- Email: [email protected]
Insight into sulphur compounds and promoter effects on Molybdenum-based catalysts for selective HDS of FCC gasoline
Applied Catalysis A: General 388 (2010) 188–195
Contents lists available at ScienceDirect
Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata
Insight into sulphur compounds and promoter effects on Molybdenum-based catalysts for selective HDS of FCC gasoline Celine Fontaine a,∗ , Yilda Romero a , Antoine Daudin b , Elodie Devers b , Christophe Bouchy b , Sylvette Brunet a a b
UMR CNRS 6503, Catalyse en Chimie Organique, Université de Poitiers, Faculté des Sciences Fondamentales et Appliquées, 40 avenue du Recteur Pineau, 86022 Poitiers Cedex, France IFP Energies nouvelles, Direction Catalyse et Séparation, Rond-point de l’échangeur de Solaize, BP 3, 69360 Solaize, France
a r t i c l e
i n f o
Article history: Received 6 April 2010 Received in revised form 23 August 2010 Accepted 24 August 2010 Available online 23 September 2010 Keywords: Hydrodesulphurisation Hydrogenation FCC gasoline Molybdenum sulphide-based catalyst Nickel Cobalt Selectivity Activity E(M-S) HDS HYD
a b s t r a c t The effect of the nature of sulphur compounds (H2 S and 2-methylthiophene) over unsupported molybdenum-based sulphide catalysts (promoted or not by nickel or cobalt) for the transformation of a model FCC feed (hydrodesulphurisation of 2-methylthiophene and hydrogenation of 2,3-dimethylbut2-ene) was investigated. The activities of the various catalysts were compared between each reactant alone and the full model mixture. The promoting effect of Co and Ni is found for both hydrodesulphurisation (HDS) and hydrogenation of olefins (HYD) reactions. However, the hydrogenation activity strongly depends on the amount of sulphur compound present in the feed. Nickel, used as a promoter, seems more sensitive than cobalt, particularly for the hydrogenation reaction. The presence of 2-methylthiophene in the feed induces a stronger inhibiting effect than H2 S on HYD reaction. This result highlights the different adsorption constant ratio between olefins and sulphur molecules depending on the sulphur compounds. Even though an impact of H2 S can be observed on the NiMo and CoMo catalysts, no major modification occurs on the volcano-shaped curves obtained in previous work, in terms of activity or selectivity, as a function of the metal–sulphur bond energy. Whatever the amount of H2 S added, NiMo and CoMo catalysts fit into the curve, NiMo being the most active and the most selective under these conditions. © 2010 Elsevier B.V. All rights reserved.
1. Introduction In order to meet the always more severe European regulations regarding the amount of sulphur in gasoline [1], hydrotreating catalysts must achieve a deep hydrodesulphurisation (HDS) with a minimum of olefins hydrogenation (HYD), i.e. get to a better selectivity (hydrodesulphurisation/olefins hydrogenation: HDS/HYD) criteria. There has been many studies dealing with HDS and focusing on the understanding of the various phenomenon involved in this reaction, in order to improve the selectivity [2–5]. To achieve such a goal, the optimisation of conventional catalysts or the elaboration of new ones are required. Processes and catalysts allowing the removal of sulphur must be more and more efficient. Commercial gasoline is composed of different fractions coming from reforming, isomerisation and fluid catalytic cracking (FCC) units. The gasoline fraction produced from the FCC process represents 30–50% of the commercialised motor fuel but it contains up to 85–95% of the sulphur impurities. It is mainly composed of aromatics (30 vol.%), isoalkenes (20–40 vol.%) and sulphur compounds
∗ Corresponding author. Tel.: +33 5 49 45 38 39; fax: +33 5 49 45 38 97. E-mail address: [email protected] (C. Fontaine). 0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2010.08.049
such as alkylthiophenes (max. 5000 wt ppm) [6–9]. The nature of sulphur compounds is now well-known and has been reviewed by Brunet et al. [2]: mainly benzothiophenes, alkylthiophenes, thiols and sulphides. Besides, the olefins amount is high (20–40 vol.%) and helps to improve the octane number. Therefore, it is important to remove the sulphur compounds while limiting the olefins hydrogenation in order to keep a high octane number (selective HDS). Conventional operating conditions are quite mild in terms of temperature and pressure. Catalysts used for such hydrotreating reactions are transition metal sulphides (TMS), supported or not [10]. Brunet et al. [2] have reviewed the different parameters influencing the activity and the selectivity such as the presence of a promoter, the reactions and mechanisms involved in HDS reactions. It is known that there is a competitive competition between the sulphur and the olefin compounds on the catalyst depending on the catalyst nature, which could modify the activity and the selectivity. Daudin et al. [11] and Miller et al. [12] showed that mixed unsupported sulphides (NiMo and CoMo) had a much better HDS activity and HDS/HYD selectivity than the non-promoted catalyst MoS2 under model conditions. Recently, experimental results were coupled with theoretical models (mainly DFT calculations) in order to explain more precisely the mechanisms involved in HDS reactions [3,13–16]. Daudin et al.
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
[11,14] have shown that there was a relation between the ab initio calculated metal–sulphur bond energy (E(MS)) of unsupported transition metal sulphides and their activity for the HDS and HYD reactions. They generalised the volcano-shaped curve to unsupported transition metal catalysts, showing that the maximum of activity (HDS and HYD) or selectivity (HDS/HYD) could be obtained for intermediate E(MS). For supported catalysts, it has been shown that a more basic support would account for a higher selectivity [2,16,17]. For example, Mey et al. [16] showed a decrease in HYD on a CoMo supported catalyst doped with potassium. Another way of improving the selectivity was also to selectively poison the hydrogenation sites [4,5]. When the hydrogenation activity was inhibited by carbon deposition or adsorption of basic nitrogen compounds such as pyridine, the activity of the catalyst for the HYD reaction was more affected than the HDS, thus increasing the selectivity. Indeed, HDS/HYD selectivity is the key parameter for an efficient gasoline hydrotreating catalyst. As explained above, selectivity can be influenced by many parameters (nature of the active phase, acidic–basic properties of the support, use of selective poison) but might also depend on the amount of H2 S in the feed. The influence of H2 S in the feed has been the subject of numerous works as it is obtained as a by-product in HDS reactions [2,18]. In particular the influence of H2 S has been studied on diesel HDS [18], olefins hydrogenation reaction [19] and very recently on model FCC feeds [16,20]. It has been admitted that Mo-based catalysts promoted by Ni are more affected by the presence of H2 S than catalysts promoted by Co while studying the HDS of dibenzothiophene or 4,6-dimethyldibenzothiophene, but the results depended on the experimental conditions [18]. Dos Santos et al. [21] used a commercial CoMo/Al2 O3 catalyst under mild conditions of temperature and pressure with a model FCC feed (3-methylthiophene and hex-1-ene diluted in n-heptane), and showed that there was an inhibiting effect of H2 S on both HDS and HYD reactions and an effect of PH2 S on the selectivity explained by a variation in the rate determining step. Only a few papers are available in the literature dealing with the influence of sulphur compounds (H2 S) on the olefins alone [19,21,22]. Lamic et al. [19] showed the negative impact of H2 S on different unsupported molybdenum-based catalysts for the transformation of a model olefin (2,3-dimethylbut-2-ene), the NiMo catalyst being more inhibited than the CoMo or the MoS2 one by small amounts of H2 S. Dos Santos et al. [21] discussed a model for the competitive reactions happening during the deep HDS of FCC gasoline, using a CoMo/Al2 O3 catalyst and under model conditions, and found that there was a competitive adsorption of 3-methylthiophene and hex-1-ene on the catalyst surface as well as an inhibiting effect of H2 S on both reactions (HDS and HYD). For the HYD reaction, the authors showed that the olefin and the sulphur molecule were in competitive adsorption on the same active site, and that a strong influence of the olefin was shown on the HDS reaction. This paper will deal with the transformation of model molecules, either alone or in mixture, representative of sulphur and olefinic compounds found in FCC gasoline. The influence of PH2 S on the performances of unsupported Mo-based catalysts will be described in terms of HDS and HYD activities and HDS/HYD selectivity. The influence of the nature of the promoter, cobalt or nickel, will be particularly studied. The impact on the olefin reactivity of two sulphur compounds: H2 S, already known as an important parameter for hydrotreatment reactions, and 2-methylthiophene, one of the reactants, will be addressed. This study will be extended to previous results correlating reactivity as a function of the metal–sulphur bonding energy (E(MS)) in order to observe the impact of the sulphur compound nature on the volcano-shaped curves established for transition metal sulphides.
