Ind. Eng. Chem. Prod. Res. Dev., Vol. 18, No. 3, 1979
SO2, Pt-Ni-SO,, Pt-HzS and Ru-H,S) lower activation energies for NzO formation suggest that conversion of NzO is occurring a t the higher temperatures required for reaction with sulfur-deactivated catalyst. The activation energy for SO2-poisoned Ru/AlZO3is about twice that for the other systems, and the low value of Nz/N20 (1.25) with this catalyst a t 536 K suggests that some NH, decomposition may be taking place. The high tolerance to SO2 of Ru/AlZO3and Pt-Ni/A1203 results in a sharp reversal of the reactivity order for these catalysts in NO reduction by NH3 At 473 K with SOz-free Ni, feed the reactivity order is Pt Pd >> Ru > Pt-Ni whereas with 50 ppm of SO2at 673 K the order is Ru >>> Pt-Ni >> Pt, Pd, Ni. Ru was about lo6 times more active than Pt or Pd at the higher temperature with SOz although it was about two orders of magnitude less active than Pt a t 473 K without SOz in the feed gas. The Pt-Ni activity was several orders of magnitude greater than observed for either of the metals alone and the catalyst showed a tolerance to higher SO2 levels similar to that observed with Pt but a t a much higher activity level. Acknowledgment The authors wish to recognize and thank the General Motors Research Laboratories for their support of this work. The HzS data were taken by Kevin Keenley as part of an undergraduate work-study project. Literature Cited
-
-
Anderson, H. S.,Geen, W. T., Steele, D. R., Ind. Eng. Chem., 53, 199 (1961). Bart&, W., Crawford, A. R., Cunningham, A. R., Hall, H. J., Manny, E. H., Skopp, A., "Systems SWiy of Nibogen Oxide control Memods for StatioMIy Swces", Final Report, Vol. 11, prepared under contract Ph-22-68-55 for National Air Pollution Control Administration by Esso Research and Engineering Co., Nov 20, 1969.
179
Bauerle, G. L., Wu, S.C., Nobe, K., Ind. Eng. Chem. Prod. Res. Dev., 14. 123 (1975). Campau, R . M., Stefan, A., Hancock, E. E., SAE Paper 720488 (May, 1972). Environmental Protection Agency, "Air Quality Criteria for Nitrogen Oxides", Air Pdlution Control office Publition No. AP-84, US. Government Printing office, Washington, D.C., Jan 1971. Gagliirdi, J. C., Smith, C. S..Weaver, E. E., Paper No. 63-72 at Midyear A.P.I. Meeting, New York, May 8-1 1, 1972. bid, R. M., Huss, A., B.S. Thesis, Department of Chemical Engineering, University of Delaware, 1974. Hartley, E. M., Matteson, M. J., Ind. Eng. Chem. Fundam., 14, 67 (1975). Jackson, H. R., McArthur, D. P., Simpson, H. D., SAE Paper 730568 (May, 1973). Jones, J. H., Kummer, J. T., Otto, K., Sheief, M., Weaver, E. E., Environ. Sci. Techno/., 5 , 790 (1971). Katzer. J. R., in "The Catalytic Chemistry of Nitrogen OxUes", R. L. Klimish and J. G. Larson, Ed., p 133, Plenum Press, New York, 1975. Koutsoukos, E. P., Blumenthal, J. L., Ghassemi, M., Bauerie, G., "Assessment of Cataksts for Control of NO, from Stationary Power Plants", Phase 1, Vol. 1, U S . Dept. of Commerce, PB 239745 Jan 1975. Lauder, A., DuPont Co., Wilmington, Del., personal communication, 1976. Markvart, M., Pour, V., J . Catal., 7, 279 (1967). McArthur, D. P., Union Oil Co.. &ea, Calif., personal communication, 1973. Meguerian, G. H., Rakowski, F. W., Hirschberg, E. H., Lang, C. R., Schock, D. N.. SAE Paper 720480, SAE National Engineering Meeting, Detroit, Mich., May 22-23, 1972. Michailova, E. A., Acta Physicochim. U.R.S.S., I O , 653 (1939). Nonnenmacher, H., Kame, K., U S . Patent 3 279 384 (1966). Otto, K., Shelef, M., Kummer, J. T., J . Phys. Chem., 74, 269 (1970). Pusateri, R. J.. Katzer, J. R., Manogue, W. H.. AIChE J . , 20, 219 (1974). Saleh, J. M., Trans. Faraday Soc., 64, 796 (1968). Saleh, J. M., Trans. Faraday Soc., 66, 242 (1970). Saleh, J. M., Trans. Faraday Soc., 67, 1830 (1971). Shelef, M., Kummer, J. T., Chem. Eng. hog. Symp. Ser., 67 (115), 74 (1971). Shelef. M., Cafal. Rev., 11, 1 (1975). Sullivan, D. R.. Katzer, J. R., Manogue, W. H.. to be submitted to Ind. Eng. Chem. Prod. Res. Dev. (1979). Taylor, K. C., Sinkevitch. R . M., Klimisch, R. L., J . Cafal., 35 (1974). Tsai. J., Agrawal. P. K., Foley, J. M., Katzer, J. R., Manogue, W. ti.,submitted to J . Cafal. (1979a). Tsai, J., Agrawal, P. K., Sullivan, D. R., Katzer, J. R., Manogue, W. H., submitted to J . Cafal. (1979b). Wu. S. C., Nobe, K., [nd. Eng. Chem. Prod. Res. Dev., 16, 136 (1977).
