co and La2O3 species. La2O3 might promote the formation of cobalt carbides (Co2C) which are postulated to play an important role in the syntheses of the mixed linear aalcohols [26]. Albeit such species formation was still a route led to losses in activity (and perhaps caused permanent deactivation) by excluding fractions of available carbons on the surface as active sites while also increased in methane selectivity [27]. Thus, a compensating (i.e., optimum) amount of La2O3 doped onto the catalyst was indeed necessary to prolong its lifetime through making it more deactivation resistant. On the other hand, high Co dispersion and an appropriate ratio of Co2+/Co0 can enhance the activity of CO hydrogenation [26]. In another investigation [25], alumina was doped with La2O3 by either impregnation or co-precipitation. The doped and undoped alumina were then used as supports for the preparation of cobalt catalysts by incipient wetness impregnation. The results of reactor test and physicochemical characterizations indicated that catalysts prepared from La2O3-doped support showed better reducibility, activity and product selectivity [25]. The major goal of the present study was to investigate the effects of La doping on g-alumina supports before Co impregnation on the catalyst lifetime. For comparison, an in-house recipe for catalyst preparation [11,16] was utilised as a baseline. Four different catalysts with different lanthanum loadings were prepared and tested in a microreactor and characterised by techniques including ICP, TPR and N2 porosimetry. 2. Materials and methods 2.1. Catalyst preparation Lanthanum nitrate hexa-hydrate (La(NO3)36H2O, Art. No. 105326, assay 99%) obtained from Merck Chemicals, was dissolved in distilled water. The g-alumina support was produced from calcination of a bohmite sample provided by the Research Institute of Petroleum Industry (RIPI). To investigate the effects of lanthanum loading, four different catalysts were prepared by dry impregnation (C.1, C.7, C.8 and C.9 with 0.61, 0.0, 1.1 and 2.7 wt% of lanthanum, respectively). Lanthanum impregnation was followed by drying at 120 8C (from ambient temperature using a 1 8C/min ramp) for 2 h and subsequent calcination at 450 8C (2 8C/ min ramp) for 3.5 h to complete the support preparation process.
705
The prepared support was cooled to room temperature prior to cobalt impregnation. Cobalt (II) nitrate hexa-hydrate was obtained from Merck Chemicals (Co(NO3)26H2O, Art. No. 102534, assay 97%) and ruthenium (III) nitrosyl nitrate solution, which provided the activation promoting effects of ruthenium on the final catalyst, was obtained from Acros (Ru(NO)(NO3)x(OH)y, x + y = 3, Art. No. 365041). Cobalt (II) nitrate hexa-hydrate and ruthenium (III) nitrosyl nitrate solutions were diluted in concentrated nitric acid and dissolved in distilled water; yielding a solution suitable for making a 10 wt% cobalt catalyst (based on the catalyst weight) with 0.5 wt% ruthenium (based on the existing cobalt on the catalyst). After 3 impregnations of this solution onto the support, the final catalyst was expected to be loaded with 14–17 wt% Co and 0.5 wt% Ru. Because of the low solubility limit of cobalt nitrate in water at ambient temperature, the solution was heated to 70 8C to ensure complete dissolution. All impregnations were performed using the dry impregnation (incipient wetness) method. The drying and calcination procedures were similar to those performed after lanthanum impregnation. 2.2. Reactor test All catalysts were tested in a fixed bed microreactor for more than 200 h, to obtain sufficient experimental data to measure their primary catalyst activity, decrease in catalyst activity with timeon-stream, methane selectivity, and liquid product yields, including water and hydrocarbons. The fixed bed microreactor was a 3/8 in. stainless steel tube, into which the catalyst was poured and reduced in situ. The reactor was placed in a molten salt bath with a stirrer, to ensure uniform temperature along the catalyst bed. A schematic of the microreactor laboratory setup is shown in Fig. 1. The reactor was loaded with 3 g of fresh catalyst, which was dried at 70 8C in an oven for 2 h and milled until it was able to pass through a 30-mesh, but not through a 60-mesh sieve. Catalyst powder was diluted before loading into the reactor by inert materials with 1:3 ratio to facilitate heat transfer and preventing hot spots to occur. The catalyst bed was placed in the middle of the reactor in between layers of inert materials to ensure a uniform temperature. To reduce the catalyst, H2 (60 cc/min g.cat) at atmospheric pressure was passed through the reactor, while the reactor temperature was increased from ambient temperature to 400 8C using a 1–2 8C/min ramp. Once the final temperature was reached, temperature, H2 flow and pressure were kept constant for at least 8 h to complete the reduction reaction. Then, while maintaining H2 flow, the electric heater was turned off and the reactor temperature was allowed to drop from 400 8C to the reactor test temperature of 230 8C over a period of approximately 10–12 h. Although 220 8C has been reported to be the optimal temperature for FTS with this catalyst, the higher temperature of 230 8C was used in this study to accelerate the deactivation process, thereby reducing data acquisition time. Once the reactor temperature was stabilised at 230 8C, a feed with an H2/CO molar ratio of 2.0 and a total flow of 60 cc/min g.cat was introduced into the reactor at 20 bars of pressure, while the backpressure valve simultaneously blocked the outlet flow until the reactor pressure reached 20 bar. At this pressure, continuous flow of the feed into the reactor was maintained for between 170 and 220 h, depending on the deactivation rate of the catalyst. The start-of-the-run was considered to be the instant when the first flow of gaseous products from the reactor was detected. During the reaction, all products from the reactor were directed into a hot trap at 95 8C to collect the heavy components (which were nearly solid at room temperature), followed by a cold trap at 0 8C to collect the light liquid fractions. The remaining gaseous products were directed into a gas chromatograph instrument for further analysis. A Varian 3800 Gas Chromatograph with three channels used for the
706
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
Fig. 1. Schematic of the microreactor test setup constructed and used in this study.
