Model test set up methodology for HDS to improve the understanding of reaction pathways in HDT catalysts. Chemical Engineering

Transcription

1 Model test set up methodology for HDS to improve the understanding of reaction pathways in HDT catalysts David Manuel Paulo Negreiro Thesis to obtain the Master of Science Degree in Chemical Engineering Supervisors Dr. Bertrand Guichard (IFPEN) Prof. Francisco Manuel da Silva Lemos (IST) Examination Committee President: Prof. José Manuel Félix Madeira Lopes (IST) Supervisor: Prof. Francisco Manuel da Silva Lemos (IST) Members of the Committee: Prof. Maria Filipa Gomes Ribeiro (IST) October 2015

2 This page was intentionally left blank.

3 Acknowledgements Firstly, I would like to start by thanking to Prof. Filipa Ribeiro the great opportunity provided for doing my master thesis at IFPEN. I would also like to thank Joana Fernandes for her support and help during the beginning of this internship. To my IFPEN supervisor Dr. Bertrand Guichard, for your availability, patience and helpful advises for my professional career. I am also very thankful to Véronique Delattre and Nathalie Lett, for all the formation they gave me and, above all, I am grateful for their kindness, joy, teaching ability and good humour which made this experience much enriching. To the people from the Catalysis by Sulfides Department (R066S), for all support they gave me and for the very good working environment. I want to express my gratitude to Prof. Francisco Lemos, for his support and for believing in my capabilities. I would like to specially thank to Fabien, Leonor, Rubén, Sónia, Mafalda, Svetan, Mathieu, Ana Rita, Max, Leonel, Marisa and Alberto for their support. I also thank to Larissa and Alexis for the amazing moments we shared together. A big thank you! to my portuguese friends, Ana, Loios, Joana, Casinhas, Catarina, Solange and Diogo. Thank you for your support, your friendship and, above all, for the great moments we shared together. You have become undoubtedly my second family. I have also to thank Pedro, for your words of wisdom about IFPEN and helping me along my internship. Finally, I would like to thank my family, especially my mother and brother. Your encouragements and cheering words over these six months made easier the fact of being far away from home. To Susana, thank you for everything, because even far away you made everything much easier. It would not be the same without you by my side iii

4 This page was intentionally left blank.

5 Abstract In this present work, the hydrodesulfurization (HDS) of 4,6-dimethyldibenzothiophene (4,6-DMDBT) was studied over three CoMo/Al2O3 catalysts (dried, calcined and additive impregnated) in a fixed-bed reactor under standard conditions close to those usually used in diesel fuel hydrotreating following particularly the HYD and DDS pathways behaviors. The main focus was to identify some strong differences in behavior between the various catalysts and evaluate the effect of H 2 S, NH 3 and H 2 partial pressures on their relative catalytic performances. It was found by experimental and modelling results that, at standard conditions, the additive impregnated catalyst performs better and was less impacted by H 2 S adsorption than dried and calcined. Though, in the presence of high amounts of H 2 S, the additive impregnated showed to be the one differing mostly from H 2 S partial pressure. In addition, the study on the impact of nitrogen-based compounds (quinoline) revealed that all three catalysts are similar inhibited. In the same way, modifying the partial pressure of H 2 was found to enhance the activity of all catalysts, especially the HYD pathway. A more detailed study and additional experimental tests should be performed in order to improve the understanding on the relation between quinoline and H 2 S within the deep HDS of 4,6-DMDBT and to further develop the kinetic model created. Keywords: Diesel, middle distillates, 4,6-DMDBT, CoMo/Al2O3, inhibition effect v

6 This page was intentionally left blank.

7 Resumo Neste trabalho, estudou-se a hidrodessulfurização do 4,6-dimetildibenzotiofeno em três tipos de catalisadores CoMo/Al 2 O 3 (seco, calcinado e aditivado), num reactor de leito fixo em condições standard próximas das usadas no hidrotratamento do gasóleo de forma a estudar particularmente as duas reacções principais do processo hidrogenação (HYD) e dessulfurização directa (DSD). O principal objectivo foi identificar as maiores diferenças entre os três catalisadores e avaliar o efeito da pressão parcial de H 2 S, NH 3 e H 2 nas performances catalíticas. A partir dos resultados experimentais e de modelação cinética obtidos, a condições standard, o catalisador aditivado mostrou ser o catalisador com maior actividade e menor impacto pela adsorção de H 2 S comparativamente com os catalisadores seco e calcinado. No entanto, para grandes quantidades de H 2 S, o catalisador aditivado foi o mais afectado. Adicionalmente, do estudo do impacto de compostos azotados (quinolina), determinou-se que os três catalisadores são inibidos de forma idêntica. Da mesma forma, da alteração da pressão parcial de H 2 determinou-se que a actividade de todos os catalisadores é aumentada, em particular na via de hidrogenação. Um estudo mais detalhado e mais testes experimentais devem ser efectuados de forma a melhor compreender a relação intrínseca entre a quinolina e o H 2 S na reacção de hidrodessulfurização do 4,6-DMDBT e progredir no desenvolvimento do modelo cinético estabelecido. Palavras-chave: Gasóleo, destilados médios, 4,6-DMDBT, CoMo/Al2O3, efeito inibitório vii

8 This page was intentionally left blank.

9 List of Contents List of Figures... xi Abbreviation List... xiv 1 Introduction Bibliographic Study Context Overview on hydrotreatment process Diesel: specifications and characteristics HDS Catalysts Sulfidation process Non-promoted catalysts: MoS 2 /Al 2 O Promoted catalysts: CoMo/Al 2 O Main compounds in HDT Reactivity of sulfur compounds ,6-DMDBT HDS pathways Inhibition effect Ammonia Hydrogen sulfide Methodology Experimental Part Catalysts preparation Unit T Unit loading Sulfidation Model feedstock Operating conditions Data analysis Kinetic Study Results and Discussion Comparison of Catalysts in Standard Conditions Impact of H 2 S partial pressure ix

10 4.2.1 Apparent comparison Kinetic comparison Impact of NH 3 partial pressure Apparent comparison Kinetic comparison Impact of H 2 partial pressure Apparent comparison Kinetic comparison Summary of Results Conclusion and Future Perspectives Bibliography Appendix 1 GC Chromatogram Appendix 2 Kinetic model x

11 List of Figures Figure 1 Yearly evolution of world consumption of primary energy [1]... 3 Figure 2 World`s percentage shares of oil demand by sector in 2011 and 2040, [2]... 4 Figure 3 Representation of the maximum sulfur limit for diesel all over the world (2014) [2]... 4 Figure 4 Schematic of a typical oil refinery [5]... 5 Figure 5 Once-through hydroprocessing unit: two separators and recycle gas scrubber, [7]... 7 Figure 6 Scheme of the top of a hydrotreatment reactor, [7]... 7 Figure 7 Typical shapes of catalysts (A and B) trilobe and cylindrical pellets, (C) spheres, (D) rings, [9]... 9 Figure 8 Different steps of hydrotreating catalyst synthesis and life [12]... 9 Figure 9 Schematic representation of the sulfidation process of a CoMo/γ-Al 2 O 3 catalyst, [13].. 10 Figure 10 Evolution of the activity versus time on stream during the HDS of DBT on NiMo/Al 2 O 3 [14] Figure 11 Top and side views of a MoS 2 cluster [17] Figure 12 STM images of triangular (A) and hexagonal (B) MoS 2 nanocluster [19] Figure 13 Structural illustration of different structures present in a sulfided CoMo/Al 2 O 3 catalyst [12] Figure 14 Co distribution on the sulfide CoMo/Al 2 O 3 catalyst [21] Figure 15 Schematic of the (a) MoS 2 and (b) CoMoS active phases (adapted). The yellow spheres represent sulfur atoms, purple spheres represent molybdenum, and finally, green spheres represent cobalt [23] Figure 16 HDS rate of DBT as function of the computed E M-S [24] Figure 17 Organosulfur compounds converted in HDT reactions [28] Figure 18 Main organic-nitrogen compounds found in pre-treated crude oil. Underlined compounds represent the non-basic nitrogen compounds and the others represent basic-nitrogen compounds Figure 19 Organosulfur reactivity in HDS process as function of aromatic ring sizes and positions of alkyl substitutions [30] Figure 20 Scheme of the reaction pathways of the hydrodesulfurization of 4,6-DMDBT HYD on the left and DDS on the right, [31] Figure 21 Scheme reaction for HYD pathway adapted from [20] Figure 22 Scheme reaction for DDS pathway adapted from [20] Figure 23 Schematic examples of a hydrogenation (A) and C-S bond cleavage (B) sites [20] Figure 24 HDS of 4,6-DMDBT (green) DDS and HYD concentration product (blue and red, respectively), [33] Adapted Figure 25 Molecular modeling results for various sulfur- and nitrogen-containing organic compounds. The bond order value in bold next to a green symbol indicates the bond with highest bond order, while the underlined blue number indicates the net electronic charge on the heteroatom in a given molecule.[36] xi