Received for review February 1, 1979 Accepted June 5, 1979
Nitrogen Removal from a Coal-Derived Liquid. 1. Effect of Catalyst Support Properties R. Sivasubramanian and B. L. Crynes' School of Chemical Engineering, Oklahoma State University, Stillwafer, Oklahoma 74074
Hydrodenitrogenation of raw anthracene oil, a coal-derived liquid, has been studied in a trickle bed reactor using Co-Mo-alumina catalysts at temperatures 340,371,and 399 OC (650, 700,and 750 O F ) at 1.03X lo7 Pa pressure (1500 psig) and at liquid volume hourly space times of 0.48,0.92,and 1.84 h. The physical properties of one of the supports were varied using a steam-treating technique, and the effects of support properties on hydrodenitrogenation were studied. The variation in total surface area of the supports has a greater effect on the denitrogenation activity of the catalysts than did variations in catalyst pore size under the conditions tested. Catalysts with a bimodal pore distribution did not show any advantage over catalysts with a monomodal pore distribution.
Introduction With ever-increasing demands on fossil fuels as a source of energy and decreasing supplies of petroleum and natural gas within the United States, this country is presently refocusing its attention on the conversion of coal into convenient gaseous and liquid forms of energy. Coal liquefaction is likely to be commercialized within the next decade. Coal-derived liquids contain significantly larger amounts of sulfur and nitrogen than petroleum feedstocks. These sulfur and nitrogen concentrations should be reduced for environmental reasons as well as to increase the 0019-7890/79/1218-0179$0 1.OO/O
quality of the product. Catalytic hydroprocessing is commonly used in the petroleum industry to reduce the concentration of heteroatoms and to increase the quality of the product. With increased interest in coal conversion processes, the potential need for catalysts for hydrotreating coal-derived liquids has also increased. In view of the time constraints, first generation hydrotreating catalysts for coal-derived liquids are likely to be based on the existing catalyst technology, which is largely an outgrowth of petroleum processes. Much effort has been given to catalyst development for hydrotreating of petroleum 0 1979 American
Chemical Society
180
Ind. Eng. Chem. Prod. Res. Dev., Vol. 18, No. 3, 1979
Table I. Raw Anthracene Oil Feed Properties carbon hydrogen sulfur nitrogen oxygena ash API gravity @ 60 "F initialb 1 0 vol % 30 50 70 90
90.65 5.76 0.48 1.06 2.05 nil -7 1 9 3 "C (380 O F ) 232 "C (450 O F ) 297 "C (570 OF) 343 "C (650 OF) 371 "C (700 O F ) 435 "C (815 "F)
a By difference. Normal boiling point determined from ASTM D1160 data.
Table 11. Effect of Steam Treatment o n Alumina Surmort property
orig. support materiala
steam-treated support
l h
10 h
*
surface area, 311 1 0 281 * 1 0 271 * 1 0 m'k pore volume, 0.647 * 0.05 0.674 f 0.05 0.686 i 0.05 cm3/g most frequent pore radius, A 33 t 1 37i 1 38 f 1 Ketjen 007-1-5E.
stocks, whereas relatively little has been directed toward those for coal liquids. However, the recent flurry of federal funding has spawned several university and industrial projects directed toward improved hydrotreating catalysts for coal-derived liquids. Problems and limitations may surface when the existing catalyst technology is applied to hydrotreating coal-derived liquids. To surmount these difficulties, second generation processes will require new and improved catalyst materials and concepts. This study is part of a larger program at the Oklahoma State University School of Chemical Engineering given to tailoring hydrotreating catalysts especially for coal-derived liquids. Hydrodenitrogenation of raw anthracene oil, a coal-derived liquid, has been studied in a trickle-bed reactor using several Co-Mo-alumina catalysts. The properties of the raw anthracene oil are shown in Table I. This oil was obtained from Reilly Tar and Chemical Corp. and served as a liquid analogous to a solvent fraction produced from a solvent refined coal process. This is a process stream that may require rehydrogenation and cleanup in some processes.