analysis. The first channel utilised for hydrogen analysis. It was equipped with a Hayesep Q and molsieve 5A columns utilising the front TCD detector. The permanent gas channel comprised three columns including; Hayesep N, Hayesep Q and molsieve 13X which used the middle TCD detector to analyse the CO and CO2. This channel was also capable of detecting the N2. The third channel was for hydrocarbon analysis with a Petrocol DH column which utilised the rear FID. Reactor pressure and temperature were kept constant during the entire reactor tests. Although it is believed that pressurising the reactor with syngas might lead to some permanent loss in primary catalyst activity, the main focus of the current study was to investigate the long-term deactivation behaviour, as the initial loss in catalyst activity would be similar for all catalyst samples. 2.3. Catalyst characterisation techniques All catalysts were characterised by N2 porosimetry, TPR and ICP (Inductively Coupled Plasma). N2 porosimetry analyses were performed using a Micromeritics ASAP 2010 system and N2 at 77 K. TPR was performed using a Micromeritics TPD-TPR 2900
system. This system is equipped with a TCD detector which is an integral part of the system and calculated the H2 consumption rate. 3. Results and discussion 3.1. Reactor tests results For each reactor test run, catalyst activity (defined as the fractional conversion of CO) and methane selectivity versus timeon-stream were obtained from the experimental data. For FTS, it is desirable to keep the methane selectivity as low as possible, while maintaining selectivity towards C5+ liquid products as high as possible. Fig. 2 shows the catalyst activity and methane selectivity for catalyst C.1 over a 217 h reactor test run. As can be seen, C.1 catalyst activity versus time-on-stream has two distinct stages. During the first stage, the loss of C.1 catalyst activity was very rapid, decreasing from a very high value of about 0.9 to between 0.45 and 0.6 within the first 4–10 h of operation. Following this initial rapid activity loss, a second stage with a smooth and gradual loss of activity develops. The first stage can be attributed to the enhanced diffusion limitations experienced
Fig. 2. C.1 catalyst activity. (^) initial activity during the first hours of the experiment; (&) activity after the first hours of the experiment; (~) methane selectivity for 217 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
707
Fig. 3. C.7 catalyst activity. (^) initial activity during the first hours of the experiment; (&) activity after the first hours of experiment; (~) methane selectivity for 310 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
by the gaseous feed components diffusing into the catalyst pores and the liquid products diffusing out of the catalyst pores, as catalyst pores become filled with liquid products. Also some permanent deactivation was introduced to the catalysts due to sintering and other causes. Once the process of pore-filling by the liquid products is complete, the catalyst activity drops to a so-called primary activity level. Thereafter, the gradual loss in activity during the second stage is due to deactivating factors, including: aging, sintering, chemical deactivation, and cobalt aluminate formation. Fig. 2 also suggests that methane selectivity for C.1 is constant over the entire period of the run; however, regression of the experimental data revealed a slight decrease in methane selectivity with timeon-stream. Catalyst activity and methane selectivity versus time-on-stream for catalysts C.7, C.8, and C.9 are presented in Figs. 3–5, respectively. The overall performance of catalysts C.1 and C.7–C.9, with respect to primary activity, slope of activity loss with time-onstream, average activity for the entire test period, primary
methane selectivity, slope of methane selectivity change with time-on-stream, and average methane selectivity for the entire test period, are reported in Table 1. Catalyst C.9 had the best performance in 4 out of 6 of the above categories. The addition of more lanthanum makes the catalyst more resistant towards deactivation as indicated by a decrease in the negative slope of activity versus time-on-stream with increasing lanthanum loadings, possibly due to a reduction in cobalt alumina interactions that would result in spinel formation. Table 2 shows TOS (time-on-stream), liquid production rate and ICP results of the all prepared catalysts. Catalyst C.1 displayed the best liquid production rates over the entire TOS. Liquid production rate is the sum of hydrocarbon phases collected up at the hot and cold trap locations. Although during product venting from the vessels all of the products were in liquid phase, but after cooling the containers down to room temperature, a solid phase prevailed due to freezing of high melting temperature materials. Indeed, this fraction contained the most valuable products of the FTS in the present study which in essence was the C5+ material.
Fig. 4. C.8 catalyst activity. (^) initial activity during the first few hours of the experiment; (&) activity after the first few hours of experiment; (~) methane selectivity for 310 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
708
Fig. 5. C.9 catalyst activity. (^) initial activity during the first few hours of the experiment; (&) activity after the first few hours of the experiment; (~) methane selectivity for 310 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
Table 1 Performance comparison between catalysts C.1 and C.7–C.9. Catalyst
Primary activitya
C.1 C.7 C.8 C.9
0.56b 0.49 0.50 0.52
a b
Slope of activity change 9 10 4 10 10 4 5 10 5 3 10 5b
Average activity
Primary methane selectivitya
0.48 0.45 0.48 0.50b
6.32 10 6.29 10 5.77 10 4.68 10
2 2 2 2b
Slope of selectivity change 0.4 10 5 10 5 2 10 5 4 10 5
5b
Average methane selectivity 5.73 10 5.41 10 5.11 10 3.68 10
2 2 2 2b
Final Score 2/6 0/6 0/6 4/6
Extrapolated by the linear regression. This is the intercept of the correlated line. Best result amongst all catalysts.
Experimental data on liquid production rates (Table 2) indicated that both C.8 and C.9 catalysts had approximately the same liquid production rates as catalyst C.1. Although addition of lanthanum beyond 2.7 wt% is expected to slightly improve catalyst performance, there is a trade-off between lanthanum doping costs, which rise with increased loading, and the incremental increase in catalyst lifetime obtained. Fig. 6 shows the absolute value of the slope of the activity versus time-on-stream as a function of lanthanum loading. After an initial rapid fall, a lower limit plateau is reached, indicating that additional lanthanum loading would not result in any meaningful change in the slope of activity loss. Although data presented in Tables 1 and 2 indicated that catalyst C.9 had the best performance, nonetheless, Fig. 6 introduced some doubts into this justification regarding higher lanthanum doping. The most significant change in the performance of the catalysts occurred when the La loading was increases from 0.61 to 1.1 wt%. There are some postulations regarding the role of lanthanum doping on the deactivation rate of cobalt-based FTS catalysts with alumina support. The strongest and most logical explanation is the
production of a mixed oxide phase, comprising aluminium, lanthanum and oxygen, which is highly resistant toward cobalt spinel formation. Cobalt can react with alumina support leading to the formation of a highly stable cobalt aluminate phase which could not be reduced to metallic cobalt under regular reduction conditions. The mixed oxide phase seems to be more resistant toward deactivation by cobalt spinel formation. The results of the present study indicated that this effect was limited to a lanthanum loading of 1.1 wt% and beyond this value no significant changes were seen. 3.2. Catalyst characterisation with N2 porosimetry and TPR techniques BET surface areas, pore volumes, and mean pore diameters were obtained from N2 porosimetry analysis and are reported in Table 3
Table 2 TOS, liquid production rate and ICP results of the prepared catalysts. Catalyst
TOS (h)
Liquid production rate (g/g.cat day)a
Co loading (wt%)
La loading (wt%)
C.1 C.7 C.8 C.9
217 310 522 530
4.99 4.47 4.97 4.66
14.0 14.8 15.2 15.5
0.61 0.0 1.1 2.7
a This liquid phase products only appeared in hot and cold traps, comprise C5+ hydrocarbons and are inherently a mixture of liquid and solid wax. C2–C7 hydrocarbons also appeared in the gas phase products leaving the cold trap. In addition, traces of water soluble hydrocarbons existed in aqueous phase of hot and cold traps.
Fig. 6. Effect of lanthanum doping (wt%) on the absolute value of the slope of activity versus time-on-stream (h 1).