12 Figure 26 Global activity as function of H 2 S partial pressure, for 4,6-DMDBT transformation on CoMo and NiMo catalysts [31] Figure 27 Overview of the T033 unit Figure 28 Schematic of a VALCO valve with six ways Figure 29 Representation of the micro-reactor used in the T033 unit Figure 30 Temperature program for the sulfidation process, before each catalytic test Figure 31 Temperature program for the catalytic test Figure 32 Schematic of the two reaction pathways for the HDS of 4,6-DMDBT, adapted [32] Figure 33 Kinetic fitting obtained for CoMo-C at standard conditions Figure 34 Selectivity DDS/HYD as function of the total HDS conversion, at 290 C Figure 35 Selectivity DDS/HYD as function of the total HDS conversion, at 300 C Figure 36 Selectivity DDS/HYD as function of the total HDS conversion, at 310 C Figure 37 Ratio DMDCH/MCHT as function of the conversion of DMBPh (HDA), at 310 o C Figure 38 Selectivity DDS/HYD as function of the temperature Figure 39 HYD conversion of 4,6-DMDBT as function of temperature for the three catalysts prepared. In the graph, full lines represent the 1,2 wt.% DMDS feed and gapped lines represent the 2 wt.% DMDS Figure 40 DDS conversion of 4,6-DMDBT as function of temperature for the three catalysts prepared. In the graph, full lines represent the 1,2 wt.% DMDS feed and gapped lines represent the 2 wt.% DMDS Figure 41 Kinetic fitting obtained for CoMo-B, for Feedstock-2 conditions (2 wt.% DMDS) Figure 42 ln(k o ) as function of ln(pph 2 S) for HYD Figure 43 ln(k o ) as function of ln (pph 2 S) for HDA Figure 44 ln(k o ) as function of ln(pph 2 S) for DDS Figure 45 H 2 S inhibition factor for CoMo-A, CoMo-B and CoMo-C. These results took into account the results obtained at standard conditions and Feedstock Figure 46 HYD conversion of 4,6-DMDBT as function of temperature for the three catalysts prepared. In the graph, full lines represent the 0,5 wt.% Quinoline model approximation and gapped lines represent the 1,0 wt.% Quinoline (standard conditions) Figure 47 DDS conversion of 4,6-DMDBT as function of temperature for the three catalysts prepared. In the graph, full lines represent the 0,5% Quinoline model approximation and gapped lines represent the 1,0% Quinoline (standard conditions) Figure 48 Kinetic fitting obtained for the HDS of 4,6-DMDBT using CoMo-B - 0,5% quinoline Figure 49 Kinetic fitting obtained for the HDS of 4,6-DMDBT using CoMo-C - 1,5% quinoline Figure 50 ln(k o ) as function of ln(ppnh 3 ) for HYD Figure 51 ln(k o ) as function of ln(ppnh 3 ) for HDA Figure 52 ln(k o ) as function of ln(ppnh 3 ) for DDS Figure 53 Overall NH 3 inhibition factor comparing standard conditions (1 wt.% Quinoline) with both 0,5 wt.% and 1,5 wt.% Quinoline xii

13 Figure 54 HYD pathway conversion as a function of the temperature for different catalysts and total pressures Figure 55 DDS pathway conversion as a function of the temperature for different catalysts and total pressures Figure 56 Kinetic fitting obtained for the HDS of 4,6-DMDBT using CoMo-C, at 40 bar Figure 57 Effect of H 2 partial pressure on the selectivity of HDS over the prepared catalysts, for 310 C Figure 58 ln(k o ) as function of ln(pph 2 ) for HYD Figure 59 ln(k o ) as function of ln(pph 2 ) for HDA Figure 60 ln(k o ) as function of ln(pph 2 ) for DDS Figure 61 H 2 activation factor on the three catalysts tested (global activity, see Eq. 27) Figure 62 Relative differences in activity for HYD reaction between the three catalysts Figure 63 Relative differences in activity for the DDS reaction between the three catalysts Figure 64 Example of a GC chromatogram Figure 65 - Kinetic fitting obtained for the HDS of CoMo-B at standard conditions Figure 66 - Kinetic fitting obtained for the HDS of CoMo-A at standard conditions Figure 67 Kinetic fitting obtained for the HDS of CoMo-C at 2 wt.% DMDS Figure 68 Kinetic fitting obtained for the HDS of CoMo-A at 2 wt.% DMDS Figure 69 Kinetic fitting obtained for the HDS of CoMo-A at 0,5 wt.% Quinoline Figure 70 - Kinetic fitting obtained for the HDS of CoMo-C at 0,5 wt.% Quinoline Figure 71 Kinetic fitting obtained for the HDS of CoMo-A at 1,5 wt.% Quinoline Figure 72 Kinetic fitting obtained for the HDS of CoMo-B at 1,5 wt.% Quinoline Figure 73 - Kinetic fitting obtained for the HDS of CoMo-B at 40 bar Figure 74 - Kinetic fitting obtained for the HDS of CoMo-A at 40 bar xiii

14 This page was intentionally left blank.

15 Abbreviation List CoMo/Al 2 O 3 CoMo catalyst supported over alumina DBT Dibenzothiophene DDS Direct desulfurization DFT Density Functional Theory DMBPh Dimethylbiphenyl DMDCH - Dimethyldicyclohexyl DMDS Dimethyl disulfide E 2 Elimination reaction GC Gas Chromatography HDA Hydrodearomatization HDN Hydrodenitrogenation HDS Hydrodesulfurization HYD Hydrogenation H 2 S Hydrogen sulfide LHSV Liquid Hourly Space Velocity MCHT Methylcyclohexyltoluene pph 2 S Hydrogen sulfide partial pressure pph 2 Hydrogen partial pressure ppnh 3 Ammonia partial pressure ppquinoline Quinoline partial pressure SiC Silicon carbide SCR Selective Catalytic Reduction STM Scanning Tunneling Microscopy 4,6-DMDBT 4,6-Dimethyldibenzothiophene wt.% Percentage weight fraction xv

16 This page was intentionally left blank.

17 1 Introduction In order to decrease pollution caused by automobile vehicles, the sulfur content in diesel fuel has been drastically reduced over the years as witnesses the restrictive regulations. However, refining industries are processing heavier feedstocks and facing an increasing demand of diesel/gasoline, so the need for more efficient hydrodesulfurization (HDS) catalysts is required, i.e. low sulfur product at even highest LHSV (the HDS process taking place in fixed bed reactors). The commercial catalysts commonly used for HDS reactions are molybdenum sulfides promoted by cobalt or nickel and supported over alumina. Those catalysts are well known and the way to prepare them is well monitored. Nevertheless, if one aims to improve their performances, it is now necessary to identify precisely the limitations. It can be achieved by characterizing deeply the catalyst, using powerful tools or by carrying a kinetic study to identify the main limitation/inhibiting/activating parameters. To do so, it is necessary to evaluate the catalyst in representative conditions witnessing the way it will have to work in real conditions. The feedstocks being too much complicated, the matrix needs to be simplified and adapted to the goals. The main compounds contained into the feedstock are generally aromatics, olefins, nitrogen and sulfides and the catalysts will simultaneously have to perform hydrogenation, hydrodesulfurization and hydrodenitrogenation reactions. All those reactions improve the feed quality but could also be competitive ones. That is why if one aims to study the reactivity, all those compounds would need to be introduced in the model feed. Indeed studying separately the HDS could lead to a wrong interpretation. In the HDS of middle distillates, as sulfur conversion increases the remaining species are mostly dibenzothiophenes. These compounds, such as 4,6-dimethyldibenzothiophene (4,6-DMDBT), are the most refractory compounds, since they are very difficult to decompose. These compounds are converted through two distinct pathways namely, Hydrogenation (HYD) and Direct Desulfurization (DDS).Furthermore, the presence of the methyl groups on 4,6-DMDBT highly limits the reactivity and leads the HDS to selectively process through the hydrogenating route compared to the DDS one. Moreover, there is a high interest to obtain the best catalysts and operating conditions to achieve HDS trough the DDS pathway since it consumes less hydrogen than HYD. The objective of this work is focused on the comprehension of the HDS mechanism on various representatives CoMo catalyst types in order to study the deep HDS of middle distillates in the range of operating conditions dedicated to the low pressure HDS (i.e. around bar with CoMo catalysts). The inhibiting and activating effects should be taken into account by modifying some of the working conditions (temperature, LHSV, pressure). The partial pressure in H 2 S and NH 3 will be monitored trough DMDS and quinoline incorporation into the prepared feedstocks, respectively. For every conditions, there was a mixture of model molecules, in the presence of 4,6-DMDBT, with the purpose to discriminate the catalytic performance of each CoMo-based catalyst and to compare it to the well-established ranking provided by the real feed evaluation. The aim is to point out some strong differences in behavior between the various catalysts and to go deeper in the comparison than 1

18 the direct ranking provided by diesel HDS, i.e. supplying activation energy, inhibition effects and changes according to the HDS conversion or partial pressure. Thus, this work should lead to propose the best conditions for each type of catalyst and also to improve their way of working induced by the preparation methodology. 2

