Reactivities, Reaction Networks, and Kinetics in High-Pressure Catalytic Hydroprocessing

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 38

I n d . Eng. Chem. Res.

1991,30, 2021-2058 2021

REVIEWS

Reactivities, Reaction Networks, and Kinetics in High-pressure


Catalytic Hydroprocessing
Michael J. Girgis*J**
and Bruce C. Gates*it
Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware,
Newark, Delaware 19716, and Central Research Laboratory, Mobil Research and Development Corporation,
Princeton, New Jersey 08543

This paper is a critical and thorough (but not exhaustive) review of the literature of catalytic
hydroprocessing reactions. It includes data characterizing thermodynamics, reactivities, reaction
networks, and kinetics of hydrogenation of aromatic hydrocarbons, hydrodesulfurization, hydro-
denitrogenation, and hydrodeoxygenation. The subject is organized by reaction type, and there are
separate sections on competitive reaction and inhibition effects. The results summarized are judged
to be the most reliable and useful available to guide process modeling for hydroprocessing of heavy
fossil fuels. The review is designed for use as a reference; most of the quantitative results are
summarized in tables and figures, and a table of contents and a chart with names and structures
of the organic compounds are also included.

I. Introduction 2021 VIA. Effects of Additives on Aromatic 2045


11. Hydrogenation of Aromatic Hydrocarbons 2022 Hydrogenation
IIA. Thermodynamics 2022 VIAL Sulfur Compounds 2045
IIB. Reactivities 2023 VIAS. Oxygen Compounds 2045
IIC. Reaction Networks and Kinetics 2024 VIA3. Nitrogen Compounds 2046
IIC1. Monoaromatics 2024 VIB. Effect of Additives on HDS 2046
IIC2. Aromatics with Two Fused Rings 2025 VIB1. Aromatic Hydrocarbons 2046
IIC3. Aromatics with Three Fused Rings 2025 VIB2. Sulfur Compounds 2048
IIC4. Aromatics with Four Fused Rings 2026 VIB3. Oxygen Compounds 2048
IIC5. Five-Carbon-Membered-Ring 2026 VIB4. Nitrogen Compounds 2048
Aromatics VIC. Effect of Additives on HDO 2050
111. Hydrodesulfurization (HDS) 2027 VIC1. Sulfur Compounds 2050
IIIA. Thermodynamics 2027 VICB. Oxygen Compounds 2050
IIIB. Reactivities 2027 VIC3. Nitrogen Compounds 2051
IIIC. Reaction Networks and Kinetics 2029 VID. Effect of Additives on HDN 2051
IIIC1. Thiophene 2029 VID1. Aromatic Hydrocarbons 2051
IIIC2. Benzothiophene 2030 VID2. Sulfur Compounds 2052
IIIC3. Dibenzothiophene 2031 VID3. Oxygen Compounds 2053
IIIC4. Benzonaphthothiophene 2032 VID4. Nitrogen Compounds 2054
IV. Hydrodenitrogenation (HDN) 2032 VII. Multicomponent Mixtures 2054
IVA. Thermodynamics 2032 VIII. Conclusions 2055
IVB. Reactivities 2033 IX. Literature Cited 2056
IVC. Reaction Networks and Kinetics 2034
IVC1. Pyridine 2034 1. Introduction
IVC2. Quinoline 2036 Catalytic hydroprocessing ranks in importance with
IVC3. Acridine and Benzoquinolines 2038 other large-scale petroleum refining processes (cracking
IVC4. Nonbasic Nitrogen Compounds 2039 and reforming), and its application is growing to meet
IVC5. Aniline Derivatives 2040 several needs, including the processing of heavier feeds,
V. Hydrodeoxygenation (HDO) 2041 the production of high-performance lubricants, and the
VA. Thermodynamics 2041 introduction of cleaner burning fuels. In the long term,
VB. Reactivities 2041 the anticipated processing of liquids from tar sands, coal,
VC. Reaction Networks and Kinetics 2042 and shale will lead to major new hydroprocessing appli-
VC1. Phenolic Oxygen Compounds 2042 cations.
VC2. Heterocyclic Oxygen Compounds 2043 The following review is meant to provide a summary of
VC3. Ethers 2044 quantitative results characterizing the thermodynamics,
VI. Binary Mixtures 2044 reactivities, reaction networks, and reaction kinetics of
catalytic hydroprocessing. The emphasis is on results of
University of Delaware. high-pressure (55-170 atm) experiments with compounds
* Mobil Research and Development Corp. representative of the least reactive constituents of heavy
0888-5885/91/2630-2021$02.50/0 0 1991 American Chemical Society
2022 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

feeds, Le., the compounds most pertinent for process Chart I. Structures of Selected Organic Compounds
modeling and design. The reactant compounds considered 1, Aromatic Hydrocarbons
are aromatic, including polycyclics found in heavy fractions
derived from petroleum and other fossil fuels. Data 5 6
7
characterizing reactivities of industrial feedstocks (e.g., 8
2 7
overall conversions of compound classes, and product 7
6 3 6 3 2
physical properties) are omitted here because the relevant 5 4

quantitative information is lacking in the open literature. Naphthalene Anthracene Phenanthrene Pyrene
The premise of this review is that, because of the great
difficulty in fundamental interpretation of results obtained 3 4
2 5
with commercial feeds, pure compound data are the best 1 6
source available for developing a sound understanding of 3
4 5
6
the chemistry of hydroprocessing and provide the best 2 7
1 9 6
available quantitative foundation for process modeling and
catalyst development. Fluoranthew Fluorene

Several classes of reactions occur simultaneously in


hydroprocessing-hydrogenation, hydrodesulfurization 2. Organosulfur Compounds
(HDS), hydrodenitrogenation (HDN), and hydro-
deoxygenation (HD0)-and these must all be considered
if the reactivity patterns of commercial feedstocks are to
be understood. In addition, inhibition by some reactants
and products markedly affects the reactivity. Accordingly,
in this review we fist consider separately hydrogenation
of aromatic hydrocarbons, hydrodesulfurization, hydro-
denitrogenation, and hydrodeoxygenation. A discussion 10
9
of inhibition effects follows, with emphasis on the modeling 1 11 10
8
2 9 2
of reaction inhibition resulting from competitive adsorp- 3 8 3
tion. The few investigations carried out with mixtures
containing three or more compounds are considered next.
The consideration of all of the major reaction classes and
the inhibition effects is a distinguishing feature of this
review; previous reviews (Schuit and Gates, 1973; Katzer 3. Organonitrogen Compounds
and Sivasubramanian, 1979; Furimsky, 1983; Vrinat, 1983;
Ho, 1988) have focused on single reaction types. Hydro-
cracking (by dual-function catalysts) and hydro- 8 9 1

demetalation are beyond the scope of the review; the latter


has been reviewed recently (Quann et al., 1988).
:Q: 7
6
2
3

Almost all the available results have been obtained with Pyrdine hinoline Aciiine 5,6-Eenzoquinoline
undeactivated commercial hydroprocessing catalysts typ-
ified by sulfided Co0-Mo03/r-A1203 (Co-Mo/A1203),
commonly used for hydrodesulfurization, sulfided NiO-
Mo03/y-A1203(Ni-Mo/A1203),which is similar to the
former but has higher activity for hydrogenation and is
typically applied for HDN, and sulfided NiO-W03/y-
A1203 (Ni-W/AI2O3). The selections made from the vast 7.8-Eenrcqumoline Carbazole
literature represent the authors' judgement of which data
appear to be most thorough, reliable, and internally con-
sistent. The reactivities and inhibition parameters are all
4.Organo-oxygen Compounds
represented in the same units, insofar as possible; however,
some literature data are reported incompletely and do not
allow determination of rate constants. For example, results
are sometimes given as pseudohomogeneous rate constants,
e.g., in units of s-l rather than L/(g of cata1yst.s).
We stress that not all the results presented here are
mutually consistent. Typically, the catalyst used by one 1-Naphthol Eenzoluran Dibenzoluran

group of investigators is different from that used by an- seems useful to present comparisons of the reactions of
other. Nonetheless, the results are valuable in providing various compounds in a group before presenting the
a comprehensive summary of hydroprocessing chemistry quantitative models of reaction networks and kinetics; the
and they offer valuable insights for modeling the kinetics; latter are available for only a small number of compounds.
this is the first compilation of quantitative results of this Most of the quantitative results are presented in tables and
kind. figures.
This review is organized for easy reference. Chart I
includes structures and names of the more complex organic 11. Hydrogenation of Aromatic Hydrocarbons
compounds, and the table of contents will help the reader IIA. Thermodynamics. Hydrogenations of aromatic
go back and forth in the document. For each class of hydrocarbons are reversible, with equilibrium conversions
reaction, the results are presented in sequence in the of hydrocarbons often being less than 100% under prac-
following categories: thermodynamics, reactivities, reaction tical processing conditions. These reactions are exother-
networks, and kinetics. These categories overlap, but it mic; the extent of reaction at equilibrium decreases with
Ind. Eng. Chem. Res., Vol. 30,No. 9,1991 2023

Table I. Equilibrium Constants for Hydrogenation of mations for reactions occurring in the liquid phase because
Aromatic Hydrocarbons (Frye, 1962; Frye and enthalpies and entropies of reaction, which affect the value
Weitkamp, 1969) of the equilibrium constant most strongly, change only
mo,,0 slightly upon a change from the gas to the liquid phase.
kcal/mol Accurate computation of equilibrium concentrations in
log K at of organic liquid-phase reactions also requires knowledge of activity
naphthalene + 2H2 - 300 "C 350 "C 400 "C reactant
-1.13 -2.03 -2.80 -30
coefficients for all species, including hydrogen. Such data
tetralin
phenanthrene + H2 - -1.57 -1.94 -2.25 -12
are generally unavailable for a given hydrocarbon and the
products of its hydrogenation. Activity coefficients for the
-
9,lO-dihydrophenanthrene
fluorene + 3H2
cis-hexahydrofluorene
-2.08 -4.17 -5.27 -42
reactants may be estimated with group contribution
methods such as UNIFAC (Reid et al., 1987). A correlation
for hydrogen solubility and liquid-phase fugacity in several
OThe enthalpies are derived from the slopes of the van't Hoff hydrocarbons and coal liquid fractions has been proposed
plots and are assumed to be standard enthalpies of reaction, al- (Sebastian et al., 19811,but its applicability requires fur-
though they are not listed as such by the authors. ther examination.
increasing temperature. Thus increasing the temperature IIB. Reactivities. Most of the quantitative reactivity
to give higher rates of other reactions results in lower data characterize reactions of individual aromatic hydro-
equilibrium conversions in aromatic hydrogenations carbons with hydrogen. A few investigations have been
and-because the hydrogenations are often fast enough reported with mixtures of hydrocarbons chosen so that
inhibition effects were negligible.
to be nearly equilibrated-lower actual conversions as well.
In practice, the lower equilibrium conversions are espe- Pseudo-first-order rate constants for hydrogenation of
cially significant at lower pressures. benzene, biphenyl, naphthalene, and 2-phenylnaphthalene
There are few thermodynamics data for polyaromatic catalyzed by Co-Mo/A1203 were determined to be 2.81 X
hydrocarbons and the products of their hydrogenation that lo*, 3.00 X lo4, 5.78 X lo-&,and 7.06 X lo-& L/(g of
are sufficiently accurate to allow useful calculations of catalystss), respectively (Sapre and Gates, 19811,under the
equilibrium concentrations of partially hydrogenated conditions shown in Table 11. These data suggest the
products. Frye (1962)and Frye and Weitkamp (1969) following generalization: naphthalene and substituted
measured the compositions of several aromatic hydro- naphthalenes are an order of magnitude more reactive than
carbon-hydrogen mixtures at several temperatures and benzene and substituted benzenes. The greater reactivity
pressures after equilibrium was attained. All species were of naphthalene and 2-phenylnaphthalene was speculated
in the vapor phase. The hydrocarbons included biphenyl, to be a consequence of the greater resonance stabilization
indene, naphthalene, phenanthrene, acenaphthene, and of diaromatic surface species formed by a-bonding of the
fluorene. Results for naphthalene, phenanthrene, and naphthalene moieties to exposed Mo cations, which were
fluorene (Table I) show that the hydrogenation equilibrium postulated to be the catalytic sites. The relatively high
constants are less than unity at typical hydroprocessing reactivity of naphthalene moieties evidently extends to
temperatures (usually greater than 340 OC for heavy monoaromatic and diaromatic moieties that are fused with
feedstocks). Consequently, operation at high hydrogen five-carbon-membered rings; for example, hydrogenation
partial pressures is necessary to hydrogenate aromatic of the naphthalene moiety in fluoranthene is an order of
hydrocarbons to an appreciable extent. For example, an magnitude faster than hydrogenation of the benzene
increase in the hydrogen partial pressure from 9.7 to 37 moiety in fluorene (Lapinas et al., 1991). Comparable
atm increases the equilibrium conversion of naphthalene quantitative data for aromatic hydrocarbons with three
from 17 to 84% at 396 OC (Frye, 1962;Frye and Weitkamp, or more fused aromatic rings are lacking.
1969). Although data for other multiring compounds are The results of Sapre and Gates also demonstrate that
also available (Reid et al., 19871,the corresponding data the effect of the phenyl substituent on the reactivity is
for the hydrogenated products are lacking. The data weak. This weak effect of single hydrocarbon substituents
available for benzene, some alkylbenzenes, and the prod- is consistent with results (Aubert et al., 1988)showing that
ucts of their hydrogenation are more extensive (Stull et the relative rate constants for hydrogenation of benzene,
al., 1969). ethylbenzene, biphenyl, cyclohexylbenzene, diphenyl-
Thermochemical data are needed; in their absence, methane, and (cyclohexylmethy1)benzene(under the con-
group contribution methods can be used to estimate the ditions shown in Table 11) were 2, 2, 6, 2, 3, and 2, re-
equilibrium constants (Reid et al., 1987). Benson and spectively. The small effect of single Substituents suggests
co-workers (Shaw et al., 1977;Stein et al., 1977)extended that neither steric nor electronic factors are large. Com-
their group contribution method (Benson et al., 1969)for parable data characterizing reactivities of polysubstituted
the estimation of standard enthalpies, entropies, and monoaromatic hydrocarbons are lacking.
ideal-gas heat capacities of selected aromatic compounds Using literature data, Nag (1984)observed a linear
and their hydrogenation products. The group contribution correlation between the logarithm of the hydrogenation
methods provide only rough approximations. For example, rates of aromatic hydrocarbons under hydroprocessing
the uncertainty in the standard enthalpy of formation of conditions and their ionization potentials. Assuming that
hydrogenated aromatic hydrocarbons is typically as high reaction intermediates in hydrogenations are a-complexes
as 3 kcal/mol, resulting in uncertainties of more than an on the catalyst surface formed by the transfer of electrons
order of magnitude in the computed equilibrium constants. from the highest occupied molecular orbital (HOMO) of
Equilibrium constants for liquid-phase-hydrogenation the hydrocarbon to an unoccupied d-orbital on the metal,
reactions are not readily available. The application of the Nag reasoned that a higher HOMO energy should result
gas-phase equilibrium constants obtained from the sources in a more facile electron transfer and hence a more stable
cited above to hydroprocessing reactions occurring in the a-complex. On the basis of the inverse proportionality
liquid phase in industrial tricklebed reactors is not strictly between the HOMO energy and the ionization potential,
correct. However, Shaw et al. (1977)pointed out that the correlation between the logarithm of the hydrogenation
gas-phase equilibrium constants provide good approxi- rate and the ionization potential was developed. The
2024 Ind. Eng. Chem. Res., Vol. 30,No. 9,1991

Table 11. Reaction Conditione in Network Studies of Hydrogenation of Aromatic Hydrocarbons


reactant conditions reference
1. Monoaromatics
benzene, biphenyl batch reactor, 325 "C, 75 atm, n-hexadecane solvent, Sapre and Gates, 1981
Co-Mo/AlZOS catalyst
benzene, biphenyl, ethylbenzene, batch reactor, 340 "C, 70 atm, n-decane solvent, Aubert et al., 1988
cyclohexylbenzene, diphenylmethane, Ni-Mo/A1203 catalyst
(cyclohexylmethyl)benzene
2. Two-Fused-Ring Aromatics
naphthalene batch reactor, 325 "C, 75 atm, n-hexadecane solvent, Sapre and Gates, 1981
Co-Mo/AlZOs catalyst
batch reactor, 35 atm, 350 "C, white oil solvent, Bhinde, 1979
Ni-Mo/AlzOs catalyst
2-methylnaphthalene liquid-phase flow reactor, 280-360 "C, 40-128 atm, Ho et al., 1988
Ni-Mo/A120Bcatalyst
l-methylnaphthalene trickle-bed reactor, 310-350 "C, 50 atm, Patzer et al., 1979
Ni-Co-Mo/A1203 and proprietary catalysts
2-phenylnaphthalene batch reactor, 325 "C, 75 atm, n-hexadecane solvent, Sapre and Gates, 1981
Co-Mo/Alz03 catalyst
3. Three-Fused-Ring Aromatics
anthracene batch reactor, 220-435 "C, 102 atm, Ni-W catalyst Wiser et al., 1970
(unsupported)
phenanthrene batch reactor, 200-380 "C, 102 atm, 2 h reaction time, Shabtai et al., 1978a
Ni-W/Al2Oa catalyst
flow reactor, 402-543 "C, 136-204 atm, nonacidic Wu and Haynes, 1975
alumina chromia
flow reactor, 316-538 "C, 103-171 atm, 1-4 WHSV, Huang et al., 1977
Co-Mo/A1203 catalyst
batch reactor, 430 "C, 100 atm, Ni-Mo/A1203 catalyst Lemberton and Guisnet, 1984
4. Four-Fused-Ring Aromatics
batch reactor, 341 "C, 69-205 atm, 0.6 WHSV, Shabtai et al., 1978a
Ni-W/Al2O3 catalyst
batch microreactors, 375-425 "C, 18-72 atm, Johnston, 1984
Co-Mo/A1203 catalyst
batch reactor, 348-400 "C,35-137 atm, n-hexadecane Stephens and Chapman, 1983;
solvent, Ni-Mo/A1203 catalyst Stephens and Kottenstette, 1985
5. Five-Carbon-Membered-Ring-Containing Aromatics
fluorene batch reactor, 310-380 "C, 153 atm, n-hexadecane Lapinas et el., 1991
solvent, Ni-W/AI2O3 catalyst
fluoranthene batch reactor, 310-380 "C, 153 atm, n-hexadecane Lapinas et al., 1987

-0-
2.81

I
x lo4

x 10.'
1.0 x
solvent, Ni-W/A1203 catalyst
10.6
Hydrocarbon
action conditions were similar to those employed to de-
termine the results of Figure l, they did not detect isom-
erization products and reported only relative rate con-
stants.
Aubert et al. also reported reaction networks for
ethylbenzene, cyclohexylbenzene, diphenylmethane, and
(cyclohexylmethy1)benzene(Table 11). The networks all
involve hydrogenation of the benzene ring. In the case of
diphenylmethane, the two monoaromatic rings are hy-
drogenated sequentially, as in the biphenyl network.
Hydrocarbons Isomerization products were not reported, and no reverse
Figure 1. Reaction networke for benzene and biphenyl propoeed by reactions were included.
Sapre and Gates (1981) (Table 11). The numbers over the arrows are Inferring that biphenyl is representative of the least
pseudo-first-order rate constants in L/(g of catalystd at 325 "C. reactive class of aromatic hydrocarbons, Sapre and Gates
(1982)chose biphenyl hydrogenation for measurement of
correlation is evidently valid only for aromatics that have kinetics. They used a flow reactor operated under dif-
collinear aromatic ring centers (e.g., anthracene) without ferential conversion conditions (biphenyl conversion less
bulky substituents. than 16% ), without characterizing the reverse reaction
IIC. Reaction Networks and Kinetics. IIC1. Mo- quantitatively. Experimenta were conducted at 300-375
noaromatics. Reaction networks for hydrogenation of "C and at several hydrogen, biphenyl, and hydrogen sulfide
benzene and biphenyl (Table 11, Figure 1) proposed by concentrations (66-328,53-26, and 8.5-38 mmol/L, re-
Sapre and Gates (1981)account for the reversibility, which spectively) with a Co-Mo/A120, catalyst. The sole reaction
was confirmed for biphenyl hydrogenation in experiments product was cyclohexylbenzene. An empirical rate equa-
with cyclohexylbenzene as a reactant. The unidentified tion proposed for biphenyl hydrogenation is shown in
secondary products were thought to be formed by isom- Table I11 with parameter values at two temperatures. The
erization of saturated six-membered rings to give me- rate increased with feed biphenyl concentrations and ap-
thylcyclopentane rings. Products containing partially proached a saturation value. The rate increased more than
hydrogenated rings (e.g., cyclohexadienes or cyclohexenes) linearly with hydrogen concentration in the liquid. Hy-
were not formed in measurable amounts. Aubert et al. drogen sulfide inhibited the reaction. The parameters that
(1988)reported similar networks, but although their re- indicate adsorption equilibrium constants according to the
Ind. Eng. Chem. Res., Vol. 30, No. 9,1991 2025

Table 111. Rate Equation Proposed by Sapre and Gates


(1982) for Biphenyl Hydrogenation, Parameter Values at
Two Temperatures, and Corresponding Activation
EnergiedHeats of Adsorptiona
- BZCCB
BICBPCH,S
' I
r=
(1 &CBP + BdCHfis)2(1+ Bijc~J'

300
temp, "C
350
E, or AH,
kcal/mol I R
I
El,L4/(mo13.gof catalystd 0.015 0.024 34
Bo L/(g of catalyst-s) 3.3 X 10" 5.0 X lo4 not given
Ea, L/mol 23.0 4.6 -14
E,, L/mol 25.6 12.0 -16 Figure 3. Reaction pathways for anthracene hydrogenation and
E5,L/mol 32.2 27.2 -5 hydrocracking suggested by the results of Wiser et al. (1970) (Table
OThe subscripts BP and CB denote biphenyl and cyclohexyl- 11).
benzene, respectively. sequence shown in Figure 2. Extensive dealkylation
leading to removal of small substituents in commercial
feedstocks would markedly increase light gas yields and
hydrogen consumption, and a better understanding of the
extent of dealkylation as a function of substituent type is
needed.
The reaction network determined for 2-phenyl-
naphthalene (Table 11, Figure 2) is analogous to that for
naphthalene. The secondary products consist mostly of
phenyl-substituted decalins. The absence of cyclohexyl-
naphthalenes in the products is consistent with the slow
hydrogenation of monoaromatics relative to diaromatics.
The hydrogenation of the ring in 2-phenylnaphthalene to
which the phenyl substituent is bonded proceeds more
rapidly, possibly because of electron-donationeffects from
6.1 X 10
the phenyl substituents.
IIC3. Aromatics with Three Fused Rings. Com-
pounds with three fused rings, anthracene and phenan-
threne, have been investigated (Table 11), but quantitative
reaction networks have not been reported. The proposed
Hydrocarbons reaction pathways for anthracene are shown in Figure 3.
Although the initial rate of formation of 1,2,3,4-tetra-
Figure 2. Reaction networks for naphthalene and 2-phenyl- hydroanthracene appeared to be nonzero, it was not pos-
naphthalene hydrogenation proposed by Sapre and Gates (1981)
(Table 11). The numbera over the arrows are pseudo-first-orderrate sible to establish whether it was a primary product. At
constants in L/(g of catalyst4 at 325 OC. 435 "C, alkyl-substituted naphthalenes were detected in
the reaction products, demonstrating the occurrence of
Langmuir-Hinshelwood formalism are roughly equal for cracking of the outer ring, presumably following isomer-
biphenyl, hydrogen sulfide, and hydrogen at the lower ization. The formation of 9,lO-dihydroanthracene is sig-
temperature. nificant because of the potential for hydrocracking 1mol
IIC2. Aromatics with Two Fused Rings. Quanti- of anthracene to give 2 mol of valuable gasoline-boiling-
tative reaction networks have been reported for naph- range monoaromatic products.
thalene hydrogenation (Table 11, Figure 2). The hydro- Phenanthrene hydrogenation has been investigated by
genation is sequential, with the rate of tetralin hydrogen- several workers (Table 11), but the pathways proposed all
ation being an order of magnitude less than that of differ and generalizations are difficult.
naphthalene hydrogenation and about the same as that Having measured product selectivities while varying the
of benzene hydrogenation. The reversibility of naphtha- temperature at constant reaction time and pressure,
lene hydrogenation was confirmed by using tetralin as Shabtai et al. (1978a) proposed that 9,lO-dihydro-
reactant. The dehydrogenation rates of the two decalins phenanthrene and 1,2,3,4-tetrahydrophenanthrenewere
are negligible. Because of differences in the conditions primary products. 9,lO-Dihydrophenanthrenewas postu-
used by the two groups of investigators listed in Table 11, lated to be relatively unstable because of the strain in its
quantitative comparison of Co-Mo/A1203 and Ni-Mo/ central hydrogenated ring and was assumed to undergo
A1203 catalysts for naphthalene hydrogenation activity is dehydrogenation rapidly to give phenanthrene. Per-
not possible. hydrophenanthrene was formed only slowly, presumably
These results are generally consistent with those ob- because of the sterically hindered central ring of ita pre-
tained with methyl-substituted naphthalenes (Table 11). cursor, sym-octahydrophenanthrene. The latter was pro-
Ho et al. (1984) investigated the hydrogenation of 2- posed to be formed from 1,2,3,4-tetrahydrophenanthrene.
methylnaphthalene and proposed a network involving the Some edgewise adsorption of sym-octahydrophenanthrene
sequential hydrogenation of the reactant to give two me- is inferred to have occurred, as producta were detected with
thyltetralins, which were subsequently hydrogenated to a partially hydrogenated central ring. The suggested
give methyldecalins. However, in experiments with 1- pathways, though not inconsistent with the data, would
methylnaphthalene, Patzer and co-workers (1979) found be more convincing if the extrapolation of the selectivities
that l-methylnaphthalene was first hydrodealkylated to were performed using isothermal data. Moreover, the
give naphthalene, with the latter reacting further by the reactant conversions were not specified; inference of the
2026 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

pathways based on selectivity extrapolations is more first-order rate constants were estimated for the hydro-
plausible at lower conversions. genation of pyrene to give 4,5-dihydropyrene and the re-
Haynes and co-workers (1975) proposed reaction path- verse reaction only. The equilibrium constant for the
ways similar to those suggested by Shabtai et al., but an hydrogenation was determined, from which the standard
additional reaction was included whereby 9,lO-dihydro- enthalpy of hydrogenation was estimated to be -10.6
phenanthrene was hydrogenated reversibly to give kcal/mol, in good agreement with Johnston's determina-
1,2,3,4-tetrahydrophenanthene; the reasons for including tion (-10 kcal/mol).
this reaction are not clear. Another investigation (Wu and From measurements of product selectivities at constant
Haynes, 1977) indicated the equilibrium limitation of the temperature and space velocity and various hydrogen
hydrogenation. Cracking of the saturated outer ring oc- partial pressures, Shabtai et al. (1978a) deduced that
curred at temperatures exceeding 427 O C , forming mostly 4,5-dihydropyrene was a primary product that was con-
n-butyltetralins; cracking at the central ring did not occur verted to asym-hexahydropyrene at higher pressure. The
appreciably. small yield of 4,5,9,10-tetrahydropyrene,a presumed hy-
9,10-Dihydrophenanthrene, 1,2,3,4-tetrahydro- drogenation product of 4,5-dihydropyrene, is in agreement
phenanthrene, and sym-odahydrophenanthene were also with results of Johnston and Stephens et al. and was at-
the predominant products at lower phenanthrene con- tributed to its instability arising from its two strained
versions, as observed by Lemberton and Guisnet (1984). hydroaromatic rings. Perhydropyrene formation was
These authors postulated that the decreasing yields of proposed to occur via hydrogenation of
9,lO-dihydrophenanthreneat higher phenanthrene con- 1,2,3,3a,4,5,5a,6,7,8-decahydropyrene, the latter being
versions were the result of conversion of the dihydro- formed by the presumed hydrogenation of asym-hexa-
phenanthrene to give 1,2,3,4-tetrahydrophenantheneand hydropyrene. Perhydropyrene was formed in appreciable
sym-octahydrophenanthrene; it is not evident why other amounts only at pressures greater than 137 atm, presum-
pathways (e.g., 9,10-dihydmphenanthrenedehydrogenation ably because the proposed hydrogenation of the hindered
to give phenanthrene) were ruled out. The higher yields aromatic ring in decahydropyreneis slow. The proposed
of cracking products (e.g., butylnaphthalenes) and isom- pathways are open to some question for the same reasons
erization products (having methylcyclopentane rings) as given earlier in reference to these workers' proposed
well as the absence of perhydrophenanthrene isomers at phenanthrene reaction pathways.
the higher conversions agree qualitatively with the results The results of these investigations all imply that equi-
of Haynes and co-workers. librium considerations take on increasing importance with
In summary, the proposed pathways suggest that 9,lO- an increasing number of rings in fused-ring aromatics.
dihydrophenanthreneand 1,2,3,4tetrahydrophenanthrene Thus, higher hydrogen partial pressures are essential for
are probably primary products. Interconversion of the converting pyrene (and higher molecular weight poly-
latter is possible but has not been established. The se- aromatics) appreciably, especially to give fully hydrogen-
lectivities to these products, the pathways to secondary ated products. These high pressures imply a need for
products, and the occurrence of central-ring opening all developing more active catalysts so that the thermody-
require further investigation. Kinetics data are especially namic advantage of lower temperatures can be realized.
needed. Furthermore, the predominant formation of 4,5-dihydro-
IIC4. Aromatics with Four Fused Rings. Quanti- pyrene implies that it is an appropriate model compound
tative networks have not been reported for hydrogenation for investigations of ring opening by dual-function cata-
of four-ring aromatic hydrocarbons. The reported inves- lysts. Also, the rapid hydrogenation to give dihydro speciea
tigations have focused on the determination of product may also apply to more highly fused aromatics (e.g., co-
distributions under various reaction conditions. ronene), and the rapid hydrogenation of such bulky species
Several investigations of pyrene hydrogenation have may be mass transfer controlled.
been conducted with thorough product analyses, but only IIC5. Five-Carbon-Membered-RingAromatics.
a few of these have been performed at high pressures with Quantitative reaction networks for the hydrogenation of
typical hydroprocessing catalysts (Table 11). Johnston fluoranthene and fluorene, each containing a five-carbon-
(1984) demonstrated that equilibrium was attained be- membered ring, have been reported (Lapinas et al., 1987,
tween pyrene and 4,5-dihydropyrene, the predominant 1991) (Table 11). The fluoranthene network proposed by
hydrogenation product at longer reaction times at 375 OC these investigators is shown in Figure 4 with representative
and 72.5 atm. At 425 OC, a constant concentration ratio values of the pseudo-first-order rate constants at 380 OC.
was attained but more rapidly and with a smaller value The dehydrogenations of 1,2,3,lObtetrahydrofluoranthene
than at the lower temperature. asym-Hexahydropyrene, and sym-decahydrofluorantheneto give fluoranthene and
an unidentified hexahydropyrene isomer, and small 1,2,3,1Ob-tetrahydrofluoranthene,respectively, were ki-
amounts of 4,5,9,10-tetrahydropyrenewere also detected. netically significant under these conditions. Hexahydro-
The reversibility of the pyrene hydrogenation reaction was fluoranthene and asym-decahydrofluoranthene were
c o n f i i e d by the formation of pyrene from its hydrogen- formed in appreciable amounts at the higher temperatures
ation products when the latter were allowed to react under only, as expected from the slow hydrogenation of mono-
nitrogen. aromatic moieties.
Stephens and co-workers (Stephens and Chapman, 1983; The network proposed for fluorene (Figure 4) involves
Stephens and Kottenstette, 1985)ale0 found that the yields the sequential hydrogenation of the aromatic rings. The
of pyrene and the major product, 4,5-dihydropyrene, ap- dehydrogenation reactions were apparently insignificant
proached constant values with increasing space time. under these conditions, in contrast to the situation for
Other hydrogenation products included 4,5,9,10-tetra- fluoranthene. The two isomerization products were formed
hydropyrene, sym-hexahydropyrene, and asym-hexa- only in trace amounts, even at the highest temperature
hydropyrene. Decahydropyrenes and perhydropyrenes (380 "0,and consequently rate constants were not esti-
were also formed, but in trace amounts. The latter mated for their formation. The slow hydrogenation of
products were not obeerved by Johnston, probably because fluorene relative to that of fluoranthene (to give
of the lower pressures employed in his work. Pseudo- 1,2,3,10b-tetrahydrofluoranthene)is consistent with the
Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2027

-
1.4

7.2
x

x
10.~

10.6 2.3 X 10
1.01 x 10.'
Table IV. Equilibrium Constants for the HDS of Selected
Organosulfur Compounds (Speight, 1981)
mom.'
kcal/m;i of
log K at organoeulfur
Y 2-propanethiol + H2 - 227 O C 427 O C
6.05 4.45
reactant
-13
propane + H a
thiacyclohexane + 2H2 - 9.22 5.92 -27
n-pentane + H2S
thiophene + 4H2
n-butane + H2S
- 12.07 3.85 -68

@Theenthalpies are assumed to be standard enthalpies of reac-


tion, although they are not listed as such by the author.

