Mechanism of Oxidative Regeneration of Molybdate Catalyst

The regeneration of cobalt molybdate catalyst used during hydrotreatment of Athabasca bitumen was carried out in fixed bed and in thermogravimetric re...
3 downloads 0 Views 722KB Size
Ind. Eng. Chem. Res. 1987,26, 657-662

657

Mechanism of Oxidative Regeneration of Molybdate Catalyst Edward Furimsky* and Yuji Yoshimura Energy Research Laboratories, Canada Centre for Mineral and Energy Technology, Energy, Mines and Resources Canada, Ottawa, Canada K I A OGI

The regeneration of cobalt molybdate catalyst used during hydrotreatment of Athabasca bitumen was carried out in fixed bed and in thermogravimetric reactors. During temperature programmed burnoff in air, two distinct maxima for SO2production were observed. One, a t about 250 "C, resulted from oxidation of sulfidic sulfur, whereas the other, at about 450 "C, resulted from oxidation of organic sulfur. The maximum for oxidation of carbon always coincided with the second maximum. The formation of SO2,CO, and CO, during treatment in N2 was explained on the basis of participation of Moos species in oxidation reactions. The mechanism of oxidation was proposed to explain the oxygen uptake by coke a t temperatures lower than 400 "C observed in the thermogravimetric reactor. Catalyst activity declines during catalytic treatment of petroleum fractions. The decline can be offset by a change of processing parameters such as temperature, pressure, and residence time. In systems where the catalyst is continuously fed to and withdrawn from the reactor, the activity may be maintained by periodically withdrawing a portion of the used catalyst and replacing it by fresh catalyst. The deactivated catalyst may be regenerated and returned to the operation. In the case of fluid catalytic cracking (FCC) units, catalyst regeneration is part of the continuous operation and as such has reached a commercial stage. This suggests that the regeneration of FCC catalysts, i.e., silica-alumina or zeolite. catalysts, is the best understood. Indeed, the most advanced stage of modeling of regeneration processes was reached for this type of catalyst (Hashimoto et al., 1984; Hughes et al., 1985). Among hydrotreating catalysts, molybdate supported on alumina and promoted by either cobalt or nickel are the most common. A great deal of attention has been given to the loss of activity of these catalysts. Most of the available information suggests that the formation of carbonaceous deposits is the most common cause of the activity decline. It is generally understood that the rate of deposit formation and its structure depends on the type of hydrotreated feedstock. Thus, nitrogen bases and some aromatic structures are known coke precursors (Furimsky, 1980). Organometallic species containing transition metals also contribute to catalyst fouling (Furimsky, 1978). Information on regeneration of hydrotreating catalysts is limited, and only recently were oxidative burnoff studies of molybdate catalysts reported (Nalitham et al., 1985; Klusacek et al., 1985; Hertan et al., 1985). The regeneration of sulfidic and oxidic forms of Co-Mo/A120Bcatalyst used during hydrodeoxygenation of a phenol solution was reported by Yoshimura and Furimsky (1986), who observed that a smaller amount of coke was deposited on the sulfidic catalyst compared with that deposited on the oxidic catalyst under the same hydrotreating conditions. Also, regeneration of the sulfidic catalyst occurred more readily than that of the oxidic catalyst. It was further observed that during oxidative burnoff sulfided forms of Mo and Co are converted to their oxidic forms. The maximum of this conversion occurred at about 250 "C. At this temperature, the conversion of coke was insignificant. To achieve high levels of coke conversion, a temperature of about 500 "C was required. In the present study, the regeneration of Co-Mo/A1203 catalyst was expanded to include the catalyst used to hydrotreat Athabasca bitumen. The work was performed in fixed bed and thermogravimetric reactors. Special atten-

Table I. Chemical Composition of Catalysts a n d O t h e r Solid Samples wt%

sample fresh presulfided catalyst fresh presulfided THF extracted catalyst spent THF extracted catalyst spent toluene extracted catalyst oil sand coke toluene insoluble of virgin pitch toluene insolubles of thermally hydrocracked pitch

C 0.6 11.4 12.7 82.3 81.0 81.5

H 2.3 2.1 2.1 5.1 5.2

N

S

0.5 0.5 1.9

5.98 5.82 4.37 4.66 7.32 6.96 7.90

2.0 2.2

tion was given to different types of sulfur present in spent catalysts and to their effect on the regeneration process. Attempts were made to elucidate the effects of pretreatment of spent catalysts on oxidative burnoff. The impact of the experimental results on development of models to follow burnoff kinetics of hydrotreating catalysts in relation to FCC catalysts was assessed as well.