Experience in the petroleum industry reveals that the properties of the catalyst support material do affect performance of the hydrotreating catalysts (Livingston, 1973). The major objective of this paper is to present the effects of support properties of the hydrotreating catalysts on hydrodenitrogenation of raw anthracene oil. Catalysts Support Material Ideally, in order to determine the influence of catalyst support properties on hydrodenitrogenation one should have sets of catalysts exactly similar in every respect except for the support properties. Earlier studies (Satchell, 1974) used commercial catalysts and the chemical composition of the supports changed considerably between the different catalysts. However, in the present study, catalyst supports were obtained from commercial vendors and impregnated with active metals in our laboratory. Schlaffer et al. (1965) exposed alumina to steam a t moderate to high temperatures and found that there is a considerable decline in surface area accompanied by little or no loss of pore volume. Hence, one of the supports used in this study (Ketjen 007-1.5E) was subjected to steam treatment at about 538 "C (1000 OF). Two sets of commercial supports were steam treated, one for a period of 1 h and the other for a period of 10 h. Table I1 presents the properties of the supports before and after steam treatment. The table shows that some modest changes resulted in surface area and most frequent pore radius. Most frequent pore radius, as used in this study, is the pore radius corresponding to the distribution peak when dV/d(ln r ) is plotted against pore radius, where V is the cumulative pore volume in cm3/g and r is the pore radius in A. The surface areas were determined by the BET method and the pore volume and pore size determined by a mercury penetration method. These were made by commercial analytical laboratories. In order to estimate the reproducibility of the laboratories that performed these analyses, blind duplicate samples were sent on different occasions. The degree of reproducibility is shown in Table 11. In addition to the Ketjen 007-1.5E support that was steam treated (10 h), two other supports (Ketjen 000-3P and Conoco Catapal HP-20) were also used in this study. All supports were impregnated with oxides of cobalt and molybdenum in our laboratories. In order to estimate'the effects of impregnation on the support properties, impregnated samples were also sent for analyses. Table 111 presents the support properties before and after impregnation for the four supports used in this study. Impregnation does not seem to affect the micropore size,
on Alumina Supports Table 111. Effect of Impregnation - ._ support properties before impregnation
series CAT KEC KET KEP KER KDC KDT KDP a
Micropores.
support used Conoco Catapal HP-2 0 Ketjen 007-1.5E steam-treated (10 h ) Ketjen 007-1.5E Ketjen 000-3P Ketjen 007-1.5E Ketjen 007-1.5E steam treated (10 h ) Ketjen 007-1.5E Ketjen 000-3P Macropores.
surface area, m'k
pore vol., cm3/g
after impregnation
most frequent pore radius,
a
surface area m2/g 218
pore vol., cm3k 0.875
most frequent pore radius, A
40," 1900b
244
1.01
40:
31 1 271
0.647 0.686
33 38
291 230
0.503 0.540
31 38
254 311 311 271
0.834 0.647 0.647 0.686
34,a 1 5 0 b 33 33 38
199 243 252 221
0.625 0.452 0.467 0.499
36: 33 31 39
254
0.834
34,a 1 5 0 b
220
0.448
33,a 180b
llOOb
160b
Ind. Eng. Chem. Prod. Res. Dev., Vol. 18, No. 3, 1979 181 Table IV. Operating Conditions variable pressure, total temperature
H,
DETECTOR
ii
liquid volume hourly space time, cm3/cm3h SAMPLE BOMB 1
SYMBOLS TEMPERATURE INMCQTCX
e
TEMPERATURE CONTROLLER
PRESSURE CONTROLLER PRESSURE GAUGE
U
CHECK VALVE
@ VARIAC
SAMPLE BOMB 2
CAUSTIC SOLUTION
I c
MIXTURES ::
=N20R HZS-H2
PRODUCT
U VALVE
* RUPTURE DISK Figure 1. Schematic flow diagram of the experimental system.
whereas there is an increase in macropore size and the surface area decreases after impregnation. Yen et al. (1976) found the same results in their study using monolithic catalysts, The pore volume decreases after impregnation. These impregnated supports were then used as catalysts in a trickle-bed reactor to hydrotreat raw anthracene oil. A brief description of the experimental equipment and procedure follows.