Table V. Reactivities of Several Heterocyclic Sulfur


1.3 x 10.41
Compoundso Determined by Nag et al. (1979)
pseudo-first-
order rate
const, L/
reactant structure (g of catdy8t.s)

15 x 10.41 thiophene 1.38 X lo-*

benzothiophene 8.11 X lo-'


053 V a
Figure 4. Reaction networks for fluoranthene and fluorene pro- dibenzothiophene 6.11 X lod
posed by Lapinas et al. (1987,1990) (Table 11). The numbers next
to the arrow8 are pseudo-firsborder rate constants in L/(g of cata-
1yst.s) at 380 O C . benzo[blnaphtho- 1.61 X lo-'
[2,3-d]thiophene
slower hydrogenation of monoaromatic moieties than of
naphthalene moieties (because the methylene group in 7,8,9,10-tetrahydro- 7.78 X lod
fluorene effectively destroys the coplanarity of its two benzo[blnaphtho-
aromatic rings). The relatively fast hydrogenation of the [2,3-d]thiophene
biphenyl-like 1,2,3,1Ob-tetrahydrofluorantheneto give Reaction conditions: batch reactor using n-hexadecane solvent
sym-decahydrofluoranthene in comparison with the hy- (0.25 mol % reactant concentration), 300 O C , 71 atm, Co-Mo/
drogenation of fluorene was attributed to the relaxation A1208 catalyst, each compound reacted individually.
of ring strain upon hydrogenation of 1,2,3,10b-tetra-
hydrofluoranthene. equilibrium constants all decrease with increasing tem-
Quantitative comparisons of the reactivities of fluorene perature, consistent with the exothermicity of HDS, and
and fluoranthene with those of fused-ring aromatics have approach values much less than 1 only at temperatures
not been reported. Such comparisons would be especially considerably higher than those required in practice.
instructive for determining whether the five-carbon-mem- Thermodynamics data for organosulfur compounds
bered ring hydrocarbons should be considered as a separate present in higher boiling fractions (Le., multiring hetero-
reaction class for purposes of modeling hydrogenation cyclics) are unavailable, except for recent data for di-
reactions of aromatic hydrocarbons. benzothiophene HDS (Vrinat, 1983). The latter results
indicate that dibenzothiophene HDS to give biphenyl is
111. Hydrodesulfurization (HDS) also favored at temperatures representative of industrial
HDS has long been a major refining process. One of its practice and is exothermic (AHo = -11 kcal/mol). Ex-
purposes is the removal of sulfur from naphtha reformer trapolation of the latter results suggests that the HDS of
feedstocks; another is the removal of sulfur from heavier higher molecular weight organosulfur compounds (e.g.,
fractions to minimize the emissions of sulfur oxides from benzonaphthothiophenes) is also favored.
refineries and power plants. With continuing interest in As will be discussed subsequently, sulfur removal occurs
the hydroprocessing of high-boiling petroleum fractions either with or without hydrogenation of the heterocyclic
(e.g., residua) and with increasingly stringent air-quality ring. The pathways involving prior hydrogenation of the
regulations demanding removal of even the least reactive ring can be affected by thermodynamics because hydro-
organosulfur compounds, recent research in HDS has been genation of the sulfur-containing rings of organosulfur
directed at the chemistry of thiophenic compounds, be- compounds is equilibrium-limited at practical HDS tem-
cause these are the least reactive organosulfur compounds peratures. For example, the equilibrium constant for
in petroleum and other fossil fuels (Gates et al., 1979). hydrogenation of thiophene to give tetrahydrothiophene
IIIA. Thermodynamics. The HDS of organosulfur is less than unity at temperatures above 350 O C (Vrinat,
compounds is exothermic and essentially irreversible under 1983). Thus, sulfur-removal pathways via hydrogenated
the reaction conditions employed industrially (e.g., 340-425 organosulfur intermediates may be inhibited at low pres-
"C and 66170 atm) (Gates et al., 1979; Speight, 1981; sures and high temperatures because of the low equilib-
Vrinat, 1983). Representative values of the gas-phase- rium concentrations of the latter species.
hydrodesulfurization equilibrium constants of several or- IIIB. Reactivities. The reactivities of a number of
ganosulfur compounds, including a mercaptan, a sulfide, organosulfur compounds, reported as the pseudo-first-
and a heterocyclic compound, are listed in Table IV with order rate constants for the disappearance of each at 300
the corresponding standard enthalpies of reaction. The OC and 71 atm, are given in Table V (Naget al., 1979);the
2028 Ind. Eng. Chem. Res., Vol. 30, No. 9,1991

experiments were conducted with a batch reactor and a Table VI. Reactivities of Selected Methyl-Substituted
Co-Mo/A1203 catalyst, with the reactants dissolved in Dibenrothiophener' As Determined by Houalla et al. (1980)
n-hexadecane solvent. With the exception of benzo[b]- pseudo-first-
naphtho[2,3-d]thiophene,the organosulfur reactants were order rate
converted to give hydrocarbon products, implying that the const, L/
reactant structure (g of catalystvs)
rates of disappearance are equal to the rates of sulfur
removal for these reactants. The three-ring compounds dibenzothiophene 7.38 x 106
are 1 order of magnitude less reactive than the two-ring
compounds, but the reactivity is similar for compounds . . ~-
having three or more rings. However, other patterns of l,&dimethyldibenzo- / 6.72 X loa
reactivity have also been reported. Kilanowski et al. (1978) thiophene
found approximately equal HDS reactivities for thiophene,
benzothiophene, and dibenzothiophene in experiments 3,7-dimethyldibenzo- 3.53 x 10-5
with a pulse reactor at 1 atm and 450 OC and with a Co- thiophene
Mo/A1203 catalyst. In experiments at 5-30 atm with
reactants in the vapor phase, Van Parijs et al. (1986)found 4,6-dimethyldibenzo- 4.92 X 10"
sulfur removal from thiophene to be substantially slower thiophene
than that from benzothiophene. Benzothiophene and
dibenzothiophene were found to have similar HDS re- 4-methyldibenzothio- 6.64 X 10"
activities in experiments with a trickle-bed reactor at 59 phene
atm with strong inhibitors such as quinoline present in the
feed (Rollmann, 1977).
The contrasting patterns of reactivities in HDS suggest Reaction conditions: flow reactor, n-hexadecane carrier oil,
that they depend strongly on the reaction conditions. HDS each reactant allowed to react individually at 300 "C at 102 atm in
rates are functions of several variables (Tables X and XI), the presence of a Co-Mo/A1203 catalyst.
including concentrations of organic reactants and products,
hydrogen, and hydrogen sullide. Comparison of reactivities slowly converted in the HDS of heavy fossil fuels. One of
should thus be made when inhibition by other organic the challenges for future technology is to find catalysts and
compounds is weak and when concentrations of hydrogen processes to desulfurize them. Compounds such as 4-
and hydrogen sulfide are similar to those used commer- methyldibenzothiophene are evidently the most appro-
cially. The data in Table V, referring to the reactants in priate model compounds for characterization of new cat-
a liquid phase, are probably most representative of in- alysts and processes for heavy feeds.
dustrial conditions. The observation that the reactivity of the methyldi-
The high-pressure reactivity data in Table V imply that benzothiophenes substituted at positions a to the sulfur
dibenzothiophene is one of the compounds most repre- atom is only 1 order of magnitude smaller than those of
sentative of the unreactive organosulfur compounds in the other substituted dibenzothiophenes (Table VI) in-
higher boiling fractions of fossil fuels. Because it is readily dicates that there is not a significant steric effect and
available commercially, it is a good model compound for therefore suggests that adsorption of the sulfur compound
characterizing the HDS chemistry of heterocyclic sulfur does not proceed in an end-on manner through the sulfur
compounds. atom (Lipsch and Schuit, 1969). If end-on adsorption
The effect of methyl substituents on the reactivity of predominated, the rate constant for HDS of 4,6-di-
dibenzothiophenehas been investigated by Houalla et al. methyldibenzothiophene would be expected to be several
(1979)(Table VI). The HDS of the reactants listed in orders of magnitude less than that for 4-methyldibenzo-
Table VI is described by pseudo-first-order kinetics. To thiophene. A multipoint adsorption model was therefore
a first approximation, the HDS pathways of the 2,8-and proposed by Kwart et al. (1980)involving the adsorption
3,7-dimethyldibenzothiophenesare as follows:
dimethyl-substituted dibenzothiophene
dimethyl-substituted biphenyl
-- of the organosulfur compound through the sulfur atom and
also through the C1-Cp bond (i.e., the bond between the
carbon atoms that are in the a-and @-positionsrelative
to the sulfur atom). Consistent with this picture, Sauer
dimethyl-substituted cyclohexylbenzene et al. (1989)proposed rekction mechanisms for thiophene
The methyl substituents in these positions decrease the HDS on the basis of presumed analogies to molecular
reactivity only slightly. organometallic chemistry.
In contrast, methyl substitution at the 4-position or at A multipoint adsorption mechanism is also consistent
the 4- and 6-positions results in an order of magnitude with the results of Singhal et al. (1981). The conversions
decrease in reactivity (Table VI). In addition, the yield of the hindered compounds tetramethylthianthrene and
of the respective substituted cyclohexylbenzene is higher tetraphenylthiophene (Table VII) were found to be similar
than that of the corresponding substituted biphenyl, in- to that of the unhindered thianthrene, which suggests that
dicating that methyl substituents in the 4- or the 4- and adsorption of the organosulfur compounds cannot involve
6-positions in dibenzothiophene cause hydrogenation to just the coordination of the sulfur atom; adsorption of the
occur to a larger extent than in the other substituted di- organosulfur reactants through their aromatic rings must
benzothiophenes before the sulfur is removed. Because also occur. Moreover, the sulfur atom in tetraphenyl-
hydrogenation of biphenyl is very slow under these con- thiophene is especially electron deficient because of the
ditions, the higher yield of cyclohexylbenzene implies that electron-withdrawing effect of the phenyl groups. Con-
hydrogenation of the substituted dibenzothiophenes oc- sequently, tetraphenylthiophene adsorption on the catalyst
curs, with the cyclohexylbenzene produced predominantly via the sulfur atom alone is even less likely. The higher
from the hydrogenolysis of the hydrogenated dibenzo- reactivity of the thianthrenes relative to dibenzothiophene
thiophenes. is attributed to the folded structure of the thianthrenes,
These methyl-substituted dibenzothiophenes are now which have more localized lone-pair electrons on the sulfur
recognized as the organosulfur compounds that are most atoms, which are in turn more exposed. Thus,coordination
Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2029

Q
---
Table VII. Conversions of Heterocyclic Organosulfur
Commundso Invertiloted by Sinehal et al. (1981)
96 / \r

thianthrene
reactant structure
conver-
sion p
\
1
/"
p\/
1,4,6,9-tetramethylthianthrene Figure 5. Thiophene reaction pathways suggested by early low-
pressure studies (Gates et al., 1979).

benzene with the reactivities of benzenes having a variety


of heteroatom substituents, including thiophenol and di-
tetraphenylthiophene phenyl sulfide. Compounds having electron-withdrawing
substituents such as sulfur underwent hydrogenolysis re-
actions preferentially, i.e., scission of the bond between the
substituent and the aromatic carbon atom predominated
over the hydrogenation of the aromatic ring; the hydro-
genolyses of thiophenol and diphenyl sulfide were more
dibenzothiophene 40-45 rapid than hydrogenation of benzene by more than 2 or-
ders of magnitude with a Ni-Mo catalyst at 340 O C and
70 atm. These results are in qualitative agreement with
O Reaction conditions: vapor-phase flow reactor, 310 "C,41 atm,
those of Broderick et al. (1982), who found that dibenzo-
space velocity of 6 g/(g of catalystah), Co-Mo/A1203catalyst with thiophene HDS was more rapid than benzene or biphenyl
large stoichiometric excess of hydrogen present. hydrogenation with a Co-Mo catalyst. Broderick et al. ala0
reported that benzonaphthothiophene conversion was
Table VIII. Comparison of Reactivities of Compounds with more rapid than hydrogenation of 2-phenylnaphthalene.
Folded Structums with that of Dibenzothiophene (Aubert However, benzonaphthothiophene conversion involved
et al., 1986)O
both hydrogenation and hydrogenolysis occurring in par-
re1 allel, and the rate of benzonaphthothiophene HDS was
pseudo-first-order comparable to that of 2-phenylnaphthalene hydrogenation.
reactant structure HDS rate const
Therefore, the results of Moreau et al. cannot be extended
thianthrene 12.6 to multiring aromatics.
IIIC. Reaction Networks and Kinetics. Experiments
leading to the proposal of reaction networks have been
phenothiazine H 11.7 carried out for thiophene, benzothiophene, dibenzo-
thiophene, and two isomers of benzonaphthothiophene. In
several investigations, the effects of the reactants and
products on the reaction rate were determined, leading to
phenoxathiin 10.8 the formulation of rate expressions.
IIIC1. Thiophene. Early thiophene HDS investiga-
tions, the great majority of which have been performed at
thioxanthene 9.2 atmospheric pressure, have been reviewed (Gates et al.,
1979; Vrinat, 1983). Thiophene HDS evidently proceeds
by two parallel pathways (Figure 5), although the reaction
dibenzothiophene 1.0 network is not completely understood, with results of
different investigators being in less than full agreement.
Rate equations from the low-pressure experiments indicate
O Reaction conditions: batch reactor, 340 O C , 70 atm, n-decane that thiophene inhibits its own HDS; in some investiga-
solvent, Ni-Mo/A120s catalyst, each reactant investigated sepa- tions, hydrogen sulfide has been found to inhibit HDS.
rately.
The possibility of inhibition by hydrogen is difficult to
of the sulfur atom is evidently important for the rapid assess because only small ranges of hydrogen partial
HDS of the thianthrenes. pressure have been used (Vrinat, 1983). Hydrogenation
The HDS reactivities of other folded thianthrene-like and hydrogenolysis are inferred to occur on two different
compounds containing different heteroatoms are summa- kinds of sites since hydrogen sulfide affects rates of
rized in Table VI11 (Aubert et al., 1986). The high re- thiophene hydrogenolysis and olefin hydrogenation dif-
activities of these folded compounds (in comparison with ferently (Gates et al., 1979).
that of the planar dibenzothiophene) are attributed to the In a thorough set of experiments characterizing HDS of
relatively small energy difference between the transition thiophene in the vapor phase (Van Parijs and Froment,
state and the adsorbed state of the folded molecules. The 1986),conditions were varied over a wide range (Table E).
higher reactivity of phenothiazine is inferred not to be a A reaction network was proposed in which thiophene is
consequence of a higher electron density on the benzenoid hydrodesulfurized to give 1-butene and cis- and tram-2-
rings [as had been suggested (Kwart et al., 1982; Geneste butene, followed by hydrogenation of the butenes to give
et al., 1983)] as its reactivity is about the same as those butane. Neither tetrahydrothiophene nor butadiene
of compounds such as thioxanthene that have nondonor (which was proposed to be hydrogenated rapidly to give
substituents. the three butenes) was detected. Two different kinds of
Some comparisons of the reactivities of organosulfur catalytic sites were assumed, one for thiophene hydro-
compounds with thoae of aromatic hydrocarbons have been genolysis and one for butene hydrogenation, in accordance
reported. Moreau et al. (1987) compared the reactivity of with results of low-pressure experiments (Vrinat, 1983).
2030 Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991

Table IX. Reaction Conditions for HDS Network Studies


reactant conditions reference
thiophene vapor-phase flow reactor, 260-350 "C, 2-30 atm, molar H2/thiophene V& Parijs and Froment, 1986
ratio of 4-9, Co-Mo/A120t catalyst
benzothiophene batch reactor, liquid phase, 200-400 "C, 86 atm, Co-Mo/Al203 Daly, 1978
catalyst
batch reactor, liquid phase, 200-300 "C, 50 atm, Co-Mo/A1203 Geneste et al., 1980
catalyst
vapor-phase flow reactor, 240-300 "C, 2-30 atm, H2/benzothiophene Van Parijs et al., 1986
mole ratio of 4-9, n-heptane solvent, Co-Mo/A1208 catalyst
dibenzothiophene two reactor systems with Co-Mo/A1203catalyst: (a) flow reactor, 300 Houalla et al., 1978
"C, 102 atm, n-hexadecane solvent (0.15 mol % reactant); (b) batch
reactor, 300 "C, 71 atm, n-hexadecane solvent (0.37 mol % reactant)
flow reactor, 275-325 "C, 177 atm, n-hexadecane solvent, 0.2-4 mol % Broderick and Gates, 1981
reactant, Co-Mo/A1203 catalyst
batch reactor, 350 "C, 35 atm, n-hexadecane solvent, Ni-Mo/A1203 Bhinde, 1979
catalyst
benzo[b]naphtho[2,3-d]thiophene batch reactor, 300 "C, 71 atm, n-hexadecane solvent, Co-Mo/A1203
~~
Sapre et al., 1979
catalyst
benzo[b]naphtho[ 1,2-d]thiophene batch reactor, 250 "C, 40 atm, Ni-Mo/A1203 catalyst Vrinat, 1983

Table X. Rate Equations for Thiophene Hydrogenolysis


and Butene Bydragenation Assuming Dissociatively
Adsorbed Hydrogen (VanParijs and Froment, 1986)
1. Thiophene hydrogenolysis on u sites:

rT,. =
~T,DKT,$HJ'TPH~

[1 + (KHg,8HJ1'* + K T . p T + K H f i , p H & u / P H 2 1 s
2. Butene hydrogenation on T sites:
1
kB,$8,$H~P$H~
rA,r W
[1 + (KH,?H2)"2 + KA,PA + K B , r h 1 2 Figure 6. Reaction network for benzothiophene proposed by Van
where the subscripta T , A, and B denote thiophene, butane, and Parijs et al. (1986). Reaction conditions are given in Table IX.
butene, respectively.
Parameter values at 260 "C are as follows: pressures (Vrinat, 1983). The reported high-pressure ex-
kT,, = 5.22 X 10' exp(-29.9/RT) 2.82 X lo4 periments with liquid-phase reactants (Table IX)lead to
K T ,=~ 6.60 X lo-' exp(lO.'l/RT) = 13.7 contradictory conclusions regarding the network; whereas
KH2,0 = 0.536 Daly (1978) reported that benzothiophene HDS gives
K ~ f i 91.2 ethylbenzene by either of two parallel pathways, Geneste
ke,? = 2.21 x 10" exp(-38.1/RT) = 5.16 x lo4 et al. (1980) reported that ethylbenzene is formed from a
Kkr = KB,, = 4.07 X lo-' erp(lO.6/RT) = 8.87 dihydrobenzothiophene intermediate. The only rate con-
KH~,,= 8.88 X lo-'' e~p(-28.4/RT) = 6.02 X
where the rate parameters k and adsorption parameters K have stants reported were for benzothiophene disappearance.
unita of mol/(g of catalysts) and atm-', respectively. The A quantitative network (Figure 6) has been reported for
adsorption parameter for hydrogen sulfide is dimeneionlees. conversion of vapor-phase benzothiophene (Van Parijs et
These resulta pertain to the temperature range 260-350 OC. al., 1986) at higher pressures (Table E). Styrene was not
detected and was assumed to be hydrogenated rapidly to
The data were correlated best by either of two sets of give ethylbenzene after being formed from bemthiophene.
Langmuir-Hinshelwood rate equations, one based on a As in the thiophene study reported by the same authors
rate-determining surface reaction between adsorbed (Van Parijs and Froment, 1986), two different types of
thiophene and molecularly adsorbed hydrogen and the catalytic sites were assumed: one for hydrogenolysis of
other differing only in that hydrogen is assumed to be bemthiophene and of 1,2-dihydrobenzothiopheneand one
diasociatively adsorbed, which is a more realistic postulate. for hydrogenation of benzothiophene. A sequential design
In each case, it was assumed that hydrogen sulfide was of experiments was used to minimize the number of data
formed by reaction of gas-phase hydrogen with adsorbed required for model discrimination and parameter estima-
sulfur by an Eley-Rideal mechanism. The rate equations tion. A best-fit rate equation (Table XI) was based on a
based on the assumption of dissociatively adsorbed hy- mechanism in which the surface reaction is the rate-lim-
drogen are given in Table X with the parameter values for iting step on both types of sites, with dissociative ad-
reaction at 260 OC. The existence of two different kinds sorption of hydrogen and the rate-limiting step involving
of sites was justified a posteriori by the large difference simultaneous addition of two hydrogen atoms. The values
between adsorption equilibrium constants for hydrogen of the parameters in Table XI indicate strong inhibition
on the two sites (Table X). The adsorption parameters of benzothiophene hydrogenation by ethylbenzene. Hy-
of the organic compounds exceed those for hydrogen by drogenation of benzothiophene is reversible. Inhibition
1-2 orders of magnitude. Hydrogen sulfide is a strong by hydrogen on the hydrogenation sites is negligible.
inhibitor of hydrogenolysis but does not inhibit hydro- Hydrogen sulfide inhibits hydrogenolysis, but not hydro-
genation. The data were also fitted to a model in which genation, as for thiophene.
the two kinds of sites are interconverted at a rate de- The functional forms of the rate equations based on the
pending on the hydrogen sulfide partial pressure, but the assumption of dissociative adsorption of hydrogen are
fit was not substantially better with this more complicated identical for bemthiophene and thiophene (Tables X and
model. XI). This consistency suggests equivalent mechanisms.
IIIC2. Benzothiophene. There have been few quan- Moreover, the adsorption parameters have similar values,
titative investigations of benzothiophene HDS at high which suggests that the parameters have a physical
Ind. Eng. Chem. Res., Vol. 30, No. 9,1991 2031
Table XI. Rate Equations and Parameter Values for
Benzothiophene Network (Figure6) Proposed by Van
P a d s et al. (1986)
~

1. Hydrogenolyeis of benzothiophene on u sites:


~ B , ~ B , J H ~ B B P H ~

[1 + ( K H s ~ H , ) " ' + ( K H f i , p H f i / P H d + KB,u(PB+ P D ) l 3


2. Hydrogenolyeis of 1,2-dihydrobenzothiopheneon u sites:
~D,~(B,JH~B#H~
rD,u =
[ l + ( K H ~ 8 H , ) ' / 2 + ( K H f i , p H f i / p H , ) + KB,u(PB+ P D ) l 3 1 1 x 10.4 \ J4 7 x 10.6

3. Hydrogenation of benzothiophene on T sites:

rB,r =
~+E,&E,#'BPH, - p ~ / K i )
(1 + K E , , ( ~+ B PD)+ KE,PEI'
0-0
where B, D,E, and K1denote, respectively, benzothiophene,
lI2-dihydrobenzothiophene, ethylbenzene, and the equilibrium
constant for benzothiophene hydrogenation giving
I sbw

