Selective ring-opening of methylcyclopentane on platinum-based bimetallic catalysts

Applied Catalysis A: General 369 (2009) 104–112

Contents lists available at ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Selective ring-opening of methylcyclopentane on platinum-based bimetallic catalysts P. Samoila, M. Boutzeloit, C. Especel, F. Epron *, P. Mare´cot Laboratoire de Catalyse en Chimie Organique, UMR 6503, CNRS – Universite´ de Poitiers, 40 avenue du recteur Pineau, 86022 Poitiers Cedex, France

A R T I C L E I N F O

A B S T R A C T

Article history: Received 9 June 2009 Received in revised form 23 July 2009 Accepted 7 September 2009 Available online 12 September 2009

Several monometallic catalysts supported on alumina were tested in methylcyclopentane ring-opening under pressure. Among the monometallic catalysts tested (Ru, Re, Rh, Pt and Ir), iridium and rhodium catalysts were the most active but iridium was by far the most selective in ring-opening (RO) products (2-methylpentane, 3-methylpentane and n-hexane), the formation of C1–C5 products being negligible, as already reported in the literature. Thereafter, platinum-based bimetallic catalysts supported on alumina were prepared by redox surface reaction in order to favor the metal–metal interaction. The aim was to obtain bimetallic catalysts leading to selectivity towards RO products similar to that of the iridium catalyst. Two types of modifiers were studied, namely (i) inactive species such as copper and germanium and (ii) active promoters for hydrogenolysis reactions, such as ruthenium and rhodium. It was shown that with inactive metals, the parent platinum catalyst undergoes mainly a dilution of its active phase. An increase of the activity is observed for the Pt-Ru/Al2O3 systems compared to the parent one, but also an increase of the C1–C5 products. On the contrary, the addition of Rh allowed us to increase the activity of the platinum parent catalyst and to obtain bimetallic catalysts with selectivity towards RO products similar to that obtained with iridium in same conditions. ß 2009 Elsevier B.V. All rights reserved.

Keywords: Methylcyclopentane Ring-opening Bimetallic catalysts Redox reactions

1. Introduction To answer to the increasingly strict regulations for the protection of environment, it is necessary to produce cleaner and cleaner automobile fuels. In particular, aromatic compounds must be reduced and replaced by paraffins, substituted in the case of the gasoline, or linear in the case of diesel oil. The production of such fuels supposes the total hydrogenation of the aromatic compounds in naphthenic compounds, then the selective opening of the latter to paraffins while avoiding the secondary reactions in particular deep hydrogenolysis leading to light hydrocarbons. The ring-opening (RO) of methylcyclopentane (MCP) (monocyclic naphthene) has been widely used as a catalytic probe for the particle size, especially for platinum catalysts [1]. The RO reaction of MCP produces nhexane (n-C6), 2-methylpentane (2MP) and 3-methylpentane (3MP). Nevertheless, this test reaction is usually performed at atmospheric pressure, and the distribution of products is discussed for conversions generally lower than 10%. In these conditions, Irbased catalysts are reported to be the most active and selective ones among the studied metals for MCP hydrogenolysis i.e. the C1–C5 cracking products are in a very small quantity.

* Corresponding author. Tel.: +33 5 49 45 48 32; fax: +33 5 49 45 37 41. E-mail address: fl[email protected] (F. Epron). 0926-860X/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2009.09.006

Among the studies concerning the monocyclic naphthene RO, few ones [2–7] report the catalytic behavior of monometallic catalysts under high pressure conditions, and only one is partly devoted to the MCP RO [2]. Moreover, the comparison of the catalytic performances of the catalysts is difficult since they are often evaluated in very different conditions. In this way, McVicker et al. [2] studied the selective ring-opening of one ring naphthenes to alkanes. The model molecules chosen were five- or six-memberedring naphthenes. The hydrogenolysis of MCP was studied at 28.5 bar on Pt, Ni, Ru and Ir catalysts supported on silica or alumina in order to obtain the best selectivity in ring-opening products. Depending on the catalyst studied, the space velocity and the temperature were varied. The most active and selective catalyst was 0.9 wt.% Ir/Al2O3, which allowed reaching 52% conversion at 275 8C with 99% selectivity in ring-opening products, whereas at the same temperature, C1–C5 cracking products were favored in the presence of 1.5 wt.% Ru/SiO2. Pt-based catalysts were less active in hydrogenolysis reactions than iridium and ruthenium, and consequently they were tested [2] at higher temperature (350 8C) and contact time. However, at this temperature, only 24% conversion was obtained but with an acceptable selectivity in ring-opening products, with only 0.3% of cracking products. The use of bimetallic catalysts may result in important changes in the activity and selectivity of MCP RO [8–14]. Contrary to monometallic catalysts, to the best of our knowledge, no publication reports the study of MCP ring-opening at high pressure