Equipment A schematic diagram of the equipment is shown in Figure 1. The reactor used in this study was a 1.27 X m (0.5 in.) i.d. tubular packed bed operated in the trickle-flow regime. Typical catalyst bed lengths of 0.498 to 0.505 m (19.6 to 19.9 in.) were maintained and catalyst particles of 8-10 mesh were used. The catalyst bed was placed in the middle of the reactor and the two ends of the reactor were packed with inert material to serve as preand post-heat zones. Excellent temperature control and linear profiles were maintained by using massive aluminum heating blocks around the reactor. These blocks were grooved and wound with resistance heating wires. The measure of the reactor bed temperature was taken as that of the thermocouple which could be traversed along the axial thermowell from catalyst bed top to bottom. Thermocouples placed at corresponding tube and wall positions indicated that radial bed differences were less than 1.5 "C (3 O F ) . A typical temperature profile over the catalyst bed would vary by only f1.5 "C (3 O F ) at a temperature of 399 "C (750 O F ) . The feed oil was pumped and metered into the reactor with a Ruska positive displacement pump, and hydrogen flow was used from bottles on a once-through basis. The liquid and hydrogen streams were combined in a tee at the reactor top and passed cocurrently down the reactor bed. The flow rate of reactor effluent gases was monitored on a bubble meter. Disengaging and sampling bombs at the bottom of the reactor provided for gas-liquid separation and liquid isolation for sample removal. Operational Procedure After the catalyst was packed in the reactor, the catalyst bed temperature was raised slowly to 232 "C (450 O F ) . Prepurified nitrogen was allowed to flow through the reactor for a period of 12 h. A t the end of this calcining period, the catalyst was presulfided at 232 "C (450 O F ) with a 5.0 vol % H2S in hydrogen mixture for a period of 90 min. After presulfiding, the system was purged with nitrogen for about 10 min and the temperature of the catalyst bed was raised. When the temperature of the bed
range
1.03 x lo' Pa (1500 psig) 340 "C (650 O F ) 371 "C (700 OF) 399 "C (750 OF) 0.46 0.92 1.84
reached about 27 "C (50 O F ) below the operating temperature, both oil and hydrogen were allowed to flow through the reactor. The system was then brought to reaction conditions and was stabilized for a t least 48 h of oil-on-catalyst operation before sampling for data. Oil samples were taken at specified intervals, and the catalyst activity level was assessed periodically at a fixed set of operating conditions throughout any given reactor start-up. No significant activity decay was noted over a period of 130 h (continuous) on oil. A complete set of data was taken on each reactor start-up. Operational conditions are shown in Table IV for the results presented in this paper. The fluid dynamics of the trickle-bed reactors and their effects on nitrogen removal from raw anthracene oil have been presented by Satchel1 (1974) and Sivasubramanian (1977). Briefly, hydrogen flow, oil flux, particle size, back mixing, and channeling effects were shown to have little importance under the conditions tested. The effectiveness factor for denitrogenation of raw anthracene oil was shown to be one. Liquid product oils were first isolated in the sample bomb and purged for hydrogen sulfide removal before analyses. Nitrogen determinations were made on a Perkin-Elmer Model 240 elemental analyzer. Multiple analyses were always made on each sample. Effect of Support Properties on Catalyst Activity The catalyst support material provides a structural framework for the active metal components and increases the available surface area per unit weight of metal above that of the unsupported metal. In doing so, the properties of the support materials play an important part in the performance of the catalyst. Pore size distribution, which is a measure of the pore volume as a function of the pore radius, and the surface area available for reaction are two important properties of the catalyst supports. The availability of the internal surface area can depend upon the size of the openings since in a given catalyst preparation, the distribution of pore sizes may be such that some of the catalyst is completely inaccessible to large reactant molecules and, furthermore, may restrict the rate of conversion of the reactants by restricting the diffusion of reactants in the internal pore structure. Changes in pore structure are almost always accompanied by changes in surface area. Increasing pore sizes may offer an advantage with large reactant molecules since this may improve the efficiency of contact and increase the diffusion rate through the pores. Also, with ash containing feeds, larger macropores can offer an advantage in their ash tolerance while still maintaining activity. However, increasing pore sizes results in a decrease in surface area which may offset the advantages gained by the increase in pore sizes. Hence, the effect of support properties on catalyst activity is discussed in two sections, namely, the effects of pore size and pore size distribution and the effect of surface area. (a) Pore Size and Pore Size Distribution. Catalysts made from the Ketjen support 007-1.5E untreated with steam (KEC series) and those made from the same support with steam treating (10 h) provide a measure of comparison of pore size and distribution. Steam treating has shifted
182
Ind. Eng. Chem. Prod. Res. Dev., Vol. 18, No. 3, 1979
PRESSURE
TEMPERATURE = 3 4 0 C l 6 5 0 F ) PRESSURE = I 0 3 x IO’ PASCALS I1500 PSIG) OKEC SERIES KET SERIES
05 IO VOLUME HOURLY SPACE T I M E , HR
I5
e 2. Effect of steam treating at reactor temperature of 650 O F .
~
103 X IO’ PASCALS I1500 P S I G I
n
0
F
:
05 IO VOLUME HOURLY SPACE TIME, HR
15
Figure 5. Comparison of two bidispersed catalyst supports.
