Bayraktar, 2002 - Characterization Of Coke On Equilibrium Fluid Catalytic Cracking Catalysts By Temperature-programmed Oxidation

Applied Catalysis A: General 233 (2002) 197–213

Characterization of coke on equilibrium fluid catalytic cracking catalysts by temperature-programmed oxidation Oguz Bayraktar, Edwin L. Kugler∗ Department of Chemical Engineering, CEMR, P.O. Box 6102, West Virginia University, Morgantown, WV 26506-6102, USA Received 17 August 2001; received in revised form 27 February 2002; accepted 6 March 2002

Abstract Characterization of coke on equilibrium, fluid catalytic cracking (FCC) catalysts contaminated with metals is investigated using temperature-programmed oxidation (TPO) and temperature-programmed hydrogenation (TPH). TPO spectra of spent equilibrium catalysts from cracking of sour imported heavy gas oil (SIHGO) and ASTM standard-gas–oil feed were deconvoluted into four peaks by fitting them into Gaussian-type functions. The four peaks are assigned to different types of coke on the catalyst. The first peak is produced by hydrocarbons desorbing from the coke. Traditionally, this is called cat-to-oil coke. The second peak is contaminant coke produced by contaminant-metal reactions. The third peak is conversion coke produced by acid-catalyzed reactions. A graphite-like coke that is related to both feedstock properties and catalyst activity produces the last peak. The TPO spectra of spent catalysts from cracking n-hexadecane are deconvoluted into three peaks, corresponding to the first three peaks observed with gas–oil cracking. The graphite-like coke is not observed after n-hexadecane cracking. TPO peak area is proportional to the amount of coke on catalyst. The amount of contaminant coke correlates with contaminant-metal concentration. The sum of the conversion coke and the graphitic coke correlates with catalyst activity. This sum can be used to characterize the coking tendency of a feedstock. Feedstock comparisons at 50% conversion show that SIHGO feed produces twice as much coke as ASTM feed and nearly five times as much coke as n-hexadecane. TPH results are less useful in characterizing coked catalysts. The peaks in TPH spectra can be correlated with only the first three peaks of TPO spectra. The high temperature peak assigned to graphitic coke in TPO is not observed in TPH spectra. This situation occurs because graphitic coke is more reactive with oxygen than with hydrogen, so that oxidation occurs within the temperature range of experimental equipment, while graphitic coke hydrogenation occurs at higher temperatures, beyond the range of the TPH apparatus. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Temperature-programmed oxidation (TPO); FCC; Catalysts; Nickel; Vanadium; Coke characterization; Coke measurement

1. Introduction Temperature-programmed techniques have found many applications for characterizing coke on spent catalysts. Coke on several catalyst systems has been ∗ Corresponding author. Tel.: +1-304-293-2111x2414; fax: +1-304-293-4139. E-mail address: [email protected] (E.L. Kugler).

characterized with respect to its reactivity towards hydrogen (temperature-programmed hydrogenation (TPH)) [1–10] and oxygen (temperature-programmed oxidation (TPO)) [1,3–5,7–14]. Carbon dioxide has been tried, also [5]. The location and nature of coke on spent catalyst affect coke-oxidation kinetics, which are important factors in catalyst regeneration [15,16]. TPO is a very important tool for characterizing spent catalysts. TPO can determine the total amount of

0926-860X/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 0 2 ) 0 0 1 4 2 - 4

198

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

coke, the hydrogen-to-carbon ratio in the coke, and the location of coke deposited on the spent catalyst. Brown and coworkers [17–19] published detailed TPO studies of coked fluid catalytic cracking (FCC) catalysts. They examined catalyst coked with 1-hexene, cyclohexane and 1-octene as well as FCC catalyst coked in commercial operation. They measured CO, CO2 and H2 O products and modeled TPO results with a single-site reaction mechanism. Experimental data fit the model well at higher temperatures. However, lower temperature data with catalyst from commercial operation contained a peak in the residual analysis that did not fit the model. This peak was attributed to either saturated coke (rather than aromatic coke) or contaminant-metal effects. Doolin et al. [15] and Dimitriadis et al. [16] studied how nickel and vanadium influence coke oxidation with FCC catalysts. Doolin studied the effect of metals on CO2 /CO ratio and Dimitriadis studied coke-oxidation kinetics. Neither group ran TPO spectra with metal contaminated FCC catalyst. However, metal effects have been reported in the TPO spectra of coked supported-metal catalysts where coke located on or near metal particles can be distinguished from coke away from metals on the catalyst support [20–23]. The present study, focuses on the influence of contaminant metals on the TPO spectra of spent, equilibrium FCC catalysts. Three different equilibrium catalysts are used with low, medium and high concentrations of nickel + vanadium. Spent catalysts are prepared by cracking hexadecane and two different gas–oil feeds in a microactivity test unit. The TPO spectra are deconvoluted into multiple peaks for correlation with catalyst activity, feed composition and contaminant-metal concentration.

