Catalyst Deactivation during Upgrade of Light Catalytic Cracking Gas

Mar 27, 2008 - To whom correspondence should be addressed. Telephone: (979) 218-1903 . E-mail: [email protected]. Cite this:Energy Fuels 22, ...
0 downloads 0 Views 749KB Size
Energy & Fuels 2008, 22, 1509–1518

1509

Catalyst Deactivation during Upgrade of Light Catalytic Cracking Gas Oil to Ultralow-Sulfur and Low-Aromatic Diesel Roberto R. Galiasso Tailleur* Simon BoliVar UniVersity, Department of Thermodynamics Sartenejal, Baruta, Miranda, Venezuela ReceiVed NoVember 28, 2007. ReVised Manuscript ReceiVed February 12, 2008

Diesel fuels containing 50 and 15 ppm sulfur were produced in a pilot plant by upgrading light catalytic cracking gas oil (LCO) during 10 weeks of continuous operation. At the end of the run, the catalysts were characterized before and after soluble coke extraction by CH2Cl2. The cokes were characterized by 13C NMR, TPO, GC-MS, and elemental analysis. Catalyst surfaces were characterized by XPS, CO adsorption, pyridine adsorption, and chemical reactions. The results indicate important differences in the amount and composition of soluble coke recovered from the two deactivated catalysts. In the two cases, the soluble coke affected the accessibility of catalytic active sites in different ways. Catalyst deactivation was higher, and the rate of ring opening was lower, under the more severe hydrotreatment conditions needed to produce fuel with 15 ppm sulfur, compared to the conditions required for production of fuel with 50 ppm sulfur.

1. Introduction A considerable number of recent papers have examined the benefits of the new hydrodesulfurization (HDS) catalysts and the potential modifications of the processing units to achieve the target of 15 ppm sulfur.1–4 However, only limited attention has been devoted to achieving low aromatics content in the new catalyst design and to determining the amounts of cracked diesel components that should be incorporated into the pool. The newest generation of HDS catalysts shows an activity more than 3-fold higher than that of the previous generation. Still, the newgeneration catalysts require about twice the amount of active catalyst to produce premium diesel fuel with reduced density and polyaromatics content and improved cetane number. The key process in polyaromatic hydrogenation is the ring-opening reaction, which improves the cetane number. The upgrading of light catalytic cracking gas oil (LCO) and straight run gas oil (SRGO) fractions to produce a low-sulfur, high-cetane diesel is a growing area of interest. The chemistry and associated catalysis for deep desulfurization and deep dearomatization (hydrogenation) were recently reviewed by Song and Ma,5 who noticed that polyaromatics and nitrogen compounds inhibit sulfur removal when the desired product is fuel with 15 ppm sulfur. The addition of LCO to SRGO increases the aromatic content and substantially reduces the apparent reaction rates.4 Among other studies, ref 4 discusses the effects of LCO on activity, selectivity, and stability of the catalyst. In hydroprocessing of petroleum products, catalyst deactivation by coke deposition is one of the major concerns for the * To whom correspondence should be addressed. Telephone: (979) 2181903. E-mail: [email protected]. (1) Topsøe, H.; Massoth, F. E.; Clausen, B. S. Hydrotreating Catalysis. In Catalysis-Science and Technology; Anderson, J., Boudart, M., Eds.; Springer: Berlin, Germany, 1996; p 11. (2) Murali Dhar, G.; Shrinivas, B. N.; Rana, M. S.; Manoj, K.; Maity, S. K. Catal. Today 2003, 86, 45–60. (3) Bej Sh, K. Fuel Process. Technol. 2004, 1503–1517. (4) Ancheyta-Juarez, J.; Aguilar Rodriguez, E.; Salazar Sotelo, D.; Betancourt River, G.; Leiva-Nuncio, M. Appl. Catal., A 1999, 180, 95– 205. (5) Song, C. H.; Ma, X. Appl. Catal., B 2003, 41 (1–2), 207–238.

