Effects of Catalyst Properties on Hydrodesulfurization Activity for Sulfur

Aug 4, 2016 - Earth and Mineral Sciences (EMS) Energy Institute and Department of Energy and Mineral Engineering, The Pennsylvania State University, 1...
1 downloads 13 Views 2MB Size
Download PDF
Article pubs.acs.org/EF

Effects of Catalyst Properties on Hydrodesulfurization Activity for Sulfur Removal from Fluid Catalytic Cracking Decant Oils Joseph P. Abrahamson, Ronald T. Wincek, and Semih Eser* Earth and Mineral Sciences (EMS) Energy Institute and Department of Energy and Mineral Engineering, The Pennsylvania State University, 114A Hosler Building, University Park, Pennsylvania 16802, United States ABSTRACT: Removing sulfur from larger ring systems in fluid catalytic cracking decant oils used as needle coke feedstock is the most effective way of reducing the needle coke sulfur content. The large sulfur compounds found in decant oil are incorporated into coke in larger proportions than smaller sulfur compounds upon carbonization. The desirable outcome of decant oil hydrodesulfurization is, therefore, removing sulfur selectively from large polyaromatic ring systems with minimum hydrogen consumption. This study investigates the effects of catalyst properties on hydrodesulfurization activity to remove sulfur from decant oils. Two decant oils (DO-HS and DO-LS) representing a high (2.5 wt %) and low (0.9 wt %) sulfur content decant oil and their vacuum distillation fractions were hydrotreated in a fixed-bed flow reactor. Four catalysts (with varying average pore sizes, promoter atoms, and supports) were prepared with sequential incipient wetness impregnation to evaluate their activities for hydrodesulfurization and hydrogenation of decant oils. An increase in the average pore diameter from 7 to 14 nm for CoMo catalysts supported on Al2O3 proved capable of meeting the desired requirements for hydrodesulfurization of decant oil used in needle coke production. Of the four catalysts evaluated, CoMo supported on TiO2 outperformed the other three catalysts supported on Al2O3; however, focus was placed on the Al2O3-supported catalysts as a result of the superior mechanical integrity and proven longevity of Al2O3 in hydrodesulfurization reactors. It was shown by proton nuclear magnetic resonance that promoting Mo supported on Al2O3 with Ni instead of Co results in equivalent hydrogenation activity and decreased desulfurization. Upon carbonization of treated oils, the sulfur content of the resulting coke increased from the feed treated with a CoMo catalyst supported on Al2O3 with an average pore diameter of 7 nm, whereas coke produced from feeds treated over the CoMo catalyst supported on Al2O3 with an average pore diameter of 14 nm had a lower sulfur content compared to the feed. Therefore, with a proper catalyst design, sulfur in decant oil that tends to be retained in the coke can selectively be removed. Thus, hydrodesulfurization can favor the direct desulfurization route over the hydrogenation route by employing high reaction temperatures and modest hydrogen pressures.

■

INTRODUCTION Delayed coking of decant oils (DOs) leads to the development of a carbonaceous mesophase, a liquid crystalline phase, that is responsible for the structural anisotropy of needle coke.1 Because of the anisotropic microstructure, needle coke is used as the filler for manufacturing graphite electrodes for electric arc furnaces. When the sulfur content of needle coke is high (>0.8 wt %), an irreversible volume expansion (puffing) takes place when extruded electrodes are subjected to graphitization heat treatment.2 Puffing in needle coke results primarily from the evolution of sulfur in the form of CS2 and H2S.3 Puffing causes the formation of microcracks, reducing the quality of electrodes. The most practical and efficient process to control the sulfur levels in needle cokes is catalytic hydrodesulfurization (HDS) of the feedstock, DO, prior to coking.4 As the trend of processing heavier crude oils containing higher levels of sulfur continues, needle coke feedstocks will require processing to reduce the sulfur content to acceptable levels. The higher the sulfur concentration in the starting crude oil, the higher the sulfur concentration will be in the DO produced by fluid catalytic cracking during petroleum refining. DO is the most common feedstock for needle coke production and represents 3−7% of the total products obtained from a fluid catalytic cracker (FCC).4 Refiners do not typically hydrotreat FCC feed, and this results in DO with a high sulfur content. 4 Hydrotreating DO offers the potential to increase the available © XXXX American Chemical Society

