You are on page 1of 17

Znd. Eng. Chem. Res.

1995,34, 101-117 101

Polynuclear Aromatic Hydrocarbons Hydrogenation. 1.


Experimental Reaction Pathways and Kinetics
Styliani C. Korret and Michael T. Klein
Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716

Richard J. Quann
Mobil Research and Development Corporation, Paulsboro Research Laboratory, Paulsboro, New Jersey 08011

The relationship between molecular structure and hydrogenation reactivity in heavy oil
hydroprocessing was sought via the elucidation of the controlling reaction pathways and kinetics
of one-, two-, three-, and four-fused ring compounds. Hydrogenation reactions of o-xylene,
tetralin, naphthalene, phenanthrene, anthracene, pyrene, and chrysene and their multicompo-
nent mixtures were studied in cyclohexane solvent using a presulfided CoMo/AlzOs catalyst in
an l-L batch autoclave a t PH,= 68.1 atm and T = 350 C.Quantitative network analysis allowed
estimation of 45 hydrogenation rate parameters and a n equal number of equilibrium ratios for
36 aromatic and hydroaromatic compounds. These values were then used in the evaluation of
five adsorption parameters for aromatic ring number-based lumps from the multicomponent
mixture experiments. Three classes of hydrogenation were discerned, based on the magnitude
of numerator rate parameters. The adsorption parameters clearly increased with increasing
aromatic ring number.

Introduction 1972; Qader, 1973; Qader and McOmber, 1975; Salim


and Bell, 1984) hydrogenation have been examined in
The increased attention paid to catalytic hydropro- considerable detail as well. Pyrene hydrogenation has
cessing in recent years is due in part to stricter been analyzed in the context of its hydrogen-donating
environmental legislation and the public demand for ability for coal liquefaction (Qader and Hill, 1972;
cleaner fuels. Hydroprocessing technology is unique in Shabtai et al., 1978; Haynes et al., 1983; Stephens and
that it can handle the sulfur and nitrogen heteroatoms Chapman, 1983; Johnston, 1984; Stephens and Kotten-
in heavy feeds while increasing the hydrogen to carbon stette, 1985). Polynuclear aromatic hydrocarbons con-
ratio and thus the fuel value of the oil. Hydroprocessing taining five-membered rings (fluoranthene, fluorene)
chemistry involves bifunctional catalysts, where a metal were examined more recently under hydrocracking
function promotes hydrogenation and an acidic function conditions by Lapinas and co-workers (Lapinas et al.,
promotes isomerization, ring opening, and dealkylation. 1987; Lapinas, 1989). Finally, the equilibrium-related
The hydrogenation function is important from the point thermochemistry of these molecules has also been
of view of product hydrogen to carbon ratio and in the examined (Frye, 1962; Frye and Weitcamp, 1969; Shaw
establishment of the balance of olefin (aromatic) and
et al., 1977; Stein et al., 1977).
alkane (naphthenic) characters of the molecule.
Heavy oils can contain an appreciable amount of In spite of this impressive literature base, the avail-
polynuclear aromatic hydrocarbons (PNAs). This has able data are often fragmented and catalyst dependent.
motivated keen interest in the discernment of hydro- Few generalizations with respect to reactivity can be
genation pathways, kinetics, and mechanisms through made, since the impact of different conditions is fre-
the use of model or pure component experiments. quently unclear. For example, there appears to be
Moreover, since hydrogenations are exothermic, higher agreement that hydrogenation reactivity increases with
temperatures applied to enhance acid center transfor- the number of aromatic rings (Girgis and Gates, 1991;
mations, for example, benefit dehydrogenations more Neurock and Klein, 1993), but the effect of alkyl
than hydrogenations. It is therefore of interest to substituents and/or naphthenic rings is not clear. Op-
examine not only rate but also saturation equilibrium posing suggestions have been offered (Shabtai et al.,
issues in heavy oil catalytic hydroprocessing. 1978; Moreau and Geneste, 1990). The effect of com-
The field has been recently reviewed (Moreau and petitive adsorption and subsequent inhibition (Bhinde,
Geneste, 1990; Girgis and Gates, 1991). A large body 1979; LaVopa and Satterfield, 1988) has not been
of information for single-ring aromatics hydrogenation pursued at length. There is often controversy sur-
exists (Aubert et al., 1988; Moreau et al., 1990). Naph- rounding the pathways of even the most extensively
thalene (Qader and Hill, 1972; Qader, 1973; Salim and studied molecules, such as the direct conversion of di-
Bell, 1982; Zeuthen et al., 1987), phenanthrene (Qader to tetrahydrophenanthrene. Finally, the results re-
et al., 1973; Qader and McOmber, 1975; Wu and ported often do not contain detailed quantitative kinet-
Haynes, 1975; Huang et al., 1977; Shabtai et al., 1978; ics.
Haynes et al., 1983; Lemberton and Guisnet, 1984; Thus, the field is far from settled in terms of reactivity
Salim and Bell, 19841, and anthracene (Qader and Hill, information. Quantitative networks are needed under
consistent reaction conditions for a wide array of
* Author to whom correspondence should be addressed aromatics and hydroaromatics. This motivated the
(klein@he.udel.edu). current study of hydrogenation of compounds bearing
+ Present address: E x o n Research and Engineering Com- from one to four aromatic rings, such as o-xylene,
pany, P.O.Box 101,Florham Park, NJ 07932. tetralin, naphthalene, phenanthrene, anthracene, pyrene,
0888-5885/95/2634-0101$09.00/0 0 1995 American Chemical Society
102 Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995

