Efficient and Reusable Palladium Catalysts Supported on Activated

Feb 25, 2005 - Pham Dung Tien, Tetsuya Satoh,* Masahiro Miura, and Masakatsu Nomura*. Department of Applied Chemistry, Faculty of Engineering, Osaka ...
0 downloads 0 Views 467KB Size
Energy & Fuels 2005, 19, 731-735

731

Efficient and Reusable Palladium Catalysts Supported on Activated Carbon Fibers for Dehydrogenation of Tetrahydronaphthalene Pham Dung Tien, Tetsuya Satoh,* Masahiro Miura, and Masakatsu Nomura* Department of Applied Chemistry, Faculty of Engineering, Osaka University, Suita, Osaka 565-0871, Japan Received September 10, 2004. Revised Manuscript Received December 29, 2004

Catalytic hydrogen evolution from neat tetrahydronaphthalene occurs under normal liquidphase conditions effectively, using palladium catalysts supported on activated carbon fibers (ACFs). ACFs that have relatively high surface areas seem to be suitable for use as supports. Although the co-product, naphthalene, accumulated in the reaction system retards the reaction, it can be removed easily, so that the initial catalyst activity recovers almost completely. Thus, the catalysts stably exhibit high activity in the reuse cycles at least five times.

Introduction The reversible reaction pair, dehydrogenation-hydrogenation of carbocyclic compounds, is now recognized to be one of the potential candidate methods to safely supply and store hydrogen for PEM fuel cell,1,2 as well as for energy storage systems.3 Of the reaction pair, the dehydrogenation usually requires the use of effective transition-metal catalysts under rather severe conditions for efficient operation, because of its endothermic nature. Besides the thermodynamic problem, site-blocking by the corresponding dehydrogenated aromatic coproducts, which suppresses the approach of the starting saturated substrates to the catalysts, may intervene. Recently, several groups have reported that the dehydrogenation of cyclohexane, methylcyclohexane, and decahydronaphthalene can be conducted effectively in the presence of platinum catalysts under “liquidfilm”1 or “wet-dry multiphase” conditions,2 in which the substrate/catalyst ratio must be kept low. Meanwhile, it is conceivable that tetrahydronaphthalene (1) may be used as an effective hydrogen source, because it has an aromatic ring that may allow advantageous adsorption on catalyst. Indeed, we reported that 1 can be dehydrogenated smoothly, even under normal liquid-phase conditions with a high substrate/catalyst ratio, and palladium catalysts supported on activated * Author to whom correspondence should be addressed. E-mail address: [email protected]. (1) (a) Hodoshima, S.; Arai, H.; Saito, Y. Int. J. Hydrogen Energy 2003, 28, 197-204. (b) Hodoshima, S.; Arai, H.; Takaiwa, S.; Saito, Y. Int. J. Hydrogen Energy 2003, 28, 1255-1262. (c) Shinohara, C.; Kawakami, S.; Moriga, T.; Hayashi, H.; Hodoshima, S.; Saito, Y.; Sugiyama, S. Appl. Catal., A 2004, 266, 251-255. (d) Hodoshima, S.; Saito, Y. J. Chem. Eng. Jpn. 2004, 37, 391-398. (2) (a) Kariya, N.; Fukuoka, A.; Ichikawa, M. Appl. Catal., A 2002, 233, 91-102. (b) Kariya, N.; Fukuoka, A.; Utagawa, T.; Sakuramoto, M.; Goto, Y.; Ichikawa, M. Appl. Catal., A 2003, 247, 247-259. (3) (a) Maria, G.; Marin, A.; Wyss, C.; Muller, S.; Newson, E. Chem. Eng. Sci. 1996, 51, 2891-2896. (b) Newson, E.; Haueter, T. H.; Hottinger, P.; von Roth, F.; Scherer, G. W. H.; Schucan, T. H. H. Int. J. Hydrogen Energy 1998, 23, 905-909. (c) Scherer, G. W. H.; Newson, E.; Wokaun, A. Int. J. Hydrogen Energy 1999, 24, 1157-1169.