1,2-dihydrobenzothiophene.
Parameter values at 260 "C are as follows:
G-0
k~,. 9.40 X l@exp(-17.6/RT) = 5.65 X lob Figure 7. Reaction network for dibenzothiophene propod by
k ~ , ,= 2.01 X 108 exp(-31.3/RT) = 2.89 X lob Houalla et al. (1980). Reaction oonditions are given in Table X The
numbers are the pseudo-first-order rate constants in L/(g of cata-
KB,, KD,, 19.3 1yst.s) at 300 O C .
KHfl.0 = 211
KH2,s = 0.358
k*~,, 8.84 X 10' exp(-24.1/RT) 1.14 X lob alyst than with Co-Mo/AlzO3 catalyst.
KB,? 2.05 A similar network for dibenzothiophene hydro-
KR, 294 desulfurization and hydrogenation was also reported
where the rate constants k have units of mol/(g of catalyst-s)and (Bhinde, 1979) for a Ni-Mo/Alz03 catalyst (Table 1x1.
the adsorption parameters K have units of atm-'. The rate The hydrogenolysis of dibenzothiopheneto give biphenyl
constant in the rate expression for benzothiophene was only 2.5 times more rapid than its hydrogenation. The
hydrogenation has units of mol/(g of catalyst-watm). These
results pertain to the temperature range 240-300 O C . lower HDS selectivity (increased hydrogenation) in this
network relative to that shown in Figure 7 was attributed
meaning in the context of Langmu+Hinshelwood models. to the greater hydrogenation activity of the Ni-Mo/A1203
Methyl substituents on benzothiophene reduce its rate catalyst in comparison with that of Co-Mo/Al2O9. How-
of hydrogenation to give the dihydro species (Geneste et ever, a contribution of a higher hydrogen concentration
al., 1980): in Bhinde's experiments (it was approximately 50% greater
than that used by Houalla et al.) to the higher selectivity
benzothiophene (1)> 2-methylbenzothiophene (0.4)> to hydrogenated dibenzothiophenes cannot be ruled out.
3-methylbenzothiophene (0.2) > The more rapid hydrogenation of dibenzothiophene sug-
2,3-dimethylbenzothiophene(0.1) gests that the contribution of this pathway contributes
more strongly at higher hydrogen concentrations, resulting
where the numbers in parentheses denote relative values in higher hydrogen consumptions for HDS at high pres-
of pseudo-first-order disappearance rate constants. The sures.
decrease in the rate is much less than expected from steric The kinetics of dibenzothiophene disappearance was
hindrance; the authors attributed the results to electronic investigated by several authors (Vrinat, 1983;O'Brien et
effects, as a good correlation was obtained between the rate al., 1986). Rate equations for dibenzothiophene disap-
constant for hydrogenation and the first ionization po- pearance were given, some associated with Langmuir-
tential. Hinshelwood mechanisms with hydrogen sulfide being an
IIIC3. Dibenzothiophene. Houalla et al. (1978)pro- inhibitor. However, as Figure 7 show, dibenzothiophene
posed a detailed network for dibenzothiophene HDS disappearance includes parallel hydrogenolysis and hy-
(Figure 7)on the basis of data from two laboratory reactor drogenation. In the work of Broderick and Gates (1981),
types (Table IX). Experiments were performed with di- rate equations were given for each pathway (Table XII).
benzothiophene and also with its products as reactants, Experiments were carried out with differential conversions
including lI2,3,4-tetrahydrodibenzothiophene, 1,2,3,4,5,6- of liquid-phase reactants in a flow reactor at 275-326 OC.
hexahydrodibenzothiophene,and biphenyl. Each of the The rates of hydrogenolysis and hydrogenation were taken
hydrogenated dibenzothiophenes was rapidly converted as the respective rates of formation of biphenyl and of the
to the other, and for purposes of a kinetics analysis, the lump consisting of cyclohexylbenzene and the two hy-
two hydrodibenzothiophenes were lumped. Figure 7 in- drogenated dibenzothiophenes. The rate equation for
dicates that dibenzothiophene conversion proceeds selec- hydrogenolysis (Table XII) corresponds to a Langmuir-
tively via the path of least hydrogen consumption and that Hinshelwood mechanism whereby the rate-limiting step
hydrogenation of biphenyl and cyclohexylbenzene is com- is a surface reaction between molecularly adsorbed hy-
paratively slow. However, the network shown is based on drogen on one type of site and dibenzothiophene on an-
data from a flow reactor in which the hydrogen sulfide other. Dibenzothiophene and hydrogen sulfide both in-
concentrations were low. The selectivity to the hydro- hibit hydrogenolysis, the latter more strongly, in agreement
genated dibenzothiophenes is higher when the hydrogen with the kinetics of thiophene and benzothiophene HDS
sulfide concentration is higher. The latter selectivity also (Tables X and XI). The adsorption parameter for hy-
depends on the catalyst composition, as Houalla et al. drogen could not be estimated accurately as a function of
found that the yield of cyclohexylbenzene at a given con- temperature (as evidenced by the positive enthalpy of
version was about 3 times higher with Ni-Mo/A1203 cat- adsorption). Biphenyl is not an inhibitor. With the ex-
2032 Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991
Table XII. Rate Equations for Dibenzothiophene
Hydrogenolyris and Hydrogenation Proposed by Broderick
and Gates (198l)O
1. Dibenzothiophene hydrogenolysis:

r=
~KDTKH,CDTCH,
(1 + K m C m + K H & H , s ) ~ (+~KH,CHJ
1 5.3 x 1 0 . ~
1 1.3 x 1 0 . ~
1 8.4 x 1 0 . ~

2. Dibenzothiophene hydrogenation:
k X'mKf~,CmCH2
r=
1 + K'mCm
(A) The numbers next to the arrows denote pseudo iirst-order rale constants
where the subscript DT denotes dibenzothiophene. in V(g of catalystas) at 308 C

Parameter values at 275 O C are as follows:


k = 7.87 X 105 exp(-30.1/R!?') = 7.67 x lo-'
KDT= 0.18 exp(4.5/RT) = 11
Ku. = 4 X 108 ex~(-8.4/RT)= 1.8
K Z = 0.70 exp(L5.3/RT) = 87
kf& = 4.22 X lo-' exp(-27.7/RT) = 2.78 X lo4
KtDT= 2.0 exp(l.l/RT) = 7.5
ONotes: (1) Reaction conditions are given in Table IX. (2) Rate
1 260 11.9

constants k are in mol/(g of catalystas); adsorption parameters K


are in L/mol. (3) The results pertain to the temperature range
275-325 "C.
ception of the dependence of the rate on hydrogen con-
centration, the equation is almost identical in form with (8) The numbers next to the arrows are relatlve values 01 the pseudo lirst.orde1
equations for hydrogenolysk of thiophene and benzo- rate constants at 250' C

thiophene (Tables X and XI). Figure 8. Reaction networks proposed for (A) benzo[b]naphtho-
The difference in form between the dibenzothiophene [2,3-d]thiophene by Sapre et al. (1980) and (B)benzo[b]naphtho-
hydrogenolysis rate equation and that for its hydrogenation [1,2-d]thiophene (Vrinat, 1983). Reaction conditions are given in
(Table XII)is consistent with the notion that the two types Table IX.
of reactions take place on separate kinds of catalytic sites.
Also, as stated above, hydrogen sulfide does not inhibit Another difference between the two benzonaphtho-
hydrogenation, in agreement with resulta stated earlier for thiophene networks concerns the relative rates of forma-
thiophene and benzothiophene. The difference in form tion of the hydrogenation products of the corresponding
between the dibenzothiophene hydrogenolysis equation phenylnaphthalenes. In the benzo[ b]naphtho[ 2,3-d]-
and the thiophene and benzothiophene hydrogenolysis thiophene network (Figure 8A), hydrogenation of 2-
equations raises a question of the nature of the hydrogen phenylnaphthalene to give 2-phenyltetralin is faster than
adsorption. It does not seem plausible that hydrogen is that giving 6-phenyltetralin, whereas in the benzo[ b]-
molecularly adsorbed, as the equation for dibenzo- naphtho[ 1,2-d]thiophene network, 5-phenyltetralin is
thiophene hydrogenolysis would imply if it were inter- formed more rapidly than 1-phenyltetralin from 1-
preted according to the LangmubHinshelwood formalism. phenylnaphthalene. No Langmuir-Hinshelwood rate ex-
The question of whether hydrogen and organic molecules pressions for reactions in any of the benzonaphtho-
compete for the same adsorption sites remains open. thiophene networks have been reported.
IIICI. Benzonaphthothiophene. Benzonaphtho-
thiophene is the heaviest organosulfur compound inves- IV. Hydrodenitrogenation (HDN)
tigated, with reaction networks having been reported for The increasing interest in converting petroleum residua,
benzo[b]naphtho[2,3-d]thiophene (Sapre et al., 1980) and coal, and shale to liquid fuels has spurred research in the
benzo[b]naphtho[l,a-d]thiophene (Vrinat, 1983) (Table M, chemistry of HDN. Nitrogen in these feedstocks is present
Figure 8). In both networks, the rate of hydrogenation predominantly in heterocyclic aromatic compounds
of the sulfur-containing reactant is comparable to that of (Katzer and Sivasubramanian, 1979). Nonheterocyclic
its hydrogenolysis. For benzo[b]naphtho[2,3-d]thiophene, organonitrogen compounds such as aliphatic amines and
the hydrogenolysis rate of the sulfur compound having a nitriles are also present, but in considerably smaller
saturated ring adjacent to the sulfur atom is only about amounts, and they are denitrogenated much more rapidly
4 times greater than that of benzo[b]naphtho[2,3-d]- than the heterocyclic compounds (Katzer and Sivasubra-
thiophene. However, for benzo[b]naphtho[l,2-d]- manian, 1979). Consequently, nonheterocyclic organo-
thiophene, the corresponding reactivity difference is more nitrogen compounds are less important for purposes of
than 2 orders of magnitude and thus comparable to that elucidating the nitrogen-removal chemistry occurring in
observed in the dibenzothiophene network. The reasons the hydroprocessing of heavy feeds.
for this difference are not known, but may be related to The nitrogen heterocyclic compounds in these feeds are
differing reaction conditions. classed as basic and nonbasic. In the latter (e.g., indole
The substantial ring hydrogenation that accompanies and carbazole), the lone-pair electrons on the nitrogen
hydrogenolysis of dibenzothiophene and benzonaphtho- atom are delocalized around the aromatic ring and are
thiophene explains in part why hydrogen consumptions unavailable for donation to a Lewis acid (Streitwieser and
of heavy feedstocks exceed the stoichiometric amount Heathcock, 1976). The opposite is true for basic hetero-
needed for sulfur removal in HDS (Gatea et al., 1979). The cyclic compounds such as quinoline and acridine.
relatively rapid hydrogenation of the benzonaphtho- IVA. Thermodynamics. Nitrogen removal from
thiophenes also suggests that these compounds are ad- heterocyclic organonitrogen compounds requires hydro-
sorbed flat on the catalyst to form r-bonded species to genation of the ring containing the nitrogen atom before
which hydrogen may be added. hydrogenolysis of the carbon-nitrogen bond occurs, at least
Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2033

Table XIII. Equilibrium Constantsa for Selected Table XIV. Reactivitieso for the HDN of Selected Basic
Hydrogenation, Hydrogenolysie, and Overall Nitrogen Heterocyclic Compounds (Mathur et al., 1982)
Nitrogen-Removal Reactions in the Quinoline and Indole pseudo-first-
Reaction Networks (Cocchetto and Satterfield, 1976; concn order rate
Cocchetto and Satterfield, 1981) in feed, const, L/
AHo,, compound structure wt % (g of catalystss)
kcal/iol
log K at of organic quin o1ine 1.0 9.39 x 10-4

- 300OC 4WoC reactant


quinoline + 2H2
-
1,2,3,4-tetrahydroquinoline
quinoline + 2H2
5,6,7&tetrahydroquinoline
1,2,3,4-tetrahydroquinoline+ 3H2 -
-1.4
4.7
-2.8
-3.2
-3.0
-5.4
-32
-41
-46
acridine

benz[clacridine
m 0.54 6.56

5.72
X

X
lo-'

lo-'
decahydroquinoline
5,6,7&tetrahydroquinoline + 3H2 - -3.5 -5.6 -37
Qc8
decahydroquinoline
1,2,3,4-tetrahydroquinoline+ H2 - 4.3 3.0 -23 0*47
o-propylaniline
decahydroquinoline + 2H2 - 6.3 7.9 -28
benz[a]acridine 0.42 4.03 X lo4

-
propylcyclohexane + NH8
o-propylaniline + H2 6.0 5.6 -7
quinoline + 4H2-
propylbenzene + NH3
propylbenzene + NH3
7.0 3.3 -65 dibenz[c,h]-
acridine
0.41 1.41 x 10-3

--
indole + H2 indoline -2.7 -3.3 -11

-
indoline + H2 o-ethylaniline
o-ethylaniline + H2
4.7
5.8
3.3
5.0
-25
-14 a Reactanta allowed to react separately in a batch reactor at 376

-
ethylbenzene + NH3
indole + 3H2 ethylbenzene + NH3 7.8 5.0 -49
a The values of the equilibrium constants in the table were read
OC and 136 atm using a Ni-Mo/A1203catalyst and paraffinic white
oil solvent.

from the van't Hoff plota given by the authors.


hydrogenolysis reactions were about 1order of magnitude.
More accurate data are available for the hydrogenations
with the catalysts now used, typically Ni-Mo/A1203. of heterocyclic compounds in the quinoline network; the
Hydrogenation of the heteroring is required to reduce the equilibrium constants were determined from experimental
relatively large energy of the carbon-nitrogen bonds in results obtained from a network study of quinoline HDN
such rings and thus permit more facile carbon-nitrogen (Satterfield and Cocchetto, 1981).
bond scission. The energies of carbon-nitrogen double Illustrative values are given in Table XI11 for selected
bonds and single bonds are 147 and 73 kcal/mol, respec- reactions occurring in the quinoline and indole reaction
tively (Streitwieser and Heathcock, 1976). If the carbon- networks. The data show that the equilibrium constants
nitrogen bond energy in a typical heterocyclic compound are less than unity for ring saturations and become smaller
is close to that of a carbon-nitrogen double bond, a reac- with increasing temperature, as the ring hydrogenations
tion breaking the carbon-nitrogen bond is expected to have are exothermic. The hydrogenolysis and overall HDN
a high activation energy. equilibria are favorable, however, even at temperatures as
The requirement that ring hydrogenation occur before high as 500 "C. Although the hydrogenation equilibria are
nitrogen removal also implies that the position of the unfavorable, the effect of higher hydrogen partial pres-
equilibrium of the hydrogenation reactions can affect ni- sures, such as those used in industrial hydroprocessing, is
trogen-removal rates if the rates of the hydrogenolysis to force the equilibria considerably toward the hydro-
reactions are significantly lower than the rates of hydro- processing products, and HDN is virtually irreversible
genation. An unfavorable hydrogenation equilibrium under typical reaction conditions.
would result in low concentrations of hydrogenated ni- In their thermodynamic analysis of the quinoline net-
trogen compounds undergoing hydrogenolysis; HDN rates, work, Cocchetto and Satterfield also calculated the dis-
at least at low surface coverages of organonitrogen com- tribution of the heterocyclics at equilibrium. From these
pounds, could thus be lowered. High hydrogen partial results, 5,6,7,8-tetrahydroquinolinewas found to be more
pressures can be used to increase the equilibrium con- favored thermodynamically than 1,2,3,4-tetrahydro-
centrations of saturated heteroring compounds to obtain quinoline. As shown below, the pathway for nitrogen re-
larger HDN rates. moval from 5,6,7,84etrahydroquinolineinvolves hydro-
Equilibrium constants for the gas-phase hydrogenation genation to give decahydroquinoline (i.e., perhydro-
and hydrogenolysis reactions of several heterocyclic orga- quinoline). The latter pathway involves a large hydrogen
nonitrogen compounds have been calculated (Cocchetto consumption for nitrogen removal in the quinoline net-
and Satterfield, 1976,1981); the reactants included pyri- work. Consequently, the thermodynamics favors a path-
dine, quinoline, isoquinoline, acridine, pyrrole, indole, and way for HDN that requires more than the minimum hy-
carbazole. The equilibrium constanta were calculated from drogen consumption.
the standard Gibbs free energies of formation at 1 atm. IVB. Reactivities. Because nitrogen heterocyclic
Group contribution methods were used to estimate the compounds must first be hydrogenated before the nitrogen
standard Gibbs free energies of formation of those com- is removed, there are numerous intermediate products that
pounds for which free energies were unavailable. Because contain nitrogen. Therefore, the conversion of a nitro-
the less accurate method of van Krevelen and Chermin gen-containing compound must be distinguished from ita
(1951) was used for estimating the standard Gibbs free HDN conversion (Le., that giving hydrocarbon products
energies for the heterocyclics, errors in the equilibrium and ammonia).
constants for ring saturation reactions were estimated to The rates of HDN of several basic nitrogen heterocyclic
be 2-3 orders of magnitude, whereas those for C-N bond compounds were determined (Mathur et al., 1982) (Table
2034 Ind. Eng. Chem. Res., Vol. 30, No. 9,1991

Table XV. Reactivitiesaof Quinoline and


Dimethyl-Substituted Quinolines (Bhinde. 1979)
H t
pseudo-first-
order rate
const, Lj I
compound structure (g of catalysbs)
quinoline
m 3.81 X 10”

-
2,6-dimethylquinoline
ax 3 x 10-5
+ NH3

Figure 9. Reaction pathways for pyridine HDN proposed by Han-


lon (1987). Reaction conditions are given in Table XVI.

m
2,7-dimethylquinoline 2 x 10-6

quinoline HDN being 4.19 X lo4 and 3.81 X L/(g of


2,8-dimethylquinoline

p 3 x 10-5

a Batch reactor, 350 O C , 34 atm, Ni-Mo/A120, catalyst, n-hexa-


catalyst-s), respectively. (The initial molar concentration
of indole was 10% leas than that of quinoline.) The similar
reactivities could arise because nonbasic organonitrogen
compounds are converted to basic ones upon hydrogena-
tion (see discussion of indole network below). However,
decane solvent. additional data are needed to resolve these reactivity
differences between basic and nonbasic organonitrogen
XIV) in experiments in which the concentrations of all the compounds, especially for higher molecular weight com-
organonitrogen species were measured. The rate of HDN pounds such as carbazole.
of each compound was pseudo-firsborder in the s u m of the The rates of conversion of three nonbasic nitrogen
concentrations of the reactant and its organonitrogen compounds, pyrrole, indole, and carbazole, were deter-
products. The comparison of the reactivities is rough mined with a batch reactor operated at 350 O C and 68 atm
because of differences in initial reactant concentrations, with reactants in n-hexadecane solvent (Stern, 1979). The
and these affect the HDN rate because of strong self-in- reactanta were present in a mixture in equimolar amounts.
hibition by nitrogen compounds (described below). The The relative first-order disappearance rate constants of
data show that the reactivities are the same within an order pyrrole, indole, and carbazole were 3.30, 1.16, and 0.12,
of magnitude; nitrogen-removal rates for three- and respectively, with a Ni-Mo catalyst. These rate constants
four-ring compounds are not markedly different from each indicate hydrogenation and not nitrogen removal. The rate
other. The intrinsic quinoline reactivity is greater than decreases with the degree of substitution on the nitro-
the value indicated in the table, as the initial quinoline gen-containing five-membered ring. No explanation was
concentration was higher than those of the other organo- offered for this effed; it is difficult to reconcile these results
nitrogen compounds. with those of Mathur et al. because the latter reported
Mathur et al. inferred that the rates of hydrogenation HDN reactivities.
were higher than those of C-N bond scission. The results IVC. Reaction Networks and Kinetics. Reaction
imply approximately equal HDN rates and suggest similar networks have been reported for pyridine, quinoline,
rates of the C-N bond scission reactions. Hence, the steric acridine, indole, and several anilines. The discussion be-
effects involved in the C-N scissions were inferred to be gins with pyridine and progresses to multiring Compounds.
nearly equivalent for all the nitrogen compounds listed in IVC1. Pyridine. Investigations of the network of va-
Table XIV. Reactions of basic nitrogen compounds with por-phase pyridine reacting in the presence of catalysts
Lewis acids are strongly affected by steric hindrance (Ho, in the sulfided (McIlvried, 1971;Hanlon, 1987)and oxidic
1988),and the equivalence of the steric effects implies that (Sonnemans et al., 1973)forms have been reported. Re-
end-on adsorption through the nitrogen atom does not action conditions for the experiments with sulfided cata-
occur since otherwise the HDN rates for the potentially lysts are shown in Table XVI. The results indicate that
more hindered compounds, such as benz[c]acridine and pyridine HDN occurs via the sequential pathway shown
dibenz[c,h]acridine, would be expected to have been orders in Figure 9. The side reaction involving alkyl transfer
of magnitude less than those for compounds such as occurs only to a minor extent (Sonnemans et al., 1973;
acridine. Hanlon, 1987). Pyridine hydrogenation is irreversible
The HDN of dimethylquinolines has been reported under these conditions, as evidenced by its absence as a
(Table XV). Pseudo-fmt-order behavior was followed for product in experiments with piperidine as the reactant;
the HDN of all the methyl-substituted quinolines. The high hydrogen partial pressures are used to counter the
nitrogen-removal rate is similar for each dimethylquinoline unfavorable hydrogenation equilibrium (Table XIII). The
and slightly less than that for quinoline. The similarity HDN of n-pentylamine is an order of magnitude more
of these HDN rates implies that the steric effects are rapid than that of pyridine.
equivalent for all these compounds. The latter conclusion Quantitative generalizations about pyridine HDN are
in turn implies that end-on adsorption through the ni- not warranted because of the different assumptions made
trogen atom is unlikely and suggests an adeorption through to model the kinetics. Exact comparisons of the HDN
the r-bonds of the aromatic rings, consistent with the rates measured in the various experiments (Table XVI)
results of Table XIV. are not possible because the rates were expressed some-
Comparisons of reactivities of basic and nonbasic orga- times in terms of catalyst weight and sometimes catalyst
nonitrogen compounds are few. From investigations of volume. The paragraphs below summarize the major
reaction networks of quinoline and indole conducted under conclusions of the different investigations.
conditions similar to those shown in Table XV, Bhinde From experiments with piperidine as the reactant,
(1979)found that the nitrogen-removal rates were similar, McIlvried (1971)found that its HDN followed pseudo-
with the pseudo-first-order rate constants for indole and first-order kinetics. The HDN rate decreased with in-
Ind. Eng. Chem. Res., Vol. 30,No. 9,1991 2035

Table XVI. h a c t i o n s Conditions in HDN Network Investigations


reactant conditions reference
pyridine vapor-phase flow reactor, 316 "C, 51-102 atm, pyridine, McIlvried. 1973
piperidine partial pressure 0.01-0.6 atm, Ni-Mo/AlZO3
catalyst
vapor-phase flow reactor, 310 "C, 12-99 atm, pyridine, Hanlon, 1987
piperidine partial pressure 0.3-0.6 atm, hydrogen sulfide
partial pressure 0.05-0.3 atm, Ni-Mo/Alz03 catalyst
quinoline batch reactor, 0.1-1 wt % quinoline in paraffinic white oil, Shih et al., 1977; Bhinde, 1979
300-375 "C, 21-136 atm, Ni-Mo/Alz03 catalyst
vapor-phase reactants, 330-420 OC, 35 and 69 atm, Satterfield and Cocchetto, 1981;
Ni-Mo/Alz03 catalyst Satterfield and Giiltekin, 1981
trickle-bed reactor, 1-5 wt % quinoline in paraffinic white oil, Satterfield and Yang, 1984; Yang
350-390 OC, 69 atm, Ni-Mo/Alz03 catalyst and Satterfield, 1984
trickle-bed reactor, 2-13 wt % quinoline in white oil, 330 and Miller and Hineman, 1984
375 "C, 69 atm, Co-Mo/Alz03 and Co-Mo/USY-Al,03
catalysts
batch reactor, 0.47-0.57 wt % quinoline in n-hexadecane Gioia and Lee, 1986
solvent, 350 "C, 10-150 atm, Ni-Mo/Alz03 catalyst
acridine batch reactor, 317-365 "C, 54-172 atm, white oil solvent, Zawadski et al., 1982
0.5-1.0 wt % acridine in feed, Ni-Mo/Al2O3 catalyst
5,6-benzoquinoline batch reactor, 330 "C, 79-171 atm, n-dodecane solvent, 0.03 Shabtai et al., 1988
mol/L reactant concentration, Co-Mo/Alz03 (3 and 6 wt %
Co), Co-Mo on fluorided AlzO3, Ni-Mo/AlzOs catalysts
batch reactor, 340 "C, 70 atm, n-decane solvent, 0.1 mol/L Moreau et al., 1988
reactant concentration, Ni-Mo/AlZO3 catalyst
7,8-benzoquinoline same aa for 5,6-benzoquinoline above Moreau et al., 1988
batch reactor, 275 "C, 170 atm, n-decane solvent, Malakani et al., 1987
Ni-Mo/AlzOB catalyst
batch reactor, 200-380 "C, 198 atm, 2 h reaction time, Shabtai et al., 1978b
Ni-W/A1203 catalyst
indole batch reactor, 350 "C, 34 atm, n-hexadecane solvent, 0.5 wt % Bhinde, 1979
reactant concentration, Ni-Mo/Alz03 catalyst
batch reactor, 350 "C, 68 atm, n-hexadecane solvent, 0.06-0.12 Stern, 1979
mol/L reactant concentration, Ni-Mo/AlZO3 and
Co-Mo/AlzOS catalysts
batch reactor, 250-350 "C, 34-70 atm, n-decane or n-dodecane Olive et al., 1985
solvent, Ni-Mo/A1203, Co-Mo/AlZO3,and Ni-W/Al2O3
catalysta
trickle-bed reactor, 250-350 "C, 69 atm, 0.001-0.0075 mol/L Odebunmi and Ollis, 1983
reactant concentration, Co-Mo/Alz03 catalyst
aniline and o-ethylaniline batch reactor, 320 and 350 "C, 34 atm, n-hexadecane solvent, Mathur et al., 1982b
1wt % reactant concentration, Ni-Mo/Alz03 catalyst
N-ethylaniline, diphenylamine, batch reactor, 350 "C, 59 atm, n-decane solvent, Ni-W/Al2O3 Finiels et al., 1986
dicyclohexylamine, N-phenylcyclo- catalyst
hexylamine, 3-methyldiphenylamine
creasing feed piperidine concentration, implying similar tions of the network shown in Figure 9,with all nitrogen
degreea of inhibition by the organic reactant and ammonia. compounds assumed to adsorb. From experiments per-
(Inhibition by n-pentylamine was neglected, presumably formed with pyridine, piperidine, and n-pentylamine as
because of its low surface coverage owing to its high re- reactants, adsorption parameters were determined for
activity.) Piperidine HDN was modeled by the equation ammonia (5000 atm-l), n-pentylamine (105 atm-l), piper-
idine (6 X lo4atm-'), and pyridine (3 X lo4 atm-'); these
values are larger by 1-2 orders of magnitude than those
found by McIlvried, despite the fact that experimental
conditions in the two investigations were similar (Table
where the subscript 0 denotes initial piperidine partial XVI). It is difficult to determine which set of values is
pressure. The adsorption parameter K, inferred to be the more reliable since different forms of rate expressions were
same for piperidine and ammonia, was equal to 23 atm-' used by Hanlon and McIlvried.
at 316 "C. Hydrogen and hydrogen sulfide have different effects
The pyridine hydrogenation rate decreased with con- on pyridine hydrogenation and piperidine hydrogenolysis.
version, consistent with product inhibition. Ammonia was Hanlon (1987)found that the rate constant for pyridine
assumed to be the sole inhibitor, but it is not clear why hydrogenation increased linearly with hydrogen partial
inhibition by piperidine was neglected. The following rate pressure, but that hydrogen sulfide partial pressure had
expression was used: no effect on the pyridine hydrogenation rate constant (the
pattern is similar to that for hydrogenation of dibenzo-
thiophene and other organosulfur compounds, described
above). The piperidine hydrogenolysis rate constant de-
creased 2-fold as the hydrogen partial pressure increased
with K' = 150 atm-' at 316 "C. The differences in the from 1 to 10 atm at a constant hydrogen sulfide partial
ammonia adsorption parameters in these two equations pressure of 0.08 atm. However, at constant hydrogen
suggest that hydrogenation and hydrogenolysis occur on partial pressure, the piperidine hydrogenolysis rate con-
different types of sibs. stant doubled as the hydrogen sulfide partial pressure
In contrast to McIlvried, Hanlon (1987)used a one-site increased by a factor of 6. An equilibrium involving two
Langmuir-Hinshelwood rate expression for all the reac- catalyst sites, hydrogen, and hydrogen sulfide was propoeed
2036 Jnd. Eng. Chem. Res., Vol. 30, No. 9, 1991

3 75 x 10.~

50x10

1 2ox10.6

1 fast
It

- All Pressures
0" Light
Hydrocarbons

- Low Pressures
4
_ _ _ _ - High Pressures
--------------______________.