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112

in the presence of bimetallic catalysts. Thus, the aim of the present study is to modify a Pt/Al2O3 catalyst by addition of another metal in order to obtain Pt-based bimetallic catalysts with selectivity comparable to those of Ir catalysts for MCP hydrogenolysis in the same conditions. For that purpose, platinum-based bimetallic catalysts supported on alumina will be evaluated in a wide range of temperature but at the same space velocity and pressure. Two types of modifiers will be studied, namely (i) inactive species such as copper and germanium and (ii) active promoters for hydrogenolysis reactions, such as ruthenium and rhodium. Monometallic catalysts will be also evaluated in the same conditions to serve as reference for bimetallic catalysts. 2. Experimental 2.1. Catalyst preparation The support was a g-alumina (AXENS) with a specific surface area (BET method) of 215 m2 g1. The support was crushed and sieved, in order to retain particles with sizes between 0.25 and 0.40 mm, and calcined in flowing air at 450 8C for 4 h. Monometallic catalysts were prepared by impregnation of the precursor salt (H2PtCl6, H2IrCl6, RhCl3 or RuCl3) in an acidic solution (HCl 0.1 M). The Re/Al2O3 catalyst was prepared by impregnation of HReO4 in water. After evaporation of the solvent, the catalysts were activated by calcination at 300 8C (except for the Pt/Al2O3 catalyst prepared from H2PtCl6 that was calcined at 450 8C) under air for 4 h and reduced at 500 8C for 4 h. Bimetallic Pt-X/Al2O3 catalysts were prepared by deposition of the X modifier by a surface redox reaction, namely either catalytic reduction or refilling method. A given amount of the parent platinum catalyst was placed in a reactor, outgassed with nitrogen and then reduced under hydrogen (60 mL min1) at 500 8C (heating rate 10 8C min1) for 1 h. Then, the catalyst was cooled down to room temperature under nitrogen (refilling method) or hydrogen (catalytic reduction), and a volume of water was added to moisten it. For the catalytic reduction, a solution of hydrochloric acid (pH 1) containing a well-defined amount of the modifier salt (CuCl2 or GeCl4) was added and maintained in contact with the catalyst under hydrogen flow for 15 min. For the refilling method, the suspension of catalyst was at first maintained under hydrogen flow in order to preadsorb hydrogen on the platinum surface and then placed under nitrogen flow for 15 min to remove dissolved or weakly adsorbed hydrogen. Afterwards the solution of hydrochloric acid (pH 1) containing the modifier salt (RhCl3 or RuCl3) was added and let under nitrogen flow in contact with the catalyst for 1 min. Then, hydrogen was introduced in the reactor during 14 min. In both cases (catalytic reduction and the refilling method), the redox reaction between the Pt parent catalyst and the X modifier salt can be schematized as the following: 2Pt þ H2 ! 2Pt-H

(1)

nPt-H þ Xnþ ! Ptn -X þ nHþ

(2)

Whatever the deposition method, the solution was filtered out after reaction and the catalyst was dried overnight under N2 before to be reduced under hydrogen flow at 500 8C for 1 h (2 8C min1 heating rate). 2.2. Catalysts characterization The metal content was determined at the ‘‘Service central d’analyse’’ of the CNRS. The metal dispersion was calculated from the H2 chemisorption measurement, carried out in a pulse chromatographic system,

105

using a stoichiometry H/Metal equal to one. It is known [15] that this may lead to erroneous relative metallic dispersions but we have checked that the values of the particle size estimated by the hydrogen chemisorption measurements are in accordance with those determined from the TEM pictures. For all the catalysts, the H2 chemisorption measurement was carried out at room temperature. The metal particle size was calculated from the dispersion D (in %) using the classical hypothesis of cubic particles with one face in contact with the support (five faces exposed to the gases), and considering an equidistribution of the (1 0 0), (1 1 0) and (1 1 1) faces. The model reaction of cyclohexane dehydrogenation was performed in order to study the effect of the X modifier deposit on the Pt metallic phase. This reaction was carried out in a continuous flow reactor at 270 8C under atmospheric pressure on 10 mg of catalyst. Injection of cyclohexane was made using a calibrated motor-driven syringe. The partial pressures were 97 and 3 kPa for hydrogen and cyclohexane, and the corresponding flow rates 6 L h1 and 2 mL h1, respectively. The resulting space velocity was V H2 ¼ 100 cm3 min1 . Analysis of the reaction products was performed by gas chromatography with a flame ionization detector (Varian 3400X) on a HP-PLOT Al2O3 ‘‘KCl’’ column. The only detected product was benzene. TEM measurements were carried out with a Philips CM120 electron microscope operating at 120 kV with a resolution of 0.35 nm. 2.3. Catalytic test The catalysts were evaluated in methylcyclopentane hydrogenolysis under a total pressure of 28.5 bar in a temperature range from 200 to 475 8C. The experiments were carried out in a fixedbed, continuous reactor. Typically, 255 mg of catalyst were mixed with 245 mg of a-Al2O3 and placed inside the reactor (a stainless steel tube of 1.3 cm inner diameter) and the reactant mixture with a molar ratio H2/MCP = 7.5 was injected at a flow rate adjusted in order to obtain a weighted hourly space velocity (WHSV) of 12 h1. For Pt/Al2O3, an experiment was also performed at a WSHV of 6 h1 in order to increase the conversion in the same range of temperature. For the same reason, the tests were performed at a WSHV of 6 h1 for bimetallic catalysts with low activity in MCP conversion. In this case only, the WHSV is reported in the figure caption. These reaction conditions are similar to those chosen by McVicker et al. [2] for the study of the performances of Pt/Al2O3 catalysts. For ruthenium, nickel and iridium-based catalysts, these authors chose a WHSV equal to 30 h1. The methylcyclopentane flow was controlled using a calibrated motor-driven syringe. Effluent products were analyzed by an on-line chromatograph (Varian 3400) using a FID and equipped with a CP-Sil 5 capillary column. The conversion was varied by changing the reaction temperature: in a typical experiment, the reaction was started at the lower temperature, chosen as a function of the catalyst activity, in order to obtain a conversion lower than 10% and then the temperature was increased in steps of 25 8C up to a conversion of 80–90%. The duration of each step was 2 h and four measurements were performed at each temperature. Conversion, defined as the percentage of MCP converted, was determined as a function of temperature. The only reaction products were 2-methylpentane, 3-methylpentane, n-hexane, C1–C5 products and 2,3-dimethylbutane. Neither cyclohexane nor benzene was detected as reaction products. As the maximum yield in 2,3-dimethylbutane was 1%, this product was neglected and the selectivity of the catalysts for ring-opening (hydrogenolysis of only one endocyclic C–C bond) was estimated from the yield in C1–C5 products resulting from deep hydrogenolysis: the lowest the C1–C5 yield, the highest the RO products yield and then the highest the MCP ring-opening