DOs that can be used as low-sulfur feedstock for needle coke production by removing sulfur in the form of H2S while essentially maintaining the carbon skeleton of the constituent compounds. The major sulfur compounds found in DO are polyaromatic sulfur compounds (PASCs). These are mainly alkyl derivatives of benzothiophene (BT), dibenzothiophene (DBT), benzonaphthothiophene (BNT), and larger condensed thiophenes.5 Because the larger thiophenes are not separated by chromatographic methods, their exact molecular structures are unknown.6,7 During HDS, the overall reaction rate is known or considered to be limited by pore diffusion.8 Several researchers have studied catalyst pore diffusion of heavy oils with techniques such as diaphragm cell,9 sorptive,10 and kinetic11 methods. It should be noted, however, that industrial CoMo HDS catalysts are typically supported on Al2O3 with a broad pore size distribution. The broad pore size distribution and tortious pore connectivity makes it difficult to distinguish intrinsic catalytic kinetics from pore diffusion effects.12 It is found that intrinsic diffusivity decreases significantly with an increasing molecular weight, and diffusion deviates from bulk Received: June 14, 2016 Revised: August 2, 2016

A

DOI: 10.1021/acs.energyfuels.6b01441 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels Table 1. Catalyst Properties property

catalyst A

catalyst B

catalyst C

catalyst D

metal loading (wt %) promoter atom (wt %) support material pore diameter (nm) surface area (m2/g)

20.5 MoO3 4.3 CoO γ-Al2O3 7 (81% at 5−10 nm) 214

20.5 MoO3 4.3 CoO γ-Al2O3 14 (86% at 12−16 nm) 110

20.5 MoO3 4.3 CoO γ-Al2O3 14 (86% at 12−16 nm) 106

16.6 MoO3 3.5 CoO anatase TiO2 14 (78% at 12−16 nm) 92

the reactor. The addition of hydrogen to aromatic rings during HDS is an exothermic reaction, giving rise to increased catalyst bed temperatures. Catalyst bed dilution with inert glass beads helps to reduce the temperature gradient and increases the contact efficiency. All experiments used a fixed oil feed flow rate of 1 mL/min, which corresponds to a liquid hourly space velocity (LHSV) of 1.0 h−1. To achieve a LHSV of 1.0 h−1, the reactor was packed with 60 mL of catalyst. A total of 60 mL of catalyst is comprised of 30.1 g of Al2O3 support or 39.5 g of TiO2 support, 8.2 g of MoO3, and 1.7 g of CoO or 1.7 g of NiO. Metal oxides were converted to HDS active metal sulfides prior to HDS following the previously reported procedure.19 The hydrogen gas flow rate was fixed at 1000 mL/min for all experiments. Hydrogen pressures of 3.45 and 6.89 MPa were used in the experiments at 350 and 400 °C. Upon exiting the reactor, the treated feed was cooled using a water-cooled heat exchanger before flowing through a backpressure regulator. More details about the operation of the reactor can be found in a previously submitted paper.19 Catalyst Preparation and Activation. Four catalysts were used in this study. Catalyst A was supplied by an industrial catalyst manufacture. Catalyst A is advertised as a hydrotreating catalyst with high activity for DDS and low activity for HYD and catalytic cracking; therefore, vatalyst A meets the objective requirements. However, like most commercially available hydrotreating catalysts on the market, catalyst A has an average pore diameter of 7 nm. The commercially available hydrotreating catalysts with larger pore diameters are bifunctional hydrotreating and hydrocracking catalysts. This comes as no surprise because upgrading heavy oil is performed to produce transportation fuels by hydrocracking and hydrotreating rather than just hydrotreating the constituent compounds. Because cracking is not desired in HDS of DO, catalysts B, C, and D were prepared in the laboratory using incipient wetness impregnation. Table 1 compares the properties of the commercial catalyst A with prepared catalysts B−D, where the surface area and pore diameters given in Table 1 were found using a previously reported procedure19 and measured using a Micromeritics ASAP 2020 surface area and porosimetry analyzer. The Al2O3 support used for catalysts B and C and the TiO2 support used for catalyst D were purchased from Alfa Aesar and have equal surface areas of 150 m2/g, prior to active metal loading. Catalysts B−D were loaded with cationic metal salt precursors in aqueous solution. Cationic metal precursors were selected because the zero point charge of TiO2 is 6 and that of Al2O3 is 7. Therefore, in aqueous solution, the adsorbents on TiO2 will be cationic, and because Al2O3 is amphoteric, the adsorbents can be cationic or anionic. Metals were loaded sequentially using metal salt precursors. Ammonium heptamolybdate tetrahydrate, (NH4)6Mo7O24·4H2O, was weighed out to 10.04 g and mixed into a 23.18 or 15.01 mL aqueous solution. The total pore volume of 30.1 g of Al2O3 support used for catalysts B and C is 23.18 mL, and the total pore volume of 39.5 g of TiO2 support used for catalyst D is 15.01 mL. The solution is just enough to fill the pores, and distribution is aided by the capillary effects caused by adding just enough solution to wet the internal volume. The solution was added dropwise to the support and continuously stirred. The impregnated support was then air-dried overnight at 105 °C. Following drying, the catalysts were calcined in a box furnace at 430 °C for 3 h. Upon cooling, the procedure was repeated for the promoter atoms. Catalysts were designed to isolate individual differences that allowed for an investigation of pore size, promoter atom, and support material effects on the hydrotreatment of DO. Catalyst B was prepared to study the effects of the pore size on HDS and HYD of DO compared to catalyst