petroleum and petrochemical industries, from both economic and technological points of view.6–10 Coke deposition occurs in the pores and on the surface of the catalysts and always leads to the loss of activity and product selectivity. Thus, investigators have first looked to develop coking-resistant catalysts as a way to avoid deactivation. To obtain the information necessary to optimize these catalysts and the design processes around them, a great deal of effort has been devoted to studying the chemistry of coke formation with emphasis on the nature and composition of coke. However, many details are missing from our understanding of how coke forms, largely as a result of the challenges associated with the comprehensive structural characterization of insoluble organic matter present in relatively low concentrations. Significant advances have come with new analytical techniques applied to coke characterization (for example, see Song and Ma,5 among others). In spite of these advances, the complex nature of the active sites and the complexity of the coke itself have prevented a thorough understanding of how coke affects the activity of hydrotreatment catalysts. In previous work examining commercial operation of the hydrotreating units that process LCO, changes in the relative proportions of the hydrogenation, hydrogenolysis, isomerization, and cracking reactions were observed to occur in the stream11 over time. The temperature was increased along the cycle to maintain a constant level of sulfur in the product. Both the metal and acid sites were deactivated at different rates, but the deactivation of the acid site function had more impact on the product quality than the metal sites did. The deactivation of the former sites is responsible for the naphthenic ring opening, the paraffin cracking, and the dealkylation reactions that (6) Vogelaar, B. M.; Steiner, P.; Dick van Langeveld, A.; Eijsbouts, S.; Moulijn, J. A. Appl. Catal., A 2003, 251 (1), 85–92. (7) Muegge, B. D.; Massoth, F. E. Fuel Process. Technol. 1991, 29 (1– 2), 19–30. (8) Ramaswamy, A. V.; Sharma, L. D.; Singh, A.; Singhal, M. L.; Sivasanker, S. Appl. Catal. 1985, 13 (2), 311–319. (9) Koizumi, N.; Urabe, Y.; Inamura, K.; Itoh, T.; Yamada, M. Catal. Today 2005, 106 (1–4), 211–218. (10) Weisman, J. G.; Edwards, J. C. Appl. Catal., A 1996, 142, 289– 314. (11) Galiasso Tailleur, R. Catal. Today, in press.

10.1021/ef700713v CCC: $40.75  2008 American Chemical Society Published on Web 03/27/2008

1510 Energy & Fuels, Vol. 22, No. 3, 2008

produced an increase in aromatics content and a reduction in cetane number as a function of time on stream. This behavior was confirmed in the pilot plant test where pure LCO was tested in a three-month cycle by using a WNiPd/TiO2Al2O3 catalyst.12 The present work focuses on understanding the effect of hydrotreatment severity on upgrading LCO to an ultralow-sulfur diesel, focusing on coke deposition on the catalyst and on catalytic selectivity in converting polyaromatics into high-cetane compounds. This study also aimed to identify the effect of coke on acid sites present on a new generation of WNiPd/TiO2Al2O3 catalyst. 2. Experimental Section 2.1. Catalyst Preparation and Characterization. The TiO2Al2O3 support was prepared by controlled coprecipitation of aluminum and titanium hydroxides, and it was calcinated in air. Small pellets of 0.001 m in diameter and length were prepared by extrusion, and the additional metals were incorporated by using two steps of incipient wetness impregnation. First, an aqueous solution of tungsten salt was added; after drying, the solids were exposed to an aqueous solution of nickel and palladium salt. After impregnation, the solid was dried in air and calcinated at 550 °C. Other details of catalyst preparation and characterization are reported by Galiasso Tailleur and Ravigli Nascar.13 Two identical samples of the catalyst were installed in a fixedbed microreactor and presulfided in situ. One of the samples was used to process 100% of the LCO at standard HDS conditions during 10 weeks on stream to produce 50 ppm of low-sulfur diesel. The other sample was used to process 100% of the LCO to produce 15 ppm of ultralow-sulfur diesel during 10 weeks of operation. The shutdown of the microplant and the subsequent handling of the catalyst were carried out by using procedures applied successfully to other catalysts. The methodology eliminates the adsorbed hydrocarbons in the spent samples, maintains the metals in their reduced state, and preserves the hydrocarbon content of the coke. The catalysts were transferred from the microreactor to the laboratory under an inert atmosphere, washed in xylene for 24 h, and dried in nitrogen to a constant weight. The dried samples were stored under an inert atmosphere in a vacuum chamber to avoid catalyst reoxidation. The solids were analyzed by using the following techniques. Physical Method. Surface, pore volume, and average pore diameter were measured by using standard nitrogen adsorption (Micromeritics 250) and mercury porosimetry methods. CH2Cl2 Extraction and GC-MS Analysis. Samples of deactivated catalyst were washed with xylene, packed as a powder into a column, and then extracted with dichloride methane. The eluted material was concentrated by evaporation and then injected into a GC-MS system comprised of an HP 5890 Series II gas chromatograph with an HP-5 column (internal diameter 0.28 mm and length 20 m with inner surface coated by methylsiloxane) and an HP 5972 mass spectrometer. A 1 µL sample was injected twice for every catalyst sample. The whole spectrum was used as a fingerprint to characterize “soluble” coke. In the MS signal, 70% of the compounds in the sample could be identified by using pure molecules and the mass spectrometry library (National Institute of Science and Technology, USA). The elemental analysis of the insoluble coke remaining on the surface was performed by dissolving the inorganic matrix with fluorhydric acid, washing the remaining solid with water and ether, drying in nitrogen, and analyzing the carbonaceous solid. Chemical Characterization. The concentrations of C, H, N, and S were determined by a LECO (D5373) carbon analyzer and by XRF (D1757). In all cases, the amounts of the elements are reported as weight percent. (12) Galiasso Tailleur, R. WNiPd/TiO2Al2O3 catalyst deactivation during the upgrading of LCO. Fuel, submitted. (13) Galiasso Tailleur, R.; Ravigli Nascar, J. Appl. Catal., A 2005, 282 (12), 227–235.