19 2 Bibliographic Study 2.1 Context Today, energy is a mix of multiple resources, as we are able to convert them in many ways in order to power high-consuming societies. In the past few decades, the world energy market has entered in a period of dynamic changes due to the economic growth in developed and developing countries. It led to a rapid growth in primary energy. As it can be seen in Figure 1, the resource that plays a vital role in order to successfully satisfy this demand is oil, as it was representing 33% of the world s energy consumption in 2013 [1]. Figure 1 Yearly evolution of world consumption of primary energy [1] Therefore, as crude oil exploitation is one of the most pollutant activities worldwide, environmental protection, cleaner fuels have been required. Practically, the sulfur released to the atmosphere has to be controlled and lowered as much as possible because this compound is responsible for many environmental problems such as production of acid rains which cause the acidification of soils, lakes and streams, and accelerates corrosion of buildings and monuments. This environmental phenomenon is produced by the reaction of water molecules, present in the atmosphere, with SO x produced within the diesel engine. The reaction is described as follows: (Eq. 1) (Eq. 2) 3

20 Furthermore, as one can see in Figure 2, the percentage shares of oil demand is mainly constituted by the transportation and industrial sectors. In the next decades the percentage share taken by the industrial sector will suffer a minor decrease of 2%, contrasting with the 4% increase within the transportation sector. Figure 2 World`s percentage shares of oil demand by sector in 2011 and 2040, [2] Moreover, post-combustion catalyst used to reduce NO x emission and massively introduced in the automobile engines (SCR systems) are very sensitive to poisoning by sulfur so it strengthened the need to reduce the sulfur amount in commercial diesels. Taking into account that crude oil quality decreases over the years, the sulfur content tends to increase thus regulatory specifications will be harder to satisfy. Furthermore, in Figure 3, it is possible to see that in developed countries the maximum sulfur limit allowed is from 10 to 15 ppm (m/m%) but the same level is expected in the coming years or decades in developing countries [2]. Figure 3 Representation of the maximum sulfur limit for diesel all over the world (2014) [2] 4

21 2.2 Overview on hydrotreatment process As crude oil quality decreases the need for new technologies capable of producing cleaner and better fuels is a continuous challenge for process engineering companies. Indeed, in modern refineries HDT units are the most common process units. The typical composition of unprocessed crude oil is shown in Table 1. Table 1 Typical composition of crude oil, [3] Element Percentage (%) Carbon Hydrogen Nitrogen 0,1 1,0 Oxygen 0,1-0,5 Sulfur 0,5-6 Metals < 0,1 As one can see in Table 1, sulfur is the main contaminant within crude oil. On one hand, sulphurous compounds are poisonous but, on the other hand, as reported by Rana et al.[4], nitrogenous compounds lead to catalyst inhibition even at minute concentrations. Therefore, hydrotreatment to remove sulfur and nitrogen is usually applied in many sections in the refinery as shown in Figure 4. Figure 4 Schematic of a typical oil refinery [5] 5

22 Usually, the HDS process takes place in a catalytic fixed-bed reactor in presence of hydrogen gas and the liquid feedstock to be processed. The typical operating conditions for the hydrotreatment process depend on the feedstock reactivity, composition and with the product`s specifications. In Table 2 are shown the operating conditions used in past as well as the conditions used nowadays. As one can see, the conditions used in current days are much more aggressive (especially for hydrogen pressure and LHSV) due to the increasing amount of sulfur and other contaminants present in crude oil, as already mentioned, and to the lower limits that are imposed by legislation. Table 2 General hydroprocessing conditions used in industry, [6] A typical flow diagram of a two reactors HDT process in which the feedstock and hydrogen gas are supplied from the top of the reactor is shown in Figure 5. There are two types of hydrotreatment processes, the single and the two or multiple-staged processes [7]. With the increasingly stringent regulations on diesel oil, a lot of attention has been paid to reduce sulfur content of distillate fuels. The two-stage process is an upgrade of the conventional hydrotreatment process, since it removes the hydrogen sulfide and ammonia produced in the first reactor, enhancing the reactivity within the second reactor. Therefore, with staged processes, high decomposition of sulfur and nitrogen-based compounds are easier to achieve. 6

23 Figure 5 Once-through hydroprocessing unit: two separators and recycle gas scrubber, [7] As there are many types of compounds to be decomposed or removed, the choice of the catalyst is therefore crucial to the process. So, to meet the required specifications, HDT catalysts have to be efficient in order to accomplished hydrodesulfurization (HDS), hydrodenitrogenation (HDN) and hydrodearomatization (HDA) reactions. Typically, the reactor (Figure 6) consists in more than one catalytic bed depending on the impurities found in the feedstock and operating conditions such as the liquid hourly space velocity (LHSV). In these reactors several reactions are found in order to remove the sulfur and nitrogen-based compounds. In order to avoid cracking reactions and to maximize the quality of the liquid fuels produced, quenching hydrogen gas is commonly injected at various points along the reactor to cool down the reaction temperature because HDT reactions are highly exothermic. Liquid-phased products are additionally fractionated according to their boiling points into the required products in a column according to their boiling points. Figure 6 Scheme of the top of a hydrotreatment reactor, [7] 7

24 2.3 Diesel: specifications and characteristics To better understand why HDT is important to produce a high quality diesel, it is essential to know the main characteristics and specifications for this fuel. Diesel is a fuel produced from crude oil and consists mainly of aliphatic and some aromatics hydrocarbons comprising, normally, carbon atoms with boiling points in the range of C [8]. Moreover, diesel is an oil fraction heavier than gasoline, with lower H/C mass ratio. The properties and reactivity of diesel feeds, composed mainly of paraffins and aromatics, are deeply dependent on their source. In Europe, the latest specifications imposed in 2009 were mainly pointed to decrease the sulfur content on diesel fuels. In Table 3 are shown the main specifications for diesel composition. The other one is directly linked to its use in the diesel engine. Table 3 Specifications for diesel fuel [9] Generally, diesel quality is essentially related to its cetane number. Hence, the higher n-paraffinic and naphthenic content, the greater will be the quality of the diesel produced. Moreover, additives are generally added to reach better properties depending on the purpose and country, such as to lower the freezing point, which is essential in some countries where the temperatures are very low during winter. 8

25 2.4 HDS Catalysts The catalysts used in the hydrotreatment process contain a metal sulfide usually molybdenum promoted with either nickel or cobalt, supported over a refractory oxide carrier (e.g. alumina). To minimize diffusion limitations and the process pressure drop, these catalysts must have a certain shape. The most commonly used are generally trilobe shaped pellets, spheres and rings (Figure 7). (A) (B) (C) (D) Figure 7 Typical shapes of catalysts (A and B) trilobe and cylindrical pellets, (C) spheres, (D) rings, [10] For industrial purposes, the most commonly used support is γ-alumina since: - it provides a greater surface area ( m 2 /g) than other supports [11]; - it allows to maximize the dispersion of the active phase, due to its acid-basicity properties; - and exhibits a high mechanical strength [12]. The preparation of a HDT catalyst (Figure 8) involves several steps, including: Impregnation the impregnation solution is added to the support; Maturation guarantees that the solution is well dispersed into the support pores; Drying remove the excess of solvent from the support. Figure 8 Different steps of hydrotreating catalyst synthesis and life [8] 9

26 These steps can be complemented by an optional calcination which was frequently performed before but is less common nowadays. An additivation step is also usual to be realized in order to promote the catalytic activity of the HDT catalyst. This step can be accomplished, for instance, with a glycol molecule [8]. Then, the catalysts are sulfided, leading to the active state Sulfidation process HDT catalysts might be subjected to a sulfidation process in order to form the active phase. This step of transformation from oxide to sulfide (and with molybdenum reduction) plays a crucial role in what concerns to the catalytic activity and the catalysts stability during hydrotreatment reactions. As this transformation is exothermic, temperature has to be carefully controlled in order to avoid poisonous side reactions, i.e. metallic oxide reduction by hydrogen and coke formation, which would reduce the catalytic activity (Figure 9). Figure 9 Schematic representation of the sulfidation process of a CoMo/γ-Al 2O 3 catalyst, [13] Due to handling and loading problems associated with the active sulfided form, hydrotreating catalysts are typically produced and shipped in their inactive form. Then, in order to be used these catalysts must be first activated by a sulfidation agent, promoting the O S exchange. This process is performed using either a gas mixture of H 2 S/H 2, an organo-sulfidation agent like dimethyl disulfide (DMDS) or even directly the diesel feedstock to be desulfurized [14]. When DMDS is used, it decomposes into CH 4 and H 2 S, which the latter acts as the actual sulfidation agent. As one can see in Figure 10, Texier et al. [14] observed that using organo-sulfide compounds like DMDS slightly increases the catalytic specific activity. Hence, it has been stated that a correct activation of hydrotreating catalysts depends greatly on temperature and H 2 S proportion. 10