Figure 11. Reaction network for quinoline proposed by Gioia and


Figure 10. Reaction network for quinoline HDN proposed by Sat- Lee (1986).
twFeld acd Yang (1984). The numbers next to the arrows are
pseudorate constants at 375 "C with 5 wt 9i quinoline and 0.59 w t lone-pair electrons; this interaction "ties up" the lone-pair
% CS3 iq the feed in units of mol/(g of cata1yst.s). electrons that may be involved in adsorption of 1,2,3,4-
tetrahydroquinoline (Shih et al., 1977). Most of the ni-
to account for these effects, but quantitative evidence was trogen removal therefore proceeds via decahydroquinoline,
not given. with the pathway through 5,6,7,8-tetrahydroquinoline
There is thus a need to resolve and quantify the effects dominating. The path of high hydrogen consumption is
of hydrogen and hydrogen sulfide more precisely by using thus favored kinetically.
Lmgmuir-Minshelwood rate expressions as was done for The activation energies of the hydrogenation reactions
hydrodesulfurization, although representing the inhibition are lower than those of the hydrogenolysis reactions (Shih
by hydrogen sulfide with such expressions may not be et al., 1977; Satterfield and Yang, 1984; Yang and Sat-
strictly valid (section VIA1). The applicability of one-site terfield, 1984). Under severe conditions (high space times
versus two-site expressions also needs to be addressed to and temperatures), high-molecular-weight products are
provide re!isb!e estimates of adsorption parameters. formed having structures that suggest the random linking
Qumtlfying these effects for the pyridine network would of fragments on the catalyst surface (e.g., 1,l'-bicyclohexyl).
be much simpler than for the networks of more complex At least some nitrogen removal from o-propylaniline ap-
polycyclic nitrogen compounds. The results for the simpler pears to require its hydrogenation to give a reactive pro-
compounds might then be used as guidelines for modeling pylcyclohexylamine intermediate, as the propylcyclo-
the kinetics of the more complex reactants. hexenes detected among the reaction products (Figure 10)
IVC2. Quinoline. The complex reaction network of are not formed from propylbenzene and propylcyclohexane
quinoline is more representative of those of higher mo- under the reaction conditions. Furthermore, the apparent
lecular weight organonitrogen compounds found in heavier activation energy for o-propylaniline conversion is close
hydroprocessing feeds (Le., those boiling above 340 "C) to the values for hydrogenation reactions; this result pro-
than that of pyridine. Most of the quantitative network vides another basis for suggesting that o-propylaniline
determinations with quinoline have been done for liquid- HDN proceeds via a hydrogenated intermediate.
phase reactants (Table XVI). The liquid-phase investi- There is disagreement regarding the kinetic significance
gatiom are emphasized below since the conditions are most of the dehydrogenations of the hydrogenated quinolines,
nearly comparable to those of hydroprocessing of com- with some investigators (Shih et al., 1977; Bhinde, 1979)
mercial feeds. having been able to correlate their data successfully by
The networks proposed by most investigators (Table neglecting the dehydrogenation reactions. Furthermore,
XVI)can be represented by that shown in Figure 10 and pathways for nitrogen removal in addition to those shown
summarized as follows. Quinoline HDN requires the prior in Figure 10 are claimed by Gioia and Lee (1986) (Figure
hydrogenation of one or both rings. Quinoline hydrogen- 11). The additional reactions shown in Figure 11giving
ation to give 1,2,3,4-tetrahydroquinolineis much more light hydrocarbons (which could not be analyzed because
rapid than that giving 5,6,7,8-tetrahydroquinolineand of their volatility) were assumed to occur because good
approaches equilibrium at lower quinoline conversions. mass balances could not be obtained at lower pressures.
The rapidity of 1,2,3,4-tetrahydroquinolineformation is (High-molecular-weightproducts, too heavy to be
associated with the high ?r-electron density of the quinoline could also explain the lack of closure of the mass balances.)
heteroring. Hydrogenolysis of 1,2,3,4-tetrahydroquinoline Mechanistic evidence was not offered by Gioia and Lee
to give o-propylaniline is slower than ita hydrogenation to (1986) for the reactions in which 1,2,3,4-tetrahydro-
give decahydroquinoline. The slow hydrogenolysis of quinoline was converted to propylbenzene and propyl-
1,2,3,4-tetrahydroquinolinecould be due to the resonance cyclohexane. However, good fits to the data were obtained
interaction between the benzenoid ring and the nitrogen- only when it was assumed that these two reactions oc-
Ind. Eng. Chem. Res., Vol. 30, No. 9,1991 2037
Table XVII. Rate Equations Proposed in Quinoline strong inhibition by the nitrogen compounds, as evidenced
Reaction Network InvestiirationtP by data obtained with varied feed quinoline concentrations.
1. Satterfield and Yang (1984);Yang and Satterfield (1984): No generalizations can be made concerning the rate form
kj,KjCj listed in Table XVII or the corresponding parameter
rij = values. In almost all cases,alternate rate forms in any one
K M C M + KSACSA+ KACA investigation were not considered. No single investigation
KdKM = has been reported in which the effects of concentrations
K A / K ~ 0.7 of the organic reactants, hydrogen, and hydrogen sulfide
The subscripts A, SA, and AA denote ammonia, were quantified, as with organosulfur compounds. Instead,
decahydroquinoline, and aromatic amines, respectively;
quinoline, 1,2,3,4-tetrahydroquinoline, each investigation typically focused on the effects of only
5,6,7,8-tetrahydroquinoline, and o-propylaniline comprise the a few of the foregoing variables on the rate. A synopsis
aromatic amines, and are assumed to have the same adsorption of the various rate forms is given below.
parameter KAA. CM denotes the s u m of the concentrations of Satterfield and Yang (1984) performed experiments with
the aromatic amines. Adsorption parameter ratios are
assumed temperature independent. each of the organonitrogen reactants shown in Figure 10
and were able to quantify the inhibition by each of the
2. Shih et al. (1977): organonitrogen intermediates in the network. A one-site
A. Hydrogenation Langmuir-Hinshelwood rate equation was used for both
hydrogenation and hydrogenolysis reactions (Table XVII),
but the analysis was limited to reactions involving orga-
nonitrogen compounds. The term of unity in the denom-
Adsorption parameter values at 342 "C are as follows: inator was neglected, as quinoline HDN followed zero-
K H ,= 0.05 atm-l order kinetics, probably because of the high surface cov-
KN = 6.4 X 108 g of oil/mol erage of the organonitrogen compounds under reaction
B. Hydrogenolysis conditions. All nitrogen compounds were assumed to ad-
sorb and were divided into three groups (Table XVII).
Only relative values of the adsorption parameters (mumed
to be independent of temperature) were determined,
where n equals 1 and 0 for the respective hydrogenolyses of making extension of the results to mixtures impossible.
1,2,3,4-tetrahydroquinolineand decahydroquinoline.
The secondary amine group (decahydroquinoline) was the
Adsorption parameter values at 342 OC are as follows: most strongly adsorbed, but the relative values for the
KH, = 0.05 atm-l several compounds are not greatly different, being equal
KN = 6.4 X l@g of oil/mol
within a factor of 3.
3. Gioia and Lee (1986): Satterfield and Yang also quantified the effect of hy-
ki jC j
1 I
drogen sulfide on hydrogenation and hydrogenolysis rate
rij = constants. (Hydrogen sulfide was produced in situ from
1 + KlCl + K2Cz
carbon disulfide added to the feed.) Hydrogenation re-
where the subscripts 1 and 2 denote quinoline and actions were inhibited slightly by hydrogen sulfide, as the
1,2,3,4-tetrahydroquinoline,respectively. rate constants decreased by about 10%. Hydrogenolysis
Adsorption parameter values at 350 O C are as follows: reactions, in contrast, were enhanced with increasing hy-
K 1 = Kz = l@ g of solution/mol drogen sulfide concentration, as' the rate constants in-
Variation of the rate constants with the hydrogen partial creased as much as 6-fold. These results are similar to
pressure is given by: Hanlon's (1987) for pyridine HDN. The different effect
of hydrogen sulfide on hydrogenation and hydrogenolysis
led the authors to propose that the two types of reactions
occurred on two different types of sites, but only one type
with m = 0 for dehydrogenation reactions and m = n = hydrogen of catalytic site was assumed in their kinetics model.
stoichiometric coefficient for hydrogenation reactions.
Two kinds of sites were assumed by Shih et al. (1977),
4. Miller and Hineman (1984): one catalyzing hydrogenations and the other hydrogen-
kCN olyses. Because the reactions of the quinoline network
followed pseudo-first-order kinetics at a given feed quin-
= 1 + KNCN
'" oline concentration, all nitrogen-containing species, in-
Respective adsorption parameter values at 330 and 375 O C are cluding ammonia, were assumed to have the same ad-
KN = (7.S9.6)X l@ g of solution/mol sorption parameter. The pseudo-first-order rate constants
KN = (1.7-2.3) X 1@ g of solution/mol were correlated with a Langmuir-Hinshelwood rate form
OThe subscripts i j denote reaction of component i in reaction j . (Table XVII). The 100-fold differences in adsorption
parameters for the nitrogen compounds on each site are
curred. Although the postulate of these reactions appar- consistent with the two-site assumption.
ently contradicts earlier results (Figure lo), the issue is Gioia and Lee (1986) analyzed thoroughly the effect of
open because previous workers did not provide quantita- hydrogen concentration on the rates of the various reac-
tive results pertaining to the rates of formation of the tions. The rate expression shown in Table XVII was fitted
individual hydrocarbons. Evidently not all the reactions to each reaction of their proposed network. Including other
shown in Figure 11 occurred at each pressure. The net- terms in the denominator did not improve the fit. Only
work reduces to one of two limiting forms at low (<31 atm) an order of magnitude estimate could be obtained for the
and high (>78 atm) pressures (Figure 11). adsorption parameters. The effect of hydrogen partial
The kinetics of the reactions in the quinoline network pressure on the rate of each reaction was estimated by
and the overall HDN have been modeled with Lang- fitting to determine each of the pseudorate constants kjj
muir-Hinshelwood rate expressions (Table XVII). as shown in Table XVII. The pseudorate constants cor-
Peeudo-first-order kinetics are inadequate because of the responding to hydrogenolysis reactions did not vary with
2038 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991
hydrogen partial pressure. In contrast, the pseudorate
constants corresponding to hydrogenation and dehydro-
genation reactions increased and decreased, respectively,
-
0 13

with hydrogen partial pressure. The latter results are in


qualitative agreement with those of Shih et al. for quinoline I 0.13

and those of Hanlon for pyridine HDN. The quinoline


HDN rate increased with hydrogen partial pressure, but
the hydrogen consumption for a given HDN conversion
also increased with pressure because of the increasing
contribution to the overall HDN of the pathway through
decahydroquinoline (Figure 11).
[m-
-W-T H - H

The rate expression proposed by Miller and Hineman


(1984) applies only to the overall quinoline HDN. Quin- I0.0190

i 0.0440

oline, 1,2,3,4-tetrahydroquinoline,and decahydroquinoline


were assumed to have the same adsorption parameter, but
adsorption by ammonia was neglected. In contrast to the
results of Satterfield and Yang (1984),Miller and Hineman
retained the term of unity in the denominator, even though
- Hydrocarbons
+ NY

Figure 12. Acridine reaction network proposed by Zawadski et al.


the quinoline concentration was comparable in the two (1982). The numbers next to the arrows are pseudo-first-orderrate
investigations. Adsorption parameter values for a zeolitic constants in (g of carrier oil)/(g of catalyst4 at 367 "C,137 atm, and
catalyst were within a factor of 1.3 of those for an 0.5 w t 76 acridine in the feed.
amorphous catalyst, and the values are within an order of
magnitude of those of Shih et al. and Gioia and Lee. petitive adsorption by hydrogen on hydrogenolysis sites
at the higher pressures. The HDN of acridine required
A comparison of quinoline HDN with reactants in the considerably higher pressures than that of quinoline; for
vapor and in the liquid phase is provided by the results example, at 342 "C and at approximately the same initial
of Satterfield and co-workers (references are cited in Table molar concentration of acridine and quinoline, a hydrogen
XVI). Although the product distributions were similar in partial pressure of 136 atm was required to obtain a value
the two cases, the adsorption parameters of the nitrogen of the pseudo-first-order rate constant for acridine HDN
compounds were grouped differently, with 1,2,3,4-tetra- equal to that for quinoline HDN at a hydrogen partial
hydroquinoline and decahydroquinolineclassified as sec- pressure of 34 atm.
ondary amines in the vapor phase. The ratios of KsA/KAA The acridine network has several features in common
and KA/KAA(terms are defined in Table X W )were found with the quinoline network. The primary hydrogenation
to be 6 and 0.25,respectively, with the vapor-phase reac- product formed by hydrogenation of the pyridine ring is
tants; for comparison, the values are 2 and 0.7 for the formed more rapidly than the primary product formed by
liquid-phase reactants, respectively. This comparison led hydrogenation of another ring, presumably because of the ,
to the proposal that the carrier solvent reduced the dif- higher *-electron density on the heteroring. At short r e
ferences in the adsorption parameters for liquid-phase action times, the ratio of the concentration of 9,lO-di-
reactants. hydroacridine to that of acridine was nearly constant,
IVC3. Acridine and Benzoquinolines. Reaction which suggests that the latter compounds were in equi-
networks have been reported for acridine, 5,6-benzo- librium. The slower hydrogenolysis of asym-octahydro-
quinoline, and 7,&benz~quinoline.The networks are more acridine relative to ita hydrogenation implies that the path
complex than those discussed above and include as many of low hydrogen consumption is not favored kinetically.
as six organonitrogen compounds and more than 12 re- Steric effects may account for the relative rates of re-
actions in some cases. Analysis of the products is also actions in the acridine network. Such effects are probably
difficult because of their larger numbers. Quantitative stronger in the acridine network than in the quinoline
treatments have been limited to the assumption of pseu- network because of acridine's three-ring structure. For
do-first-order kinetics in the organic reactant for a given example, the authors attributed the insignificantly slow
reaction, although the rates of individual reactions and formation of asym-octahydroacridine from 1,2,3,4-tetra-
overall nitrogen removal are surely strong functions of hydroacridine to the presence of the bulky fused cyclo-
organonitrogen reactant concentration (see above). hexane ring. The slow hydrogenation of sym-octahydro-
The acridine network proposed by Zawadski et al. (1982) acridine was ascribed to the steric hindrance arising from
(Table XVI, Figure 12) is the simplest one giving a good the two puckered cyclohexane rings fused to the central
fit to the data. Hydrocarbon HDN products were not ring.
identified. Pseudo-fmt-orderkinetics in the corresponding Two networks have been proposed for 5,6-benzo-
organic reactant was assumed for each reaction. Overall quinoline (Table XVI). Shabtai et al. (1989) reported the
nitrogen removal also followed first-order kinetics (in the more detailed of these (Figure13). Product identification
sum of the organonitrogen compound concentrations). was carried out by NMR, IR, and GC/MS. In contrast
Both hydrogenations and hydrogenolyses were kinetically to the HDN networks shown above, only one primary
important. The acridine HDN rate increased with hy- hydrogenation product, 1,2,3,4-tetrahydro-5,6-benzo-
drogen partial pressure at lower hydrogen partial pressures quinoline, was formed, that being the product of hydro-
(i.e., the HDN rate constant increased almost 6-fold as genation of the nitrogen-containing ring. The slow hy-
hydrogen partial pressure increased from 60 to 100 atm) drogenolysis of the latter was attributed to its strong
but approached a limiting value at higher pressures (the resonance-stabilized C-N bond (cf. the dow hydmgenolysis
HDN rate constant being only a factor of 1.3 greater at 170 of 1,2,3,4-tetrahydroquinoline).In contrast to the quino-
atm than at 100 atm). The authors attributed this be- line case, however, no aniline derivative8 were detected.
havior to two oppoaing effects: (a) an increasing reaction The dominant nitrogen removal pathway occurs via the
rate with increasing hydrogen concentration at the lower 1,2,3,4,7,8,9,10-octahydro-5,6-benzoquinoline and
pressure and (b) an apparent saturation because of com- 1,2,3,4,11,12,13,14-octahydro-5,6-benzoquinoline to give
Ind. Eng. Chem. Res., Vol. 30, No. 9,1991 2039

t 7 t Y
t 3

I7.2 X
I i3
9.3X10.6
4

Figure 14. Reaction network for 7,8-benzoquinolineproposed by


I B x 10.~ Moreau et al. (1988). The numbers next to the arrows are relative
values of the pseudo-first-order rate constants at 340 "C.

H
&
fast
@
H
3 4 2 X 10: x2
Figun, 13. Reaction network for 5,6-benz.oquinoline proposed by
Shabtai et al. (1989). The numbers next to the arrows are the
pseudo-first-orderrate constants in L/(g of catalystd at 300 "C. I
Hydrocarbons

propyltetralins. This is not the path of minimal hydrogen Other + NH3


consumption; it is analogous to the dominant nitrogen- Products

removal path (via decahydroquinoline) in the quinoline Figure 15. Indole HDN network proposed by Bhinde (1979) (Table
network at higher pressures. Apparent activation energies XVI). The numbers next to the arrows are the pseudo-first-order
were reported and, consistent with earlier results for rate constants in L/(g of catalysbs) at 350 "C.
quinoline (Shih et al., 1977; Satterfield and Yang, 1984;
Yang and Satterfield, 19841, the values for hydrogenation pressures, but a definitive assessment is not possible.
reactions (12-13 kcal/mol) were lower than those corre- A partidy quantitative network for 7,&benzoquinoline
sponding to hydrogenolysis reactions (24-49 kcal/mol), has also been reported by Moreau et al. (1988) (Figure 14).
except for saturation of the terminal monoaromatic ring. Three primary hydrogenation products were observed, in
In agreement with results of Satterfield and Yang (1984), contrast to the observations for 5,6-benzoquinoline, but
pseudo-first-order rate constants for hydrogenation and the dominant hydrogenation product was found to result
hydrogenolysis reactions decreased and increased, re- from hydrogenation of the pyridine ring, in agreement with
spectively, with increasing hydrogen sulfide concentration. earlier nonquantitative studies using 7,8-benzoquinoline
The data given above allow some conclusions about the (Shabtai et al., 1978b; Malakani et al., 1987). The relative
effects of catalyst promoters (Table XVI). Increasing the amounts of hydrocarbons formed differ among the various
cobalt content from 3 to 6 w t % gave higher ring hydro- investigations. Whereas propylnaphthalene was reported
genation activity but similar hydrogenolysis activity. At to be the hydrocarbon formed in greatest yield in two of
the same 3 wt % loading, the nickel-promoted catalyst was the investigations (Moreau et al., 1988; Shabtai et al.,
more active for hydrogenation but only slightly more active 1978b), propyltetralin was reported to be the major hy-
for hydrogenolysis than cobalt-promoted catalysts. Sur- drocarbon product in the other (Malakani et al., 1987).
prisingly, a fluoride-promoted Co-Mo catalyst did not This disagreement might again be attributed to the dif-
exhibit higher hydrogenolysis activity, but gave better fering hydrogen partial pressures and suggests a need for
hydrogenation activity than the same catalyst without investigationsquantifjing hydrogen partial pressure effects
fluorine. on product selectivities.
The network reported by Moreau et al. (1988) differs IVCI. Nonbasic Nitrogen Compounds. The only
from that shown in Figure 13 in that two primary hydro- nonbasic nitrogen compound for which quantitative re-
genation products were observed, one formed by hydro- action networks have been reported is indole; several
genation of the pyridine ring and the other by hydrogen- networks have been reported (Table XVI). The most
ation of the adjacent benzenoid ring. Hydrogenation of detailed of these was proposed by Bhinde (1979) (Figure
the terminal benzenoid ring was not observed. Also in 15). Pseudo-first-order kinetics was followed for both the
contrast to the results of Shabtai et al., the dominant overall indole HDN (in the s u m of the concentrations of
nitrogen removal pathway occurred via hydrogenolysis of the organonitrogen compounds) and for each reaction of
the 1,2,3,4-tetrahydro-5,6-benzoquinoline to give propyl- the network (in the respective organic reactant). The
naphthalene without an aniline intermediate. This rapid hydrogenation of indole to give indoline was rapid, and the
hydrogenolysis,which the authors found to be more rapid reaction attained virtual equilibrium; in the kinetics
than that of anilines, was attributed to the weaker reso- analysis, the latter two compounds were lumped. No or-
nance interaction between the nitrogen atom's lone-pair ganonitrogen products were detected in which the ben-
electrons and the aromatic ring arising from the puckered zenoid ring was saturated, which implies the preferential
configuration of the terminal saturated ring. The differ- hydrogenation of the nitrogen-containing ring, as for the
ences in the latter network and that shown in Figure 13 other organonitrogen compounds. Two hydrocarbon
could be attributed to the different hydrogen partial products were detected, ethylcyclohexane and ethyl-
2040 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

benzene, and these were also lumped in the kinetics


analysis. Cycloolefins were not detected although their
formation is expected on the basis of the quinoline network
investigations carried out under similar conditions. The
initial rate of formation of ethylcyclohexane could not be O N H *0S *-0 7.5 x 104

accounted for by the rate of disappearance of o-ethylaniline 2.86 x 10-6


+ NH3
alone. A second pathway was therefore assumed whereby
ethylcyclohexane was formed from the indole/indoline
pseudocomponent by way of a postulated perhydroindole r 1
intermediate (which was not detected); such a pathway is
analogous to the formation of propylcyclohexane via de-
cahydroquinoline in the quinoline network (Figure 10). In
addition, a reaction of indole to high-molecular-weight
products was assumed, as the material balance closure with
the products detected was low. Figure 16. (A) Reaction network determined by Mathur et al.
Stern (1979) and Olive et al. (1985) proposed indole (1982b) for aniline HDN (Table XVI). The numbers next to the
reaction networks similar to that shown in Figure 15. In mows are pseudo-first-order rate constants in L/(g of catalystd at
350 "C. (B) Suggested mechanism for aniline HDN.
contrast to Bhinde, the latter authors proposed that the
hydrocarbon products were formed exclusively through
o-ethylaniline. However, as in Bhinde's study, the rates mediates may occur. The critical parameter appears to
of formation of the hydrocarbons were not determined. In be the hydrogen concentration. Mathur et al. (1982b)
both investigations, ethylbenzene was formed in consid- examined the HDN of several anilines (Table XVI). The
erably lower yields than ethylcyclohexane, but because of proposed reaction network for the HDN of aniline (Figure
the slow hydrogenation of ethylbenzene under the con- 16A) shows that the primary product was an unsaturated
ditions employed, ethylcyclohexane was assumed to be hydrocarbon. The primary hydrogenolysis products of
formed via 2-ethylcyclohexylamine and ethylcyclohexene, o-ethylaniline and diphenylamine were also unsaturated
a reaction analogous to the deamination of o-propylaniline products; o-ethylaniline gave ethylbenzene, whereas di-
(Figure 10). Olive et al. detected neither ethylcyclohexene phenylamine gave benzene and aniline. The results in-
nor 2-ethylcyclohexylamine, but ethylcyclohexenes were dicate that full hydrogenation of the ring was not necessary
detected in Stem's experiments. Stem also demonstrated for nitrogen removal. However, the authors proposed that
that 2-ethylcyclohexylaminewas converted rapidly under the formation of the primary aromatic products did not
reaction conditions to give ethylcyclohexene, ethylbenzene, occur as a result of the hydrogenolysis of the C-N bond,
and ethylcyclohexane. The reversibility of the indole but instead that a dihydro intermediate was formed from
hydrogenation to indoline was substantiated in both in- aniline. The formation of this intermediate presumably
vestigations by experiments in which indoline was the destroyed the aromatic resonance interaction with the
reactant; by the time the batch reactor had been heated lone-pair electrons of the nitrogen atoms, thereby weak-
to reaction temperature, at least some indoline was de- ening the C-N bond. The scission of the C-N bond oc-
hydrogenated to indole, even in the absence of catalyst. curred rapidly from the dihydro intermediate as a 6-elim-
The indole-indoline equilibrium was also observed in the ination (Figure 16B), driven by the regaining of the aro-
trickle-bed experiments of Odebunmi and Ollis (1983). matic resonance energy upon formation of the aromatic.
Different conclusions were reached, however, by Finiels
Quantitative network information for higher molecular et al. (1986), who examined the HDN of several aniline
weight nonbasic nitrogen compounds is lacking. Some derivatives and secondary amines (Table XVI). In their
information characterizing the HDN pathways of 3- experiments with diphenylamine, the following products
ethylcarbazole has been reported (Ho, 1988) on the basis were formed: aniline, benzene, cyclohexene, N-phenyl-
of experiments with a Ni-Mo/A1203 catalyst at 390 "C, 69 cyclohexylamine, and cyclohexane, with the latter formed
atm, and 1.4 LHSV. The nitrogen-containing compounds in substantial amounts. The concentration of benzene was
formed in highest yields are hydrogenated 3-ethyl- considerably lower than that of aniline at shorter reaction
carbazoles, with tetrahydro-3-ethylcarbazolebeing the times, which implies that benzene and aniline were not
dominant product, which suggests that hydrogenation of formed exclusively by the pathway involving the dihydro
the latter is slow. Neither amines nor anilines were formed intermediate proposed by Mathur et al. (If aniline and
in appreciable yields, which implies that they underwent benzene had been formed only via the dihydro interme-
relatively rapid hydrogenolysis. diate, their concentrations would have been equal at the
IVC5. Aniline Derivatives. Aniline derivatives are short reaction times.) The high aniline concentration
present in almost all the HDN networks of nitrogen het- relative to that of benzene further demonstrates that the
erocyclics. The reactions removing nitrogen from the an- dihydro pathway was not the only one operative, since
iline derivatives are among the most important in the aniline was hydrogenated more rapidly than benzene. The
networks because the pathways involving these reactions authors therefore proposed that the predominant pathway
are the ones requiring the lowest hydrogen consumptions. for nitrogen removal from aniline derivatives required f d
Furthermore, the direct hydrogenolysis of the C-N bond hydrogenation of one of the aromatic rings before breaking
in anilines is believed to require prior hydrogenation (Oliv6 of the C-N bond.
et al., 1985; Satterfield and Yang, 1984; Stern, 1979), but The greater importance of the pathway requiring full
it is not clear whether full hydrogenation of the aromatic hydrogenation observed by Finiels et al. is probably a
ring bonded to the amine group is required. The hydrogen consequence of the higher hydrogen partial pressures a p
required per mole of nitrogen removed (i.e., the total hy- plied in their experiments (59 atm vs 34 atm for M a t h
drogen consumption) depends strongly on whether full et al.). Higher pressures would lead to faster hydrogena-
hydrogenation of the aromatic ring occurs. tion of the benzene ring and thus result in a smaller con-
Investigations of the HDN of anilines indicate that both tribution of the pathway involving the dihydro interme-
direct hydrogenolysis and HDN via hydrogenated inter- diate. Thus,in agreement with the conclusion of Gioia and
Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2041

Table XVIII. Equilibrium Constants and Standard Table XIX. Pseudo-First-OrderRate Constants k i n L/(g
Enthalpy of Reaction for the HDO of Organooxygen of catalyst s) for Disappearance of Components in Weak
Compounds (Edelman et el., 1988: Furimsky, 1983b) Acid Fraction of SRC-I1 Coal Liquid (Li et al., 1988; Grandy
AH0,? et al., 1986)O
kcal/mol of k detd
log K at organic by high-
350°C 400°C reactant k detd resoln
furan + 4Hz 7 11.4 9.2 -84 compound byGCb GC/MS'
n-butane + HzO
tetrahydrofuran + 2Hz - 11.4 10.2 -20
5,6,7,8-tetrahydro-l-naphthol
2-phenylphenol
1.9 X lo-' 4.2 X lo-'
8.3 X lod 7.2 X lod
n-butane + HzO
benzofuran + 3H2 -
ethylbenzene + H20
10 9.3 -25
methylphenylphenol
4-cyclohexylpi~enol
dimethyldihydroxyindan
1.5 X lo-' 2.0 X lo-'
4.4 X lo-' 6.2 X lo-'
2.3 X lo-'
methyltetrahydronaphthol 4.5 x 104
OThe values of the standard enthalpies of reaction in the table methylphenyl or methylphenoxybenzene 1.4 X lo-'
were read from the van't Hoff plots given by the authors. 1-naphthol 5.4 x 10-4
phenolic oxygen in weak acid fractiond 1.6 X lo4
Lee (1986) for the quinoline network, higher hydrogen
partial pressures result in higher hydrogen consumption "Experiments performed with a fixed-bed flow reactor at
300-350 "C and 120 atm with a Ni-Mo/Alz03 catalyst. The reac-
per mole of nitrogen removed. tants were dissolved in cyclohexane saturated with hydrogen a t
room temperature and ca. 100 atm. *Li et al., 1985. 'Grandy et
V. Hydrodeoxygenation (HDO) al., 1986. Determined by infrared spectroscopy.
Relatively few investigations of HDO have been carried
Table XX. Pseudo-First-Order Rate Constants for
out because the oxygen content of petroleum is consider- Disappearance of Methyl-Substituted Phenols in HDO
ably less than the concentrations of sulfur and nitrogen
(Allen et al., 1985; Whitehurst, 1978). However, since t", k, L/(g
reactant "C of catalyst-e) reference
oxygen is the most abundant heteroatom in many coal
liquids, hydrodeoxygenation is expected to assume in- 2-methylphenol" 350 1.6 X lo-' Odebunmi and
Ollie, 1983a
creasing importance. Organooxygen compounds in com- 3-methylphenol 350 3.1 X lo-'
mercial hydroprocessing feedstocks fall predominantly into 4-methylphenol 350 9.4 X lo4
two classes: (a) phenol and naphthol derivatives and (b) 2-methylphenolb 300 2.4 X lob Gevert et al., 1987
heterocyclic oxygen compounds (Le., furan and its deriv- 4-methylphenol 300 4.9 x 10-6
atives) (Allen et al., 1985; Whitehurst, 1978). Alcohols, 2,4-dimethylphenol 300 2.8 X low6
carboxylic acids, and ketones are also present, but in 2,6-dimethylphenol 300 6.3 X 10"
2,4,64rimethylphenol 300 8.3 X lo4
smaller amounts (Allen et al., 1985).
VA. Thermodynamics. Equilibrium constants for the OTrickle-bed reactor, 350-400 OC, 68 atm, n-hexadecane solvent,
gas-phase HDO of furan and tetrahydrofuran (Furimsky, aged Co-Mo/A1203 catalyst. bBatch reactor, 300 "C, 50 atm, n-
1983b) and reactions in the benzofuran HDO network hexadecane solvent, Ni-Mo/Alz03 catalyst.
(Edelman et al., 1988) have been calculated from ther- having a pseudo-first-order rate constant for HDO of 10
mochemical data (Reid et al., 1987; Stull et al., 1969) and relative to that for naphthalene hydrogenation. The rel-
group contribution methods (Benson et al., 1969). Rep- ative rate constanta for 4-methylphenol,2-ethylphenol, and
resentative results (Table XVIII) indicate that, at typical 2-phenylphenol were 5.2, 1.2, and 1.4, respectively, indi-
hydroprocessing temperatures, the exothermic HDO is cating lower reactivities for the ortho-substituted phenols.
virtually irreversible. However, the hydrogenation of or- The benzofuran reactivity (1.0) was comparable to that of
ganooxygen compounds, which may precede oxygen re- the substituted phenols. Dibenzofuran was the least re-
moval, is reversible; for example, the equilibrium constant active, with a relative rate constant of 0.4.
for the reaction
-
benzofuran + H2 2,3-dihydrobenzofuran
is 0.4 at 300 O C . Equilibrium constants for higher mo-
The rates of disappearance of several phenolic com-
pounds present in the "weak acid" and "very weak acid"
fractions of a SRC-I1 coal liquid, prepared by separating
the coal liquid by liquid chromatography (Petrakis et al.,
lecular weight organooxygen compounds have not been 19831, have also been reported (Li et al., 1985b; Grandy
reported. Group contribution methods can be used to et al., 1986). The reactivities of a number of compounds
obtain rough estimates. The group contribution method identified by GC/MS were determined by GC and high-
of Benson et al. (1969) has been adapted for pyrans (Shaw resolution GC/MS (Table XIX), and the rate of disap-
et al., 1977). pearance of the phenolic hydroxyl groups was determined
VB. Reactivities. Reactivities of single organooxygen by infrared spectroscopy. The results indicate that the
compounds in the absence of other compounds have not reactivities of the different phenolic compounds are the
been reported. However, the reactivities of organooxygen same within one order of magnitude. The ortho-substi-
compounds present in mixtures with other types of com- tuted 2-phenylphenol is less reactive than the para-sub-
pounds have been determined. The results indicate that stituted 4-cyclohexylphenolby a factor of 5-10; the latter
benzofuran and dibenzofuran are much less reactive than uncertainty shows the difficulty in quantifying relative
phenols and that substituent effects are important in the reactivities using even well-characterized commercial feeds.
HDO of the latter. The lower reactivity of ortho-substituted phenols relative
The reactivities of substituted phenols, cis-2-phenyl-l- to para- and meta-substituted phenols has also been ob-
hexanol, benzofuran, and dibenzofuran in mixtures in- served in other investigations (Table XX). The presence
cluding sulfur and nitrogen compounds and aromatic hy- of an additional methyl substituent ortho to the hydroxyl
drocarbons have been determined (Rollmann, 1977) in group reduced the reactivity further. The results of Table
experiments with a trickle-bed reactor and Co-Mo/A120B XX also show that methyl substitution at an unhindered
catalyst at 344 "C and 50 atm. cis-2-Phenyl-1-hexanolwas position increased the reactivity somewhat. [The rate
by far the most reactive oxygen compound among these, constants reported by Odebunmi and Ollis (1983a) in
2042 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