2. Experimental 2.1. Materials Gas mixtures, such as 3% oxygen-in-helium and 8% hydrogen-in-argon were obtained from Matheson and used as received. The calcium oxalate monohydrate (CaC2 O4 ·H2 O, >98% purity) used for calibration of the TPO apparatus was obtained from Fluka and used as received.

Table 1 Nominal properties of equilibrium catalysts from Ashland Inc. Catalyst type

ECat-LOW

BET surface area (m2 /g) Matrix surface area (m2 /g) Zeolite surface area (m2 /g)

178

Ecat-INT 160

Ecat-HIGH 115

63

63

45

115

97

70

Metals (ppm) Nickel Vanadium

300 700

900 1700

2600 6700

Microactivity

71

69

62

2.2. Catalysts and catalyst preparation In this study, three equilibrium FCC catalysts (used catalysts from commercial operation) were supplied by Ashland, Inc. They are named as ECat-LOW, ECat-INT and ECat-HIGH based on their contaminant-metals concentration. Characterization data on these equilibrium catalysts are provided in Table 1. Spent catalysts were prepared by cracking sour imported heavy gas oil (SIHGO, Davison Chemical), n-hexadecane (99.3% purity, Fisher Scientific), and ASTM standard-gas–oil (National Institute for Standards and Technology, RM-8590) in a microactivity test (MAT) unit. Standard test conditions were 5 g of catalyst, 500 ◦ C reactor temperature and 3.0 catalyst/oil ratio. The MAT unit, described in ASTM method D-3907, was manufactured by Industrial Automated Systems (Parlin, NJ). A 1% platinum-alumina (Pt-Al2 O3 ) catalyst was used downstream of the TPO reactor to oxidize CO and any hydrocarbons desorbed from the catalyst. This oxidation catalyst was prepared by impregnating platinum chloride hexahydrate (H2 PtCl6 ·6H2 O, VWR Co.) on aluminum oxide (Al2 O3 12–20 mesh size, VWR Co.). After impregnation, the catalyst was dried in air at 100 ◦ C overnight and calcined at 500 ◦ C for 6 h. Before using, it was activated with 8% hydrogen-in-argon at 500 ◦ C for 6 h. 2.3. Temperature-programmed oxidation The TPO equipment consists of a Hewlett-Packard 5890 gas chromatograph (GC) with a thermal

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

conductivity detector (TCD) and flame-ionization detector (FID) in series, and an external furnace connected to a temperature-program controller (Automated Test Systems, Butler, PA). This apparatus can be used in multiple modes by changing the reactant–gas mixture. In TPO experiments, samples are exposed to a 3% oxygen-in-helium mixture flowing at 40 cm3 /min. The oxygen consumption is measured as a function of time (temperature) in order to obtain the TPO spectrum. In a TPO experiment, a 50 mg spent catalyst sample is placed in a quartz U-tube reactor and surrounded with quartz chips. The TPO sample forms a fixed bed inside the reactor. The reactor is placed in a furnace where temperature is increased linearly from room temperature to 865 ◦ C at a rate of 10 ◦ C/min. On reaching 865 ◦ C, the temperature is held constant for about 15 min. The exit stream from the reactor goes directly to the 1% platinum-alumina oxidation catalyst operating at 595 ◦ C. The platinum-alumina catalyst converts the CO formed and any desorbing hydrocarbons to CO2 and H2 O. The product gas mixture from the platinum oxidation catalyst passes through two cold traps in series in order to trap H2 O (dry-ice/acetone