diffusivity when the ratio of the molecular diameter to the pore diameter is larger than 0.04. Below 0.04, diffusion is considered to be near free diffusion.12 However, free diffusion rates do not maximize HDS. The highest HDS activity occurs when molecular diameter/pore diameter is between 0.06 and 0.1.12 HDS of heavy oils, such as DO, therefore, requires a large mesoporous catalyst to be effective.13 Typical DOs have an average boiling point between 380 and 440 °C according to Escallon et al.14 Common polyaromatic hydrocarbons (PAHs) within this boiling range found in DO are chrysene and pyrene. Chyrsene and pyrene have dimensions of 1.38 × 0.8 nm and 1.16 × 0.92 nm, respectively.15 The average of their longest dimension is 1.27 nm. According to the ideal molecular diameter/pore diameter, the ideal HDS catalyst pore diameter for DO is between 13 and 21 nm. It has been demonstrated that there is a strong relationship between carbonaceous mesophase development and the aromatic/aliphatic hydrocarbon ratio in delayed coking feedstocks.16 Therefore, hydrogenation (HYD) during HDS should be kept to a minimum to maintain the high hydrogen aromaticity of DO. Sulfur reduction from HDS is not expected to impact mesophase development because organic sulfur has not been shown to impact mesophase development when under 3 wt %.17 The ideal HDS catalyst for treatment of DO used as needle coke feedstock should possess a high activity for direct desulfurization (DDS) and a low activity for HYD. Additionally, the catalyst should selectively remove sulfur from the heavy ends of DO because these compounds are most likely to become incorporated into the coke. Therefore, this study represents a novel approach for selecting an optimum catalyst for specifically targeting high-molecular-weight sulfur species for sulfur removal without any undesired HYD activity. The objective of this work is to investigate the effects of catalyst properties, including pore size, promoter atom, and support material, on HDS and HYD activities. A companion paper18 investigates the effects of HYD on the subsequent mesophase development and coke yield.

■

EXPERIMENTAL SECTION

Two parent FCC DOs, DO-HS-P and DO-LS-P, representing a high (2.5 wt %) and low (0.9 wt %) sulfur content DO that may be available from U.S. refineries were selected for this study. Vacuum distillation fractions were also evaluated. DOs were separated into fractions using a simple vacuum distillation setup with accordance to ASTM method D1160-03 (Standard Test Method for Distillation of Petroleum Products at Reduced Pressure). Additional details about the vacuum distillation procedure are provided in a companion paper.18 Both parent oils and their middle fraction from distillation, DO-HS and DO-LS, account for the four feedstocks hydrotreated in this study. Operation of the Laboratory-Scale Fixed-Bed Hydrotreater. The fixed-bed hydrotreater reactor used in the study has been described in a previous paper.19 This reactor enabled individual variables to be studied under conditions relevant to the operation of an industrial-scale HDS unit. The 316 stainless-steel reactor is 79.4 cm long and has an internal diameter of 2.12 cm. The DO and hydrogen mixtures traverse through a preheat coil that wraps the upper length of B

DOI: 10.1021/acs.energyfuels.6b01441 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

Figure 1. Hydrogen functionality of 1H NMR DO spectra. fluidized sand bath. Carbonization experiments were carried out at 500 °C for a duration of 5 h. After carbonization, semicoke bars were extracted and washed with dichloromethane to remove any liquid product. Additional details about the carbonization procedure have been previously published in refs 5, 6, 7, and 19.