and chrysene. Hydrogenation yields the respective nmes


hydroaromatics, and carefhl product identification has
enabled estimation of their rate and equilibrium pa- i=l
rameters from experiments with the aromatics as
reactants. Finally, multicomponent mixture experi- The Langmuir-Hinshelwood-Hougen-Watson
ments enabled the evaluation of adsorption constants (LHHW)kinetics formalism was used to account for the
for aromatic ring number-based lumps. competitive adsorption of all aromatic hydrocarbons
(Froment and Bischoff, 1990). The rate law reflected
Experimental Procedure the model of a reversible bimolecular surface reaction
as the rate-determining step (Qader, 1973; Landau,
Materials and Equipment. Hydrogenation experi- 1991). Nevertheless, since there was no statistical
ments were performed in a l-Lbatch autoclave, equipped difference between models for a unimolecular and a
with a spinning catalyst basket, reactant loader for good bimolecular surface reaction, the simpler case of uni-
zero time determination, on-line data acquisition, and molecular reaction was adopted. The reaction order in
continuous sampling capabilities. A detailed description hydrogen was not considered explicitly, since its pres-
is available elsewhere (Landau, 1991). Reactions took sure was the same for all experiments and was kept
place in the presence of 420 g of cyclohexane as a solvent constant throughout a run (Bhinde, 1979).
and 68.1 atm of Hz at 350 "C. The total pressure, Individual estimation of adsorption parameters for
including cyclohexane vapor pressure at 350 "C, was each compound examined was statistically uncertain
191.6 atm. Hydrogen back-pressure was kept constant because of the relatively low initial concentration of
throughout the reaction. The selected conditions were reactants combined with the multitude of products.
intended to emphasize hydrogenation rate measure- Thus the adsorption parameters, Km, were lumped
ments. Thus, sampling concentrated on the early according t o the number 0 I1 I 4 of aromatic rings
hydrogenation stages. Nevertheless, long-time experi- (Neurock and Klein, 1993). This required estimation
ments were also scheduled to provide hydrogenation of only five adsorption parameters, KL, for the whole
equilibrium information. series of experiments. Implicit is the approximation
The reacting mixture was in a one-phase supercritical that the adsorption parameter is weakly dependent on
state in all cases, as verified both by calculations the alkyl substituents, at least for the narrow range
employing the Virial equation of state (Reid et al., 1987) examined. This approximation leads to the inevitable
and also from the actual experimental vapor pressure loss of detail during the estimation of numerator rate
data. The excellent material balance closures further parameters, k ~ .They thus implicitly contain an adsorp-
attest to the absence of crystallization. Cyclohexane tion constant dependence. The validity of this
was subject to limited isomerization (x .e 2 x to approximation is assessed in detail in a following
methylcyclopentane during heatup, which continued publication (Korre et al., 1995).
during the typical reaction time of 3 h without any Equation 2 summarizes the foregoing discussion in
apparent effect of the presence of the aromatic reactant. terms of the operative LHHW rate law used to reduce
An average additional conversion of x = 8 x was the experimental data to rate and adsorption param-
observed over the course of the reaction. eters:
The catalyst was 10 g of CoMo/AlzOs (1.7% Co, 7.0%
Mo) in particles of 0.3 cm and was presulfided in the
autoclave for 135 min at 400 "C in a stream of 10%HzS
in Hz. CSz (1mL) was added with each experiment to
keep the catalyst in a sulfided state. The catalyst was
equilibrated for 10 h at 350 "C and 68.1 atm of Hz
during three naphthalene hydrogenation experiments In eq 2, r~ (mol kg,,t-l s-l) is the rate of conversion of
in cyclohexane (Landau, 1991). All experiments were compound i to compoundj, Ci (mol L-l) is the con-
performed using the same catalyst charge, and the centration, k~ (L kg,,-l s-l) is the numerator rate
steady-state catalytic activity was periodically verified parameter, including surface reaction and adsorption
by naphthalene hydrogenation kinetics (Korre, 1994). contributions, KU (molj/mol i) is the equilibrium ratio,
Routine sample analysis was performed immediately Km (L mol-l) is the adsorption parameter of individual
after sampling on a Hewlett-Packard 5880A flame compounds, and KL(L mol-l) is the adsorption param-
ionization gas chromatograph employing a J&W Scien- eter of aromatic ring-number based lumps. It follows
tific DB-5 fused silica capillary column. A Hewlett- that the overall equilibrium constant is K,$ = K~~/(PHJ~,
Packard 5970 mass selective detector connected t o the where a is the reaction stoichiometry w t h respect t o
outlet stream of a Hewlett-Packard 5890 gas chromato- hydrogen. This is because the forward rate parameters
graph was also used for product identification. Yields contain a hydrogen partial pressure dependence. The
less than 0.1% could be safely detected at all cases. overall rate of production of a given compound Ci was
Response factors were established for the commercially obtained as the summation of all the conversion rates
available reactants and several of the products. For the involving species i.
bulk of the hydroaromatics, response factors were set The rate parameters and equilibrium ratios were
equal to that of the aromatic with the same number of regressed individually for each compound, with the
rings. constraint of non-negativity. Equilibrium ratios were
Parameter Estimation. Nonlinear parameter es- further constrained during parameter estimation by
timation was performed using a Simplex optimization application of the Denbigh rule concerning equivalent
algorithm (Press et al., 1986) and the LINPAC DASSL pathways in a network.
predictor/corrector method ODE solver. The objective The parameter estimation strategy examined both
function F was the sum of the squares of the differences relative and absolute rates, in turn. For a given
between experimental and calculated yields: compound, pure component and mixture experimental
Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995 103
Table 1. Experimental Plan for Pure Compound those structures yielded the respective hydroaromatics.
Hydrogenation Specifically, the hydrogenation reactions of o-xylene
NA' reactant loading (g) (h)
tmax (Aldrich, 99%), tetralin (Aldrich, 99%), naphthalene
1 o-xylene 34.18 5.993 (Aldrich, 99+%), phenanthrene (Aldrich, 98+%), an-
1 tetralin 29.75 3.000 thracene (Aldrich, 97%), pyrene (Aldrich, 99+%), and
2 naphthalene 23.90 3.147 chrysene (Aldrich, 98%) were studied.
3 phenanthrene 15.18 3.000 The mixture experiments summarized in Table 2
3 anthracene 2.06 3.000 allowed quantitative assessment of competitive adsorp-
4 PPene 5.75 3.000 tion and inhibition interference. Single-ring aromatics
4 chrysene 2.0 4.733
were represented by o-xylene, two-ring aromatics by
a N A = number of aromatic rings. naphthalene, three-ring aromatics by phenanthrene,
Table 2. Initial Aromatics Distribution (wt %) for
and four-ring aromatics by pyrene and chrysene. The
Mixture Experiment@ initial distributions for the PNA mixture experiments
summarized in Table 2 were selected so as to mimic
NA* reactant GAUS LCO SR TCFD EQUIL aspects of the aromatic distributions in both straight-
1 o-xylene 13.00 13.01 20.00 24.00 0.00 run and processed oil fractions (van der Eijk et al.,
2 naphthalene 20.04 61.51 14.01 19.00 27.80 1990). The LCO distribution is triangular, centered at
3 phenanthrene 33.96 22.02 14.00 26.00 27.56 two-ring aromatics (i.e., naphthalene). The SR distribu-
4 pyrene 20.11 3.50 32.00 19.01 27.92
4 chrysene 12.99 0.00 20.01 12.00 16.72 tion was approximated by an inverted triangular dis-
tribution. The TCFD and EQUIL distributions had
loading(g) 20.02 20.01 20.00 20.00 17.94 approximately equal amount of PNA's with two to four
tmax (h) 6.939 7.000 7.638 6.295 9.574
rings, but EQUIL did not include single-ring aromatics.
a GAUS = Gaussian, LCO = light cycle oil, SR = straight run. The gaussian (GAUS) distribution was centered at
TCFD = thermally cracked flashed distillate, EQUIL = equilib- three-ring aromatics (Korre, 1994).
rium run. NA = number of aromatic rings.
The cumulative wt % distribution is also of interest.
data (where available) were used simultaneously. As The SR distribution had the highest content of four-ring
a first step, parameter estimation was performed using aromatics, with 52.01 wt %, and the EQUIL distribution
the product yield vs reactant conversion data involving had the highest combined content in three- and four-
a particular compound (ri VS.XI). These relative rates ring aromatics, with 72.20 wt % (Table 2). The LCO
eliminated the LHHW denominators and provided rate distribution had the lowest content in both cases (3.50
and equilibrium parameters that were independent of wt % four-ring aromatics, 25.52 wt % three- and four-
the mixture composition (eq 3): ring aromatics combined).
These experiments are considered below on a com-
pound by compound basis. Product identification issues
are addressed first. Then the product rank in the
network, as provided mainly by the yield and selectivity
vs conversion information (Delplots), is considered
(Bhore et al., 1990). Subsequent parameter estimation
using the kinetics data with the proposed network
affords rate, equilibrium, and adsorption parameters,
In eq 3, CO(mom) is the initial reactant concentration. as discussed above.
The rate parameters k~ and equilibrium ratios KG
regressed by eq 3 were then held constant, and eq 2 was Kinetics and Thermodynamics Results
used for the estimation of adsorption parameters, Kl,
using the yield vs time data, simultaneously for all o-Xylene. Reaction of o-xylene in 420. g of cyclohex-
experiments involving a particular compound and for ane solvent was a t 68.1 atm of H2 and 350 "C. The
all compounds in the mixture. This assured maximum conditions for the experiments involving o-xylene are
agreement and between experiments with varying presented in Tables 1-3. Reaction duration was 4.637
initial compositions (Korre, 1994). tf < 7.638 h, and the ultimate conversion was 0.138
Experimental Design. The hydrogenation experi- < XM < 0.183. The major products were truns-1,2-
mental plan is outlined in Tables 1 and 2. Pure- dimethylcyclohexane, cis-1,2-dimethylcyclohexane, m-
compound experiments were complemented by polynu- xylene, and p-xylene. Traces of cis- and truns-1,3-
clear aromatics mixture experiments at different initial dimethyl- and -1,4-dimethylcyclohexaneswere also
loadings. detected. All products were identified by co-injections
The model compounds of Table 1were selected so as and mass spectral information.
to represent fundamental structural characteristics The kinetics of conversion corresponding t o the mix-
encountered in heavy oils. The structural parameter ture experiments in Table 2 are presented in Figure 1.
varied for the reactant selection was the aromatic ring The highest conversion for a particular time was
number (one to four aromatic rings). Hydrogenation of observed for experiments with the LCO distribution
Table 3. Mass Balances and Maximum Conversions for Compounds in the Mixture Experiments
o-xylene naphthalene phenanthrene PFene chrvsene
MB closure (%) xmax MB closure (%) xmax MB closure (%) xmax MB closure (%) xmax MB closure (%) xmax
Pure 0.984f 2.07 0.183 0.981 f 3.101 0.952 0.990 f 2.564 0.977 0.945f 1.900 0.858 0.969f 2.826 0.999
Gauss 0.997 f 2.01 0.151 0.976f 3.447 0.984 0.973f 1.381 0.959 0.966 f 4.585 0.832 0.950 f 7.035 0.992
LCO
SR
*
0.989 0.79
0.994f 1.34
0.156
0.155
0.958f 3.987 0.976 0.988f 3.676 0.959 0.993f 6.536
0.978f 4.007 0.982 0.970f 3.517 0.953 0.944f 4.707
0.851
0.855
-
0.962 f 3.834
-
0.984
TCFD 1.005 f 1.67 0.145 0.985 f 7.062 0.989 0.994f 7.297 0.959 0.973f 5.546 0.887 0.956f 10.68 0.992
EQUIL - - 0.972f 4.763 0.998 0.978f 7.923 0.996 0.952f 8.866 0.915 0.976i 7.067 1.000
104 Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995