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112

106

selectivity. The yield of C1–C5, defined as the percentage of MCP converted into C1–C5 products, was reported as a function of the conversion. In the following C1–C5 products will be also called ‘‘cracking’’ products. It was verified that the yield in C1–C5 obtained for a given conversion is independent of the reaction conditions by varying the WHSV instead of the temperature. 3. Results and discussion 3.1. Monometallic catalysts Various monometallic catalysts, based on platinum, rhodium, ruthenium, rhenium and iridium were prepared in order to obtain the same metal loading of 0.6 wt.%. The characteristics of these catalysts, i.e. the chlorine content, the metal dispersion and the metal particles size, are summarized in Table 1. Whatever the catalyst, the chlorine content is in the 1.0–1.2% range except for Pt/Al2O3, which contains 1.4 wt.% of chlorine. The Re/Al2O3 catalyst has no chlorine since it was prepared by impregnation of a nonchlorinated precursor salt in water. Depending on the catalyst, the values of the dispersion vary between 85% and 30% and consequently the particle size is between 1.1 and 3.1 nm. The performances of the monometallic catalysts were evaluated in methylcyclopentane ring-opening. Fig. 1 presents the evolution of MCP conversion versus temperature for the catalyst series. This figure shows that the iridium, rhodium and rutheniumbased catalysts are the most active. Even tested at a lower WHSV, i.e. a higher contact time, platinum is not active before 300 8C, whereas the conversion reached with iridium, rhodium and ruthenium-based catalysts is higher than 70% at this temperature. The activity at 300 8C of the catalysts calculated per mol of metal in the sample (Table 1) is as follows:

Fig. 2. Yield in C1–C5 products on (a) Ru/Al2O3 (~) and Re/Al2O3 (); (b) Pt/Al2O3 with WHSV = 6 h1 (^), Pt/Al2O3 (&), Ir/Al2O3 (^), and Rh/Al2O3 (~) catalysts as a function of the MCP conversion.

Ir=Al2 O3 > Rh=Al2 O3 > Ru=Al2 O3 > Re=Al2 O3  Pt=Al2 O3 Table 1 Characteristics and activity for MCP conversion of monometallic catalysts. Catalyst

wt.% Cl

Dispersion (%)

d¯ a (nm)

Ab (h1)

Pt/Al2O3 Rh/Al2O3 Ru/Al2O3 Re/Al2O3 Ir/Al2O3

1.4 1.0 1.2 – 1.2

75 56 41 30 85

1.3 1.5 2.0 3.1 1.1

47 2307 1807 1596 4583

A = number of converted MCP mol per hour per mol of metal in the sample. a Mean particle diameter calculated from the dispersion value. b Activity A for MCP conversion given at 300 8C.

This trend is in the same order as that obtained by McVicker et al. [2] who compared the activity of iridium, platinum, nickel and ruthenium also under pressure. The evolution of the C1–C5 products obtained in the presence of these catalysts is reported in Fig. 2 as a function of conversion. This figure shows that whatever the catalyst, the C1–C5 yield increases with the conversion, i.e. when the temperature is increased. The highest C1–C5 yields are obtained with the rhenium and ruthenium-based catalysts, which promote deep hydrogenolysis contrary to the Ir/Al2O3 catalyst. One can note that for platinum, the evolution of the yield in C1–C5 products as a function of conversion follows the same trend whatever the WHSV used for the experiment. The evolution observed in the presence of Rh/Al2O3 is similar to that observed with platinum. On iridium, less than 2% of cracking products are produced up to 90% of conversion. Consequently, the Ir/Al2O3 catalyst is the most selective in RO products, in accordance with the results reported in the literature [2,16–18]. The RO selectivity is in the following order: Ir=Al2 O3 > Pt=Al2 O3  Rh=Al2 O3 > Ru=Al2 O3 > Re=Al2 O3 :

Fig. 1. Conversion of MCP on Pt/Al2O3 with WHSV = 6 h1 (^), Pt/Al2O3 (&), Rh/ Al2O3 (~), Ru/Al2O3 (~), Re/Al2O3 (), and Ir/Al2O3 (^) as a function of temperature.

Product distribution is very often analyzed at low MCP conversion since high conversions should favor secondary reactions that could modify the product distribution [19–21]. In the present study, the RO product distribution and the molar ratio of nhexane to 3-methylpentane (n-C6/3MP) and 2-methylpentane to 3-methylpentane (2MP/3MP) are reported for each catalyst at various temperatures and then conversions (Table 2). The results show that, for conversions higher than 10%, the product distribution does not vary significantly as a function of the conversion. Gault and co-workers [18] showed that the ring-opening product distribution results from three possible mechanisms.