189
2. Experimental 2.1. Catalysts preparation The preparation method of unsupported molybdenum sulphide catalysts has already been reported in previous works [11,14,19–24]. Molybdenum sulphide preparation was carried out by thermal decomposition of ammonium thiomolybdate at 673 K under H2 S/H2 (10 mol% H2 S) during 4 h. First, the ammonium tetrathiomolybdate (NH4 )2 MoS4 (ATM) was obtained by reaction between ammonium heptamolybdate (NH4 )6 [Mo7 O24 ], 4H2 O (4 g in 20 cm3 of distilled water) with 50 wt% of ammonium sulphide (NH4 )2 S in aqueous solution at 353 K. The ATM precursor precipitated as red crystals by cooling down the solution in iced water during 3 h. Then, the precipitated red crystals were thoroughly washed with isopropanol and dried. Molybdenum-based materials promoted by nickel or cobalt in order to obtain NiMo and CoMo catalysts were prepared according to the method reported by Fuentes et al. [25] and are the same than those used by Lamic et al. [19]. (NH4 )2 MoS4 (prepared as mentioned below) was used as a support to impregnate Co(NO3 )2 or Ni(NO3 )2 diluted in acetone in order to obtain an atomic ratio of Ni(Co)/(Ni(Co) + Mo) = 0.3. The bimetallic organosulphides precursors were obtained by removing the solvent at room temperature. Finally, NiMo and CoMo catalysts were prepared by thermal decomposition of the corresponding bimetallic organosulphide precursor at 673 K under H2 S/H2 (10 mol% H2 S) flow during 4 h. All catalysts were characterised before and after catalytic activity measurements by TEM combined with EDX (Philips CM 120 kV), X-ray diffraction (Bruker D5005), BET surface area (Micromeritics ASAP 2010) and elemental analysis (CE Instruments NA2100 Protein) at the University of Poitiers (LACCO) in order to verify the exact nature of the transition metal sulphide (TMS) and that there was no modification of the catalyst during the test, according to Lamic et al. [19,27]. XPS analyses have also been performed at IFP-Lyon and are reported elsewhere [19] and confirmed the formation of the desired sulphide catalysts. Characterisation of the catalysts before and after catalytic test (Tables 1 and 2 [19]) showed a decrease in the SBET especially for the MoS2 catalyst (from 76 to 10 m2 g−1 ) while the promoted catalysts were less affected (from 13 to 7 m2 g−1 for CoMo and from 14 to 12 m2 g−1 for NiMo). The particles size is small for all catalysts (8, 9 and 6 nm for MoS2 , NiMo and CoMo, respectively) and the size of the particles was almost not modified after catalytic test (8, 8 and 5 nm for MoS2 , NiMo and CoMo, respectively). Quantitative analyses of XPS are reminded in Table 2, showing that the ratios of Co and Ni in the mixed phase were found to be different: only 18% of Co was engaged in CoMoS phase instead of 45% for the nickel. Consequently, the NiMoS phase seems to be more favoured than CoMoS phase under sulpho-reductive conditions. It was also shown that the global sulphidation level was very high for all catalysts. For the CoMo catalyst the value was about 95% whereas it was 80% for NiMo in accordance with the NiOx presence detected by XPS. 2.2. Activity measurements Catalytic activity measurements were carried out in a fixedbed reactor at 523 K under a total pressure of 2 MPa. The catalyst was presulphided at 673 K for 10 h with a mixture of 10 mol% H2 S/H2 under atmospheric pressure and then cooled down to the reaction temperature in the presence of the sulphiding mixture. The desired reaction conditions were adjusted (T = 523 K, P = 2 MPa, H2 /feed = 360L/L) and the model feed was injected into the reactor. Three feeds were studied in order to compare the results for each reactant taken separately or in mixture:
190
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
Table 1 Physico-chemical properties of the unsupported molybdenum sulphide-based catalysts determined from elemental analysis (S/Metal atomic ratio (Me = Mo or Mo + Ni(Co)) and Ni(Co)/(Ni(Co) + Mo) atomic ratio), specific surface area (before and after reaction) and particles size (before and after reaction) [19]. Sulphides
S/metal (ICP)
Ni(Co)/(Ni(Co) + Mo) atom (before reaction)
TMS
SBET (m2 g−1 )
SBET after reaction (m2 g−1 )
Particles size (nm)
Particles size after reaction (nm)
Mo NiMo CoMo
2.0 1.7 1.9
– 0.25 0.27
MoS2 NiMoS, MoS2 , NiS CoMoS, MoS2 , Co9 S8
76 14 13
10 12 7
8 9 6
8 8 5
Table 2 XPS characterisation of unsupported molybdenum sulphide-based catalysts: binding energy (eV) of the various elements (S 2p3/2 , Mo 3d5/2 , Ni 2p3/2 and Co 2p3/2 ) from [19]. Relative contribution (in brackets) of the different phases (NiS, NiMoS, NiOx , Co9 S8 , CoMoS).
MoS2 CoMo NiMo
S 2p3/2 (eV)
Mo 3d5/2 (eV)
162.2 162.2 162.2
229.4 229.4 229.4
Ni 2p3/2 (eV)
Co 2p3/2 (eV)
NiS
NiMoS
NiOx
Co9 S8
CoMoS
– – 852.9 (34%)
– – 853.9 (45%)
– – 856.0 (21%)
– 778.4 (82%) –
– 779.1 (18%) –
• The thiophenic feed, composed of 0.3 wt% (1000 wt ppm S) of 2-methylthiophene (2MT) in n-heptane (solvent); this sulphur molecule is representative of the sulphur compounds present in the real FCC gasoline feedstock, • The olefinic feed, with 20 wt% of 2,3-dimethylbut-2-ene (23DMB2N) in n-heptane; representing the branched olefins existing in the FCC gasoline feedstock, • The full model mixture was composed of 0.3 wt% of 2MT, 20 wt% of 23DMB2N and 30 wt% of ortho-xylene, representative of an aromatic molecule, in n-heptane. The residence time has been varied from 0.5 to 80 s in order to compare the catalysts selectivity at a same conversion.
The partial pressure of H2 S varied from 0 to 2 kPa corresponding to the equivalent of 1000 wt ppm S maximum in the feed. The concentration of H2 S added to the feed was monitored by varying the ratio (H2 S–H2 )/pure H2 using a 1 mol% H2 S in H2 cylinder. Experiments were carried out as followed: different amounts of sulphur from 1000 wt ppm to 0 ppm were successively injected and back points at 1000 wt ppm were performed for all catalysts. In order to observe the impact of the addition of H2 S on the transformation of 2MT, the HDS yield was maintained at about 15 mol% for the experiments using the thiophenic molecule alone or in mixture. When the transformation of the olefins was studied, the conversion of the alkenes was also set to 15 mol% alone or in mixture (corresponding under these experimental conditions to an HDS yield of about 30–40 mol%). Ortho-xylene and n-heptane were not transformed under these experimental conditions according to previous works [16]. The reaction products were injected on-line by means of an automatic sampling valve into a 3380 Varian gas chromatograph equipped with a PONA capillary column, a flame ionisation detector and a cryogenic system. The identification of the products was performed by GC–MS coupling (Table 3). Preliminary work has shown that the response factors between the sulphur compounds and the hydrocarbons from the transformation of 2MT and olefins were close and could be considered equal to 1 for all products. This is
Scheme 1. Transformation of 23DMB2N over sulphide catalysts. 23DMB2N: 2,3-dimethylbut-2-ene, 23DMB1N: 2,3-dimethylbut-1-ene, 23DMB: 2,3dimethybutane.
why the area of the GC peaks has been directly linked to the molar concentration of the compounds. The detailed mechanism for the transformation of 2MT is already well known ([10] and references therein) and the reaction products are shown in Table 3. The main products are pentenes, pentane, small quantities of cracking products and 2methyltetrahydrothiophene (2MTHT). These products have already been described by several authors [10,16]. The detailed reaction scheme for the transformation of 23DMB2N has also been reported and discussed in the literature ([11,16] and references therein). Scheme 1 shows the main reactants and product as a reminder. The main reaction product is 2,3-dimethylbutane (23DMB) in a large majority (>98 mol%), the other products being isomers. The isomerisation of 23DMB2N to 23DMB1N is known to be very fast so that the mixture composed of 23DMB2N and 23DMB1N is considered as one reactant when measuring the hydrogenation activity of the catalysts [16,26]. Hydrogenation activities (HYD) of the catalysts are defined as the number of moles of 2,3-dimethylbutane formed per second
Table 3 Transformation of 2-methylthiophene. Products identified [2]. Hydrodesulphurisation (HDS) products:
Sulphur compound:
C4 , C3 , C<3 , Crack
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
191
per square meter of catalyst considering the specific surface area after reaction and adjusted to the amount of sulphide phases determined by XPS (MoS2 : 100%, NiMo: 80% and CoMo: 100%) [27]. Desulphurised products, resulting from the transformation of 2MT were designated as HDS products. They were measured after catalyst stabilisation and under conditions where a linear relationship between conversion and residence time was obtained (conversion rate below 40 mol%). The selectivity of the reaction is given by the ratio between hydrodesulphurisation and olefin hydrogenation (HDS/HYD) and corrected by the amount of sulphur phase obtained in XPS. HYD and HDS activities, selectivities and the synergy coefficient are calculated as described below. Activity a for the transformation of 2MT or olefins (23DMB1N + 23DMB2N) is presented in Eq. (1) a [mol m−2 s−1 ] =
XF mSs wt% of sulphur phase
(1) (mol s−1 ),
where F is the molar flow of the considered reactant X is the conversion and X[%]= 100 × [1 − Areactant /(Areactant + i Ai )], m is the mass of catalyst (g) and SS is the specific surface of the solid (m2 g−1 ). For a selected reaction (transformation of 2MT or 23DMB2N), the synergy coefficient is the ratio between the activity of the promoted catalyst by Ni or Co and the activity of the unpromoted catalyst (MoS2 ). Finally, the selectivity is calculated as the ratio between the molar yields of HDS and HYD (Eq. (2)) and the molar yield is calculated as shown in Eq. (3): Selectivity =
HDS HYD
HDS or HYD [%] = 100 ×
(2)
A i i
Areactant +
A i i
(3)
where Ai is the GC peak area for the transformation products considered and Areactant is the area of the remaining reactant. 3. Results and discussion The performances of the molybdenum-based sulphides (MoS2 , CoMo and NiMo) were measured for the transformation of 2MT alone, of 23DMB2N alone and of the model feed (a mixture of 2MT, 23DMB2N and ortho-xylene), with different H2 S partial pressures. 3.1. Transformation of 2MT alone or in the model feed Fig. 1 presents the H2 S impact on the transformation of 2MT alone or in the model mixture and especially the synergy coefficient as a function of the total amount of H2 S in the feed. The synergy coefficient is indeed the ratio between the activities of the catalysts with higher activity and the catalyst with the lower activity (which is MoS2 when tested with the full model mixture in all case). The total amount of H2 S represents the quantity of H2 S added to the feed plus the amount of H2 S produced by HDS reaction (about 150 ppm S, as the work has been carried out at about 15 mol% of conversion of the 2MT in the thiophenic or the model feed). As awaited [19], a promoting effect of Co and Ni on the HDS activity compared to MoS2 was found, no matter the feed nature. Using the thiophenic mixture, without H2 S added to the feed, the synergy coefficient of the NiMo and the CoMo catalysts were about 15 and 17 (meaning that their activities were 15 and 17 times higher than MoS2 ). When H2 S was progressively added to the feed, the synergy coefficient decreased for the promoted (NiMo and CoMo) catalysts, showing the negative impact of the sulphur molecule for the reaction. For 1000 wt ppm of added H2 S, the synergy coefficient of the promoted catalysts were only 9 and
Fig. 1. Transformation of 2MT over (䊉) NiMo, () CoMo and () MoS2 alone or in mixture. Synergy coefficient (ratio of the activity between the promoted catalyst (NiMo or CoMo) and the activity of MoS2 catalyst) versus total amount of H2 S (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Thiophenic feed: dotted lines (empty symbols) and full model mixture: plain lines (full symbols).