the selectivity vs conversion Delplots (Bhore et al., 1990)


e 0.14
of Figure 2(ii).
0 A replicate experiment produced identical results, as
'E
El
0.12 is depicted in Figure 2(ii). Thus the apparent zero
:: 0.1 initial selectivity to hydrogenation, relative to isomer-
u" 0.08 ization, is telling. Perhaps subtleties in the differences
in adsorption of the xylene isomers are manifested as
=,Ic 0.06 apparent autocatalytic kinetics. This would have caused
8 0.04 the dimethylcyclohexaneproduction rate to increase as
the o-xylene concentration decreased. However, the
8 0.02 maximum conversion is not likely enough to account for
0 the observed production acceleration. This is reflected
0 120 240 360 480 in a small regressed value of adsorption parameter
Time (min) (maximum 1.364 m3 kmol-I, with eq 21, which cannot
Figure 1. Effect of reacting mixture composition on o-xylene account for the observed kinetics.
SR; (A)TCFD; (-1
conversion kinetics. ((0)GAUS; (+) LCO; (0) The apparent secondary rank of the dimethylcyclo-
Est.) hexanes could also result from the autocatalytic action
of hydrogen transfer from the hydrogenated product,
0.16
1i1 I according to the combined unimolecular and bimolecular
0.14 rate expression of eq 4. The plausibility of this scheme
0.12 derives from the several conjugated structures possible
0. I
after abstraction of a tertiary hydrogen. This somewhat
0 empirical scheme fit the experimental data extremely
.I0.08
+ well. This is demonstrated by the continuous lines of
0.06 Figure 2(i).
0.04
0.02
n
0 60 120 180 240 300 360
Time (min)

1 (4)

0.8
However, the quantitative aspects of eq 4 undermined
2! the hydrogen-transfer scenario. Note that the best fit
*
-
.- 0.6
.-*
equilibrium ratio is less than 1 (0.175). This implied
that o-xylene would be more abundant than 1,2-di-
-g
m
Y
0.4 methylcyclohexane at equilibrium, contradicting previ-
ous studies that reported equilibrium ratios for single-
0.2 ring hydrogenations well over 10 at 350 "C(Frye, 1962;
Frye and Weitcamp, 1969). This suggested that the
0 critical slopes in Figure 2(i) that led to the hydrogen-
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 transfer hypothesis were overemphasized a t the low
ortho-xylene conversion x conversions.
Figure 2. Kinetics of o-xylene hydrogenation. (i) Yield vs time This suspicion motivated a set of experiments aimed
and (ii) selectivity vs conversion plots. (i) Solid curves represent a t broadening the range of conversions to probe the
the network correlation with eq 4.Dotted curves represent the hydrogen-transfer hypothesis further. A series of ex-
network correlation of Figure 4.F = 1.636 x estimated periments at conditions identical to those for o-xylene
deviation = 5.720 x ((+) o-xylend7; (0)cis- and truns-l,2- were thus performed where o-xylene was reacted in a
dimethylcyclohexane; (W) metu- and para-xylene; (A) 1,3-dimethyl-
and 1,4-dimethylcyclohexane.)(ii) Open symbols represent repli-
mixture with 1,2-dimethylcyclohexanes. Figure 3a,b
cate experiment; (0,0) truns-l,2-dimethyl- plus cis-1,2-dimeth- presents results from 0 wt %, 50 wt %, and 80 wt %
ylcyclohexane; (W, 0 ) m-plus p-xylene; (A,A) 1,3-dimethyl- plus 1,2-dimethylcyclohexanesin o-xylene. The net 1,2-
1,4-dimethylcyclohexane. dimethylcyclohexaneyields (Figure 3a) exhibited similar
behavior at t = 0 for all three experiments. Their
because of the lower loading in aromatics with higher subsequent evolution with time, however, was much
ring numbers. Equivalently, the lowest conversions different than the equilibrium predicted based on the
were observed for the SR hydrogenation, with the GAUS autocatalytic rate law (eq 4) and its parameters.
and TCFD distributions inhibiting in similar degree. Parameter estimation on eq 2 was thus performed for
all three experiments simultaneously. The results are
The evolution of products with time for the reaction summarized in the numerical values of the combined
of pure o-xylene is presented in Figure 2(i). Clearly, unimolecular and bimolecular rate expression of eq 5
m- and p-xylene were primary products, and their and the fair fit of Figure 3a,b.
concentrations reached constant values after 30 min.
cis- and truns-l,2-dimethylcyclohexanes were the major dy1,2 - r3.16 x
-- - y1,2/5.61 x
10-*(Yortho lo4) + 2.69 x
hydrogenation products observed, in very similar yields. dt
Relative to the behavior of the xylene products, the
dimethylcyclohexanes, including both cis- and truns-1,2-
dimethyl, formed with what appeared to be zero slope
at t = 0 in Figure 2(i). This is exposed more clearly by
Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995 106