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112 Table 2 Yield in n-C6, 2MP, 3MP, n-C6/3MP and 2MP/3MP ratio on 0.6% M/Al2O3 and Pt-Rh/ Al2O3 catalysts. Catalyst

T (8C)

Conversion (%)

2MP (%)

3MP (%)

n-C6 (%)

n-C6/ 3MP

2MP/ 3MP

Ir/Al2O3

225 250 275 290

8 31 73 94

58 67 68 69

25 28 29 29

17 5 3 3

0.7 0.2 0.1 0.1

2.3 2.4 2.4 2.4

Pt/Al2O3

300 325 350 375 400

1 2.5 9 22 50

15 37 42 45 45

9 19 22 23 23

77 44 36 32 32

8.8 2.3 1.7 1.4 1.4

1.7 1.9 1.9 2.0 1.9

Ru/Al2O3

250 275 300

35 48 75

60 62 65

36 33 33

4 5 2

0.1 0.1 0.1

1.7 1.9 2.0

Rh/Al2O3

225 250 275 300

11 30 64 94

66 65 62 61

27 28 30 32

7 7 8 7

0.3 0.2 0.3 0.2

2.4 2.3 2.1 1.9

Re/Al2O3

250 275 300

3 12 36

47 50 51

23 27 35

30 23 14

1.3 0.8 0.4

2.0 1.8 1.4

Pt-0.17Rh/Al2O3

300 325 350

14 59 95

69 70 68

23 24 24

8 6 8

0.3 0.2 0.3

3.0 2.9 2.8

According to a non-selective mechanism, there is an equal probability of breaking cyclic bonds and then the distribution is statistical with 2/5 of n-C6, 2/5 of 2MP and 1/5 of 3MP, corresponding to n-C6/3MP and 2MP/3MP ratios both equal to 2. A completely selective mechanism will produce only 2MP and 3MP, the breaking of substituted endocyclic C–C bonds being impossible. Then, the n-C6/3MP ratio is equal to 0 and the 2MP/ 3MP ratio is equal to 2. The third type of mechanism is partly selective, yielding to a product distribution intermediate between those obtained with the statistical mechanism and the selective one. According to the work of Weisang and Gault [22] the hydrogenolysis of methylcyclopentane on iridium-based catalysts yields selectively to the formation of 2MP and 3MP. The same result was obtained by McVicker et al. [2] and confirmed by the results presented in Table 2, where very few n-hexane is observed, except at 225 8C, although the reaction conditions are completely different from those used in [22] especially the pressure and the H2/MCP ratio. On platinum-based catalysts, the distribution of ring-opening products from the hydrogenolysis of methylcyclopentane is known to be sensitive to the size of the metal particles [18,23,24], small platinum particles favoring the non-selective mechanism and the larger ones the selective mechanism, especially at low temperatures. On large particles but at high temperature, the mechanism is partly selective. In the same conditions as us, McVicker et al. [2] observed a non-selective mechanism on their 0.3 wt.% Pt/Al2O3, where platinum is likely to be well dispersed due to the low metal loading. They obtained a statistical distribution with 41% of nhexane, 42% of 2-methylpentane and 17% of 3-methylpentane at 350 8C. The product distribution obtained in the present study on 0.6 wt.% Pt/Al2O3 at the same temperature (350 8C) is slightly different from that of McVicker et al. [2], with a little less n-hexane and a little more 3-methylpentane. However, except at very low conversion (1%) because of a lack of precision, the product distribution is close to 2/5 of n-C6, 2/5 of 2MP and 1/5 of 3MP. Consequently, the RO product distribution obtained with this catalyst can be considered as resulting from a statistical mechan-

107

ism, in accordance with the high dispersion value (75%) of 0.6 wt.% Pt/Al2O3, as measured by H2 chemisorption. Contrary to what was demonstrated for platinum-based catalysts, the particle size of rhodium does not affect significantly the product distribution resulting from the methylcyclopentane hydrogenolysis. However, the turnover frequency is noticeably changed by the variation of particle size [25]. Results presented in Table 2 show that the percentage of n-hexane is very low and the product distribution is very similar to that obtained with Ir/Al2O3 corresponding to a selective mechanism. These results are in accordance with those of Teschner et al. [26] and Coq et al. [25] obtained in different conditions, at a pressure lower or equal to 1 bar. The product distribution obtained with the Ru/Al2O3 catalyst is similar to what is obtained with the Rh/Al2O3 catalyst. This result is in accordance with those presented in Refs. [17,27,28] obtained on Ru/SiO2 catalyst (chloride precursors) but at much lower temperature (160 8C) and pressure. On Re/Al2O3 catalyst, the RO product distribution varies a lot as a function of temperature: an increase in the temperature yields to a decrease in n-C6 and an increase in 3MP whereas 2MP is not strongly affected. The product distribution seems to result from a partly selective mechanism, contrary to what is reported in [17,29] for Re catalysts tested under atmospheric pressure. 3.2. Bimetallic catalysts The parent Pt/Al2O3 catalyst was modified by metal additives i.e. either germanium or copper with the aim of poisoning platinum metallic sites responsible for deep hydrogenolysis or (ii) ruthenium or rhodium in order to increase the activity of the catalyst with the hope of also creating new selective metallic sites. 3.2.1. Effect of Ge or Cu addition In reforming processes, addition of metals such as tin or germanium to monometallic Pt/Al2O3 catalysts provides better stability, reduces the sintering effect, improves selectivity by inhibition of the hydrogenolytic effect of platinum, and decreases isomerization and coke deposition, while aromatization is increased [30–33]. Two metals were studied, i.e. germanium and copper. They were both added by catalytic reduction onto the parent Pt/Al2O3 catalyst. This method of preparation may favor the deposition of the modifier onto platinum particles but a part is likely to be deposited onto the support. However, these isolated species will not play a role in the reaction since monometallic copper and germanium species are inactive for methylcyclopentane RO. The composition of the different bimetallic catalysts is summarized in Table 3. Whatever the modifier and the amount added, the deposition is complete and the chlorine content is roughly equal to 1.4%. The complete deposition is due to (i) the preparation conditions, i.e. the catalytic reduction, in which the reducing agent of the modifier salt, hydrogen, is continuously provided at the platinum surface, (ii) the nature of the modifiers, Cu and Ge being easily deposited in a trimensional structure [34,35], (iii) to the low amount of modifier introduced or (iv) the deposition of the modifier onto the support. In order to evidence the interaction between the modifier and platinum, the activity of the bimetallic catalyst was determined for cyclohexane dehydrogenation at 270 8C. The relative activity, corresponding to the ratio between the activity of the bimetallic catalysts and the activity of the parent catalyst, is reported in Table 3. It can be seen that copper and germanium induce a decrease in the activity of the parent catalyst. However, this effect is more marked for germanium. It can be inferred from these results that a large part of the modifiers is deposited onto the platinum particles thus inhibiting its activity for cyclohexane dehydrogenation. In the following, the behavior of Pt-Cu and Pt-Ge catalysts will be examined separately.