Table 1. Characterization Data for Activated Carbon Fibers (ACFs) ACF

BET surface area (m2/g)

pore volume (cm3/g)

mean pore radius (nm)

ACF-1 ACF-2 ACF-3

957 1776 2020

0.40 0.87 1.13

0.84 0.98 1.10

carbon fibers (ACFs) exhibit good activity.4 Thus, for this reaction, less-expensive palladium can be used effectively in place of platinum, whereas the use of the latter metal is essential for the dehydrogenation of cyclohexanes and decahydronaphthalene.1-3 However, even using such active palladium catalysts, the rate of hydrogen evolution decreased gradually as the reaction progressed. There seem to be two major factors for the aforementioned rate decrease. One is for self-retardation by the product(s), hydrogen and/or naphthalene (2), formed through the dehydrogenation. The other is for irreversible degeneration of the used catalyst itself, due to sintering, poisoning, or leaching of metal. Consequently, we have investigated the dehydrogenation of 1 further to make the influence of these factors clearer. Reuse of the Pd/ACF catalysts then is undertaken. Experimental Section Catalyst Preparation. Three types of ACFssdenoted as ACF-1, ACF-2, and ACF-3swere supplied by Osaka Gas Co., Ltd. The Brunauer-Emmett-Teller (BET) surface area, pore volume, and mean pore diameter for each are shown in Table 1. A conventional impregnation method was applied to a HCl aqueous solution containing PdCl2 by stirring at 80 °C for 3 h, followed by a reduction treatment in water under hydrogen at room temperature for 16 h.5 The palladium contents, as determined by inductively coupled plasma-atomic emission (4) Tien, P. D.; Morisaka, H.; Satoh, T.; Miura, M.; Nomura, M.; Matsui, H.; Yamaguchi, C. Energy Fuels 2003, 17, 658-660. (5) Mozingo, R. In Organic Syntheses, Collective Volume 3; Wiley: New York, 1955; pp 685-690.

10.1021/ef040083v CCC: $30.25 © 2005 American Chemical Society Published on Web 02/25/2005

732

Energy & Fuels, Vol. 19, No. 3, 2005

Tien et al.

Figure 2. Time course of the turnover number (TON, expressed as mmol H2/mmol Pd) in the dehydrogenation of 1, using (0) Pd/ACF-1, (O) Pd/ACF-2, and (+) Pd/ACF-3 as catalysts. Reaction conditions: 8 cm3 of 1 and 0.004 mmol of palladium catalyst, under nitrogen (1 atm) at 300 °C.

Figure 1. Experimental apparatus used for the dehydrogenation of tetrahydronapththalene (1). spectroscopy (ICP-AES), and BET surface areas of Pd/ACF-1, Pd/ACF-2, and Pd/ACF-3 prepared by this method were 0.84, 0.71, and 0.70 wt % and 574, 1629, and 1221 m2/g, respectively. Pd/ACF-1 and Pd/ACF-3 were observed before and after use by a transmission electron microscopy (TEM) system (Hitachi model FE-TEM HF-2000), using an accelerating voltage of 200 kV. Dehydrogenation of Tetrahydronaphthalene (1). The volume of evolved hydrogen was measured by a gas buret (2 L). In a typical experiment, the reaction of 1 (8 cm3) was conducted in the presence of a Pd/ACF catalyst (palladium content of 0.004 mmol) at 300 °C (the heater temperature) under a nitrogen atmosphere for 90-540 min, using an apparatus that is shown in Figure 1. After cooling, n-hexadecane (ca. 800 mg, as an internal standard) was added to the reaction mixture, and then the amounts of produced 2 and unconsumed 1 were determined by gas chromatography (GC) analysis.

Results and Discussion First, neat 1,2,3,4-tetrahydronaphthalene (1, 8 cm3) was treated using three palladium catalysts that were supported on ACFssdenoted as Pd/ACF-1, Pd/ACF-2, and Pd/ACF-3 (net amount of palladium for each run: 0.004 mmol)sat 300 °C (external temperature) in a batch reactor under nitrogen (Figure 2). In each case, a dehydrogenation of 1 was conducted to produce hydrogen and 2.6 During the reaction, the volume of evolved hydrogen was monitored by a gas buret. In contrast to the reaction under “liquid-film”1 or “wet-dry multiphase” conditions,2 the present reaction was conducted under normal liquid-phase conditions, in which the catalyst was sunken in the liquid completely. Despite the higher external heating temperature, that of the liquid-phase was kept between 207 and 218 °C (the boiling points of 1 and 2, respectively). Among the three catalysts, Pd/ACF-3 gave the highest turnover number (TON) for hydrogen production, which reached 1.2 × 104 (6) It was confirmed that ACFs themselves could not induce the dehydrogenation.