Tailleur Solid 13C NMR Spectra. The analysis of the carbon deposited on solids was performed by using a Varian spectrometer operating at a frequency of 50.576 MHz, with cross-polarization using adamantine as a reference. The technique allowed us to detect the amount of aromatics and paraffinic carbon present on the surface. The signals were recorded, deconvoluted, and integrated, and their relative amount was calculated for further analysis. Temperature-Programmed Oxidation (TPO). The samples of catalyst were subject to TPO by using a PerkinElmer TGA 8 system with a gas mixture containing air (35 mL/min) in helium heated from 30 to 600 °C at a linear rate of 10 °C/min. The CO2 and SO2 produced were recorded as a function of time. Pyridine Thermal Programmed Desorption (TPD). Pyridine TPD analysis was used to characterize the total acidity strength of the deactivated catalyst, and FTIR analysis was used to determine the Brönsted and Lewis content. A McBain microbalance was used to measure out 1 mg samples (thin layer) and transfer them into IR cells specifically designed for this purpose. The catalysts were pretreated with argon (3 L/min) and heated from 30 to 200 or 300 °C at 10 °C/min until a constant weight was achieved; weight losses were recorded. Once a constant weight was reached, the temperature was kept constant at 200 or 300 °C for 2 h and then cooled at a rate of 10 °C/min. A stream containing pyridine (0.001 M) diluted in argon was passed through the catalyst at room temperature (30 °C) for 2 h. After that, pure argon was flowed at the same temperature for an additional 2 h. The catalyst sample was then analyzed by FTIR spectroscopy (main bands: Lewis at 1445 cm-1, Brönsted at 1540 cm-1). Adsorption intensities were calibrated by using the internal standard and compared with those of fresh catalyst. CO Adsorption. Pd dispersion was measured by using the same IR equipment used for pyridine adsorption. CO reactant (Linde) was at 99.99% purity. The solids were compressed in a thin-layer film and dried in argon at 350 K and 1000 Pa for 3 h. Then, the cell was evacuated to 13 Pa at room temperature before the FTIR measurements were started. The metallic function was characterized by stepwise adsorption of small doses of carbon monoxide at room temperature until saturation was achieved at 990 Pa (200 scans with 2 cm-1 spectral resolution). XPS. A Bruker 300 apparatus (Al cathode) using Al KR, 1486.6 eV radiation, and 200 W of power was used. A Shirley-type integral background was used to measure peak areas.14 When multiple components were present under a given XPS envelope, a nonlinear least-squares-curve-fitting routine was implemented (Levenberg– Marquardt damping method). All peaks were fitted using a Voigth function with 20% Lorentzian character. Curve fitting of the W4f region (42–32 eV) was carried out according to the methodology described previously.13 The XPS parameters (peak position, peak width) relative to W4+ were obtained by curve fitting the spectra of the oxide samples. These parameters were kept constant when fitting the spectra of the sulfide specimens. The parameters for the “sulfide” species were obtained by curve fitting the sample sulfide at 375 °C and allowing the peak position and the full-width halfmaximum (FWHM) to relax to their local minima. The results are reported as exposed IW/(ITi + IAl) and ITi/(ITi + IAl). The dispersion was analyzed by using different angles of beam incidence on the sample to check the coke attenuation of the signals. 2.2. Feed and Product Characterization. Analysis of the Feedstock and the HDT Products. The LCO was characterized by using a programmed-temperature GC technique coupled with a lowintensity mass spectrometry method. In parallel, preparative HPLC was also performed by using a hexane carrier and an ultraviolet detector. The fractions eluted by toluene and methanol were analyzed by using a LECO elemental analyzer, 1H NMR spectroscopy (Varian 400 MHz spectrometer, 10 mm broad-band probe), the alumina percolation method (using tetrahydrofuran as solvent), and a high-resolution mass spectrometer (HRMS-Philips 2341). In (14) Practical Surface Analysis by Auger and X-ray Spectroscopy; Briggs, D., Seah, M. P., Eds.; John Wiley & Sons, Ltd.: Chichester, U.K., 1987; pp 511–32.

