Catalytic hydrodenitrogenation of quinoline in a ... - ACS Publications

Jan 1, 1984 - Peter A. Fox, Michael A. Bruck, Steven D. Gray, Nadine E. Gruhn, Carina ... Steven D. Gray , Keith J. Weller , Michael A. Bruck , Paula ...
1 downloads 0 Views 703KB Size
20

Ind. Eng. Chem. Process Des. Dev. 1984, 2 3 , 20-25

Catalytic Hydrodenitragenation of Quinoline in a Trickle-Bed Reactor. Effect of Hydrogen Sulfide Shan Hsl Yangt and Charles N. Satterfleld" Department of Chemical Engineering, Massachusetts Institute of Technoicgy, Cambridge, Massachusetts 02 139

Studies at 350,375,and 390 O C and 6.9 MPa showed that the presence of H,S (generated in situ from CS,) in the overall reaction network somewhat inhibits hydrogenation and dehydrogenation reactions but markedly accelerates hydrogenolysis reactions, for a net increase in the overall rate of hydrodenitrogenation (HDN). These effects are similar to those observed previously in vapor-phase reactions. H,S has little effect on the activation energies for the hydrogenation and dehydrogenation reactions, but it significantly reduces those for the hydrogedysis reactions. Certain plugging problems encountered upon reacting 5,6,7,84etrahydroquinoline are attributed to formation of a trimer imine as an intermediate. The presence of Tetralin in the liquid inhibits the overall hydrodenitrogenation reaction, indicating that inhibition by competitive adsorption is more important than the accelerating hydrogen-donor capability that Tetralin exhibits in some homogeneous reactions.

In an associated paper (Satterfield and Yang, 1984) we have reported on studies of quinoline HDN in a trickle bed reactor using an NiMo/A120, catalyst which was presulfided but in which no sulfur compound was added to liquid or gas during reaction. In previous vapor-phase studies we have shown that the rate of catalytic hydrodenitrogenation (HDN) of pyridine or quinoline is increased in the presence of H2S (Satterfield and Giiltekin, 1981) and we have interpreted this in terms of the comparative effect of H2S on the hydrogenation vs. the hydrogenolysis reactions that occur in the overall HDN reaction. In the present study we have analyzed the effects of the presence of H2Son the reaction in the liquid phase, in an analogous fashion to our earlier vapor-phase studies. Experimental Section The studies reported on here were carried out in the same trickle-bed apparatus, using samples of the same catalyst and procedures previously described, except that CS2was added to the feed in various concentrations. CS2 rapidly reads with hydrogen to generate HzSin situ. Most of the data here were taken with catalyst charge no. 3. Catalyst charge no. 2 was used here primarily for some comparisons with no. 3. These same two catalyst charges were also used to generate the kinetic data in the absence of H2S, published in the accompanying paper. All the data reported here were obtained after the catalyst charges had reached steady-state activity. Catalyst charge no. 1 was used for sdme preliminary studies and charge no. 4 was used to study the effect of a change in the activation procedure. The time required for steady state to be reached was less, the higher the quinoline concentration in the feed and the lower the space-time (faster flow rates) used. As characterized by a standardized run (see below), steady state was essentially reached in about 150 h. The acronyms for the various compounds present are the same as those used previously and are given in Figure 2 of our accompanying paper (Satterfield and Yang, 1984). Results Reproducibility. Figure 1 shows the activity history of four samples of the catalyst, used for the present study and the previous study in the absence of added CS2. A 'Amoco Production Co., Tulsa, OK 74136.