.,
0.14
0.12

Figure 4. Proposed network for xylenes hydrogenation. (k in


0 60 120 180 240 300 360 0 0.2 0.4 0.6 0.8 1 L/(kg,t s). Numerator rate parameters underlined).
Time (min) Yield
0.12 . Interconversion between 1,2- and 1,3/1,4-dimethyl cy-
0.1 / clohexanes is also included. p-Xylene yields were less

1
than 10% of m-xylene yields and thus not always
quantitatively detectable. Consequently, m- and p -
0
* 0 0.06
2 . O 8 i m n , xylene were lumped into one component. The same was
0.04 true for 1,Cdimethyl- and 1,3-dimethylcyclohexanes.
Parameter estimation to this network exploited the
0.02
0
0
I30

60
m n
,
120
~ ~

180 240 300


., 360 420
range of available experimental compositions. Thus, the
pure o-xylene data, the 50 w t % data, the 80 w t % 1,2-
dimethylcyclohexane in o-xylene data, as well as the
Time (min)
m-xylene hydrogenation experimental data combined to
Figure 3. Initial composition dependence of xylene hydrogenation give the parameter set of Figure 4. Isomerization of o-
product distributions. (a) Net 1,2-dimethylcyclohexaneyield vs to m-xylene was approximately 1 order of magnitude
time for different initial compositions. (Curves represent param-
eter estimation results with eq 5. F = 2.439 x estimated slower than hydrogenation of o-xylene to 1,a-dimethyl-
deviation = 1.030 x low2.(0,0 ) 100 w t % o-xylene;(A) 50 w t % cyclohexanes. Hydrogenation of o-xylene proceeded
o-xylene, 50 wt % 1,2-dimethylcyclohexanes; (H) 20 w t % o-xylene, with a higher rate parameter than hydrogenation of the
80 wt % 1,2-dimethylcyclohexanes).(b)Reaction time as calculated m- and p-xylene lump; both reactions were virtually
from o-xylene hydrogenation yields for different initial composi- irreversible. Finally, the rate parameter for isomeriza-
tions. (Curves represent parameter estimation results using eq 5: tion of 1,2-dimethyl- to 1,3-and 1,4-dimethylcyclohex-
(W) o-xylene; (A) 1,2-dimethylcyclohexanes.)(c) 1,3-Dimethylcy-
clohexane yield from m-xylene for different initial compositions
anes was negligible. This parameter set provided the
((HI Reactant, pure m-xylene;(0) reactant, 50 w t % m-xylene, 50 calculated yield profiles corresponding to conversion
w t % 1,2-dimethylcyclohexanes.) interval 0-0.181 presented as the dotted curves in
Figure 2(i).
Note that the bimolecular rate parameter is two orders The reaction patterns discovered above were repeated
of magnitude lower than in eq 4, and the equilibrium for o-xylene hydrogenation in the PNA mixture (Table
ratio is much larger than 1. These results essentially 2). No shifts in selectivity were observed, although
nullified the hydrogen-transfer hypothesis. The best- o-xylene conversion was inhibited in the presence of
fit total yield profiles are presented in Figure 3b as the higher aromatics. This allowed for evaluation of ad-
curves, along with the experimental data. The time for sorption parameters for the aromatic ring number-based
a desired conversion was calculated from eq 5, and was lumps, as discussed below.
28.0 h for 0.470 conversion and 67.2 h for 0.818 Tetralin. Reaction of 29.75 g of tetralin took place
conversion. Note that, on the scale of Figure 3b, the in 420. g of cyclohexane solvent at 68.1 atm of Hz and
autocatalytic behavior at very low conversions is at 350 C. Conversion after 3 h was 0.137. The major
most a boundary layer phenomenon. products were trans-decalin, cis-decalin, and naphtha-
The absence of hydrogen-transfer reactions was also lene, which accounted for 0.991 f 0.011 material
demonstrated by two additional experiments involving balance closure. The balance consisted of traces of
m-xylene as the reactant, with and without the candi- decalin isomers, 1-and 2-methylindans and ring-open-
date hydrogen transfer additive 1,2-dimethylcyclohex- ing products (total max. 1.1%).Including those prod-
ane. Figure 3c demonstrates that the yields of 1,3- ucts, the material balance closure was 1,000 f 0.009.
dimethylcyclohexane were almost identical in both All products were identified by coinjection and mass
cases. Thus, the hydrogen-transfer hypothesis as an spectral information.
explanation of the apparent secondary rank of the Figure 5 summarizes the evolution of product yields
dimethylcyclohexanes was abandoned. and selectivities with tetralin conversion. Naphthalene
The bifunctional catalyst thus promoted both isomer- was the product with the highest initial selectivity,
ization and hydrogenation reaction pathways. The followed by trans- and cis-decalin. Notice the similari-
isomerizations to m- and p-xylene were substantially ties between the Delplots for tetralin in Figure 5 and
faster than hydrogenation; recall that the products of those for o-xylene in Figure 2. Decalins appeared to
the latter reaction appeared with a low initial rate. be fairly unreactive under these conditions, while
Although the xylene isomers yields were low-the naphthalene hydrogenated back into tetralin. All three
maximum yield to m- and p-xylene was 0.0131-they major products were thus considered primary in the
were achieved within the first 30 min of reaction. This network of Figure 6. The interconversion between cis-
suggested that the rates were high. Indeed, Figure 2b and trans-decalin was also included for completeness.
indicated that the initial selectivity to m- and p-xylene Parameter estimation to the network of Figure 6
was close to unity. using the tetralin yield vs conversion data and eq 3 is
This motivated construction of the network of o-xylene summarized as the smooth curves in Figure 5. The
reactions presented in Figure 4. o-Xylene hydrogenates equilibrium ratio for isomerization of trans- t o cis-
to 1,2-dimethylcyclohexanesin parallel with its isomer- decalin was obtained as the ratio of the two hydrogena-
ization to m- and p-xylene. The latter in turn hydro- tion equilibrium ratios, according to Denbighs rule, and
genate to 1,3-dimethyl- and 1,4-dimethylcyclohexanes. is denoted in italics.
106 Ind. Eng. Chem. Res., Vol. 34,No. 1, 1995

0.08 .
0.07 -
0.06 .
0.05 -
0.04 .