1.2
24
d 20-
FEED DOCTORED RAW ANTHRACENE OIL PRESSURE 1500 PSIG TEMPERATURE 371C (7OOF) OKOT SERIES A K D C SERIES U K D P SERIES
K 0
iz
0.4-
-
o’2-
L
W
a
T E M P E R A T U R E = 371 C 1700 F l P R E S S U R E = 1.03 x IO’ PASCALS ( 1 5 0 0 P S I G l
E
W
O K E C SERIES 0 K E T SERIES
I.o V O L U M E H O U R L Y SPACE T I M E , H R
0.5
15
4 04
VOLUME HOURLY SPACE TIME, HR
Figure 3. Effect of steam treating at reactor temperature of 700 O F .
Figure 6. Nitrogen response for quinoline doctored runs.
4
activity. Similar results were obtained when raw anthracene oil doctored with quinoline was used as a feedstock over the catalysts prepared from the support Ketjen 007-1.5E and the 10 h steam-treated Ketjen 007-1.5E. Addition of quinoline increased the nitrogen content of the feedstock from 1.06 wt 5% to 1.89 wt %. Figure 6 presents a comparison of the results obtained from these two series. As noted with undoctored raw anthracene oil feedstock, the steam-treated support used in the KDT series tends to be less active than the untreated support. In hydrodesulfurization studies, using raw anthracene oil and similar catalysts, Sooter (1974) found that increasing the pore radius tends to increase sulfur removal. However, Satchel1 (1974) found that increasing the pore radius from 25 to 33 A had no effect on nitrogen removal from raw anthracene oil. Van Zoonen and Douwes (1963) studied the effect of volume average pore radius with Co-Mo-alumina catalysts on the rate of denitrogenation of a Middle East gas oil. Their study showed that for two catalysts with similar surface areas (190 and 240 m2/g) but with different average pore radii (33 and 73 A), the average pore radius had a negligible effect on the rate of denitrogenation. However, for desulfurization, they found that pore sizes influence conversion. With both surface area and the pore radius changing, predicting the effects on nitrogen removal becomes difficult. However, from Figure 5 where two catalysts with different pore properties produce almost similar nitrogen removal and taking Satchell’s and Van Zoonen’s study into consideration, nitrogen removal from raw anthracene oil was relatively insensitive to pore radius changes in the range between 25 and 40 A. Pore size distribution, a measure of the pore volume as a function of the pore radius, is often neglected in the
TEMPERATURE 3 9 9 C (750F) PRESSURE = 1.03 I 0 3 x IO’ PASCALS I1500 PSI G )
!L 0 1.0
OKEC SERIES O K E T SERIES
g 08
I \
L t
a
\
021 04
4
05 IO VOLUME HOURLY SPACE TIME, H R
I5
Figure 4. Effect of steam treating at reactor temperature of 750 O F .
the most frequent pore radius from 33 to 38 A, only a change of 15% relative to the smaller size. The surface area of the steam treated catalyst was smaller than the base catalyst (230 vs. 291 m2/g). However, the two catalysts have the same chemical composition. Figures 2, 3, and 4 compare activity of the two catalysts for nitrogen removal at the three temperatures used in this study. The steam-treated catalyst was considerably less active than the nontreated catalyst at all temperature levels as measured by the total nitrogen remaining in the product oil. Figure 5 shows a comparison of two bidispersed catalysts having about the same surface area (199 and 218 m2/g) but with different pore radii, especially macropores (36 and 160 8,vs. 40 and 1900 A). As can be seen from the figure, the difference in their activity was very small, with the higher surface area catalyst showing a slightly greater
Ind. Eng. Chem. Prod. Res. Dev., Vol. 18, No. 3, 1979 12
Table V. Pore Properties of Catalysts Used in this Study run series
support used
Df,A
KEC Ketjen 007-1.5E KET steam-treated ( 1 0 h ) Ketjen 007-1.5E KEP Ketjen 000-3P KER Ketjen 007-1.5E KDC Ketjen 007-1.5E KDT steam-treated (10 h ) Ketjen 007-1.5E KDP Ketjen 000-3P CAT Conoco Catapal HP-20
62 76
ADD A
183
PRESSURE = I 0 3 X IO' PASCALS ( 1 5 0 0 P S I G )
PD
18 6.9 30 1 7 . 3
72,b 32OC 44a 22.8 20 8.7 66 62 16 6.2 78 28 17.0 66,b 36OC 52" 22.7 380OC 44" 28.2
w n
-
Most frequent pore diameter spread for micropores. Micropores. Macropores.