standardized test was applied from time to time, using 5 wt % quinoline in the liquid feed without CS2addition at a space-time of 269 h g of cat./mol. of Q and at 6.9 MPa and 375 "C. With catalyst charges no. 3 and 4 twice the quantity of catalyst (1.6 g) was used than in charges 1and 2 (0.8 g), and the liquid and gas flow rates were correspondingly doubled to keep space-time and gas/liquid mole ratio constant. Studies with charges no. 2 and 3 showed identical behavior in the absence of CS2. Doubling of flow rate, keeping space-time and gas/liquid ratio constant, was without effect, which is consistent with our previously stated conclusion that the reactor operated with good contacting. The percent HDN for charge no. 1for the first 330 h are for 1 wt % quinoline and at a space-time of 693 h g of cat./mol. of Q. At 330 h this charge was subjected to the standard reaction conditions and it reached the same steady-state activity shown by charges 2 and 3 within a further 100 h. The designations 0, 000, and 00 indicate the approximate times at which data were taken on catalysts 2 and 3 in the absence of CS2 and 1,2, and 3 when data were taken on catalyst 3 in the presence of CS2. Activation Procedure. The oxide catalyst was first activated under a flow of 10% H$ and 90% H2following a specified heating procedure described by Satterfield and Yang (1984). Subsequently, the catalyst was resulfided after each 10-15 h run as a precaution, although we did not establish that this was necessarily required. At each resulfiding the catalyst was also flushed with a mixture of inert carrier and xylene for 1 / 2 h to help dissolve any possible accumulated high molecular weight material. Routinely a total pressure of 0.24 MPa was used in the initial sulfiding and resulfiding conditions. One batch of catalyst (charge no. 4) was initially sulfided and resulfided at a total pressure of 0.33 MPa of 10% H2S and 90% H2,in contrast to the usual pressure of 0.24 MPa. A higher level of catalytic activity was obtained as shown in Figure 1. The effects of changes in presulfidation procedures on subsequent catalytic activity are discussed elsewhere (Yang and Satterfield, 1983). Effect of CSz Concentration. Figure 2 shows that addition of a small amount of CS2 to the quinoline markedly increases the percent HDN under representative reaction conditions, but the percent HDN reaches a broad maximum plateau value at about 1.5 wt % CS2in the feed.

0196-4305/84/1123-0020$01.50/00 1983 American Chemical Society

Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 1, 1984

MOLE

RATIO H z S / H z

0.59 1.47 r l I

201

~ A uI O ); ~IOWX11(2) ; l ” ; ( 3Q) 1001 l

O

Figure 1. Activity history of four catalyst charges: 5 wt % Q,269 (h g of cat.)/mol of Q, 6.9 MPa, 375 “ C , no CS2 added.

z 9

MOLE

1

I

j

100 200 300 4 0 0 500 600 TIME, HOURS

-

XlOO

5.88

I

!

P

21

RATIO, H2S / H 2 x l O O

0

1 ’

i 40tr-----!

2

3

4

5

6

IN F E E D

WT % C S 2

Figure 3. H2S enhances ring isomerization of propylcyclohexene. .-

I

d

I

I

I

I

6 9 MPa

1.6 8

037 0.42

1

2 6 9 h r g cat /mol Q

-

$

40

450 O-PROPYiAluILI:.IE IOPA !

500

600

550

TIME OF CATALYST HOURS

ON

STREAM

Figure 4. The effect of CS2 on % HDN is reversible.

0

1

WT

2

3

% CS2

4 5 6 IN F E E D

Figure 2. Increase in CS2 concentration in feed above 1.5 w t % shifts intermediate product distribution markedly, but with little effect on % HDN; 6.9 MPa, 375 O C ; charge no. 3. 269 (h g of cat.)/mol of Q.

60

501 0

vOpo PPCSO aa0Ctor 1500hrg c o t /mol Q I

1-

:n T r i c k l o

However, the product distribution continues to be affected, largely with respect to OPA and DHQ. As CSz is added to the feed, the amount of DHQ produced markedly decreases while the amount of OPA increases. Apparently two opposite effects are occurring that result in about the same overall degree of HDN. It is known from our previous vapor-phase study (Satterfield and Giiltekin, 1981) that hydrogen sulfide, if present during hydrodenitrogenation, enhances hydrogenolysis reactions significantly and inhibits hydrogenation reactions to some degree. This suggests that hydrogen sulfide increases the rate of DHQ hydrogenolysis to form propylcyclohexylamine, which is readily converted to hydrocarbons and ammonia, and increases the hydrogenolysis of PyTHQ to form OPA. However, the denitrogenation of OPA is rate-controlled by the hydrogenation of its aromatic ring which is inhibited by the presence of hydrogen sulfide. As mentioned in the associated paper, the amount of methylpropylcyclopentane (MPCP) isolated in the quinoline HDN reaction is almost undetectable in the absence of CS2 in the feed. However, the ratio of MPCP/PCH is markedly increased upon addition of CS2 to the feed, as shown in Figure 3. The formation of MPCP is through the intermediate ring isomerization of propylcyclohexene, which is enhanced by the acidity produced by adsorptive dissociation of H2S, as discussed below. The low ratio of

I

40 .’