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112

108

Table 3 Composition of 0.6 wt.% Pt-X/Al2O3 (X = modifier, Cu, Ge, Ru or Rh) and catalytic activities. Catalyst

wt.% X theoretical

wt.% X deposited

wt.% Cl

Relative activity in MCP hydrogenolysis (at T, 8C)a

Relative activity in cyclohexane dehydrogenationa

Pt/Al2O3

0.00

0.00

1.4

1.00

1.00

Pt-Cu/Al2O3

0.10 0.20 0.50

0.10 0.20 0.50

1.4 1.3 1.4

0.80 (350) 0.60 (350) 0.32 (350)

0.52 0.43 0.34

Pt-Ge/Al2O3

0.05

0.05

1.3

0.08 (375)

0.41

Pt-Ru/Al2O3

0.05 0.20

0.03 0.10

1.3 1.4

1.10 (350) 2.70 (350)

– –

Pt-Rh/Al2O3

0.10 0.20 0.30 0.40 0.60 0.80

0.04 0.06 0.08 0.11 0.14 0.17

1.2 1.4 1.4 1.3 1.4 1.4

2.33 (350) 3.77 (350) 8.55 (350) 8.44 (350) 9.77 (350) 10.55 (350)

0.59 0.56 0.54 0.45 0.37 0.27

a

Relative activity = activity of the bimetallic catalyst divided by the activity of Pt/Al2O3 catalyst (activity calculated per gram of catalyst).

3.2.1.1. Addition of germanium. The conversion as a function of temperature and the evolution of C1–C5 products obtained in the presence of the Pt-Ge/Al2O3 catalyst are presented in Fig. 3a and b, respectively. This figure as well as the relative activity for MCP RO reported in Table 3 for a given temperature (375 8C) clearly show that the addition of a very low amount of germanium (0.05 wt.%) onto platinum strongly deactivates the parent catalyst. This observation is in accordance with the result obtained in cyclohexane dehydrogenation presented also in Table 3. Moreover, an increase of the yield in C1–C5 products is obtained (Fig. 3b), thus

a decrease of the RO selectivity. Then, germanium is not an appropriate modifier to improve the selectivity of platinum in ring-opening. 3.2.1.2. Addition of copper. The conversion as a function of temperature and the evolution of C1–C5 products obtained in the presence of the Pt-Cu/Al2O3 catalysts with various copper loadings are presented in Fig. 4a and b, respectively. The results presented in Fig. 4a and Table 3 show that whatever the copper content, the activity of the bimetallic catalyst is lower than that of the parent monometallic Pt/Al2O3 catalyst. The addition of copper leads to a decrease in the activity of the catalyst compared to that of the parent Pt/Al2O3 catalyst at the same WHSV, which is all the more important as copper loading increases. This decrease in activity may be explained by a geometrical effect of copper, which is inactive for MCP ring-opening. This result is in accordance with the result obtained in cyclohexane dehydrogenation presented in Table 3 As far as the yield in cracking products is concerned (Fig. 4b), the behavior of the Pt-Cu catalysts is similar to that of the parent Pt/Al2O3 catalyst. Then, it can be concluded that the addition of modifiers such as copper or germanium onto platinum yields a deactivation of the catalyst without improving its selectivity in ring-opening. 3.2.2. Effect of Ru or Rh addition It was shown in Section 3.1 that ruthenium and rhodium are very active for methylcyclopentane hydrogenolysis but their strong hydrogenolytic activity yields not only ring-opening products but also non-valuable cracking products resulting from deep hydrogenolysis. In this study, ruthenium and rhodium were added onto a parent Pt/Al2O3 catalyst by a surface redox reaction, according to the refilling method, with the aim of increasing the activity of platinum catalyst and also modifying its selectivity in methylcyclopentane ring-opening to obtain properties similar to those of iridium. A strong interaction between the active modifier and the parent metal is of major importance since the presence of Ru or Rh isolated species on the support would favor the production of cracking products as it was observed with the monometallic Ru/Al2O3 and Rh/Al2O3 catalysts.

Fig. 3. (a) Conversion of MCP as a function of temperature; (b) yield in C1–C5 products as a function of the MCP conversion for Pt/Al2O3 with WHSV = 6 h1 (^), Pt/Al2O3 (&), and Pt-Ge/Al2O3 with WHSV = 6 h1 ().