A, with a key focus on desulfurization of large polyaromatic sulfur compounds (PASCs). Catalyst C was used to evaluate a promoter atom change from Co to Ni compared to catalyst B. Catalyst D was prepared to study the effects of support material, TiO2 for catalyst D and γ-Al2O3 for catalyst B, on HDS and HDY. Gas Chromatography (GC)−Pulsed Flame Photometric Detector (PFPD). Sulfur compound analysis was performed on an Agilent 6890 series GC fitted with an OI analytical model 5380 sulfurselective PFPD. Model sulfur compounds used for GC−PFPD calibration included BT, DBT, and BNT. Only a few standards were used as a result of limited availability of large PASCs. Samples were injected in splitless mode into a Restek RXI column. The injector was held at 290 °C. The GC oven was programmed to increase in temperature from 25 to 140 °C at a rate of 10 °C/min and then increased from 140 to 330 °C at a rate of 5 °C/min. Sample preparation followed the previously reported procedure.19 Total Sulfur Determination. The total sulfur content was determined with a model SC 144-DR sulfur−carbon analyzer manufactured by Leco Corporation. Proton Nuclear Magnetic Resonance (1H NMR). The extent of HYD was measured by analyzing the hydrotreated products with NMR. The 1H NMR data for parent oils, fractionated oils, and HDS products were collected on a Bruker DRX 400 spectrometer. Samples were dissolved in deuterochlorofrom. Chloroform-d, CDCl3, was rated at 99.8 atom % D. The 0.2 atom % H is known to have a solvent residual peak of 7.26 ppm. Both the solvent residual peak at 7.26 ppm and tetramethlysilane (TMS) peak at 0 ppm were used as internal standards. Peak assignments are given in Table 3 in our companion paper18 and displayed here visually in Figure 1. The peak assignments in Figure 1 were adapted from Rodriguez et al.20 The water peak at 1.59 ppm is from contamination of the deuterated chloroform solvent, and the dichloromethane peak at 5.29 ppm is an additional contaminate. NMR data were processed using Bruker Topinspin 2.1 software. Carbonization of Needle Coke Feedstocks. Semicoke production permits the evaluation of the feedstocks relative propensity to yield anisotropic domains during carbonization. Vertical tubing bomb reactors were filled with 4 g of feedstock and purged of oxygen with a nitrogen manifold prior to being plunged into a preheated

■

RESULTS AND DISCUSSION It is the high hydrogen aromaticity of DO that gives rise to a well-developed mesophase that is responsible for producing the premium-grade needle coke.16 Therefore, emphasis was placed on identifying the catalyst and conditions that are capable of reducing the sulfur content down to acceptable levels while minimizing HYD. An additional focus was on the possibility of selectively removing sulfur that tends to be retained in the resulting coke. Sulfur compounds likely to become incorporated into the coke are the larger sulfur-containing compounds. This section aims to identify the catalyst properties and conditions that promote DDS, minimize HYD, and selectively desulfurize large PASCs. The extent of HYD was determined by the 1H NMR signal reduction from the aromatic hydrogen region (6−9.3 ppm). This approach assumes that the observed decrease in aromatic hydrogen content comes solely from HYD of the aromatic rings, ignoring any possible catalytic cracking and coke deposition onto the catalyst. It was observed that the decrease in aromatic signal is largely accounted for by the increase in the hydroaromatic signal in the region of 1.7−2.0 ppm. The increase in hydroaromatics is directly related to the decrease in total aromatic hydrogen content, i.e., HYD of aromatic rings. This trend is illustrated in Figure 2. The naming scheme used hereafter is oil-catalyst-temperature-pressure; for instance, DO-HS-B-400-6.89 represents the middle distillation fraction of the high-sulfur DO treated over catalyst B at 400 °C and 6.89 MPa. The total signal loss from the aromatic region of DO-HS-B-400-6.89 compared to untreated DO-HS is 5.25%, while the hydroaromatic region C

DOI: 10.1021/acs.energyfuels.6b01441 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

Table 2. Sulfur Reduction and HYD Extent Ratio (SR/HYD) for DO-HS sample

aromatic signal (%) (6.0−9.3 ppm)

DO-HS

Figure 2. 1H NMR spectra from DO-HS treated over catalyst B with varying conditions.