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14


x Tetralin og I
0 120 240 360 480
Time (min)
Figure 7. Effect of reacting mixture composition on naphthalene
GAUS; (+) LCO; (0)SR (A)TCFD; (A)
conversion kinetics. ((0)
EQ; (-1 Est.)

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14


x Tetralin
Figure 5. Kinetics of tetralin hydrogenation. Yield (i) and 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
selectivity (ii) vs conversion plots. (Curves represent the parameter x Naphthalene x Naphthalene
estimation results of Figure 6 ((0)trans-decalin; ( 0 )cis-decalin;
(W naphthalene).) 0.25
trans
-
e
.-
0.2

-x. c3
1
1u 0.361
?
.m

h
0.15

0.1

0.05

0
0 0.2 0.4 0.6 0.8 1 o 0.2 oi 0.i 0.i i
x Naphthalene x Naphthalene
m
Figure 6. Proposed network for tetralin hydrogenation. (k in
Figure 8. Kinetics of naphthalene hydrogenation. Yield (i) and
selectivity (ii) vs conversion plots for tetralin (a) and cis-decalin
L/(kg,,t 9). Numerator rate parameters underlined. F = 1.75 x (b). (Curves represent the parameter estimation results of Figure
estimated deviation = 0.009 35.) 9. (W Pure; (0) GAUS; (+I LCO; ( 0 )SR (A)TCFD; (A) EQ; (-)
Est.)
Figure 6 also summarizes the kinetics. The overall
rate parameter for hydrogenation of tetralin was slightly lowest conversions were observed for the SR and EQUIL
higher than that for hydrogenation of o-xylene. Con- hydrogenation experiments. This may be attributed to
sidering the similar position of the alkyl substituents the fact that although EQUIL distribution had a lower
with respect to the aromatic ring, this enhancement content in four-ring aromatics than SR, it had a higher
may be attributed to the presence of the saturated ring, combined content of three- and four-ring aromatics.
either by additional inductive effect or by improved These inhibition effects imply a higher adsorption
adsorption properties. These results also show that the parameter for the three-ring than for the two-ring
naphthalenic moiety hydrogenated 2 orders of magni- aromatics lump.
tude faster than the benzenic moiety. To quantify this The reaction network was revealed by the selectivity
effect further, naphthalene was employed as a reactant. vs conversion Delplots of Figure 8(ii). The initial
Naphthalene. Reaction of naphthalene in 420.g of selectivities show that tetralin was the sole primary
cyclohexane solvent was at 68.1 atm of H2 and 350 "C. product, whereas cis- and trans-decalins were clearly
The conditions for the experiments involving naphtha- secondary. Note that the data of Figure 8 coincide for
lene are presented in Tables 1-3. Reaction duration all experiments. This shows that the numerator rate
was 3.147 < tf < 9.574 h, and maximum conversion and equilibrium parameters are concentration-indepen-
varied between 0.952 < XM 0.998. The major hydro- dent and attests t o the validity of eq 3 and the
genation products were tetralin, trans-decalin, and cis- parameter estimation approach.
decalin, accounting for more than 95% of the initial Parameter estimation to the implied network of
weight of naphthalene. Traces of methylindans, alkyl- Figure 9 was performed using the rate expression of eq
benzenes, and alkylcyclohexaneswere detected as well. 3. The equilibrium ratio for isomerization of trans- to
Figure 7 summarizes the kinetics of conversion for cis-decalin was constrained as above. Figure 8 shows
the mixture experiments of Table 2. The highest that the resulting best-fit numerator parameters of
conversion for a particular time was observed for Figure 9 provide good agreement between model and
experiments with the LCO composition distribution; the data. Note that the kinetic parameters are quite close
Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995 107
hWlS The selectivity vs conversion plots of Figure ll(ii)
M
provided information about the reaction network. Di-
and tetrahydrophenanthrenes were clearly primary
products, the former appearing with higher initial
selectivity (projected 0.7 vs 0.3). On the other hand,
both octahydrophenanthrenes exhibited secondary be-
havior. Perhydrophenanthrene appeared very late in
cis the reaction network, and only with low yields.
Figure 9. Proposed network for naphthalene hydrogenation. (k The phenanthrene network contains an ambiguity
in I.4kg-t s). Numerator rate parameters underlined. F = 0.123; concerning the hydrogenation pathway and reactivity
estimated deviation = 0.0428.) of dihydrophenanthrene. This is still an open question
in the literature. Lemberton et al. (Lemberton and
Guisnet, 1984) and Wu et al. (Wu and Haynes, 1975)
suggest interconversion of di- to tetrahydrophenan-
threne. The driving force behind this reaction may be
energetic, i.e., the recovery of the two fused aromatic
ring system, in place of the two isolated aromatic rings.
Additionally, the dihydrophenanthrene molecule is ster-
ically strained, as the hydrogenation of the middle ring
forces the aromatic rings out of coplanarity (Shabtai et
al., 1978). In contrast, more recent studies (Girgis and
Gates, 1991), employing dihydrophenanthrene as the
reactant a t comparatively milder conditions than the
0% present study, suggest that its main reaction was
0 120 240 360 480 dehydrogenation to phenanthrene.
Time (min) The foregoing motivated the present studies of the
Figure 10. Effect of reacting mixture composition on phenan- catalytic reaction of dihydrophenanthrene with CoMol
GAUS;(+) LCO; (0)
threne conversion kinetics. ((0) S R (A)TCFD A1203 and the standard conditions of 350 "C and 68.1
(A) EQ;(-1 Est.)
atm of H2. As shown in Figure 12a(ii), the initial
selectivity to phenanthrene approached unity at zero
to those estimated from the experiment with tetralin conversion (0.58 conversion at the first 15 min), while
as a reactant (Figure 6). the initial selectivity to tetra- and octahydrophenan-
Phenanthrene. Reaction of phenanthrene in 420. threnes approached zero. The underlying driving force
g of cyclohexane solvent was a t 68.1 atm of H2 and 350 appeared to be thermodynamic, as demonstrated in
"C. The conditions for the experiments involving phenan- Figure 12b, where the evolution of the dihydrophenan-
threne are presented in Tables 1-3. Reaction duration threndphenanthrene ratio with time was plotted for the
varied between 3.147 tf < 9.574 h, and ultimate two experiments. In both cases, the ratio approached
conversions of 0.953 < X M < 0.996 were observed. The a constant value of approximately 0.6, which was very
identified products were dihydrophenanthrene, tetrahy- close to the equilibrium concentration ratio. Thus these
drophenanthrene, symmetric and asymmetric octahy- results suggest that a reasonable phenanthrene network
drophenanthrene, and perhydrophenanthrene. These would include a fast reversible reaction of phenanthrene
accounted for more than 97% of the initial weight of to dihydrophenanthrene with no further reaction of
phenanthrene at all cases. Traces of methylbiphenyls dihydrophenanthrene to deeper hydrogenation products.
and biphenyl were also detected (total max. 3.5%).
The kinetics of phenanthrene conversion at the mix- These observations led to the construction of the
ture compositions of Table 2 are presented in Figure network of Figure 13. Hydrogenation proceeded in a
10. The inhibition due to the presence of higher ring-by-ring manner, phenanthrene t o tetra- to octa-
aromatics is revealed in the lowest conversion for the and perhydrophenanthrene. Secondary hydrogenation
SR and EQUIL hydrogenation mixtures and the highest of dihydrophenanthrene does not occur in Figure 13.
conversion for the LCO mixture. Figure lla-c contains the results of parameter
Resolution of the reaction pathways required careful estimation as the smooth curves. The rate expression
analysis of the product spectra. Identification of 9,lO- of eq 3 and the phenanthrene network of Figure 13 were
di-, 1,2,3,4-tetra-, and perhydrophenanthrene was used with all experimental data involving phenanthrene
straightforward, as they were the only GC peaks cor- and dihydrophenanthrene simultaneously. The as-
responding to molecular weight 180, 182, and 192, octahydrophenanthrene to perhydrophenanthrene equi-
respectively. Five GC peaks of molecular weight 186 librium ratio was constrained by the three other equi-
were detected, correspondingto octahydrophenanthrene librium ratios in the network to account for the reaction
isomers. The three earlier-eluting GC peaks exhibited path degeneracy and is denoted in italics in Figure 13,
identical MS fragmentation patterns, including frag- along with the best-fit parameters. There is very good
ments of molecular weight 43 and 57-signifying two agreement between experimental and calculated yield
fused saturated rings-and were associated with 1,2,3,4,- profiles, as shown in Figures 11 and 12.
4a,9,10,10a-octahydrophenanthrene (as-octahydro- The parameter estimation results invite scrutiny of
phenanthrene) conformational isomers. The two later- qualitative structureheactivity relationships. Phenan-
eluting GC peaks exhibited a distinct MS fragmentation threne hydrogenation at the middle ring was kinetically
pattern with fragments of molecular weight 28-signify- favored over that at the terminal ring. This may be
ing a saturated ring fused with an aromatic ring-and attributed to the increased reactivity of the 9,lO-bridge
were attributed to 1,2,3,4,5,6,7,8-octahydrophenan- (Moreau and Geneste, 1990). Notice that the equilib-
threne (s-octahydrophenanthrene)conformational iso- rium ratio for this reaction was less than one (0.5821,
mers (McLafTerty, 1980). which indicates that more phenanthrene than dihydro-
108 Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995

