Catalytic Hydroprocessing of SRC-I I Heavy Distillate Fractions. 5

May 31, 1984 - 5. Conversion of the Acidic Fractions Characterized by Gas. Chromatography/Mass Spectrometry. David W. Grandy and Leonidas Petrakisat...
0 downloads 0 Views 1MB Size
40

Ind. Eng. Chem. Process Des. Dev. 1986, 25, 40-48

Nukiyama, S.; Tanasava, Y. Trans. Soc.Mech. Eng. (Jpn.) 1939, 5 , 63. Lapple, C. E.; Kamack, H. S. Chem. Eng. h o g . 1955, 51, 133. Pyla Naidu, MSc. Thesis, Indian Institute of Science, Bangalore, India,1980. Semaru, K. T.; Marunowski, C. W.; Lude, K. E.; Lapple. C. E. Ind. Eng. Chem. 1958, 50, 1615. Seymour Calvert AIChE J. 1970, 16. 392.

Virkar, P. D.; Sharma, M. M. a n . J . Chem. E ~ Q .1975, 13. 512. Volgln, B. P.; Efimova, T. F.; Gofman, M. S.Int. Chem. Eng. 1968, 8 , 113.

Receiued f o r review May 31, 1984 Accepted April 23, 1985

Catalytic Hydroprocessing of SRC-I I Heavy Distillate Fractions. 5. Conversion of the Acidic Fractions Characterized by Gas Chromatography/Mass Spectrometry David W. Grandy and Leonidas Petrakisat Gulf Research and Development Company, Plttsburgh, Pennsylvania 15230

Cheng-Lie LI and Bruce C. Gates' Center for Catelyfic Science and Technobgy, Department of Chemical Engineerlng, University of Delaware, Newark, Delaware 19716

Kinetics data have been determined for the catalytic hydroprocessing of the acidic fractions of a heavy distillate of a liquid derived from Powhatan No. 5 coal. A commercial, sulfided Ni-Mo/y-Al,O, catalyst was used in the experiments, carried out at 350 OC and 120 atm with the coal liquid fractions dissolved in cyclohexane. The feed and hydrotreated products were analyzed by gas chromatography/mass spectrometry. The data were analyzed with group-type methods for compound classes, and results were also obtained for some individual organooxygen compounds. Catalytic hydroprocessing leads to a large increase in the number of compounds and a shift to lower boiling ranges. The data are broadly consistent wlth reaction networks determined with pure compounds; the most important reactions include aromatic ring hydrogenation, hydrodeoxygenation, and hydrodemethylation. Pseudofirst-order rate constants for conversion of the predominant organooxygen compounds are on the order of L/(g of catalyst-s); the reactivity decreases in the order cyclohexylphenol > dimethylhydroxyindan > tetrahydronaphthol > phenylphenol > I-naphthol.

Hydrodeoxygenation is a class of catalytic hydroprocessing reaction that has received scant attention because the oxygen content of petroleum is almost negligibly low. But the need for alternative energy sources has focused attention on coal and shale,which have much higher oxygen contents. The liquid fuels derived from these sources contain compounds typified by 1-naphthol and dibenzofuran (Grandy et al., 1984);reaction networks for catalytic hydrogenation/ hydrodeoxygenation of a few of these compounds have been recently reported (Krishnamurthy et al., 1981; Li et al., 1985A; Furimsky, 1983). Our approach to the determination of the reactivities of organooxygen compounds in synthetic fuels has involved (1)liquid-chromatographic fractionation of a heavy distillate derived from hydroliquefied Powhatan No. 5 coal into classes of compounds distinguished by chemical types (Petrakis et al., 1983A) and (2) investigation of the reactions of the individual fractions with hydrogen under carefully controlled conditions in a high-pressure catalytic microreactor. the acidic fractions of the coal liquid contain high concentrations of substituted naphthols and related compounds (Petrakis et al., 1983B; Grandy et al., 1984). The first experiments characterizing the reactivities of the weak- and very-weak-acid fractions have been reported (Li et al., 1985B). Characterization of the reactants and Present address: Chevron Research Co., Richmond, CA 94802. 0196-4305/86/1125-O040$0 1.5010