10, respectively for NiMo and CoMo compared to the unpromoted catalyst (meaning that the activities were 9 and 10 times higher for the promoted catalysts compared to MoS2 ). The same observations could be made when using the model mixture. Indeed, when the amount of H2 S in the feed increased from 0 to 1000 ppm of added H2 S, the synergy coefficient for the NiMo and for the CoMo catalysts dropped, respectively from 18.2 to 12.3 and from 14.4 to 10.8. A slight impact of the nature of the feed (thiophenic feed alone or the model mixture) could be observed mainly on the performances of the NiMo catalyst. The trend of the plot was the same, no matter the feed nature. However, it seemed that whatever the amount of H2 S in the feed, the synergy coefficient was about 1.2 times higher when using the model mixture on the NiMo catalyst. The difference between the nature of feeds was less important with the CoMo or MoS2 catalysts. Previous works on the promoting effect of Co and Ni compared to MoS2 have been published, comparing experimental results and DFT calculations [28]. The authors compared the promoted and unpromoted catalysts in terms of activity (HDS of DBT, HYD of biphenyl or toluene) with a calculated model. The impact of the promoter could be explained by the modification of the electron density of the molybdenum when the promoter was added to the catalyst, leading to a weakening of the Mo metal–sulphur bond energy (E(MS)) [29]. The promoters also seemed to facilitate the formation of vacancies, and would then increase the number of active sites as shown by Nielsen et al. [30]. A change in the shape of the crystallite has also been shown by Lauritsen et al. [31–33] that could explain the change in activity. The authors showed, using scanning tunnelling microscopy (STM) and density functional theory (DFT), that the MoS2 crystallite had either a triangular shape [33] or a hexagonal one depending on the sulphiding conditions of gold supported MoS2 clusters [31]. Co–Mo–S crystallites also adopted a hexagonal shape while the NiMo particles shape depended on the nanoclusters size. The smaller Ni–Mo–S particles are more likely dodecagonal, while the larger Ni–Mo–S particles exhibited a truncated triangular shape similar to that observed for Co–Mo–S nanoclusters [31]. These results were confirmed by Raybaud with DFT calculations [28]. It was seen in the literature by Choi et al. [34] that cobalt enhanced significantly 3-methylthiophene conversion compared to MoS2 catalyst. In the literature, using different experimental conditions, typical of diesel hydrotreatment using DBT and 46DMDBT (models diesel molecules), an inhibiting effect of H2 S was observed on the HDS reactions with NiMo and CoMo catalysts by Rabarohoela-
192
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
Fig. 3. Transformation of 23DMB2N alone or in mixture over (䊉) NiMo, () CoMo and () MoS2 . Synergy coefficient (ratio of the activity between the promoted catalyst (NiMo or CoMo) and the activity of MoS2 catalyst) versus total amount of H2 S (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Olefinic feed: dotted lines (empty symbols) and full model mixture: plain lines (full symbols).
Fig. 2. Transformation of 2MT. Selectivity towards formation of (a) pentane nC5 , (b) pentenes and (c) 2MTHT over (䊉) NiMo, () CoMo and () MoS2 (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L).
Rokotovao et al. [18]. They also showed that their catalyst containing Ni (NiMoP) was more sensitive to H2 S than the CoMoP one. Hence a decrease of the activity was observed when the amount of sulphur in the feed increased, as already seen in other reactions with the same sulphur DBT and 46DMDBT molecules [35,36]. The negative impact of H2 S could be explained by a heterolytic dissociation of H2 S (leading to the formation of H+ and SH− species) on the catalyst surface (M-edges and S-edges) [11,37]. The SH− species would adsorb on the vacancies thus increasing the covering rate of H2 S and therefore leaving less vacancies available for the cleavage of the C–S bonds (involving mainly S-edge), leading to a decrease in activity while PH2 S increased. It was observed that the amount of H2 S in the feed slightly modified the selectivity towards pentane (nC5 ) (Fig. 2a), pentenes (Fig. 2b) (the HDS products of 2MT) and 2-methyl-
tetrahydrothiophene (2MTHT) (the hydrogenation product of 2MT) (Fig. 2c). Generally, whatever the amount of H2 S in the feed, the selectivity towards pentane was higher for the MoS2 than for the promoted catalysts, the formation of pentenes was the lowest and more 2MTHT was formed on MoS2 than on the other two. The opposite could be observed for the CoMo catalyst regarding pentane and pentenes with a lower formation of 2MTHT. As to the NiMo catalyst, the selectivities were in between the two other catalysts for pentane and pentenes selectivity, while no 2MTHT was observed. For the pentane formation (Fig. 2a), two different catalysts behaviours were observed. When the amount of H2 S increased, the selectivity towards pentane (nC5 ) decreased on MoS2 while it slightly increased on both promoted catalysts. At first, when no H2 S was added to the feed, the selectivity of the alkane was higher for the unpromoted catalyst (62%) compared to NiMo (41%) and CoMo (20%). However, for 1000 ppm of added H2 S, the selectivity using MoS2 or NiMo was the same (46%) and was higher than the one obtained on CoMo for the production of pentane. The selectivity towards pentenes was shown in Fig. 2(b) and when the amount of H2 S added to the feed increased, the selectivity using MoS2 increased while it slightly decreased on both promoted catalysts. When no sulphur compound was added to the feed, the selectivity was thus higher on CoMo (75%) followed by NiMo (59%) and finally the MoS2 catalyst (9%). When the amount of H2 S increased, the trend for the three catalysts remained the same. Finally, in Fig. 2(c) the selectivity towards 2MTHT (sulphur compound issued of the hydrogenation of 2MT) was reported as a function of the total amount of H2 S present in the feed for the three catalysts. This compound was never observed on the NiMo catalyst whatever the amount of H2 S in the feed. For the two other catalysts though, the amount of 2MTHT increased when H2 S increased to 1000 ppm, from 24% to 36% for MoS2 , and from 6% to 17% for the CoMo catalyst. These results showed that there were differences among the three catalysts as the CoMo catalyst would be more selective for pentenes. As to the NiMo catalyst, no 2MTHT was observed but this result would be in accordance with the literature where it is well known that NiMo catalysts are more hydrogenating [39]. The 2MTHT intermediate would be quickly transformed into pentenes, which would explain the higher amount of pentane observed for the NiMo catalyst compared to CoMo. 3.2. Transformation of 23DMB2N alone or in mixture The transformation of 23DMB2N alone has also been studied (Fig. 3 [19]). Regarding the olefinic feed alone, the synergy coef-
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
ficient was higher on NiMo than on the CoMo or MoS2 catalyst. When no H2 S was added to the feed, the synergy coefficient for the Ni promoted catalyst was higher than for MoS2 (corresponding to an activity 37 times higher for NiMo), CoMo being 9 times more active than the unpromoted catalyst. When H2 S was added to the feed, a spectacular decrease of the synergy coefficients (and thus the activities) could be observed on the NiMo catalyst, as 200 ppm of sulphur were enough to induce an important decrease compared to the two other catalysts. NiMo catalyst finally lost 60% of its activity with 1000 ppm of H2 S in the feed. The activity of the CoMo catalyst slightly decreased, while MoS2 was still the less active. While studying the impact of H2 S on the catalysts with the full model mixture, the synergy coefficients were much lower for the promoted catalysts than when the feed contained the olefins alone. Indeed for the NiMo catalyst, while using the olefinic feed, an important decrease could be observed when the amount of H2 S increased (synergy coefficient from 37 to 12) followed by a stabilisation. In the case of the model mixture, the synergy coefficient was already very low (about 5) without any added H2 S and seemed to decrease slightly when PH2 S increased. The CoMo catalyst seemed to be a little bit more active than the NiMo. With 1000 ppm of H2 S in the feed, the relative activities for the promoted catalysts were about 5 and 2 times higher for CoMo and NiMo, respectively compared to MoS2 , showing the strong impact of 2MT on HYD reaction. These experiments confirmed the promoting effect of nickel and cobalt for the hydrogenation of the 23DMB2N under hydrotreating (HDT) conditions, this synergetic effect being more pronounced with nickel, as previously observed in other reactions [38] such as HDS and HYD of a commercial FCC naphtha [20], hydrogenation of propene and cyclopentene [39]. However, the HYD activity decrease was more significant for NiMo than CoMo and MoS2 and in particular the NiMo catalyst was much more sensitive to the presence of H2 S at low concentrations (200 ppm of sulphur). This negative impact of H2 S, more pronounced for the NiMo catalyst, could be explained by a decrease of the Mo–S bond energy due to the presence of Ni, inducing the creation of more acidic sites and therefore an increase of adsorption of the SH− species, as already mentioned by several authors [19,22]. Lamic et al. [19] also showed that the promoting effect of nickel was more important than cobalt, thus confirming that the NiMo catalyst was more hydrogenating but also more sensitive to PH2 S .
193
Fig. 4. Transformation of 2MT over various unsupported monometallic sulphides with model feed (×). HDS activity versus monometallic bulk sulphur–metal bond energy E(MS) (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Impact of H2 S on the HDS activity for three catalysts (NiMo, CoMo and Mo) (Model feed: () 500 ppm and () 1000 ppm H2 S added, Thiophenic feed: () 0 ppm, (+) 500 ppm and (ⵦ) 1000 ppm H2 S added).
3.3. Impact of sulphur compound nature on the relation between HDS and HYD activities versus E(MS)
MoS2 catalytic systems were added to the data previously obtained (Figs. 4 and 5). The addition of H2 S (500 and 1000 ppm) did not change the shape of the volcano curves, for HDS or HYD in the range of E(MS) of the three catalysts (NiMo, CoMo and MoS2 ). Fig. 4 shows that no matter the feed and whatever the amount of H2 S, NiMo, CoMo and MoS2 catalysts fitted into the volcano shaped curves. Two different feeds were used (thiophenic and model mixture) with several amounts of H2 S (0, 500 and 1000 ppm added) and the global shape of the volcano curve was not modified. It is interesting to notice that under these operating conditions the olefin has a very little influence on the transformation of 2MT. On the contrary, a strong effect of 2MT could be observed on the HYD volcano shaped curve compared to H2 S effect, but only for high E(MS) (Fig. 5). The difference could be explained by a stronger adsorption of 2MT than H2 S, these sulphur molecules being in competitive adsorption with the olefin on the catalyst. The competitive adsorption on the same sites would be in favour of 2MT, since its covering level on the catalyst surface would be higher than for the olefin. This result was confirmed by theoretical calculations of the adsorption constants done by Krebs et al. [3] who showed that Kads (2MT) = −145 kJ/mol was inferior to Kads (23DMB2N) = −107 kJ/mol on the S-edges, confirming that 2MT was more strongly adsorbed on the catalyst. The high E(MS) would
In order to go further in the interpretation, the impact of the sulphur compound nature on the results previously found by Daudin et al. [11,14] have been observed. The authors obtained volcano type relationships between the ab initio calculated metal–sulphur bond energies E(MS) of various TMS (transition metal sulphides) and the activity in the hydrodesulphurisation of 2MT (HDS) on the one hand, and the hydrogenation of 23DMB2N (HYD) on the other hand (Figs. 4 and 5, plain and dotted lines). As previously described [11,14] for the HDS reaction (Fig. 4), the lowest activities corresponded to Ni3 S2 which had the lowest E(MS) and to MoS2 which had the highest E(MS). The maximum in activity was obtained with the Rh2 S3 and NiMo catalysts, which have intermediate E(MS). As to the olefin hydrogenation reaction (HYD), the same observations were made (Fig. 5), only this time the maximum of activity was for the Rh2 S3 catalyst. It was interesting to observe that CoMo and NiMo promoted catalysts had intermediate E(MS) which were consistent with the high activity observed for both HDS and HYD. These results show that it is possible to generalise the volcano curves to bimetallic systems, thus to conventional hydrotreatment catalysts. The values obtained in the present work with the CoMo, NiMo and
Fig. 5. Transformation of 23DMB2N. Olefinic feed alone over various unsupported monometallic sulphides ( and plain line) and model mixture ( and dotted line). HYD activity versus monometallic bulk sulphur–metal bond energy E(M–S) (T = 523 K, P = 2 MPa, H2 /feed = 360L/L). Impact of H2 S on the HYD activity for three catalysts (NiMo, CoMo and Mo) (: 500 ppm H2 S added and ×: 1000 ppm S added).