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
709
Table 3 N2 porosimetry results for catalysts and the alumina support. Catalyst
BET surface area (m2/g)
Pore volume (cm3/g)
Mean pore diameter (Nm)
C.1 C.7 C.8 C.9 g-Al2O3
171 180 172 171 271
0.40 0.39 0.37 0.36 0.70
8.3 8.7 8.6 8.4 10.3
for all catalysts as well as for the alumina support. As expected, the catalyst surface area and pore volume decreased with increasing numbers of calcination steps due to sintering. Catalysts C.1, C.8, and C.9 experienced two calcinations; but catalyst C.7 experienced only one calcination step. Comparison of the specific surface areas and pore volumes for different catalysts with those for the alumina support indicated that the loss in pore volume and surface area due to sintering was most pronounced during the first calcination step. The loss in surface area and pore volume due to the additional calcination was only marginal. Catalyst surface area and pore volume also slightly decreased with increasing lanthanum loading. The effect of various parameters on the TPR analysis of aluminasupported cobalt FTS catalysts has been reported in previous studies [6,28,29]. Here the focus was on the location of peaks for cobalt species present in the catalysts, and on the total amount of H2 consumed, as a measure of the quantity of cobalt available for reaction. Any low temperature peak below 500 K would be due to cobalt nitrate species remaining on the catalyst surface after calcination. The next medieval temperature peaks (after 500 K) belonged to the two step reduction of the Co3O4 phase present in the catalyst pores to metallic cobalt. In the first step, the Co3O4 species was reduced to the CoO after which this material reacted with hydrogen to result in the metallic cobalt [26,27]. It should be noted that larger crystals shall reduce faster than the smaller ones which are highly of dispersed cobalt oxide phase [6]. Activity promoters such as ruthenium can play a role in making these crystals reduce earlier. The final peak above 1000 K was attributed to a cobalt aluminate phase which is very hard to reduce [6,7,28,29]. Data presented in Table 4 indicated that most of the hydrogen consumption form TPR analysis was from the second peak due to small and finely dispersed crystals. Note that there are some differences in the hydrogen consumption data between catalysts C.1 and catalysts C.7–C.9, with the former displaying lower values. Catalyst C.1 had the lowest cobalt loading and the least total hydrogen consumption. Better dispersion of cobalt over the surface of alumina is one factor which would lead to higher H2 consumption observed for C.9. In the TPR diagram for catalysts C.7–C.9, an early peak from cobalt nitrate was observed, although the H2 consumption for this peak was not significant. Incorporation of lanthanum most probably caused the high temperature peak of the cobalt aluminate to be shifted to lower temperatures. This phenomenon was rather more exaggerated at higher lanthanum loadings, such that for the highest lanthanum loaded sample this peak might have either slightly remained or really diminished. This
issue was evidenced by looking at Fig. 7. The C.7 catalyst (i.e., without any lanthanum doping) showed a high temperature peak which gradually diminished through the C.1, C.8 and C.9 catalysts. Although it is believed that total amount of H2 consumed in TPR in particular (and of course, other peaks in general) is directly proportional to the average activity of a given catalyst however, other factors are also of considerable importance. For instance, in the present results, the trend of increase in H2 consumption did not completely correspond with those of average activity change, but the discrepancy might be justified regarding the interaction of different parameters including; (i) the total amount of loaded cobalt and lanthanum, (ii) aluminate phase formation and fraction of total cobalt which is consumed by the aluminate and (iii) share of large and small clusters of cobalt. All of these parameters and their impacts on the apparent activity were shown in Table 5 where the ranking of different effects presented. This table clearly indicated the importance of cobalt and lanthanum loading amongst all. The trend of activity changes thoroughly fitted with that of lanthanum doped amount. Although the H2 consumption is directly a sign of total available cobalt for the reaction, but this cobalt phase should be of appropriate size to be suitable for the reaction [21]. Also it should not be from aluminate phase (i.e., with the high temperature peak symptom). The main reason for the present results observed is that all of the cobalt species available on the catalysts surfaces were detected by the TPR method however; just part of these remained active during a long period of reactor tests. A possible rational for the recent phenomena is oxidation of small clusters and aluminate phase formation which would be prevented by doping of lanthanum. Thus, for better correlation between the total H2 consumption and average activity, one should be reminded of the effects of lanthanum and total loaded cobalt. This issue may completely justify the present TPR results and their relation with the average activity.
Table 4 H2 consumption for different catalysts from TPR analysis.
Table 5 Ranking different affecting factors in the average activity.
Fig. 7. TPR diagram for catalysts C.1–C.3 and C.7–C.9.
Catalyst
First peak temp. (K)
Second peak temp. (K)
H2 from first peak (mmole/g)
H2 from second peak (mmole/g)
Total H2 (mmole/g)
Second Total H2 Catalyst Loaded Loaded First Average peak H2 name Co La peak H2 consumption activity consumption consumption
C.1 C.7 C.8 C.9
572 579 596 599
805 887 906 783
4.58 6.04 7.08 6.16
8.84 12.64 12.18 11.92
13.42 18.68 19.26 18.08
C.1 C.7 C.8 C.9
4 3 2 1
3 4 2 1
4 3 1 2
4 1 2 3
4 2 1 3
3 4 2 1
710
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
Nonetheless, looking at this table, still a point remains to be ratified in it, namely, the justification for less H2 consumption for C.9 catalyst which has the most amount of loaded cobalt. The rational for this is found in Table 4. Although C.8 presented the most H2 consumption; but this difference was not so large (i.e., 3% for the C.7 and 6% for the C.9 materials based upon the C.8 catalyst). Nonetheless, the point is that for the C.9 material, the high temperature peak has most probably shifted to a lower temperature (distancing from the cobalt aluminate peak temperature) which is a significant factor for the C.9 catalysts and could justify its appropriate average activity.
[10]
[11]
[12]
[13]
4. Conclusions
[14]
To investigate the effects lanthanum doping on the physiochemical behaviour and durability of FTS catalysts, four catalyst samples were made and evaluated. All samples were supported on g-alumina with nearly the same chemical formulation expect for the amount of doped lanthanum. The prepared catalysts were subsequently analysed by ICP, TRR and N2 porosimetry techniques and their performance was tested in a FTS reactor. The objective was to prepare catalysts which were resistant to deactivation. Results showed that increasing the amount of lanthanum led to superior performance in terms of catalyst primary activity, average activity during time, activity change with time-on-stream, average selectivity, and selectivity change with time-on-stream. Lanthanum doping alters the chemical state of the support by forming mixed oxides with alumina that are more resistant to oxidation. This is reinforced by TPR results as the high temperature peak due to the cobalt aluminate phase diminished with increased loading of lanthanum. However, an increase in lanthanum doping beyond 1.1 wt% only marginally improved the deactivation resistance of the catalyst.
[15]
[16]
[17]
[18]
[19]
[20]
[21]
References [22] [1] Iglesia E. Design, synthesis, and use of cobalt-based Fischer-Tropsch synthesis catalysts. Applied Catalysis A General 1997;161:59–78. [2] Li T, Yang YZ, Wan H, An X, Zhang C, Xiang H, Li Y. Effect of sulfate on an iron manganese catalyst for Fischer-Tropsch synthesis. Journal of Natural Gas Chemistry 2007;16:354–62. [3] Sehabiague L, Lemoine R, Behkish A, Heintz YJ, Sanoja M, Oukaci R, Morsi BI. Modeling and optimization of a large-scale slurry bubble column reactor for producing 10,000 bbl/day of Fischer-Tropsch liquid hydrocarbons. Journal of the Chinese Institute of Chemical Engineers 2008;39:169–79. [4] Dalai AK, Davis BH. Fischer-Tropsch synthesis: a review of water effects on the performances of unsupported and supported Co catalysts. Applied Catalysis A General 2008;348:1–15. [5] Li J, Zhan X, Zhang Y, Jacobs G, Das T, Davis BH. Fischer-Tropsch synthesis: effect of water on the deactivation of Pt promoted Co/Al2O3 catalysts. Applied Catalysis A General 2002;228:203–12. [6] Hilmen AM, Schanke D, Hanssen KF, Holmen A. Study of the effect of water on alumina supported cobalt Fischer-Tropsch catalysts. Applied Catalysis A General 1999;186:169–88. [7] Krishnamoorthy S, Tu M, Ojeda MP, Pinna D, Iglesia E. An investigation of the effects of water on rate and selectivity for the Fischer-Tropsch synthesis on cobalt-based catalysts. Journal of Catalysis 2002;211:422–33. [8] Tsakoumis NE, Ronning M, Borg Ø, Rytter E, Holmen A. Deactivation of cobalt based Fischer-Tropsch catalysts: a review. Catalysis Today 2010;154:162–82. [9] Saib AM, Moodley DJ, Ciobıˆca IM, Hauman MM, Sigwebela BH, Weststrate CJ, Niemantsverdriet JW, Van de Loosdrecht J. Fundamental understanding of