199

trap) and CO2 (liquid nitrogen trap). Oxygen does not condense in the liquid nitrogen trap because the 3% concentration is an order-on-magnitude below its saturation vapor pressure. Since CO and hydrocarbons are converted to CO2 and both CO2 and H2 O are trapped, only oxygen consumption is measured by the TCD. Oxygen consumption is recorded in the TPO profile as a negative peak (see Fig. 1a). At the end of the temperature programming sequence, the liquid nitrogen trap is removed and solid carbon dioxide is allowed to evaporate into the carrier gas flow. This causes a pressure surge and sudden flow increase through the TCD. To dampen this pressure surge, a 0.5 m GC column (Carboxen 1000, Supelco) operating at 50 ◦ C is located before the TCD. This column holds the CO2 long enough for the flow disturbances to disappear before CO2 is detected. The CO2 produced by oxidizing the coke deposit is measured as a positive peak (see Fig. 1a). Water produced during the TPO experiment is not measured but remains condensed in the dry-ice/acetone trap until the experiment is over. Fig. 1 shows the calibration for the TPO system with calcium oxalate monohydrate. Under helium

Fig. 1. Spectra for calcium oxalate decomposition: (a) with 3% oxygen-in-helium; and (b) in helium without oxygen.

200

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

Fig. 2. TPH spectra for equilibrium FCC catalyst ECat-INT: (a) after cracking SIHGO feed; (b) after cracking n-hexadecane feed; and (c) calcined equilibrium catalyst.

flow, CaC2 O4 ·H2 O was decomposed into H2 O, CO and CO2 by heating linearly from room temperature to 865 ◦ C at a rate of 10 ◦ C/min. A dry-ice/acetone trap was used to retain H2 O. Fig. 2b shows the resultant spectrum. The first peak is a mixture of CO and CO2 , and the second peak is only CO2 . The chemical composition of eluted gases was confirmed qualitatively using a mass spectrometry (Varian Saturn 3 GC–MS) at the exit of the TCD detector. The thermal decomposition of CaC2 O4 ·H2 O was repeated in 3% oxygen-in-helium with the Pt-Al2 O3 oxidation catalyst operating downstream of the TPO reactor. The dry-ice/acetone trap and liquid nitrogen trap were used in series. The H2 O, CO and CO2 products should be produced at the temperatures observed in Fig. 1b. Fig. 1a shows the results. The CO produced

is oxidized to CO2 on the Pt-Al2 O3 oxidation catalyst located at the exit of the reactor. The oxidation of CO produces a negative peak located near 450 ◦ C. The CO2 formed is removed from the oxygen-in-helium mixture with the liquid nitrogen trap. At the end of the TPO experiment, the liquid nitrogen bath is removed from the second trap releasing the CO2 product. This evaporated CO2 produces the positive peak at the end of the run. The peak areas in the TPO spectrum are proportional to the amount of O2 consumed (negative peak) and CO2 formed (positive peak). The calibration constants for O2 and CO2 are obtained by running the TPO experiment with a known amount of CaC2 O4 ·H2 O. The TPO results measuring carbon on spent catalysts are similar to results from a commercial

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

201

Table 2 Comparison of quantitative analysis results from TPO and commercial laboratory Sample number

Sample 1 Sample 2 Sample 3 a

Combustion analysis resultsa

TPO analysis results Carbon (wt.%)

Hydrogen (wt.%)

Carbon (wt.%)

Hydrogen (wt.%)

0.79 1.21 1.07

0.05 0.08 0.09

0.83 1.18 1.09

<0.5 <0.5 <0.5

Galbraith laboratories.

laboratory (Table 2). Hydrogen in the coke was calculated by material balance using the amount of O2 consumed and CO2 formed. 2.4. Temperature-programmed hydrogenation Carbon on FCC catalysts can be characterized according to its reactivity to form methane. The TPH configuration is shown in Fig. 1. In converting from TPO to TPH experiments, the

platinum-alumina oxidation catalyst is bypassed so that the reactive-gas mixture flows directly from the TPH reactor to the dry-ice/acetone trap. An 8% hydrogen-in-argon gas mixture is passed through the system at 35 cm3 /min. The dry-ice/acetone trap is used to condense water. The liquid nitrogen trap is omitted since argon freezes to form a solid in a liquid nitrogen bath. A FID, which is very sensitive to hydrocarbons, detects methane formation.