194
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
system might be more selective than the CoMo catalyst in our specific experimental conditions. These observations on HDS, HYD and selectivity should be confirmed for catalysts with lower E(MS) in order to obtain a better description of the relationship. Nevertheless, CoMo and NiMo catalysts are the most used catalysts in that kind of application. 4. Conclusion
Fig. 6. Transformation of a model feed. Selectivity HDS/HYD over various unsupported monometallic sulphides () (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Impact of H2 S on the selectivity for three catalysts (NiMo, CoMo and Mo) (: 500 ppm H2 S added and ×: 1000 ppm S added).
also lead to a stronger adsorption of 2MT thus limiting the adsorption of the olefin and its hydrogenation. These results highlight the competitive adsorption between the olefin and the sulphur compounds which seems to depend on the compound nature. 3.4. Impact of sulphur compound nature on the relation between selectivity (HDS/HYD) and E(MS) The influence of the amount of H2 S in the feed was also studied for the HDS/HYD selectivity criteria (Fig. 6). A bell shaped curve was also found for the selectivity. These results were obtained for a same conversion of 2MT and olefins (HDS yield of about 30–40 mol% and alkene yield of about 15 mol%) and calculated as the ratio of HDS and HYD. Surprisingly, the unsupported NiMo catalyst was found to be the most selective under these conditions, which is quite unusual in gasoline selective HDS [2,40]. As previously observed for HDS and HYD activities, there was no major change in the bell-shaped curve when H2 S was added to the model feed. Regarding the selectivity, NiMo catalyst would be more selective than CoMo, especially depending on the amount of sulphur in the feed. Krebs et al. [3,13,41] showed that the adsorption selectivity of the reactants (determined by DFT calculations) was representative of the evolution of the selectivity HDS/HYD. More precisely, they pointed out on CoMo and NiMo catalysts that the Co content on the edges hardly influenced the adsorption selectivity (of thiophene derivatives) for CoMo, while the Ni edge content was a key parameter for NiMo. Therefore, a partial decoration of the edges (maximising mixed Ni–Mo sites) should increase the adsorption selectivity of thiophene derivatives on the NiMo active phase and thus increase the selectivity. They also showed that the S-edge was more selective than the M-edge (regarding the calculated adsorption constants for all molecules). It means that the selectivity HDS/HYD would increase when increasing the ratio S-edge/M-edge. Experimentally the amount of NiMoS or CoMoS phases (called “promotion rate”) is accessible by X-ray photoelectron spectroscopy for supported and unsupported catalysts [19,42,43]. For the unsupported NiMo catalyst [19], the promotion rate is quite low (45%) and could explain a partial decoration of the edges and thus an increase in selectivity if compared to the supported transition metal sulphides (promotion rate of 55–60% [43]). Moreover, on the NiMo catalyst, the amount of S-edges would increase with the quantity of H2 S [13], which would lead to an increase in selectivity. However, the amount of Ni on the edges would also increase, then decreasing the number of mixed Ni–Mo sites. These trends highlight that NiMo catalysts are much more sensitive to operational conditions and this could explain why this
These results confirmed that H2 S was an important parameter and had a negative impact in the transformation of model molecules representative of FCC gasoline as reported previously in the treatment of other feeds. However, it was the first time that H2 S was studied for the hydrogenation (HYD) and the hydrodesulphurisation (HDS) reactions separately and in mixture. It has been shown under these operating conditions that the presence of 2methylthiophene in the feed induced a stronger inhibiting effect than H2 S on the HYD reaction. The promoting effect of Co and Ni for HDS and HYD reactions was confirmed and the NiMo catalyst appeared to be the most efficient. However, NiMo catalysts performances are also more sensitive to operational conditions. These results, using unsupported Mo-based catalysts, seemed to fit the previous ones obtained on supported transition metal sulphides where the activity and selectivity were linked to the metal–sulphur bond energy (E(MS)). The impact of H2 S was also observed on the activity and the selectivity but it appeared not to affect the trends of the volcano-shaped curves obtained in previous works. References [1] Off. J. Eur. Commun. L76 (2003) 10–19. [2] S. Brunet, D. Mey, G. Pérot, C. Bouchy, F. Diehl, Appl. Catal. A: Gen. 178 (2005) 143–172. [3] E. Krebs, B. Silvi, A. Daudin, P. Raybaud, J. Catal. 260 (2008) 276–287. [4] S. Hatanaka, Catal. Surv. Asia 9 (2005) 87–93. [5] S. Hatanaka, O. Sadakane, US Patent 6,120,679, Mitsubishi Oil Corporation, 2000. [6] T.G. Kaufmann, A. Kaldor, G.F. Stuntz, M.C. Kerby, L.L. Ansell, Catal. Today 62 (2000) 77–90. [7] C. Marcilly, Stud. Surf. Sci. Catal. 135 (2001) 37–60. [8] C. Song, X. Ma, Appl. Catal. B: Environ. 41 (2003) 207–238. [9] I.V. Babich, J.A. Moulijn, Fuel 82 (2003) 607–631. [10] H. Topsøe, B.S. Clausen, F.E. Massoth, Hydrotreating Catal. Sci and Tech., Springer Verlag, 1996, 11. [11] A. Daudin, A.F. Lamic, G. Pérot, S. Brunet, P. Raybaud, C. Bouchy, Catal. Today 130 (2008) 221–230. [12] J.T. Miller, W.J. Reagan, J.A. Kaduck, J. Kropf, J. Catal. 193 (2000) 123–131. [13] E. Krebs, B. Sylvi, P. Raybaud, Catal. Today 130 (2008) 160–169. [14] A. Daudin, S. Brunet, G. Perot, P. Raybaud, C. Bouchy, J. Catal. 248 (2007) 111–119. [15] N. Guernalec, C. Geantet, P. Raybaud, T. Cseri, M. Aouine, M. Vrinat, Oil Gas Sci. Technol.-Rev. IFP 61 (2006) 515–535. [16] D. Mey, S. Brunet, C. Canaff, F. Maugé, C. Bouchy, F. Diehl, J. Catal. 227 (2004) 436–447. [17] P.S.E. Dai, D.E. Sherwood Jr., R.H. Petty, US Patent 5,340,466, Texaco, 1994. [18] V. Rabarohoela-Rokotovao, S. Brunet, G. Pérot, F. Diehl, Appl. Catal. A: Gen. 306 (2006) 34–44. [19] A.-F. Lamic, A. Daudin, S. Brunet, C. Legens, C. Bouchy, E. Devers, Appl. Catal. A: Gen. 344 (2008) 198–204. [20] P. Ghosh, A.T. Andrews, R.J. Quann, T.R. Halbert, Energy Fuels 23 (2009) 5743–5759. [21] N. Dos Santos, H. Dulot, N. Marchal, M. Vrinat, Appl. Catal. A: Gen. 352 (2008) 114–123. [22] M. Badawi, L. Vivier, G. Pérot, D. Duprez, J. Mol. Catal. A: Chem. 293 (2008) 53–58. [23] G. Berhault, A. Mehta, A.C. Pavel, J. Yang, L. Redon, M.J. Yacaman, L.C. Araiza, A.D. Moller, R.R. Chianelli, J. Catal. 198 (2001) 9–19. [24] Y. Iwata, K. Sato, T. Yoneda, Y. Miki, Y. Sugimoto, A. Nishijma, H. Shimada, Catal. Today 45 (1998) 353–359. [25] S. Fuentes, G. Diaz, F. Fedraza, H. Rojas, N. Rosas, J. Catal. 113 (1988) 535–539. [26] D. Mey, S. Brunet, G. Pérot, F. Diehl, S. Kasztelan, Am. Chem. Soc. Prep. Div. Pet. Chem. 47 (2002) 69–72. [27] Y. Romero, C. Fontaine, S. Brunet, A. Daudin, C. Bouchy, E. Devers, C. Legens, Am. Chem. Soc. Prep. Div. Pet. Chem. 53 (2) (2008) 5–8. [28] P. Raybaud, Appl. Catal. A: Gen. 322 (2007) 76–91. [29] R.R. Chianelli, Catal. Rev.-Sci. Eng. 26 (1984) 361–393.
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195 [30] L.P. Nielsen, S.V. Christensen, H. Topsoe, B.J. Clausen, Catal. Lett. 67 (2000) 81–85. [31] J.V. Lauritsen, J. Kibsgaard, G.H. Olesen, P.G. Moses, B. Hinnemann, S. Helveg, J.K. Nørskov, B.S. Clausen, H. Topsøe, E. Lægsgaard, F. Besenbacher, J. Catal. 249 (2007) 220–233. [32] J.V. Lauritsen, M.V. Bollinger, E. Lægsgaard, K.W. Jacobsen, J.K. Nørskov, B.S. Clausen, H. Topsøe, F. Besenbacher, J. Catal. 221 (2004) 510–522. [33] S. Helveg, J.V. Lauritsen, E. Lægsgaard, I. Stensgaard, J.K. Nørskov, B.S. Clausen, H. Topsøe, F. Besenbacher, Phys. Rev. Lett. 84 (2000) 951–954. [34] J.S. Choi, C. Petit-Clair, D. Uzio, Stud. Surf. Sci. Catal. 143 (2002) 585–594. [35] F. Bataille, J.L. Lemberton, P. Michaud, G. Pérot, M. Vrinat, M. Lemaire, E. Schulz, M. Breysse, S. Kasztelan, J. Catal. 191 (2000) 409–422. [36] M. Egorova, R. Prins, J. Catal. 225 (2004) 417–427.
195
[37] S. Kasztelan, Hydrotreating Technology for Pollution Control, Dekker, New York, 1996, pp. 29–45. [38] R. Candia, O. Sorensen, J. Villadsen, N.-Y. Topsoe, B.S. Clausen, H. Topsoe, Bull. Soc. Chim. Belg. 93 (1984) 763–766. [39] M. Brémaud, L. Vivier, G. Pérot, V. Harlé, C. Bouchy, Appl. Catal. A: Gen. 289 (2005) 44–50. [40] M. Toba, Y. Miki, T. Matsui, M. Harada, Y. Yoshimura, Appl. Catal. B: Environ. 70 (2007) 542–547. [41] E. Krebs, A. Daudin, P. Raybaud, Oil Gas Sci. Technol.-Rev. IFP 64 (2009) 707–718. [42] A.D. Gandubert, C. Legens, D. Guillaume, S. Rebours, E. Payen, Oil Gas Sci. Technol.-Rev. IFP 62 (2007) 79–89. [43] K. Marchand, C. Legens, D. Guillaume, P. Raybaud, Oil Gas Sci. Technol.-Rev. IFP 64 (2009) 719–730.