Figure 3. Time course of TON (expressed as mmol H2/mmol Pd) in the dehydrogenation of 1 with Pd/ACF-3 at 300 °C under (b) nitrogen and (O) hydrogen atmospheres.

after 540 min. [Note: TON ) amount of evolved hydrogen (mmol)/bulk Pd amount added (mmol).] The reaction using Pd/ACF-2 proceeded smoothly with a comparable reaction rate up to initial 180 min, although the accumulated TON for 540 min is somewhat lower.7 In the case using Pd/ACF-1, both the initial rate and the final TON were considerably low. In all cases, the reaction efficiency decreased gradually as the reaction progressed.8 It is possible that accumulation of the products (hydrogen and 2), in the reaction system induces the decrease in the reaction rate. Of the two products, the effect of hydrogen was first examined. Thus, the reaction of 1 was conducted using Pd/ACF-3 under hydrogen instead of nitrogen. As shown in Figure 3, however, the time course of TON under hydrogen was determined to be very similar to that under nitrogen. The results indicate that the partial pressure of hydrogen in the reaction system is not the significant factor that retards the dehydrogenation, at least for the reaction period that has been measured. Next, the effect of another product (2) was investigated. In our previous paper, we reported that the dehydrogenation was completely inhibited by the addition of 10% of 2 to substrate 1 before starting the (7) It is noted that Pt/ACF-2 (platinum content, 0.72 wt %; BET surface area, 1686 m2/g) was less active than Pd/ACF-2. The TONs for 90 min were 2300 and 3700, respectively. (8) As noted in the Experimental Section, the surface areas of the ACFs decreased after the metal impregnation, and it was significant with ACF-1 and ACF-3 (ca. 40% reduction). Nevertheless, Pd/ACF-3 gave the best result. The reason for this is not clear at the present stage.

Dehydrogenation of Tetrahydronaphthalene

Figure 4. Time course of the amount of 2 in the treatment of (O) neat 1, (b) a 9:1 mixture, and (]) a 7:3 mixture of 1 and 2, using Pd/ACF-3 at 300 °C under a nitrogen atmosphere. (Total amount of 2 expressed as mmol H2 evolved/2 + mmol 2 added.)

reaction.4 In that case, a purchased sample of 2 was used after recrystallization from ethanol. Commercially produced 2 is known to be contaminated by sulfurcontaining impurities such as benzothiophene.9 It is possible that the impurities, which could not be removed by recrystallization, acted as catalyst poison. In the present work, therefore, we reinvestigated the influence using sulfur-free 2, which was obtained by the dehydrogenation of 1. Actually, this is known to be a method to obtain highly purified 2.9 As expected, the addition of 10% or 30% of 2 to 1 did not inhibit but rather retarded the reaction, to a reasonable extent. Figure 4 shows the results using Pd/ACF-3. The profile in each added case is comparable to that of the reaction of pure 1 after 10% or 30% conversion. This indicates that the rate retardation observed in Figure 2 is caused, at least in part, by the accumulation of 2, which may lead to site blocking on the catalyst, to suppress the approach of 1. It was observed that the catalyst activity can be readily recovered only by the removal of 2, by rinsing the catalyst with hexane. Thus, after the reaction of 1 using Pd/ACF-3 for 90 min, the mixture of unconsumed 1 and produced 2 was removed via a cannula, leaving the catalyst in the reactor. After the catalyst was washed by hexane at room temperature twice (20 cm3, 15 min for each)10 and dried under a N2 flow at 100 °C for 15 min, substrate 1 (8 cm3) was added to start the next reaction. By this treatment, the catalyst exhibited almost the same activity as that of the fresh catalyst. Even after the process was repeated five times, the high reaction efficiency was maintained and the accumulated TON for 540 min reached ca. 2.2 × 104 (Figure 5). Pd/ ACF-2 also showed comparable performance to give a similar accumulated TON (Figure 6). However, Pd/ ACF-1 did not maintain its catalytic activity. In the case of Pd/ACF-1, the TON accumulated for each 90 min gradually decreased, as shown in Figure 7 (first run: TON ) 3.0 × 103/90 min, sixth run: TON ) 2.3 × 103/ 90 min). After the sixth run, it was confirmed that 99.5 wt % of the catalyst remained in the reactor, and that (9) (a) Sato, Y.; Yamamoto, Y.; Kanda, N.; Yamada, M.; Suda, Y.; Tanimichi, N. Jpn. Kokai Tokkyo Koho 1994, JP 06048967. (b) Yamada, M.; Suda, Y.; Kanda, N.; Tanimichi, J. Jpn. Kokai Tokkyo Koho 1995, JP 07133237. (10) In the second fluid, 2 could no longer be detected by gas chromatography.

Energy & Fuels, Vol. 19, No. 3, 2005 733

Figure 5. Time course of TON (expressed as mmol H2/mmol Pd) in the dehydrogenation of 8 cm3 of 1, using fresh and recovered Pd/ACF-3. Each run was performed at 300 °C under a nitrogen atmosphere for 90 min. Open circles (O) represent the time course of TON; dotted line (- - -) shows accumulated TON.

Figure 6. Time course of TON (expressed as mmol H2/mmol Pd) in the dehydrogenation of 8 cm3 of 1, using fresh and recovered Pd/ACF-2. Each run was performed at 300 °C under a nitrogen atmosphere for 90 min. Open circles (O) represent the time course of TON; dotted line (- - -) shows accumulated TON.