Fig. 3. Analysis of TPO spectrum for ECat-LOW after cracking SIHGO feed: overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

202

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

Fig. 4. Analysis of TPO spectrum for ECat-INT after cracking SIHGO feed: Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

3. Results and discussion 3.1. Characterization of coke by temperature-programmed hydrogenation The gasification of carbon with hydrogen requires high temperatures especially for the less reactive forms of carbon like graphite [2]. For this reason, all types of carbons may not be observed when using TPH over a reasonable temperature range. On the other hand, reactivity of oxygen and carbon at relatively low temperatures makes TPO an attractive technique for coke characterization. In this work, both TPH and TPO have been used to characterize coke on equilibrium FCC catalysts. Fig. 2 shows the TPH spectra of ECat-INT both in calcined form and as spent catalyst after cracking hydrocarbon feeds in the MAT unit. The signal from the FID comes mostly from methane formed during the hydrogenation of coke. Fig. 2a shows the TPH profile of spent catalyst obtained by cracking SIHGO on

ECat-INT. Four peaks can be detected visually from this curve, without any deconvolution. Most of the coke reacts with hydrogen around 500 ◦ C with a little shoulder around 300 ◦ C. A high temperature peak occurs at 865 ◦ C with a broad front shoulder around 750 ◦ C. The last maximum in the TPH spectrum at 865 ◦ C is caused by the end of the heating program and the beginning of a constant temperature period at the end of the analysis. Fig. 2b shows the TPH profile of the spent catalyst obtained by n-hexadecane cracking, using the same ECat-INT catalyst. In this figure, three smaller but nicely separated peaks are observed near 280, 540 and 865 ◦ C. Fig. 2b indicates that different types of coke can form on the catalyst from only one reactant. As in the previous case, the end of the heating program causes the highest temperature peak. The calcined FCC catalyst does not create a very significant signal with the FID (Fig. 2c) since it should not contain carbon. The TPH data from ECat-INT indicates that at least three types of coke form on spent catalysts. The

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

203

Fig. 5. Analysis of TPO spectrum for ECat-HIGH after cracking SIHGO feed. Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

low temperature peak, near 280 ◦ C with hexadecane, is probably produced by light hydrocarbons released by the coke deposit. Metal-catalyzed hydrogenations are probably not involved, since little hydrogen consumption occurs at this low temperature. Light hydrocarbons may be produced by thermal cracking of hydrocarbon deposits or by the release of absorbed products. Absorbed hydrocarbons in the coke may be reaction products from the MAT reaction, or contaminants from air exposure, or contaminants from sample handling. Heavy hydrocarbons, like hexadecane from the feed, are not expected in the TPH spectrum since hexadecane would be removed in the dry-ice acetone trap located between the reactor and the FID. The peak near 540 ◦ C is probably produced by metal-catalyzed hydrogenation of the coke product. The temperature-programmed reduction (TPR) spectra of calcined equilibrium catalysts show several temperatures where catalyst reduction occurs [24–28].

The TPR spectra reported in the literature show a low temperature reduction around 500 ◦ C, so that some deposited metals should be reduced and active for coke hydrogenation at 540 ◦ C. The TCD spectra of coke hydrogenation (not shown) are complex above 450 ◦ C. The complexity is caused by opposing TCD responses for hydrogen and methane. Hydrogen consumption gives a negative response while methane production gives a positive response. The TCD spectrum above 450 ◦ C is noisy but positive, suggesting that methane is formed. The features of the TCD spectrum above 450 ◦ C track the FID spectrum. The amount of hydrogen consumed in methane production cannot be measured. The high temperature peak in the TPH spectra of coked catalyst is probably caused by coke hydrogenation on non-metal surfaces. Evidence for this assignment is provided by TPO data. Discussion of the high temperature peak from TPH spectra is given in the following sections.

204

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

Fig. 6. Effect of contaminant-metal concentration on (a) peak areas; and (b) peak positions. Data obtained with SIHGO feed. TPO spectra deconvolution shown in Figs. 3–5.

3.2. TPO of coke on equilibrium FCC catalysts TPO spectra were measured for spent equilibrium catalysts used for cracking SIHGO feed, ASTM feed, and n-hexadecane providing results for three different catalysts with different contaminant-metal concentrations. When the contaminant-metal concentration increases, the temperature for maximum oxygen consumption decreases. This can be attributed to the catalytic effect of vanadium and nickel on the coke-oxidation reaction. Decreases in temperature for maximum oxygen consumption have been reported for FCC [15,16,18] and other catalysts [20–22] with metal contaminants. TPO profiles usually contain several peaks. Unfortunately, more often than not, these peaks are not well resolved. Deconvolution is needed in order to isolate and identify individual peaks and assign them to different types of coke.