discussion of the effect of catalyst support properties on hydrotreating. One of the reasons may be that it is too complex. Simple parameters such as most frequent pore radius cannot completely characterize a pore distribution curve. Anderson et al. (1959) presented in a patent some arbitrary methods of characterizing pore size distributions to quantify their effects on hydrotreating based on results obtained from hydrodesulfurization of gas oil feedstocks. Because of the absence of standard methods for characterizing pore size distribution effects, the methods recommended by Anderson et al. (1959) will be illustrated for discussion. Briefly they recommend a most frequent pore diameter (Of)of above 60 8, and a spread of the most frequent pore diameter (AD,) of at least 10 8, for hydrodesulfurization. Also a pore distribution factor (PD), defined as PD =
X
AD,
104
(where Df is the most frequent pore diameter, A, and AD, is the range of the more frequent diameters, 8,) of at least 5.0 is preferred by Anderson et al. Table V lists these variables for the catalysts used in this study. As can be seen from Table V, all the catalysts used in this study met the specifications presented in the patent. Sooter (1974), in his study on hydrodesulfurization of raw anthracene oil, found that two of his catalysts that were short of the necessary minimum AD, and P D values described in the patent performed poorly compared to a catalyst that did meet the specifications. However, Satchel1 (1974), using the same catalysts, found no such effects for denitrogenation. In the present study, catalysts having a bimodal pore distribution did not offer any obvious advantages over catalysts having a unimodal pore distribution. Two of the catalyst supports used in this study had bimodal pore distribution (CAT, KEP, and KDP series). CAT and KEP series used raw anthracene oil as feedstock while KDP series used doctored raw anthracene oil as feedstock. Both CAT and K E P catalysts were less active than the monodispersed catalyst used in the KEC series a t all temperatures. For example, at 371 "C (700 O F ) and at a space time of 1.84 h, the percent nitrogen removal for the KEC, KEP, and CAT catalysts were 77, 58, and 63%, respectively. When raw anthracene oil doctored with quinoline was used as the feedstock the bidispersed catalyst used in the KDP series was comparable to the monodispersed catalyst used in the KDC series, but not more active. Hence, pore size distribution does not seem to affect nitrogen removal from raw anthracene oil over the range of parameters tested. However, it is possible that for nitrogen removal, the minimum values of the parameters used (AD, and PD) may
be different, possibly lower than those employed for sulfur removal. Also with heavier feedstocks, pore size distribution may begin to have an effect on nitrogen removal. Further studies should be made using heavier feedstocks and with catalysts having smaller pore sizes and narrow pore size distributions to find out whether there exists a minimum value for these values, above which pore size and pore size distribution cease to have an effect on nitrogen removal. (b) Surface Area. The total and active surface area available for reaction may be an important parameter to be considered when comparing catalyst supports. Catalysts made from Ketjen 007-1.5E, untreated with steam (KEC series) and those made from the same support with 10 h steam treating (KET series) provide a measure of the effect of the change in surface area on nitrogen removal. Steam treating the support reduces the BET surface area while increasing the most frequent pore radius. The catalyst used in the KET series had a surface area of 230 m2/g, while the catalyst used in the KEC series had a surface area of 291 m2/g; a 27% change relative to the lower value. Referring back to Figures 2, 3, and 4, if pore radius had little or no effect on nitrogen removal, then the lower surface area of the catalyst used in the KET series could account for the lower activity of the catalyst. Similar results were obtained when raw anthracene oil doctored with quinoline was used as a feedstock over the catalysts prepared from the supports Ketjen 007-1.5E untreated with steam (KDC series) and 10 h steam treated Ketjen 007-1.5E (KDT series). The KDC catalyst had a surface area of 252 m2/g, while the catalyst used in the KDT series had a surface area of 220 m2/g. Figure 6 presents a comparison of the results obtained from these two series. As noted with undoctored raw anthracene oil feedstock, the steam-treated support used in the KDT series tends to be less active than the untreated support. Another possibility for different catalyst performances is that the intrinsic activity of the catalysts is different. However, this is not likely because a more conclusive proof of the effect of surface area on nitrogen removal can be found by examining Figure 7 , which compares the two series KEC and KER. The catalysts used for the KEC and KER series were prepared from the same support (Ketjen 007-1.5E) and were designed to be replicates. However, the surface areas of the two catalysts were different (291 vs. 243 m2/g). This large difference in the surface areas of the two catalysts brings up the question of the ability of catalyst property duplication since the catalysts were prepared in our laboratories. Each series of catalysts was prepared independently and Table VI presents a comparison of catalysts property duplication for the supports used in this study. Table VI reveals that except for KEC
184
Ind. Eng. Chem. Prod. Res. Dev., Vol.