27 Figure 10 Evolution of the activity versus time on stream during the HDS of DBT on NiMo/Al 2O 3 [14] Moreover, Hallie [15] reported that the use of organo-sulfide agents (such as DMDS) increases the HDS activity of VGO (Vacuum Gas Oil) by as much as 60% when compared to a gas-phase H 2 /H 2 S sulfidation procedure particularly for CoMo/Al 2 O 3 catalysts. All these results underline the key role of the sulfidation step in the catalytic performances Non-promoted catalysts: MoS 2 /Al 2 O 3 In HDT processes, most part of the catalysts used are based in molybdenum sulfide. This is why one first describes the non-promoted catalysts. These catalysts are constituted by a well dispersed active phase of MoS 2 on an alumina surface whose the primary unit cell consists in a single hexagonal slab. Every single slab exhibits the same structure, where molybdenum ions are coordinated with six sulfur ions in a trigonal-prismatic configuration. Also, the slabs interact with each other by Van der Waals forces, creating a layered structure with interposed molybdenum between two layers of sulfur atoms. In addition, depending on the crystallographic plan terminating the obtained structure exhibits two types of edges, either Mo-edges or S-edges as evidenced in Figure 11. Figure 11 Top and side views of a MoS 2 cluster [16] 11

28 Both edges can lose a sulfur atom by reaction with hydrogen. The resulting vacancies lead to the exposure of Mo-cations, and are known as coordinately unsaturated sites (CUS). CUS are deficient in electrons and thus interact with electron donor compounds (Eq. 3). (Eq. 3) These sites are capable of adsorbing organosulfur compounds, which will bond to the unsaturated Mo ions creating a metal-sulfur bond, becoming more active in HDS reactions [8]. Nevertheless, some thermodynamic calculations show that this is not the only way to react for sulfur compounds. Raybaud et al. [17] studied the morphology of MoS 2 catalysts and observed that it depends on the sulfidation conditions, such as temperature and partial pressure of H 2 and H 2 S. It is the relative thermodynamic stability of the two types of edges (under specific conditions) that determines the morphology of the MoS 2 nanoclusters. Under strongly sulfidation conditions (high H 2 S partial pressures) triangular-shaped MoS 2 particles should be obtained, whereas under more reducing conditions the MoS 2 particles might exhibit a hexagonal shape. Lauritsen et al. [18] observed these two possible shapes for MoS 2 nanoclusters by STM imaging of MoS 2 /Au (Figure 12). Figure 12 STM images of triangular (A) and hexagonal (B) MoS 2 nanocluster [18] Promoted catalysts: CoMo/Al 2 O 3 Since the activity of CoMo catalysts will be the main subject of this work, they will be described in more detail. As observed by Bataille et al. [19], for the decomposition of DBT and 4,6-DMDBT (the most difficult compounds to decompose) the overall catalytic activity increases from 0,4 to 7,2 and 0,65 to 2,3 mol.h - 1.kg -1, respectively, when Co is added to Mo/Al 2 O 3. Moreover, this increase is not homogenous for both compounds since their main reaction pathway, as further discussed, is different. However, cobalt by itself does not present any activity, which is why it is considered a promoter of MoS 2 activity [20]. It is also believed that the substitution of Mo by Co atoms at S-edges enhances the formation of sulfur vacancies, CUS [21]. Although there are many attempts to describe the structure of these Co promoted catalysts, Topsøe et al. [6] have proposed the mixed phase "CoMoS" model, which is currently the most accepted model. 12

29 As one can see in Figure 13, cobalt and molybdenum are dispersed into different structures at the catalyst surface [8]. During the sulfidation process, there are cobalt atoms, which react with sulfur atoms resulting on Co 9 S 8 crystallites (Co-sulfide phase). Other cobalt atoms, influenced by the calcination stage of the catalyst preparation, occupy tetrahedral sites inside the catalyst support (Co/Al 2 O 3 ). Finally, there is also the molybdate species, which remain from the precursor, will be dispersed along the catalyst surface. Figure 13 Structural illustration of different structures present in a sulfided CoMo/Al 2O 3 catalyst [8] The mentioned Co 9 S 8 crystallites do not present any catalytic activity, therefore their formation has to be minimized in order to produce selectively the active catalytically structure CoMoS. The formation of all these structures must be controlled all along the catalysts preparation steps, including impregnation and maturation. In Figure 14, one can observe how the quantity of cobalt impregnated in the support influences the formation of each individual phase of cobalt. Figure 14 Co distribution on the sulfide CoMo/Al 2O 3 catalyst [20] As can be seen, adding Co to a given support leads to an increasing amount of the CoMoS phase up to a certain Co/Mo ratio. As Co increases, the edge positions will be occupied until certain point when all these positions are completely filled and, then Co atoms will start to form Co 9 S 8 crystallites. This is a complex set of transformations, which is difficult to monitor, as there are many geometrical 13

30 and structural constraints, which will limit the CoMoS phase amount that is formed. Therefore, for high Co/Mo ratios the HDS activity decreases. Raybaud et al. [22] have proposed, based on DFT simulations, that the final morphology of MoS 2 structures is influenced by cobalt atoms since their presence changes the shape of the MoS 2 slabs. Actually, it would be explained by the fact that cobalt would be incorporated into the S-edges of MoS 2 particles, which would enhance the stabilization of the S-edges relatively to the situation of pure MoS 2. In Figure 15, are shown the structural modifications on the active phase from a MoS 2 to a CoMoS nanocluster. Figure 15 Schematic of the (a) MoS 2 and (b) CoMoS active phases (adapted). The yellow spheres represent sulfur atoms, purple spheres represent molybdenum, and finally, green spheres represent cobalt [22] Concerning the added promoter, it is believed that the substitution of Mo by Co atoms at S-edges increases the formation of sulfur vacancies and creates new and more active sites. Indeed, it is assumed that Co-S bond is weaker than the Mo-S bond, thus vacancy formation is expected to be much easier. Also, it is known that Co increases the electronic density on the sulfur atoms, enhancing the basicity of specific S 2- centers important to HDS reactions. Indeed, it is possible to relate the S-Metal bonding energy (E M-S ) with catalytic activity, using the volcano curve (Figure 16) proposed by Raybaud et al. [23]. Figure 16 HDS rate of DBT as function of the computed E M-S [23] 14

31 In Figure 16, it is possible to verify that CoMoS phase present an optimal E M-S (which corresponds to the maximum activities), contrary of what happens to Co 9 S 8. Moreover, Besenbacher et al. [21] reported that the support interacts with the active phase, influencing the catalyst activity. Further studies suggested a relation between the catalyst structure and activity [24]. Thus, it was proposed that CoMoS has two distinct types - Type I (with low catalytic activity) and Type II (with high catalytic activity). On one hand, Type I structures are known to be incompletely sulfided, presenting bonds with the support. These bonds correspond to the interaction between Mo and the surface of the alumina support which produces monolayer-type structures thus influencing the catalytic properties of CoMoS. On the other hand, Type II structures have the same interactions with the support, as Type I, however they are much weaker, therefore the sulfidation of Mo and Co is easier. This CoMoS-type presents a multilayered slab structure. Furthermore, it has been reported the existence of S-H groups as an active site, created due the adsorption of hydrogen on the S-edge. It is believed that these sites play an important role in hydrogenation and hydrogenolysis reactions since the evidence of adsorption of sulfur containing molecules, as well as the dissociation reaction of H 2 seems to occur [25] [26]. 2.5 Main compounds in HDT In the HDT process, all reactions take place in the liquid phase. The compounds to be converted diffuse through the liquid feed, filing the catalyst pores and adsorb on the catalyst surface where reactions take place. Generally, crude oil contains a huge amount of organosulfur compounds as they can be divided into two main families: the non-heterocyclic and the heterocyclic compounds. On one hand, nonheterocyclic compounds include mercaptans, sulfides and disulfides. On the other hand, heterocyclic compounds have sulfur atoms within the cyclic structure (e.g. thiophene) and others with adjacent aromatic rings and alkyl groups. In Figure 17 are shown examples of non-heterocyclic and heterocyclic structures, which are converted within HDT reactions. Figure 17 Organosulfur compounds converted in HDT reactions [27] Besides sulfur compounds, organic-nitrogenous compounds are also found in crude oil. Depending on its origin, crude may contain amounts of these compounds between 0,1% and 1,0% (wt.%) [28]. This nitrogen content appears in crude oil especially in the form of nitrogen-containing polycyclic aromatic rings, such as quinoline, indole, acridine and carbazole (Figure 18). These nitrogen-based 15

32 compounds are divided in basic and non-basic which thus influences their reactivity (basic compounds adsorb easily and will compete severely with the sulfur compound to be desulfurized). Figure 18 Main organic-nitrogen compounds found in pre-treated crude oil. Underlined compounds represent the nonbasic nitrogen compounds and the others represent basic-nitrogen compounds Reactivity of sulfur compounds As illustrated in Figure 19, the reactivity of the sulfur compounds depends highly on the molecule structure. Thus, thiols, sulfides and disulfides are easier to be converted compared to heterocyclic compounds. Moreover, the overall reactivity decreases with the increasing number of aromatic rings. Figure 19 Organosulfur reactivity in HDS process as function of aromatic ring sizes and positions of alkyl substitutions [29] 16