3.2.2.1. Addition of ruthenium. The composition of the prepared bimetallic Pt-xRu/Al2O3 catalysts (x = weight percentage of Ru in the catalyst) is detailed in Table 3. Contrary to what was observed

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112

Fig. 4. (a) Conversion of MCP as a function of temperature; (b) yield in C1–C5 products as a function of the MCP conversion for Pt/Al2O3 with WHSV = 6 h1 (^), Pt/Al2O3 (&), and Pt-Cu/Al2O3 catalysts with WHSV = 6 h1 with: 0.1 wt.% Cu (~), 0.2 wt.% Cu () and 0.50 wt.% Cu (~).

with copper or germanium, the deposition of Ru is not complete. However, the amount deposited increases with the theoretical amount, corresponding to the amount of precursor salt introduced for the preparation. The method used to deposit ruthenium onto platinum is the refilling method: during the first minute, the solution containing Ru3+ cations is contacted with the platinum catalyst under hydrogen. Thus Ru3+ cations may be reduced by the hydrogen preadsorbed onto platinum. Taking into account the accessibility of platinum (75% as shown in Table 1), and that three hydrogen atoms are needed to reduce one Ru3+, the maximum amount of ruthenium that can be deposited during this first minute is approximately 0.08 wt.% onto 0.6 wt.% Pt/Al2O3, considering a Pt/H stoichiometry of 1. After this first step, hydrogen is bubbled in the medium and the remaining ruthenium cations that have not been yet deposited can be reduced either by the hydrogen adsorbed onto the metallic particles or in homogeneous phase by hydrogen dissolved in water. However, the amount of ruthenium that can be deposited onto the parent catalyst using this preparation method is limited by the fact that Ru probably poorly catalyzes its own reduction since the H2 chemisorption rate on Ru is very low at room temperature [36] and three adjacent hydrogen atoms are needed to deposit one Ru metal atom. Then, the difference between the experimental and theoretical ruthenium content could result from the difficulty of finding three hydrogen neighbors, which is all the more important that the ruthenium coverage increases in platinum surface, even if Ru probably segregates on particular Pt sites. The same evolution was observed by Pieck et al. during the deposition of rhenium on platinum catalysts using the same type of preparation method (redox reaction) [37].

109

Fig. 5. (a) Conversion of MCP as a function of temperature; (b) yield in C1–C5 products as a function of the MCP conversion for Pt/Al2O3 with WHSV = 6 h1 (^), Pt/Al2O3 (&), Ru/Al2O3 (*), and Pt-Ru/Al2O3 catalysts: Pt-0.03Ru () and Pt-0.10Ru (~).

The conversion as a function of temperature and the evolution of C1–C5 products obtained in the presence of the Pt-xRu/Al2O3 catalysts are presented in Fig. 5a and b, respectively, and compared to those of the monometallic platinum and ruthenium catalysts. Fig. 5a and relative activities given in Table 3 show that the addition of a small amount of Ru (0.03 wt.%) does not significantly increase the activity of the Pt/Al2O3 in the same experimental conditions, even if ruthenium is much more active than platinum for methylcyclopentane hydrogenolysis. For the higher Ru content, the conversion increases with the amount of Ru. As far as the evolution of the C1–C5 products is concerned (Fig. 5b), the Pt-xRu/ Al2O3 catalysts favor the deep hydrogenolysis whatever the Ru content compared to the behavior of the monometallic Pt/Al2O3 catalyst. The intermediate catalytic behavior exhibited by the bimetallic Pt-Ru systems compared to the two monometallic ones seems to indicate that Pt and Ru are little or even not alloyed in the bimetallic samples. In conclusion, the addition of ruthenium on a monometallic platinum-based catalyst allows one to increase its activity but also the selectivity to cracking products that is not the objective of this study. 3.2.2.2. Addition of rhodium. The parent Pt/Al2O3 was modified by addition of various amounts of rhodium by the refilling method. The composition of the Pt-xRh catalysts is summarized in Table 3. As for ruthenium, the deposition of rhodium is not complete and the same reasons may be invoked. Moreover, it was verified that some rhodium reduction by hydrogen dissolved in solution may occur.

110

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112

The conversion as a function of temperature and the evolution of C1–C5 products obtained in the presence of the Pt-xRh/Al2O3 catalysts (x = weight percentage of rhodium) are presented in Fig. 6a and b, respectively, and compared to those of the monometallic platinum, rhodium and iridium catalysts. In order to favor the comparison of the behavior of bimetallic catalyst, containing 0.6 wt.% Pt and less than 0.3 wt.% Rh, to those of the monometallic ones, the monometallic Rh catalyst presented in this figure contains 0.3 wt.% Rh. As far as the conversion of MCP (Fig. 6a and Table 3) is concerned, one can see that the activity of the bimetallic catalysts is higher than that of the monometallic platinum catalyst tested in the same experimental conditions. The activity increases with the rhodium content to tend to that of the monometallic rhodium catalyst for the highest rhodium content. As far as the yield in C1–C5 products is concerned (Fig. 6b), one can see that the addition of a small amount (0.04 wt.%) of Rh does not significantly modify the behavior of the monometallic Pt catalyst. When the rhodium content is increased, the C1–C5 yield decreases and, from 0.08 wt.% of Rh, the production of C1–C5 products as a function of the conversion is very close to that of Ir. The RO product distribution and the molar ratio of n-hexane to 3-methylpentane (n-C6/3MP) and 2-methylpentane to 3-methylpentane (2MP/3MP) obtained with the 0.6% Pt-0.17% Rh/Al2O3 catalyst is reported in Table 2 at various temperatures and then conversions. Whereas the amount of rhodium deposited onto the parent Pt/Al2O3 catalyst is rather low, the RO product distribution resembles more to that obtained with rhodium (very similar to that of iridium) than that obtained with platinum. The low yield in C1–C5 products obtained with the Pt-Rh catalysts indicates that the deposition of Rh onto a Pt/Al2O3

Fig. 6. (a) Conversion of MCP as a function of temperature; (b) yield in C1–C5 products as a function of the MCP conversion for Pt/Al2O3 with WHSV = 6 h1 (^), Pt/Al2O3 (&), 0.3 wt.% Rh/Al2O3 (*), and bimetallic Pt-Rh/Al2O3 catalysts [Pt0.04Rh (), Pt-0.06Rh (+), Pt-0.08Rh (), Pt-0.11Rh (*), Pt-0.14Rh (&) and Pt-0.17Rh (~)] compared to 0.6 wt.% Ir/Al2O3 (^).