Sed

Reoctor

1 4 9 4 h r j c o t I mol 0

!

10

0

0.4

0.8

1.2

MOLE RATIO, H 2 S / t i 2

16

xlOO

Figure 5. The enhancement effects of H2Sare very similar in liquidand vapor-phase reaction.

MPCP/PCH when CS2was provided at the extremely high concentration of 5.89 wt ?& was caused by a polymerization side reaction (see later). The change in catalyst activity by the presence or absence of CS2 during reaction is completely reversible, as shown in Figure 4. The dashed line reproduces the results of the standardized test in the absence of CS2on charges no. 2 and 3 as shown in Figure 1. With both catalysts the addition of CS2 to the feed causes an increase in the percent HDN, but the conversion drops to its former value when CS2 is absent. The effect of CS2 should be primarily associated with the partial pressure of H2Spresent or the mole ratio of H2S to Hp If it is assumed that all CS2is converted to H2Sand H2 consumption is neglected, the data points for charge no. 3 in Figure 2 for 0.59, 1.47 or 5.89 wt 70CS2in the feed correspond to mole ratios of H2Sto H2 of 0.0017,0.0042, and 0.0168, respectively. In Figure 5 data for charge no. 3 for these three CS2concentrations, obtained at a longer

22

Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 1, 1984

Table I. Rate Constants of Quinoline Reaction Network (mol/g of cat. h ) (1)5 w t % Q + 0.59 wt % CS,

(0) 5 w t % Q

(2) 5 wt 7% Q + 1.47 wt % CS,

(3) 5 wt % Q + 5.89 w t % CS,

350°C

375°C

39OoC

350°C

375°C

390°C

350°C

375°C

390°C

350°C

375°C

0.00025 0.0038 0.0088 0.0026 0.013 0.000 7 5 0.00045 0.165 0.0076 0.00013 0.00005

0.0009 0.0070 0.016 0.0040 0.016 0.0030 0.001 5 0.135 0.018 0.0007 0.0003

0.0019 0.01 1 0.024 0.0050 0.01 5 0.0065 0.0040 0.12 0.027 0.00 20 0.0008

0.0008 0.0037 0.0090 0.0025 0.013 0.001 0.0013 0.165 0.0080 0.00013 0.00005

0.0019 0.0071 0.016 0.0040 0.016 0.0032 0.0035 0.135 0.01 8 0.0007 0.0003

0.0028 0.011 0.040 0.0045 0.01 8 0.0080 0.0055 0.100 0.025 0.0020 0.0008

0.0010 0.0030 0.0080 0.0020 0.01 5 0.0010 0.0016 0.165 0.0078 0.0001 3 0.00005

0.0025 0.0066 0.016 0.0035 0.016 0.0035 0.0050 0.140 0.017 0.0007 0.0003

0.0040 0.0096 0.033 0.0037 0.018 0.011 0.0040 0.100 0.024 0.00 20 0.0008

0.00085 0.0020 0.0070 0.0014 0.0080 0.0 0045 0.0028 0.17 0.0065 0.0 0013 0.0 0005

0.0040 0.0031 0.014 0.0020 0.013 0.0018 0.0095 0.12 0.01 5 0.0007 0.0003

MOLE R A T I O , H2S / H2 X l O O 0.17 0.42

0,017

z

J

i.

0

'0°r- - -7

375'c

80-

0 1 2 3 M.T % CS2 I N

4

5

5

FEED

Figure 7. CS2 reduces hydrogenation and dehydrogenation rate constants; 6.9 MPa, 375 "C.

U

100 MOLE RATIO, H2S / H 2 x 1 0 0

Data

80

I

PyTHQ

50

40

oory

-

20

4 0 r0, 0 4 Y 5 4 r

0 0

100

SPACE

300 500 700 T I M E , h r g - c a t /mol Q f a a d

Figure 6. Comparison of simulated product distribution with experimental data; feed: 5 wt 70 quinoline plus 1.49 wt % CS,; 6.9 MPa.