18, No. 3, 1979
Table VI. Catalyst Property Duplication surface pore run area, volume series m2/g cm3/g
most frequent pore radius, A
Ketjen 007-1.5E
KEC KER KDC
291 243 252
0.503 0.452 0.467
31 33 31
steam-treated Ketjen 007-1.5E
KET KDT
230 221
0.540 0.499
38 38
Ketjen 000-3P
KEP KDP
200 219
0.625 0.448
36,a 160* 33," 180b
support
a
Micropores.
Macropores.
and KER series, the surface area changes for the others were within the precision of measurement of the analytical laboratories (*lo m2/g). At this point, no explanation can be offered for the poor reproducibility of the surface areas for the catalysts used in the KEC and KER series. However, since these two catalysts have identical support properties except for total surface area, a comparison of the denitrogenation activities of these two catalysts provides an excellent measure of the effect of the changes in surface area on nitrogen removal. The properties of these two catalysts can be found in Table VI. The changes in pore volume and most frequent pore radius are negligible and within the precision of measurement of these variables. The changes in surface area fall outside the range of precision in measuring the surface area (*lo m2/g). Figure 7 reveals that the catalyst used in the KER series was consistently less active than the catalyst used in KEC series a t all three temperature levels. The chemical composition of the two catalysts was similar in every respect. However, the surface areas were different; this reduction of surface area resulted in lower conversion of organonitrogen species. If the conclusion that average pore radius has a negligible effect on nitrogen removal is valid, then Figure 5 offers further evidence that reduction in nitrogen removal follows reduction in surface area. The figure shows that the catalyst used in the CAT series with a surface area of 218 m2/g was more active than the catalyst used in the KEP series which had a surface area of 200 m2/g. While the activity differences are small, the greater activity is consistently favored by the greater surface area. Reaction rate constants based on unit surface area were calculated in order to obtain further confirmation of the surface area effects on denitrogenation. The reaction rate constants were calculated using the relation CAO k,S In - = -(LHSV)-" CAf
vr
where CAois the initial nitrogen content of liquid product, weight percent, CAfis the final nitrogen content of liquid product, weight percent, k, is the first-order reaction rate constant based on unit surface area, cm/(h)", LHSV is the liquid hourly space velocity, (cm3/(h)(cm3of catalyst), x is an empirical parameter, V , is the volume of the reactor, cm3, and S is the total surface area, m2. This model was found to fit the experimental data more closely than other
simple nth-order models tested (Sivasubramanian, 1977). The model has been suggested by Mears (1974) and accounts for catalyst particle wetting in a trickle bed. Table VI1 presents the k, values for all the catalysts used in this study. In order to compare the k, values between different catalysts, an uncertainty analysis was made and the uncertainties are presented in the table along with the k, values. If denitrogenation is sensitive to total available surface area and if the intrinsic activities are the same and if changes in pore sizes are unimportant, the k, values should remain the same. The catalyst used in the KEC and KET series were essentially the same except for changes in physical properties of the supports caused by steam treating and the catalyst used in the KER series was a replicate of the KEC series. A comparison of the k, values for the KEC, KET, and KET series reveals that k, values do remain constant within experimental precision. These observations c o n f m that the lower activity of the catalysts used in the KET series can be attributed to the lower surface area of the KER catalyst (230 m2/g) as compared to the KEC catalyst (290 m2/g). A similar observation is true when KEC and KER series are compared. The k, values remain within experimental uncertainty, for the KDC and KDT series which used doctored raw anthracene oil as feedstock. The catalysts used in these two series were the same except for the physical properties of the support. Hence, nitrogen removal from the particular coal liquid seems to be more sensitive to changes in surface area of the catalyst support materials than changes in pore radius over the range of variables tested and for these catalysts. The k, values for the other two supports used in this study, Ketjen 000-3P and Conoco Catapal HP-20, are slightly higher than the k, values for the catalysts used in the KEC, KET, and KER series, as can be seen from Table VII. These two supports had lower surface areas (190 and 218 m2/g) compared to the KEC (290 m2/g) or the KET (230 m2/g) catalysts. Still the activities of these two catalysts seem to be higher on a per unit surface area basis. Both these catalysts had a bimodal pore distribution. However, earlier it was shown that bidispersed catalysts did not offer any advantage over monodispersed catalysts for the coal liquid used. Results obtained from distilling the product oil into several fractions and analyzing the individual