33 Song et al. [29] reported that when sulfur level in diesel is reduced to 30 ppm, the residual compounds are essentially alkyl-dibenzothiophenes, such as 4-MDBT and 4,6-DMDBT. These compounds exhibit a low HDS reactivity. In fact, the low reactivity of 4,6-DMDBT could be explained by the steric hindrance caused by the methyl groups and also by the electronic factors around the sulfur atom even if the demonstration has never been provided. Ultra-deep HDS of diesel fuel is thus a huge challenge since the lower is the sulfur composition of the crude oil, the more difficult is the HDS and because the heavier feedstocks that have to be used nowadays contain the highest alkyl-dibenzothiophenes proportions. Therefore it becomes important to improve the catalysts activity towards refractory compounds such as 4,6-DMDBT ,6-DMDBT HDS pathways As established by many authors [6] [19], it is known that the HDS of DBT-type compounds occurs by two parallel pathways hydrogenation (HYD) and direct desulfurization (DDS) (Figure 19). HYD pathway implies that the molecule undergoes numerous hydrogenation reactions before the intended sulfur removal, while DDS pathway goes through the direct elimination of sulfur, producing biphenyl-type compounds. Figure 20 Scheme of the reaction pathways of the hydrodesulfurization of 4,6-DMDBT HYD on the left and DDS on the right, [30] 17

34 The final products obtained for the hydrogenation pathway are methylcyclohexyltoluene (MCHT) and dimethyldicyclohexyl (DMDCH). For the DDS pathway the main final product is dimethylbiphenyl (DMBPh) [30]. There are several intermediate products involved in the 4,6-DMDBT conversion but they are not always observed due to their high reactivity. Bataille et al. [19] have suggested that 4,6-DMDBT conversion starts with its partial hydrogenation to form a dihydrointermediate. This first step is considered to be the most difficult step as it partially saturates the benzene ring. In general, nine isomers of dihydrointermediates may be formed by 1,2 or 1,4-addition of two hydrogen atoms. However, compounds formed by 1,4-addition are not favored as the double bonds present in their aromatic ring are not conjugated which means that electrons cannot be delocalized over the electronic system. In Bataille et al. [19] proposed mechanism, there is a same first intermediary compound. For the HYD pathway, the different steps which 4,6-DMDBT undergoes are shown in Figure 21. Figure 21 Scheme reaction for HYD pathway adapted from [19] 1. Firstly, dihydroisomers is hydrogenated, producing tetrahydroisomers; 2. Secondly, an elimination reaction leads to the first C-S bond cleavage; 3. The second aromatic ring is partially hydrogenated (1,2-addition of two hydrogen atoms); 4. Then, a second elimination reaction breaks the second C-S bond, thus forming MCHT; 5. Finally, hydrogenation of MCHT can occur, producing DMDCH. In the DDS route, the vicinity of the sulfur atom must not contain a double bond in order to directly break the C-S bond by an elimination step. To respect this configuration, only two of the dihydrointermediates over the nine possible can be converted through this route. The next steps can be considered for the DDS pathway, as illustrated on Figure 22. Figure 22 Scheme reaction for DDS pathway adapted from [19] 1. An elimination reaction leads to the first C-S bond cleavage; 2. The second aromatic ring is partially hydrogenated (1,2 addition of two hydrogen atoms); 3. A final elimination reaction performs the second C-S bond cleavage, consequently forming DMBPh. Finally, Bataille et al. [19] suggested that the catalytic sites, responsible for the hydrogenation and the C-S bond cleavage, are basically the same, namely that they are made of sulfur vacancies 18

35 associated to neighboring sulfur anions (Figure 23). In this hypothesis, HYD and DDS sites would only differ in the availability of adsorbed hydrogen and in the basicity of the associated sulfur anions. Hydrogenation site C-S bond cleavage site (A) (B) Figure 23 Schematic examples of a hydrogenation (A) and C-S bond cleavage (B) sites [19] On one hand, the hydrogenation site would be composed by: - a vacancy (CUS); - associated with a SH group - and with a hydrogen atom (adsorbed on a Mo atom). So, the vacancy should adsorb the substrate and the neighbouring SH - group associated with the H atom should undergo an immediate hydrogenation reaction. On the other hand, the C-S bond cleavage site, which may be involved in the elimination step of the proposed DDS route, would be then composed by: - two vacancies (CUS) - associated with a S 2- anion. Subsequently, one vacancy should adsorb the substrate (e.g. 4,6-DMDBT), as the second retains the sulfur atom. Subsequently, the sulfur anion should act as a basic site to favour the elimination reaction and directly breaks the C-S bond. 19

36 2.6 Inhibition effect According to Prins et al.[31], HDS reactions are commonly inhibited by compounds such as H 2 S and NH 3. On one hand, the presence of H 2 S during diesel hydrotreatment is inevitable, since it is produced from sulfur compounds decomposition and is necessary to remain in a sulfided form. On the other hand, NH 3 is a product of the hydrodenitrogenation (HDN) of compounds such as quinoline and carbazole, which occurs within the HDT reactors and so is often inevitable Ammonia As reported by Kwak et al. [32], HDS is markedly suppressed by the presence of nitrogen-type compounds such as quinoline and carbazole, even at low concentrations. In Figure 24 is reported the concentration of products formed in the HDS of 4,6-DMDBT in presence of different concentrations of basic and non-basic nitrogenous compounds, carbazole and quinoline, respectively. Figure 24 HDS of 4,6-DMDBT (green) DDS and HYD concentration product (blue and red, respectively), [32] Adapted As one can see, the 4,6-DMDBT conversion decreases when small amounts of both nitrogen compounds are present and drops to near 40% and 38% when using 650 ppm of carbazole or 500 ppm of quinoline in the feedstock, respectively. The inhibiting effect of quinoline is higher than the one imposed by carbazole, as smaller quantities of the basic-nitrogen compound is added have nearly the same effect on 4,6-DMDBT conversion when using higher concentrations of the non-basic one. Hence, as reported by literature [30] [33] [34], basic-nitrogen compounds inhibit much more HDS reactions than basic-nitrogen compounds. To further explain how nitrogen-based compounds influence each HDS pathway (HYD and DDS), Ma et al. [34] have correlated by molecular modeling calculations the bond order and net electronic 20

37 charge on the heteroatoms for nitrogen- and sulfur-containing organic compounds typically found in diesel and jet fuel feedstocks. The bond order and net electronic charge on the heteroatom have been associated to the activity for hydrogenation and hydrogenolysis, respectively. Among a set of molecules, it is expected that: - the molecule having a bond with the highest order is expected to have the greatest reactivity for hydrogenation. - the molecule having the highest electronic charge is expected to have the greatest reactivity for hydrogenolysis. Figure 25 depicts the bond order for each molecules considered. Figure 25 Molecular modeling results for various sulfur- and nitrogen-containing organic compounds. The bond order value in bold next to a green symbol indicates the bond with highest bond order, while the underlined blue number indicates the net electronic charge on the heteroatom in a given molecule.[35] The decreasing order of this bond is reflected in the following classification of the molecules in Figure 25: Acridine > Quinoline > Carbazole > 4,6-DMDBT Quinoline, by virtue of having a bond with the highest bond order as compared to those in carbazole and 4,6-DMDBT, for instance, could be expected to have the highest reactivity for hydrogenation. In other words, quinoline would be the first to undergo HYD reaction. Furthermore, quinoline could also be expected to adsorb first and more strongly inhibit a hydrogenation site in a catalyst. Moreover, Satterfield et al. [36] also reported that secondary amines (reaction intermediates) produced by the decomposition of quinoline have a strong adsorption into the catalyst hydrogenation sites. 21

38 order: The net charge on the heteroatom in the molecules listed in Figure 25 decreased in the following 4,6-DMDBT > Carbazole > Acridine > Quinoline The sulfur atom in 4,6-DMDBT possesses a higher electronic charge than the nitrogen atoms in carbazole and quinoline. Consequently, 4,6-DMDBT would have a much higher tendency to undergo hydrogenolysis as compared to carbazole or quinoline. So DDS would be poorly inhibited by quinoline. Thus, due to the net charge, DDS would be less inhibited by NH 3 than HYD. To summarize the available data, in Table 4 are reported the various observations and conclusions from the literature on the inhibition effect caused by NH 3 on HDT catalysts. It is clear from this Table that nitrogen compounds are very strong inhibitors for HDS of DBT or 4,6-DMDBT. There are still some disagreements in the literature concerning the inhibition of HYD or DDS pathways. It could probably be due to many experimental differences (feedstock composition, operating conditions and catalysts). 22

39 Table 4 List of the results obtained in literature for NH 3 inhibition Publication Reactant studied Type Catalysts Composition Operating Conditions Observations 9 NiMo/Al 2 O 3 CoMo/Al 2 O 3 3,0%/4,0% NiO/CoO 16,0%/19,0% MoO 3 2,6% P 2 O 5 P = 25, 40, 55 bar Temperature = 340 C DDS centers are less sensitive to nitrogen-basic compounds than HYD sites. 4,6-DMDBT [37] CoMo/Al 2 O 3 4,0% CoO 16,0% MoO 3 2,6% P 2 O 5 P = 30 bar Temperature = 330ºC In presence of quinoline, HDS is greatly inhibited. Quinoline undergoes HYD easily thus inhibits hydrogenation sites faster. [35] CoMo/Al 2 O 3 5,8% CoO 27,0% MoO 3 P = 45 bar Temperature = 350 C Quinoline inhibits HYD sites primarily than DDS sites. [38] DBT 4,6-DMDBT CoMo/Al 2 O 3 NiMo/Al 2 O 3 3,0% CoO/NiO 16,0% MoO 3 P = 50 bar Temperature = 300 to 340 C Amines strongly decrease the 4,6-DMDBT global HDS rate and especially the HYD pathway. [32] DBT 4-MDBT 4,6-DMDBT CoMo/Al 2 O 3 4,0% CoO 17,0% MoO 3 P = 40 bar Temperature = 320 C Basic-nitrogen compounds inhibit 4,6-DMDBT global HDS even at low concentrations. Inhibition of quinoline is higher on DDS than HYD. 23

