Polyhedron Vol. 16, No. 18, pp.

3213 3217, 1997

© 1997 Elsevier Science Ltd
~ Pergamon All rights reserved. Printed in Great Britain
PII : S0277-5387(97)00074-0 0277 5387/97 $17.00+0.00

Reactivities in deep catalytic

hydrodesulfurization: challenges, opportunities,
and the importance of 4-methyldibenzothiophene
and 4,6-dimethyldibenzothiophene
B. C. Gates a* and Henrik T o p s o e b

"Departmentof Chemical Engineering and Materials Science, University of California, Davis, CA 95616,
U.S.A.
b Haldor Topsoe Research Laboratories, Nymollevej 55, 2800 Lyngby, Denmark

Abstract--The organosulfur compounds present in fossil fuels vary widely in their reactivities in catalytic
hydrodesulfurization. In naphtha, thiophene is so much less reactive than the thiols, sulfides, and disulfides
that the latter can be considered to be virtually infinitely reactive in practical high-conversion processes.
Similarly, in gas oils and petroleum residua, the reactivities of (alkyl-substituted) 4-methyldibenzothiophene
and 4,6-dimethyldibenzothiophene are much less than those of other sulfur-containing compounds. Conse-
quently, in deep hydrodesulfurization, the conversion of these key substituted dibenzothiophenes largely
determines the required conditions. Because of the growing technological importance of deep desuifurization
of heavy feedstocks, we infer that 4-methyldibenzothiophene and 4,6-dimethyldibenzothiophene are the most
appropriate compounds for investigations of candidate catalysts and reaction mechanisms. © 1997 Elsevier
Science Ltd

Keywords: hydrodesulfurization; dibenzothiophene; 4-methyldibenzothiophene; 4,6-dimethyldibenzo-

thiophene; reactivities in hydrodesulfurization.

REACTIVITIES IN HDS : WHAT IS KNOWN depend strongly on the number of rings in the reactant
(Fig. 1), with thiophene being the most reactive and
The goals of this note are to review briefly what is dibenzothiophene the least. In the family of (sub-
known about reactivities of organosulfur compounds stituted) dibenzothiophenes, which are present in the
in fossil fuels and to draw inferences about how these heavier fractions of oil (e.g., residua), methyl groups
results may guide precess modelling, testing of new in the 4 and in the 4 and 6 positions lead to order-of-
catalysts, and fundamental investigations of reaction magnitude reductions in the reactivity, whereas
networks, kinetics, and mechanism. methyl groups in the other positions have only little
Years of research on catalytic hydrodesulfurization effect on the reactivity (Table 1) [5].
(HDS) [1,2] have led to technologically valuable gen- The foregoing reactivity pattern has been observed
eralizations about how the reactivity of a compound
undergoing HDS depends on its structure. Under con-
ditions typical of industrial HDS, the reactivities of
compounds predominant in naphtha (thiols, sulfides, 9 1
and disulfides) are much greater than those of thio-
phenic compounds [3]. The reactivities of the latter
7 3
6 4
5
Dibenzothiophene
* Author to whom correspondence should be addressed.
3213
3214 B. C. Gates and H. Topsoe
140 effects also depend on the type of molecule being
desulfurized. For example, high concentrations
~' 120 of aromatics (e.g., naphthalene) exert a moderate
c~ inhibiting effect on dibenzothiophene conversion, but
tO0 only a weak inhibiting effect on the conversion of 4,6-
dimethyldibenzothiophene [13].
80 * A reaction network determined for dibenzo-
thiophene conversion with H2 in the presence of Co-
60 Mo/A1203 [14] (Fig. 2) accounts well for the available
¢-

data [1,2], indicating competitive, hydrogenation

40
(prehydrogenation and saturation of an aromatic
x ring) and direct hydrogenolysis (cleavage of C-S
20-
bonds). The data available for 4-methyldibenzo-
I I I I thiophene and 4,6-dimethyldibenzothiophene indicate
0 I 2 3 4 that both the hydrogenolysis and hydrogenation path-
Number of rings in reactant ways are significant but that the rate of hydrogenolysis
Fig. 1. Dependence of reactivity on the number of rings relative to the rate of hydrogenation is less for these
in thiophenic compounds. The reactants were thiophene, substituted dibenzothiophenes than for dibenzothio-
benzothiophene, dibenzothiophene, and benzo[b]naph- phene itself [5, 16-18]. The data of Table 2 represent
tho[2,3-dlthiophene. Reactivities were determined in a batch
reactor with each individual reactant in n-hexadecane solvent
at 300°C and 71 bar; the catalyst was Co-Mo/y-A1203 [4].