increases by 4.86% of the total signal. Therefore, 92+% of the hydrogenated aromatics appear to end up in the hydroaromatic region. The remaining hydrogenated product is in the form of naphthenic structures and observed by the small signal increase between 1.1 and 1.6 ppm. A similar correlation exists for all hydrotreated samples, as displayed in Figure 2. The hydrogen aromatic signal decrease was compared to the carbon aromatic signal decrease with 13C NMR and was in good accordance. The 13C NMR spectra in Figure 3 shows that catalyst A has a higher HYD activity than that of catalyst B.

sample

aromatic signal decrease (%)

29.0 sulfur reduction (%)

ratio of SR/HYD

DO-HS-A-350-3.45 DO-HS-B-350-3.45 DO-HS-A-350-6.89 DO-HS-B-350-6.89 DO-HS-A-400-3.45 DO-HS-B-400-3.45 DO-HS-A-400-6.89 DO-HS-B-400-6.89 DO-HS-C-400-6.89 DO-HS-D-400-6.89

20.0 14.8 33.7 30.6 18.4 14.1 30.9 28.8 28.3 24.4

85.6 84.6 87.8 84.9 93.3 91.9 96.1 93.0 90.5 97.5

4.28 5.72 2.43 2.77 5.07 6.52 3.11 3.23 3.20 4.00

any catalyst. Catalyst D has both the highest HDS and lowest HYD activity. The increase in HDS activity of TiO2-supported CoMo over Al2O3-supported CoMo is believed to be due to an increase in active sulfur vacancy sites on the TiO2-supported catalyst.21 It is believed that a fraction of Ti4+, the naturally occurring species of Ti in TiO2, is reduced under reaction conditions to Ti3+.21 When in intimate contact with the MoS2 slabs, the excess electron in Ti3+ can be injected into the Mo 3d conduction band, creating a sulfur vacancy and active HDS site.21 The lower HYD activity of TiO2-supported CoMo over Al2O3-supported CoMo may be due to the higher Brønsted acid strength of Al2O3. It has been demonstrated by Busca22 that TiO2 has very few Brønsted acid sites compared to Al2O3, and at high temperatures, Al2O3 can act in Brønsted acidic catalysis. Therefore, Al2O3 will have a higher HYD and hydrocracking activity than that of TiO2. However, there was no recognizable increase in the 1H NMR signal from treated oils between 0.5 and 1.1 ppm (aliphatic CH3 or CH2 attached γ or further to the ring and paraffinic compounds). It can, therefore, be inferred that hydrocracking was not prevalent during the hydrotreatment of samples, albeit coke deposition on the catalyst was not analyzed nor were the product gases. Even though TiO2 support material has been developed with surface areas comparable to Al2O3, its use as a hydrodesulfurization catalyst carrier has not been widely employed as a result of the inferior mechanical integrity of TiO2 compared to that of Al2O3. In HDS catalysts, Ni and Co are often interchanged or used in combination as a promoter for Al2O3-supported Mo catalysts. It is widely reported in the literature23,24 that NiMo supported on Al2O3 possesses a higher HYD activity than that of CoMo on Al2O3. However, it has been reported by Azizi et al.25 that the higher HYD activity of Ni is more pronounced for small, 1−2 member, PAHs and levels off for larger PAHs. The DOs used here are comprised mainly of 3+ member PAHs as determined by gas chromatography/mass spectrometry (GC/ MS) in a previously submitted paper,19 and HYD activity was found to be similar between both Co (catalyst B) and Ni (catalyst C) promoted Mo catalysts supported on Al2O3 as suggested by Azizi et al.25 As seen in Table 2, the SR/HYD ratio of Ni-promoted catalyst C is slightly less than that of the Co-promoted catalyst B. The Ni catalyst has similar HYD extents but appreciably less sulfur reduction than that of the Co catalyst. Of the three catalysts supported on Al2O3, catalyst B

Figure 3. 13C NMR spectra of (top) DO-HS-P-B-400-6.89 and (bottom) DO-HS-P-A-400-6.89.

The hydroaromatic and aromatic regions in the 13C NMR spectra have chemical shifts of 14−35 and 110−150 ppm, respectively. Catalyst Activity. Table 2 provides the SR/HYD ratio for the high-sulfur DO middle fraction (DO-HS), where SR is defined as the total sulfur reduction in weight percent based on the feed sulfur content and HYD is the aromatic hydrogen signal decrease as a percentage of the aromatic signal of the original DO, as measured by 1H NMR. The higher the SR/ HYD ratio, the more selective the catalyst is to the DDS pathway. Table 2 shows that, although catalyst A has a higher overall desulfurization activity, it generally gives the lowest SR/ HYD ratio as a result of a high HYD activity. Catalyst D provides by far the highest direct sulfur reduction selectivity of D