Contents lists available at ScienceDirect
Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata
Insight into sulphur compounds and promoter effects on Molybdenum-based catalysts for selective HDS of FCC gasoline Celine Fontaine a,∗ , Yilda Romero a , Antoine Daudin b , Elodie Devers b , Christophe Bouchy b , Sylvette Brunet a a b
UMR CNRS 6503, Catalyse en Chimie Organique, Université de Poitiers, Faculté des Sciences Fondamentales et Appliquées, 40 avenue du Recteur Pineau, 86022 Poitiers Cedex, France IFP Energies nouvelles, Direction Catalyse et Séparation, Rond-point de l’échangeur de Solaize, BP 3, 69360 Solaize, France
a r t i c l e
i n f o
Article history: Received 6 April 2010 Received in revised form 23 August 2010 Accepted 24 August 2010 Available online 23 September 2010 Keywords: Hydrodesulphurisation Hydrogenation FCC gasoline Molybdenum sulphide-based catalyst Nickel Cobalt Selectivity Activity E(M-S) HDS HYD
a b s t r a c t The effect of the nature of sulphur compounds (H2 S and 2-methylthiophene) over unsupported molybdenum-based sulphide catalysts (promoted or not by nickel or cobalt) for the transformation of a model FCC feed (hydrodesulphurisation of 2-methylthiophene and hydrogenation of 2,3-dimethylbut2-ene) was investigated. The activities of the various catalysts were compared between each reactant alone and the full model mixture. The promoting effect of Co and Ni is found for both hydrodesulphurisation (HDS) and hydrogenation of olefins (HYD) reactions. However, the hydrogenation activity strongly depends on the amount of sulphur compound present in the feed. Nickel, used as a promoter, seems more sensitive than cobalt, particularly for the hydrogenation reaction. The presence of 2-methylthiophene in the feed induces a stronger inhibiting effect than H2 S on HYD reaction. This result highlights the different adsorption constant ratio between olefins and sulphur molecules depending on the sulphur compounds. Even though an impact of H2 S can be observed on the NiMo and CoMo catalysts, no major modification occurs on the volcano-shaped curves obtained in previous work, in terms of activity or selectivity, as a function of the metal–sulphur bond energy. Whatever the amount of H2 S added, NiMo and CoMo catalysts fit into the curve, NiMo being the most active and the most selective under these conditions. © 2010 Elsevier B.V. All rights reserved.
1. Introduction In order to meet the always more severe European regulations regarding the amount of sulphur in gasoline [1], hydrotreating catalysts must achieve a deep hydrodesulphurisation (HDS) with a minimum of olefins hydrogenation (HYD), i.e. get to a better selectivity (hydrodesulphurisation/olefins hydrogenation: HDS/HYD) criteria. There has been many studies dealing with HDS and focusing on the understanding of the various phenomenon involved in this reaction, in order to improve the selectivity [2–5]. To achieve such a goal, the optimisation of conventional catalysts or the elaboration of new ones are required. Processes and catalysts allowing the removal of sulphur must be more and more efficient. Commercial gasoline is composed of different fractions coming from reforming, isomerisation and fluid catalytic cracking (FCC) units. The gasoline fraction produced from the FCC process represents 30–50% of the commercialised motor fuel but it contains up to 85–95% of the sulphur impurities. It is mainly composed of aromatics (30 vol.%), isoalkenes (20–40 vol.%) and sulphur compounds
∗ Corresponding author. Tel.: +33 5 49 45 38 39; fax: +33 5 49 45 38 97. E-mail address: [email protected] (C. Fontaine). 0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.apcata.2010.08.049
such as alkylthiophenes (max. 5000 wt ppm) [6–9]. The nature of sulphur compounds is now well-known and has been reviewed by Brunet et al. [2]: mainly benzothiophenes, alkylthiophenes, thiols and sulphides. Besides, the olefins amount is high (20–40 vol.%) and helps to improve the octane number. Therefore, it is important to remove the sulphur compounds while limiting the olefins hydrogenation in order to keep a high octane number (selective HDS). Conventional operating conditions are quite mild in terms of temperature and pressure. Catalysts used for such hydrotreating reactions are transition metal sulphides (TMS), supported or not [10]. Brunet et al. [2] have reviewed the different parameters influencing the activity and the selectivity such as the presence of a promoter, the reactions and mechanisms involved in HDS reactions. It is known that there is a competitive competition between the sulphur and the olefin compounds on the catalyst depending on the catalyst nature, which could modify the activity and the selectivity. Daudin et al. [11] and Miller et al. [12] showed that mixed unsupported sulphides (NiMo and CoMo) had a much better HDS activity and HDS/HYD selectivity than the non-promoted catalyst MoS2 under model conditions. Recently, experimental results were coupled with theoretical models (mainly DFT calculations) in order to explain more precisely the mechanisms involved in HDS reactions [3,13–16]. Daudin et al.
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
[11,14] have shown that there was a relation between the ab initio calculated metal–sulphur bond energy (E(MS)) of unsupported transition metal sulphides and their activity for the HDS and HYD reactions. They generalised the volcano-shaped curve to unsupported transition metal catalysts, showing that the maximum of activity (HDS and HYD) or selectivity (HDS/HYD) could be obtained for intermediate E(MS). For supported catalysts, it has been shown that a more basic support would account for a higher selectivity [2,16,17]. For example, Mey et al. [16] showed a decrease in HYD on a CoMo supported catalyst doped with potassium. Another way of improving the selectivity was also to selectively poison the hydrogenation sites [4,5]. When the hydrogenation activity was inhibited by carbon deposition or adsorption of basic nitrogen compounds such as pyridine, the activity of the catalyst for the HYD reaction was more affected than the HDS, thus increasing the selectivity. Indeed, HDS/HYD selectivity is the key parameter for an efficient gasoline hydrotreating catalyst. As explained above, selectivity can be influenced by many parameters (nature of the active phase, acidic–basic properties of the support, use of selective poison) but might also depend on the amount of H2 S in the feed. The influence of H2 S in the feed has been the subject of numerous works as it is obtained as a by-product in HDS reactions [2,18]. In particular the influence of H2 S has been studied on diesel HDS [18], olefins hydrogenation reaction [19] and very recently on model FCC feeds [16,20]. It has been admitted that Mo-based catalysts promoted by Ni are more affected by the presence of H2 S than catalysts promoted by Co while studying the HDS of dibenzothiophene or 4,6-dimethyldibenzothiophene, but the results depended on the experimental conditions [18]. Dos Santos et al. [21] used a commercial CoMo/Al2 O3 catalyst under mild conditions of temperature and pressure with a model FCC feed (3-methylthiophene and hex-1-ene diluted in n-heptane), and showed that there was an inhibiting effect of H2 S on both HDS and HYD reactions and an effect of PH2 S on the selectivity explained by a variation in the rate determining step. Only a few papers are available in the literature dealing with the influence of sulphur compounds (H2 S) on the olefins alone [19,21,22]. Lamic et al. [19] showed the negative impact of H2 S on different unsupported molybdenum-based catalysts for the transformation of a model olefin (2,3-dimethylbut-2-ene), the NiMo catalyst being more inhibited than the CoMo or the MoS2 one by small amounts of H2 S. Dos Santos et al. [21] discussed a model for the competitive reactions happening during the deep HDS of FCC gasoline, using a CoMo/Al2 O3 catalyst and under model conditions, and found that there was a competitive adsorption of 3-methylthiophene and hex-1-ene on the catalyst surface as well as an inhibiting effect of H2 S on both reactions (HDS and HYD). For the HYD reaction, the authors showed that the olefin and the sulphur molecule were in competitive adsorption on the same active site, and that a strong influence of the olefin was shown on the HDS reaction. This paper will deal with the transformation of model molecules, either alone or in mixture, representative of sulphur and olefinic compounds found in FCC gasoline. The influence of PH2 S on the performances of unsupported Mo-based catalysts will be described in terms of HDS and HYD activities and HDS/HYD selectivity. The influence of the nature of the promoter, cobalt or nickel, will be particularly studied. The impact on the olefin reactivity of two sulphur compounds: H2 S, already known as an important parameter for hydrotreatment reactions, and 2-methylthiophene, one of the reactants, will be addressed. This study will be extended to previous results correlating reactivity as a function of the metal–sulphur bonding energy (E(MS)) in order to observe the impact of the sulphur compound nature on the volcano-shaped curves established for transition metal sulphides.
189
2. Experimental 2.1. Catalysts preparation The preparation method of unsupported molybdenum sulphide catalysts has already been reported in previous works [11,14,19–24]. Molybdenum sulphide preparation was carried out by thermal decomposition of ammonium thiomolybdate at 673 K under H2 S/H2 (10 mol% H2 S) during 4 h. First, the ammonium tetrathiomolybdate (NH4 )2 MoS4 (ATM) was obtained by reaction between ammonium heptamolybdate (NH4 )6 [Mo7 O24 ], 4H2 O (4 g in 20 cm3 of distilled water) with 50 wt% of ammonium sulphide (NH4 )2 S in aqueous solution at 353 K. The ATM precursor precipitated as red crystals by cooling down the solution in iced water during 3 h. Then, the precipitated red crystals were thoroughly washed with isopropanol and dried. Molybdenum-based materials promoted by nickel or cobalt in order to obtain NiMo and CoMo catalysts were prepared according to the method reported by Fuentes et al. [25] and are the same than those used by Lamic et al. [19]. (NH4 )2 MoS4 (prepared as mentioned below) was used as a support to impregnate Co(NO3 )2 or Ni(NO3 )2 diluted in acetone in order to obtain an atomic ratio of Ni(Co)/(Ni(Co) + Mo) = 0.3. The bimetallic organosulphides precursors were obtained by removing the solvent at room temperature. Finally, NiMo and CoMo catalysts were prepared by thermal decomposition of the corresponding bimetallic organosulphide precursor at 673 K under H2 S/H2 (10 mol% H2 S) flow during 4 h. All catalysts were characterised before and after catalytic activity measurements by TEM combined with EDX (Philips CM 120 kV), X-ray diffraction (Bruker D5005), BET surface area (Micromeritics ASAP 2010) and elemental analysis (CE Instruments NA2100 Protein) at the University of Poitiers (LACCO) in order to verify the exact nature of the transition metal sulphide (TMS) and that there was no modification of the catalyst during the test, according to Lamic et al. [19,27]. XPS analyses have also been performed at IFP-Lyon and are reported elsewhere [19] and confirmed the formation of the desired sulphide catalysts. Characterisation of the catalysts before and after catalytic test (Tables 1 and 2 [19]) showed a decrease in the SBET especially for the MoS2 catalyst (from 76 to 10 m2 g−1 ) while the promoted catalysts were less affected (from 13 to 7 m2 g−1 for CoMo and from 14 to 12 m2 g−1 for NiMo). The particles size is small for all catalysts (8, 9 and 6 nm for MoS2 , NiMo and CoMo, respectively) and the size of the particles was almost not modified after catalytic test (8, 8 and 5 nm for MoS2 , NiMo and CoMo, respectively). Quantitative analyses of XPS are reminded in Table 2, showing that the ratios of Co and Ni in the mixed phase were found to be different: only 18% of Co was engaged in CoMoS phase instead of 45% for the nickel. Consequently, the NiMoS phase seems to be more favoured than CoMoS phase under sulpho-reductive conditions. It was also shown that the global sulphidation level was very high for all catalysts. For the CoMo catalyst the value was about 95% whereas it was 80% for NiMo in accordance with the NiOx presence detected by XPS. 2.2. Activity measurements Catalytic activity measurements were carried out in a fixedbed reactor at 523 K under a total pressure of 2 MPa. The catalyst was presulphided at 673 K for 10 h with a mixture of 10 mol% H2 S/H2 under atmospheric pressure and then cooled down to the reaction temperature in the presence of the sulphiding mixture. The desired reaction conditions were adjusted (T = 523 K, P = 2 MPa, H2 /feed = 360L/L) and the model feed was injected into the reactor. Three feeds were studied in order to compare the results for each reactant taken separately or in mixture:
190
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
Table 1 Physico-chemical properties of the unsupported molybdenum sulphide-based catalysts determined from elemental analysis (S/Metal atomic ratio (Me = Mo or Mo + Ni(Co)) and Ni(Co)/(Ni(Co) + Mo) atomic ratio), specific surface area (before and after reaction) and particles size (before and after reaction) [19]. Sulphides
S/metal (ICP)
Ni(Co)/(Ni(Co) + Mo) atom (before reaction)
TMS
SBET (m2 g−1 )
SBET after reaction (m2 g−1 )
Particles size (nm)
Particles size after reaction (nm)
Mo NiMo CoMo
2.0 1.7 1.9
– 0.25 0.27
MoS2 NiMoS, MoS2 , NiS CoMoS, MoS2 , Co9 S8
76 14 13
10 12 7
8 9 6
8 8 5
Table 2 XPS characterisation of unsupported molybdenum sulphide-based catalysts: binding energy (eV) of the various elements (S 2p3/2 , Mo 3d5/2 , Ni 2p3/2 and Co 2p3/2 ) from [19]. Relative contribution (in brackets) of the different phases (NiS, NiMoS, NiOx , Co9 S8 , CoMoS).