products by combined gas chromatography/mass spectrometry (GC/MS) determined quantitative reactivities of the several major organooxygen compounds in the presence of sulfided Ni-Mo/yA1203 catalyst under hydroprocessing conditions representative of potential industrial applications. In the work described here, we have taken the feed and product samples obtained in the above-mentioned work and characterized them exhaustively by GC/MS. The additional data have been analyzed by group-type methods, the results determining reactivities of several classes of compounds; further, improved GC/MS capability has provided reactivity information for individual compounds which was not attainable in the earlier experiments. Experimental Section Materials. The catalyst, a standard commercial NiOMo03/yA1203(Houalla et al., 1978), and the coal-liquid fractions (Petrakis et al., 1983A,B) have been described in detail. The two coal-liquid fractions (the weak acids and the very weak acids) were dissolved in cyclohexane and used as feeds to an isothermal, piston-flow microreactor. Catalytic Hydroprocessing. The acidic fractions were used as 0.25 wt % solutions in cyclohexane, and these feed solutions were saturated with hydrogen at room temperature. CS2 (0.1 wt %) was added to the feeds to ensure that the catalyst remained in the sulfided state. The reaction conditions were 350 "C and 120 atm. Space ve0 1985 American Chemical Society

Ind. Eng. Chem. Process Des. Dev., Vol. 25, No. 1, 1986 41

locities ranged from 0.05 to 3.2 g of acid fraction/(g of catalyst-h). A more detailed description of the reactor system and its operation has been published (Li et al., 1985B). The samples characterized in this work were the ones described previously by Li et al. (1985B). Analysis of the Reactants and Products. The Finnigan GC/MS instrument was equipped with a Model 9600 gas chromatograph. The products were resolved in a J & W fused silica capillary column, 30 m X 0.24 mm, with an SE-54 liquid phase. The injection and transfer line(s) was maintained at 250 OC. The column temperature was programmed from 50 to 220 “C, with a 2 OC/min ramp and a 5-min hold at 220 “C. Ultrahigh purity helium carrier gas flowed at a rate of 1.5 cm3/min. The liquid sample volume was typically 1 pL, without splitting. The Finnigan 4510 mass spectrometer, which has a mass range of 33-400 with a resolution of about 2000, was operated with an electron energy of 70 eV, an emission current of 0.25 mA, and a multiplier voltage of 1200 V. The preamplifier sensitivity range was A/V, and the preamplifier filter was set at 1000 amu/s. The scanning was done at 0.95 s/scan with a 0.04-s hold. About 10 mass spectra were obtained per GC peak. A Nova 4/IN COS 2000 system was employed for handling and storage of the data. The NBS search library of about 31 OOO compounds was used for peak identification. The search routine was done in the “purity” mode, whereby the hierarchy of fit was determined by the number of peaks in the unknown that matched the library spectrum, with intensity ratios considered. This mode is different from the “fit” mode, whereby the match of peaks in the library to those in the unknown (without regard to intensities or extra peaks in the unknown) determines the goodness of fit, and is different from the “reserve fit” mode, whereby the unknown spectrum is matched to the library spectrum in the same way. In the analysis of the GC/MS data, a peak assignment was considered to be correct, provided that (1)the number of points characterizing quantitatively the goodness of match between the experimental spectrum and the library spectrum was high, typically 800 or more out of lOOO, (2) the next best fits, if they were essentially the same, were isomers of the same compound, and (3) the assignment made good chemical sense. The assignment of an isomer to an empirical formula was considered to be correct if on an overall scale of 1000, the next best fit was 100 less. In some instances, assignments were corroborated by spiking experiments and GC analysis. Often compounds could not be identified, and the best that could be done was the assignment of an empirical formula; for example, it was not possible to determine whether a GC peak represented an isomer of methyltetralin or dimethylindan. These incomplete identifications are still helpful because they allow comparison of the feed and the products in terms of the carbon number and degree of saturation. Since the data-handling package on the GC/MS computer did not provide integrated intensities, it was necessary to match the ion chromatograms with the FID trace from a separate run on the gas chromatograph, which did provide intensity data. n-Dodecane was added to each sample prior to injection at a concentration of 2 wt % as an internal intensity standard. The determination of the concentrations of the individual components was limited in many instances, for one or more of the following reasons: (1)There was no reasonable peak assignment by GC/MS; (2) there was no calculated intensity for a given peak determined by FID/GC as a consequence of the peak’s being too small, (3) it was not possible to match the GC/MS peak