Table XXI. Reaction Conditions in HDO Network Investigations


reactant conditions reference
1-naDhthol batch reactor. 2W275 "C. 22-34-atm H, Dartial Dreesure. n-hexadecane Li et al.. 1985a
solvent, reactant concentration 0.016 gjL, Ni-M0/Alzd3 catalyst
4-methylphenol batch reactor, 300 "C, 50 atm, n-hexadecane solvent, Co-Mo/A&OS catalyst Gevert et al., 1987
2-hydroxydiphenylmethane batch reactor, 300-330 "C, 69 atm, n-hexadecane solvent, reactant Petrocelli and Klein, 1987
concentration 0.06 mol/L, Co-Mo/Al9O1 catalyst
benzofuran vapor-phase flow reactor,'3W400 "C, 3b k m , N&Mo/A120a catalyst Edelman et al., 1988
trickle-bed reactor, 220-340 "C, 69 atm, 0.15 mol/L reactant concentration, Lee and Ollis, 1984a
Co-Mo/AlZO3 catalyst
dibenzofuran batch reactor, 343-376 "C, 105 atm, n-dodecane solvent, Ni-Mo/A1203 Krishnamurthy et al., 1981
catalyst
vapor-phase flow reactor, 350-390 "C, 69 atm, n-hexadecane solvent, reactant LaVopa and Satterfield, 1987
partial pressure of 0.24 atm, Ni-Mo/AlZO3 catalyst
diphenyl ether batch reactor, 280-320 "C, 69 atm, n-hexadecane solvent, 0.05-0.7 mol/L Petrocelli and Klein, 1987
reactant concentration, Co-Mo/A1203 catalyst

Table XX are smaller by 2-3 orders of magnitude than


those reported by Gevert et al. (1987); such a large dis-
crepancy is surprising considering the similarity of the
reaction conditions and may be due to the aged catalyst
used by Odebunmi and Ollis.]
Tables XIX and XX indicate that the reactivity dif-
ferences between ortho- and para-substituted species are
within a factor of 3. [The differences are smaller than the
order of magnitude difference observed between the HDS
of dibemthiophene and 4methyldibemthiophene (Table
1.40 x 1 0 . ~ 1.16X10.7
VI).] One might expect that if oxygen removal from or-
ganooxygen compounds occurred solely by coordination t A 8 " //
of the oxygen atom with the catalyst surface, larger re-
activity differences should result (see section IIIB). Crackiryl
Products
W
The electronic effect of the hydroxyl substituent on the
reactivity is substantial, however. Moreau et al. (1987) Figure 17. Reaction network for 1-naphthol proposed by Li et al.
(1985a) (Table XXI). The numbers next to the arrows are pseudo-
compared the reactivities of phenol and diphenyl ether first-order rate constants in L/(g of catalystes) at 200 "C.
with that of benzene (they used a Ni-Mo/A1203 catalyst
and a batch reactor at 340 O C and 70 atm) and found that 5,6,7,8-tetrahydro-l-naphthol was demonstrated in ex-
phenol was by far the most reactive compound. Whereas periments with the latter as a reactant; these experiments
benzene was hydrogenated to give cyclohexane with a also showed that naphthalene was formed by tetralin
relative rate constant of 1,phenol reacted by two parallel dehydrogenation. Decalin formation directly from
routes to give benzene (relative rate constant of 1) and 5,6,7,8-tetrahydro-l-naphthol (expected on the basis of the
cyclohexanol (relative rate constant of 16), with the latter rapid hydrogenation of phenols reported above) was not
hydrodeoxygenated rapidly to give cyclohexane. The rapid considered. Unidentified products formed in small yields
hydrogenation of phenol to give cyclohexanol was attrib- were all assumed to be cracking products.
uted to the electron-donating hydroxyl substituent. Ox- Apparent activation energies were estimated for some
ygen removal from diphenyl ether also proceeded by of the reactions. The values for two hydrogenolysis re-
parallel pathways, but because the substituent was less actions, 1-naphthol giving naphthalene and 5,6,7,&tetra-
electron donating, hydrogenation and hydrogenolysis rates hydro-1-naphthol giving naphthalene, were larger than
were similar (relative rate constant of 3 for each). those corresponding to the hydrogenation of 1-naphthol
VC. Reaction Networks and Kinetics. Reaction to give either 1-tetralone or 5,6,7,&tetrahydro-l-naphthol
networks have been reported for both phenolic and furanic (32-34 vs 11-24 kcal/mol); the trend is similar to that
organooxygen compounds. The networks for the former observed for quinoline HDN. Thus, lower hydrogen con-
class are generally simpler and have a greater degree of sumptions for HDO could occur at higher temperatures.
quantitative detail. The networks indicate that oxygen The apparent activation energy for the pathway leading
removal can occur either with prior ring saturation (as in to the formation of the presumed cracking products was
HDN) or directly (HDO), suggesting the need for devel- uncharacteristically low (18 kcal/mol) for hydrogenolysis
oping catalysts selective for the latter pathway to reduce reactions, suggesting that the assumed pathway may be
hydrogen consumption. incorrect. The appreciable tetralin dehydrogenation rate,
VC1. Phenolic Oxygen Compounds. Li et al. (1985a) the higher yield of trans-decalin than of cis-decalin, and
proposed a reaction network for the HDO of 1-naphthol the much faster hydrogenation of naphthalene than of
(Table XXI, Figure 17). Oxygen removal occurred via four tetralin are all consistent with reported naphthalene
pathways, including direct oxygen removal and ring hy- networks (Bhinde, 1979; Sapre and Gates, 1981).
drogenation prior to oxygen removal. The fastest deoxy- A reaction network for the HDO of 4-methylphenol
genation pathway at 200 O C occurred through a 1-tetralone (Table XXI, Figure 18A) has been reported (Gevert et al.,
intermediate and required 3 mol of H2/mol of 0 removed, 1987). As for 1-naphthol, the pathways include direct
implying high hydrogen consumptions for HDO. Deoxy- oxygen removal as well as ring saturation before hydro-
genation of 1-tetralone to give tetralin was assumed. The genolysis. However, an additional pathway was proposed
reversible formation of 1-tetralone was presumed to pro- leading to the direct formation of a nonaromatic hydro-
ceed by hydrogenation of 1-naphthol to give an undetected carbon (methylcyclohexane)via hydrogenated phenols.
dihydro intermediate followed by a rapid keto-enol in- The partially hydrogenated phenols (Le., methylcyclo-
terconversion, giving 1-tetralone. Tetralin formation from hexenole) were not detected and were assumed to be highly
Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991 2043

e3 \
OH ’ 1
- 9 d

o^ 0
Figure 19. Benzofuran reaction network proposed by Edelman et
al. (1988) (Table XXI).

A formed after cracking into two monoaromatic species.


PH /” Y HDO pathways of additional phenolic lignin model com-
pounds, including 4-methylguaiacol, 4-methylcatechol,
eugenol, vanillin, and o,o’-biphenol, have been reported

\
1 (Petrocelli and Klein, 1987).
VC2. Heterocyclic Oxygen Compounds. Two groups
investigated the HDO of benzofuran (Edelman et al., 1988;
Q+O
Figure 18. (A) Reaction network for 4-methylphenol proposed by
Lee and O b ,1984a) (Table XXI). In the network shown
in Figure 19, oxygen removal occurs only after partial
hydrogenation of benzofuran; direct hydrogenolysis of the
Gevert et al. (1987) (Table XXI). (B) Simplified network used for latter to give ethylbenzene was postulated not to occur,
kinetics analysis, in contrast to the situation with the sulfur analogue ben-
reactive intermediates since the HDO of methylcyclo- zothiophene (Figure 6). However, the data are incomplete
hexanol to give methylcyclohexenes in the presence of and the hydrocarbon formation sequence open to question.
y-Al,O, is fast. The postulated formation of the me- Dealkylation reactions occurred, with phenol and toluene
thylcyclohexenes from partially hydrogenated phenols is being produced at 300 O C , the lowest temperature used.
questionable since the former can be formed by the de- Dealkylation reactions have also been observed in inves-
hydration of methylcyclohexanol. Furthermore, no ketones tigations of l-methylnaphthalene conversion (Patzer et al.,
were detected, in contrast to the l-naphthol network 1979) and HDO of acidic coal liquid fractions (Grandy et
(Figure 17). al., 1986). Although benzofuran hydrogenation to give
A simplified network was used for kinetics analysis 2,3-dihydrobenzofuran followed pseudo-fmt-orderkinetics
(Figure lSB), with the slow hydrogenation of toluene to (with an apparent activation energy of 11.8 kcal/mol), the
give methylcyclohexane and methylcyclohexenes being HDO did not. The poor fit to fmt-order kinetics for HDO
neglected. 4-Methylphenol disappearance did not follow was attributed to the weaker inhibition by water. The
pseudo-first-order kinetics, and the rate decreased with HDO data were fitted to a simplified Langmuir-Hin-
increasing initial reactant concentrations, suggesting shelwood rate expression with an order of -1 in total or-
weaker inhibition by water than by organooxygen com- ganooxygen compounds (implying high coverage by the
pounds. Two kinds of catalytic sites were assumed: hy- latter). A similar network had been reported by Lee and
drogenolysis sites on which 4methylphenol was converted Ollis (1984a), except that dealkylation was not observed,
to toluene and hydrogenation sites on which 4-methyl- possibly because of the generally lower temperatures used
cyclohexanol and the methylcyclohexenols were formed. (Table XXI).
The same rate form was used for each site: A quantitative reaction network for dibenzofuran has
been reported by Krishnamurthy et al. (1981). Dibenzo-
furan HDO was proposed to occur by the following four

Inhibition by the products was assumed to be negligible.


The intrinsic rate constants kj for hydrogenolysis and
(3) pathways:

- - -
dibenzofuran 2-phenylphenol biphenyl
dibenzofuran 2-cyclohexylphenol-
hydrogenation were 2 X lo-‘ and 2.3 X L/(g of cata-
lyst-s), respectively; the values of the adsorption equilib-
rium constants characterizing the hydrogenolysis and hy- dibenzofuran -
2-cyclohexylphenol-
cyclohexylbenzene

bicyclohexyl + single-ring products


drogenation sites were 23 and 12 L/mol, respectively.
Thus,direct hydrogenolysis was the faster pathway under
the conditions employed. Since only two initial reactant
concentrations were used to estimate the parameters, the
-
dibenzofuran 1,2,3,4-tetrahydrodibenzofuran
6-phenyl-l-hexanol- (unspecified) hydrocarbons
-
estimates are imprecise. The phenolic compounds listed in the above pathways were
A partial network has been reported for a higher mo- not detected. The pathways involving 2-phenylphenol and
lecular weight phenolic compound, 2-hydroxydiphenyl- 2-cyclohexylphenol were proposed on the basis of exper-
methane (Table XXI) (Petrocelli and Klein, 1987). The iments done with the latter as reactants. These pathways
products were not all identified. The HDO proceeded via evidently do not occur strictly in parallel. For example,
two major pathways: direct hydrogenolysis to give di- the authors did not rule out additional reactions involving
phenylmethane and oxygen removal from phenolic species the intermediates such as hydrogenation of 2-phenylphenol
2044 Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991
Products

2,049
t
a- 56x10.’
Single Ring Hydrocarbons

19.31X 0-0
Figure 21. RBaction network for dibemfuran proposed by LaVopa
and Satterfield (1987) (Table XXI).
Products Products

Figure 20. Simplified dibenzofuran network proposed by Krishna- at a partial pressure at 0.137 atm gave a decrease in the
murthy et al. (1981) for kinetics analpis (Table XXI). The numbers rate of dibenzofuran HDO. The selectivity to two-ring
next to the arrows are pseudo-first-order rate constants in L/(g of compounds decreased with increasing feed hydrogen sul-
cata1yst.s) at 343 O C .
fide concentration. The authors interpreted the latter
to give 2-cyclohexylphenol; such a hydrogenation is cer- finding to mean that hydrogen sulfide inhibited hydro-
tainly expected on the basis of the rapid hydrogenation genolysis more than hydrogenation, as observed in the
of phenols (Section VB). The proposed network (Figure HDS of dibenzothiophene by Broderick and Gates (1981).
20) is thus only approximate. Dibenzofuran conversion However, the analogy with the HDS results is not entirely
occurs by direct oxygen removal as well as ring saturation.
The most rapid pathway, designated as “dibenzofuran
products”, represents the reaction sequence forming cy-
- valid; Broderick and Gates considered cyclohexylbenzene
to be a hydrogenation product of dibenzothiophene (since
hydrogenolysis of the hydrogenated dibenzothiophenes was
much more rapid than their formation). The rate of cy-
clohexylcylcohexanol,phenol, and cyclohexane through the
formation of 2-cyclohexylphenol. The oxygen-removal clohexylbenzene formation from dibenzothiophene was
reaction sequence from phenols was not elucidated. Figure weakly affected by hydrogen sulfide; in contrast, the se-
20 indicates that all the dibenzofuran HDO reactions were lectivity to cyclohexylbenzene (a two-ring product) in
considerably slower than the hydrogenations of biphenyl LaVopa and Satterfield‘s study evidently decreased with
and cyclohexylbenzene. In the network for dibenzo- increasing hydrogen sulfide concentration.
thiophene HDS proposed by Houalla et al. (1978) (Figure The inhibition by hydrogen suKde was reversible. The
71, the rates of biphenyl and cyclohexylbenzene hydro- disappearance of dibenzofuran followed pseudo-first-order
genation were markedly lower than that of dibenzo- kinetica when hydrogen sulfide was present in the feed (at
thiophene hydrogenolysis. The comparison suggests that 0.14-atm partial pressure), with the rate constant at 350
dibenzofuran HDO is an order of magnitude slower than “C being 1.2 X lo+ mol/(g of catalystvs).
dibenzothiophene HDS. VC3. Ethers. A proposed network for HDO of di-
A dibenzofuran network was also reported by LaVopa phenyl ether (Petrocelli and Klein, 1987) indicates only
and Satterfield (1987) (Table XXI,Figure 21). Some a direct hydrogenolysis pathway for oxygen removal, giving
experiments were performed with organooxygen interme- phenol and benzene. Ring saturation prior to oxygen re-
diates as the reactants, including 2-phenylphenol, 2- moval did not occur, in contrast to the network for the
cyclohexylphenol, 2-phenylcyclohexanol, 2-cyclohexyl- structurally similar dibenzofuran. At higher conversions,
cyclohexanol, and 1,2,3,4-tetrahydrodibenzofuran.Rate phenol was converted to benzene and cyclohexane, but the
constants for the individual reactions were not determined. formation sequence was not given. There is a lack of
The product distribution was qualitatively similar to that comparable quantitative networks for other ethers.
observed by Krishnamurthy et al. in that (a) yields of
single-ring HDO products were larger than yields of two- VI. Binary Mixtures
ring compounds (with the former comprising about 75% Most investigations of the hydroprocessing reactivities
of the products), (b) cyclohexylbenzene was formed in the of mixtures have been done with pairs of organic reactants,
highest yields among the two-ring compounds, and (c) whereby the concentration of at least one reactant was
oxygen-containing products were not detected with di- varied to determine inhibition of the other’s reactivity. A
benzofuran feed. few investigations have been done with three or more
In contrast to the results of Krishnamurthy et al., La- reactants, but because the quantitative analysis of the
Vopa and Satterfield did not detect 6-phenyl-l-hexanol. products is more difficult, the results are limited to com-
The latter authors found a product having a GC retention parisons of either the rates of conversion of the reactants
time similar to that of 6-phenyl-1-hexanol, namely, (cy- or the conversions in the presence and absence of inhib-
clopentylmethyl)benzene,and they questioned Krishna- itors, Binary mixture results are addressed first.
murthy et al.’s claim of identifying 6-phenyl-l-hexanol. For purposes of the discussion, a compound added to
Furthermore, LaVopa and Satterfield found no evidence the feed to determine its effect on another’s reactivity is
for the formation of the intermediate compounds 2- termed an “additive“. The effects of additives consisting
phenylcyclohexanol and 2-cyclohexylcyclohexanol. of organic sulfur, oxygen, and nitrogen compounds on the
Addition of hexanethiol (which was rapidly converted reactivities of aromatic hydrocarbons and organosulfur,
to give hydrogen sulfide at reaction conditions) to the feed organooxygen, and organonitrogen compounds are con-
Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991 2046

Table XXII. Reaction Conditions for Investieations of Inhibition of Aromatic Hydrogenation


reactant additive conditions reference
1. Sulfur Additives
biphenyl, 5.8-26 mmol/L H2S, 8.5-38 mmol/L liquid-phase flow reactor, 300-375 "C, 70-200 atm, Sapre and Gates, 1982
n-hexadecane solvent, Co-Mo/A120s catalyst
propylbenzene, 0.13 atm H2S, 0 . 1 3 atm vapor-phase flow reactor, 375 "C, 68 atm, Gtiltekin et al., 1984
Ni-Mo/A120S catalyst
naphthalene, 4-6 w t % dibenzothiophene, batch reactor, 350 "C, 35 atm, n-hexadecane solvent, Lo, 1981
0-2.5 wt % Ni-Mo/A1203 catalyst
2. Oxygen Additives
biphenyl" H20,b0.1 wt % batch reactor, 350 "C,102 atm, n-dodecane solvent, Krishnamurthy and Shah, 1982
Ni-Mo/Al2O3 catalyst
3. Nitrogen Additives
naphthalene, 4.4 wt % quinoline, 0.2-2 w t % batch reactor, 350 "C, 35 atm, n-hexadecane solvent, Bhinde, 1979; Lo, 1981
Ni-Mo/A120S catalyst
2-methylnaphthalene, 2,4-lutidine, 0-1 wt % flow reactor, 280-360 "C, 39-128 atm, n-hexadecane Ho et al., 1984
20 wt % solvent, Ni-Mo/A1203 catalyst
naphthalene, 0.9 w t % indole, 0-4 wt % batch reactor, 350 "C, 35 atm, n-hexadecane solvent, Lo, 1981
Ni-Mo/A1203 catalyst
propylbenzene, 0.13 atm NH3, 0 . 1 3 atm vapor-phase flow reactor, 375 "C, 69 atm, Gtiltekin et al., 1984
Ni-Mo/A120s catalyst
"Produced from dibenzothiophene (1w t % ) HDS. bProduced from in situ HDO of o-cyclohexylphenol (1 wt %).

sidered. Effects of ammonia, water, and hydrogen sulfide genation rates of naphthalene and tetralin in terms of a
are also evaluated. modified pseudo-first-order rate constant:
VIA. Effects of Additives on Aromatic Hydrogen-
ation. Reaction conditions for investigations of the effecta k: = ki
of the various additives on reactivities of aromatic hy- (4)
1 + KHCCHC
+ KDTCDT
drocarbons are summarized in Table XXII.
VIAL. Sulfur Compounds. Modeling of the inhibition where the subscripts HC and DT denote, respectively, the
by sulfur compounds, especially hydrogen sulfide, is com- aromatic hydrocarbons and dibenzothiophene. All the
plicated by the fact that hydrogen sulfide can modify the aromatic hydrocarbons were assumed to be represented
catalyst surface (e.g., by increasing the density of Bransted by one value, KHc,and hydrogenated dibenzothiophenes
acid sites), especially a t higher concentrations. Further- were assumed to be represented by K m Values of ki were
more, hydrogen sulfide (with hydrogen) is evidently re- determined independently from experiments with uaph-
quired to maintain the catalysts in the form of sulfides, thalene in mixtures with quinoline. The results (Khc =
rather than oxides (Broderick et al., 1982). Thus, Lang- 2.7 L/mol, KDT= 12 L/mol) show that inhibition by di-
muir-Hinshelwood rate expressions that do not account benzothiophene is comparable (i.e., within 1 order of
for interconversion of catalytic sites may not be theoret- magnitude) to that by the aromatic hydrocarbons, at least
ically valid over the whole range of hydrogen sulfide con- when the molecular weights are similar. At the concen-
centrations. tration ranges employed, the KC products for both the HC
Sapre and Gates (1982)expressed hydrogen sulfide in- and DT groups were of order 1, which implies that in-
hibition of biphenyl hydrogenation with a modified hibition by dibenzothiophene was not severe. Inhibition
Langmuir-Hinshelwood rate form (Table 111);the data by hydrogen sulfide produced from dibenzothiophene HDS
were collected in differential conversion experiments was not considered, although the ratio K&KHc (4.4) was
(Table XXII). Hydrogen sulfide was always present in within a factor of 2 of the value for sulfi&,ipi,enyl
their experiments, with its concentration varied by a factor (2.6)reported by Sapre and Gates (1982)a t the same
of 4. A t 350 OC, the adsorption equilibrium constant for temperature but with a different form of rate equation
hydrogen sulfide (12L/mol) was slightly larger than that (Table 111).
for biphenyl (4.6L/mol), implying that inhibition by hy- VIA%. Oxygen Compounds. Inhibition effects of or-
drogen sulfide was comparable to self-inhibition by bi- ganooxygen compounds and water on hydrogenations of
phenyl. In the range of concentrations examined (Table aromatic hydrocarbons have not been reported. However,
XXII), the inhibition was mild, with the product of the an investigation of the effects of water on the reactions of
adsorption parameter and the concentration being of order the dibenzothiophene network (Table XXII) allowed es-
1. timation of the change in the pseudo-first-order rate con-
Comparable inhibition of propylbenzene hydrogenation stant for biphenyl hydrogenation as water was added to
to give propylcyclohexane was observed by Gultekin et al. the feed (Krishnamurthy and Shah, 1982). With 1 w t %
(19841,who measured the effect of hydrogen sulfide con- dibenzothiophene in the feed, the pseudo-first-order rate
centration on the pseudo-first-order rate constant for hy- constant for biphenyl hydrogenation decreased from 0.0056
drogenation (Table XXII). At 375 O C and 68.1atm, the min-I to 0.0005 min-* upon addition of 1 wt % o-cyclo-
rate constant decreased by 56% from a value of 3.9 X lo-' hexylphenol (which was rapidly hydrodeoxygenated to gh-e
mol/(g of cata1yst.s) as feed hydrogen sulfide partial water). However, this apparent strong inhibition of bi-
pressure increased from zero to 0.13atm. In this work, phenyl hydrogenation is inconsistent with the observation
however, there was no hydrogen sulfide in the feed in the that the pseudo-first-order rate constant for conversion of
base case, which means that the catalyst may have been dibenzothiophene to give cyclohexylbenzene increased by
in a less active (nonsulfided) form a t the outset. about 20% upon addition of the same amount of water;
Inhibition of the two hydrogenation reactions in the recall that conversion of dibenzothiophene to give cyclo-
naphthalene network (Figure 2) by dibenzothiophene has hexylbenzene proceeds by slow hydrogenation of di-
been reported by Lo (1981),who expressed the hydro- benzothiophene (Houalla et al., 1978). Additional exper-
2046 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

iments are needed to resolve the issue. than those estimated by Lo on the basis of data obtained
VIAS. Nitrogen Compounds. The role of organo- at 350 O C . The data of Table XXII show that the higher
nitrogen compounds as inhibitors of aromatic hydrogen- concentrations of inhibitors gave lower values of the in-
ation reactions has been known for years (Desikan and hibition parameters. Only one inhibitor concentration was
Amberg, 1963), but the effects have been quantified only employed by Ho et al., and it is not clear how sound the
recently (Table XXII). statistical basis of their estimates might be. It is also not
Strong inhibition of the hydrogenation reactions of the known how firm a basis there is for comparison of the data
naphthalene network by quinoline and its products has from various investigators.
been reported (Bhinde, 1979; Lo, 1981). Inhibition by The effect of the nonbasic nitrogen compound indole on
quinoline and the three products of its hydrogenation the naphthalene reaction network has also been reported
(Figure 10) is apparently greater than that resulting from (Lo, 1981). The rate constants ki were determined from
the other nitrogen-containing species (o-propylaniline and experiments with naphthalene and quinoline in the feed,
ammonia) formed at longer reaction times, as evidenced and an equation analogous to eq 4 was used to represent
by the lower rate of naphthalene disappearance at shorter kinetics of the naphthalene hydrogenation in the presence
times. Two approaches were used to model the inhibition. of indole. Indole, being nonbasic, was assumed to have the
In the first, rate constants for each reaction in the naph- same adsorption parameter as naphthalene. Indoline and
thalene network were calculated separately for "early" and o-ethylaniline, the basic reaction products, were charac-
"late" reaction times (determined somewhat arbitrarily) terized by the parameter KBN. The ammonia parameter
and correlated with the feed quinoline concentration: was determined from experiments with naphthalene-
quinoline mixtures. KBN was estimated to be 710 L/mol,
(5) which is close to the value of KeBNestimated for quinoline
and its hydrogenated analogues (lo00 L/mol). It appears,
where the subscript denotes initial quinoline concen- then, that the best fmt approximation justified by the few
tration. The rate constants decreased as the amount of available data is to lump the basic organonitrogen com-
quinoline in the feed increased, and the rate of decrease pounds into one group of strong inhibitors.
was largest for the first increments of quinoline addition. The inhibition of aromatics hydrogenation by ammonia
The values of KQ at the "emliern reaction times (1000-1300 has been reported (Table XXII) (Giiltekin et al., 1984).
L/mol) were somewhat larger than those at "later" times At 375 O C and 69 atm, the pseudo-fmborder rate constant
(210-370 L/mmol), implying stronger inhibition by quin- for propylbenzene hydrogenation decreased by a factor of
oline and ita hydrogenated derivatives since the latter are 7 when the ammonia partial pressure increased from 0 to
mostly primary products (Figure 10). 0.13 atm, which was equal to the propylbenzene partial
A more rigorous kinetics analysis was also performed by pressure in the feed. Ammonia was thus a stronger in-
assuming a Langmuir-Hinshelwood rate expression of the hibitor of hydrogenation than hydrogen sulfide, as the
form latter reduced the pseudo-first-order rate constant by a
factor of only 2.3.
k;;C; In summary, nitrogen compounds are the strongest in-
hibitors of aromatics hydrogenation. Inhibition by orga-
nonitrogen compounds is stronger than that by ammonia,
and estimating the parameters directly from the concen- but quantitative agreement between different investiga-
tration-time data. Quinoline, ita hydrogenated derivatives, tions is lacking. There are also inconsistencies concerning
and o-propylaniline were grouped into the term designated the absolute values of the adsorption parameters charac-
as KQ~Nand ammonia (A) was represented by Kk The terizing the inhibition by the organonitrogen compounds.
parameter estimation was complicated by statistical cor- These results suggest that a more fruitful approach to
relation between the adsorption parameters (in the de- modeling the inhibition would be to conduct experiments
nominator) and rate constants (in the numerator), re- using organonitrogen compounds in the feed having sim-
quiring a trial-and-error procedure whereby adsorption pler networks (e.g., amines that form ammonia rapidly),
parameters were estimated independently of the rate estimate the corresponding adsorption parameters, and
constants. The values KQBN = lo00 L/mol and K A= 400 then repeat the process using organonitrogen reactants
L/mol show that inhibition parameters characteristic of having more complex networks (e.g., progress to anilines
the nitrogen compounds, including ammonia, are at least and then quinolines).
2 orders of magnitude larger than that characteristic of VIB. Effect of Additives on HDS. HDS networks
naphthalene and its products (KHC = 2.7 L/mol). Small include hydrogenation as well as hydrogenolysis reactions,
amounta of organonitrogen compounds in the feed there- which evidently occur on different kinds of catalytic sites
fore reduce hydrogenation rates markedly. (Desikanand Amberg, 1964; Satterfield and Roberts, 1968;
Strong inhibition of 2-methylnaphthalene hydrogenation Broderick and Gates, 1981). The effecta of additives have
by 2,4-lutidine (2,4-dimethylpyridine) has also been re- been determined for both types of reactions, with the re-
ported (Table XXII) (Ho et al., 1984). The proposed rate sults interpreted in terms of their competitive adsorption
expression for hydrogenation of the aromatic hydrocarbon on the two kinds of sites.
is the same in form as that used by Lo, except that the VIB1. Aromatic Hydrocarbons. Weak inhibition by
ammonia adsorption was neglected:
naphthalene of reactions in the dibemthiophene network
~HCKHCCHC (Figure 7) has been reported by Lo (1981) (Table XXIII).
rHC = (7) Naphthalene inhibits both the hydrogenation and hydro-
1 + KNCN+ KHCCHC genolysis reactions of the dibenzothiophene network
The organonitrogen compounds formed, 2,4-lutidine and mildly, with selectivities to the various products being
2,4-dimethylpiperidine, were each assumed to be charac- virtually unchanged with naphthalene in the feed. In the
terized by KN.At 310 "C and 69 atm, the estimated values kinetics analysis, hydrogenations of dibenzothiophene,
of the parameters were K H C = 0.085 L/mol and KN = 0.9 1,2,3,4-tetrahydrodibenzothiophene, and aromatic hydro-
L/mol. These values are smaller by 2 orders of magnitude carbons were all assumed to occur on the same kinds of
Ind. Eng. Chem. Res., Vol. 30, No. 9,1991 2047