[23]
[24]
[25]
[26]
[27]
[28]
[29]
deactivation and regeneration of cobalt Fischer-Tropsch synthesis catalysts. Catalysis Today 2010;154:271–82. Zhou W, Chen JG, Fang KG, Sun YH. The deactivation of Co/SiO2 catalyst for Fischer-Tropsch synthesis at different ratios of H2 to CO. Fuel Processing Technology 2006;87:609–16. Hemmati MR, Kazemeini M, Zarkesh J, Khorasheh F. Effect of preparation parameters on catalytic performance of alumina supported Co catalysts utilized in Fischer-Tropsch synthesis. In: Proc. of the 7th European congress of chemical engineering and 19th international congress of chemical and process engineering; CHISA 2010, Prague, Czech Republic, August 2010. Borg Ø, Eri S, Blekkan EA, Storsæter S, Wigum H, Rytter E, Holmen A. FischerTropsch synthesis over g-alumina-supported cobalt catalysts: effect of support variables. Journal of Catalysis 2007;248:89–100. Rane S, Borg Ø, Yang J, Rytter E, Holmen A. Effect of alumina phases on hydrocarbon selectivity in Fischer-Tropsch synthesis. Applied Catalysis A General 2010;388:160–7. Jacobs G, Das TK, Li J, Luo M, Patterson PM, Davis BH. Fischer-Tropsch synthesis: influence of support on the impact of co-fed water for cobalt-based catalysts. Catalysts and Catalysis 2007;217–53. Martinez A, Prieto G, Rollan J. Nanofibrous g-Al2O3 as support for Co-based Fischer-Tropsch catalysts: pondering the relevance of diffusional and dispersion effects on catalytic performance. Journal of Catalysis 2009;263:292–305. Karimi A, Nakhaei Pour A, Torabi F, Hatami B, Tavasoli A, Alaei M, Irani M. Fischer-Tropsch synthesis over ruthenium-promoted Co/Al2O3 catalyst with different reduction procedures. Journal of Natural Gas Chemistry 2010;19: 503–8. Burakorn T, Panpranot J, Mekasuwandumrong O, Chaisuk C, Praserthdam P, Jongsomjit B. Characterization of cobalt dispersed on the mixed nanoscale alumina and zirconia supports. Journal of Materials Processing Technology 2008;206:352–8. Prieto G, Martı´nez A, Concepcio´n P, Moreno-tost R. Cobalt particle size effects in Fischer-Tropsch synthesis: structural and in situ spectroscopic characterisation on reverse micelle-synthesised Co/ITQ-2 model catalysts. Journal of Catalysis 2009;266:129–44. Borg Ø, Dietzel PDC, Spjelkavik AI, Tveten EZ, Walmsley JC, Diplas S, Eri S, Holmen A, Rytter E. Fischer-Tropsch synthesis: cobalt particle size and support effects on intrinsic activity and product distribution. Journal of Catalysis 2008;259:161–4. Radstakea PB, den Breejenb JP, Bezemerb GL, Bitterb JH, de Jong KP, Frsetha V, Holmen A. On the origin of the cobalt particle size effect in the Fischer-Tropsch synthesis. Natural Gas Conversion 2007;VIII:85–90. Bezemer GL, Bitter JH, Kuipers HPCE, Oosterbeek H, Holewijn JE, Xu X, Kapteijn F, van Dillen AJ, de Jong KP. Cobalt particle size effects in the Fischer-Tropsch reaction studied with carbon nanofiber supported catalysts. Journal of American Chemical Society 2006;128:3956. van Steen E, Viljoen EL, van de Loosdrecht J, Claeys M. Evaluation of molybdenum-modified alumina support materials for Co-based Fischer-Tropsch catalysts. Applied Catalysis A General 2008;335:56–63. Bedel L, Roger AC, Rehspringer JL, Zimmermann Y, Alain Kiennemann A. La(1 y)Co0.4Fe0.6O3-d perovskite oxides as catalysts for Fischer-Tropsch synthesis. Journal of Catalysis 2005;235:279–94. Wang T, Ding Y, Lu¨ Y, Zh H, Lin L. Influence of lanthanum on the performance of Zr-Co/activated carbon catalysts in Fischer-Tropsch synthesis. Journal of Natural Gas Chemistry 2008;17:153–8. Cai Z, Li J, Liew K, Hu J. Effect of La2O3-dopping on the Al2O3 supported cobalt catalyst for Fischer-Tropsch synthesis. Journal of Molecular Catalysis A General 2010;330:10–7. Jiao G, Ding Y, Zhu H, Li X, Li J, Lin R, Dong W, Gong L, Pei Y, Lu Y. Effect of La2O3 doping on syntheses of C1–C18 mixed linear a-alcohols from syngas over the Co/AC catalysts. Applied Catalysis A General 2009;364:137–42. Gruver V, Zhan X, Engman J, Robota HJ, Suib SL, Polverjan M. Deactivation of a Fischer-Tropsch catalyst through the formation of cobalt carbide under laboratory slurry reactor conditions. Preprints—American Chemical Society. Division of Petroleum Chemistry, Fischer-Tropsch: Materials, Theories, and Practice. Symposium. 2004;49(2):192–94. Su H, Zeng S, Dong H, Du Y, Zhang Y, Hu R. Pillared montmorillonite supported cobalt catalysts for the Fischer-Tropsch reaction. Applied Clay Science 2009;46: 325–9. Zeng S, Du Y, Su H, Zhang Y. Promotion effect of single or mixed rare earths on cobalt-based catalysts for Fischer-Tropsch synthesis. Catalysis Communication 2011;13:6–9.
Contents lists available at SciVerse ScienceDirect
Journal of the Taiwan Institute of Chemical Engineers journal homepage: www.elsevier.com/locate/jtice
Effect of lanthanum doping on the lifetime of Co/g-Al2O3 catalysts in Fischer-Tropsch synthesis Mohammad Reza Hemmati a,b, Mohammad Kazemeini a,*, Jamshid Zarkesh b, Farhad Khorasheh a a b
Department of Chemical and Petroleum Engineering, Sharif University of Technology (SUT), Tehran, Iran Research Institute of Petroleum Industry (RIPI), National Iranian Oil Co (NIOC), Tehran, Iran
A R T I C L E I N F O
A B S T R A C T
Article history: Received 24 November 2011 Received in revised form 22 January 2012 Accepted 2 March 2012 Available online 10 April 2012
Cobalt-based catalysts were prepared on gamma alumina supports, and their behaviour for different Fischer-Tropsch synthesis (FTS) conditions was assessed. Although Co/g-Al2O3 is a well-known FTS catalyst, its durability ought to be improved to make the industrial process economically feasible. The effect of lanthanum doping on the catalyst lifetime was examined utilising reactor tests and catalyst characterization techniques including TPR, ICP and N2 porosimetry. Reactor test results revealed that an optimum amount of lanthanum improved catalyst activity and selectivity. Increasing amounts of lanthanum doping up to about 1.1 wt% seemed to modify the chemical composition of the support resulting in improved catalyst selectivity and lifetime. Further increase in lanthanum doping up to 2.7 wt% only marginally enhanced the catalyst selectivity and lifetime. TPR results revealed that the high temperature peak due to cobalt aluminate phase shifted to lower temperatures with increasing amount of doped lanthanum possibly due to the formation of lanthanum/aluminium mixed oxides. Crown Copyright ß 2012 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Cobalt catalyst Fischer-Tropsch synthesis Deactivation Lanthanum doping
1. Introduction Cobalt and iron are the most widely used active metals for the preparation of the Fischer-Tropsch synthesis (FTS) catalysts [1]. Iron catalysts are more selective toward the olefin and gasoline cuts, produce more CO2, have lower durability and a lower price. The research field is still wide open to investigate all aspects of iron-based FTS catalysts including the effects of feed impurities that are of considerable importance in commercial plants [2]. Both catalysts can be used in different reactors and recent investigations have reported the benefits and disadvantages of various reactors for each catalyst [3]. Cobalt-based catalysts result in a wide range of product distributions, produce water as a major product, and are more expensive than iron-based catalysts. Thus, cobalt-based catalysts need to have a longer lifetime to overcome their negative price trade-off and be economically competitive with iron-based catalysts. The most important deactivation agent for cobalt catalysts is water with both reversible and irreversible effects [4–7]. Another important deactivating agent for the system
* Corresponding author at: Department of Chemical and Petroleum Engineering, Sharif University of Technology, Azadi Avenue, Tehran 11365-9465, Iran. Tel.: +98 21 66165425; fax: +98 21 66164823. E-mail addresses: [email protected] (M.R. Hemmati), [email protected] (M. Kazemeini), [email protected] (J. Zarkesh), [email protected] (F. Khorasheh).