42x 104,/ \2 x 10,

for pure compounds reacting with H 2 in the presence
of Co-Mo/A1203 or Ni-Mo/A1203 catalysts [1-3].
Similar results have been found for the same com-
pounds reacting in various oil fractions [7-12] and
coal liquids [6].
/4,xl0
The reactivities summarized in Table 1 for com-
pounds in a mixture of the neutral oils fraction of
a coal-derived liquid (containing largely polycyclic I SIow
aromatic hydrocarbons and nonpolar oxygen-
containing compounds such as dibenzofuran) are less
than those determined for the individual compounds
O--O
reacting in n-hexadecane. The reactivity of a com- Fig. 2. Reaction network for dibenzothiophene hydro-
desulfurization and hydrogenation catalyzed by Co-Mo/
pound with U 2 in a mixture is typically less than that
?-A1203 [14]. The pseudo-first-order rate constants, shown
of the compound in a nearly inert solvent such as n-
next to the arrows, are given in units of L/(g of catalyst s).
hexadecane, and the difference is often attributed to The temperature was 300°C and the pressure 102 bar. These
the competitive adsorption of the mixture components results were obtained in the near absence of H2S; HaS
on the catalyst; aromatic compounds containing changes the values of the rate constants. For a recent state-
nitrogen are the strongest inhibitors [2]. The inhibiting ment of the kinetics in this network, see [15].

Table 1. Reactivities of methyl-substituted dibenzothiophenes

Temperature Pseudo-first-order rate constant Reference

Reactant/solvent (°C) (L/(g of catalyst s))

Dibenzothiophene/n-hexadecane 300 7.38 x 10 -5 5 (b)

Dibenzothiophene/neutral oils 355 5.8 × 10-5 6
2,8-Dimethyldibenzothiophene/n-hexadecane 300 6.72 × 10 5 5 (b)
3,7-Dimethyldibenzothiophene/n-hexadecane 300 3.53 × 10 5 5 (b)
4,6-Dimethyldibenzothiophene/n-hexadecane 300 4.92 x 10 -6 5 (b)
4-Methyldibenzothiophene/n-hexadecane 300 6.64 × 10 -6 5(b)
4-Methyldibenzothiophene/neutral oils 355 1.8 × 10-5 6

Reaction conditions : flow reactor, reactants present in liquid phase with H2 at 102 bar when the solvent was n-hexadecane
and 120 bar when the solvent was neutral oils; the catalyst was Co-Mo/7-A1203 in the former case and Ni-Mo/7-AI203 in
the latter.
 Reactivities in deep catalytic hydrodesulfurization 3215
Table 2. Pseudo-first-order rate constants for HDS of dibenzothiophene, 4-methyldibenzothiophene, and 4,6-dimethyl-
dibenzothiophene catalyzed by CoMo/AI203 at 350°C and 50 bar [18].

Hydrogenolysis Hydrogenation Total HDS

Reactant rate constant (h -~) (h -1)
rate c o n s t a n t rate constant (h ~)

Dibenzothiophene 123 15 138

4-Methyldibenzothiophene 26 15 41
4,6-Dimethyldibenzothiophene 6 11 17

HDS of dibenzothiophene, 4-methyldibenzothiophene, organosulfur compound is 4,6-dimethyldibenzo-