Table XXIII. Reaction Conditions for Investigations of Inhibition of HDS


reactant additive conditions reference
1. Hydrocarbon Additives
dibenzothiophene, naphthalene, 0-5.9 wt % batch reactor, 350 "C, 35 atm, n-hexadecane Lo, 1981
0.7-2.5 wt % solvent, Ni-Mo/A1203 catalyst
thiophene, 0.24 atm naphthalene, vapor-phase flow reactor, 360 "c,69 atm LaVopa and Satterfield, 1988
phenanthrene, 0 . 2 4
atm
dibenwthiophene, 5 phenanthrene, 0-2 w t % vapor-phase flow reactor, 300 "C, 100 atm, xylene Nagai and Kabe, 1983
wt% solvent, Mo/A1203 catalyst
2. Sulfur Additives
thiophene, H#, unspecified vapor-phase flow reactor, 260-350 "C, 2-30 atm, Van Parijs and Froment, 1986
unspecified concentration n-heptane solvent, Co-Mo/Alp03 catalyst
concentration
range, 0.4-6 atm
estimated
benzothiophene, 0.96 H@, unspecified vapor-phase flow reactor, 240-300 "C, 2-30 atm, Van Parijs et al., 1986
wt% concentration n-heptane solvent, Co-Mo/AlzOS catalyst
dibenzothiophene, H a , 0.001-0.068 mol/L flow reactor, 275-325 "C, 177 atm, n-hexadecane Broderick and Gates, 1981
0.008-0.13 mol/L solvent, Co-Mo/A120S catalyst
dibenzothiophene, 5 HZS, 0-2 wt % vapor-phase flow reactor, 300 "C, 100 atm, xylene Nagai and Kabe, 1983
wt% solvent, Mo/Al,03 catalyst
3. Oxygen Additives
benzothiophene, 0.15 m-cresol, 0 . 1 5 mol/L trickle-bed reactor, 375-400 OC, 69 atm, Odebunmi and Ollis, 1983b
mol/L n-hexadecane solvent, Co-Mo/Aln03, catalyst
dibenzothiophene, benzofuran, 0 . 1 5 mol/L trickle-bed reactor, 325 "C, 69 atm, n-hexadecane Lee and Ollis, 1984b
0.15 mol/L solvent, Co-Mo/Al,03 catalyst
dibenzothiophene, 5 phenol, dibenzofuran, vapor-phase flow reactor, 300 "C, 100 atm, xylene Nagai and Kabe, 1983
wt% xanthene, 0-2 w t % solvent, Mo/A1203 catalyst
dibenzothiophene, 1 dibenzofuran, H,O, 0-1 w t batch reactor, 350 "C,102 atm, n-dodecane solvent, Krishnamurthy and Shah, 1983
wt% % Ni-Mo/A1,03 catalyst
4. Nitrogen Additives
dibenzothiophene, 0.7 quinoline, 0-2 w t % batch reactor, 350 "C, 35 atm, n-hexadecane Bhinde, 1979;Lo, 1981
wt% solvent, Ni-Mo/A1203 catalyst
thiophene, 0.03-0.2 pyridine, 0 . 1 atm vapor-phase flow reactor, 200-500 "C, Satterfield et al., 1975
atm Co-Mo/Alz03, Ni-Mo/A1203, and Ni-W/A1203
catalysts
thiophene, 0.12 atm pyridine, 0.12 atm vapor-phase flow reactor, 200-400 "C, 11-69 atm, Satterfield et al., 1980
Ni-Mo/A1203 catalyst
dibenwthiophene, 5 acridine, carbazole, vapor-phase flow reactor, 300 "C, xylene solvent, Nagai and Kabe, 1983
wt% phenothiazine, Mo/A1303catalyst
dicyclohexylamine, 0-2
wtw
sulfur in petroleum see Table XXIII vapor-phase flow reactor, 292 "C, 11 atm, Gutberlet and Bertolacini, 1983
naphtha Co-Mo/Al2O3 catalyst
thiophene, 0.24 atm see Table XXIV, 0 . 2 4 vapor-phase flow reactor, 360 "C,69 atm, LaVopa and Satterfield, 1988
atm Ni-Mo/A1,03 catalyst
thiophene quinoline, pyridine, vapor-phase stirred balance microreactor 350 "C, 1 Miciukiewicz et al., 1984
aniline, piperidine, atm, Co-Mo/A1,03 catalyst
methyl-substituted
derivatives
dibenzothiophene, 5 see Table XII, 0 . 5 wt % flow reactor, 300 "C, 100 atm, xylene solvent, Nagai et al., 1986b
wt% Ni-Mo/Al,03 catalyst
sites. Hydrogenolysis reactions were assumed to occur on ation and hydrogenolysis. The inhibition was represented
these sites and an additional type of site. A two-site model by a rate expression with thiophene adsorption neglected
for hydrogenolysis reactions was selected because of the
good fit to the data, especially in modeling inhibition by
organonitrogen compounds (see below). To reduce the
number of parameters in the model, the inhibition pa-
rameters for the aromatic hydrocarbons and for dibenzo- At 360 "C, adsorption parameters for naphthalene and
thiophene and its hydrogenation products were assumed phenanthrene were 11 and 65 atm-', respectively. The
to be equal for both types of sites. Consequently, in the 6-fold increase in KHCas the number of rings increased to
absence of organonitrogen compounds, the rate expression three suggests that the adsorption parameters of com-
reduced to eq 4. The values of the adsorption parameters pounds having four fused rings could be as much as 1 order
were the same as those obtained in determining the effect of magnitude greater than that of naphthalene, but
of dibenzothiophene on the naphthalene network, namely, quantitative evidence for this suggestion is lacking.
KHC= 2.7 L/mol and K m = 12 L/mol. Inhibition by Other investigations, however, indicate that inhibition
naphthalene is thus quantitatively similar to self-inhibition of hydrogenation of organosulfur compounds by aromatic
by dibenzothiophene. hydrocarbons is stronger. In contrast to Lo's results, Nagai
Mild inhibition of thiophene conversion (equal to its and Kabe (1983)(Table XXIII) observed in differential
HDS in this case) by two aromatic hydrocarbons has also conversion experiments that addition of up to 2 wt %
been reported (LaVopa and Satterfield, 1988) (Table phenanthrene to a solution of 5 wt % dibenzothiophene
XXIII), but no distinction was made between hydrogen- decreased the yield of cyclohexylbenzene but not that of
2048 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991
biphenyl. Phenanthrene was therefore assumed to adsorb hibition of biphenyl and cyclohexylbenzene formation was
solely on hydrogenation sites. Inhibition of the hydro- accounted for with expressions of the same form, with that
genation reactions was represented by a rate expression corresponding to biphenyl formation being
of the same form as eq 8 giving the rate of formation of ~KDTKHFDTPH~
1,2,3,4-tetrahydrodibenzothiophenefrom dibenzo- rBP = (10)
thiophene; the phenanthrene inhibition term was assumed 1 + KOPO
to be much larger than those of dibenzothiophene and its where the subscript 0 denotes an oxygen compound.
hydrogenated derivatives. This assumption is consistent Adsorption parameter values for dibenzofuran, xanthene,
with Van Parijs et al.’s finding that ethylbenzene is a and phenol in the biphenyl rate expression were 11.7,11.2,
strong inhibitor of benzothiophene hydrogenation (Table and 9.2 atm-’, respectively, at 300 “C. The near equality
XI). The value of the phenanthrene adsorption parameter of values suggests that inhibition by one-ring and multirii
estimated by Nagai and Kabe was equal to 10 atm-’ and oxygen compounds is similar. These values were also
comparable to that obtained for thiophene HDS inhibition within a factor of 2 of those in the cyclohexylbenzene rate
by LaVopa and Satterfield. equation. Data obtained at higher concentrations of the
In summary, inhibition of HDS by aromatic hydro- oxygen compounds deviate from the predictions of eq 10,
carbons is not negligible. There is disagreement regarding presumably because of structural changes in the catalyst
the degree to which aromatic hydrocarbons inhibit the surface induced by product water.
conversion of organosulfur compounds. Nevertheless, the Weak inhibition of dibenzothiophene HDS by di-
inhibition is much weaker than that by basic organo- benzofuran and water has also been reported (Krishna-
nitrogen compounds, as discussed below. murthy and Shah, 1982) (Table XXIII). Addition of 0.75
VIB2. Sulfur Compounds. The effects of hydrogen wt % dibenzofuran to 1wt % dibenzothiophene reduced
sulfide on HDS networks were stated above. Briefly, hy- the pseudo-first-order rate constant for dibenzothiophene
drogen sulfide inhibits hydrogenolysis reactions but not disappearance by only 25%. Addition of 1 wt % o-
hydrogenations (Broderick and Gates, 1981; Van Parijs and cyclohexylphenol (which was rapidly hydrodeoxygenated
Froment, 1986; Van Parijs et al., 1986). Similar conclusions to form water) to 1 wt % dibenzothiophene reduced the
were reached by Nagai and Kabe (1983) (Table XXIII), pseudo-first-order rate constant for the hydrogenolysis of
who found that the yields of biphenyl and cyclohexyl- dibenzothiophene 2-fold.
benzene decreased with increasing amounts of hydrogen VIB4. Nitrogen Compounds. Basic organonitrogen
sulfide (formed rapidly in situ from carbon disulfide) compounds are among the strongest inhibitors of HDS and
added to the dibenzothiophene feed. Using the same rate therefore the most widely investigated, but only recently
form as eq 8, a value of 12.3 atm-’ was calculated for the have quantitative inhibition effects been reported.
hydrogen sulfide adsorption parameter. Inhibition by A thorough analysis of the inhibition by quinoline of the
hydrogen sulfide is thus similar to that by aromatic hy- reactions in the dibenzothiophene network (Figure 7) was
drocarbons. carried out by Bhinde (1979) (Table XXIII). Quinoline
VIB3. Oxygen Compounds. Oxygen compounds strongly inhibited all the reactions of the network. As for
mildly inhibit the reactions of organosulfur compounds. inhibition of naphthalene hydrogenation, the inhibition
Odebunmi and Ollis (1983b) investigated m-cresol inhib- was stronger at earlier reaction times when quinoline and
ition of benzothiophene HDS at nearly differential con- its hydrogenated products were the predominant nitro-
versions (Table XXIII). The inhibition of the HDS re- gen-containing species. In the kinetics analysis, Bhinde
action was represented by divided the reaction period into “early” and “late” segments
(the latter occurring when the only nitrogen-containing
~KBTCBT compounds remaining were o-propylaniline and ammonia).
r= (9)
1 + KBTCBT~
+ KcrCcro Pseudo-first-order rate constants for each reaction in the
dibenzothiophene network were calculated separately for
where the subscripts BT and Cr denote, respectively, each region and correlated with the feed quinoline con-
benzothiophene and cresol and the additional subscript centration. Hydrogenation reactions were inhibited more
0 denotes a feed concentration. At 375 O C , the estimated than hydrogenolyses, and the approximate 2-fold difference
values of the adsorption parameters were Kcr = 8.4 L/mol in the decrease of the rate constants was interpreted as
and KBT = 16.5 L/mol. Inhibition of HDS by water was evidence of separate sites for hydrogenation and hydro-
not considered because little water was formed at the low genolysis.
conversions examined. Inhibition by hydrogen sulfide Using Bhinde’s data, Lo (1981) developed a model based
formed from the benzothiophene was not considered either. on Langmuir-Hinshelwood rate expressions for each re-
Inhibition of dibenzothiophene HDS by benzofuran has action of the dibenzothiophene network; an advantage of
been reported (Lee and Ollis, 1984b) (Table XXIII). this approach was the simultaneous modeling of HDS and
Determination of the inhibition of individual reactions in HDN. It was found that a single-site model for the hy-
the dibenzothiophene network was not possible because drogenolysis reactions in the dibenzothiophene network
experiments were performed a t differential dibenzo- (such as the one used to model naphthalene hydrogena-
thiophene conversions. Dibenzothiophene disappearance tion) gave a poor fit to the data. A two-site model, with
was represented by an expression of the same form as eq more parameters, gave a better fit:
9. At 325 “C,the values of the inhibition parameters were
comparable: K m = 11.9 L/mol and KBF = 61.7 L/mol. rij,tod = rij,I + rij,II (11)
The effects of single-ring and multiring oxygen com- kij,ICi
pounds on dibenzothiophene reactivity have been exam- rij,r = (12)
ined (Nagai and Kabe, 1983) (Table XXIII). Oxygen 1 + KHCCHC + KsCs + K Q B N J ~ Q+BKA,ICA
N
compounds decreased yields of both biphenyl and cyclo- kij,&i
hexylbenzene. Because the selectivities to biphenyl and rij,II =
cyclohexylbenzene were independent of the amount and + KsCs + K Q B N J I ~ Q
1 + KHCCHC +BKA,IICA
N
type of oxygen compound added, poisoning of both hy- (13)
drogenolysis and hydrogenation sites was inferred. In- The form of the equations is suggestive of the two-site
Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2049

Table XXIV. Adsorption Parameters for Nitrogen sociated with resonance interactions. Inhibition of HDS
Compounds at 292 "C Determined by Gutberlet and by ammonia (produced by rapid HDN of benzylamine) was
Bertolacini (1983) comparable to that characterizing the unhindered heter-
compound K, atm-l ocyclics, in qualitative agreement with Lo's results.
4-ethylpyridine 0.23 The relatively weak inhibition by sterically hindered
3,5-dimethylpyridine 0.16 species has also been reported for thiophene HDS by
2,4-dimethylpyridine 0.012 LaVopa and Satterfield (1988) (Table XXIII). As with the
2,6-dimethylpyridine 0 work of Gutberlet and Bertolacini, it is not clear whether
4-methylaniline 0.012 the HDS here involves hydrogenolysis of the thiophene
ammonia (produced from benzylamine) 0.099
directly or desulfurization of a hydrogenated thiophene.
model proposed years earlier by Desikan and Amberg The pseudo-first-order rate constant k' for thiophene
(1964), with the first term in eq 11corresponding to a more disappearance in the presence of several inhibitors was
active hydrogenolysis site that is more susceptible to correlated by the following equation:
poisoning and the second term corresponding to a less k' = k
(15)
active site that is more resistant to poisoning. The values 1 + KIP1
of the adsorption parameters for the less active sites
where the subscript I denotes an inhibitor. At 360 "C,the
(KBBN = KA,II= 20 L/mol) were found to be more than values of KI for pyridine, 1-methylpyridine, and 2,6-di-
an order of magnitude smaller than those for the more
active sites. The inhibition parameters for the hydro- methylpyridine were 430,263, and 110 atm-', respectively,
carbons and the sulfur compounds were assumed to be the in qualitative agreement with the data of Gutberlet and
same for both kinds of sites to reduce the number of pa- Bertolacini. There is apparently a lack of quantitative
rameters to be estimated (note that the adsorption pa- agreement between the two sets of results, however.
rameters corresponding to hydrocarbons and the sulfur A relatively weak inhibition by sterically hindered or-
compounds are represented by KHCand Ks,respectively, ganonitrogen compounds has even been reported for low-
on both kinds of sites). This choice was defended only on pressure experiments (Miciukiewicz et al., 1984) (Table
the basis of the goodness of fit, determined a posteriori. XXIII). The adsorption of nitrogen compounds on the
The number ?f parameters in eqs 12 and 13 is still large, catalyst a t steady state was determined with feeds con-
however, and the representation of such a complex set of taining a nitrogen compound and either thiophene or 1-
phenomena is less well developed than would be desirable. hexene. The inhibition was expressed in terms of ratios
The assumption of two kinds of sites might be criticized of pseudo-first-order rate constants for the disappearance
as being a computational expedient to give a better fit to of thiophene (or 1-hexene) with and without nitrogen
the data. However, earlier studies by Satterfield and co- compounds. The hindered nitrogen compounds were more
workers (Satterfield et al., 1975, 1980) (Table XXIII) of weakly adsorbed than the others (as measured gravime-
inhibition of thiophene HDS by pyridine, among others trically) and inhibited the thiophene HDS less. Surpris-
cited above, suggest that two kinds of sites are indeed ingly, the hindered nitrogen compounds inhibited 1-hexene
operative, although the data have been insufficient for hydrogenation more than thiophene HDS.
construction of a model. The correlation of the degree of steric hindrance with
Investigations of inhibition of the HDS of some sulfur the inhibition parameter (identified with the adsorption
compounds by various nitrogen compounds have also been equilibrium constant of the inhibitor) for sulfur removal
reported. These results are useful for correlating molecular suggests that the nitrogen compounds inhibit reaction by
structure with inhibition, and they provide insights into adsorption through the sp2 lone-pair electrons of the ni-
inhibition mechanisms. trogen atom. This proposal at first might appear to con-
The degree of inhibition by organitrogen compounds as tradict the hypothesis of Mathur et al. (1982) that heter-
a function of steric hindrance around the nitrogen atom ocyclic nitrogen compounds are adsorbed flat through their
was investigated by Gutberlet and Bertolacini (1983) *-electrons (see above). However, in the inhibition ex-
(Table XXIII), who measured inhibition of the HDS of a periments just described, hydrogenation of the nitrogen
petroleum naphtha by various methyl- and ethyl-substi- compounds did not occur to a significant extent, and the
tuted pyridines (Table XXIV). In this work, distinction adsorption leading to inhibition might well have been
between sulfur removal via direct hydrogenolysis of the different from that leading to catalytic hydrogenation of
organosulfur species or by its prior hydrogenation was not the inhibitors. Moreover, as Kwart et al. (1982) have
possible. Under the conditions employed (Table XXIII), pointed out, 2,6-dimethylpyridine is hydrogenated at
the organonitrogen compounds added to the naphtha un- higher temperatures and pressures than those of the in-
derwent little HDN. The naphtha HDS data were fitted hibition experiments, and the 2,6-dimethylpiperidine thus
to the following equation: formed is a strong inhibitor of HDS because its sp3 lone-
pair electrons are not hindered owing to the puckered
kPa nature of the piperidine ring.
, -
(1 + KHlSpHfi) (1 + KAPAO.' + KBpBO" + KCPcO.') The foregoing results suggest relations between the
basicity of unhindered organonitrogen compounds and the
(14) strength of their inhibition for conversion of organosulfur
where S, A, B, and C denote, respectively, sulfur compo- compounds. Such relations were sought by LaVopa and
nents of the naphtha, ammonia, nitrogen components of Satterfield (1988) for thiophene HDS and by Nagai et al.
the naphtha, and nitrogen compounds added to the (1986) for dibenzothiophene hydrogenation to give hexa-
naphtha. The data of Table XXIV show that the inhib- hydrodibenzothiophene (Table XXIII). LaVopa and
ition parameters correlate well with the degree of steric Satterfield represented the inhibition with eq 15; param-
hindrance about the nitrogen atom; the hindered nitrogen eters are listed in Table XXV. At the relatively low
compounds were characterized by relatively low values of temperature of 260 "C,Nagai et al. observed that nitrogen
the inhibition parameters. The weak inhibition by the compounds inhibited hydrogenation but not hydrogenol-
unhindered 4-methylaniline (Table XXIV) was attributed ysis of dibenzothiophene. The inhibition of the formation
to the partial delocalization of its lone-pair electrons as- of hexahydrodibenzothiophenewas accounted for with a
2050 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

Table XXV. Adsorption Parameters at 360 OC of Nitrogen weaker inhibition of dibenzothiophene hydrogenation by
Compounds in Rate Equation for Thiophene HDS phenanthrene (adsorption parameter equal to 10 atm-')
Determined by LaVopa and Satterfield (1988) compared to nitrogen inhibitors and (b) the weaker in-
compound K,atm-' hibition by dicyclohexylamine consistent with steric hin-
ammonia 65 drance about its nitrogen atom.
aniline 95 LaVopa and Satterfield (1988) demonstrated that the
2-ethylaniline 100 approach of representing the inhibition by eq 15 has some
2,6-dimethylpyridine 110 generality. Using the inhibition terms for quinoline and
1-methylpyridine 260 its nitrogen-containing products determined in the HDS
pyridine 430
1,2,3,4-tetrahydroquinoline 465 experiments, they were able to generate concentration
carbazole 510 profiles for dibenzofuran conversion that matched well
piperidine 590 data from dibenzofuran conversion with quinoline in the
4-methylpyridine 690 feed.
quinoline 990 VIC. Effect of Additives on HDO. Investigations of
5,6,7,&tetrahydroquinoline 2000
decahydroquinoline 2000 the effects of aromatic hydrocarbons on HDO are lacking.
l,&bia(NJV-dimethy1amino)naphthalene 3000 The following discussion is therefore limited to additives
(proton sponge) containing sulfur, oxygen, or nitrogen.
VIC1. Sulfur Compounds. Organoaulfur compounds
Table XXVI. Adsorption Parameters at 260 "C inhibit HDO only weakly. From results of experimenta
Characterizing Nitrogen Compounds in Rate Equation for with mixtures composed of oxygen and sulfur compounds
Hexahydrodibenzothiophene Formation Determined by
Nag& et al. (1986) (Table XXIII), Ollis and co-workers (Lee and Ollis,1984b;
Odebunmi and Ollis, 1983b) found that the same form of
compound K,atm-' the Langmuir-Hinshelwood rate expression describes the
acridine 1.98 mutual inhibition of HDO by organosulfur compounds and
quinoline 1.31 of HDS by organooxygen compounds. Inhibition of m-
pyridine 1.07
piperidine 0.85 cresol conversion by benzothiophene was expressed by eq
cyclohexylamine 0.77 9, with Kcr = 8.4 L/mol and KBT= 16.5 L/mol at 375 O C ,
y -picoline 0.63 as stated earlier.
dicyclohexylamine 0.45 Under some conditions (with a deactivated and poorly
aniline 0.19 sulfided catalyst), sulfur compounds promote HDO, con-
rate expression identical with eq 8. (Tetrahydrodibenzo- sistent with the suggestion that a sulfided catalyst is more
thiophene was virtually in equilibrium with dibenzo- active than one in the oxidic form (Lee and Ollis, 1984b)
thiophene at 260 "C.)The parameters are shown in Table (Table XXVII). In experiments with varying amounts
XXVI. In both studies, the estimated adsorption pa- of dibenzothiophene in the feed, the rate of benzofuran
rameters correlated poorly with basicities of the nitrogen HDO increased at feed dibenzothiophene concentrations
compounds in aqueous solution (i.e., pK values). Instead, up to 0.075 mol/L, but decreased at higher concentrations.
these parameters generally correlated well with gas-phase The higher benzofuran HDO rate at lower dibenzo-
proton affinities, which evidently better represent the in- thiophene concentrations was attributed to promotion by
teraction between an organic base and a surface catalytic hydrogen sulfide produced from the dibenzothiophene. A t
site. Data characterizing hindered species did not follow higher feed dibenzothiophene concentrations, the domi-
the correlation, as expected. As the gas-phase proton nant effect was competitive adsorption of the organic
affinities are enthalpy changes for proton-transfer reac- species. No rate equation was given. The promotion by
tions, it was speculated that the hydrogenation sites poi- hydrogen sulfide was not mentioned by Odebunmi and
soned by the nitrogen compounds were Brernsted acid sitee. Ollis (198313) in the report concerning m-cresol and ben-
The inhibition effect of the nonbasic carbazole appears zothiophene, perhaps because only a fresh catalyst was
unexpectedly strong. Table XXV, for example, indicates used.
that the adsorption parameter characterizing carbazole is Weak inhibition of dibenzofuran conversion by di-
comparable to that characterizing much more basic species benzothiophene has also been reported (Krishnamurthy
(e.g., piperidine). Strong inhibition by carbazole has also and Shah, 1982) (Table XXVII). Addition of 0.24 wt %
been observed, but for dibenzothiophene hydrogenation, dibenzothiophene to a 1.0 wt % dibenzofuran feed de-
by Nagai and Kabe (1983) (Table XXIII). Inhibition of creased the pseudo-first-order rate constant for dibenzo-
dibenzothiophene hydrogenation to give tetrahydrodi- furan disappearance by only 5%.
benzothiophene was represented with a Langmuir-Hin- The effect of hydrogen sulfide on HDO has not been
shelwood rate expression of the form of eq 8. The inhib- established clearly. In their investigation of dibenzofuran
ition parameters were determined to be 690,650,62, and HDO, LaVopa and Satterfield (1988) observed that an
41 atm-l for acridine, carbazole, phenothiazine, and di- increase of hydrogen sulfide partial pressure from 0 to 0.05
cyclohexylamine, respectively. The high value of the ad- atm in the feed decreased the HDO conversion of di-
sorption characterizing carbazole was attributed to prod- benzofuran from 60 to 44%. Furthermore, the presence
ucts of a rapid hydrogenation converting it to basic com- of hydrogen sulfide in the feed reduced the selectivity to
pounds, although the identities of the basic compounds two-ring products. On the other hand, Krishnamurthy et
and the kinetics of their formation were not determined. al. (1982) found that the addition of 0.026 wt 70 carbon
This hypothesis appears reasonable in light of the networks disaide (which was rapidly converted to hydrogen sulfide
of nonbasic nitrogen compounds (section IVC4). Thus, the under reaction conditions) increased the rate of dibenzo-
distinction between nonbasic and basic organonitrogen furan HDO.
compounds, as far as inhibition is concerned, is lost after VIC2. Oxygen Compounds. Self-inhibition of HDO
hydrogenation of the nonbasic species to give basic ones. by organooxygen compounds is weak. Odebunmi and Ollis
Additional results of Nagai and Kabe's study are in (198310) fitted the HDO rate of m-cresol to an equation of
agreement with other investigations including (a) the much the form of eq 9; the mild inhibition is indicated by the
Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2061