0.7
E
0.6
e5 0.5
0.4
0.3
oh
0.2
0.1
c
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x Phenanthrene x Phenanthrene

P 0.35 >
0.14
I
8 0.12
I 0.1
9 0.08
8e 0.06
p" 0.04
2
E 0.02
h O
0 0.2 0.4 0.6 0.8 I 0 0.2 0.4 0.6 0.8 1
x Phenanthrene x Phenanthrene
$ 0.25
C
9P 0.2
e
oh 0.15
-
2

6 0.1

0.05

$ 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 I
x Phenanthrene x Phenanthrene
Figure 11. Kinetics of phenanthrene hydrogenation. Yield (i) and selectivity (ii) vs conversion plots for dihydrophenanthrene (a),
tetrahydrophenanthrene (b), and as-octahydrophenanthrene(c). (Curves represent the parameter estimation results of Figure 13. (W)
Pure; (0)GAUS; (+) LCO; (0) SR; (A)TCFD; (A) EQ; (-1 Est.)

::;Ji.
I * 1,
(a,)

0.7 '

0 50 100 150 200 0 0.5 1


T h e (min) x Dihydrophenanthrene Figure.13. Proposed network for phenanthrene hydrogenation.
(k in L4kg-t s). Numerator rate parameters underlined. F =
1, 0.0648; estimated deviation = 0.0334.)
g 0.8
5C 0.7 for the hydrogenation of a single aromatic ring. The
g . 0.6 0
* * * * e *o o* presence of saturated rings also had an effect on
0.5 -
o o
2 *O o o o o o reactivity. For example, the hydrogenation of octahy-
$ 0.4 . drophyanthrenes was much faster than that of either
0.3 .
2 0.2 . * tetralin or o-xylene. These results are in qualitative
0.1 - agreement with those of Aubert et al. (19881, who
04 1 suggested an anchoring effect of the naphthenic sub-
0 50 100 I50 200 stituents on the catalyst surface. This is contrary to
Time (min) the results reported by Shabtai et al. (19781, who
Figure 12. Kinetics of dihydrophenanthrene hydrogenation. (a) suggested that the bulkiness of the naphthenic rings
Yield (i)and selectivity (ii) vs conversion plots. (Curves represent would result in lower hydrogenation rates than less
the parameter estimation results of Figure 13. ( 0 )Phenanthrene; substituted molecules.
(A) dihydro-, (W) Tetrahydro-, (0)s-octahydro-, (0)as-octahydro-,
(A) perhydrophenanthrene.) (b) Equilibrium between phenan- Another revelation of the parameters in Figure 13 is
threne and dihydrophenanthrene. ((0)Dihydrophenanthrene re- the difference in hydrogenation rates between terminal
actant; ( 0 )phenanthrene reactant.) and middle rings for the "ring-substituted" benzenic and
naphthalenic moieties. The terminal ring in tetrahy-
phenanthrene is present at equilibrium. Rate param- drophenanthrene (naphthalenic moiety) hydrogenated
eters for hydrogenation of both two- and three-fused faster than the middle ring. This may be attributed to
aromatic rings were 1 order of magnitude higher than inaccessibility of the internal ring t o hydrogenation due
Ind. Eng. Chem. Res., Vol. 34,No. 1, 1995 109
1
0.6 -, - 1 I
0.9
0.8

c1
0.4 1I 0 .e- I
-5E 0.7
$ 0.6
C
8 0.5

0.5 0.6 0.7 0.8 0.9 1


x Anthracene 0 120 240 360 480
Time (min)
0.6
Figure 16. Effect of reacting mixture composition on pyrene
.
X
0.5