undertaken is carbon deposition for which mechanism of composition as well as origin and effects are considered elsewhere in detail [8,9]. Other factors such as reactor type and H2/CO ratio have a lesser effect on the deactivation of cobalt catalysts [10]. Catalyst preparation conditions, support properties, support modifications prior to impregnation, conditions and number of calcination steps, and the type of precursor are other important factors affecting the catalyst selectivity and lifetime. Due to repetition of calcination steps, for example, porosimetric properties and also particle size and particle size distribution deteriorated [11]. The type and properties of alumina and silica, which are most commonly used as the support for cobalt-based catalysts, alter the deactivation behaviour of the catalyst [12–15]. The reduction procedure can also affect the primary activity and the deactivation behaviour [16]. The cobalt dispersion and crystal size are important properties affecting the behaviour of the catalyst [17– 20]. There is an optimum range for cobalt crystal size as smaller crystals deactivate rapidly by oxidative effect of water and larger crystals lead to lower cobalt dispersion. It should be noted that cobalt crystal oxidation is a function of some interacting parameters such as the H2O/H2 (molar) ratio in the reaction media [9] as well as the type of support. Furthermore, crystal size also has effects on activity and selectivity of the catalysts. In fact, studies showed that smaller crystals led to more methane and less C5+ species. On the other hand, smaller crystals might have not provided the required active sites for a reaction to take place [21].
1876-1070/$ – see front matter . Crown Copyright ß 2012 Taiwan Institute of Chemical Engineers. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.jtice.2012.03.002
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
Ultimately, smaller crystals might be more prone toward sintering and their reduction could be more troublesome. This investigation deals with possible modification of the support prior to cobalt impregnation. The properties of g-alumina support may be modified by doping with elements such as molybdenum [22] and lanthanum [23–26]. In addition to modifying the alumina support, lanthanum can also be utilised as activity and selectivity promoter for cobalt based FTS catalysts. The influence of La loading on Zr-Co/activated carbon (AC) catalysts was reported recently [24]. The experimental results showed that when a low La loading of 0.2 wt% was added to the ZrCo/AC catalyst, CO conversion and C5+ selectivity increased while methane selectivity decreased. However, high loadings of La resulted in a decrease in catalyst activity and C5+ selectivity and an increase in methane selectivity. The results also indicated that lanthanum loading had no significant effect on cobalt dispersion. The addition of a small amount of La increased the reducibility of the Zr-Co/AC while excess amount of La led to the decrease of the reducibility of Co catalyst thus resulted in higher methanation activity [24]. These types of catalysts were also used to investigate the effect of La on the production of a-alcohols in the C1–C18 range under mild processing conditions. The selectivity towards alcohols was improved by La doping. The reducibility of the catalyst decreased and the Co dispersion improved due to the strong interaction between Co and La2O3 species. La2O3 might promote the formation of cobalt carbides (Co2C) which are postulated to play an important role in the syntheses of the mixed linear aalcohols [26]. Albeit such species formation was still a route led to losses in activity (and perhaps caused permanent deactivation) by excluding fractions of available carbons on the surface as active sites while also increased in methane selectivity [27]. Thus, a compensating (i.e., optimum) amount of La2O3 doped onto the catalyst was indeed necessary to prolong its lifetime through making it more deactivation resistant. On the other hand, high Co dispersion and an appropriate ratio of Co2+/Co0 can enhance the activity of CO hydrogenation [26]. In another investigation [25], alumina was doped with La2O3 by either impregnation or co-precipitation. The doped and undoped alumina were then used as supports for the preparation of cobalt catalysts by incipient wetness impregnation. The results of reactor test and physicochemical characterizations indicated that catalysts prepared from La2O3-doped support showed better reducibility, activity and product selectivity [25]. The major goal of the present study was to investigate the effects of La doping on g-alumina supports before Co impregnation on the catalyst lifetime. For comparison, an in-house recipe for catalyst preparation [11,16] was utilised as a baseline. Four different catalysts with different lanthanum loadings were prepared and tested in a microreactor and characterised by techniques including ICP, TPR and N2 porosimetry. 2. Materials and methods 2.1. Catalyst preparation Lanthanum nitrate hexa-hydrate (La(NO3)36H2O, Art. No. 105326, assay 99%) obtained from Merck Chemicals, was dissolved in distilled water. The g-alumina support was produced from calcination of a bohmite sample provided by the Research Institute of Petroleum Industry (RIPI). To investigate the effects of lanthanum loading, four different catalysts were prepared by dry impregnation (C.1, C.7, C.8 and C.9 with 0.61, 0.0, 1.1 and 2.7 wt% of lanthanum, respectively). Lanthanum impregnation was followed by drying at 120 8C (from ambient temperature using a 1 8C/min ramp) for 2 h and subsequent calcination at 450 8C (2 8C/ min ramp) for 3.5 h to complete the support preparation process.
705
The prepared support was cooled to room temperature prior to cobalt impregnation. Cobalt (II) nitrate hexa-hydrate was obtained from Merck Chemicals (Co(NO3)26H2O, Art. No. 102534, assay 97%) and ruthenium (III) nitrosyl nitrate solution, which provided the activation promoting effects of ruthenium on the final catalyst, was obtained from Acros (Ru(NO)(NO3)x(OH)y, x + y = 3, Art. No. 365041). Cobalt (II) nitrate hexa-hydrate and ruthenium (III) nitrosyl nitrate solutions were diluted in concentrated nitric acid and dissolved in distilled water; yielding a solution suitable for making a 10 wt% cobalt catalyst (based on the catalyst weight) with 0.5 wt% ruthenium (based on the existing cobalt on the catalyst). After 3 impregnations of this solution onto the support, the final catalyst was expected to be loaded with 14–17 wt% Co and 0.5 wt% Ru. Because of the low solubility limit of cobalt nitrate in water at ambient temperature, the solution was heated to 70 8C to ensure complete dissolution. All impregnations were performed using the dry impregnation (incipient wetness) method. The drying and calcination procedures were similar to those performed after lanthanum impregnation. 2.2. Reactor test All catalysts were tested in a fixed bed microreactor for more than 200 h, to obtain sufficient experimental data to measure their primary catalyst activity, decrease in catalyst activity with timeon-stream, methane selectivity, and liquid product yields, including water and hydrocarbons. The fixed bed microreactor was a 3/8 in. stainless steel tube, into which the catalyst was poured and reduced in situ. The reactor was placed in a molten salt bath with a stirrer, to ensure uniform temperature along the catalyst bed. A schematic of the microreactor laboratory setup is shown in Fig. 1. The reactor was loaded with 3 g of fresh catalyst, which was dried at 70 8C in an oven for 2 h and milled until it was able to pass through a 30-mesh, but not through a 60-mesh sieve. Catalyst powder was diluted before loading into the reactor by inert materials with 1:3 ratio to facilitate heat transfer and preventing hot spots to occur. The catalyst bed was placed in the middle of the reactor in between layers of inert materials to ensure a uniform temperature. To reduce the catalyst, H2 (60 cc/min g.cat) at atmospheric pressure was passed through the reactor, while the reactor temperature was increased from ambient temperature to 400 8C using a 1–2 8C/min ramp. Once the final temperature was reached, temperature, H2 flow and pressure were kept constant for at least 8 h to complete the reduction reaction. Then, while maintaining H2 flow, the electric heater was turned off and the reactor temperature was allowed to drop from 400 8C to the reactor test temperature of 230 8C over a period of approximately 10–12 h. Although 220 8C has been reported to be the optimal temperature for FTS with this catalyst, the higher temperature of 230 8C was used in this study to accelerate the deactivation process, thereby reducing data acquisition time. Once the reactor temperature was stabilised at 230 8C, a feed with an H2/CO molar ratio of 2.0 and a total flow of 60 cc/min g.cat was introduced into the reactor at 20 bars of pressure, while the backpressure valve simultaneously blocked the outlet flow until the reactor pressure reached 20 bar. At this pressure, continuous flow of the feed into the reactor was maintained for between 170 and 220 h, depending on the deactivation rate of the catalyst. The start-of-the-run was considered to be the instant when the first flow of gaseous products from the reactor was detected. During the reaction, all products from the reactor were directed into a hot trap at 95 8C to collect the heavy components (which were nearly solid at room temperature), followed by a cold trap at 0 8C to collect the light liquid fractions. The remaining gaseous products were directed into a gas chromatograph instrument for further analysis. A Varian 3800 Gas Chromatograph with three channels used for the
706
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
Fig. 1. Schematic of the microreactor test setup constructed and used in this study.