DOI: 10.1021/acs.energyfuels.6b01441 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

The largest GC amenable sulfur compounds show a similar trend as determined by GC−PFPD. The chromatogram from DO-HS-P is given in Figure 4. DO-HS-P contains small concentrations of BT and methylated BT. No BTs were present in the low-sulfur parent DO. BT was readily desulfurized by all four catalysts. Standard PASCs used in the GC-PFPD calibration included BT, DBT, and BNT. Only a few standards were used as a result of limited availability of large PASCs and because GC−PFPD is used here to demonstrate the catalyst activities toward the larger PASCs found in DO and not for specific species identification. Therefore, the methyl-substituted PASC peak labeling in Figures 4 and 5 is a best guess approach based on previous work,5 and additional sulfur species are likely present. A central focus of this study was the possibility of selectively removing sulfur that tends to be retained in the resulting coke. Sulfur compounds likely to become incorporated into the coke are the larger sulfur-containing compounds. The larger pore size of catalyst B was more effective toward the HDS of these larger compounds than that of the smaller pore size of catalyst A. The chromatogram in Figure 5 starts with the sterically hindered tetramethyl-DBT, and as seen, the activities between the two catalysts for this compound are similar. However, catalyst B has higher HDS activity toward the larger, 4+ member, polyaromatic sulfur compounds. This is believed to be caused by limited pore diffusion of these large PASCs through catalyst A. In an effort to investigate the possible benefits of selectively removing sulfur from the larger compounds that are likely to become incorporated into the coke, coke sulfur levels were compared to feed sulfur levels. Resulting Semicoke Sulfur Content. Selected samples of semicokes produced by carbonization were analyzed for the total sulfur content. The results from total sulfur analysis are given in Table 4. The sulfur contents of semicokes produced from parent DOs (DO-HS-P and DO-LS-P) are similar to the starting sulfur content in the corresponding liquid feed. The increase in coke sulfur from middle fractions (DO-HS and DO-LS) and VBs (VB-HS and VB-LS) suggests that large sulfur compounds found in these fractions are incorporated into the semicoke in larger proportions. The increase in coke sulfur from treated middle fractions is also apparent (DO-HS-A-350-1000 and DOHS-B-350-1000). Of most interest is the different sulfur contents in semicokes from parent oils treated with catalysts A and B. Semicokes produced from parent oils treated over catalyst B show a slight decrease in the sulfur content compared to that of the feed, whereas semicokes from the oils treated with catalyst A have a higher sulfur content than the feed. As shown in Table 3, catalyst B has an increased activity toward larger sulfur compounds found in the VBs and as shown here in Table 4. Catalyst B selectively removes sulfur that would have been incorporated into the coke. In contrast, catalyst A has a higher activity toward smaller compounds that are less likely to become incorporated into the coke and more likely to end up in the coker gas oil.

gave the highest SR/HYD ratio and would be desirable to meet the objectives that maximize DDS and minimize HYD extent. Effects of the Operation Temperature and Hydrogen Pressure on Desulfurization and HYD Activity. As displayed in Table 2, an increase in the temperature from 350 to 400 °C (with everything else being the same) results in a greater sulfur reduction and suppression of HYD in all cases. The observed reduction of HYD with an increased temperature is not surprising considering that the reaction is exothermic and restricted by thermodynamic equilibrium and resonance stabilization of the PAHs under high temperatures. Farag et al.26 reported similar findings that there is little formation of hydrogenated compounds from HDS of 4,6-dimethyldibenzothiophene over CoMo supported on Al2O3 when temperatures are in excess of 380 °C. The effect of increasing the hydrogen pressure from 3.45 to 6.89 MPa also increases HDS activity, albeit small in comparison to the effect of increasing the temperature, and is accompanied by a substantial increase in HYD. Any added sulfur reduction brought about by an increase in pressure is quite small, and it would appear that pressures above 3.45 MPa are not needed for desulfurization of DO. Additionally, increasing pressure results in a sharp increase in HYD over all catalysts. Therefore, a high reaction temperature and low pressure are the preferred operating conditions to limit the undesirable HYD of aromatic ring systems. Desulfurization of Large Polycyclic Aromatic Sulfur Compounds. To evaluate the sulfur reduction on vacuum bottoms (VBs) that contain the largest sulfur compounds, treated parent oils were vacuum-distilled and the treated VBs with boiling points above 490 °C were collected and analyzed. VBs obtained from vacuum distillation prior to HDS treatment were not hydrotreated as a result of their high melting points and inability to flow through the HDS reactor. The VBs represent the largest sulfur-containing compounds, which are most likely to experience pore diffusion limitations. The desulfurization activity toward these large PASCs is greater for catalyst B than catalyst A, as shown in Table 3. However, Table 3. Sulfur Content (wt %) of Hydrotreated Parent High-Sulfur DO and VBs sample

temperature

pressure

catalyst

sulfur (±0.01, wt %)