40 2.6.2 Hydrogen sulfide Concerning H 2 S, Rabarihoela-Rakotovao et al.[30] have clearly established the unavoidable impact of this molecule as it is a by-product of HDT reactions. Moreover, H 2 S is essential to maintain the sulfided state of HDT catalysts. Generally, H 2 S is admitted to have an inhibition effect on hydrotreating reactions however, discrepancies remain on its influence on the two consider routes (HYD and DDS) of HDS of DBT-type compounds. Which types of active sites are subjected to H 2 S poisoning and in which step of the reactions is not clear either. Some point out electronic changes or active sites variations with H 2 S partial pressure, related to nature or/and number, as well as mechanism and kinetics reasons have been proposed in order to explain the influence of H 2 S, and, thereby theoretical investigations were also carried out [19][39][40]. As shown by Rabarihoela-Rakotovao et al. [30] in Figure 26, the activity for NiMo or CoMo catalysts differ depending on the H 2 S partial pressure used on the catalytic test. Figure 26 Global activity as function of H 2S partial pressure, for 4,6-DMDBT transformation on CoMo and NiMo catalysts [31] However, there may be two ways for H 2 S to inhibit the HDS reactions. As reported by Besenbacher et al. [21], H 2 S is possible adsorbed on the sulfur vacancies, and, as they seem to be the sites for the C-S bond break, DDS pathway is then more impacted than HYD. The second effect is related to the available S 2- atoms in the S-edge, which will be protonated by H 2 S, which has Brönsted acid properties, lowering their basicity and consequently their reactivity [19]. 24

41 Finally, in relation with the objectives of deep HDS, the effect of H 2 S on the activity of hydrotreating catalysts is a very important issue in industrial practice. As already seen, depending on the sulfur content in the feed and on the operating conditions, the choice of the catalyst may be crucial. And the choice could also depend on the position into the reactor. To further describe the understanding on the influence of H 2 S partial pressure on the 4,6-DMDBT HDS, other observations are reported Table 5 and Table 6. From those results, DDS is always more inhibited by H 2 S compared to HYD and no promoting effect has ever been observed. 25

42 Table 5 List of the results obtained in literature for H 2S inhibition Publication Reactant studied Type Catalysts Composition Operating Conditions Observations [41] NiMo/Al 2 O 3 2,9% NiO 15,3% MoO 3 P = 50 bar pph 2 S = 0 to 0,88 bar Temperature = 200 to 320 C H 2 S may be adsorbed on to the hydrogenolysis sites (DDS) of 4,6-DMDBT more strongly than into hydrogenation sites (HYD). HDS reactivity, and the selectivity between DDS and HYD [42] DBT 4,6-DMDBT CoMo/Al 2 O 3 NiMo/Al 2 O 3 3,0% CoO/NiO 16,0% MoO 3 P = 50 bar pph 2 S = 0 to 1,00 bar Temperature = 340 C pathways depend on the competitive adsorption between the reactant (DBT or 4,6-DMDBT) and H 2 S. Some adsorption conformation data are provided to explain that DDS should be more inhibited than HYD. H 2 S would adsorb preferentially on the DDS centers of [30] CoMo/Al 2 O 3 NiMo/Al 2 O 3 3,0%/4,0% NiO/CoO 16,0%/19,0% MoO 3 2,6% P 2 O 5 P = 25, 40, 55 bar pph 2 S = 0,058 to 1,00 bar Temperature = 300 to 340 C DBT-type compounds. The centers could be identical but owing to the fact that both reactions have not necessarily the same rate-limiting steps, the reactions would be altered differently by H 2 S partial pressure. 26

43 Table 6 List of the results obtained in literature for H 2S inhibition (continuation) Catalysts Publication Reactant studied Operating Conditions Observations Type Composition [19] DBT 4,6-DMDBT NiMo/Al 2 O 3 CoMo/Al 2 O 3 3,1% CoO/NiO 14,0% MoO 3 P = 30 to 50 bar pph 2 S = 0 to 1,00 bar Temperature = 340 C Steric effects upon adsorption on the catalyst active sites could not be responsible for differences in reactivity of DBTs. 27

44 This page was intentionally left blank.

45 3 Methodology 3.1 Experimental Part Catalysts preparation For this work, three types of catalysts were prepared CoMo-A, CoMo-B and CoMo-C. All the prepared catalysts were CoMo trilobe extrudates supported on γ-alumina. In order to process to their catalytic evaluation, their length has been calibrated between 2 and 4 mm for hydrodynamic considerations. These catalysts were prepared by three common stages: active phase impregnation (dry impregnation), maturation (in air atmosphere for 1,5 hours) and drying (at 90 C for 24 hours). On one hand, CoMo-A was the only catalyst used directly on the catalytic tests after drying phase. On the other hand, CoMo-B was calcined under air at 450 C, for 120 minutes, and to produce CoMo- C catalyst, an organic solution was added as an additive, at pore volume, which was then dried under nitrogen flow at 140 C and two hours to preserve the organic compound but eliminating the solvent. In every test performed, it was used the exactly same volume of catalyst (V cat = 4 cm 3 ). Hence, the mass needed for each test was obtained by the density of the catalytic bed (Densité Rempli Tassé - DRT). Along with this characteristic, some other features were also analyzed. In Table 7 are shown these characteristics. Table 7 Characteristics obtained for each catalyst Catalyst DRT (g/cm 3 ) Oxide Density (g/cm 3 ) Support surface area (m 2 /g) MoO 3 (wt.%) Co/Mo CoMo-A 0,89 0,83 CoMo-B 0,85 0, X Y CoMo-C 1,17 0,84 29

46 3.1.2 Unit T033 To simulate as real as possible the industrial conditions of HDS, every catalytic test was carried out in the T033 unit (Figure 27), at IFP Energies nouvelles. Figure 27 Overview of the T033 unit The T033 unit consists in a fixed-bed reactor, under hydrogen pressure. The reactor was fed up with a liquid (feedstock) and gas supply (hydrogen). On one hand, the liquid stream was constantly fed to the reactor by a HPLC pump and controlled by a Quantim TM. For this flow meter, the liquid passes through a U-shaped tube which vibrates in an angular harmonic oscillation. Oscillating forces will then deform the tube and a further vibration component gets added to the already oscillating tube. This added vibration results in a phase shift or twist in few parts of the tubes. This phase shift which is directly proportional to the liquid mass flow rate is measured with the help of sensors. The measured information is further transferred to the electronics unit where it gets transformed to a voltage proportional to mass flow rate. This high performance instrument allows the simultaneous measurement of mass flow, volumetric flow, density and temperature of the fluid. Nonetheless, this device only operates in the presence of a differential pressure. This differential pressure is engaged by a manual valve and at the same time allows pumping the liquid stream at high pressure. On the other hand, the gas streams, nitrogen and hydrogen, were provided by the local networks at high and low pressure. Before introducing both gas and liquid into the reactor, the streams are combined and mixed. Then, the mixture enters into the reactor in a down-flow mode. Furthermore, to heat up the reactor an oven is used and a multipoint cane monitors the increasing temperature with four thermocouples attached, two in the middle and one at each end. The operative pressure is measured by a Keller TM indicator and regulated by a Kammer TM valve. To avoid the effluent condensation the reactor outline is thermally insulated since if effluent temperature reaches values below the dew point and, as the solvent is not liquid, it could lead to precipitation and could have to face top plugging of pipes. 30

47 To fully control the unit during the reaction process, every line has flow indicators, manual control valves and even particles filters. Finally, in order to analyse the reactor effluent, a nitrogen stream is added at the outlet line and then fed to the gas chromatographer. All effluent samples are automatically injected in the chromatographer by a VALCO valve with six ways. Moreover, as one can see in Figure 28, this valve alternately rotates acquiring two distinct positions. Figure 28 Schematic of a VALCO valve with six ways First, the valve assumes the balayage position when the effluent analysis is not required. Then, the injection position is taken when an effluent sample is injected within the GC. The initial temperature of the GC column is 50 C, which then rises to 67 C throw a 15 C/min heating rate, followed by another increase until 290 C (30 C/min heating rate) Unit loading As mentioned before, all catalytic tests were made in the same reactor which has 10 mm of diameter and 18,2 cm of height. To load the reactor three main steps have to be taken: 1. Fill-in 4 cm 3 of inert silicon carbide (SiC); 2. Load 4 cm 3 of SiC mixed with 4 cm 3 of the chosen catalyst; 3. Load again with SiC until the top of the reactor. Nevertheless, between each loading step the reactor has to be shaken to minimize the void spaces along the reactor (intra-particular void). Indeed, the feed flow must be homogeneously distributed along the reactor to minimize the risk of preferred path and ensure the wetting of all the grain of catalysts. To prevent any leak, the reactor is topped with a porous joint and sealed with a torque tool (80 N.m -1 ). In Figure 29 is shown a schematic of the micro-reactor used to perform the catalytic tests. 31