Catalyst DeactiVation during Upgrade of LCO this way, it was possible to analyze families of aromatics, naphthenes, and paraffins. The cetane number was determined by using the D613 ASTM method. Microplant Long Run Tests. The microplant had a conventional, isothermal down-flow fixed-bed reactor11 containing 25 cm3 of catalyst diluted at 50% with inert material (glass, particle diameter 0.1 cm). Two parallel tests were performed. Both took place with the temperature varying between 350 and 378 °C to maintain constant sulfur content, while the other operating variables (molar ratio of H2/HC ) 4, total pressure ) 10.5 MPa) were kept constant during the 10 weeks of operation. In one of the tests, a 0.2 h-1 space velocity was used to produce a 50 ppm sulfur product, while a 0.1 h-1 space velocity was used in the other to generate a 15 ppm sulfur product. The WNiPd/TiO2Al2O3 catalyst was loaded in the reactor diluted with 50% of inert material. The catalyst was preactivated in situ with straight-run light gas oil containing 1% CS2 at 300 °C for 6 h. Then, the microreactor was loaded with LCO and operated continuously for 10 weeks on stream under the operating conditions described above, generating two spent catalyst samples at the end of the cycle. These catalysts were called Spent 1 (used to produce 50 ppm sulfur fuel) and Spent 2 (used to produce 15 ppm sulfur fuel). During the tests, duplicate samples of hydrotreated LCO were taken daily, and mass balance was performed to check the operational stability of the unit. Once the period on stream was ended, the catalysts were cooled to 200 °C, washed with xylene for 12 h, and then dried with nitrogen at 200 °C for another 12 h. The catalysts were ready to be tested by using the synthetic feed. Test with Synthetic Feed (Probe Molecules). The spent catalysts were tested by using a blend of 30% of methylnaphthalene, 10% of methyltetralin, and 1.5 wt % of sulfur as dibenzothiophene in hexadecane (Feed 2). The temperature needed to obtain a product with 15 ppm sulfur was 340 °C on fresh catalyst, 372 °C on Spent 1, and 380 °C on Spent 2; the other operating conditions were 0.5 h-1 of LHSV, H2/HC molar ratio of 4, and 10.5 MPa of pressure. Other space velocities were also explored for Spent 1 catalyst at 372 °C and for Spent 2 catalyst at 380 °C. Finally, the catalysts were washed by continuous pumping of xylene at 200 °C for 24 h, dried in nitrogen for another half-day (120 °C), and stored in sealed ampules in a dried-inert chamber. The two samples of spent catalyst were then characterized.