Figure 7. Time course of TON (expressed as mmol H2/mmol Pd) in the dehydrogenation of 8 cm3 of 1, using fresh and recovered Pd/ACF-1. Each run was performed at 300 °C under a nitrogen atmosphere for 90 min. Open circles (O) represent the time course of TON; dotted line (- - -) shows accumulated TON.

the palladium content of the used catalyst was essentially the same as that of the fresh catalyst. Consequently, the deactivation of Pd/ACF-1 may be attributed not to leaching but to the growth of the size of the metal particle, which has been previously observed in the platinum-catalyzed dehydrogenation of cyclohexane.2b In harmony with this, the representative comparison of TEM micrographs for fresh and spent samples of Pd/

734

Energy & Fuels, Vol. 19, No. 3, 2005

Tien et al.

Figure 8. Transmission electron microscopy (TEM) images for (a) fresh Pd/ACF-1, (b) used Pd/ACF-1, (c) fresh Pd/ACF-3, and (d) used Pd/ACF-3.

ACF-1 and Pd/ACF-3 showed that the size of the palladium particle on Pd/ACF-1 increases more conspicuously during the six runs than that on Pd/ACF-3 (Figure 8). Thus, particles >100 nm in size could be observed only on the used Pd/ACF-1 (Figure 8b) among the four samples, at least within the ranges observed. The support ACF-1, which possesses the lowest surface area among the three supports, may readily allow metal aggregation on it. However, the details are not clear at the present stage. ACF has recently been regarded as a promising catalyst support,11 because of its advantageous characteristics, in comparison to traditional activated carbons, such as (a) a morphological network formed by short micropores with a narrow size distribution, ensuring (11) For example: (a) Matatov-Meytal, Y.; Sheintuch, M. Appl. Catal., A 2002, 231, 1-16. (b) deMiguel, S. R.; Vilella, J. I.; Jablonski, E. L.; Scelza, O. A.; Salinas-Martinez de Lecea, C.; Linares-Solano, A. Appl. Catal., A 2002, 232, 237-246. (c) Joannet, E.; Horny, C.; KiwiMinsker, L.; Renken, A. Chem. Eng. Sci. 2002, 57, 3453-3460. (d) Landau, M. V.; Kogan, S. B.; Tavor, D.; Herskowitz, M.; Koresh, J. E. Catal. Today 1997, 36, 497-510. (e) Suzuki, M. Carbon 1994, 32, 577586.

Figure 9. Time course of TON (expressed as mmol H2/mmol Pd) in the dehydrogenation of 8 cm3 of 1, using fresh and recovered commercial Pd/AC (5 wt %). Each run was performed at 300 °C under a nitrogen atmosphere for 90 min. Open circles (O) represent the time course of TON; dotted line (- - -) shows accumulated TON.

fast adsorption/desorption, and (b) a fibrous form that enables easy separation from the liquid phase without a time-consuming filtration step. For comparison, the

Dehydrogenation of Tetrahydronaphthalene

same treatment for the catalyst reuse was applied to a purchased palladium catalyst supported on activated carbon (Pd/AC) powder12 (Figure 9). As reported in the previous paper,4 the catalyst showed low activity, and it diminished rapidly after two cycles of reuse. It was so difficult to keep the powder catalyst in the reactor through the reuse processes that 11 wt % of the catalyst was totally lost. The observed decrease of activity was, by far, more severe than that expected from the catalyst loss. Therefore, irreversible degeneration of the catalyst itself (such as sintering) would also occur in this case.

Energy & Fuels, Vol. 19, No. 3, 2005 735

carbon fibers (ACFs) that possess high specific surface areas are adequate as the supports. Although coproduced naphthalene (2) retards the reaction, it can be removed easily by a simple process. Consequently, the catalysts are regenerated to be ready for their reuse. In addition, the fibrous Pd/ACF catalysts are known to be applicable to a fixed-bed flow reactor without any modification and may have advantages such as low resistance to the passage of fluid and high fluid permeability.11 Accordingly, a further study of the present reaction with such a type of reactor for continuous hydrogen production will be conducted in our laboratory.

Conclusion In summary, we have shown that the hydrogen evolution from tetrahydronaphthalene (1) proceeds smoothly under normal liquid-phase conditions in the presence of supported palladium catalysts. Activated (12) Pd/AC (5 wt %) was purchased from Wako Pure Chemicals.

Acknowledgment. We thank Dr. Hisaji Matsui and Dr. Chiharu Yamaguchi, both of Osaka Gas, for supplying the activated carbon fibers and for helpful discussion. EF040083V