analysis. The first channel utilised for hydrogen analysis. It was equipped with a Hayesep Q and molsieve 5A columns utilising the front TCD detector. The permanent gas channel comprised three columns including; Hayesep N, Hayesep Q and molsieve 13X which used the middle TCD detector to analyse the CO and CO2. This channel was also capable of detecting the N2. The third channel was for hydrocarbon analysis with a Petrocol DH column which utilised the rear FID. Reactor pressure and temperature were kept constant during the entire reactor tests. Although it is believed that pressurising the reactor with syngas might lead to some permanent loss in primary catalyst activity, the main focus of the current study was to investigate the long-term deactivation behaviour, as the initial loss in catalyst activity would be similar for all catalyst samples. 2.3. Catalyst characterisation techniques All catalysts were characterised by N2 porosimetry, TPR and ICP (Inductively Coupled Plasma). N2 porosimetry analyses were performed using a Micromeritics ASAP 2010 system and N2 at 77 K. TPR was performed using a Micromeritics TPD-TPR 2900
system. This system is equipped with a TCD detector which is an integral part of the system and calculated the H2 consumption rate. 3. Results and discussion 3.1. Reactor tests results For each reactor test run, catalyst activity (defined as the fractional conversion of CO) and methane selectivity versus timeon-stream were obtained from the experimental data. For FTS, it is desirable to keep the methane selectivity as low as possible, while maintaining selectivity towards C5+ liquid products as high as possible. Fig. 2 shows the catalyst activity and methane selectivity for catalyst C.1 over a 217 h reactor test run. As can be seen, C.1 catalyst activity versus time-on-stream has two distinct stages. During the first stage, the loss of C.1 catalyst activity was very rapid, decreasing from a very high value of about 0.9 to between 0.45 and 0.6 within the first 4–10 h of operation. Following this initial rapid activity loss, a second stage with a smooth and gradual loss of activity develops. The first stage can be attributed to the enhanced diffusion limitations experienced
Fig. 2. C.1 catalyst activity. (^) initial activity during the first hours of the experiment; (&) activity after the first hours of the experiment; (~) methane selectivity for 217 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
707
Fig. 3. C.7 catalyst activity. (^) initial activity during the first hours of the experiment; (&) activity after the first hours of experiment; (~) methane selectivity for 310 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
by the gaseous feed components diffusing into the catalyst pores and the liquid products diffusing out of the catalyst pores, as catalyst pores become filled with liquid products. Also some permanent deactivation was introduced to the catalysts due to sintering and other causes. Once the process of pore-filling by the liquid products is complete, the catalyst activity drops to a so-called primary activity level. Thereafter, the gradual loss in activity during the second stage is due to deactivating factors, including: aging, sintering, chemical deactivation, and cobalt aluminate formation. Fig. 2 also suggests that methane selectivity for C.1 is constant over the entire period of the run; however, regression of the experimental data revealed a slight decrease in methane selectivity with timeon-stream. Catalyst activity and methane selectivity versus time-on-stream for catalysts C.7, C.8, and C.9 are presented in Figs. 3–5, respectively. The overall performance of catalysts C.1 and C.7–C.9, with respect to primary activity, slope of activity loss with time-onstream, average activity for the entire test period, primary
methane selectivity, slope of methane selectivity change with time-on-stream, and average methane selectivity for the entire test period, are reported in Table 1. Catalyst C.9 had the best performance in 4 out of 6 of the above categories. The addition of more lanthanum makes the catalyst more resistant towards deactivation as indicated by a decrease in the negative slope of activity versus time-on-stream with increasing lanthanum loadings, possibly due to a reduction in cobalt alumina interactions that would result in spinel formation. Table 2 shows TOS (time-on-stream), liquid production rate and ICP results of the all prepared catalysts. Catalyst C.1 displayed the best liquid production rates over the entire TOS. Liquid production rate is the sum of hydrocarbon phases collected up at the hot and cold trap locations. Although during product venting from the vessels all of the products were in liquid phase, but after cooling the containers down to room temperature, a solid phase prevailed due to freezing of high melting temperature materials. Indeed, this fraction contained the most valuable products of the FTS in the present study which in essence was the C5+ material.
Fig. 4. C.8 catalyst activity. (^) initial activity during the first few hours of the experiment; (&) activity after the first few hours of experiment; (~) methane selectivity for 310 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
708
Fig. 5. C.9 catalyst activity. (^) initial activity during the first few hours of the experiment; (&) activity after the first few hours of the experiment; (~) methane selectivity for 310 h after the start of the run (3 g of catalyst, 230 8C, 20 bars, 60 cc/(min g.cat) and H2/CO = 2.0).
Table 1 Performance comparison between catalysts C.1 and C.7–C.9. Catalyst
Primary activitya
C.1 C.7 C.8 C.9
0.56b 0.49 0.50 0.52
a b
Slope of activity change 9 10 4 10 10 4 5 10 5 3 10 5b
Average activity
Primary methane selectivitya
0.48 0.45 0.48 0.50b
6.32 10 6.29 10 5.77 10 4.68 10
2 2 2 2b
Slope of selectivity change 0.4 10 5 10 5 2 10 5 4 10 5
5b
Average methane selectivity 5.73 10 5.41 10 5.11 10 3.68 10
2 2 2 2b
Final Score 2/6 0/6 0/6 4/6
Extrapolated by the linear regression. This is the intercept of the correlated line. Best result amongst all catalysts.
Experimental data on liquid production rates (Table 2) indicated that both C.8 and C.9 catalysts had approximately the same liquid production rates as catalyst C.1. Although addition of lanthanum beyond 2.7 wt% is expected to slightly improve catalyst performance, there is a trade-off between lanthanum doping costs, which rise with increased loading, and the incremental increase in catalyst lifetime obtained. Fig. 6 shows the absolute value of the slope of the activity versus time-on-stream as a function of lanthanum loading. After an initial rapid fall, a lower limit plateau is reached, indicating that additional lanthanum loading would not result in any meaningful change in the slope of activity loss. Although data presented in Tables 1 and 2 indicated that catalyst C.9 had the best performance, nonetheless, Fig. 6 introduced some doubts into this justification regarding higher lanthanum doping. The most significant change in the performance of the catalysts occurred when the La loading was increases from 0.61 to 1.1 wt%. There are some postulations regarding the role of lanthanum doping on the deactivation rate of cobalt-based FTS catalysts with alumina support. The strongest and most logical explanation is the
production of a mixed oxide phase, comprising aluminium, lanthanum and oxygen, which is highly resistant toward cobalt spinel formation. Cobalt can react with alumina support leading to the formation of a highly stable cobalt aluminate phase which could not be reduced to metallic cobalt under regular reduction conditions. The mixed oxide phase seems to be more resistant toward deactivation by cobalt spinel formation. The results of the present study indicated that this effect was limited to a lanthanum loading of 1.1 wt% and beyond this value no significant changes were seen. 3.2. Catalyst characterisation with N2 porosimetry and TPR techniques BET surface areas, pore volumes, and mean pore diameters were obtained from N2 porosimetry analysis and are reported in Table 3
Table 2 TOS, liquid production rate and ICP results of the prepared catalysts. Catalyst
TOS (h)
Liquid production rate (g/g.cat day)a
Co loading (wt%)
La loading (wt%)
C.1 C.7 C.8 C.9
217 310 522 530
4.99 4.47 4.97 4.66
14.0 14.8 15.2 15.5
0.61 0.0 1.1 2.7
a This liquid phase products only appeared in hot and cold traps, comprise C5+ hydrocarbons and are inherently a mixture of liquid and solid wax. C2–C7 hydrocarbons also appeared in the gas phase products leaving the cold trap. In addition, traces of water soluble hydrocarbons existed in aqueous phase of hot and cold traps.