48 Figure 29 Representation of the micro-reactor used in the T033 unit Sulfidation For industrial purposes, the sulfidation of HDT catalysts is accomplished by using organo-sulfide compounds as an activating agent [14]. For practical reasons, organo-sulfide compounds (e.g. DMDS) are much easier to handle than H 2 S and might deliver sulfur gradually to the catalyst through a control of their kinetics of decomposition. In this work, it was used DMDS. Its decomposition is carried out in accordance with the following reaction: (Eq. 4) To begin the test, the catalyst was activated in order to produce the CoMoS phase. Hence, the sulfidation was made in situ with a liquid feedstock composed by DMDS, xylene and cyclohexane. The composition of this feedstock is presented in Table 8. Then, in Table 9 are presented the operating conditions which the catalysts went through in the sulfidation step. These operating conditions were used for almost every test. In fact, the sulfidation pressure used to study the impact of H 2 was the pressure at which the actual catalytic test was performed (40 bar). Table 8 Mass composition of the liquid feed used for the sulfidation process Compound wt.% DMDS 5,88 Xylene 20,00 Cyclohexane 74,12 32

49 Table 9 Operating conditions for the sulfidation stage Parameter Value Pressure (bar) 30,0 LHSV (h -1 ) 4,0 As can be seen in Figure 30, the sulfidation process is formed by 4 steps: - It began with a temperature ramp from 40 C to 350 C (1). - Then this temperature was maintained in order to produce the sulfide form of the catalyst (2). - Then the temperature was decreased until testing value and all the sulfidation operative conditions are changed for the conditions of the catalytic test, respectively (3 and 4). 1,7 C/min 1 (2) 2 3 1,5 C/min 4 (1) (3) (4) Figure 30 Temperature program for the sulfidation process, before each catalytic test Thus, one used always the same activating conditions in order to be sure that the catalysts performances were only depending on the test conditions and not the activation procedure. 33

50 3.1.5 Model feedstock In the first part of the study, the aim was to simulate, as real as possible, the last part of a HDT reactor diesel feed with high sulfur content to study the HDS activity of each CoMo catalyst prepared. As already mentioned, the most refractory sulfur compound to decompose is 4,6-DMDBT. Hence, this was the model molecule chosen to study in this work. The composition of the standard model feedstock (Feedstock-1) prepared is shown below (Table 10): Table 10 Model liquid feed characteristics, Feedstock-1 Compound Weight Fraction (wt.%) ppm S or N Cyclohexane 57,14 - DMDS 1, Quinoline 1, ,6-DMDBT 0, Xylene 40,00 - On one hand, quinoline was chosen as a model basic-nitrogen compound to evaluate how these nitrogen-based species influence the HDS. In this study, it was found that quinoline is not fully converted into NH 3, whatever the temperature. On the other hand, xylene was added to the liquid feedstock to increase the dissolution of 4,6- DMDBT, while cyclohexane was added to decrease the boiling point of the mixture. The mixture of both an aromatic solvent and a non-aromatic one is also more representative of a diesel. Two other feedstocks were prepared in order to determine the conversion of 4,6-DMDBT with higher H 2 S and NH 3 partial pressures, changing the mass composition of DMDS from 1,2 wt.% to 2 wt.% (Feedstock-2) and quinoline from 1,0 wt.% to 1,5 wt.% (Feedstock-3), respectively. However, to go even further in the study of the influence of NH 3, experimental results from previous works were taken into account. These previous tests were performed with a feedstock with a lower concentration of quinoline than the one considered in this present study (i.e. 0,5 wt.%). All feedstocks were prepared in a vessel by adding 4,6-DMDBT to xylene as it is commercialized in a powder form. Then, cyclohexane, quinoline and DMDS (all liquids) were added to these components. 34

51 3.1.6 Operating conditions Regarding the operating conditions of the catalytic test, the goal was to use similar conditions as industrial HDS. The applied conditions are summarized in Table 11. Table 11 Operating standard conditions used in the catalytic tests Parameter Value Pressure (bar) 30,0 Temperature ( C) 290 to 310 Catalyst volume (cm 3 ) 4 H 2 /feed ratio (NL/L) 240 The LHSV conditions were changed in order to to evaluate the model feedstock in the same range of HDS conversion, i.e. not too high (to avoid some saturation and reduce the crossing between the HDS pathways). In this way, it was possible to evaluate the DDS/HYD selectivity without being influenced by supposed thermodynamic effects. In Table 12 is shown the value of LHSV input for each catalyst. The values were chosen taking into account the activity obtained in some previous tests (before this present work). Table 12 LHSV used for each catalyst Catalyst LHSV (h -1 ) CoMo-A 4,0 CoMo-B 3,0 CoMo-C 6,5 During the catalytic test, for each tested temperature, 11 samples of the reactor effluent were analysed within a regular time interval of 45 minutes (1). The temperature program of the catalytic test is shown in Figure 31: 35

52 (1) 1 (2) 2 0,8 C/min (3) 3 Figure 31 Temperature program for the catalytic test Along the process, a nitrogen stream (15 NL/h) is injected at the reactor outlet in order to dilute the effluent before enter the GC. Afterwards, the reactor is washed with xylene and then dried (at the final test temperature) (2) and cooled (3) with a nitrogen and hydrogen stream at descending temperature until 40 C, thus avoiding the catalyst being stuck to the walls of the reactor and to improve the downloading Data analysis As mentioned in the bibliographic study, the HDS of 4,6-DMDBT produces many intermediary compounds and products. However, in this study, to simplify the analysis, one considers that 4,6- DMDBT is converted into two main HYD products (MCHT and DMDCH) and one DDS product (DMBPh). DMBPh MCHT DMDCH Figure 32 Schematic of the two reaction pathways for the HDS of 4,6-DMDBT, adapted [32] 36

53 The analytic results obtained from the GC Galaxie software allowed to determine the conversion of the liquid feed. Hence, the catalytic performance of each catalyst was evaluated according to the HDS conversion of 4,6-DMDBT ( ): (Eq. 5) In addition, the conversion of 4,6-DMDBT through both HYD and DDS pathways were calculated by the following equations: (Eq. 6) (Eq. 7) Where, (Eq. 8) (Eq. 9) In Eq. 8 and Eq. 9, and represent the number of moles of HYD products and DDS product produced during the catalytic test, respectively. 37

54 3.2 Kinetic Study In order to obtain the kinetic model and the parameters from the studied reactions in this work it was used the software ReactOp Cascade. Hence, the software allows the user to create its own set of reactions to better evaluate and estimate the kinetic parameters of a complex mechanism, based on available sets of experimental data. Although it was considered, for the first data analysis, just two main reactions (HYD and DDS), for the kinetic modeling it was added an equilibrium reaction concerning the hydrogenation of DDS products leading to enhance the apparent proportion of HYD products. This reaction was taken into consideration because, at the mentioned operating conditions, DMBPh may be hydrogenated into DMDCH or MCHT and vice-versa (see experimental results chapter 4.1, Figure 37). Additionally, this kinetic study was made in order to better understand the different catalytic performances and how they are influenced by the operating conditions, i.e. H 2 S, NH 3 and H 2 partial pressures. For the decomposition of 4,6-DMDBT, the global kinetic model is: (Eq. 10) As already mentioned, for this part of the study, three reactions were taken into account HYD, DDS, HDA/HDAe. Thus, the kinetic equation for both main pathways (HYD and DDS) is the following: (Eq. 11) Where, (Eq. 12) (Eq. 13) 38

55 In addition, the reaction constant rates for both HYD and DDS are influenced by H 2, H 2 S and NH 3 partial pressures. So, the reaction constant rates can also be written as: (Eq. 14) (Eq. 15) (Eq. 16) With reaction constant rate (h -1 ), pre-exponential constant rate (h -1-m-s-p ), activation energy (J/mol), gas constant (J/mol.K), temperature (K), kinetic partial order for H 2, H 2 S and NH 3, respectively, and pressure (bar). For all the reactions the order relatively to reactant is supposed to be 1. Considering this, the new kinetic model has been created by selecting the ReactOp Cascade tool Model Wizard and then, the following reactions were introduced into the software: (Eq. 17) (Eq. 18) (Eq. 19) With A 4,6-DMDBT, B HYD products and C DDS product. Then, in order to input the experimental results, other software tool had to be selected, namely Experiment Wizard. The experimental data was introduced in the software (Table 13), for a given temperature, as follow: Table 13 Example on how the experimental results, fixing a given temperature, were introduced in Experimental Wizard (ReactOp software) Time (hour) A (mol/100g feed) B (mol/100g feed) C (mol/100g feed) Temperature (K) 0,00 0,300 0,000 0, ,25 0,250 0,042 0, ,33 0,200 0,080 0,