MoS2 CoMo NiMo
S 2p3/2 (eV)
Mo 3d5/2 (eV)
162.2 162.2 162.2
229.4 229.4 229.4
Ni 2p3/2 (eV)
Co 2p3/2 (eV)
NiS
NiMoS
NiOx
Co9 S8
CoMoS
– – 852.9 (34%)
– – 853.9 (45%)
– – 856.0 (21%)
– 778.4 (82%) –
– 779.1 (18%) –
• The thiophenic feed, composed of 0.3 wt% (1000 wt ppm S) of 2-methylthiophene (2MT) in n-heptane (solvent); this sulphur molecule is representative of the sulphur compounds present in the real FCC gasoline feedstock, • The olefinic feed, with 20 wt% of 2,3-dimethylbut-2-ene (23DMB2N) in n-heptane; representing the branched olefins existing in the FCC gasoline feedstock, • The full model mixture was composed of 0.3 wt% of 2MT, 20 wt% of 23DMB2N and 30 wt% of ortho-xylene, representative of an aromatic molecule, in n-heptane. The residence time has been varied from 0.5 to 80 s in order to compare the catalysts selectivity at a same conversion.
The partial pressure of H2 S varied from 0 to 2 kPa corresponding to the equivalent of 1000 wt ppm S maximum in the feed. The concentration of H2 S added to the feed was monitored by varying the ratio (H2 S–H2 )/pure H2 using a 1 mol% H2 S in H2 cylinder. Experiments were carried out as followed: different amounts of sulphur from 1000 wt ppm to 0 ppm were successively injected and back points at 1000 wt ppm were performed for all catalysts. In order to observe the impact of the addition of H2 S on the transformation of 2MT, the HDS yield was maintained at about 15 mol% for the experiments using the thiophenic molecule alone or in mixture. When the transformation of the olefins was studied, the conversion of the alkenes was also set to 15 mol% alone or in mixture (corresponding under these experimental conditions to an HDS yield of about 30–40 mol%). Ortho-xylene and n-heptane were not transformed under these experimental conditions according to previous works [16]. The reaction products were injected on-line by means of an automatic sampling valve into a 3380 Varian gas chromatograph equipped with a PONA capillary column, a flame ionisation detector and a cryogenic system. The identification of the products was performed by GC–MS coupling (Table 3). Preliminary work has shown that the response factors between the sulphur compounds and the hydrocarbons from the transformation of 2MT and olefins were close and could be considered equal to 1 for all products. This is
Scheme 1. Transformation of 23DMB2N over sulphide catalysts. 23DMB2N: 2,3-dimethylbut-2-ene, 23DMB1N: 2,3-dimethylbut-1-ene, 23DMB: 2,3dimethybutane.
why the area of the GC peaks has been directly linked to the molar concentration of the compounds. The detailed mechanism for the transformation of 2MT is already well known ([10] and references therein) and the reaction products are shown in Table 3. The main products are pentenes, pentane, small quantities of cracking products and 2methyltetrahydrothiophene (2MTHT). These products have already been described by several authors [10,16]. The detailed reaction scheme for the transformation of 23DMB2N has also been reported and discussed in the literature ([11,16] and references therein). Scheme 1 shows the main reactants and product as a reminder. The main reaction product is 2,3-dimethylbutane (23DMB) in a large majority (>98 mol%), the other products being isomers. The isomerisation of 23DMB2N to 23DMB1N is known to be very fast so that the mixture composed of 23DMB2N and 23DMB1N is considered as one reactant when measuring the hydrogenation activity of the catalysts [16,26]. Hydrogenation activities (HYD) of the catalysts are defined as the number of moles of 2,3-dimethylbutane formed per second
Table 3 Transformation of 2-methylthiophene. Products identified [2]. Hydrodesulphurisation (HDS) products:
Sulphur compound:
C4 , C3 , C<3 , Crack
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
191
per square meter of catalyst considering the specific surface area after reaction and adjusted to the amount of sulphide phases determined by XPS (MoS2 : 100%, NiMo: 80% and CoMo: 100%) [27]. Desulphurised products, resulting from the transformation of 2MT were designated as HDS products. They were measured after catalyst stabilisation and under conditions where a linear relationship between conversion and residence time was obtained (conversion rate below 40 mol%). The selectivity of the reaction is given by the ratio between hydrodesulphurisation and olefin hydrogenation (HDS/HYD) and corrected by the amount of sulphur phase obtained in XPS. HYD and HDS activities, selectivities and the synergy coefficient are calculated as described below. Activity a for the transformation of 2MT or olefins (23DMB1N + 23DMB2N) is presented in Eq. (1) a [mol m−2 s−1 ] =
XF mSs wt% of sulphur phase
(1) (mol s−1 ),
where F is the molar flow of the considered reactant X is the conversion and X[%]= 100 × [1 − Areactant /(Areactant + i Ai )], m is the mass of catalyst (g) and SS is the specific surface of the solid (m2 g−1 ). For a selected reaction (transformation of 2MT or 23DMB2N), the synergy coefficient is the ratio between the activity of the promoted catalyst by Ni or Co and the activity of the unpromoted catalyst (MoS2 ). Finally, the selectivity is calculated as the ratio between the molar yields of HDS and HYD (Eq. (2)) and the molar yield is calculated as shown in Eq. (3): Selectivity =
HDS HYD
HDS or HYD [%] = 100 ×
(2)
A i i
Areactant +
A i i
(3)
where Ai is the GC peak area for the transformation products considered and Areactant is the area of the remaining reactant. 3. Results and discussion The performances of the molybdenum-based sulphides (MoS2 , CoMo and NiMo) were measured for the transformation of 2MT alone, of 23DMB2N alone and of the model feed (a mixture of 2MT, 23DMB2N and ortho-xylene), with different H2 S partial pressures. 3.1. Transformation of 2MT alone or in the model feed Fig. 1 presents the H2 S impact on the transformation of 2MT alone or in the model mixture and especially the synergy coefficient as a function of the total amount of H2 S in the feed. The synergy coefficient is indeed the ratio between the activities of the catalysts with higher activity and the catalyst with the lower activity (which is MoS2 when tested with the full model mixture in all case). The total amount of H2 S represents the quantity of H2 S added to the feed plus the amount of H2 S produced by HDS reaction (about 150 ppm S, as the work has been carried out at about 15 mol% of conversion of the 2MT in the thiophenic or the model feed). As awaited [19], a promoting effect of Co and Ni on the HDS activity compared to MoS2 was found, no matter the feed nature. Using the thiophenic mixture, without H2 S added to the feed, the synergy coefficient of the NiMo and the CoMo catalysts were about 15 and 17 (meaning that their activities were 15 and 17 times higher than MoS2 ). When H2 S was progressively added to the feed, the synergy coefficient decreased for the promoted (NiMo and CoMo) catalysts, showing the negative impact of the sulphur molecule for the reaction. For 1000 wt ppm of added H2 S, the synergy coefficient of the promoted catalysts were only 9 and
Fig. 1. Transformation of 2MT over (䊉) NiMo, () CoMo and () MoS2 alone or in mixture. Synergy coefficient (ratio of the activity between the promoted catalyst (NiMo or CoMo) and the activity of MoS2 catalyst) versus total amount of H2 S (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Thiophenic feed: dotted lines (empty symbols) and full model mixture: plain lines (full symbols).
