4202
Ind. Eng. Chem. Res. 2007, 46, 4202-4211
Promotion of Thiophene Hydrodesulfurization by Ammonia over Amorphous-Silica-Alumina-Supported CoMo and NiMo Sulfides Emiel J. M. Hensen, Dilip G. Poduval, and J. A. Rob van Veen* Schuit Institute of Catalysis, EindhoVen UniVersity of Technology, P.O. Box 513, 5600 MB EindhoVen, The Netherlands
Contrary to the commonly accepted notion that ammonia inhibits the hydrodesulfurization (HDS) of organosulfur compounds, an unexpected promotional effect of ammonia on the hydrodesulfurization of thiophene over amorphous-silica-alumina (ASA)-supported sulfides of CoMo (CoMo/ASA) and NiMo is reported. Addition of ammonia may increase the activity up to a factor of 1.5. The effect is specific to Coand Ni-promoted MoS2 supported on an acidic amorphous silica-alumina support. Over CoMo sulfides supported on γ-alumina and carbon, the HDS of thiophene is inhibited by ammonia. The effect is related to the presence of Brønsted acid hydroxyl groups on the amorphous silica-alumina support, as it is found that after ion-exchange with Na+ the HDS activity decreases with the NH3 partial pressure. A comparison of infrared spectra of adsorbed CO over alumina- and ASA-supported CoMo following various pretreatments (H2S/H2 vs H2S/NH3/H2) points to an electronic effect of the acidic ASA support on the active Co-Mo-S phase. It is speculated that the sulfur bond energy is increased due to this interaction. While the desulfurization of thiophene is sensitive to these changes, sulfur removal from (di)benzothiophene is not, and ammonia acts only as an inhibitor, likely due to the stronger adsorption of these reactants. Introduction Hydrodesulfurization (HDS) consisting of the removal of organic sulfur compounds from refinery streams is one of the most important catalytic processes in a modern refinery.1 Environmental regulations which require zero sulfur fuels2 are set to come in force by 2009, spurring research to further increase the performance of existing CoMo and NiMo sulfide catalysts and to develop alternative processes.3 Industrial Co(Ni)Mo catalysts consist of MoS2 slabs promoted by Co(Ni) dispersed on γ-alumina.1 The active Co-Mo-S phase is made up of atomically dispersed Co atoms located on the edges of the MoS2 particles. It is generally accepted that there are two types of Co-Mo-S structures.1,4 Type I Co-Mo-S is characterized by a lower HDS activity because of the presence of residual linkages with the subjacent alumina support. More active are fully sulfided type II Co-Mo-S structures. The strong interaction between γ-alumina and MoS2 in principle leads to type I structures unless special precautions are taken.5 Activated carbon supports have a lower tendency to interact with Mo, resulting in highly active type II structures.5 In recent years, developments have therefore been aimed at obtaining a maximum amount of type II Co-Mo-S structures6,7 in view of the need of more active alumina-supported catalysts for the production of ultralow sulfur diesel (ULSD) fuels. The sulfur tolerance of HDS catalysts is of paramount importance in industrial practice, because hydrogen sulfide decreases the rate of desulfurization of organosulfur compounds by competitive adsorption. Recently, it has been established that the choice of support has a strong influence on the extent of H2S inhibition.8,9 For sulfided CoMo and NiMo catalysts supported by activated carbon (C), γ-alumina (Al2O3), and amorphous-silica-alumina (ASA), the inhibiting effect of H2S in thiophene HDS tended to increase in the order ASA < Al2O3 < C. In addition to an electronic effect of the support, a direct * To whom correspondence should be addressed. Tel.: +31 (0)20 630 3831. Fax: +31 (0)20 630 3110. E-mail:
[email protected].