and 4,6-dimethyldibenzothiophene in the presence of thiophene.
a Co-Mo/A1203 catalyst, showing the relative impor- Thus, predictions of the overall sulfur removal must
tance of the hydrogenolysis and hydrogenation path- take into account the removal of the key components,
ways [18]. Dibenzothiophene is desulfurized pre- which can be measured by gas chromatographic
dominantly by the hydrogenolysis pathway, whereas analysis of the feed and product. The generalization
4,6-dimethyldibenzothiophene is desulfurized pre- is applicable to the heavier feedstocks (residua) as
dominantly by the hydrogenation pathway. Thus, it well. Models accounting for the reactivities of several
is not surprising that the catalysts Ni-Mo/A1203 and fractions of organosulfur compounds in residua have
Ni-Mo-P/AI203, which are generally regarded as been presented [19], with the HDS of each being rep-
superior to Co-Mo/AI203 as hydrogenation (and resented as first-order in the reactant, as expected
HDN) catalysts and inferior to Co-Mo/Al203 as HDS [2,31.
catalysts (as illustrated by their performance in HDS In deep HDS, which is becoming the rule rather
of dibenzothiophene), are better than Co-Mo/A1203 than the exception because of increasingly stringent
for the conversion of 4,6-dimethyldibenzothiophene fuel sulfur specifications needed to meet air-quality
(Table 3) [7,13]. standards, the key, relatively unreactive, compounds,
(substituted) 4-methyldibenzothiophene and (sub-
stituted) 4,6-dimethyldibenzothiophene, must be
IMPLICATIONS OF THE REACTIVITIES
substantially converted, and the more reactive organ-
The reactivity data characterizing HDS of naphtha osulfur compounds must be almost completely con-
show that at the high conversions attained in practice, verted. An important challenge is to find more active
all the compounds except thiophenes are removed vir- catalysts for deep HDS, and this challenge can be
tually completely; thus, thiophene is the key compon- stated without much oversimplification as that of
ent, and predictions of the overall sulfur removal can finding more active catalysts for HDS of (substituted)
be based to a good approximation on the removal of 4-methyldibenzothiophene and (substituted) 4,6-
thiophene [3]. Similarly, in the latter stages of deep dimethyldibenzothiophene.
desulfurization of gas oils (e.g., to give 0.05 wt% S in Another practical inference is that in the search for
the product, as is now typical in practice), the more improved catalysts it may be fruitful, particularly in
reactive compounds are virtually all converted. For the initial stages of investigations of candidate cata-
example, the data of Fig. 3 show removals of organ- lysts, to use the performance in HDS of 4-methyl-
osulfur compounds from oil measured by gas chro- dibenzothiophene and/or 4,6-dimethyldibenzo-
matography with a sulfur-specific detector. At the thiophene as a guide. By investigating the reactivities
highest conversions, the predominant remaining of these pure compounds, it may be possible to deter-
mine the reaction networks and relative rates of
hydrogenation and hydrogenolysis and thus to gain
insights into the workings of the candidate catalysts.
Table 3. Comparison of catalysts for HDS of dibenzo- Furthermore, it may be possible to modify the reac-
thiophene, 4-methyldibenzothiophene, and 4,6-dimethyl- tivities of compounds such as 4-methyldibenzo-
dibenzothiophene at 350°C and 50 bar [18] thiophene and 4,6-dimethyldibenzothiophene by
converting them in pretreatment processes. Routes
Rate constant for reaction that have been pursued recently for conversion of the
catalyzed by Ni-Mo-P/AI203 relatively unreactive organosulfur compounds include
relative to that for reaction isomerization, dealkylation, and hydrocracking prior
Reactant catalyzed by Co-Mo/A1203
to desulfurization [20-22].
0.76 The marked differences in reactivities of the various
Dibenzothiophene
4-Methyldibenzothiophene 1.01 organosulfur components in residua have important
4,6-Dimethyldibenzothiophene 1.61 implications for the recognition of optimal processing
conditions and of optimal catalysts or catalyst corn-
3216 B. C. Gates and H. Tops~e

Fig. 3. Gas chromatographic traces showing the removal of organosulfur compounds from oil [10].The upper trace represents
the feed, and the traces in descending order represent products with increasing conversions in HDS. The data illustrate the
relatively low reactivity of 4,6-dimethyldibenzothiophene.