Table XXVII. Reaction Conditions for Investigations of Inhibition of HDO


reactant additive conditions reference
1. Sulfur Compounds
benzofuran, 0.15 dibenzothiophene, trickle-bed reactor, 325 "C, 69 atm, n-hexadecane Lee and Ollis, 1984b
mol/L 0 . 1 5 mol/L solvent, Co-Mo/AlzO3 catalyst
dibenzofuran, 1 dibenzothiophene, carbon batch reactor, 350 O C , 102 atm, n-dodecane solvent, Krishnamurthy and Shah, 1982
wt% disulfide, 0 . 2 4 w t % Ni-Mo/AlzOS catalyst
dibenzofuran, hydrogen sulfide, 0 . 1 4 atm vapor-phase flow reactor, 360 "C, 69 atm, LaVopa and Satterfield, 1988
0.24 atm Ni-Mo/Alz03 catalyst
2. Oxygen Compounds
l-naphthol, 0.002 methanol, 0 . 0 0 3 mmol/g flow reactor, 225 "C, 120 atm, cyclohexane solvent, Vogelzang et al., 1983
mmol/g Ni-Mo/AlzOs catalyst
3. Nitrogen Compounds
m-ethylphenol, quinoline, o-ethylaniline, trickle-bed reactor, 375 "C, 69 atm, n-hexadecane Satterfield and Yang, 1983
o-ethylphenol, 0 . 4 mmol/g solvent, Ni-Mo/A1203 catalyst
benzofuran,
dibenzyl
ether,
benzodioxane,
0.4 mmol/g
dibenzofuran, 1 7,8-benzoquinoline, 0 . 2 5 w t % batch reactor, 350 "C, 102 atm, n-dodecane Kriahnamurthy and Shah, 1982
wt% solvent, Ni-Mo/AlzOS catalyst
m-cresol, 0.15 indole, 0 . 1 5 mol/L trickle-bed reactor, 250-350 "C, 69 atm, Odebunmi and Ollii,1983c
mol/L n-hexadecane solvent, C0-Mo/A120s catalyst

value of the m-cresol adsorption parameter, Kcr = 8.4 Table XXVIII. Inhibition Parameters Calculated by
L/mol. Odebunmi and Ollis (1983~)from Binary m-Cresol and
Investigating dibenzofuran HDO, LaVopa and Satter- Indole Experimentso
field (1988) (Table XVII) found that a partial pressure of temp, KC,, KN, k m o , mol/(g ~HDN.mol/(g
0.24 atm of water in the feed (with hydrogen sulfide also "C L/mmol L/mmol of catalystnh) of catalyst-h)
present) did not affect the dibenzofuran HDO rate at 69 250 0.0211 0.0037 1.34 X lod 1.1 X lo-'
atm; the temperature and space time were not given, but 300 0.0017 0.0037 2.55 X 5.3 X lo-'
may correspond to the authors' standard reaction condi- "Reaction conditions are given in Table XVII.
tions: 360 O C and 160 (hag of catalyst)/(mol of dibenzo-
furan). Vogelzang et al. (1983) (Table XVII) also observed inhibited m-cresol HDO, the inhibition being greater at
that water weakly inhibits l-naphthol HDO. The HDO lower temperatures because of the larger ratio of indoline
conversion was stabilized after approximately 20 h on to indole concentration at lower temperatures arising from
stream in the flow reactor without water in the feed. On the equilibrium between the latter. A rate expression of
addition of 0.00312 mmol of methanol/g of feed (methanol the form of eq 9 was proposed. Inhibition terms corre-
was rapidly deoxygenated to form water under reaction sponding to (a) m-cresol and (b) indole and its nitrogen-
conditions), both the steady-state HDO conversion and the containing producta were included. The indole adsorption
product distribution became similar to those obtained parameter was assumed to be equal to that of the other
without water in the feed. However, the catalyst attained nitrogen-containing products (Le., indoline and o-ethyl-
its steady-state activity faster with water present in the aniline). In light of the discussion of HDS inhibition by
feed. carbazole (section VIB4), lumping indole with ita basic
VIC3. Nitrogen Compounds. Inhibition of the HDO products is an oversimplification since the inhibition is
of several compounds by two basic nitrogen compounds likely to be dominated by the latter; the approach em-
was examined by Satterfield and Yang (1983) (Table ployed by Lo (1981) to model naphthalene hydrogenation
XXVII). The effects of quinoline and o-ethylaniline were with indole in the feed is probably more appropriate
determined with equimolar binary mixtures containing one (section VIA3). As shown in Table XXVIII, the inhibition
oxygen compound and one nitrogen compound. Quinoline parameter representing indole and its products is com-
and o-ethylaniline strongly inhibited the HDO of m- parable to that for m-cresol at 300 O C , but much larger at
ethylphenol, o-ethylphenol, benzofuran, and dibenzyl the lower temperatures, a result attributed to the higher
ether, but the inhibition was leas severe with o-ethylaniline. concentration of the basic compound indoline in the re-
A kinetics analysis was not presented; representative re- actor at the lower temperatures.
sults are as follows: at a space time of 50 (g of catalyst. VID. Effect of Additives on HDN. VIDl. Aromatic
h)/(mol of m-ethylphenol), the HDO conversion of m- Hydrocarbons. The inhibition of both the hydrogenation
ethylphenol was 98%. Addition of quinoline to the feed and the hydrogenolysis reactions of the quinoline network
decreased the HDO conversion to about 60%,and addition by aromatic-hydrocarbons is minimal. Bhinde (1979) and
of o-ethylaniline decreased the conversion to approximately Lo (1981) reached this conclusion in investigations of the
79%. influence of quinoline on the naphthalene network dis-
The strong inhibition of HDO by basic organonitrogen cussed earlier (Tables XXII and XXIX). For example,
compounds is also evident from the resulta of Krishna- Bhinde found that the pseudo-first-order rate constants
murthy and Shah (1982) (Table XXVII). Addition of 0.25 for his proposed quinoline network (similar to that shown
wt % 7,&benzoquinoline to 1wt % dibenzofuran reduced in Figure 10) were lower by 10-15% when 5.9 wt %
the pseudo-first-order rate constant for disappearance of naphthalene was added to a feed containing 0.2 wt %
the latter by a factor of 3.7. quinoline in hexadecane. This result is consistent with the
Inhibition of the HDO of m-cresol by the nonbasic ni- conclusion that the inhibition parameters for naphthalene
trogen compound indole was investigated by Odebunmi and ita hydrogenation products are 2 orders of magnitude
and Ollis (1983~)at nearly differential conversions. Indole smaller than those for ammonia and the organonitrogen
2052 Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991

Table XXIX. Reaction Conditions for Investigations of Inhibition of HDN


reactant additive conditions reference
1. Aromatic Hydrocarbons
quinoline, 0.2 wt % naphthalene, 0-5.9 wt % batch reactor, 350 "C, 35 atm, Bhinde, 1979 Lo, 1981
n-hexadecane solvent, Ni-Mo/A1203
catalyst
2,4-lutidine, 1wt % 2-methylnaphthalene, 20 wt % flow reactor, 280-360 OC, 39-128 atm, Ho et al., 1984
n-hexadecane solvent
2. Sulfur Compounds
quinoline, 0.5 wt % dibenzothiophene, 0 . 7 wt % batch reactor, 350 "C, 35 atm, Bhinde, 1979; Lo, 1981
n-hexadecane solvent, Ni-Mo/A1203
catalyst
acridine, 0.25 wt % thiophene, benzothiophene, flow reactor, 280-360 "C, 99 atm, Nagai et al., 1986a
dibenzothiophene, dimethyl xylene solvent, Mo/A1203(sulfided
sulfide, ethanethiol, 0-1 wt and unsulfided forms)
%
pyridine, 0.1 atm thiophene, 0.034.2 atm vapor-phase flow reactor, 200-500 "C, Satterfield et al., 1975
Co-Mo/Alz03, Ni-Mo/A1203, and
Ni-W/A1203 catalysts
pyridine, 0.12 atm thiophene, 0.12 atm vapor-phase flow reactor, 200-400 OC, Satterfield et al., 1980
11-69 atm, Ni-Mo/A1203 catalyst
3. Oxygen Compounds
quinoline, o-ethylaniline, m-ethylphenol, o-ethylphenol, trickle-bed reactor, 375 "C, 69 atm, Satterfield and Yang, 1983
0.4 mmol/g benzofuran, dibenzyl ether, n-hexadecane solvent, Ni-Mo/Alz03
0.4 mmol/g catalyst
quinoline, 5 wt % H20, H2S, 0 . 1 2 atm trickle-bed reactor, 375 "C,69 atm, Satterfield et al., 1985; Satterfield
n-hexadecane solvent, Ni-Mo/A1203 and Smith, 1986
catalyst
7,8-benzoquinoline, 1 wt % dibenzofuran, o-cyclohexyl- batch reactor, 350 "C, 102 atm, Krishnamurthy and Shah, 1982
phenol, 0-1 wt % n-dodecane solvent, Ni-Mo/A1203
catalyst
indole, 0.15 mol/L m-cresol, 0 . 1 5 mol/L trickle-bed reactor, 250-350 "C, 69 atm, Odebunmi and Ollis, 1983c
n-hexadecane solvent, Co-Mo/A1203
catalyst
acridine, 0.25 wt % xanthene, 0-1 wt % flow reactor, 280-360 "C, 99 atm, Nagai et al., 1986a
xylene solvent, Mo/Al2O3 (sulfided
and unsulfided forms)
4. Nitrogen Compounds
quinoline, 0.5 wt % indole, quinoline, 0.2-2 wt % batch reactor, 350 OC, 35 atm, Bhinde, 1979
n-hexadecane solvent, Ni-Mo/A1203
catalyst

compounds in the quinoline network (Lo, 1981). Ho et al. Table XXX. Inhibition Parameters Determined by Nagai et
(1984) reached similar conclusions. al. (1986a) for Perhydroacridine and Selected Sulfur
VID2. Sulfur Compounds. Sulfur compounds are Compoundsa
also weak inhibitors of HDN. Bhinde (1979) (Table additive K A , atm-' KpA, atm-'
XXIX) determined pseudo-first-order rate constants for dibenzothiophene 16.9 1350
the quinoline network from data for feeds containing either benzothiophene 6.9 720
0.5 wt % quinoline or 0.5 w t % quinoline and 0.7 wt 90 thiophene 5.2 560
dibenzothiophene. The rate constants for the overall ni- dimethyl sulfide 3.2 910
trogen removal and for the individual reactions in the ethanethiol 1.0 830
quinoline network were almost unchanged by the addition See Table XXIX for reaction conditions. The subscripts A and
of dibenzothiophene. The only discernible effect of the PA denoted "additive" and 'perhydroacridine", respectively.
dibenzothiophene was a 25% enhancement in the pseu-
do-first-order rate constant for the hydrogenolysis of de- data were represented with the following equation, which
cahydroquinoline. In a more rigorous analysis of Bhinde's reflects the additional assumption that perhydroacridine
data, Lo (1981) estimated the inhibition parameter for was the sole nitrogen compound adsorbed on the hydro-
quinoline to be 2 orders of magnitude greater than that genolysis sites:
for dibenzothiophene.
Nagai et al. (1986a) investigated the effect of several
sulfur compounds (Table XXIX) on the HDN of acridine
at HDN conversions less than 20% (assumed to be dif- where the subscripts DCM, PA, and S denote dicyclo-
ferential). The addition of the sulfur compounds inhibited hexylmethane, perhydroacridine, and the sulfur additive,
acridine HDN, with the yield of dicyclohexylmethane respectively. The inhibition parameters at 360 OC were
decreasing with increasing concentration of the sulfur determined separately for each binary mixture. This
compound. In the kinetics analysis, the slow reaction in procedure unfortunately gave different estimates for KPA,
nitrogen removal was assumed to be the hydrogenolysis one value for each sulfur additive investigated (Table
of perhydroacridine to give dicyclohexylmethane. This XXX). KpA was 2 orders of magnitude larger than KS for
assumption may be criticized since the acridine network any of the sulfur compounds, in qualitative agreement wlth
investigation discussed earlier (Zawadski et al., 1982) Bhinde's and Lo's results. Sulfur compounds having two
showed that rates of hydrogenation and hydrogenolysis or more rings were considerably stronger inhibitors than
reactions in the network were comparable. The kinetics mercaptans or sulfides. The inhibition parameter for
Ind. Eng. Chem. Res., Vol. 30,No. 9,1991 2053
hydrogen was also found to be 3 orders of magnitude Table XXXI. Effect of Hydrogen Sulfide and Water on
smaller than any value of Ks. Quinoline HDNO (Satterfield et al., 1985; Satterfield and
The effect of thiophene on the HDN of pyridine was Smith, 1986)
characterized in the investigations by Satterfield and co- P H ~atm , atm ?& quinoline HDN
workers (Satterfield et al., 1975, 1980) with feeds con- 0 0 42
taining thiophene, pyridine, or mixtures of the two. In 0.063 0 54
both low- and high-pressure experiments at temperatures 0 0.12 78
below 325 OC, the HDN of pyridine was less when 0.063 0.12 94
thiophene was added to the feed. The authors postulated "Conditions: 375 "C, 69 atm, space time = 365 (g of catalyst.
competitive adsorption of thiophene and pyridine on the h)/(mol of quinoline). See Table XXIX for additional details.
hydrogenation sites. At temperatures exceeding 325 "C,
however, pyridine HDN was greater when thiophene was without hydrogen sulfide in the feed caused a small en-
added to the feed. The effect was attributed to the pro- hancement in the HDN rate. The enhancement was less
motion of piperidine hydrogenolysis by hydrogen sulfide than that obtained with hydrogen sulfide in the absence
formed from thiophene. It was speculated that hydrogen of water. The addition of water with hydrogen sulfide in
sulfide kept the catalyst in a sulfded state (no other source the feed increased the HDN rate markedly. Removal of
of sulfur was present in the feed), or perhaps the hydrogen hydrogen sulfide from the feed resulted in a rapid decrease
sulfide aided in removing basic nitrogen compounds from in HDN, whereas removal of water from the feed (with
the surface. The data were insufficient for construction hydrogen sulfide present) resulted in a more gradual de-
of a kinetics model. cline in the HDN rate.
The effect of hydrogen sulfide on HDN was addressed Satterfield and Smith also determined the effects of
in the discussion of HDN (see above). The studies by hydrogen sulfide and water addition on the HDN kinetics,
Satterfield and co-workers (Table XVI) of quinoline HDN assuming rate expressions identical with those used earlier
demonstrated that hydrogen sulfide enhanced the hydro- (Table XVII). Each additive was absent or present a t a
genolysis of C-N bonds and inhibited hydrogenation partial pressure of 0.13 atm. The only effect of water
slightly (but see below). The overall effect was to enhance addition was to increase the hydrogenolysis rate constant
the HDN. The effect of hydrogen sulfide has not been for decahydrequinolineby 20%. The addition of hydrogen
presented in terms of a Langmuir-Hinshelwood rate ex- sulfide enhanced the rates of some of the hydrogenation
pression, but if the effect of hydrogen sulfide is to modify reactions as well as those of the hydrogenolysis reactions.
the catalytic surface in some way, then such a simple rate With both water and hydrogen sulfide in the feed, the rate
expression would not be appropriate. constants for all the reactions were larger than when either
water or hydrogen sulfide alone was present. The rate of
VID3. Oxygen Compounds. The effects of m-ethyl- the hydrogenolysis of decahydroquinoline was most sen-
phenol, o-ethylphenol, benzofuran, and dibenzyl ether on sitive to the presence of the additives; the addition of
the HDN of quinoline and o-ethylaniline were investigated water, hydrogen sulfide, and both caused the following
by Satterfield and Yang (1983)(Table XXIX). In ex- percentage increases in the hydrogenolysis rate constant:
periments with quinoline or a mixture of quinoline and 20,167,and 197,respectively. The enhancement of the
equimolar amounts of an oxygen compound, the quinoline hydrogenation and hydrogenolysis rates was suggested to
HDN conversion increased when the oxygen compound be related to increased acidity of the catalyst surface.
was added, and the enhancement was the same for each Somewhat different results were reported for the effect
compound. (hydrogen sulfide was also present.) For ex- of dibenzofuran and water, as measured by the rate of
ample, a t a space time of 300 (g of catalyst.h)/(mol of disappearance of 7,8-benzoquinoline(Krishnamurthy and
quinoline), the HDN conversion increased from 33% with Shah, 1982) (Table XXIX). Addition of 1 wt % di-
no oxygen compound present to 60% when an oxygen benzofuran to 1 wt % 7,8-benzoquinoline reduced the
compound was present. The yield of decahydroquinoline pseudo-first-order rate constant by only 5%. However, the
was less without oxygen compounds, suggesting that hy- addition of 1 wt % o-cyclohexylphenol (which was rapidly
drogenolysis was enhanced by the oxygen compounds. All converted to give water) reduced the rate constant by 15%.
the oxygen compounds had nearly the same effect on the In either case, the inhibition was very mild. The differ-
HDN. This result and the observation that the HDO was ences in the effect of water observed by the two groups of
rapid led the authors to propose that the water formed was investigators may be associated with differences in catalyst
the cause of the HDN promotion. This hypothesis was pretreatment conditions. The effect of water might be
confirmed by addition of 0.01 wt % water to the feed, especially difficult to determine because precautions must
which had the same effect. also be taken to ensure the exclusion of moisture in
The effect of the oxygen compounds on the HDN of base-case experiments.
o-ethylaniline was different. The ethylphenols increased The inhibition effects of m-cresol on the HDN of indole
the HDN conversion, but benzofuran inhibited it slightly; were determined from experiments by Odebunmi and Ollis
addition of 0.01 wt % water did not change the HDN (1983~) (Table XXIX). Their rate equation for both HDO
conversion. The authors therefore concluded that since and HDN was of the form of eq 9;the values of the pa-
the HDN conversion of o-ethylanilinewas strongly affected rameters given in Table XXVIII show that m-cresol in-
by its rate of hydrogenation, water did not enhance hy- hibited the HDN slightly. This inhibition of HDN by an
drogenation. It was suggested that oxygen compounds oxygen compound does not agree with the results of Sat-
exerted opposing effects, a form of promotion in addition terfield and co-workers mentioned above, possibly because
to adsorption in competition with o-ethylaniline. the conversions were differential and any promotional
More thorough experiments addressing the effects of effect of water was minimized.
water on quinoline HDN have been reported (Satterfield Nagai et al. (1986a),in an investigation of various ad-
et al., 1985;Satterfield and Smith, 1986)(Table XXIX). ditives on acridine HDN (Table XXIX), also examined the
The effects of the addition of hydrogen sulfide, water, and effect of xanthene, which was found to inhibit the acridine
both of these on the HDN were investigated; representative reactions in the same manner as the organosulfur com-
results are given in Table XXXI. The addition of water pounds, Le., by retarding the hydrogenolysis of per-
2054 Ind. Eng. Chem. Res., Vol. 30,No. 9,1991
Table XXXII. Variation of Selected Pseudo-First-Order Table XXXIII. Relative Pseudo-First-OrderRate
Rata Constants in L/(g of catalyst 0s) for the Quinoline Constants for Disappearance of Reactants or Heteroatom
HDN Network for Various Feed Quinoline Concentrations' Removal in Mixturea" (Rollmann, 1977)
(Bhinde. 1979) ~ ~~

compound re1 rata const


quinoline feed concn, wt %
hydrocarbons

quinoline + 2H2 -.
reaction 0.2
3.2 X
0.5 2.0
1.8 X 10-97.8y10-'
2-methylnaphthalene
2,3-dimethylnaphthalene
1.0
1.4

decahydroquinoline
hydrocarbons + NH3
-
5,6,7,&tetrahydroquinoline
Ha
7.5 x 10-3 3.1 x 10-3 1.5 x 10-3
organosulfur compounds
dibenzothiophene HDS
benzothiophene HDS
3.6
3.4

quinoline
carbons
-Ha
+ NHS
hydro- 6.7 X lo-' 2.8 X lo-' 9.7 X loa
thiophenol HDS
organonitrogen compounds
indole HDN
o-ethylaniline HDN
210
0.9
1.1
a Reaction conditions are given in Table XXIX. quinoline HDN 1.3
organooxygen compounds
hydroacridine. Using eq 16, they calculated an adsorption benzofuran HDO 1.0
parameter of 0.18 atm-', which is more than 2 orders of dibenzofuran HDO 0.4
2-phenylphenol HDO 1.4
magnitude smaller than the corresponding parameter for cis-2-phenyl-1-cyclohexanolHDO 210
acridine. p-cresol HDO 5.2
In summary, competitive adsorption involving oxygen
and nitrogen compounds leads to mild inhibition of HDN Reaction conditions: trickle-bed reactor, 344 O C , 48.6 atm, Co-
Mo/A1203 catalyst stabilized by operation for 4 days.
at low conversions. At higher HDO conversions, when
substantial amounts of water are formed, the HDN is en- value of K was close to zero while the estimated value of
hanced slightly. the rate constant was close to the actual value of k / ( K C P .
VID4. Nitrogen Compounds. The effect of quinoline This result implies that the rate constant in the numerator
on indole HDN was investigated by Bhinde (1979) (Table and the adsorption parameter in the denominator are
XXIX). The pseudo-first-order rate constants were cal- statistically correlated and should be estimated separately.
culated for the indole network shown in Figure 15. The
addition of 0.5 wt % quinoline to 0.5 wt % indole de- VII. Multicomponent Mixtures
pressed the pseudo-first-order rate constants by one-third, There are only a few quantitative results characterizing
which indicates a substantial inhibition by quinoline and the individual hydroprocessing reactions of mixtures with
its nitrogen-containing products. However, Bhinde ob- more than two organic reactants. Rollmann (1977) de-
served that the inhibition by quinoline of the reactions in termined the relative rates of disappearance of compounds
the indole network was less severe than the inhibition of in various mixtures containing 10 organic compounds.
the reactions of the naphthalene network; in the latter case, Experiments were conducted with a trickle-bed reactor at
the same amount of quinoline reduced the pseudo-first- several space velocities with two Co-Mo/A1203 catalysts.
order rate constants by an order of magnitude. Bhinde Standard reaction conditions were 344 OC, 48.6 atm, and
inferred that indole and its nitrogen-containing products a hydrogen to organic reactant molar ratio of 8; no solvent
were more strongly adsorbed than the aromatic hydro- was used. The feed included combinations of aromatic
carbons. It was not possible to determine from the hydrocarbons as well as sulfur, nitrogen, and oxygen com-
analysis, however, whether indole itself was more strongly pounds. The major products formed from each reactant
adsorbed than the hydrocarbons. were identified, being consistent with those found in the
Bhinde also examined the effect of indole on quinoline pure-component experiments discussed above. Neither
conversion by comparing the pseudo-first-order rate con- quantitative reaction networks nor inhibition effects were
stants in the quinoline network determined in experiments considered. The disappearance of the sulfur and oxygen
with 0.5 wt % quinoline and the indole-quinoline mixture heterocyclics and nitrogen removal from the organo-
mentioned above. The indole affected the pseudo-first- nitrogen compounds followed pseudo-first-order kinetics;
order rate constants by less than 10%. This weak inhib- the hydrogenations of napthalene and 2-methyl-
ition was attributed to the weaker adsorption by indole, naphthalene were reversible, but only the hydrogenation
which is nonbasic. However, one expects that if sufficiently rate constants were determined.
large amounts of indole were present, the concentrations The reactivity of each compound was expressed in terms
of the basic products formed from it (e.g., indoline) would of a pseudo-first-order rate constant for disappearance
have been large enough to inhibit the quinoline reactions. relative to that for naphthalene. Data are given in Table
The inhibition of quinoline conversion by its own re- XXXIII. The most reactive compounds were mercaptans
action products was addressed by Bhinde (1979) in ex- and alcohols. In contrast to the results of Nag et al. (1979),
periments with mixtures of quinoline and naphthalene (or the HDS reactivities of benzothiophene and dibenzo-
dibenzothiophene). The quinoline concentration was thiophene were approximately the same. Dibenzofuran
varied from 0.2 to 2.0 wt % with the concentrations of the had a low reactivity; only mesitylene was less reactive (but
other reactants held constant (Table XXIX). Pseudo- quantitative results were not given). In agreement with
first-order rate constants were estimated for the quinoline the results of Odebunmi and Ollis (1983a), o-cresol was
reaction network; each decreased significantly with in- found to be less reactive than m-cresol.
creasing feed quinoline concentration (Table XXXII). Similar experiments were reported by Badilla-Ohlbaum
In analyzing Bhinde's data, Lo (1981) found that if the et al. (1979a,b),who used a trickle-bed reactor at 68 atm
rate constant k and absorption parameter were estimated with several Ni-Mo catalysts on different supports at 400
simultaneously in a rate expression such as and 480 "C. The space velocity was approximately0.2 (g
kC of feed)/(g of cataIyst*h),calculated on a solvent-free Mi.
r= (17) The feed consisted of dibenzothiophene (10 wt %), quin-
(1 + Kc)m oline (20 wt %), dibenzofuran (7 wt %), and phenanthrene
and the adsorption parameter was large, the estimated (10 wt %) dissolved in n-heptane. The product distribu-
Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991 2055
tion was consistent with that obtained from the individual understanding of hydroprocessing chemistry of heavy
compound experiments discussed earlier. Dibenzo- feedstocks and to guide modeling efforts. However, the
thiophene was completely desulfurized at both tempera- available data are not all consistent with each other, and
tures. Pseudo-first-order rate constants for disappearance the conditions of some of the reported experiments are not
or heteroatom removal were estimated from data at one close to those of commercial practice. The limitations in
space time for the other reactants at 400 "C; reaction in the data indicate many research opportunities.
the presence of a commercial Ni-Mo/A120s catalyst was There is a strong need for good thermodynamics data
characterized by the following relative values of the to determine reaction equilibria aecurately in the liquid
pseudo-first-order rate constants: 5.0, 3.1, and 0.68 for phase, especially for the equilibrium-limited hydrogenation
phenanthrene hydrogenation, quinoline HDN, and di- reactions. It would be of great value to obtain correlations
benzofuran HDO, respectively. The resulta are similar to indicating equilibrium concentrations of species at dif-
Rollmann's, although the conditions were different. ferent temperatures and hydrogen partial pressures for
Chu and Wang (1982) measured the reactivities of di- representative one-, two-, three-, and four-ring aromatic
benzothiophene, aniline, and naphthalene in mixtures; the hydrocarbons and selected sulfur, nitrogen, and oxygen
solvent was a toluenemesitylene mixture. Five com- compounds. Such correlations could be used to determine
mercial catalysts were used. Cumene and pyridine were accurately hydrogen partial pressure effects on the equi-
present in the feed, but their reactivities were not reported. librium position at conditions representative of industrial
The experiments were conducted with vapor-phase reac- practice. The estimates that can now be obtained with
tants at 220-375 "C and 15-55 atm. Dibenzothiophene group contribution methods are too crude for design or
HDS, aniline HDN, and naphthalene hydrogenation fol- modeling purposes.
lowed pseudo-first-order kinetics. Dibenzothiophene HDS Many of the available kinetics data are represented in
was more rapid than the other reactions. The pseudo- an oversimplified manner, usually with pseudo-first-order
first-order rate constants were 17.3, 1.4, and 1.2 h-' for rate equations assumed. More thorough kinetics inves-
dibenzothiophene HDS, naphthalene hydrogenation, and tigations are needed. Effects of temperature need to be
aniline HDN, respectively, catalyzed by Co-Mo/A1203. As investigated further, especially to characterize inhibition
observed by Rollmann (1977) and Badilla-Ohlbaum et al. effects. Because of the potentially large number of ex-
(1979a,b), HDS was faster than HDN. periments required, statistical model discrimination pro-
The rates of disappearance of several compounds cedure and optimal experimental design procedures could
present (with many others) in the neutral oils fraction of play an important role here. Results of such work, in
a coal liquid from the SRC-II process were determined by conjunction with reliable thermodynamics data, would be
Katti et al. (1983). Experiments were performed with a especially useful for modeling commercial adiabatic reac-
batch reactor operated at 355 "C and 36 atm with a Ni- tors operating with significant temperature gradients.
M0/A1203catalyst. The neutral oils fraction was composed
primarily of polynuclear aromatic hydrocarbons and sulfur Additional problems also need to be addressed. The
and oxygen heterocyclic compounds (e.g., thiophenes and complex effects of hydrogen sulfide (an inhibitor of some
furans) (Petrakis et al., 1983); inhibition effects were in- reactions, an apparent promoter of others, and unimpor-
ferred to be less severe than for reactants containing ni- tant in others) need to be explained. Mawtransfer effecta
trogen. on the rates of hydroprocessing reactions also need to be
The rate constants for disappearance of fluorene, explained; whereas most fundamental investigations are
phenanthrene, a methyl-substituted phenanthrene, and performed using small catalyst particles, the catalysts in
pyrene were estimated to be 4.4 X lo4, 2.2 X 1.7 X commercial hydroprocessing reactors consist of extrudatea,
and 6.1 X lo-' L/(g of catalyst-s), respectively. Ev- meaning that mass-transfer effects can be significant, es-
idently, the reactivity of pyrene, which has four fused rings, pecially for high-molecular-weight compounds. Quanti-
was similar to that of phenanthrene, which has three. The tative characterization of catalyst deactivation is also
reactivity of pyrene was only slightly greater than that of needed.
fluorene, which has a benzene moiety; this small difference Even with their limitations, the results presented here
in reactivities is in contrast with the trend observed by are of practical value. For example, a small fraction of the
Sapre and Gates (1981) and Lapinas et al. (1987, 1991), organosulfur compounds (exemplified by 4-methyl- and
who observed a large difference in reactivity between 4,6-dimethyldibenzothiophene)are identified as the key
compounds with a monoaromatic moiety and those with reactants in heavy fossil fuels, Le., those that are desul-
two fused rings. furized most slowly. One could envision processing cor-
The work of Katti et al. is one of several publications relations based on the reactivities of these compounds-
reporting the conversion of chemically similar fractions of and modern analytical methods now allow their quanti-
a heavy coal liquid (Grandy et al., 1986; Li et al., 1985b). tative measurement even in heavy fractions. Additionally,
In a more recent study (Gray, 19901, the reactivity of a the results indicate that organonitrogen compounds are
coker gas oil has been expressed in terms of the rates of the dominant inhibitors, with their adsorption parameters
selected functional groups measured by Nh4R spectroscopy exceeding those of other species by 2 orders of magnitude.
and elemental analysis. Such investigations are among the Estimates can be made as to which reactions are so slow
first to characterize reactivities of hydroprocessing as to be kinetically insignificant and which are so fast as
feedstocks in terms of the chemistry of their components. to attain virtual equilibrium; approximate reaction net-
Development of a systematic basis for connecting pure works and kinetics, including inhibition effects, can then
compound results to whole feedstock resulta in hydro- be formulated.
processing remains a topic of current research. Such a Models such as those envisioned for hydroprocessing
methodology has been proposed recently for the catalytic have proved their worth in other refining processes, in-
cracking of gas oils (Liguras and Allen, 1989a,b). cluding cracking (Jacob et al., 1976) and reforming (Ram-
age et al., 1980). The complexity of the industrial feeds
VIII. Conclusions and reactions suggests that progress toward process models
The data presented here are believed to represent the may be slow, but the rapid advances in the analytical
best pure-compound results available to provide a better chemistry of heavy, complex mixtures resulting from
2056 Ind. Eng. Chem. Res., Vol. 30,No. 9,1991
progress in capillary column gas chromatography, liquid of the Acidic Fractions Characterized by Gas Chromatography/
chromatography, mass spectrometry, and NMR spec- Mass Spectrometry. Ind. Eng. Chem. Process Des. Dev. 1986,25,
troscopy indicate that the opportunities are excellent. 40-48.
Gray, M. R. Lumped Kinetics of Structural Groups: Hydrotreating
of Heavy Distillate. Znd. Eng. Chem. Res. 1990,29,505-512.
IX. Literature Cited Gdtekin, S.;Ali,S. A.; Satterfield, C. N. Effecta of Hydrogen Sulfide
Allen, D. T.; Grandy, D. W.; Jeong, K.-M.; Petrakis, L. Heavier and Ammonia on Catalytic Hydrogenation of Propylbenzene.
Fractions of Shale Oils, Heavy Crudes, Tar Sands, and Coal Li- Ind. Eng. Chem. Process Des. Dev. 1984,23,179-181.
quids: Comparison of Structural Profiles. Ind. Eng. Chem. Pro- Gutberlet, L. C.; Bertolacini,R. J. Inhibition of Hydrodeaulfurization
cess Des. Dev. 1986,24,737-742. by Nitrogen Compounds. Ind. Eng. Chem. Prod. Res. Dev. 1983,
Aubert, C.; Durand, R.; Geneste, P.; Moreau, C. Hydroprocessing of 23,246-250.
Dibenzothiophene, Phenothiazine, Phenoxanthiin, Thianthrene, Hanlon, R. T. Effecta of P F ~ PHn, , and PH /PH on the Hydro-
and Thioxanthene on a Sulfided Ni0-MoO3/yAlZO3Catalyst. J. denitrogenation of Pyridine. Energy Fueg 198& 1,424-430.
Catal. 1986,97,169-176. Ho, T. C. Hydrodenitrogenation Catalysis. Catal. Rev.-Sci. Eng.
Aubert, C.; Durand, R.; Geneste, P.; Moreau, C. Factors Affecting 1988,30,117-160.
the Hydrogenation of Substituted Benzenes and Phenols over a Ho, T. C.; Montagna, A. A,; Steger, J. J. Competitive Hydro-
Sulfded NiO-MoO3/yAlZO3Catalyst. J. Catal. 1988,112,12-20. denitrogenation and Aromatics Hydrogenation. Eighth Interna-
Badilla-Ohlbaum, R.; Chadwick, D. Hydroprocessing of a Model tional Congress on Catalysis Proceedings; Verlag Chemie: Berlin,
Feedstock on a Nickel-Molybdenum Coal-Hydrogenation Cata- 1984;Vol. 11, pp 257-268.
lyst. Fuel 1979,58,834-835. Houalla, M.; Nag, N. K.; Sapre, A. V.; Broderick, D. H.; Gates, B. C.
Badilla-Ohlbaum, R.; Pratt, K. C.; Trimm, D. L. A Study of Nick- Hydrodesulfurization of Dibenzothiophene Catalyzed by Sulfided
el-Molybdate Coal-Hydrogenation Catalysts Using Model Feeds- C00-Mo03/y-Al,03: The Reaction Network. AIChE J. 1978,24,
tocks. Fuel 1979,58,309-314. 1015-1021.
Benson, S. W.; Cruikshank, F. R.; Golden, D. M.; Haugen, G. R.; Houalla, M.; Broderick, D. H.; Sapre, A. V.; Nag, N. K.; de Beer, V.
ONeil, H. E.; Rodgers, A. S.; Shaw. R.; Walsh, R. A Group Con- H. J.; Gates, B. C.; Kwart, H. Hydrodesulfurization of Methyl-
tribution Method for the Estimation of Thermochemical Prop- Substituted Dibenzothiophenes Catalyzed by Sulfided CoO-
erties of Organic Compounds. Chem. Rev. 1969,69,279-324. MOO3/7-A1.& J. Catal. 1980,61,523-527.
Bhinde, M. V. Quinoline Hydrodenitrogenation Kinetics and Reac- Huang, C.4.; Wang, K.-C.; Haynes, H. W., Jr. Hydrogenation of
tion Inhibition. Ph.D. Dissertation, University of Delaware, Phenanthrene over a Commercial Cobalt Molybdenum Sulfide
Newark, 1979. Catalyat under Severe Reaction Conditions. In Liquids from Coal;
Broderick, D. H.; Gates, B. C. Hydrogenolysis and Hydrogenation Ellington, R. T., Ed.; Academic: New York, 1977.
of DibenzothioDhene Catalyzed bv Sulfided CoO-MoOnlr-Al,O,: - " Jacob, S. M.; G r w , B.; Voltz, S. E.; Weekman, V. W., Jr. A Lumping
The Reaction Kinetics. AiChE 3. 1981,27,663-673.-' '