Power-law kinetic expressions were used by Querini and Fung [29] to deconvolute and interpret TPO data. Using a power-law model, these researchers were able to determine not only the amount of coke but also activation energies and reaction orders. However, to obtain good results, they needed to regress two independent sets of data obtained at different heating rates. Querini and Fung found that the application of power-law kinetics to single-heating-rate data resulted in poor kinetic parameters [29]. In a separate study, Larsson and coworkers [12,30] found that using power-law kinetics to deconvolute TPO data gave poor results when applied to single-heating rate, fixed bed experiments. However, Larsson got a good data fit using Gaussian-type peaks to analyze his TPO data. Since our TPO data were obtained using only one heating rate, it seemed practical to deconvolute TPO spectra using Gaussian-type peaks. The original

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

205

Fig. 7. Analysis of TPO spectrum for ECat-LOW after cracking ASTM feed: Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

negative TPO profiles from oxygen consumption were inverted into positive peaks, since only positive peaks can be fitted with our deconvolution software (PeakFit, Jandel Scientific). Two or more types of coke were assumed to be present on any spent catalyst sample. Normally, three or four peaks were needed to deconvolute TPO spectra. The final analysis used the minimum number of peaks that provided a good fit to the experimental data. Replicate TPO experiments showed that, after deconvolution, the shift in maximum temperature for any specific peak was within ±5 ◦ C, while the change of relative peak area of each peak (contribution of each peak to the total peak area) was within ±2%. Deconvolution results for SIHGO feed are presented in Figs. 3–5. Four peaks are used to fit the data in each figure. The lowest temperature peak has been designated K and the following peaks at increasing temperature have been labeled L, M and N. Data on the peak areas and temperature maximum for all deconvoluted peaks are listed in Table 3.

The areas for the peaks observed with SIHGO feed are plotted in Fig. 6a as a function of nickel+vanadium concentration. The size of peak L increases as the contaminant-metal concentration increases, while peaks K, M and N show no definitive trend. Fig. 6b shows that the temperature maxima for all of the deconvoluted peaks decrease with increasing concentration of contaminant metals. More definitive trends between peak size and contaminant-metal concentration are seen in the results with ASTM feed. The ASTM feed is highly paraffinic, containing lower concentrations of heteroatoms and multi-ring-aromatic molecules than the SIHGO feed. The deconvolution results for ASTM feed are presented in Figs. 7–9. The TPO spectra are fitted to four peaks, again labeled K, L, M, and N. As in the previous case, peak L increases with contaminant-metal concentration. Peaks K, M and N decrease with increasing metal concentration. These trends are plotted in Figs. 10a. The positions of the peak maxima decrease with increasing

206

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

Table 3 Peak areas from TPO spectra shown in Figs. 3-5, 7-9, 11-13 Feed

Catalyst

kb

a

Peak area (temperature, ◦ C)

Total area

Peak K

Peak L

Peak M

Peak N

SIHGO

ECat-LOW ECat-INT ECat-HIGH

1.38 1.33 0.96

3.26 (340) 3.78 (341) 2.79 (2.94)

5.56 (439) 5.94 (434) 10.83 (3.91)

13.00 (530) 13.91 (521) 10.83 (503)

2.24 (613) 2.90 (600) 2.02 (598)

24.06 26.53 26.47

ASTM

ECat-LOW ECat-INT ECat-HIGH

3.00 2.23 1.64

4.98 (349) 2.68 (345) 1.62 (305)

5.37 (445) 3.57 (439) 8.15 (417)

14.10 (537) 10.59 (527) 9.77 (516)

3.09 (621) 1.47 (611) 1.53 (598)

27.54 18.31 21.07

n-C16 H34

ECat-LOW ECat-INT ECat-HIGH

1.34 1.25 1.25

1.48 (333) 1.40 (318) 1.71 (307)

1.05 (442) 1.05 3.37 (425)

3.17 (551) 3.17 (551) 3.32 (539)

0.00 0.00 0.00

a b

5.70 5.69 8.40

The values in parenthesis is the temperature of each peak. Kinetic conversion, k = % conversion/(100 − conversion).

contaminant-metal concentration (Fig. 10b) as is observed with the other cracking feeds. The deconvoluted TPO spectra of spent catalysts from n-hexadecane cracking are shown in Figs. 11–13. In this case, the spectra are deconvoluted into three

peaks labeled K, L and M. Peak N is absent from coke produced from n-hexadecane feed. Fig. 14a shows that the area of peak L increases with increasing contaminant-metal concentration, as is observed with the other feedstocks. Peaks K and N remain nearly

Fig. 8. Analysis of TPO spectrum for ECat-INT after cracking ASTM feed. Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