Fig. 7. Compensation plot for the production of C1–C5 by deep hydrogenolysis onto Pt (*), Rh (&), Ir (~) and Pt-Rh (^) catalysts supported on alumina.

parent catalyst modifies strongly the selectivity of the catalyst, the behavior of the bimetallic catalyst being completely different from those of the two monometallic catalysts. In order to evidence this singular behavior, the apparent activation energies (Ea) as well as the pre-exponential factors (A) were determined from the results presented in Fig. 6. For Pt-Rh and Ir catalysts, and contrary to the monometallic Pt and Rh catalysts, a compensation effect [15,38] is observed between the apparent activation energy and the pre-exponential factor for the C1–C5 formation, as shown in Fig. 7. This suggests that the same mechanism occurs on Pt-Rh and Ir catalysts for the deep hydrogenolysis and that binary Pt-Rh sites are formed. The Pt-Rh bimetallic catalysts were also characterized by a test reaction, cyclohexane dehydrogenation, and electron microscopy. As rhodium is much less active than platinum for cyclohexane dehydrogenation [39], the activity of bimetallic Pt-Rh catalysts in cyclohexane dehydrogenation was determined and compared to that of the Pt/Al2O3 catalyst in order to evidence if rhodium is deposited onto platinum or not. Indeed, if rhodium is not deposited on platinum, the activity of the bimetallic catalyst should be slightly higher than that of the parent Pt/Al2O3 catalyst. The relative activity, corresponding to the ratio between the activity of the bimetallic catalyst and the activity of the monometallic platinum catalyst, is presented in Table 3. The values clearly show that the addition on Rh poisons the activity of platinum since the activity of bimetallic catalysts decreases with an increase in the rhodium content. This behavior proves that Rh is deposited in interaction with platinum. This modification of platinum by rhodium is also evidenced by the mean particle size of the bimetallic catalysts, characterized by transmission electron microscopy (Fig. 8). The mean particle size in the bimetallic Pt-0.14Rh/Al2O3 catalyst is 2.2 nm compared to 1.3 nm for the parent Pt catalyst (Table 1): the increase in the particle size by addition of Rh may be attributed to the real deposition of rhodium onto platinum. It may also be due to the preparation method used to deposit rhodium. It is known that the immersion of a parent catalyst in hydrochloric acid under hydrogen atmosphere favors the sintering of the metallic phase in the case of a silica or alumina support [35,40]. In order to discriminate between the particle enlargement due to platinum sintering and that due to rhodium deposition, a blank Pt monometallic catalyst was prepared by submitting the 0.6 wt.% Pt/Al2O3 catalyst to the same treatment than that used for the deposition of rhodium but without introduction of Rh salt. The particle size of this blank catalyst was then determined by H2 chemisorption and TEM. The particle size of the blank

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112

111

Fig. 8. (a) TEM picture and (b) metal particle size distribution for Pd-0.14Rh/Al2O3 bimetallic catalyst.

monometallic catalyst determined from TEM pictures (1.8 nm) was slightly higher than that calculated from the H2 chemisorption measurements (1.6 nm). This can be explained by the presence of very small particles (<1 nm) which are not visible in the TEM pictures. The comparison of the particle sizes measured on the bimetallic Pt-0.14Rh/Al2O3 catalyst with the blank Pt/ Al2O3 catalyst estimated from TEM pictures shows that the addition of Rh leads to an increase of the particle size of 0.4 nm. This value is in accordance with an increase in size estimated from the Rh loading considering that all the rhodium is deposited onto the platinum particles. The H2 chemisorption experiments onto the Pt-Rh bimetallic catalysts showed that the dispersion varies between 40% and 48%. 4. Conclusion The aim of this paper was to try to obtain bimetallic Pt-X catalysts supported on alumina with RO selectivity similar to that of Ir/Al2O3, which is known to be the most selective for the methylcyclopentane ring-opening even under near-industrial conditions. Two types of X modifiers were studied, either inactive metals, such as copper or germanium, or active ones, such as ruthenium and rhodium, for methylcyclopentane hydrogenolysis. Among the tested series, only Pt-Rh showed interesting properties: whereas platinum is not very active for this reaction, the addition of rhodium by surface redox reaction allowed us to increase its activity. More surprising, whereas neither platinum nor rhodium were as selective as iridium, both favoring the production of C1–C5 products in a different extent, the combination of the two metals in the same metal particles led to completely different selectivity, which was similar to that of iridium, in a large range of rhodium content. This may be explained by a synergetic effect between the two metals, the bimetallic particles presenting behavior completely different from that of the monometallic particles. In a following paper, we will study the influence of the preparation method on the catalytic performances of the Pt-Rh/Al2O3 catalysts for MCP ring-opening and characterize more in depth these catalyst series by various techniques in order to evidence if this particular behavior is linked to the preparation method and to precise the arrangement of platinum and rhodium at the support surface.