Broderick, D. H.; Sapre, A. V.; Gates, B. C.; Kwart, H. Hydrogena- and Reaction Scheme for Catalytic Cracking. AIChE J. 1976,22,
tion of Aromatic Compounds Catalyzed by Sulfided CoO- 701-713.
Mo03/-pAlz0,. J. Catal. 1982,73,45-49. Johnston, K. P. Hydrogenation-dehydrogenation of Pyrenes Cata-
Chu, C.-I.; Wang, I. Kinetic Study on Hydrotreating. Ind. Eng. lyzed by Sulphided Cobalt-Molybdate at Coal Liquefaction Con-
Chem. Process Des. Dev. 1982,21,336344. ditions. Fuel 1984,63,463-468.
Cocchetto, J. F.; Satterfield, C. N. Thermodynamic Equilibria of Katti, S.S.;Westerman, D. B.; Gates, B. C.; Youngleas, T.; Petrakis,
Selected Heterocyclic Nitrogen Compounds with Their Hydro- L. Catalytic Hydroprocessing of SRC-I1 Heavy Distillate Frac-
genated Derivatives. Ind. Eng. Chem. Process Des. Dev. 1976,15, tions. 3. Hydrodesulfurization of the Neutral Oils. Ind. Eng.
272-277. Chem. Process Des. Dev. 1983,23,773-778.
Cocchetto, J. F.; Satterfield, C. N. Chemical Equilibria among Katzer, J. R.; Sivasubramanian, R. Proceas and Catalyst Needs for
Quinoline and Ita Reaction Producta in Hydrodenitrogenation. Hydrodenitrogenation. Catal. Rev.-Sei. Eng. 1979,20,155-208.
Ind. Eng. Chem. Process Des. Dev. 1981,20,49-53. Kilanowski, D. R.; Teeuwen, H.; de Beer, V. H. J.; Gates, B. C.;
Daly, F. P. Hydrodesulfurization of Benzothiophene over CoO-Mo- Schuit, G. C. A.; Kwart, H. Hydrodesulfurization of Thiophene,
O3-A1,o3 Catalyst. J. Catal. 1978,51,221-228. Benzothiophene, Dibenzothiophene, and Related Compounds
Desikan, P.; Amberg, C. H. Catalytic Hydrodesulphurization of Catalyzed by Sulfided Co0-MoO3/y-Al2O3: Low-Pressure Re-
Thiophene V. The Hydrothiophenes, Selective Poisoning and activity Studies. J. Catol. 1978,55, 129-137.
Acidity of the Catalyst Surface. Can. J. Chem. 1964,42,843-850. Krishnamurthy, S.; Shah, Y. T. Interactions between Dibenzo-
Edelman, M. C.; Maholland, M. K.; Baldwin, R. M. Cowley, S. W. thiophene, 7,8-Benzoquinoline, and Oxygen Compounds During
Vapor-Phase Catalytic Hydrodeoxygenation of Benzofuran. J. Heteroatom Removal. Chem. Eng. Commun. 1982,16,109-117.
Catal. 1988,111, 243-253. Krishnamurthy, S.; Panvelker, S.; Shah, Y. T. Hydrodeoxygenation
Finiels, A,; Geneste, P.; Moulinas, C.; Olive, J.-L. Hydroprocessing of Dibenzofuran and Related Compounds. AIChE J. 1981,27,
of Secondary Amines Over NiW-AlZO3Catalyst. Appl. Catal. 994-1001.
1986,22,257-262. Kwart, H.; Schuit, G. C. A.; Gates, B. C. Hydrodesulfurization of
Frye, C. G. Equilibrium Hydrogenation of Polycyclic Aromatics. J. Thiophenic Compounds: The Reaction Mechanism. J. Catal.
Chem. Eng. Data 1962,7,592-595. 1980,61,128-134.
Frye, C. G.; Weitkamp, A. W. Equilibrium Hydrogenation of Mul- Kwart, H.; Katzer, J. R.; Horgan, J. Hydroprocessing of Pheno-
ti-Ring Aromatics. J. Chem. Eng. Data 1969,14,372-376. thiazine Catalyzed by Co-Mo/y-AlzO,. J. Phys. Chem. 1982,86,
Furimsky, E. Chemistry of Catalytic Hydrodeoxygenation. Catal. 2641.
Rev.-Sci. Eng. 1983a,25,421-458. Lapinas, A. T.; Klein, M. T.; Gates, B. C.; Macris, A,; Lyons, J. E.
Furimsky, E. Mechanism of Catalytic Hydrodeoxygenation of Tet- Catalytic Hydrogenation and Hydrocracking of Fluoranthene:
rahydrofuran. Ind. Eng. Chem. R o d . Res. Dev. 198313,22,31-34. Reaction Pathways and Kinetics. Ind. Eng. Chem. Res. 1987,26,
Gates, B. C.; Katzer, J. R.; Schuit, G. C. A. Chemistry of Catalytic 1026-1033.
Processes; McGraw-Hill: New York, 1979.
Geneste, P.; Amblard, P.; Bonnet, M.; Graffin, P. Hydro- Lapinas, A. T.; Klein, M. T.; Gates, B. C.; Macris, A.; Lyons, J. E.
desulfurization of Oxidized Sulfur Compounds in Benzothiophene, Catalytic Hyrogenation and Hydrocrackingof Fluorene: Reaction
Methylbenzothiophene, and Dibenzothiophene Series over Co- Pathways, Kinetics, and Mechanisms. Ind. Eng. Chem. Res. 1991,
0-MoO3-Al2O3Catalyst. J. Catal. 1980,61,115-127. 30, 42-50.
Geneste, P.; Oliv6, J.-L.; Biyoko, S. Hydroprocessing of Pheno- LaVopa, V.; Satterfield, C. N. Catalytic Hydrodeoxygenation of
thiazine, Dibenzothiophene, and Thianthrene over Ni-Mo/A1203 Dibenzofuran. Energy Fuels 1987,1,323-331.
Catalyst. J. Catal. 1983,83,245-247. LaVopa, V.; Satterfield, C. N. Poisoning of Thiophene Hydro-
Gevert, B. 5.;Otterstedt, J.-E.; Massoth, F. Kinetics of the HDO of desulfurization by Nitrogen Compounds. J. Catal. 1988, 1 IO,
Methyl-Substituted Phenols. Appl. Catal. 1987, 31, 119-131. 375-387.
Gioia, F.;Lee, V. Effect of Hydrogen Pressure on Catalytic Hydro- Lee, C.-L.; Ollis, D. F. Catalytic Hydrodeoxygenation of Benzofuran
denitrogenation of Quinoline. Ind. Eng. Chem. Process Des. Dev. and o-Ethylpenol. J. Catal. 1984a,87,325-331.
1986,25,918-925. Lee, C.-L.; Ollis, D. F. Interactions between Catalytic Hydro-
Grandy, D. W.; Petrakis, L.; Li, C. L.; Gates, B. C. Catalytic Hy- deoxygenation of Benzofuran and Hydrodesulfurization of Di-
droproceasing of SRC-I1Heavy Distillate Fractions 5. Conversion benzothiophene. J. Catal. 1984b,87, 332-338.
Ind. Eng. Chem. Res., Vol. 30,No. 9, 1991 2057
Lemberton, J.-L.; Guisnet, M. Phenanthrene Hydroconversion as a Odebunmi, E. 0.;Ollis, D. F. Catalytic Hydrodeoxygenation 11. In-
Potential Test Reaction for the Hydrogenation and Cracking teractions between Catalytic Hydrodeoxygenation of m-Cresol and
Properties of Coal Hydroliquefaction Catalysts. Appl. Catal. Hydrodesulfurization of Benzothiophene and Dibenzothiophene.
1984,13, 181-192. J. Catal. 1983b, 80,65-75.
Li, C. L.; Xu, Z.; Cao, Z.-A.; Gates, B. C.; Petrakis, L. Hydro- Odebunmi, E. 0.;Ollis, D. F. Catalytic Hydrodeoxygenation 111.
deoxygenation of 1-Naphthol Catalyzed by Sulfided Ni-Mol y- Interactions between Catalytic Hydrodeoxygenation of m-Cresol
A1209. AZChE J. 198Sa, 31, 17Cb174. and Hydrodenitrogenation of Indole. J. Catal. 1983c, 80,76-89.
Li, C. L.; Xu, Z.; Gates, B. C.; Petrakis, L. Catalytic Hydroproceasing Oliv6, J.-L.; Biyoko, S.; Moulinas, C.; Geneste, P. Hydroprocessing
of SRC-I1 Heavy Distillate Fractions. 4. Hydrodeoxygenation of of Indole and o-Ethylaniline over Sulfided CoMo, NiMo, and NiW
Phenolic compounds in the Acidic Fractions. Ind. Eng. Chem. Catalysts. Appl. Catal. 1985,19, 165-174.
Process Des. Deu. 1985b,24,92-97.
Liguras, D. K.; Allen, D. T. Structural Models for Catalytic Cracking. Patzer, J. F., 11; Farrauto, R. J.; Montagna, A. A. Characterization
1. Model Compound Reactions. Znd. Eng. Chem. Res. 1989a, 28, of Coal Liquefaction catalysts Using 1-Methylnaphthalene as a
665-673. Model Compound. Ind. Eng. Chem. Process Des. Deu. 1979,18,
Liguras, D. K.; Allen, D. T. Structural Models for Catalytic Cracking. 625-630.
2. Reactions of Simulated Oil Mixtures. Ind. Eng. Chem. Res. Petrakis, L.; Ruberto, R. G.; Young, D. C.; Gates, B. C. Catalytic
1989b, 28,674-683. Hydroprocessing of SRC-I1Heavy Distillate Fractions. 1. Prep-
Lipsch, J. M. J. G.; Schuit, G. C. A. The Co0-Mo03-A1203Catalysts aration of the Fractions by Liquid Chromatography. Znd. Eng.
111. Catalytic Properties. J. Catal. 1969, 15, 179-189. Chem. Process Des. Deu. 1983,22,292-297.
Lo, H. S. Kinetic Modeling of Hydrotreating. Ph.D. Dissertation, Petrocelli, F. P.; Klein, M. T. Modeling Lignin Liquefaction 1.
University of Delaware, Newark, 1981. Catalytic Hydroproceasing of Lignin-related Methoxyphenols and
Malakani, K.; Magnoux, P.; Perot, G. Hydrodenitrogenation of 7,8- Interaromatic Unit Linkages. Fuel Sci. Technol. Int. 1987, 5,
Benzoquinoline over Nickel Molybdenum Alumina. Appl. Catal. 25-62.
1987,30,371-375. Quann, R. J.; Ware, R. A.; Hung, C.-W.; Wei, J. Catalytic Hydrode-
Mathur, K. N.; Sarbak, Z.; Islam, N.; Kwart, H.; Katzer, J. R. metallation of Petroleum. Adv. Chem. Eng. 1988, 14, 95-259.
'Kinetics and Mechanism of Catalytic Hydroprocessing of Com-
ponents of Coal-Derived Liquids: Tenth and Eleventh Quarterly Ramage, M. P.; Graziani, K. R.; Krambeck, F. J. Development of
Reports for the Period August 16, 1981 to February 15, 1982"; Mobil's Kinetic Reforming Model. Chem. Eng. Sci. 1980, 35,
Prepared for Office of Fossil Energy, Department of Energy, 41-48.
Washington, DC, 1982a. Reid, R. C.; Prausnitz, J. M.; Polling, B. E. The Properties of Gases
Mathur, K. N.; Schrenk, M. D.; Kwart, H.; Katzer, J. R. and Liquids, 4th ed.; McGraw-Hilk New York, 1987.
'Development of Unique Catalysts for Hydrodenitrogenation of Rollmann, L. D. Catalytic Hydrogenation of Model Nitrogen, Sulfur,
Coal-Derived Liquids: Final Report for the Period September 15, and Oxygen Compounds. J. Catal. 1977,46, 243-252.
1978 to September 1981"; Prepared for Office of Fossil Energy, Sapre, A. V.; Gates, B. C. Hydrogenation of Aromatic Hydrocarbons
Department of Energy, Washington, DC, 1982b. Catalyzed by Sulfided Co0-Mo03/r-A1203: The Reaction Net-
McIlvried, H. G. Kinetics of the Hydrodenitrification of Pyridine. works. Ind. Eng. Chem. Process Des. Deu. 1981,20,68-73.
Ind. Eng. Chem. Process Des. Deu. 1971, 10,125-130. Sapre, A. V.; Gates, B. C. Hydrogenation of Biphenyl Catalyzed by
Miciukiewicz, J.; Zmierczak, J.; Massoth, F. E. The Effect of N- Sulfided Co0-Mo03/y-A1203: The Reaction Kinetics. Ind. Eng.
Compound Poisoning on HDS and Hydrogenation Activities over Chem. Process Des. Deu. 1982,21,86-94.
Co-Mo/A120SCatalysts. Eighth International Congress on Ca-
talysis Proceedings; Verlag Chemie: Berlin, 1984; Vol. 11, pp Sapre, A. V.; Broderick, D. H.; Fraenkel, D.; Gates, B. C.; Nag, N.
671-682. K. Hydrodesulfurization of Benzo[b]naphtho[2,3-d] thiophene
Miller, J. T.; Hineman, M. F. Non-First-Order Hydrodenitrogenation Catalyzed by Sulfided Co0-MoOS/y-Al2Os: The Reaction Net-
Kinetics of Quinoline. J. Catal. 1984,85, 117-126. work. AIChE J. 1980,26,690-694.
Moreau, C.; Durand, R.; Geneste, P.; Oliv6, J.; Bachelier, J.; Cornet, Satterfield, C. N.; Roberta, G. W. Kinetics of Thiophene Hydrogen-
D.; Duchet, J. C.; Lavalley, J. C.; Bonnelle, J. P.; Grimplot, J.; olysis on a Cobalt Molybdate Catalyst. AIChE J. 1968, 14,
Kasztelan, S.; Payen, E.; Breysse, M.; Cattenot, M.; Decamp, T.; 159-164.
Frety, R.; Gachet, C.; Lacroix, M.; Leclerq, C.; De Mourgues, L.; Satterfield, C. N.; Cocchetto, J. F. Reaction Network and Kinetics
Portefaix, J. L.; Vrinat, M.; Engelhard, P.; Gueguen, C.; Toulhoat, of the Vapor-Phase Catalytic Hydrodenitrogenation of Quinoline.
H. Chemical Evidence for the Existence of Two Types of Catalytic Ind. Eng. Chem. Process Des. Deu. 1981,20,53-61.
Sites for Hydroprocessing of Substituted Benzenes over NiW- Satterfield, C. N.; Giiltekin, S. Effect of Hydrogen Sulfide on the
(Mo)/y-A1203Hydrotreating Catalyst. Prepr.-Am. Chem. SOC., Catalytic Hydrodenitrogenation of Quinoline. Znd. Eng. Chem.
Div. Pet. Chem. 1987,32, 298-302. Process Des. Deu. 1981,20, 62-68.
Moreau, C.; Durand, R.; Zmimita, N.; Geneste, P. Hydro-
denitrogenation of Benzo(0quinoline and Benzo(h)quinoline over Satterfield, C. N.; Yang, S. H. Simultaneous Hydrodenitrogenation
a Sulfided Ni0-MoO3/y-Al2OS. J. Catal. 1988, 112, 411-417. and Hydrodeoxygenation of Model Compounds in a Trickle Bed
Nag, N. K. On the Mechanism of the Hydrogenation Reactions Oc- Reactor. J. Catal. 1983,81,335-346.
curring Under Hydroprocessing Conditions. Appl. Catal. 1984, Satterfield, C. N.; Yang, S. H. Catalytic Hydrodenitrogenation of
10, 53-62. Quinoline in A Trickle-Bed Reactor. Comparison with Vapor
Nag, N. K.; Sapre, A. V.; Broderick, D. H.; Gates, B. C. Hydro- Phase Reaction. Ind. Eng. Chem. Process Des. Dev. 1984, 23,
desulfurization of Polycyclic Aromatics Catalyzed by Sulfided 11-19.
Co0-MoO3/y-Al2OS:The Relative Reactivities. J. Catal. 1979, Satterfield, C. N.; Smith, C. M. Effect of Water on the Catalytic
57, 50!+512. Reaction Network of Quinoline. Ind. Eng. Chem. Process Des.
Nagai, M.; Kabe, T. Selectivity of Molybdenum Catalyst in Hydro- Deu. 1986,25,942-949.
desulfurization, Hydrodenitrogenation, and Hydrodeoxygenation: Satterfield, C. N.; Modell, M.; Mayer, J. F. Interactions between
Effect of Additives on Dibenzothiophene Hydrodesulfurization. Catalytic Hydrodesulfurization of Thiophene and Hydro-
J. Catal. 1983,81,440-449. denitrogenation of Pyridine. AIChE J. 1975,21, 1100-1107.
Nagai, M.; Masunaga, T.; Hana-oka, N. Selectivity of Molybdenum Satterfield, C. N.; Modell, M.; Wilkens, J. A. Simultaneous Catalytic
Catalyst in Hydrodenitrogenation, Hydrodesulfurization, and Hydrodenitrogenation of Pyridine and Hydrodesulfurization of
Hydrodeoxygenation: Effecta of Sulfur and Oxygen Compounds Thiophene. Ind. Eng. Chem. Process Des. Deu. 1980,19,154-160.
on Acridine Hydrodenitrogenation. J. Catal. 1986a, 101,284-292.
Nagai, M.; Sato, T.; Aiba, A. Poisoning Effect of Nitrogen Com- Satterfield, C. N.; Smith, C. M.; Ingalls, M. Catalytic Hydro-
pounds on Dibenzothiophene Hydrodesulfurization on Sulfided denitrogenation of Quinoline. Effect of Water and Hydrogen
NiMo/A120s Catalyst and Relation to Gas-Phase Basicity. J. Sulfide. Ind. Eng. Chem. Process Des. Deu. 1985,24,1000-1004.
Catal. 1986b, 97,52-58. Sauer, N. N.; Markel, E. J.; Schrader, G. L.; Angelici, R. J. Studies
OBrien, W. S.; Chen, J. W.; Nayak, R. V.; Carr, G. S. Catalytic of the Mechanism of Thiophene Hydrodesulfurization: Conver-
Hydrodesulfurization of Dibenzothiophene and a Coal-Derived sion of 2,3 and 2,5-Dihydrothiophene and Model Organometallic
Liquid. Znd. Eng. Chem. Process Des. Deu. 1986,25, 221-229. Compounds. J. Catal. 1989, 117, 295-297.
Odebunmi, E. 0.; Ollis, D. F. Catalytic Hydrodeoxygenation I. Schuit, G. C. A.; Gates, B. C. Chemistry and Engineering of Catalytic
Conversions of 0 - , p-, and m-Cresols. J. Catal. 1983a, 80,56-64. Hydrodesulfurization. AIChE J. 1973, 19,417-438.
2068 Ind. Eng. Chem. Res., Vol. 30,No.9,1991
Sebastian, H. M.; Lin, H. M.; Chao, K. C. Correlation of Solubility Streitwieser, A,, Jr.; Heathcock, C. H. Introduction to Organic
of Hydrogen in Hydrocarbon Solvents. AIChE J. 1981, 27, Chemistry; Macmillan: New York, 1976.
138-148. Stull, D. R.; Westrum, E. F., Jr.; Sinke, G.C . The Chemical Ther-
Shabtai, J.; Veluswamy, L.; Oblad, A. G.Steric Effects in Phenan- modynamics of Organic Compounds: Wiley: New York, 1969.
threne and Pyrene Hydrogenation Catalyzed by Sulfided Ni-W/ Van Krevelen, D.W.; Chermin, H. A. G. Estimation of the Free
AlzOa. Prepr. Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1978a,23, Enthalpy (Gibbs Free Energy) of Formations of Organic Com-
107-113. pounds from Group Contributions. Chem. Eng. Sci. 1951, 1 ,
Shabtai, J.; Veluswamy, L.; Oblad, A. G.Steric Effects in the Hy- 66-80.
drogenation-Hydrodenitrogenationof Isomeric Benzoquinolines Van Parijs, I. A.; Froment, G.F. Kinetics of Hydrodesulfurization
Catalyzed by Sulfided Ni-W/AlZOa. Prepr. Pap.-Am. Chem. on a Co-Mo/y-A120SCatalyst. 1. Kinetice of the Hydrogenolysia
SOC.,Diu. Fuel Chem. 1978b,23, 114-118. of Thiophene. Ind. Eng. Chem. prod. Res. Deu. 1986,25,431-436.
Shabtai, J.; Yeh, G. J. C.; Russell, C.; Oblad, A. G. Fundamental Van Parijs, I. A.; Hosten, L. H.; Froment, G.F. Kinetics of Hydro-
Hydrodenitrogenation Studies of Polycyclic N-Containing Com- desulfurization on a Co-Mo/y-Alp03Catalyst. 2. Kinetics of the
pounds Found in Heavy Oil. 1. 5,6-Benzcquinoline. Ind. Eng. Hydrogenolysis of Benzothiophene. Ind. Eng. Chem. Prod. Res.
Chem. Res. 1989,28,139-146. Dev. 1986,25,437-443.
Shaw, R.;Golden, D. M.; Benson, S. W. Thermochemistry of Some Vogelzang, M. W.; Li, C.-L.; Schuit, G. C. A.; Gates, B. C.; Petrakis,
Six-Membered Cyclic and Polycyclic Compounds Related to Coal. L. Hydrodeoxygenation of 1-Naphthol: Activities and Stabilities
J. Phys. Chem. 1977,81,1716-1729. of Molybdena and Related Catalysts. J. Catal. 1983,84,170-177.
Shih, S. S.;Mathur, K. N.; Katzer, J. R.; Kwart, H.; Stiles, A. B. Vrinat, M. L.; The Kinetics of the Hydrodesulfurization Process: A
Quinoline Hydrodenitrogenation: Reaction Network and Kinet- Review. Appl. Catal. 1983,6,137-158.
ics. Prepr.-Am. Chem. SOC.,Diu. Pet. Chem. 1977,22,919-940. Whitehurst, D.D. A Primer on the Chemistry and Constitution of
Singhal, G. H.; Espino, R. L.; Sobel, J. E. Hydrodesulfurization of Coal, Larsen, J. W., Ed.; ACS Symposium Series; American
Sulfur Heterocyclic Compounds. Reaction Mechanisms. J. Catal. Chemical Society: Washington, DC, 1978; Vol. 71, pp 1-35.
1981,67,446-456. Wiser, W. H.; Singh, S.; Qader, S. A.; Hill, G. R. Catalytic Hydro-
Sonnemans, J.; van den Berg, G. H.; Mars, P. The Mechanism of genation of Multiring Aromatic Coal Tar Constitutents. Id. Eng.
Pyridine Hydrogenolysis on Molybdenum-Containing Catalysts Chem. Prod. Res. Dev. 1970,9,360-357.
11. Hydrogenation of Pyridine to Piperidine. J. Catal. 1973,31, Wu, W. L.; Haynes, H. W., Jr. Hydrocracking Condenaed-Ring
220-230. Aromatics Over Nonacidic Catalysts; ACS Symposium Series;
Speight, J. G. The Desulfurization of Heavy Oils and Residua; American Chemical Society: Washington, DC, 1975;Vol. 20,pp
Marcel Dekker: New York, 1981. 65-81.
Stein, S. E.; Golden, D. M.; Benson, S. W. Predictive Scheme for Yang, S. H.; Satterfield, C. N. Catalytic Hydrodenitrogenation of
ThermochemicalProperties of PolycyclicAromatic Hydrocarbons. Quinoline in a Trickle-Bed Reactor. Effect of Hydrogen Sulfide.
J. Phys. Chem. 1977,81,314-317. Znd. Eng. Chem. Process Des. Dev. 1984,23,20-25.
Stephens, H. P.; Chapman, R. N. The Kinetics of Catalytic Hydro- Zawadaki, R.: Shih,S. S.; Reiff, E.; Katzer, J. R;Kwart, H. "Kinetics
genation of Pyrene: Implications for Direct Coal Liquefaction and Mechanism of Catalytic Hydroprocessing of Components of
Processing. Prepr. Pap.-Am. Chem. Soc., Diu. Fuel Chem. 1983, Coal-Derived Liquids: Tenth and Eleventh Quarterly Reports for
28,161-168. the Period August 16,1981 to February 15,1982";Prepared for
Stephens, H. P.; Kottenstette, R. J. The Kinetics of Catalytic Hy- Office of Fossil Energy, Department of Energy, Washington, DC,
drogenation of Polynuclear Aromatic Components in Coal Li- 1982.
quefaction Solvents. Prepr. Pap.-Am. Chem. SOC.,Div. Fuel
Chem. 1985,30,345-353. Received for review November 19,1990
Stem, E. W. Reaction Networks in Catalytic Hydrodenitrogenation. Revised manuscript received May 16,1991
J. Catal. 1979,57,39+396. Accepted May 31, 1991

You might also like