Experimental Section Catalyst. The catalyst was the extrudate form (about 1 mm in diameter) of commercial Ketjen Co-Mo/A1203 containing about 16 wt % Moo3 and about 4 wt ?' & COO. The BET surface area and pore volume of this catalyst were 179 m2/g and 0.39 mL/g, respectively. This catalyst was used for upgrading Athabasca bitumen in a series of experiments performed from 350 to 460 "C under H2 pressure of 13.9 MPa. For the fixed bed experiments, the catalyst was extracted either by toluene or THF. This was followed by vacuum treatment at 250 "C. The pretreated catalyst was crushed to a particle size of 100-200 mesh. Elemental analyses of these catalysts and of other related materials are shown in Table 1. In thermogravimetric reactor experiments, the crushed THF extracted catalyst, treated under vacuum at 250 "C, was subjected to the following additional pretreatments: (A) preoxidation in air at 250 "C for 1h, (B) heat treatment under N2 at 500 "C for 1h, and (C) preoxidation in air at 250 "C for 1 h in the thermogravimetric reactor. Burnoff System and Procedure. Experiments were carried out in fixed bed and in thermogravimetric reactors. Experimental conditions are summarized in Table 11. Fixed Bed Reactor. A schematic diagram of the experimental fixed bed system is shown in Figure 1. The 15-mm-i.d. reactor was made of 316 SS. The fixed bed (Figure 2) consisted of three layers of quartz on a perforated plate on top of which was the layer of catalyst. The catalyst layer was covered by another two layers of quartz.

0888-5885/87/2626-0657$01.50/0 Published 1987 by the American Chemical Society

658 Ind. Eng. Chem. Res., Vol. 26, No. 4, 1987 Table 11. Burnoff Conditions

250-500 350-500 0.10 0.20 100-200 100-200 air or N2 5% O2 air 0.40 0.08 weight loss yields of products

temp, o c sample size, g mesh oxidizing gas flow of gas, L/min data measurement

+

4

"81-

G A S CYLINDER DRYER 3 FLOW METER 4 MIXING COLUMN

I

2

8

t

?6

5 REACTOR 6 COOLING COIL 7 ELECTRIC FURNACE E ICE WATER TRAP

9 IO

/I I2

SO2- REMOVER CO CONVERTER M A S S SPECTROMETER SORPT-FILM FLOW METER

-

Figure 1. Schematic diagram of experimental fixed bed system. TEMPERATURE, "C Figure 3. Profiles of C 0 2 and SO2 formation under N2 and in air for toluene extracted spent catalyst.

29 of Q (32-80merh)49 of Q(32-8Omash)-

2a of Q (12-14 mash)

+IIIII

0.04 to 0 . 2 g of catalyst (100200 mesh) t0.49of Q (- 2 0 0 mesh)

-

2 9 of Q(8-10mesh)

PERFORATED

---.-)

PLATE

Figure 2. Details of fixed bed.

With this arrangement, the temperature increase in the initial stages of burnoff was insignificant, e.g., a t 500 "C the temperature increased by 4 "C but leveled off at 500 "C after about 1 min. At lower burnoff temperatures, the temperature increase was smaller than that observed a t 500 "C. Other details of the system were published elsewhere (Yoshimura and Furimsky, 1986). The oxidizing medium was prepared by mixing N2 with 20 vol % 02.The gases were UHP grade. This mixture is referred to as air. Before the experiment, the reactor was heated under N2 to the anticipated temperature of burnoff. After equilibration of the system, the N2 was replaced by air. The measurement of experimental parameters started a t this point. Thermogravimetric Reactor. The burnoff runs were performed by using a Dupont 951 TGA analyzer equipped with a Du Pont 1090 data system. Before each burnoff experiment, the sample was maintained at a corresponding temperature under N2 until no further weight change occurred. Then the N2 was replaced by the oxidizing medium. The thermocouple to measure the burnoff temperature was placed in proximity to the sample holder. Analysis. Perkin-Elmer CHN 240 and Leco analyzers were used for the determination of carbon and sulfur, respectively. An on-line Perkin-Elmer 1200 multiple mass spectroscope analyzed and continuously monitored the concen-

tration of products formed during regeneration, e.g., C02, CO, and SO2. No attempts were made to analyze H2 and H20. For most experiments, the gas exiting the reactor passed tubes filled with CuO and/or MnO, before entering the analysis system. The data log system attached to the mass spectroscope provided print outs of results every 20 s. The concentrations determined by this procedure together with the volume of flowing gas were used to calculate the number of moles of individual components in the gas exiting the reactor. The number of moles of carbon containing components and SO2 were used to calculate the amGunt of carbon and sulfur removed during burnoff.