3. Results and Discussion The main objective in characterizing the spent catalysts was to study the effect of LCO hydroprocessing on catalyst deactivation. The following discussion will first consider the nature and composition of coke obtained during the processing of LCO at two HDS levels for 10 weeks. Then, it will focus on how catalyst modification affected the different reactions occurring on the catalyst. The spent catalysts were characterized by a combination of different analytical techniques to obtain the composition of coke deposits and the characteristics of acid sites. The carbon content in the spent catalysts was ∼6.4 wt %, and the H/C ratio (wt) was 1.09 in Spent 1 and 1.08 in Spent 2. The hydrotreating temperature was adjusted to keep sulfur removal constant and in the range of 99.66% (50 ppm) or 99.9% (15 ppm) during the 10 weeks in operation. Keeping this level constant meant using a higher temperature (by 10 °C) at the beginning of the cycle and a higher temperature (by 18 °C) at the end of cycle in order to upgrade the LCO to low-sulfur (50 ppm) or to ultralow-sulfur (15 ppm), with other operating variables held constant. The two processes of upgrading LCO showed differences in paraffin, naphthene, and aromatics content, as well as in the nature of the coke deposition, which caused catalyst deactivation.12 To optimize the level of conversion and the catalyst composition for future commercial operations, it is imperative to know the impact of the coke

Energy & Fuels, Vol. 22, No. 3, 2008 1511 Table 1. Physical and Chemical Properties of Catalysts surface (m2/g) micropore volume (cm3/g) macropore volume (cm3/g) total sulfur (wt %) total carbon (wt %) soluble carbon (wt %) C/H/S soluble coke (wt %) C/H/S insoluble coke (wt %)

fresh

Spent 1

Spent2

245 0.17 0.35 5.6

112 0.08 0.29 6.1 6.2 0.99 91.0/7.66/1.34 91.5/7.13/1.24

103 0.075 0.27 6.3 6.4 1.11 91.2/7.39/1.41 91.7/6.89/1.31

Table 2. Polyaromatic Composition in Soluble Coke compound naphthalene fluorene phenanthrene anthracene fluoranthene triphenilene pyrene chrysene benzo(a)anthracene benzo(g,h,i)fluoranthene benzo(k)fluoranthene perylene benzo(a)perylene dibenzo(a,h) anthracene other

Spent 1 (wt %)

Spent 2 (wt %)

1.5 13.5 2.2 9.1 1.2 22 0.5 18 3.5 1.5 0.3

0.5 12 3.1 7.9 2.4 23 1.6 16 5.1 0.5 0.1 0.1

2.4 24.6

2.0 26.3

deposits on the active sites. Thus, we began by characterizing the catalyst and the coke. 3.1. Catalyst Composition and Physical Properties. The fresh catalyst before sulfiding contained 15% of WO3, 5.3% of NiO, and 0.2% by weight of PdO oxide species; these were reduced and sulfided in situ to generate the active metal phases. The support contained 10 wt % of TiO2 in γ-alumina. This support was treated with steam-ammonia to induce the Ti to migrate into the aluminum on the surface, thereby building a particular distribution of acid sites, as confirmed by analyses using XPS, FTIR, and 29Al NMR, as well as other data discussed previously.12 The characteristics of the fresh catalyst are included in Table 1 for reference, but the following discussion will focus on the spent catalyst that had accumulated coke during the 10 weeks on stream. Both of the spent catalysts presented completely different surface properties than the fresh catalyst. During the first week in operation with 100% LCO to produce low-sulfur diesel (not included here, see details in ref 13), the catalyst lost around 41% of its surface area and 55% of its micropores. Spent 1 had ∼5% higher micropore volume and surface area than Spent 2, a difference that is insignificant. Similarly, the total coke content (∼5%) differed only slightly between the two samples, as did the total sulfur content (∼4%) and the total H/C ratio (3%). The macropore volume and the amount of insoluble coke were approximately the same in both samples. The main difference between both samples was in the soluble coke content, which was 18% higher in Spent 2 than in Spent 1, and this amount represented around 15% of the total coke content. Soluble Coke Composition. Table 2 reports the concentration of the major polyaromatic hydrocarbons identified by mass spectrometry in the soluble coke extracted by CH2Cl2 from the Spent 1 and Spent 2 catalysts. The identification of these polyaromatic hydrocarbons was based on determining the retention times for various model polyaromatic hydrocarbons used as references. The main compounds observed in soluble coke were fluorene, anthracene, triphenilene, chrysene, and benzo(a)anthracene, which together

1512 Energy & Fuels, Vol. 22, No. 3, 2008

Tailleur

–135.0 ppm, together with two other small bands at 125 and 152 ppm due to aromatic coke. Other small bands between 30 and –50 ppm were assigned to aliphatic coke.18,19 The ratios of intensities between these deconvoluted peaks are Spent 1

Spent 1 Figure 1. 13C NMR analysis of coke before (a and c) and after (b and d) extraction.