207

Fig. 9. Analysis of TPO spectrum for ECat-HIGH after cracking ASTM feed. Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

constant, showing no particular trend with metal concentration. The positions of the peak maxima shift to lower temperatures with increasing metal concentration (Fig. 14b), as is observed with other feedstocks. Habib et al. [31] categorized the coke formed during commercial FCC processing into four groups. These are catalytic coke, which is formed during acid-catalyzed cracking; contaminant coke, which is caused by the catalytic dehydrogenation activity of metals such as nickel and vanadium; Conradson coke which correlates directly with feedstock basic nitrogen concentration, average molecular weight, and Conradson carbon residue; and cat-to-oil coke, which are hydrocarbons sorbed on spent catalyst as a result of non-ideal stripping in the commercial process. Peak K in the TPO spectra is related to desorbing hydrocarbons that are oxidized on the platinumalumina catalyst immediately downstream from the TPO reactor. The platinum-oxidation catalyst consumes oxygen to generate the TPO peak. These desorbing hydrocarbons are seen by the flame-ionization

detector as the low temperature peak in the TPH spectra of coked catalysts (see Fig. 2). Coked catalysts also produce a low temperature peak from a flame-ionization detector when helium is used as a carrier gas, showing that hydrocarbons coming from the coke are responsible for the TPO peak. In the terminology used by Habib and coworkers, this is cat-to-oil coke. A microactivity test unit uses a downflow, fixed bed reactor to evaluate FCC catalyst. The feed is vaporized in a preheater and on quartz wool at the top of the catalyst bed. After the oil is reacted, the catalyst is flushed with nitrogen for 30 min. Nitrogen strips hydrocarbons from the catalyst bed at 500 ◦ C. However, hydrocarbon stripping in the MAT unit may not be 100% effective, perhaps leaving some hydrocarbons to desorb from spent catalyst. Handling spent catalysts in air provides another source for desorbing hydrocarbons. Coke on spent catalyst may sorb hydrocarbons from the laboratory atmosphere, desorbing them during a TPO or TPH experiment. In addition, coke on

208

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

Fig. 10. Effect of contaminant-metal concentration on: (a) peak areas; and (b) peak positions. Data obtained with ASTM feed. TPO spectra deconvolution shown in Figs. 7–9.

spent catalysts may be oxidized when exposed to air. Oxidized hydrocarbon residues may react or decompose on heating to release hydrocarbons that produce peak K in the TPO spectra. The area of peak L correlates with contaminantmetal concentration. Peak L must be produced by metal-catalyzed oxidation of coke residues on the FCC catalyst. This does not mean that all of the peak area represents coke produced by metal contaminants, only that the peak area relates to metal-catalyzed coke oxidation. The coke that produces peak L is probably located both on metal contaminants and adjacent to metal contaminants. The coke on the metal must be contaminant coke. The coke adjacent to the metal contaminant may have been caused by the metal or by some other function in the catalyst. Using the terminology of Habib et al. [31], peak L is assigned to contaminant coke.

Of the two remaining peaks, one or both must be attributed to coke from acid-catalyzed cracking. Peak N was prominent in coke produced from SIHGO and ASTM feeds but did not appear in coke produced from n-hexadecane cracking. Clearly, peak N is related to feedstock. However, it cannot be produced by Conradson carbon since: (1) both SIHGO and ASTM feeds have very low Conradson carbon residue; and (2) the MAT unit would deposit Conradson carbon either on the preheater or on the quartz wool plug that protects the top of the catalyst bed. Conradson carbon would not be found on spent catalyst from a MAT unit. Feedstock factors that might affect peak N are basic nitrogen content or multi-ring-aromatic content. SIHGO feed contains more basic nitrogen and more multi-ring aromatics than ASTM feed. Peak N from SIHGO cracking is generally larger than peak N from ASTM feed cracking, except for one case. Peak N

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

209

Fig. 11. Analysis of TPO spectrum for ECat-LOW after cracking n-hexadecane. Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

was largest when ASTM feed was used with the most active catalyst. Therefore, peak N must be related to both feedstock and catalyst activity. Both relationships can exist if peak N is produced from a highly aromatic form of coke. High multi-ring-aromatic concentra-

tions for coking reactions my come directly from the feedstock or the high concentration may be the result of cracking heavy feeds that yield multi-ring-aromatic products. Highly aromatic forms of coke should assume some of the high temperature characteristics of

Fig. 12. Analysis of TPO spectrum for ECat-INT after cracking n-hexadecane feed. Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

210

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

Fig. 13. Analysis of TPO spectrum for ECat-HIGH after cracking n-hexadecane feed. Overlay of experimental curve and composite curve from analysis (top). Individual peaks from analysis (bottom).