interaction between (acid) support hydroxyl groups and the active phase was suggested.8,10 As nitrogen bases interact with the acid support hydroxyl groups, we investigate here the effect of ammonia on the thiophene HDS activity for Mo-based hydrotreating catalysts. A second reason is to understand the influence of nitrogen bases on the HDS activity of organosulfur compounds, ammonia being the simplest representative of this class of inhibitors. It is important to note here that in general inhibition of HDS reactions due to competitive adsorption on the active sites by heterocyclic nitrogen compounds is much more pronounced than by ammonia. In the present study, it will be shown that NH3 promotes the desulfurization of thiophene over sulfided Co(Ni)Mo/ASA. Besides kinetic experiments, the active metal sulfide phase will be studied as a function of the gas-phase composition by infrared spectroscopy of adsorbed CO. As shown by others,11-17 the latter approach helps to understand the constitution of the active edge surface. The main finding will be that the Brønsted acidic hydroxyl groups change the quality of the active sites of the Co-Mo-S phase. This electronic effect negatively affects the catalytic performance and can be (partly) alleviated by adsorbing ammonia to these hydroxyl groups or replacing them by Na+ ions. The relevance of this unusual effect for the industrially more important hydrodesulfurization of dibenzothiophene will also be discussed. Experimental Section Catalyst Preparation. A commercial CoMo/Al2O3 catalyst (Shell C444) was used as received. For the preparation of CoMo/ ASA, NiMo/ASA, Mo/ASA, and CoMo/C, amorphous-silicaalumina (Shell, 455 m2/g, 0.67 cm3/g, 45 wt % silica) and activated carbon (Norit RX-3 EXTRA, 1190 m2/g, 1 cm3/g) supports were used. The ASA support material was calcined at 573 K for 2 h, while the carbon support was dried at 383 K overnight. All catalysts were prepared by pore volume impregnation with solutions of ammonium heptamolybdate and cobalt (nickel) nitrate. For the carbon support, the complexing agent
10.1021/ie0611756 CCC: $37.00 © 2007 American Chemical Society Published on Web 12/21/2006
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007 4203 Table 1. Loading of the Catalysts loading (wt %)
a
catalyst
Co
CoMo/Ca CoMo/Al2O3 CoMo/ASA NiMo/ASA Mo/ASA
1.6 3.4 3.6
Ni
1.2
Mo 8 9.8 10.8 8.1 8
Prepared with nitrilotriacetic acid (NTA).5
nitrilotriacetic acid was used.5 The catalysts were dried in air at 383 K for 16 h and, with the exception of CoMo/C, calcined in static air at 723 K for 2 h. Table 1 lists the loadings of the various catalysts. Characterization. For infrared spectroscopic measurements, the catalysts were ground and pressed into self-supporting wafers (density 7.5 mg/cm2) and placed into a transmission cell with CaF2 windows. Prior to recording spectra, the catalyst was sulfided in a flow of 10 vol % H2S in H2 at a total flow rate of 100 cm3/min while heating from room temperature to 673 K at a rate of 6 K/min. After an isothermal period of 2 h, the temperature was lowered to 573 K. Typically, the catalyst was subjected to four consecutive treatments with intermittent acquisition of infrared spectra of adsorbed carbon monoxide: (i) sulfidation in a flowing mixture of H2S/H2 (volume ratio 10:90); (ii) exposure to a flowing mixture of H2S/H2 (volume ratio 1:99) for 1 h; (iii) exposure to a flowing mixture of NH3/ H2S/H2 (volume ratio 0.1:0.9:99) for 1 h; and (iv) evacuation at 673 K for 15 min. Prior to recording IR spectra of adsorbed CO, the cell was evacuated for 15 min to a pressure better than 10-4 mbar. After admitting high-purity He (5 mbar) to the cell, the catalyst was cooled to 85 K. A background spectrum was recorded at this temperature. Subsequently, infrared spectra of adsorbed carbon monoxide were recorded at increasing CO partial pressures (amounts of 200 µmol/g). Finally, the catalyst was evacuated and heated to room temperature followed by sulfidation in a mixture of 10 vol % H2S in H2 at 673 K for 15 min before the next gas treatment. After step iii, no sulfidation was carried out. For transmission electron microscopy measurements, CoMo/ Al2O3 and CoMo/ASA were sulfided in a mixture of 10 vol % H2S in H2 at a total flow rate of 60 cm3/min. The temperature was increased from room temperature to 673 K at a rate of 6 K/min. After an isothermal period of 2 h, the catalyst was cooled to room temperature to 473 K in the sulfiding mixture and further in flowing He. The catalyst was then transferred via a nitrogen-flushed glovebox into a glass ampule and transported to the TEM facility. The ampule was opened in a glovebox. A few drops of ground sample suspended in n-hexane was then mounted on a carbon polymer supported on a copper grid. The sample was transferred to the microscope in a special vacuumtransfer sample holder to exclude air to prevent reoxidation. The measurements were carried out using a Philips CM30T electron microscope with a field emission gun as the source of electrons at 300 kV. Reactivity Measurements. Atmospheric gas-phase thiophene HDS was carried out in a single-pass quartz reactor. Prior to reaction, the catalysts were sulfided in a mixture of 10 vol % H2S in H2 at a total flow rate of 60 cm3/min. During sulfidation, the catalyst was heated from room temperature to 673 K at a rate of 6 K/min followed by an isothermal period of 2 h. After reaching steady-state activity after 13 h at 673 K in a mixture of 4 vol % thiophene in H2, the temperature was lowered to 623 or 573 K. Kinetic measurements were carried out by varying
Figure 1. Thiophene HDS activity (p ) 1 bar, pthiophene ) 4 kPa, T ) 623 K) as a function of the NH3 partial pressure for CoMo/ASA at various H2S partial pressures: 0 kPa (2), 0.1 kPa (b), and 1 kPa (9).