Spent 2

Spent 2

Figure 2. TPO-CO2 from Spent 1 and Spent 2 catalysts before extraction (a and c) and after extraction (b and d), 10 °C/min.

accounted for more than 50% of the total hydrocarbons. Onequarter of the hydrocarbons could not be identified due to unavailability of suitable reference compounds. In Spent 2, similar types of compounds to those found in Spent 1 soluble coke were present, albeit in different proportions. The envelope of MS peaks can be used as a fingerprint to characterize the soluble coke. Both extracts contain a low H/C ratio, as well as a small amount of sulfur compounds. The MS envelop of soluble coke obtained from both spent catalysts differed from the spectra obtained on a commercially available hydrotreating catalyst of the NiMo type.12 Sahoo et al.15,16 reported differences in the soluble coke obtained from the HDS of residue catalyst (NiMo type), and they attributed the differences to the composition of the feed. In our study, the feed is the same, and the difference between Spent 1 and 2 soluble cokes can be attributed solely to conditions used to achieve the sulfur target in the product. Similar effects of process conditions on coke were reported by Martín et al.17 using a reforming-type catalyst. 13C NMR Analysis. Two deactivated samples before (BE) and after (AE) extraction with CH2Cl2 were analyzed by CP/MAS. The majority of the coke was found to be insoluble and located in clusters on the solids (TEM analysis not shown). The coke clusters were composed mainly of condensed polyaromatics that appear to surround the metallic areas based on microdiffraction analysis. 13C NMR was used to evaluate the degree of aromaticity in both soluble and insoluble cokes. Figure 1a and c shows that both samples contain an intense aromatic band at (15) Sahoo, S. K.; Ray, S. S.; Singh, I. D. Appl. Catal., A 2004, 278, 83–91. (16) Sahoo, S. K.; Rao, P. V. C.; Rajeshwer, D.; Krishnamurthy, K. R.; Singh, I. D. Appl. Catal., A 2003, 244, 311–421. (17) Martín, N.; Viniegra, M.; Zarate, R.; Espinosa, G.; Batina, N. Catal. Today 2005, 107-108, 719–725. (18) Callejas, M. A.; Martinez, M. T.; Blasco, T.; Sastre, W. Appl. Catal. 2001, 218, 181–188.

I152 I125 ) 0.45; ) 0.15; I135 I135 I30 ) 0.08 before extraction (BE) (1) I135 I125 I152 ) 0.67; ) 1.32; I135 I135 I30 ) 0.03 after extraction (AE) I135 I152 I125 ) 0.41; ) 0.15; I135 I135 I30 ) 0.08 before extraction (BE) I135 I152 I125 ) 0.71; ) 1.21; I135 I135 I30 ) 0.02 after extraction (AE) I135

Before extraction, Spent 2 showed higher aromaticity (152/ 135 bands ratio) and broader distribution than Spent 1. In the two spent catalysts, the aromatic bands are quite broad due to the heterogeneity of aromatic species in the coke and to the different extents of alkyl branching. Efforts were made to evaluate aromaticity by using cross-polarization, according to parameters published elsewhere.16 The time-dependent evolution of carbon magnetization during polarization transfer was higher for Spent 2 before extraction, indicating that it had more complex bonding configurations and motional constraints than Spent 1 before extraction. After extraction, both samples gave spectra characteristic of aged coke, as previously observed in spent catalysts used in other hydroprocesses. This is consistent with the fact that both samples experienced similar operating conditions during the initial deposition of coke, which was when most of the insoluble coke was produced. The calculation of aromaticity alone does not give enough information to describe the average coke molecule; more quantitative information on the different types of protonated and nonprotonated or “bridgehead” aromatic carbons is needed. Experiments are underway to measure the difference in magnitude of the dipolar coupling between aromatic carbons, which are attached to protons, and quaternary carbons. Preliminary results indicate that coke in Spent 1 was more protonated than that in Spent 2, in agreement with the H/C ratio. In summary, the two spent catalyst samples showed important differences in the amount and composition of soluble coke and only a slight difference in insoluble coke. The following discussion analyzes how these differences in coke affect the oxidation behavior of the catalyst. TPO of Soluble and Insoluble Coke. Thermal oxidation analysis of the spent catalysts, before and after extraction, is shown in Figure 2. The amount of CO2 released by the sample as a function of temperature gives information about the nature of the coke. Barbier et al.20 and Parera et al.,21 among others, recognized three different temperature regimes for reforming (19) Bonardet, J. L.; Barrage, M. C.; Fraissard, J. J. Mol. Catal. A: Chem. 1995, 96, 123–143.