Fig. 14. Effect of contaminant-metal concentration on: (a) peak areas; and (b) peak positions. Data obtained with n-hexadecane feed. TPO spectra deconvolution shown in Figs. 11–13.

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

graphite. In a departure from prior terminology, peak N is assigned to graphitic coke, which is related to both feedstock properties and catalyst activity. Peak M is assigned to catalytic coke produced by acid-catalyzed cracking. This form of coke is more reactive to oxygen than the graphitic coke that produces peak N, but less reactive than the coke associated with metal contaminants. Both peak M and N should correlate with catalyst activity. Reichle [32] pointed out many years ago that coke concentration on FCC catalysts often correlates with

211

kinetic conversion. Fig. 15 shows the relationship between kinetic conversion and the area of peaks M +N. The kinetic conversion, designated k, is defined where k = % conversion/(100 − % conversion). A 75% conversion of a particular feed would be a kinetic conversion of 3.00. A 50% conversion would be a kinetic conversion of 1.00. Kinetic conversion is listed for every data set in Table 3. Fig. 15 shows the sum of the areas for peaks M and N plotted as a function of kinetic conversion. Linear trend lines have been drawn that pass through the

Fig. 15. Relationship between kinetic conversion and area sum of peaks M and N for SIHGO feed, ASTM feed and n-hexadecane.

212

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213

origin. The data fit are only fair, but are consistent with expectations. In all cases, coke area increases with conversion for a specific feed. SIHGO feed has a steeper slope than ASTM feed, which in turn has a steeper slope than n-hexadecane feed. The heaviest, most aromatic feed makes the most coke. This coke-making tendency for different feeds can be compared at k = 1.00 (50% conversion), where the area sums are: SIHGO feed 12.1, ASTM feed 5.8, and n-hexadecane feed 2.5. This comparison shows that SIHGO feed is expected to produce twice as much coke as ASTM feed and five-times as much coke as n-hexadecane. The data in Fig. 2 show results from TPH with coked ECat-INT. It is significant because TPH data correlates directly with TPO data. Prior discussion assigns the first TPH peak to desorbing hydrocarbons and the second TPH peak to metal-catalyzed coke hydrogenation. The assignments are similar with TPO data. The first peak is assigned to desorbing hydrocarbons (cat-to-oil coke) and the second peak is assigned to metal-catalyzed coke oxidation (contaminant coke). The TPH spectrum from n-hexadecane cracking shows three distinct peaks without deconvolution. The TPO spectrum with n-hexadecane-cracked catalyst has three peaks after deconvolution, where the last peak is assigned to conversion coke produced from acid-catalyzed reactions. The third peak in the TPH spectrum must also be due to conversion coke, since TPH and TPO results agree for n-hexadecane-produced coke. The fourth peak in the TPO spectra from SIHGO and ASTM feeds has been assigned to graphitic (graphite-like) coke. This peak is not observed in TPH spectra terminated at 865 ◦ C. Very high temperatures are needed to gasify graphitic coke.

4. Conclusion TPO spectra of spent FCC catalysts can be deconvoluted into four separate peaks. These peaks represent desorbing hydrocarbons, contaminant-metal coke, acid-catalyzed reaction coke and graphitic coke. These cokes had been classified from refinery operations as cat-to-oil coke, contaminant coke, conversion coke and Conradson coke. Three names used previously are adopted in this paper for the first three

peaks in the TPO spectra. The fourth type of coke observed in TPO spectra is related to both feedstock type and catalyst activity. This has been renamed graphitic (graphite-like) coke rather than Conradson coke to emphasize that the graphite-like character of the coke is more important than the Conradson carbon concentration of the feedstock. The sizes of the TPO peaks are proportional to the amount of coke on catalyst. The amount of contaminant coke (peak L) correlates with contaminant-metal concentration. The sum of the conversion coke (peak M) and the graphitic coke (peak N) correlates with catalyst activity. This sum can be used to characterize the coking tendency of a feedstock. Feedstock comparisons at 50% conversion show that SIHGO feed produces twice as much coke as ASTM feed and nearly five times as much coke as n-hexadecane. TPH results are less useful in characterizing coked catalysts. The peaks in TPH spectra can be correlated with only the first three peaks of TPO spectra. The high temperature peak assigned to graphitic coke (peak N) in TPO spectra is not observed in TPH spectra. This situation occurs because graphitic coke is more reactive with oxygen than with hydrogen, so that oxidation occurs within the temperature range of experimental equipment, while graphitic coke hydrogenation occurs at higher temperatures, beyond the range of the TPH apparatus.