DO-HS-P DO-HS-P DO-HS-P DO-HS-P VB-HS VB-HS VB-HS VB-HS

350 400 350 400 350 400 350 400

3.45 6.89 3.45 6.89 3.45 6.89 3.45 6.89

A A B B A A B B

0.70 0.22 0.79 0.30 0.77 0.47 0.74 0.45

sulfur reduction (%) 72.1 91.2 68.5 88.0

catalyst A has a higher HDS activity for the parent oils than that of catalyst B. The higher activity of catalyst A is most likely due to an increase in active site dispersion. This is a logical assumption considering that catalyst A has a much higher surface area available for dispersion, as displayed in Table 1; however, dispersion was not explicitly measured. Therefore, because catalyst B is more active than catalyst A for reduction of the largest sulfur compounds, it can be inferred that the smaller pores do limit pore diffusion and the resulting desulfurization of large compounds.

■

CONCLUSION Two DOs and their vacuum distillate middle fractions were hydrotreated in a pilot-scale fixed-bed hydrotreater. Four catalysts were investigated for the purpose of identifying the properties and reaction conditions that promote DDS, suppress HYD, and selectively remove sulfur from large polyaromatic sulfur compounds likely to be retained in the resulting coke. Of E

DOI: 10.1021/acs.energyfuels.6b01441 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels

Figure 4. GC−PFPD high-sulfur parent DO chromatogram.

Figure 5. GC−PFPD chromatograms of C4-DBT and larger PASCs from DO-HS treated over catalysts A and B.

diameter of 14 nm. Upon carbonization of oils, the sulfur contents of the resulting cokes increased from the feeds treated with a CoMo catalyst supported on Al2O3 with an average pore diameter of 7 nm, whereas coke produced from feeds treated over the CoMo catalyst supported on Al2O3 with an average pore diameter of 14 nm had a lower sulfur content compared to the feeds. Therefore, with a proper catalyst design, sulfur that is retained in the coke can selectively be removed and the desulfurization can favor the DDS route over the HYD route by employing high reaction temperatures and modest hydrogen pressures.

Table 4. Sulfur Content (wt %) of Feedstock and Resulting Semicoke sample

feedstock (±0.01, wt % sulfur)

semicoke (±0.03, wt % sulfur)

DO-HS DO-LS DO-HS-P DO-LS-P VB-HS VB-LS DO-HS-A-350-6.89 DO-HS-B-350-6.89 DO-HS-P-A-350-3.45 DO-HS−P-B-350-3.45 DO-LS-P-A-400-6.89 DO-LS−P-B-400-6.89

2.85 0.87 2.51 0.94 1.98 1.05 0.52 0.43 0.70 0.79 0.11 0.29

2.97 0.95 2.48 0.96 2.21 1.10 0.56 0.45 0.79 0.75 0.15 0.29

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■ ■

the four catalysts evaluated, CoMo supported on TiO2 with an average pore diameter of 14 nm outperformed the other three catalysts supported on Al2O3; however, focus was placed on the Al2O3-supported catalysts as a result of the superior mechanical integrity and proven longevity of Al2O3 in HDS reactors. It was shown by 1H NMR that promoting Mo supported on Al2O3 with Ni in place of Co results in a decreased sulfur reduction and equivalent HYD. Chromatograms from GC−PFPD showed that desulfurization of large compounds was limited for smaller pored, 7 nm, CoMo supported on Al2O3 compared to a CoMo catalyst supported on Al2O3 with an average pore

ACKNOWLEDGMENTS The authors thank Dr. Frank Dorman and Gal Kreitman of Penn State for their assistance gathering the GC−PFPF data. REFERENCES

(1) Marsh, H.; Menendez, R. Fuel Process. Technol. 1988, 20, 269. (2) Brandtzaeg, S.; Oye, H. Carbon 1988, 26, 163. (3) Fujimoto, K.; Mochida, I.; Todo, Y.; Oyama, T.; Yamashita, R.; Marsh, H. Carbon 1989, 27, 909. (4) Guercio, V. Oil Gas J. 2010, 108, 96.