0.4
GAUS (e)LCO; (0)
conversion kinetics. ((0)
EQ; (-1 Est.)
S R (A)TCFD; (A)

.-"
h

2 parameters for the initial part of the anthracene net-


* 0.3
V work are larger than for phenanthrene, possibly due to
0.2
anthracene's decreased resonance stabilization energy
0. I (Moreau and Geneste, 1990). The equilibrium ratio for
the hydrogenation of the middle of the three rings is
0 larger than in the phenanthrene network, but still less
0.5 0.6 0.7 0.8 0.9 1 than unity. Deeper in the network, tetrahydroan-
x Anthracene thracene hydrogenation is overall slower than tetrahy-
Figure 14. Kinetics of anthracene hydrogenation. Yield (i) and drophenanthrene hydrogenation. The single-aromatic-
selectivity (ii) vs conversion plots. (Curves represent the parameter ring hydrogenation rate parameters observed from
estimation results of Figure 15. (A) Dihydro-, (H) tetrahydro-, (0) octahydroanthracenes' hydrogenation are also lower
s-octahydro-, (0)as-octahydro-, (A)perhydroanthracene.) than those regressed from phenanthrene network. The
preferential hydrogenation of a terminal vs an internal
ring in one- and two-aromatic ring systems was also
seen in both tetra- and octahydroanthracenes. Overall,
m Figure 14 shows that the results of parameter estima-
tion to the anthracene hydrogenation network proposed
above are quite good.
Pyrene. Reactions of pyrene took place in 420. g of
cyclohexane solvent at 68.1 atm of Hz and 350 "C. The
Figure 15. Proposed network for anthracene hydrogenation. (k conditions for experiments involving pyrene are pre-
in L/(kg,,t s). Numerator rate parameters underlined. F = 0.0164; sented in Tables 1-3. Reaction duration was 3.147 <
estimated deviation = 0.0158.) tf < 9.574 h, which afforded maximum conversions
between 0.831 < XM < 0.915. The major identified
to sterics. The same observations hold for as- vs products were dihydropyrene, tetrahydropyrene, sym-
s-octahydrophenanthrene hydrogenation. metric and asymmetric hexahydropyrene, decahydro-
Anthracene. Reaction of 2.06 g of anthracene took pyrenes A and B, and perhydropyrene. Traces of
place in 420. g of cyclohexane solvent at 68.1 atm of Hz alkylphenanthrenes and alkylbiphenyls were also de-
and 350 "C. The major products were dihydroan- tected.
thracene, tetrahydroanthracene, symmetric and asym- Figure 16 summarizes the kinetics of pyrene conver-
metric octahydroanthracene, and perhydroanthracene. sion for the mixture experiments of Table 2. A time-
These accounted for 0.968 f 0.043 of the material invariant plateau was established in all cases after
balance closure. The balance consisted of traces of approximately 4 h. Pyrene conversion rates also de-
hydrophenanthrene isomers, whose yields brought the creased with increasing heaviness of the feed. The rate
material balance closure to 1.000 & 0.048. Product was highest for the lightest feed (LCO) and lowest for
identification followed the reasoning detailed above for the heaviest feeds (SR and EQUIL). Pyrene conversion
phenanthrene; the retention times and order were very was more inhibited in the EQUIL composition distribu-
similar. tion than in the SR distribution.
The kinetics of anthracene hydrogenation are sum- Resolution of the reaction pathways required careful
marized in Figure 14. Di- and tetrahydroanthracenes analysis of the product spectra. The identification of
were clearly primary products, the former apparently pyrene, 4,5-dihydropyrene, and perhydropyrene was
appearing in higher initial selectivity, while octahy- straightforward from the mass spectra, as they were the
droanthracenes exhibited secondary behavior. This only possible products with molecular weights 202,204,
suggested the anthracene hydrogenation network to be and 218, respectively. The two hexahydropyrene iso-
as shown in Figure 15. mers were identified by co-injection of available s-
Parameter estimation using eq 3 in the network of hexahydropyrene (1,2,3,6,7,8-hexahydropyrene, Aldrich
Figure 15 and the yield vs conversion data of Figure 14 94%). The presence of 4,5,9,10-tetrahydropyrenewas
provided a quantitative summary of the kinetics. As deduced as it was the only stable tetrahydropyrene, and
was the case for phenanthrene, the equilibrium ratio in addition it eluted earlier than both hexahydropy-
for as-octahydroanthracene hydrogenation was con- renes, signifying the presence of single aromatic rings.
strained, according to Denbigh rule. The resulting rate Three GC peaks corresponding to molecular weight 212
110 Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995
0.8 ,
0.6 -
0.3 -
. 0.2 -
0.1 -
0,
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x Pyrene x Pyrene

0.06 0.14 -
e
t
0.05 -
$ 0.04 -
b
R 0.03 - $, 0.08 -
P
2 0.06-
2F 0.02 -

h 0.01 - $ 0.02 -
0 1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x Pyrene x Pyrene

0.18 0.25
I 5
2 0.16
h
P 0.14 ;0.2
*
g 0.12
I.

2a 0.1s
P 0.1
8 0.08
q 0.1
9 0.06 E
g 0.04 0.05
0.02
h O
ch O
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x Pyrene x Pyrene

Figure 17. Kinetics of pyrene hydrogenation. Yield (i) and selectivity (ii) vs conversion plots for dihydropyrene (a),tetrahydropyrene (b),
and as-hexahydropyrene (c). (Curves represent the parameter estimation results of Figure 18. (H)Pure; (0) GAUS; (e)LCO; (0) SR (A)
TCFD; (A) EQ; (-1 Est.)