Acknowledgements Support from the Turkish Ministry of National Education and the West Virginia University Department of Chemical Engineering is gratefully acknowledged. We wish to thank Mr. Howard Moore, Marathon-Ashland LLC, for his helpful discussions and for providing the equilibrium FCC catalyst samples used in this study. We also thank the West Virginia University Office of Academic Affairs for a grant to purchase the platinum oxidation catalyst used in this study. References [1] C.A. Querini, S.C. Fung, Catalysis Today 37 (1997) 277. [2] J.G. McCarty, H. Wise, J. Catal. 57 (1979) 406. [3] J. van Doorn, H.A.A. Barbolina, J.A. Moulijn, Ind. Eng. Chem. Res. 31 (1992) 101.

O. Bayraktar, E.L. Kugler / Applied Catalysis A: General 233 (2002) 197–213 [4] D. Duprez, K. Fadili, J. Barbier, Ind. Eng. Chem. Res. 36 (1997) 3180. [5] J. van Doorn, R.C.S. Verheul, L. Singoredjo, J.A. Moulijn, Fuel 65 (1986) 1383. [6] J.L. Figueiredo, Fuel 65 (1986) 1377. [7] M.A. Goula, A.A. Lemonidou, A.M. Efstathiou, J. Catal. 161 (1996) 626. [8] R.T. Baker, I.S. Metcalfe, Ind. Eng. Chem. Res. 34 (1995) 1558. [9] S.V.L.S. Teixeira, R. Frety, M. Schmal, Ind. Eng. Chem. Res. 33 (1994) 1692. [10] Å. Wrammerfors, B. Anderson, J. Catal. 147 (1994) 82. [11] E.S. Oh, Y.C. Park, I.C. Lee, H.K. Rhee, J. Catal. 172 (1997) 314. [12] M. Larsson, J. Jansson, S. Asplund, J. Catal. 178 (1998) 49. [13] J.M. Grau, J.M. Parera, Appl. Catal. 70 (1991) 9. [14] J. Barbier, E. Churin, P. Marecot, J. Catal. 126 (1990) 228. [15] P.K. Doolin, J.F. Hoffman, M.M. Mitchell Jr., Appl. Catal. 71 (1991) 233. [16] V.D. Dimitriadis, A.A. Lappas, L.A. Vasalos, Fuel 77 (1998) 1377. [17] C. Li, T.C. Brown, Energy Fuels 13 (1999) 888. [18] C.L. Minh, R.A. Jones, I.E. Craven, T.C. Brown, Energy Fuels 11 (1997) 463. [19] C. Li, C.L. Minh, T.C. Brown, J. Catal. 178 (1998) 275.

213

[20] Y. Yoshinura, E. Furimsky, Fuel 65 (1986) 1388. [21] P. Zeuthen, B.H. Cooper, Ind. Eng. Chem. Res. 134 (1995) 755. [22] J. Bartholdy, P. Zeuthen, F.E. Massoth, Appl. Catal. 129 (1995) 33. [23] S.C. Fung, C.A. Querini, J. Catal. 138 (1992) 240. [24] D.F. Tatterson, R.L. Mieville, Ind. Eng. Chem. Res. 27 (1988) 1595. [25] V. Cadet, F. Raatz, J. Lynch, Marcilly Ch., Appl. Catal. 68 (1991) 263. [26] W.-C. Cheng, M.V. Juskels, W. Suárez, Appl. Catal. 103 (1993) 87. [27] P.K. Doolin, J.F. Hoffman, M.M. Mitchell, Appl. Catal. 71 (1991) 233. [28] A. Haas, W. Suárez, G.W. Young, in: K.C. Chuang, George W. Young, M. Benslay (Eds.), Proceedings of the AIChE Symposium Series on Advanced Fluid Catalytic Cracking Technology, Vol. 88, 1992, p. 133. [29] C.A. Querini, S.C. Fung, Appl. Catal. 117 (1994) 53. [30] M. Larsson, Ph.D. Thesis, Department of Chemical Reaction Engineering, Chalmers University of Technology, Sweden, 1997. [31] E.T. Habib, H. Owen, P.W. Snyder, C.W. Streed, P.B. Venuto, Ind. Eng. Chem. Prod. Res. Dev. 16 (1977) 291. [32] A.D. Reichle, Exxon Research and Development Laboratories, Baton Rouge, LA, 1984, private communication.