56 Additionally, the value input for time is directly linked with the LHSV at what the test was performed, as established on Eq. 20 (Plug-Flow Reactor). (Eq. 20) Thus, the result obtained will be equal to the reaction contact time during the catalytic test. Finally, loading the experimental results already introduced in the software for all test temperatures on the Estimation Wizard, it was possible to determine the kinetic parameters selected for the created model ( and ) for each reaction giving a total of eight parameters. Moreover, the activation energy does not change for any reaction when increasing or lowering the partial pressure of H 2 S and NH 3. For the equilibrium reaction (HDAe) the and were fixed since they represent a thermodynamic characteristic equal for all the catalysts. In the following table (Table 14) is shown the assumptions made to fit the eight parameters. Table 14 Assumptions made to establish each kinetic parameter s model CoMo-A (Dried) CoMo-B (Calcined) CoMo-C (Additive impregnated) HYD ln(k o ) (hour -1 ) Ea (kj/mol) May change with H 2 S, NH 3 and H 2 partial pressure For each catalyst, the value should be constant independently the operating conditions DDS ln(k o ) (hour -1 ) Ea (kj/mol) May change with H 2 S, NH 3 and H 2 partial pressure For each catalyst, the value should be constant independently the operating conditions ln(k o ) (hour -1 ) May change with H 2 S, NH 3 and H 2 partial pressure HDA (equilibrated) Ea (kj/mol) ln(k o,eq ) (hour -1 ) Ea eq (kj/mol) For each catalyst, the value should be constant independently the operating conditions For all catalysts, these values were considered to be the same and would only change if the H 2 partial pressure is modified 40

57 Then, for each reaction considered, the rate constant was determined by the following equation: (Eq. 21) Moreover, in order to determine the selectivity between the two main reaction pathways (DDS and HYD) and the global activity of each catalyst the following equations were used, respectively: (Eq. 22) (Eq. 23) Furthermore, to evaluate the HYD and DDS relative activity between the additive impregnated catalyst and the other two catalysts (dried and calcined) the next equation were used, respectively: (Eq. 24) (Eq. 25) In this way, it was possible to determine in what conditions the catalytic performance of the additive impregnated catalyst would be enhanced or inhibited. Finally, an inhibition factor has been also calculated in order to understand the influence of H 2 S, NH 3 on the HDS pathways. Nevertheless, to evaluate the influence of H 2, an activation factor has been determined since in literature it is stated that H 2 does not have a negative impact in the 4,6-DMDBT HDS. Thus, the inhibition and activation factors were calculated by a direct ratio between global activities, as following: (Eq. 26) (Eq. 27) With, (h -1 ) global activity determined for a given study condition (Eq. 23) and (h -1 ) global activity determined for standard conditions. Finally, it has to be mentioned that it was not possible to add adsorption constants or inhibiting effects directly into the software model. So, for instance, the nitrogen compounds and/or NH 3 inhibition were not included in the kinetic model. 41

58 This page was intentionally left blank.

59 4 Results and Discussion 4.1 Comparison of Catalysts in Standard Conditions To evaluate the base kinetic parameters for each catalyst, one used the experimental results obtained from previous works and the experimental data accomplished during the internship. These experimental results were performed at standard conditions thus using Feedstock-1. In Table 15 are shown the experimental tests performed with CoMo-A, CoMo-B and CoMo-C and the respective LHSV used at standard conditions. Table 15 Operating conditions used in previous and present works to evaluate standard conditions Catalyst Previous work LHSV (h -1 ) Present work CoMo-A 3 and 4 5 CoMo-B 3 and 4 2,5 CoMo-C 3 5 and 6,5 The kinetic parameters were then optimized with ReactOp Cascade software. For example, the fit obtained for CoMo-C, the additive impregnated catalyst, is reported on Figure 33. The other fits are present in Appendix 2. Figure 33 Kinetic fitting obtained for CoMo-C at standard conditions 43

60 However, the fit did not simulate perfectly the experimental results. Indeed, one can see that, for high temperatures, the experimental curve obtained did not follow exactly the same tendency as for other temperatures. It can be due to various reasons: - First, the variability of the test and GC analysis; - Secondly, the average kinetic parameters could change with the conversion rate of 4,6- DMDBT owed to many changes in the mechanism such as inhibitions and kinetically limiting steps; - Also, the adsorption of 4,6-DMDBT might be stronger than expected thus changing the reaction kinetic order (lower than 1), hence denying the assumption made. Nevertheless, it was not possible to improve the model due to the absence of adsorption constants in the kinetic model. This point could be studied modifying the programmed model. To decide between the various hypotheses, it would be interesting to evaluate the kinetic and simulate it without any quinoline. Indeed, if the inhibition is responsible for the fitting error, it should be drastically improved suppressing the nitrogen inhibitor. The work aiming to study the catalysts in closed conditions compared to real HDT units (LHSV and NH 3 partial pressure). The choice was done to continue with quinoline even if the fit did not represent exactly the experimental values. In the following discussion, the parameters shown will be the average parameters obtained, keeping in mind that these are probably not the good ones nevertheless, the tendency will be discussed. The various kinetic parameters determined by the experimental results and depending on the catalyst are summarized in Table 16. Table 16 Kinetic parameters obtained for each catalyst used, at standard conditions CoMo-A (Dried) CoMo-B (Calcined) CoMo-C (Additive impregnated) HYD DDS ln(k o ) (hour -1 ) 34,6 32,2 30,0 Ea (kj/mol) 164,0 152,0 140,4 ln(k o ) (hour -1 ) 33,7 31,7 29,4 Ea (kj/mol) 166,5 156,5 143,1 ln(k o ) (hour -1 ) 19,5 19,8 20,2 HDA (equilibrated) Ea (kj/mol) 102,4 101,3 96,1 ln(k o,eq ) (hour -1 ) 6,8 Ea eq (kj/mol) 134,3 44

61 First, looking at the three catalysts, the rate constants obtained corresponding to HYD pathway were slightly different and present the following hierarchy: CoMo-A (dried) > CoMo-B (calcined) > CoMo-C (additive impregnated) The activation energy was also ranked likewise. Moreover, the activation energies found for the three catalysts correspond to the same order of magnitude as reported by literature, which is around 130 kj/mol [41]. A possible explanation to why the dried catalyst has the highest activation energy may be due to a higher adsorption of inhibitory compounds such as H 2 S, NH 3 or aromatic compounds. Indeed, considering the inhibition effect as a Langmuir-Hinshelwood-type, its strength should decrease with temperature. Practically, it could be due to the fact that the additive impregnated catalyst (CoMo-C) exhibits more active sites so, it would be less impacted by inhibitors adsorption (at same partial pressure) and as a consequence, the rate constant will be less impacted by the temperature changes. Whatever the catalyst, the values obtained for the activation energies of HYD and DDS appear to be very similar to each other (at least in the same order of magnitude), which might confirm the fact, reported by Bataille et al.[19] that, at least, HYD and DDS sites share a similar nature. Nevertheless, one can also see that for the three CoMo-based catalysts, the HYD activation energy was slightly lower than for the DDS pathway, which might additionally indicate that DDS is more sensitive to reaction temperature and could also be indicative that DDS is more impacted by the inhibitors adsorption. The second hydrogenation reaction considered (HDA reaction after to DDS) did not change much from catalyst to catalyst. However, it is indeed a secondary reaction and the differences found for the kinetic parameters are not significant enough between the catalysts to be discussed. To better visualize the differences between the catalysts, one compared the rate constants, calculated from the average kinetic parameters for the temperatures of interest (Table 17). Table 17 Kinetic rate constants obtained for both HYD and DDS reactions, at standard conditions k, rate constant (h -1 ) Temperature ( C) HYD DDS CoMo-C CoMo-B CoMo-A CoMo-C CoMo-B CoMo-A 290 1,00 (100) 0,76 (76) 0,65 (65) 0,31 (100) 0,18 (58) 0,15 (48) 300 1,69 (100) 1,34 (79) 1,19 (70) 0,53 (100) 0,32 (60) 0,29 (55) 310 2,80 (100) 2,32 (83) 2,15 (77) 0,88 (100) 0,56 (64) 0,52 (59) 330* 6,36 (100) 5,64 (89) 5,61 (88) 2,04 (100) 1,39 (68) 1,38 (68) 45

62 The values presented in parenthesis represent the relative difference between in activity between the three catalysts considering the additive impregnated rate constant as the base value. Moreover, the rate constant values obtained for 330 C were extrapolated from the experimental results and it represents the typical temperature used on HDS process. From Table 17, one can see that the difference between the additive impregnated catalyst rate constant and the other catalysts rate constant becomes lower with increasing temperature, especially for hydrogenation. For direct desulfurization, the rate constants obtained for CoMo-B are higher than the ones found for CoMo-A, at low temperatures. However, with increasing temperature, the relative difference between them disappear which could might be related with the increasing desorption of inhibitory compounds thus leading to a higher activity on the dried catalyst. Concerning selectivity, it was first calculated as the direct ratio of DDS and HYD rate constants and plotted as function of the global HDS conversion. The results at various temperatures are presented in Figure 34, Figure 35 and Figure 36 (290 C, 300 C and 310 C, respectively). Increasing contact time Figure 34 Selectivity DDS/HYD as function of the total HDS conversion, at 290 C 46

63 Increasing contact time Figure 35 Selectivity DDS/HYD as function of the total HDS conversion, at 300 C Increasing contact time Figure 36 Selectivity DDS/HYD as function of the total HDS conversion, at 310 C 47