F

DOI: 10.1021/acs.energyfuels.6b01441 Energy Fuels XXXX, XXX, XXX−XXX

Article

Energy & Fuels (5) Eser, S.; Wang, G. Energy Fuels 2007, 21, 3573. (6) Filley, R.; Eser, S. Energy Fuels 1997, 11, 623. (7) Wang, G.; Eser, S. Energy Fuels 2007, 21, 3563. (8) Chen, S.; Dong, P.; Xu, K.; Qi, Y.; Wang, D. Catal. Today 2007, 125, 143. (9) Sane, R.; Tsotsis, T.; Webster, I.; Ravi-Kumar, V. Chem. Eng. Sci. 1992, 47, 2683. (10) Chen, Z.; Xu, C.; Gao, J.; Zhao, S.; Xu, Z. AIChE J. 2010, 56, 2030. (11) Wang, G.; Chen, Z.; Lan, X.; Wang, W.; Xu, C.; Gao, J. Chem. Eng. Sci. 2011, 66, 1200. (12) Wang, Z.; Chen, S.; Pei, J.; Chen, A.; Zhang, J.; Xu, Z.; Benziger, J. AIChE J. 2014, 60, 3267. (13) Ancheyta, J.; Rana, M.; Furimsky, E. Catal. Today 2005, 109, 3. (14) Escallon, M. M.; Fonseca, D. A.; Schobert, H. H. Energy Fuels 2013, 27, 478. (15) Sander, L.; Wise, S. Polycyclic Aromatic Hydrocarbon Structure Index; National Institute of Standards and Technology (NIST): Gaithersburg, MD, 1997; NIST SP 922, DOI: 10.6028/NIST.SP.922. (16) Eser, S.; Jenkins, R. Carbon 1989, 27, 889. (17) Gu, Y.; Weinberg, V.; Sadeghi, M.; Yen, T. Effects of Sulfur and Metals on Mesophase Formation in Coal Liquid Asphaltene and Petroleum Pitch. In Polymers for Fibers and Elastomers; Arthur, J. C., Jr., Diefendorf, R. J., Yen, T. F., Needles, H. L., Schaefgen, J. R., Jaffe, M., Logothetis, A. L., Eds.; American Chemical Society (ACS): Washington, D.C., 1984; Vol. 260, Chapter 16, p 263, DOI: 10.1021/bk-1984-0260.ch016. (18) Abrahamson, J. P.; Wincek, R. T.; Eser, S. Scheme for Hydrotreatment of Fluid Catalytic Cracking Decant Oil with Reduced Hydrogen Consumption and High Needle Coke Yield upon Carbonization. Energy Fuels 2016. (19) Wincek, R. T.; Abrahamson, J. P.; Eser, S. Hydrodesulfurization of Fluid Catalytic Cracking Decant Oils in a Laboratory Flow Reactor and Effect of Hydrodesulfurization on Subsequent Coking,. Energy Fuels 2016, DOI: 10.1021/acs.energyfuels.6b00843. (20) Rodriguez, J.; Tierney, J. W.; Wender, I. Fuel 1994, 73, 1870. (21) Ramirez, J.; Macias, G.; Cedeno, L.; Gutierrez-Alejandre, A.; Cuevas, R.; Castillo, P. Catal. Today 2004, 98, 19. (22) Busca, G. Phys. Chem. Chem. Phys. 1999, 1, 723. (23) Bataille, F.; Lemberton, J.-L.; Michaud, P.; Pérot, G.; Vrinat, M.; Lemaire, M.; Schulz, E.; Breysse, M.; Kasztelan, S. J. Catal. 2000, 191, 409. (24) Gialella, R. M.; Andrews, J. W.; Cosyns, J.; Heinrich, G. Proceedings of the NPRA Annual Meeting; New Orleans, LA, March 22− 24, 1992. (25) Azizi, N.; Ali, S. A.; Alhooshani, K.; Kim, T.; Lee, Y.; Park, J.; Miyawaki, J.; Yoon, S.-H.; Mochida, I. Fuel Process. Technol. 2013, 109, 172. (26) Farag, H.; Whitehurst, D. D.; Sakanishi, K.; Mochida, I. Catal. Today 1999, 50, 9.

G

DOI: 10.1021/acs.energyfuels.6b01441 Energy Fuels XXXX, XXX, XXX−XXX