z

-

0002

c

2

0

1 WT

contact time, are presented in the mole ratio form and compared to our previous studies in the vapor phase at the same temperature, pressure, and contact time. The amount of reaction and the enhancement effect of the partial pressure of H2Son HDN is seen to be very similar in the two phases. Kinetic Analysis. The detailed kinetics of the quinoline HDN network in the presence of H2Swere studied with four sets of data, in which the CS2present in the feed was zero, 0.59 wt % , 1.47 wt % , or 5.89 wt %. These are symbolized as sets (O), (l), (2), and (3), respectively. Set (0) is taken from Satterfield and Yang (1984). The data were all taken with charge no. 3 at the times shown in Figure 1. With each set, the feed consisted of 5.0 wt 70 quinoline, 5.2 wt % BzTHQ or 4.7 w t % o-ethylaniline (OEA), which correspond to the same molar concentrations. Studies were at 6.9 MPa, at a gas-to-liquid ratio equivalent to 9OOO SCF H,/bbl of oil, and at 350,375, and 390 "C except for set 3. For set 3 data were taken at 375 "C, and partially at 350 "C, but the run then had to be discontinued because of

"/a

I

I

I

l

i

2

3

4

5

5

CS2

IN

FEED

Figure 8. CS2 markedly increases hydrogenolysis rate constants.

reactor plugging, discussed below. The reaction network and method of analysis are the same as those used in our accompanying paper, which treats studies in the absence of H2S. Detailed results are given in the thesis by Yang (1982). Table I summarizes the calculated values of the rate constants. The same values of KSA/KAA and KNHJ K U , namely 2 and 0.7, found in the absence of CS2,also provided the best fit for the present data. Figure 6 compares experimental data from set 2 with simulated product distributions calculated from the model rate constants by the Runge-Kutta method. Additional comparisons are given by Yang (1982). The comparison for set 3 data (not shown) was slightly less satisfactory, probably because the data are less accurate due to the plugging problem. The rate constants at 375 "C are plotted against the amount of CS2 in the feed in Figure 7 for the significant hydrogenation and dehydrogenation reactions and in Figure 8 for the hydrogenolysis reactions. Values of k8 and

Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 1, 1984

MOLE R A T I O , H z S / H Z X l O O

~

,@-

017

3.2

I

W

01.a T

60 -

1.68

0.42

23

I

5040-

-

'

--

R -

0

1 WT

2 %

3 CS2

4

5

6

IN FEED

Figure 9. The ratio of a hydrogenolysis rate constant to a hydrogenation rate constant is about proportional to CS, concentration. Table 11. Activation Energies, kJ/mol. Data at 6.9 MPa ~

QUINOLINE HDN CONVERSION , %

liquid phase vapor phase

reaction

no CS, added

1 2 3 4 5 6 7

174 91 85 56 18 185 187

0.59 wt % PPQ = 13.3 H a , CS, PP,,s = 13.3 kPa 107 93 85 52 28 178 123

112 104 65 73 33 176 127

k9 are not shown, since they are relatively inaccurate although their ratio could be accurately determined. The constants klo and kll, for the reverse dehydrogenation reactions, are likewise omitted because their accuracy was considerably less than that of the other constants. It is seen that when CS2is present in small amounts,e.g., 0.59 wt % or 1.49 wt %, it has little inhibiting effect on the hydrogenation reactions, but with 5.89 wt CS2 in the feed, hydrogenation reactions are significantly inhibited. The hydrogenolysis reactions are greatly enhanced with an increasing amount of CS2 in the feed. Figure 9 shows that the ratio of a hydrogenolysis rate constant to a hydrogenation rate constant is about proportional to CS2 concentration. Table I1 lists activation energies for the rate constants 1 through 7 in the presence and absence of CS2. The latter are the values determined from data set (0). These were calculated from Arrhenius plots, which showed good straight lines. Also shown are activation energies from our previous vapor-phase studies in the presence of H2S (Satterfield and Gtiltekin, 1981)with the data re-analyzed in terms of our modified model. Their accuracy is somewhat limited by the moderate temperature range that could be utilized, but it is evident that the activation energies for the hydrogenation reactions (2through 6) are not markedly affected by the presence of CS2, but those for the hydrogenolysis reactions (1and 7) are significantly reduced. The values for vapor-phase reaction are comparable to those for the liquid phase. The values of the hydrogenation rate constants relative to one another are not markedly affected by changes in H2S content or temperature, but there is a strong effect on the ratio of a hydrogenolysis rate constant to a hydrogenation rate constant. Increased temperature and increased CS2 have opposite effects on such a ratio. A t 375 "C in the absence of H2S the slowest reaction is the hydrogenolysis of PyTHQ to OPA, kl. With 5.89 wt % CS2 in the feed kl is substantially increased and hydrogenation reactions are reduced, such as the two forward reactions given by k2 and k4. Thus the amount of OPA