binations in deep HDS [23]. A process engineer should 4. Can answers to questions such as these lead to
be able to estimate the fractional sulfur removal at ideas about how to make better catalysts and develop
high HDS conversions rather well by measuring the better processes?
fractional removal of these key components and
accounting for the almost complete removal of the
others. Changes in the conversions of these key com-
pounds during operation might be useful measures REFERENCES
of the changes undergone by a catalyst as it ages.
Furthermore, it seems to be feasible to determine pro- 1. Topsoe, H., Clausen, B. S. and Massoth, F. E.,
in Catalysis, Science and Technology, eds J. R.
cessing correlations, such as those for prediction of
Anderson and M. Boudart, Vol. 11, Springer,
catalyst activity loss in HDS, in terms of the catalytic
Berlin, 1996.
activity for conversion of the key components. Kin- 2. Girgis, M. J. and Gates, B. C., Ind. En9. Chem.
etics representing the dependence of reaction rate on Res., 1991, 30, 2021.
temperature and reactant and inhibitor concen- 3. Gates, B. C., Katzer, J. R. and Schuit, G. C. A.,
trations, determined from data for conversion of the Chemistry of Catalytic Processes, McGraw-Hill,
key components, might form the basis for models of New York, 1979, Chap. 5.
process operation. 4. Nag, N. K., Sapre, A. V., Broderick, D. H. and
Furthermore, in fundamental investigations of Gates, B. C., J. Catal., 1979, 57, 509.
catalyst structure-property relationships and reaction 5. (a) Houalla, M. Broderick, D. H., de Beer, V. H.
J., Gates, B. C. and Kwart, H., Prepr. Am. Chem.
mechanisms, it would be advantageous to investigate
Soc., Div. Petr. Chem., Aug.-Sept., 1977, p. 941 ;
these relatively unreactive compounds, although,
(b) Houalla, M. Broderick, D. H., Sapre, A. V.,
under some conditions (e.g., ultrahigh vacuum), this Nag, N. K., de Beer, V. H. J., Gates, B. C. and
may not be possible. Questions to be answered include Kwart, H., J. Catal., 1980, 61, 523.
the following : 6. Katti, S. S., Westerman, D. W. B., Gates, B. C.,
1. How do the rates of hydrogenation and hydro- Youngless, T. and Petrakis, L., Ind. Eng. Chem.
Process Des. Dev., 1983, 23, 773.
genolysis of 4-methyldibenzothiophene and 4,6-di- 7. Ma, X., Sakanishi, K., Isoda, T. and Mochida, I.,
methyldibenzothiophene depend on reaction con- Ind. Eng. Chem. Res., 1995, 34, 748.
ditions (e.g., temperature, concentration of H2, con- 8. Amorelli, A., Amos, Y. D., Halsig, C. P.,
centration of H2S, and concentrations of other Kosman, J. J., Jonke, R. R. J., De Wind, M. and
inhibitors)? Vrieling, J., Hydrocarbon Process., 1992, 71,
2. How do the methyl (or other alkyl) substituents 93.
affect these reactivities? Do these substituents steri- 9. Petrakis, L. and Gates, B. C. unpublished results.
cally hinder the adsorption or subsequent steps in the 10. Shih, S. S., Mizrahi, S., Green, L. A. and Sarli,
surface reactions? Are the effects of these substituents M. S., Ind. Eng. Chem. Res., 1992, 31, 1232.
(Table 1) too small to be described as steric [2]? How 11. Kabe, T., Ishihara, A. and Tajima, H., Ind. Eng.
Chem. Res., 1992, 31, 1577,
should they be described?
12. (a) Ma, X., Sakanishi, K. and Mochida, I., Ind.
3. How do changes in the catalyst morphology (e.g., Eng. Chem. Res., 1994, 33, 218 ; (b) ibid 1996, 35,
the dimensions of CoMoS domains [1]) affect the rates 2487.
of the hydrogenation and hydrogenolysis reactions, 13. Isoda, T., Ma, X. and Mochida, I., Prepr. Am.
and do the effects of the substituents vary depending Chem. Soc., Div. Fuel Chem., 1994, 39, 584.
on this morphology? 14. Houalla, M., Nag, N. K., Sapre, A. V., Broderick,
 Reactivities in deep catalytic hydrodesulfurization 3217
D. H. and Gates, B. C., AIChEJ., 1978, 24, 1015. roleum Chemistry, American Chemical Society,
15. Van Rysselberghe, V. and Froment, G. F., Ind. Orlando, Florida, August, 1996.
Eng. Chem. Res., 1996, 35, 3311. 20. Isoda, T., Nagao, S. Korai, Y. and Mochida, I.,
16. Kabe, T., Ishihara, A. and Zhang, A., Appl. Catal. Prepr. Am. Chem. Soc., Div. Fuel Chem., 1996,
A, 1993, 97, L1. 41, 559.
17. Lamure-Meille, V., Schulz, E., Lemaire, M. and 21. Landau, M. V., Berger, D. and Herskowitz, M.,
Vrinat, M., Appl. Catal., 1995, 131, 143. J. Catal., 1996, 158, 236.
18. Ovesen, C. V., Brorson, M., Nielsen, I. V., 22. Kazuyoski, K. Takema, W., Satoru, M. and
Andersen, K. A. and Topsoe, H., unpublished Masakatsu, M., Chem. Lett., 1996, 9, 829.
results. 23. Mochida, I., Sakanishi, K., Ma, X., Nagao, S.
19. Andersen, K., paper presented at Division of Pet- and Isoda, T., Catal. Today, 1996, 29, 185.