the partial pressure of NH3 and H2S between 0 and 0.2 kPa and 0 and 1 kPa, respectively. A medium-pressure stainless-steel microflow reactor system was used to evaluate the catalytic activity in gas-phase HDS of thiophene, benzothiophene (BT), and dibenzothiophene (DBT) at 30 bar. The reactor bed consisted of an amount of 50 mg catalyst diluted with 5 g SiC. After sulfidation at atmospheric pressure in a similar fashion to the atmospheric thiophene HDS experiments, the temperature was lowered to the desired reaction temperature and the pressure was increased to 30 bar. The reactor feed consisted of hydrogen flow of 500 cm3/min into which a flow of 0.1 cm3/min liquid feed was evaporated which consisted of a mixture of DBT (1 wt %) or BT (5 wt %) in n-decane. This resulted in reactant concentrations of BT and DBT of 0.1 and 0.02 vol %, respectively. In the case of thiophene HDS, the liquid flow consisted of pure thiophene, and the rate was chosen such that the thiophene gas-phase concentration was 4 vol %. The firstorder reaction rate constants for BT and DBT were determined as a function of the NH3 partial pressures. The reaction experiments at elevated pressure were carried out in the absence of H2S in the feed. Results Figure 1 depicts the thiophene HDS activity of CoMo/ASA as a function of the NH3 partial pressure at atmospheric pressure and at a temperature of 623 K. Clearly, the HDS activity increases strongly upon addition of ammonia to the feed at the various H2S partial pressures. The effect is more pronounced when the H2S partial pressure in the feed is increased. In the absence of ammonia, the HDS activity decreases with increasing H2S partial pressure. On the contrary, at the highest NH3 partial pressure employed here the HDS activity is highest for intermediate H2S partial pressure. Moreover, when no H2S is added to the feed there is a clear maximum in the HDS activity with increasing ammonia partial pressure. In these atmospheric pressure experiments, the selectivity to C4 hydrocarbon products was over 99.9% with very small amounts of tetrahydrothiophene being produced. For comparison, Figure 2 shows the thiophene HDS activities of carbon- and alumina-supported CoMo sulfide catalysts under similar conditions. Similar to CoMo/ASA, selectivities to tetrahydrothiophene are lower than 0.1%. In the absence of NH3, the activity of CoMo/C is considerably higher than that of
4204
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007 Table 2. First-Order Reaction Rate Constants for Dibenzothiophene (DBT) HDS and Selectivities to Biphenyl (BP) and Cyclohexylbenzene (CHB) as a Function of the Ammonia Partial Pressure CoMo/ASA pammonia (kPa) kBTa 0.0 1.5 3.0 4.5 c
3.7 2.9 2.2 1.8
CoMo/Al2O3
SBPb SCHBb kBT 92 98 99 100
8 2 1 0
4.3 3.9 3.6 3.5
SBPb 98.4 99.5 100 100
CoMo/C SCHBb kBTa 1.6 0.5 0 0
SBPb
SCHBb
9.5 97.7 8.7 97.6 8.1 98.2 n.d.c n.d.c
2.3 2.4 1.8 n.d.c
a First-order rate constant in mol DBT/kg b catalyst h. Selectivities in %. Not determined.
Table 3. First-Order Reaction Rate Constants for Benzothiophene (BT) and Selectivities to Ethylbenzene (EB), Styrene (STY), and 2,3-Dihydrobenzothiophene (DHBT) as a Function of the Ammonia Partial Pressure for CoMo/ASA Figure 2. Thiophene HDS activity (p ) 1 bar, pthiophene ) 4 kPa, T ) 623 K, no H2S added) as a function of the NH3 partial pressure for CoMo/ Al2O3 (9) and CoMo/C (2).
pammonia (kPa)
kBTa
SEBb
SSTYb
SDHBTb
0.0 1.5 3.0 4.5
35 30 26 24
96.1 94.7 93.6 93
0.1 0.7 1.1 1.3
3.8 4.6 5.3 5.7
a
Figure 3. Thiophene HDS activity (p ) 1 bar, pthiophene ) 4 kPa, T ) 623 K) as a function of the NH3 partial pressure for (left) NiMo/ASA at various H2S partial pressures, 0 kPa (9), 0.1 kPa (b), and 1 kPa (2), and (right) Mo/ASA (1), without addition of H2S to the feed.