10, respectively for NiMo and CoMo compared to the unpromoted catalyst (meaning that the activities were 9 and 10 times higher for the promoted catalysts compared to MoS2 ). The same observations could be made when using the model mixture. Indeed, when the amount of H2 S in the feed increased from 0 to 1000 ppm of added H2 S, the synergy coefficient for the NiMo and for the CoMo catalysts dropped, respectively from 18.2 to 12.3 and from 14.4 to 10.8. A slight impact of the nature of the feed (thiophenic feed alone or the model mixture) could be observed mainly on the performances of the NiMo catalyst. The trend of the plot was the same, no matter the feed nature. However, it seemed that whatever the amount of H2 S in the feed, the synergy coefficient was about 1.2 times higher when using the model mixture on the NiMo catalyst. The difference between the nature of feeds was less important with the CoMo or MoS2 catalysts. Previous works on the promoting effect of Co and Ni compared to MoS2 have been published, comparing experimental results and DFT calculations [28]. The authors compared the promoted and unpromoted catalysts in terms of activity (HDS of DBT, HYD of biphenyl or toluene) with a calculated model. The impact of the promoter could be explained by the modification of the electron density of the molybdenum when the promoter was added to the catalyst, leading to a weakening of the Mo metal–sulphur bond energy (E(MS)) [29]. The promoters also seemed to facilitate the formation of vacancies, and would then increase the number of active sites as shown by Nielsen et al. [30]. A change in the shape of the crystallite has also been shown by Lauritsen et al. [31–33] that could explain the change in activity. The authors showed, using scanning tunnelling microscopy (STM) and density functional theory (DFT), that the MoS2 crystallite had either a triangular shape [33] or a hexagonal one depending on the sulphiding conditions of gold supported MoS2 clusters [31]. Co–Mo–S crystallites also adopted a hexagonal shape while the NiMo particles shape depended on the nanoclusters size. The smaller Ni–Mo–S particles are more likely dodecagonal, while the larger Ni–Mo–S particles exhibited a truncated triangular shape similar to that observed for Co–Mo–S nanoclusters [31]. These results were confirmed by Raybaud with DFT calculations [28]. It was seen in the literature by Choi et al. [34] that cobalt enhanced significantly 3-methylthiophene conversion compared to MoS2 catalyst. In the literature, using different experimental conditions, typical of diesel hydrotreatment using DBT and 46DMDBT (models diesel molecules), an inhibiting effect of H2 S was observed on the HDS reactions with NiMo and CoMo catalysts by Rabarohoela-
192
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
Fig. 3. Transformation of 23DMB2N alone or in mixture over (䊉) NiMo, () CoMo and () MoS2 . Synergy coefficient (ratio of the activity between the promoted catalyst (NiMo or CoMo) and the activity of MoS2 catalyst) versus total amount of H2 S (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Olefinic feed: dotted lines (empty symbols) and full model mixture: plain lines (full symbols).
Fig. 2. Transformation of 2MT. Selectivity towards formation of (a) pentane nC5 , (b) pentenes and (c) 2MTHT over (䊉) NiMo, () CoMo and () MoS2 (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L).
Rokotovao et al. [18]. They also showed that their catalyst containing Ni (NiMoP) was more sensitive to H2 S than the CoMoP one. Hence a decrease of the activity was observed when the amount of sulphur in the feed increased, as already seen in other reactions with the same sulphur DBT and 46DMDBT molecules [35,36]. The negative impact of H2 S could be explained by a heterolytic dissociation of H2 S (leading to the formation of H+ and SH− species) on the catalyst surface (M-edges and S-edges) [11,37]. The SH− species would adsorb on the vacancies thus increasing the covering rate of H2 S and therefore leaving less vacancies available for the cleavage of the C–S bonds (involving mainly S-edge), leading to a decrease in activity while PH2 S increased. It was observed that the amount of H2 S in the feed slightly modified the selectivity towards pentane (nC5 ) (Fig. 2a), pentenes (Fig. 2b) (the HDS products of 2MT) and 2-methyl-
tetrahydrothiophene (2MTHT) (the hydrogenation product of 2MT) (Fig. 2c). Generally, whatever the amount of H2 S in the feed, the selectivity towards pentane was higher for the MoS2 than for the promoted catalysts, the formation of pentenes was the lowest and more 2MTHT was formed on MoS2 than on the other two. The opposite could be observed for the CoMo catalyst regarding pentane and pentenes with a lower formation of 2MTHT. As to the NiMo catalyst, the selectivities were in between the two other catalysts for pentane and pentenes selectivity, while no 2MTHT was observed. For the pentane formation (Fig. 2a), two different catalysts behaviours were observed. When the amount of H2 S increased, the selectivity towards pentane (nC5 ) decreased on MoS2 while it slightly increased on both promoted catalysts. At first, when no H2 S was added to the feed, the selectivity of the alkane was higher for the unpromoted catalyst (62%) compared to NiMo (41%) and CoMo (20%). However, for 1000 ppm of added H2 S, the selectivity using MoS2 or NiMo was the same (46%) and was higher than the one obtained on CoMo for the production of pentane. The selectivity towards pentenes was shown in Fig. 2(b) and when the amount of H2 S added to the feed increased, the selectivity using MoS2 increased while it slightly decreased on both promoted catalysts. When no sulphur compound was added to the feed, the selectivity was thus higher on CoMo (75%) followed by NiMo (59%) and finally the MoS2 catalyst (9%). When the amount of H2 S increased, the trend for the three catalysts remained the same. Finally, in Fig. 2(c) the selectivity towards 2MTHT (sulphur compound issued of the hydrogenation of 2MT) was reported as a function of the total amount of H2 S present in the feed for the three catalysts. This compound was never observed on the NiMo catalyst whatever the amount of H2 S in the feed. For the two other catalysts though, the amount of 2MTHT increased when H2 S increased to 1000 ppm, from 24% to 36% for MoS2 , and from 6% to 17% for the CoMo catalyst. These results showed that there were differences among the three catalysts as the CoMo catalyst would be more selective for pentenes. As to the NiMo catalyst, no 2MTHT was observed but this result would be in accordance with the literature where it is well known that NiMo catalysts are more hydrogenating [39]. The 2MTHT intermediate would be quickly transformed into pentenes, which would explain the higher amount of pentane observed for the NiMo catalyst compared to CoMo. 3.2. Transformation of 23DMB2N alone or in mixture The transformation of 23DMB2N alone has also been studied (Fig. 3 [19]). Regarding the olefinic feed alone, the synergy coef-
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
ficient was higher on NiMo than on the CoMo or MoS2 catalyst. When no H2 S was added to the feed, the synergy coefficient for the Ni promoted catalyst was higher than for MoS2 (corresponding to an activity 37 times higher for NiMo), CoMo being 9 times more active than the unpromoted catalyst. When H2 S was added to the feed, a spectacular decrease of the synergy coefficients (and thus the activities) could be observed on the NiMo catalyst, as 200 ppm of sulphur were enough to induce an important decrease compared to the two other catalysts. NiMo catalyst finally lost 60% of its activity with 1000 ppm of H2 S in the feed. The activity of the CoMo catalyst slightly decreased, while MoS2 was still the less active. While studying the impact of H2 S on the catalysts with the full model mixture, the synergy coefficients were much lower for the promoted catalysts than when the feed contained the olefins alone. Indeed for the NiMo catalyst, while using the olefinic feed, an important decrease could be observed when the amount of H2 S increased (synergy coefficient from 37 to 12) followed by a stabilisation. In the case of the model mixture, the synergy coefficient was already very low (about 5) without any added H2 S and seemed to decrease slightly when PH2 S increased. The CoMo catalyst seemed to be a little bit more active than the NiMo. With 1000 ppm of H2 S in the feed, the relative activities for the promoted catalysts were about 5 and 2 times higher for CoMo and NiMo, respectively compared to MoS2 , showing the strong impact of 2MT on HYD reaction. These experiments confirmed the promoting effect of nickel and cobalt for the hydrogenation of the 23DMB2N under hydrotreating (HDT) conditions, this synergetic effect being more pronounced with nickel, as previously observed in other reactions [38] such as HDS and HYD of a commercial FCC naphtha [20], hydrogenation of propene and cyclopentene [39]. However, the HYD activity decrease was more significant for NiMo than CoMo and MoS2 and in particular the NiMo catalyst was much more sensitive to the presence of H2 S at low concentrations (200 ppm of sulphur). This negative impact of H2 S, more pronounced for the NiMo catalyst, could be explained by a decrease of the Mo–S bond energy due to the presence of Ni, inducing the creation of more acidic sites and therefore an increase of adsorption of the SH− species, as already mentioned by several authors [19,22]. Lamic et al. [19] also showed that the promoting effect of nickel was more important than cobalt, thus confirming that the NiMo catalyst was more hydrogenating but also more sensitive to PH2 S .
193
Fig. 4. Transformation of 2MT over various unsupported monometallic sulphides with model feed (×). HDS activity versus monometallic bulk sulphur–metal bond energy E(MS) (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Impact of H2 S on the HDS activity for three catalysts (NiMo, CoMo and Mo) (Model feed: () 500 ppm and () 1000 ppm H2 S added, Thiophenic feed: () 0 ppm, (+) 500 ppm and (ⵦ) 1000 ppm H2 S added).
3.3. Impact of sulphur compound nature on the relation between HDS and HYD activities versus E(MS)
MoS2 catalytic systems were added to the data previously obtained (Figs. 4 and 5). The addition of H2 S (500 and 1000 ppm) did not change the shape of the volcano curves, for HDS or HYD in the range of E(MS) of the three catalysts (NiMo, CoMo and MoS2 ). Fig. 4 shows that no matter the feed and whatever the amount of H2 S, NiMo, CoMo and MoS2 catalysts fitted into the volcano shaped curves. Two different feeds were used (thiophenic and model mixture) with several amounts of H2 S (0, 500 and 1000 ppm added) and the global shape of the volcano curve was not modified. It is interesting to notice that under these operating conditions the olefin has a very little influence on the transformation of 2MT. On the contrary, a strong effect of 2MT could be observed on the HYD volcano shaped curve compared to H2 S effect, but only for high E(MS) (Fig. 5). The difference could be explained by a stronger adsorption of 2MT than H2 S, these sulphur molecules being in competitive adsorption with the olefin on the catalyst. The competitive adsorption on the same sites would be in favour of 2MT, since its covering level on the catalyst surface would be higher than for the olefin. This result was confirmed by theoretical calculations of the adsorption constants done by Krebs et al. [3] who showed that Kads (2MT) = −145 kJ/mol was inferior to Kads (23DMB2N) = −107 kJ/mol on the S-edges, confirming that 2MT was more strongly adsorbed on the catalyst. The high E(MS) would
In order to go further in the interpretation, the impact of the sulphur compound nature on the results previously found by Daudin et al. [11,14] have been observed. The authors obtained volcano type relationships between the ab initio calculated metal–sulphur bond energies E(MS) of various TMS (transition metal sulphides) and the activity in the hydrodesulphurisation of 2MT (HDS) on the one hand, and the hydrogenation of 23DMB2N (HYD) on the other hand (Figs. 4 and 5, plain and dotted lines). As previously described [11,14] for the HDS reaction (Fig. 4), the lowest activities corresponded to Ni3 S2 which had the lowest E(MS) and to MoS2 which had the highest E(MS). The maximum in activity was obtained with the Rh2 S3 and NiMo catalysts, which have intermediate E(MS). As to the olefin hydrogenation reaction (HYD), the same observations were made (Fig. 5), only this time the maximum of activity was for the Rh2 S3 catalyst. It was interesting to observe that CoMo and NiMo promoted catalysts had intermediate E(MS) which were consistent with the high activity observed for both HDS and HYD. These results show that it is possible to generalise the volcano curves to bimetallic systems, thus to conventional hydrotreatment catalysts. The values obtained in the present work with the CoMo, NiMo and
Fig. 5. Transformation of 23DMB2N. Olefinic feed alone over various unsupported monometallic sulphides ( and plain line) and model mixture ( and dotted line). HYD activity versus monometallic bulk sulphur–metal bond energy E(M–S) (T = 523 K, P = 2 MPa, H2 /feed = 360L/L). Impact of H2 S on the HYD activity for three catalysts (NiMo, CoMo and Mo) (: 500 ppm H2 S added and ×: 1000 ppm S added).