SCAN TIME, s

I l l

I

. ‘13

IO00

2000

3000

4000

5000

SCAN T I M E , s

Figure 1. (A, top) Ion chromatogram of the very-weak-acid feed. (B, bottom) Ion chromatogram of the very-weak-acid product obtained at a space velocity of 0.4 g of avid fraction/(g of cata1yst.h).

with a peak on the FID/GC trace because of differences in resolution (e.g., in some cases twin peaks in the FID chromatogram having a single calculated area were resolved into two peaks with different structural assignments in the GC/MS chromatogram). Thus, in some instances, no intensity could be assigned to an identified compound. Results and Discussion Formation of Lower-Molecular-Weight Products. All the products were characterized by GC/MS, and the most detailed analyses were carried out by characterizing the two feed samples (the weak- and very-weak-acid fractions) and the products of their conversion at one intermediate space velocity, namely, 0.78 g of acid fraction/(g of catalyst-h) for the weak-acid fraction and 0.4 g of acid fraction/ (g of cata1yst.h) for the very-weak-acid fraction. The ion chromatograms of the feeds and products are shown in Figures 1and 2. These show that the feeds are devoid of any light materials ( dimethylhydroxyindan = methyltetrahydronaphthol tetrahydronaphthol > methylphenylphenol phenylphenol. Methyl substitution of an organooxygen compound increases the rate of disappearance by providing an additional pathway for reaction (hydrodealkylation). Neither this observation nor the above order of reactivity implies anything about the ease of oxygen removal. Acknowledgment We thank S . S. Starry and S. K. Banerjee for experimental assistance. This work was supported by the U.S. Department of Energy.

- -

Registry No. Cyclohexylphenol, 26570-85-4; dimethyldihydroxyindan, 98303-72-1;methyltetrahydronaphthol,62184-87-6; tetrahydronaphthol, 518,54-14-9; methylphenylphenol,88403-77-4; phenylphenol, 1322-20-9; methylphenoxybenzene, 31324-44-4; 1-naphthol, 90-15-3.

Literature Cited Furimsky, E. Catal. Rev.-Sd. Eng. 1983, 25, 421. (;randy, D. W.; Petrakis, L.; Young, D. C.; Gates, B. C. Nature (London) 1984, 308, 175. HouaHa. M.; Nag, N. K.; Sapre, A. V.; Broderick, D. H.; Gates, B. C. AIChE J 1978, 24, 1015. Krishnamurthy, S.; Panvelker, S.; Shah, Y. T. AIChE J . 1981, 2 7 , 994. Li, C.-L.; Xu, Z.R.; Gates, B. C.; Petrakls, L. I&. Eng. Chem. Process Des. Dev. 19858, 24, 92. Li, C.-L.; Xu, 2.-R.; Cao. Z. A,; Gates, B. C.; Petrakis, L. AIChE J. 1985A, 31, 170. McNeIl, R. I.; Cronauer, D. C.; Young, D. C. Fuel 1983, 62, 401. Patzer, J. F.. 11; Ferrauto, R. J.; Montagna, A. A. Ind. Eng. them. Process Des. Dev. 1979, 18, 625. Petrakis, L.; Ruberto, R. G.; Young, D. C. Ind. Eng. Chem. Process Des. Dev. 1983A. 22, 292. Petrakis, L.; Young, D. C.; Ruberto, R. G. Ind. Eng. Chem. Process Des. Dev. 19838, 22, 298. Ruberto, R. G. FuelProcess Techno/. 1980, 3 , 7. Sapre, A. V.; Gates, B. C. Ind. Eng. Chem. Process Des. Dev. 1981, 20, 68.

Received for review May 18, 1984 Accepted April 19, 1985