CoMo/Al2O3. The much higher activity of CoMo/C derives from the presence of type II Co-Mo-S structures. Moreover, the use of an inert carbon support is known to result in higher HDS activities.5 In contrast, the strong interaction between Mo and Al2O3 leads to less active type I structures in CoMo/Al2O3. The HDS activity CoMo/Al2O3 is about two times higher than that of CoMo/ASA in the absence of ammonia. CoMo/Al2O3 and CoMo/C exhibit inhibition by ammonia in a manner which strongly points to competitive adsorption of ammonia for the active sites of thiophene HDS. To establish whether the promoting effect of ammonia on thiophene HDS is specific to Co-promoted MoS2 supported on amorphous silica-alumina, similar HDS experiments were performed for sulfided NiMo/ASA and Mo/ASA catalysts. Figure 3 shows that the HDS activity of a sulfided NiMo/ASA catalyst also strongly increases with the NH3 partial pressure. Different from the observations for CoMo/ASA is that the promoting effect of ammonia is almost similar for H2S partial pressures of 0 and 0.1 kPa but is less pronounced at the highest H2S partial pressure for NiMo/ASA. For unpromoted Mo/ASA, ammonia acts as an inhibitor to thiophene HDS. Hence, the promotional effect appears to be linked to the presence of Co or Ni active sites located on the MoS2 edges rather than the active sites of MoS2 themselves. The next step is to determine whether the NH3 promotion effect is specific to the desulfurization of thiophene. Because
First-order rate constant in mol BT/kgcatalyst h. b Selectivities in %.
of its importance in industrial hydrotreating, dibenzothiophene (DBT) was selected as reactant. Benzothiophene (BT) was included because it has an intermediate reactivity. Table 2 collects the first-order rate constants and selectivities for the conversion of DBT over CoMo/ASA, CoMo/Al2O3, and CoMo/ C. The activity differences for DBT HDS in the absence of NH3 between CoMo/Al2O3 and CoMo/ASA are smaller than for thiophene HDS. Ammonia clearly inhibits the HDS activity for all three catalysts. The selectivity to cyclohexylbenzene is relatively low in the absence of ammonia indicating that the major reaction path for desulfurization is via direct hydrogenolysis of the carbon-sulfur bond. Since hydrogenation of biphenyl to cyclohexylbenzyl does not occur under these reaction conditions,9 the contribution of the hydrogenation reaction path is below 10%. This finding is in line with the notion that only for sterically hindered DBTs does prehydrogenation of one of the aromatic rings contribute more significantly to hydrodesulfurization because the direct desulfurization pathway is strongly inhibited.9,18,19 On the other hand, the selectivity data indicate that the hydrogenative pathway decreases in importance in the following order: CoMo/ASA > CoMo/C > CoMo/Al2O3. This trend agrees with differences noted earlier by Hensen et al.20 who reported that the ratio of the hydrogenation and direct pathway for DBT hydrodesulfurization is higher over Mo/ASA than over Mo/Al2O3. Upon addition of ammonia, the overall HDS activity is decreased. Especially, the hydrogenation pathway is impeded, and for CoMo/Al2O3 and CoMo/ASA no CHB is found at the highest ammonia partial pressure. The relatively strong inhibiting effect of NH3 has been noted before for DBT HDS.21 The complex inhibiting effect on aromatics hydrogenation has also been studied by the group of Kasztelan.22 Interestingly, the inhibiting effect of ammonia on the hydrogenation pathway appears to be less pronounced for CoMo/C. Table 3 summarizes the results for benzothiophene HDS over CoMo/ASA. The reactivity of BT is an order of magnitude higher than that of DBT. Again, addition of ammonia strongly inhibits the HDS activity. In the desulfurization of BT, prehydrogenation of the thiophenic ring leads to 2,3-dihydrobenzothiophene (DHBT). It is not clear whether ethylbenzene results from desulfurization of DHBT or from hydrogenation of styrene.23 However, the higher selectivity to styrene with
Ind. Eng. Chem. Res., Vol. 46, No. 12, 2007 4205
Figure 4. Thiophene HDS activity (p ) 30 bar, pthiophene ) 4 kPa, T ) 623 K, no H2S added) as a function of the NH3 partial pressure for CoMo/ ASA (2). Table 4. Average Slab Length, Stacking Degree, and MoS2 Dispersion catalyst
daverage (nm)
Saverage
fMoa
CoMo/Al2O3 CoMo/ASA
2.7 3.3
1.2 1.5
0.42 0.34
a Dispersion of MoS expressed as the fraction of Mo at the edges of 2 the MoS2 particles.20
increasing ammonia partial pressure must be due to inhibition of its hydrogenation to ethylbenzene. To ensure that the absence of promotion by ammonia is not due to the higher (H2) reaction pressure, a thiophene HDS activity experiment was carried out for CoMo/ASA at 30 bar. Figure 4 shows that the promotional effect of ammonia on the reaction rate for thiophene hydrodesulfurization is reproduced under these reaction conditions. Strikingly, the trend is very similar to that observed for thiophene HDS at atmospheric pressure under otherwise similar conditions. We note here that the selectivity to THT is much higher at elevated pressure (∼20%) than at atmospheric pressure (