194
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195
system might be more selective than the CoMo catalyst in our specific experimental conditions. These observations on HDS, HYD and selectivity should be confirmed for catalysts with lower E(MS) in order to obtain a better description of the relationship. Nevertheless, CoMo and NiMo catalysts are the most used catalysts in that kind of application. 4. Conclusion
Fig. 6. Transformation of a model feed. Selectivity HDS/HYD over various unsupported monometallic sulphides () (T = 523 K, P = 2 MPa, H2 /feed = 360NL/L). Impact of H2 S on the selectivity for three catalysts (NiMo, CoMo and Mo) (: 500 ppm H2 S added and ×: 1000 ppm S added).
also lead to a stronger adsorption of 2MT thus limiting the adsorption of the olefin and its hydrogenation. These results highlight the competitive adsorption between the olefin and the sulphur compounds which seems to depend on the compound nature. 3.4. Impact of sulphur compound nature on the relation between selectivity (HDS/HYD) and E(MS) The influence of the amount of H2 S in the feed was also studied for the HDS/HYD selectivity criteria (Fig. 6). A bell shaped curve was also found for the selectivity. These results were obtained for a same conversion of 2MT and olefins (HDS yield of about 30–40 mol% and alkene yield of about 15 mol%) and calculated as the ratio of HDS and HYD. Surprisingly, the unsupported NiMo catalyst was found to be the most selective under these conditions, which is quite unusual in gasoline selective HDS [2,40]. As previously observed for HDS and HYD activities, there was no major change in the bell-shaped curve when H2 S was added to the model feed. Regarding the selectivity, NiMo catalyst would be more selective than CoMo, especially depending on the amount of sulphur in the feed. Krebs et al. [3,13,41] showed that the adsorption selectivity of the reactants (determined by DFT calculations) was representative of the evolution of the selectivity HDS/HYD. More precisely, they pointed out on CoMo and NiMo catalysts that the Co content on the edges hardly influenced the adsorption selectivity (of thiophene derivatives) for CoMo, while the Ni edge content was a key parameter for NiMo. Therefore, a partial decoration of the edges (maximising mixed Ni–Mo sites) should increase the adsorption selectivity of thiophene derivatives on the NiMo active phase and thus increase the selectivity. They also showed that the S-edge was more selective than the M-edge (regarding the calculated adsorption constants for all molecules). It means that the selectivity HDS/HYD would increase when increasing the ratio S-edge/M-edge. Experimentally the amount of NiMoS or CoMoS phases (called “promotion rate”) is accessible by X-ray photoelectron spectroscopy for supported and unsupported catalysts [19,42,43]. For the unsupported NiMo catalyst [19], the promotion rate is quite low (45%) and could explain a partial decoration of the edges and thus an increase in selectivity if compared to the supported transition metal sulphides (promotion rate of 55–60% [43]). Moreover, on the NiMo catalyst, the amount of S-edges would increase with the quantity of H2 S [13], which would lead to an increase in selectivity. However, the amount of Ni on the edges would also increase, then decreasing the number of mixed Ni–Mo sites. These trends highlight that NiMo catalysts are much more sensitive to operational conditions and this could explain why this
These results confirmed that H2 S was an important parameter and had a negative impact in the transformation of model molecules representative of FCC gasoline as reported previously in the treatment of other feeds. However, it was the first time that H2 S was studied for the hydrogenation (HYD) and the hydrodesulphurisation (HDS) reactions separately and in mixture. It has been shown under these operating conditions that the presence of 2methylthiophene in the feed induced a stronger inhibiting effect than H2 S on the HYD reaction. The promoting effect of Co and Ni for HDS and HYD reactions was confirmed and the NiMo catalyst appeared to be the most efficient. However, NiMo catalysts performances are also more sensitive to operational conditions. These results, using unsupported Mo-based catalysts, seemed to fit the previous ones obtained on supported transition metal sulphides where the activity and selectivity were linked to the metal–sulphur bond energy (E(MS)). The impact of H2 S was also observed on the activity and the selectivity but it appeared not to affect the trends of the volcano-shaped curves obtained in previous works. References [1] Off. J. Eur. Commun. L76 (2003) 10–19. [2] S. Brunet, D. Mey, G. Pérot, C. Bouchy, F. Diehl, Appl. Catal. A: Gen. 178 (2005) 143–172. [3] E. Krebs, B. Silvi, A. Daudin, P. Raybaud, J. Catal. 260 (2008) 276–287. [4] S. Hatanaka, Catal. Surv. Asia 9 (2005) 87–93. [5] S. Hatanaka, O. Sadakane, US Patent 6,120,679, Mitsubishi Oil Corporation, 2000. [6] T.G. Kaufmann, A. Kaldor, G.F. Stuntz, M.C. Kerby, L.L. Ansell, Catal. Today 62 (2000) 77–90. [7] C. Marcilly, Stud. Surf. Sci. Catal. 135 (2001) 37–60. [8] C. Song, X. Ma, Appl. Catal. B: Environ. 41 (2003) 207–238. [9] I.V. Babich, J.A. Moulijn, Fuel 82 (2003) 607–631. [10] H. Topsøe, B.S. Clausen, F.E. Massoth, Hydrotreating Catal. Sci and Tech., Springer Verlag, 1996, 11. [11] A. Daudin, A.F. Lamic, G. Pérot, S. Brunet, P. Raybaud, C. Bouchy, Catal. Today 130 (2008) 221–230. [12] J.T. Miller, W.J. Reagan, J.A. Kaduck, J. Kropf, J. Catal. 193 (2000) 123–131. [13] E. Krebs, B. Sylvi, P. Raybaud, Catal. Today 130 (2008) 160–169. [14] A. Daudin, S. Brunet, G. Perot, P. Raybaud, C. Bouchy, J. Catal. 248 (2007) 111–119. [15] N. Guernalec, C. Geantet, P. Raybaud, T. Cseri, M. Aouine, M. Vrinat, Oil Gas Sci. Technol.-Rev. IFP 61 (2006) 515–535. [16] D. Mey, S. Brunet, C. Canaff, F. Maugé, C. Bouchy, F. Diehl, J. Catal. 227 (2004) 436–447. [17] P.S.E. Dai, D.E. Sherwood Jr., R.H. Petty, US Patent 5,340,466, Texaco, 1994. [18] V. Rabarohoela-Rokotovao, S. Brunet, G. Pérot, F. Diehl, Appl. Catal. A: Gen. 306 (2006) 34–44. [19] A.-F. Lamic, A. Daudin, S. Brunet, C. Legens, C. Bouchy, E. Devers, Appl. Catal. A: Gen. 344 (2008) 198–204. [20] P. Ghosh, A.T. Andrews, R.J. Quann, T.R. Halbert, Energy Fuels 23 (2009) 5743–5759. [21] N. Dos Santos, H. Dulot, N. Marchal, M. Vrinat, Appl. Catal. A: Gen. 352 (2008) 114–123. [22] M. Badawi, L. Vivier, G. Pérot, D. Duprez, J. Mol. Catal. A: Chem. 293 (2008) 53–58. [23] G. Berhault, A. Mehta, A.C. Pavel, J. Yang, L. Redon, M.J. Yacaman, L.C. Araiza, A.D. Moller, R.R. Chianelli, J. Catal. 198 (2001) 9–19. [24] Y. Iwata, K. Sato, T. Yoneda, Y. Miki, Y. Sugimoto, A. Nishijma, H. Shimada, Catal. Today 45 (1998) 353–359. [25] S. Fuentes, G. Diaz, F. Fedraza, H. Rojas, N. Rosas, J. Catal. 113 (1988) 535–539. [26] D. Mey, S. Brunet, G. Pérot, F. Diehl, S. Kasztelan, Am. Chem. Soc. Prep. Div. Pet. Chem. 47 (2002) 69–72. [27] Y. Romero, C. Fontaine, S. Brunet, A. Daudin, C. Bouchy, E. Devers, C. Legens, Am. Chem. Soc. Prep. Div. Pet. Chem. 53 (2) (2008) 5–8. [28] P. Raybaud, Appl. Catal. A: Gen. 322 (2007) 76–91. [29] R.R. Chianelli, Catal. Rev.-Sci. Eng. 26 (1984) 361–393.
C. Fontaine et al. / Applied Catalysis A: General 388 (2010) 188–195 [30] L.P. Nielsen, S.V. Christensen, H. Topsoe, B.J. Clausen, Catal. Lett. 67 (2000) 81–85. [31] J.V. Lauritsen, J. Kibsgaard, G.H. Olesen, P.G. Moses, B. Hinnemann, S. Helveg, J.K. Nørskov, B.S. Clausen, H. Topsøe, E. Lægsgaard, F. Besenbacher, J. Catal. 249 (2007) 220–233. [32] J.V. Lauritsen, M.V. Bollinger, E. Lægsgaard, K.W. Jacobsen, J.K. Nørskov, B.S. Clausen, H. Topsøe, F. Besenbacher, J. Catal. 221 (2004) 510–522. [33] S. Helveg, J.V. Lauritsen, E. Lægsgaard, I. Stensgaard, J.K. Nørskov, B.S. Clausen, H. Topsøe, F. Besenbacher, Phys. Rev. Lett. 84 (2000) 951–954. [34] J.S. Choi, C. Petit-Clair, D. Uzio, Stud. Surf. Sci. Catal. 143 (2002) 585–594. [35] F. Bataille, J.L. Lemberton, P. Michaud, G. Pérot, M. Vrinat, M. Lemaire, E. Schulz, M. Breysse, S. Kasztelan, J. Catal. 191 (2000) 409–422. [36] M. Egorova, R. Prins, J. Catal. 225 (2004) 417–427.
195
[37] S. Kasztelan, Hydrotreating Technology for Pollution Control, Dekker, New York, 1996, pp. 29–45. [38] R. Candia, O. Sorensen, J. Villadsen, N.-Y. Topsoe, B.S. Clausen, H. Topsoe, Bull. Soc. Chim. Belg. 93 (1984) 763–766. [39] M. Brémaud, L. Vivier, G. Pérot, V. Harlé, C. Bouchy, Appl. Catal. A: Gen. 289 (2005) 44–50. [40] M. Toba, Y. Miki, T. Matsui, M. Harada, Y. Yoshimura, Appl. Catal. B: Environ. 70 (2007) 542–547. [41] E. Krebs, A. Daudin, P. Raybaud, Oil Gas Sci. Technol.-Rev. IFP 64 (2009) 707–718. [42] A.D. Gandubert, C. Legens, D. Guillaume, S. Rebours, E. Payen, Oil Gas Sci. Technol.-Rev. IFP 62 (2007) 79–89. [43] K. Marchand, C. Legens, D. Guillaume, P. Raybaud, Oil Gas Sci. Technol.-Rev. IFP 64 (2009) 719–730.