Fig. 6. Effect of lanthanum doping (wt%) on the absolute value of the slope of activity versus time-on-stream (h 1).
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
709
Table 3 N2 porosimetry results for catalysts and the alumina support. Catalyst
BET surface area (m2/g)
Pore volume (cm3/g)
Mean pore diameter (Nm)
C.1 C.7 C.8 C.9 g-Al2O3
171 180 172 171 271
0.40 0.39 0.37 0.36 0.70
8.3 8.7 8.6 8.4 10.3
for all catalysts as well as for the alumina support. As expected, the catalyst surface area and pore volume decreased with increasing numbers of calcination steps due to sintering. Catalysts C.1, C.8, and C.9 experienced two calcinations; but catalyst C.7 experienced only one calcination step. Comparison of the specific surface areas and pore volumes for different catalysts with those for the alumina support indicated that the loss in pore volume and surface area due to sintering was most pronounced during the first calcination step. The loss in surface area and pore volume due to the additional calcination was only marginal. Catalyst surface area and pore volume also slightly decreased with increasing lanthanum loading. The effect of various parameters on the TPR analysis of aluminasupported cobalt FTS catalysts has been reported in previous studies [6,28,29]. Here the focus was on the location of peaks for cobalt species present in the catalysts, and on the total amount of H2 consumed, as a measure of the quantity of cobalt available for reaction. Any low temperature peak below 500 K would be due to cobalt nitrate species remaining on the catalyst surface after calcination. The next medieval temperature peaks (after 500 K) belonged to the two step reduction of the Co3O4 phase present in the catalyst pores to metallic cobalt. In the first step, the Co3O4 species was reduced to the CoO after which this material reacted with hydrogen to result in the metallic cobalt [26,27]. It should be noted that larger crystals shall reduce faster than the smaller ones which are highly of dispersed cobalt oxide phase [6]. Activity promoters such as ruthenium can play a role in making these crystals reduce earlier. The final peak above 1000 K was attributed to a cobalt aluminate phase which is very hard to reduce [6,7,28,29]. Data presented in Table 4 indicated that most of the hydrogen consumption form TPR analysis was from the second peak due to small and finely dispersed crystals. Note that there are some differences in the hydrogen consumption data between catalysts C.1 and catalysts C.7–C.9, with the former displaying lower values. Catalyst C.1 had the lowest cobalt loading and the least total hydrogen consumption. Better dispersion of cobalt over the surface of alumina is one factor which would lead to higher H2 consumption observed for C.9. In the TPR diagram for catalysts C.7–C.9, an early peak from cobalt nitrate was observed, although the H2 consumption for this peak was not significant. Incorporation of lanthanum most probably caused the high temperature peak of the cobalt aluminate to be shifted to lower temperatures. This phenomenon was rather more exaggerated at higher lanthanum loadings, such that for the highest lanthanum loaded sample this peak might have either slightly remained or really diminished. This
issue was evidenced by looking at Fig. 7. The C.7 catalyst (i.e., without any lanthanum doping) showed a high temperature peak which gradually diminished through the C.1, C.8 and C.9 catalysts. Although it is believed that total amount of H2 consumed in TPR in particular (and of course, other peaks in general) is directly proportional to the average activity of a given catalyst however, other factors are also of considerable importance. For instance, in the present results, the trend of increase in H2 consumption did not completely correspond with those of average activity change, but the discrepancy might be justified regarding the interaction of different parameters including; (i) the total amount of loaded cobalt and lanthanum, (ii) aluminate phase formation and fraction of total cobalt which is consumed by the aluminate and (iii) share of large and small clusters of cobalt. All of these parameters and their impacts on the apparent activity were shown in Table 5 where the ranking of different effects presented. This table clearly indicated the importance of cobalt and lanthanum loading amongst all. The trend of activity changes thoroughly fitted with that of lanthanum doped amount. Although the H2 consumption is directly a sign of total available cobalt for the reaction, but this cobalt phase should be of appropriate size to be suitable for the reaction [21]. Also it should not be from aluminate phase (i.e., with the high temperature peak symptom). The main reason for the present results observed is that all of the cobalt species available on the catalysts surfaces were detected by the TPR method however; just part of these remained active during a long period of reactor tests. A possible rational for the recent phenomena is oxidation of small clusters and aluminate phase formation which would be prevented by doping of lanthanum. Thus, for better correlation between the total H2 consumption and average activity, one should be reminded of the effects of lanthanum and total loaded cobalt. This issue may completely justify the present TPR results and their relation with the average activity.
Table 4 H2 consumption for different catalysts from TPR analysis.
Table 5 Ranking different affecting factors in the average activity.
Fig. 7. TPR diagram for catalysts C.1–C.3 and C.7–C.9.
Catalyst
First peak temp. (K)
Second peak temp. (K)
H2 from first peak (mmole/g)
H2 from second peak (mmole/g)
Total H2 (mmole/g)
Second Total H2 Catalyst Loaded Loaded First Average peak H2 name Co La peak H2 consumption activity consumption consumption
C.1 C.7 C.8 C.9
572 579 596 599
805 887 906 783
4.58 6.04 7.08 6.16
8.84 12.64 12.18 11.92
13.42 18.68 19.26 18.08
C.1 C.7 C.8 C.9
4 3 2 1
3 4 2 1
4 3 1 2
4 1 2 3
4 2 1 3
3 4 2 1
710
M.R. Hemmati et al. / Journal of the Taiwan Institute of Chemical Engineers 43 (2012) 704–710
Nonetheless, looking at this table, still a point remains to be ratified in it, namely, the justification for less H2 consumption for C.9 catalyst which has the most amount of loaded cobalt. The rational for this is found in Table 4. Although C.8 presented the most H2 consumption; but this difference was not so large (i.e., 3% for the C.7 and 6% for the C.9 materials based upon the C.8 catalyst). Nonetheless, the point is that for the C.9 material, the high temperature peak has most probably shifted to a lower temperature (distancing from the cobalt aluminate peak temperature) which is a significant factor for the C.9 catalysts and could justify its appropriate average activity.
[10]
[11]
[12]
[13]
4. Conclusions
[14]
To investigate the effects lanthanum doping on the physiochemical behaviour and durability of FTS catalysts, four catalyst samples were made and evaluated. All samples were supported on g-alumina with nearly the same chemical formulation expect for the amount of doped lanthanum. The prepared catalysts were subsequently analysed by ICP, TRR and N2 porosimetry techniques and their performance was tested in a FTS reactor. The objective was to prepare catalysts which were resistant to deactivation. Results showed that increasing the amount of lanthanum led to superior performance in terms of catalyst primary activity, average activity during time, activity change with time-on-stream, average selectivity, and selectivity change with time-on-stream. Lanthanum doping alters the chemical state of the support by forming mixed oxides with alumina that are more resistant to oxidation. This is reinforced by TPR results as the high temperature peak due to the cobalt aluminate phase diminished with increased loading of lanthanum. However, an increase in lanthanum doping beyond 1.1 wt% only marginally improved the deactivation resistance of the catalyst.