Figure 10. Degree of HDN does not affect ratio of saturated to unsaturated hydrocarbon product, 6.9 MPa, 375 " C : Dashed lines, with CS,; solid lines, without CS2.

present increases and that of DHQ decreases. The presence of CS2slightly increases the concentration of PCHE and decreases that of PB but has no noticeable effect on the ratio of saturated to unsaturated hydrocarbons in the product, as shown in Figure 10. In all cases propylcyclohexane is dominant. Plugging with 5,6,7,8-Tetrahydroquinoline (BzTHQ).In some of our studies with BzTHQ we found that the reactor pressure drop gradually increased and finally the reactor became completely plugged. This occurred only in the presence of CS2and at short space-times, for example with a solution containing 5.16 wt % BzTHQ and 1.47 wt % CS2 being fed to the reactor operating at 6.9MPa, 375 "C,and 37 or 47 g of cat. h/mol of BzTHQ. The plugging first occurred on charge no. 2, after it had been on stream for 650 h with steady activity, with feeds of Q and CS2. Upon switching to BzTHQ and CS2 feed, the catalyst lost activity within a few days. With charge no. 3,the same plugging phenomenon occurred again. During this time a mass balance on heterocyclic N compounds in and out of the reactor showed a loss of about 10%. Flushing the reactor with a solvent consisting of 50% xylene and 50% liquid carrier under hydrogen flow eliminated the plug, but simultaneously a significant amount of DHQ was detected in the exit solvent. After flushing, the reactor pressure drop reduced to its normal value, and the catalyst activity was completely restored. It appears that this plugging is caused by some kind of polymerization related to the conversion of BzTHQ to DHQ. The compound A'-piperidiene, an imine, does not exist solely in a monomeric form (Schopf et al., 1948),but it forms two trimers that are geometric isomers, termed aand @-tripiperidienes,which exist in equilibrium with the monomer.

n A similar reaction could occur from the compound 3,4,5,6,7,8,9,10-octahydroquinoline, which is a probable intermediate when BzTHQ is hydrogenated to DHQ.

24

Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 1, 1984

Plugging could well be caused by the gradual accumulation of a solid trimer of this type, which could depolymerize and dissolve in the presence of a solvent. The plugging problem occurred only with BzTHQ as a feed, with a large amount of CS2 (5.89 wt %) present and under high flow operation. With the high CS2 content, hydrogenation is significantly hampered, which could allow the formation of a greater amount of the octahydroquinoline than normal and hence increase the degree of formation of the trimer by the polymerization reaction instead of continuing hydrogenation to DHQ. A stronger acidic form of Bronsted site on the catalyst (see later) could also enhance this polymerization reaction. Upon flushing the reactor with solvent under hydrogen flow, the polymerization reaction is reversed and the hydrogen further hydrogenates octahydroquinoline to DHQ. In the five-membered ring series, Luttringhaus et al. (1959) also found that A'-pyrroline forms a trimer, tripyrroline. The trimer formed on a catalyst surface could be a coke precursor, and the polymerization of an imine in general may be one of the problems encountered when heavy oil or coal liquids are processed. Effect of a Hydrogen Donor Liquid. Tetralin is an excellent homogeneous hydrogen donor and is representative of compounds that act as hydrogen donors in coal liquefaction processes. It might be expected that it could enhance HDN reactions, so its effect on quinoline HDN was studied here using two concentrations, 5 w t % or 40 wt % in the liquid. All runs were made with 5 w t '30 quinoline and 0.74 wt % CS2 in the feed, with the same sulfided NiMo/A1203catalyst previously used and at 6.9 MPa and 375 "C. (See Figure 11). A t a space time of 500 g of cat. h/mol of quinoline, the total HDN was 65% using the usual paraffin carrier liquid, 65.5% when the carrier contained 5 w t % Tetralin and 56% when the carrier contained 40 wt % Tetralin, as shown in Figure 9. Only 4.5 to 5% of the Tetralin itself was hydrogenated. Studies were also made with OEA (0-ethylaniline) with 5 or 40 wt % Tetralin in the carrier. The HDN conversion of OEA was not noticeably affected with 5 wt % Tetralin, but decreased from 38.5% to 31.9% with 40 wt % Tetralii present. We conclude that Tetralin does not function significantly as a hydrogen donor to enhance hydrogenation reactions in the quinoline HDN network, but instead it inhibits the HDN reactions, presumably by adsorption competition with quinoline and reaction intermediates. Although Tetralin is a relatively neutral molecule, its adsorption is significant, whereas that of paraffins is negligible. The boiling point of Tetralin (207 "C at atmospheric pressure) is approximately 80 "C below that of the paraffin carrier and is less than that of quinoline (237 "C). It is anticipated that similar three-ring compounds would have an even greater inhibiting effect. Effect of Hydrogen Pressure. Shih et al. (1977) reported that the % HDN of quinoline increased to a maximum at 6.9 to 10.5 MPa and then decreased slightly at 14.0 MPa. Such an effect has not been reported elsewhere. The effect of pressure on percent HDN was studied