fractions for nitrogen removal also confirmed the above conclusion (Sivasubramanian, 1977). Hence, one had to conclude that the intrinsic activities of these two catalysts KEP and CAT had to be higher than the other two catalysts KEC and KET. In addition to bimodal dispersion, the other thing that these two catalysts had in common was the presence of silica, which is known to increase the cracking activity. However, the CAT series catalyst had very little silica (0.0088wt % ), where as the KEP catalyst had nearly 1% silica. The CAT series catalyst support material was prepared by a special method by contacting an aqueous alumina slurry with an effective amount of an organic solvent (Leach, 1971). This special method of preparation could have contributed to the CAT catalyst's higher intrinsic activity. However, no information is available on the method of preparation of the Ketjen supports used in
Table VII. Comparison of k, Values temp, "C (OF) 340 ( 6 5 0 ) 371 ( 7 0 0 ) 399 ( 7 5 0 )
k, x
___
KEC series KET series 3.41 t 0.28 3.52 i 0.36 5.09 t 0.41 4.93 f 0.47 7.83 5 0.67 7.72 i 0.68
KER series 3.06 5.11 8.41
t i f
0.29 0.43 0.72
lo',
KEP series
cm/(h)x
CAT series
KDC series
KDT series
___
KDP series
3.71 i 0.39 3.71 i 0.39 _-___ 6.17 f 0.58 6.26 f 0.57 4.17 i 0.30 3.48 i 0.28 5.06 i 0.40 --_-_-10.68 i 0.99
___
Ind. Eng. Chem. Prod. Res. Dev., Vol. 18, No. 3, 1979
all other series. The presence of silica in the KEP catalyst could have caused the higher intrinsic activity of that catalyst. Another question that arises is the effect of active metal concentration per unit surface area. Since active metal concentration is measured in terms of weight percents, the lower surface area catalyst would have a higher active metal surface concentration if the same weight percent of active metals are employed. However, the changes in active metal concentration per surface area changes are relatively small (0.00012 g of Mo03/m2for the KEC series vs. 0.00015 g of Mo03/m2 for the KET series) and published information on the optimum concentration of active metals ranges from 2 to 4 % Co as COOand 8 to 15% Mo as Moo3. Also, increasing concentration of active metals beyond these optimum values has no significance in increasing the hydrotreating activity of catalyst (Beuther et al., 1959). Hence, the changes in active metal surface concentration likely had little effect on nitrogen removal in this study. Conclusions (1)The rate of denitrogenation of raw anthracene oil, a coal liquid, is sensitive to changes in catalyst surface area available for reaction. A reduction in surface area results in a reduction of nitrogen removal. (2) The rate of denitrogenation of raw anthracene oil is not sensitive to changes in the most frequent pore radius in the range of 25 to 40 A for the Co-Mo-alumina catalyst. (3) Changes in pore size distribution do not seem to affect nitrogen removal from raw anthracene oil. Bimodal pore distribution does not seem to offer any obvious advantages over monodispersed catalysts for removal of nitrogen from raw anthracene oil (containing essentially no ash). (4) Steam-treating alumina supports considerably reduces the surface area while slightly increasing the most frequent pore radius.
185
(5) Impregnating the alumina supports with active metals results in moderate changes in pore properties. The surface area and pore volume decreases after impregnation, but impregnation does not seem to change the most frequent micropore radius, whereas there is an increase in the most frequent macropore radius. (6) These results should be assessed for other coal derived liquids. One might expect that as the feedstock becomes more difficult to hydrotreat, the increasing pore sizes should begin to offer an advantage. Also, with ash containing feeds, the macropore system could offer an advantage in its ash tolerance while still maintaining activity. These factors are being assessed in our laboratories. Acknowledgment This work was sponsored by funds made available from US.Department of Energy-Fossil Energy and from the School of Chemical Engineering, Oklahoma State University. Appreciation is expressed for these funds and for assistance from D. C. Mehta, Mushtaq Ahmed, Kerry Scott, Anthony Jones, Paul Madison, and Susan Phillips. Literature Cited Anderson, J. A., Jr., Dinwiddie, J. A,, Moseman, M. A,, Vernon, L. W., US. Patent
2890 182 (1959). Beuther, H., Fiinn, R. A., McKinley, J. B., Ind. Eng. Chem., 51, 1349 (1959). Leach, B. E., US. Patent 3 773 691 (1971). Livingston, J. Y., "Hydrotreating Catalyst Properties Do Affect Performance", paper presented at the AChE 74ih National Meeting, New Orleans, La., Mar
11-15, 1973. Mears, D. E., Adv. Chem. Ser., No. 133, 218 (1974). Satchell, D., Ph.D. Thesis, Okhhoma State University, Stillwater, Okla., 1974. Schhffer, S. G.,Adams, C. R., Wilson, J. N., J. Phys. Chem., 69, 1530 (1965). Sivasubramankn, R., Ph.D. Thesis, Oklahoma State University, Stillwater, Okla.,
1977. Swter, M. C., Ph.D. Thesis, Okhhoma State University, Stillwater, Okla., 1974. Van Zoonen, D., Douwes, J. Inst. Pet., 49,385 (1963). Yen, Y. K., Furiani, D. E., Welier, S. W., Ind. Eng. Chem., Prod. Res. D e v . , 15, 24 (1976).
Received for review July 10, 1978 Accepted June 1, 1979