[15]
[16]
[17]
[18]
[19]
[20]
[21]
References [22] [1] Iglesia E. Design, synthesis, and use of cobalt-based Fischer-Tropsch synthesis catalysts. Applied Catalysis A General 1997;161:59–78. [2] Li T, Yang YZ, Wan H, An X, Zhang C, Xiang H, Li Y. Effect of sulfate on an iron manganese catalyst for Fischer-Tropsch synthesis. Journal of Natural Gas Chemistry 2007;16:354–62. [3] Sehabiague L, Lemoine R, Behkish A, Heintz YJ, Sanoja M, Oukaci R, Morsi BI. Modeling and optimization of a large-scale slurry bubble column reactor for producing 10,000 bbl/day of Fischer-Tropsch liquid hydrocarbons. Journal of the Chinese Institute of Chemical Engineers 2008;39:169–79. [4] Dalai AK, Davis BH. Fischer-Tropsch synthesis: a review of water effects on the performances of unsupported and supported Co catalysts. Applied Catalysis A General 2008;348:1–15. [5] Li J, Zhan X, Zhang Y, Jacobs G, Das T, Davis BH. Fischer-Tropsch synthesis: effect of water on the deactivation of Pt promoted Co/Al2O3 catalysts. Applied Catalysis A General 2002;228:203–12. [6] Hilmen AM, Schanke D, Hanssen KF, Holmen A. Study of the effect of water on alumina supported cobalt Fischer-Tropsch catalysts. Applied Catalysis A General 1999;186:169–88. [7] Krishnamoorthy S, Tu M, Ojeda MP, Pinna D, Iglesia E. An investigation of the effects of water on rate and selectivity for the Fischer-Tropsch synthesis on cobalt-based catalysts. Journal of Catalysis 2002;211:422–33. [8] Tsakoumis NE, Ronning M, Borg Ø, Rytter E, Holmen A. Deactivation of cobalt based Fischer-Tropsch catalysts: a review. Catalysis Today 2010;154:162–82. [9] Saib AM, Moodley DJ, Ciobıˆca IM, Hauman MM, Sigwebela BH, Weststrate CJ, Niemantsverdriet JW, Van de Loosdrecht J. Fundamental understanding of
[23]
[24]
[25]
[26]
[27]
[28]
[29]
deactivation and regeneration of cobalt Fischer-Tropsch synthesis catalysts. Catalysis Today 2010;154:271–82. Zhou W, Chen JG, Fang KG, Sun YH. The deactivation of Co/SiO2 catalyst for Fischer-Tropsch synthesis at different ratios of H2 to CO. Fuel Processing Technology 2006;87:609–16. Hemmati MR, Kazemeini M, Zarkesh J, Khorasheh F. Effect of preparation parameters on catalytic performance of alumina supported Co catalysts utilized in Fischer-Tropsch synthesis. In: Proc. of the 7th European congress of chemical engineering and 19th international congress of chemical and process engineering; CHISA 2010, Prague, Czech Republic, August 2010. Borg Ø, Eri S, Blekkan EA, Storsæter S, Wigum H, Rytter E, Holmen A. FischerTropsch synthesis over g-alumina-supported cobalt catalysts: effect of support variables. Journal of Catalysis 2007;248:89–100. Rane S, Borg Ø, Yang J, Rytter E, Holmen A. Effect of alumina phases on hydrocarbon selectivity in Fischer-Tropsch synthesis. Applied Catalysis A General 2010;388:160–7. Jacobs G, Das TK, Li J, Luo M, Patterson PM, Davis BH. Fischer-Tropsch synthesis: influence of support on the impact of co-fed water for cobalt-based catalysts. Catalysts and Catalysis 2007;217–53. Martinez A, Prieto G, Rollan J. Nanofibrous g-Al2O3 as support for Co-based Fischer-Tropsch catalysts: pondering the relevance of diffusional and dispersion effects on catalytic performance. Journal of Catalysis 2009;263:292–305. Karimi A, Nakhaei Pour A, Torabi F, Hatami B, Tavasoli A, Alaei M, Irani M. Fischer-Tropsch synthesis over ruthenium-promoted Co/Al2O3 catalyst with different reduction procedures. Journal of Natural Gas Chemistry 2010;19: 503–8. Burakorn T, Panpranot J, Mekasuwandumrong O, Chaisuk C, Praserthdam P, Jongsomjit B. Characterization of cobalt dispersed on the mixed nanoscale alumina and zirconia supports. Journal of Materials Processing Technology 2008;206:352–8. Prieto G, Martı´nez A, Concepcio´n P, Moreno-tost R. Cobalt particle size effects in Fischer-Tropsch synthesis: structural and in situ spectroscopic characterisation on reverse micelle-synthesised Co/ITQ-2 model catalysts. Journal of Catalysis 2009;266:129–44. Borg Ø, Dietzel PDC, Spjelkavik AI, Tveten EZ, Walmsley JC, Diplas S, Eri S, Holmen A, Rytter E. Fischer-Tropsch synthesis: cobalt particle size and support effects on intrinsic activity and product distribution. Journal of Catalysis 2008;259:161–4. Radstakea PB, den Breejenb JP, Bezemerb GL, Bitterb JH, de Jong KP, Frsetha V, Holmen A. On the origin of the cobalt particle size effect in the Fischer-Tropsch synthesis. Natural Gas Conversion 2007;VIII:85–90. Bezemer GL, Bitter JH, Kuipers HPCE, Oosterbeek H, Holewijn JE, Xu X, Kapteijn F, van Dillen AJ, de Jong KP. Cobalt particle size effects in the Fischer-Tropsch reaction studied with carbon nanofiber supported catalysts. Journal of American Chemical Society 2006;128:3956. van Steen E, Viljoen EL, van de Loosdrecht J, Claeys M. Evaluation of molybdenum-modified alumina support materials for Co-based Fischer-Tropsch catalysts. Applied Catalysis A General 2008;335:56–63. Bedel L, Roger AC, Rehspringer JL, Zimmermann Y, Alain Kiennemann A. La(1 y)Co0.4Fe0.6O3-d perovskite oxides as catalysts for Fischer-Tropsch synthesis. Journal of Catalysis 2005;235:279–94. Wang T, Ding Y, Lu¨ Y, Zh H, Lin L. Influence of lanthanum on the performance of Zr-Co/activated carbon catalysts in Fischer-Tropsch synthesis. Journal of Natural Gas Chemistry 2008;17:153–8. Cai Z, Li J, Liew K, Hu J. Effect of La2O3-dopping on the Al2O3 supported cobalt catalyst for Fischer-Tropsch synthesis. Journal of Molecular Catalysis A General 2010;330:10–7. Jiao G, Ding Y, Zhu H, Li X, Li J, Lin R, Dong W, Gong L, Pei Y, Lu Y. Effect of La2O3 doping on syntheses of C1–C18 mixed linear a-alcohols from syngas over the Co/AC catalysts. Applied Catalysis A General 2009;364:137–42. Gruver V, Zhan X, Engman J, Robota HJ, Suib SL, Polverjan M. Deactivation of a Fischer-Tropsch catalyst through the formation of cobalt carbide under laboratory slurry reactor conditions. Preprints—American Chemical Society. Division of Petroleum Chemistry, Fischer-Tropsch: Materials, Theories, and Practice. Symposium. 2004;49(2):192–94. Su H, Zeng S, Dong H, Du Y, Zhang Y, Hu R. Pillared montmorillonite supported cobalt catalysts for the Fischer-Tropsch reaction. Applied Clay Science 2009;46: 325–9. Zeng S, Du Y, Su H, Zhang Y. Promotion effect of single or mixed rare earths on cobalt-based catalysts for Fischer-Tropsch synthesis. Catalysis Communication 2011;13:6–9.