1

I

I

I

I

100 200 300 400 500 600 S P A C E T I M E hr g c o t l m o l O

Figure 11. Tetralin moderately inhibits the HDN reaction, but its hydrogenation rate is slow; 6.9 MPa, 375 "C.

I

I

hF\

5Wi%3.074 CC7' I T a l -1

35

70

105

140

P R E S S U R E , M Pa

Figure 12. Quinoline conversion increases with hydrogen pressure.

here by use of a feed of 5 w t '70 quinoline, in the absence or presence of CS2 and at two space-times, with results shown in Figure 12. The catalyst charge, no. 6, had a somewhat higher activity than charges no. 2 or 3 because of the particular sulfiding procedure used (Yang and Satterfield, 1983) and had reached steady-state activity before these data were taken. The degree of conversion increases with hydrogen pressure and, in the presence of CS2 and at the longer space-times, reaches a plateau value of about 93%. Quinoline and PyTHQ are essentially in equilibrium in the overall reaction, and the leveling off may be associated with the fact that the ratio of PyTHQ/(PyTHQ + Q) approaches unity at this temperature and 14.0 MPa. Frost and Jensen (1973) reported the HDN rate constant for quinoline at 685 O F to increase with pressure in studies at 4.2,6.9, and 9.6 MPa, and Qader et al. (1968) in studying hydrotreating of a coal tar reported that the % HDN at 400 "C increased with pressure, reaching a plateau value corresponding to a high degree of conversion at 13.7 MPa. Nature of the Active Sites. We postulate the existence of two kinds of catalytic sites to explain our results: site I is a sulfur vacancy associated with the molybdenum atom while site I1 is a Bronsted acid site, either from an added promoter (e.g., phosphate) or from the dissociation of H2S. The characteristics of the two types of sites are discussed in more detail elsewhere (Y ang and Satterfield, 1983) as well as their activities for various hydrogenation and hydrotreating reactions. Briefly, we postulate that

Ind. Eng. Chem. Process Des. Dev., Vol. 23, No. 1, 1984

Figure 13.

Postulated interaction of

H2S w i t h catalyst.

sulfur vacancies are responsible for hydrogenation and hydrogenolysis reactions, while Bronsted acid sites are responsible for hydrogenolysis and ring isomerization. Adsorption and dissociation of an H2Smolecule can convert a sulfur vacancy to a Bronsted acid site plus a sulfhydryl (SH)group, as shown in Figure 13, but the adsorption is readily reversible if H2S is removed from the reaction system. Maternova (1982) has recently established the existence of SH groups on a sulfided commercial CoMo/A1203catalyst by a technique involving adsorption of silver ions from a pyridine solution. The dissociation of H2Sis similar to the dissociative adsorption of water onto a zeolite or alumina, in which a surface vacancy is converted to a Bronsted acid site. Voorhoeve (1971) demonstrated by an ESR study that the number of vacancies on a sulfided NiW/A1203 catalyst decreased with an increase of H2S partial pressure when the catalyst was exposed to an H2S/ H2 environment. Catalytic activity has been correlated with concentration of surface vacancies for a number of reactions. For example, Voorhoeve (1971) related the ESR signal of a sulfided NiW/A1,03 catalyst to activity for benzene hydrogenation. Lipsch and Schuit (1969) proposed a mechanism of thiophene hydrodesulfurization on CoMo(A1203 related to surface vacancies. Tanaka and Okuhara (1977) suggested a correlation between coordination on the surface and hydrogenation and isomerization of olefins. The hydrogenation of a heterocyclic molecule, adsorbed on a vacancy, should be facilitated by nearby chemisorbed hydrogen atoms and/or by hydrogen from the SH group. A similar vacancy could also be responsible for a hydrogenolysis reaction if chemisorbed hydrogen and/or a SH group are nearby. (There may be different kinds of vacancies having relatively different degree of activities for hydrogenation and hydrogenolysis.) Some hydrogenolysis can also occur on Bronsted acid sites. The presence of hydrogen sulfide during reaction reduces the number of sulfur vacancies and increases the number of Bronsted acid sites. In the presence of H2Swe observed a slight reduction in the hydrogenation rates but a significant increase in the hydrogenolysis rates in the quinoline HDN network. Furthermore, the activation energy for the hydrogenolysisreactions drops considerably; that of Itl is reduced from 174 kJ/mol to 107 kJ/mol, and that of k7 from 187 kJ/mol to 123 kJ/mol. This suggests that the Bronsted acid sites are more active than vacancies.