Results and Discussion Reactivity of Deposits. A temperature-programmed burnoff such as that used by Massoth (1981) is suitable for obtaining information on the reactivity of deposits. For the deactivated catalyst extracted by toluene, the burnoffs performed in air, following treatment under N2, gave profiles as shown in Figure 3. In both cases the temperature increase was maintained a t 10 "C/min. For treatment under N2,the final temperature was 500 "C and heating at this temperature continued for another 30 min. A t this point, the catalyst was cooled to near room temperature before the burnoff in air began. In this case, the temperature was increased continuously to about 700 "C. During heat treatment under Nz, about 20% S originally present in the catalyst was removed as SOz. The SO2 formation exhibited two maxima, one at 375 "C and the other at 475 "C. The second maximum coincided with that for COz formation, suggesting that this portion of SOz results from oxidative destruction of organic matter. For the THF extracted catalyst, the subsequent treatment under N2 doubled the SOz peak with a maximum at 375 "C compared with the toluene extracted catalyst. During burnoff in air, the maximum for SOz formation occurred at about 250 "C (Figure 3). At this temperature, the carbon removal was insignificant, but it increased with

Ind. Eng. Chem. Res., Vol. 26, No. 4, 1987 659 loo

!I

e x

OIL SAND COKE TI O F T H P m TI OF VP

I

1

i

FUESULPHIDED AND PREDXIDIZED AT 25OF

o I:I MIXTURE OF

UNDER

OXlDlc

AND PRESULPHIDED x-X

1

PRESLmlDED

\x

i-

"2

TEYPERbTURE KEPT bT

i \I

i

TEMPERATURE, O C

Figure 5. Effects of pretreatment of fresh presulfided catalyst on SOz yields produced under NO

200r

v

IS02FOR

0

1 X

PRESULPHIDED AND THF EXTRACTED

,

prEpT

C q FOR

PRESULPHIDED AND THF EXTRACTED

TEMPERATURE

bTmt

SO2 FOR PRESULPHIDED

TEMPERATURE,aC Figure 4. Profiles of COz and SOz formation for oil sand coke and toluene insolubles of virgin pitch and of thermally hydrocracked pitch.

o o 0 0 00

increasing temperature and reached a maximum at about 420 "C. Also, the second SOz maximum coincided with that for COz formation. To confirm with certainty the organic sulfur matter as being the origin of SOz formed around the second maximum, the burnoff experiments were performed on three materials derived from Athabasca bitumen, Le., oil sand coke from fluid coking, toluene insolubles of thermally hydrocracked pitch, and toluene insolubles of virgin pitch. Thus, as the results in Figure 4 show, for all three materials the maximum for the formation of SOz occurred near that for COzformation. For oil sand coke, the burnoff maximum was at 490 "C, whereas that for toluene insolubles was at 440 "C. This agrees with the markedly lower reactivity of coke compared with insolubles. The results for the deactivated catalyst obtained in air and in N2 (Figure 3) indicate that SOz is formed in more than one reaction. To explain this, the fresh catalyst was sulfided in a H2 + H2S mixture containing 10 vol % H2S and subsequently treated in N2and in air under conditions identical with those for the deactivated catalyst. In addition, the sulfided catalyst was preoxidized at 250 "C to remove about 50% sulfur and heated under N2 up to 500 "C. The results shown in Figure 5 suggest that the maximum for SOz formation observed at about 490 "C originated from sulfur oxidation by molybdenum and cobalt oxides. This was confirmed by a significant increase in SOz yield during heating a 1:l mixture of the sulfided and oxidic catalyst in Nz. The fresh sulfided catalyst was extracted by THF and treated in N2and in air. As expected, the THF extraction resulted in a small deposit as indicated by the formation of C 0 2 (Figure 6a). Moreover, during treatment in N2, the SO2 yield increased compared with the fresh sulfided unextracted catalyst. During burnoff in air which followed treatment in N2,the SO2yield for the extracted catalyst was lower than that for the unextracted catalyst (Figure 6b). This agrees with the observation made during the

L ~

PRESULPHIDED PRESULPHIDED AND THF EXTRACTED

TEMPERATURE, C'

Figure 6. Effects of extraction of fresh sulfided catalyst on SOz yields under N2 and in air.

burnoff of the THF extracted spent catalyst. This may be attributed to the removal of sulfur during extraction. It is speculated that some sulfur was interchanged by oxygen from THF. The higher strength of the Mo-0 bond compared with the Mo-S bond (Barin and Knacke, 1973) supports such an interchange. Figure 7 shows the results for the sulfided catalyst treated in hexadecane at 350 "C and 6.89 MPa of H2 (Furimsky et al., 1986) and subsequently extracted by either THF or toluene. During treatment in N2 (Figure 7a), the SO, yields for the THF extracted catalyst were markedly higher than those for the toluene extracted catalyst, i.e., about 70% and 15% sulfur was removed as SOz for the THF and toluene extracted catalyst, respectively. The large SOz yield for the THF extracted catalyst is attributed to organic oxygen which accumulated in the coke during extraction. Thus, the slight sulfur decrease during the THF extraction, i.e., from 5.98 to 5.82 wt % , would not be sufficient to produce enough molybdenum and cobalt oxides to produce these quantities of SO2. As one would expect, the SOz yields obtained during the burnoff in air for the toluene extracted catalyst were markedly higher than that for the THF extracted catalyst. The same catalyst was used for hydrodeoxygenation of a phenol solution in hexadecane. The dotted line in Figure 7a represents yields of SO2 of the toluene extracted catalyst

660 Ind. Eng. Chem. Res., Vol. 26, No. 4, 1987