Catalyst DeactiVation during Upgrade of LCO

Figure 3. TPO-SO2 for (a) fresh catalyst, (b) Spent 1 catalyst, (c) Spent 2 catalyst, and (d) insoluble coke from Spent 1.

catalysts, and these regimes are used extensively in the literature for characterizing coke: (1) 150-250 °C oxidation of high-inhydrogen coke present on metals, (2) 300-400 °C oxidation of coke located on the support near the metal, which contains some hydrogen, and (3) 450-520 °C oxidation of highly aromatic coke deposited on the support. We used all three regimes, and in parallel, we measured the amount of SO2 released by the samples and the amount of hydrogen, which oxidized quite quickly into water at low temperature. The ratios of the integrated area of the three CO2 peaks are Spent 1

I250 ) 0.17, I360

I470 ) 0.12 (BE); I360 I470 I250 ) 0.09, ) 0.19 (AE) I360 I360

Spent 2

I250 ) 0.16, I360

I470 ) 0.10 (BE); I360 I470 I250 ) 0.10, ) 0.21 (AE) I360 I360

The ratios of the area (I) of the peaks of CO2 released at different temperatures by the samples indicate that ∼50% of the carbon was burned at 360 °C in both samples, but the amount of coke that combusted early in the process at 250 °C was lower for the Spent 1 catalyst than for the Spent 2 catalyst. This difference may be associated with the soluble carbon deposited on the metal. After extraction, the amount of coke burned at 250 and 360 °C was reduced in both samples. In contrast, the quantity of graphitized coke that was oxidized at 470 °C was quite similar in both catalysts, before and after extraction. This information, together with previous 13C NMR analysis, indicates that graphitized coke was formed on the support during the first week of operation and subsequently “aged” as a function of time on stream. The main difference in the coke between the samples was due to the metals present. The catalyst in this study contained Pd, which seemed to help the combustion during the TPO, because the carbon combustion for all the peaks started at temperatures lower than those reported in the literature for deactivated NiMo/Al2O3 catalyst.22 SO2 was formed by combustion of sulfur species at low temperature between 230 and 260 °C (Figure 3b and c); no further SO2 was produced at 330 °C. Most of the sulfur species (90%) came from the sulfide linked to the metals, and the TPO provided little information about the nature of the sulfur in the coke itself. The combustion of the fresh catalyst without coke (20) Barbier, J. In Studies in Surface Science and Catalyst, Catalyst DeactiVation; Delmon, B., Froment, G., Eds.; Elsevier: Amsterdam, The Netherlands, 1987; p 1. (21) Parera, J. M.; Figoli, N.; Traffano, E. J. Catal. 1983, 79, 481–85. (22) van Doorn, J.; Barbolin, H.; Moujlin, J. A. Ind. Eng. Chem. Res. 1992, 31, 101–107.

Energy & Fuels, Vol. 22, No. 3, 2008 1513

showed a different pattern (Figure 3a) from that of the Spent 1 catalyst (Figure 3b), with one peak at 200 °C with a long tail and a second peak at 350 °C with respect to the Spent 1 catalyst. The difference between the patterns is due to the fact that oxygen has greater access to the metals containing sulfur on the fresh catalyst. Spent 1 and Spent 2 showed similar rates of sulfur combustion (Figure 2b vs c). The spectra do not provide further information. The combustion of insoluble coke, obtained after dissolving the support in HF, presented a very different pattern than those observed with all coke. Combustion started at 200 °C and showed two peaks. The rate of coke combustion may have been limited by how accessible the internal layers were to oxygen, since insoluble coke is a nonporous solid. The insoluble coke did not contain metal sulfides (