25

The pairs of Bronsted acid H+ and -SH group, and some of the weakly bonded sulfur can be removed upon purging under helium or under HDN conditions, as has been observed. In the presence of H2S, ring isomerization of propylcyclohexene is also enhanced; such isomerization is facilitated by the Bronsted acid sites on the catalyst surface. But when CS2was present in very high concentration, i.e. 5.89 wt %, the isomerization of propylcyclohexene is hampered by polymerization of octahydroquinoline which would occur preferentially on acid sites. To be strictly consistent with the above interpretation, a mathematical model should allow for more than one kind of site, but development of such a model from the present data did not seem to be meaningful in view of the complexity of the reaction network. Our model (Satterfield and Yang, 1984) treats only one kind of site, in which we can visualize all molecules competing for surface vacancies for hydrogenation and hydrogenolysis. The chemisorbed hydrogen is lumped into the rate constant of hydrogenation and the rate constant of hydrogenolysis is a function of vacancies as well as Bronsted acid sites, but no attempt was made to bring this into the model. Acknowledgment This work was supported by the Office of Fossil Energy,

US. Department of Energy, under Contract No. DEAC22-80PC30075. Registry No. Ni, 7440-02-0; Mo, 7439-98-7; CS2,75-15-0; H2S, 7783-06-4;N ~7727-37-9; , OPA, 1821-39-2;DHQ, 2051-2a7; MPCP, 87207-54-3;PCH, 167a92-8; PCHE, 31620-24-3; P ~ ~ H 635-46-1; Q, BzTHO, 10500-57-9; PB, 103-65-1; OEA, 578-54-1; quinoline, 91-22-5; T e t r a l i n , 119-64-2.

Literature Cited Frost, C. M.; Jensen, H. B. Am. Chem. SOC.Div. Pet. Chem. Prepr. 1973, 18(1), 119. Lipsch, J. M. J. G.; Schuit, G. C. A. J . Catal. lS89, 15, 169. Luttringhaus, A.; Jander, J.; Schneider, R. Ber. 1959,92, 1756, as cited in Cook, A. G. "Enamines: Synthesis, Structure, and Reactions": Marcel Dekker: New York, 1989,p 297. MaternovB, J. Appl. Catal. 1982,3 , 3. Qader, S. A.; Wiser, W.H.; Hill, G. R. Ind. Eng. Chem. Process Des. Dev. 1988, 7,391. Satterfield, C. N.; Gultekin, S. Ind. Eng. Chem. Process Des. Dev. 1981, 20,62. Satterfield, C. N.; Yang, S. H. Ind. Eng. Chem. Process Des. Dev. 1984-in press. Schopf, C.; Komzak, A.; Braun, F.; Jacobi. E. Ann. 1948,559, 1, as cited in Cook, A. G. "Enamines: Synthesis, Structure, and Reactions"; Marcel Dekker: New York, 1969,p 297. Shih, S.S.; Katzer, J. R.; Kwart, H.; Stiles, A. B. Am. Chem. SOC.Div. Pet. Chem. Prepr. 1977,22, 919. Tanaka, K.; Okuhara, T. Catal. Rev. Sci. Eng. 1977, 15, 249. Voorhoeve, R. J. H. J . Catal. 1971,2.3,236. Yang, S. H. Sc.D. Thesls, M.I.T.. Cambridge, MA, 1982. Yang, S. H.; Satterfield, C. N. J . Catal. 1983,81, 168.

Received for review July 13, 1982 Accepted March 7, 1983