Acknowledgements The authors thank G. Mabilon and D. Uzio (IFP) for the helpful discussions. References [1] F.G. Gault, Adv. Catal. 30 (1981) 1–95. [2] G.B. McVicker, M. Daage, M.S. Touvelle, C.W. Hudson, D.P. Klein, W.C. Baird Jr., B.R. Cook, J.G. Chen, S. Hantzer, D.E.W. Vaughan, E.S. Ellis, O.C. Feeley, J. Catal. 210 (2002) 137–148. [3] P.T. Do, W.E. Alvarez, D.E. Resasco, J. Catal. 238 (2006) 477–488. [4] M. Santikunaporn, W.E. Alvarez, D.E. Resasco, Appl. Catal. A 325 (2007) 175–187. [5] S. Dokjampa, T. Rirksomboon, D.T.M. Phuong, D.E. Resasco, J. Mol. Catal. A 274 (2007) 231–240. [6] S. Lecarpentier, J. van Gestel, K. Thomas, J.P. Gilson, M. Houalla, J. Catal. 254 (2008) 49–63. [7] S.L. Gonza´lez-Corte´s, S. Dokjampa, P.T. Do, Z. Li, J.M. Ramallo-Lope´z, F.G. Requejo, Chem. Eng. J. 139 (2008) 147–156. [8] R.J. Fenoglio, G.M. Nun˜ez, D.E. Resasco, Appl. Catal. 63 (1990) 319–332. [9] G. Espinosa, G. Del Angel, J. Barbier, P. Bosch, V. Lara, D. Acosta, J. Mol. Catal. A 164 (2000) 253–262. [10] M. Chamam, A. Wootsch, L. Pirault-Roy, I. Boghian, Z. Paa´l, Catal. Commun. 8 (2007) 686–692. [11] N. Gyo˝ rffy, A. Wootsch, S. Szabo´, I. Bakos, L. To´th, Z. Paa´l, Topics Catal. 46 (1–2) (2007) 57–64. [12] R.J. Chimenta˜o, G.P. Valenc¸a, F. Medina, J. Pe´rez-Ramı´rez, Appl. Surf. Sci. 253 (2007) 5888–5893. [13] Z. Paa´l, N. Gyo˝ rffy, A. Wootsch, L. To´th, I. Bakos, S. Szabo´, U. Wild, R. Schlo¨gl, J. Catal. 250 (2007) 254–263. [14] M. Chamam, K. La´za´r, L. Pirault-Roy, I. Boghian, Z. Paa´l, A. Wootsch, Appl. Catal. A 332 (1) (2007) 27–36. [15] G.C. Bond, Metal-Catalysed Reactions of Hydrocarbons, Springer, 2005. [16] L.M. Kustov, T.V. Vasina, O.V. Masloboishchikova, E.G. Kelkovskaya-Sergeeva, P. Zeuthen, Stud. Surf. Sci. Catal. 130 (2000) 227–232. [17] Z. Paal, P. Te´te´nyi, Nature 267 (1977) 234–236. [18] G. Maire, G. Plouidy, J.C. Prudhomme, F.G. Gault, J. Catal. 4 (1965) 556–569. [19] M. Vaarkamp, P. Dijstra, J. Van Grondelle, J.T. Miller, F.S. Modica, D.C. Koningsberger, R.A. van Santen, J. Catal. 151 (1995) 330–337. [20] J.B.F. Anderson, R. Burch, J.A. Cairns, J. Catal. 107 (1987) 351–363. [21] M.J. Dees, M.H.B. Bol, V. Ponec, Appl. Catal. 64 (1990) 279–295. [22] F. Weisang, F.G. Gault, J. Chem. Soc. Chem. Commun. 11 (1979) 519–520. [23] G. Maire, C. Corolleur, D. Juttard, F.G. Gault, J. Catal. 21 (1971) 250–253. [24] C. Corolleur, F.G. Gault, D. Juttard, J.M. Muller, J. Catal. 27 (1972) 466–468. [25] B. Coq, R. Dutartre, F. Figueras, T. Tazi, J. Catal. 122 (1990) 438–447. [26] D. Teschner, Z. Paal, D. Duprez, Catal. Today 65 (2001) 185–190. [27] G. Diaz, F. Garin, G. Maire, S. Alerasool, R.D. Gonzalez, Appl. Catal. A. 124 (1995) 33–46. [28] B. Coq, A. Bittar, F. Figueras, Appl. Catal. 59 (1990) 103–121. [29] W. Juszczyk, Z. Karpinski, Appl. Catal. A 206 (2001) 67–78. [30] J. Volter, G. Lietz, M. Uhlemann, M. Herman, J. Catal. 68 (1981) 42–50. [31] J. Beltramini, D.L. Trimm, Appl. Catal. 31 (1987) 113–118.

112 [32] [33] [34] [35]

P. Samoila et al. / Applied Catalysis A: General 369 (2009) 104–112

N. Macleod, J.R. Fryer, D. Stirling, G. Webb, Catal. Today 46 (1998) 37–54. M. Boutahala, B. Djellouli, N. Zouaoui, F. Garin, Catal. Today 89 (2004) 379–385. F. Epron, F. Gauthard, C. Pine´da, J. Barbier, J. Catal. 198 (2001) 309–318. G. Lafaye, C. Micheaud-Especel, C. Montassier, P. Mare´cot, Appl. Catal. A 230 (2002) 19–30. [36] R.A. Dalla Betta, J. Catal. 34 (1974) 57–60.

[37] C.L. Pieck, P. Mare´cot, J. Barbier, Appl. Catal. A 134 (1996) 319–329. [38] G.C. Bond, M.A. Keane, H. Kral, J.A. Lercher, Catal. Rev. Sci. Eng. 42 (2000) 323–383. [39] L.I. Ali, A.-G.A. Ali, S.M. Aboul-Fotouh, A.K. Aboul-Gheit, Appl. Catal. A 177 (1999) 99–110. [40] A. Douidah, P. Mare´cot, S. Labruque`re, J. Barbier, Appl. Catal. A 210 (2001) 111– 120.