were detected and attributed to decahydropyrene iso-


mers. The earlier eluting isomer exhibited an MS
fragmentation pattern distinctly different from the later
two, and was identified with 1,2,3,3a,4,5,9,10,10a,lOb-
decahydropyrene by the NBS-REVE mass spectral
library (decahydropyrene A), leaving the latter two GC
peaks for 1,2,3,3a,4,5,5a,6,7,8-decahydropyrene (decahy-
dropyrene B) conformational isomers.
Selectivity data for hydrogenation of pyrene and
hydropyrenes are presented in Figure 17a-c. The
selectivity vs conversion plots show clearly that dihy-
dropyrene appeared as a primary product. Tetra-, and
s- and as-hexahydropyrenes exhibited mixed primary Figure 18. Proposed network for pyrene hydrogenation. (k in
and secondary behavior. Finally, deca- and perhydro- L/(kg,, 9). Numerator rate parameters underlined. F = 0.153;
pyrenes were clearly of tertiary or higher rank. estimated deviation = 0.0513.)
These observations led t o the construction of the middle-ring hydrogenations were influenced by the
network of Figure 18. The interconnectivity of the equilibrium limitation, with the parent aromatic more
network rendered four equilibrium ratios redundant, abundant than the hydroaromatic at equilibrium.
and they were calculated in terms of the other equilib- Hydrogenation of dihydropyrene at the terminal ring
rium ratios in the network. They are presented in to as-hexahydropyrene was faster than the formation
italics in Figure 18. of tetrahydrophenanthrene from phenanthrene. as-
Subsequent parameter estimation provided an indica- Hexahydropyrene production directly from pyrene was
tion of structurelreactivity trends. Pyrene hydrogena- negligible. s-hexahydropyrene was a minor primary
tion at the middle ring was faster than hydrogenation hydrogenation product of pyrene, as well as an isomer-
a t the terminal ring. Further hydrogenation of the ization product of as-hexahydropyrene. This is in
phenanthrenic moiety in dihydropyrene at the middle agreement with the pathways involving consecutive
ring proceeded with a rate parameter very close to that rearrangements proposed in previous works (Johnston,
for phenanthrene itself. The equilibrium ratio for both 1984; Girgis, 1988).
middle-ring hydrogenations was less than unity. As The secondary reactions of s-hexahydropyrene were
was the case for phenanthrene, this indicates that notably slow. The rate constant for its further hydro-
Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995 111
1 The identification of di-, tetra-, and perhydrochrysene
0.9 was straightforward from the mass spectra, as they
= 0.8 were the only possible products with molecular weights
.-E 0.7 230, 232, and 246, respectively. The abundance of a
al
2 0.6 single peak a t mle 234, which could be either a hexahy-
drochrysene or a butylphenanthrene, was too low to
8 0.5
allow decisive discrimination. However, the formation
f 0.4
of butylphenanthrene was considered unlikely, because
5 0.3 of the low acidity of the catalyst and the lack of further
8 0.2 cracking products such as phenanthrene. Moreover, the
0.1 mle 234 peak appeared early in the network. This
,~ suggested a hexahydrochrysene product and, on the
0 120 240 360 480 basis of the pattern of selective terminal vs internal ring
Time (min) hydrogenation established so far, 1,2,3,4,5,6-hexahy-
Figure 19. Effect of reacting mixture composition on chrysene
drochrysene was selected against 5,6,11712,14,17-hexahy-
GAUS; (0)S R (A)TCFD; (A) EQ; (-)
conversion kinetics. ((0) drochrysene. The latter would have lower stability, as
Est.) a product of two consecutive internal ring hydrogena-
tions. The two octahydrochrysene isomers were differ-
genation to B-decahydropyrene was small compared to entiated by their fragmentation patterns: as-octahy-
values for similar moieties in other host molecules. The drochrysene was characterized by a prominent propyl-
location of the aromatic rings in internal parts of the enic fragment (236/193). The dodecahydrochrysene
molecule could adversely affect hydrogenation reactiv- isomers also exhibited two distinct fragmentation
ity. Indeed, in the as-hexahydropyrene isomer, hydro- patterns: one contained a propylenic fragment (240/197)
genation of the equivalent internal ring also proceeds in addition to the pronounced presence of an ethylenic
much more slowly than hydrogenation of the terminal fragment (240/212),and the other an M - 82 (2401158)
ring. peak in the absence of an ethylenic fragment. The first
A possible explanation for these trends could involve pattern implicated dodecahydrochrysene B, because of
steric as well as thermodynamic arguments. Assuming the presence of one saturated ring attached t o an
that s- and as-hexahydropyrene need to adsorb flatly aromatic. This left the other pattern for dodecahydro-
on the catalyst for the addition of all four hydrogen chrysene A.
atoms to proceed to the innermost loa- and 10b- The kinetics of chrysene conversion for the conditions
positions, the availability of sites would be reduced of Table 2 are presented in Figure 19. Chrysene
(compared to tetrahydrophenanthrene or naphthalene) conversion rates appeared t o be the fastest and most
due to the size and bulkiness of naphthenic substitu- insensitive to the reactant mixture composition of all
ents. In that respect, the sterics of their saturation previously examined PNAs. This may imply that chry-
would approximate single-ring saturations and should sene is the most strongly adsorbed of all model com-
therefore proceed with equivalent rate constants. The pounds examined.
equilibrium ratios in both unfavored hydrogenations are
also less than unity. This indicates a strong dehydro- The conversion dependence of the yields and selectivi-
genation trend of the resulting single-ring hydroaro- ties of the initial chrysene hydrogenation products are
matic. This is distinctly different from all other naph- summarized in Figure 20a-c. Di- and tetrahydrochry-
thalenic moieties and can be used to explain the senes were primary products, with approximately equal
equilibrium limitations in pyrene hydrogenation sug- projected initial selectivity (0.45),while hexahydrochry-
gested in the kinetics data of Figure 16. sene exhibited mixed primary and secondary behavior.
Rate constants for the hydrogenations of the benzenic Octa-, dodeca-, and perhydrochrysenes were clearly of
moieties in the decahydropyrenes were slightly higher rank 2 or higher.
than for the benzenic moieties in phenanthrene. It can The richness of the product spectrum and the pos-
be speculated that the presence of two more methylene sibilities opened by the secondary hydrogenation of
groups further enhanced hydrogenation reactivity. Ter- hexahydrochrysene complicated the chrysene network.
minal ring hydrogenations were again preferred to Thus, using the phenanthrene and naphthalene net-
internal ring hydrogenations. works as guides, all possible reaction pathways were
Figure 17a-c shows the goodness of fit of the optimal included in the chrysene reaction network. Parameter
rate and equilibrium parameters to the network of estimation t o the resulting network of Figure 21 using
Figure 18. Overall, the parameter estimation results, the relative rate expression of eq 3 reduced some of the
represented by the smooth curves of Figures 16 and 17, complexity caused by the phenanthrene-like reaction
represent the observed pyrene conversion kinetics well. path degeneracy. Six of the equilibrium ratios were
Chrysene. Reactions of chrysene in 420. g of cyclo- redundant due to the high degree of interconnectivity
hexane solvent were at 68.1 atm of Hz and 350 "C. The in the network and were constrained by the other
conditions for these experiments are presented in Tables equilibrium ratios. They are denoted in italics in Figure
1-3. Reaction duration was 3.147 < tf < 9.574 h, and 21.
this produced maximum conversions of 0.962 < X M Scrutiny of the rate law parameters of the chrysene
1.000. The major identified products were dihydrochry- hydrogenation network led to reactivity trends analo-
sene, tetrahydrochrysene, hexahydrochrysene, symmet- gous to those drawn from the preceding experiments.
ric and asymmetric octahydrochrysene, dodecahydro- Middle-of-threeand middle-of-fourring hydrogenations
chrysenes A and B, and perhydrochrysene. These major in tetrahydrochrysene and chrysene proceeded with
products accounted for more than 95% of the initial equilibrium ratios less than 1, as was the case for the
weight of chrysene in all cases. Traces of perhydro- corresponding moieties in phenanthrene and pyrene
chrysene isomers, alkylphenanthrenes, and alkylbiphe- (i.e., chrysene t o dihydrochrysene and tetrahydrochry-
nyls were also detected. sene to hexahydrochrysene). Hydrogenation at the
112 Ind. Eng. Chem. Res., Vol. 34,No. 1,1995
0.2 0.45
0.18 A
A
2 0.16 m
0.14
; 0.12
; 0.25
e 0.1
R 0.08
g 0.06
0.04 0.1
0.02 'h 0.05
0I
0 0.2 0.4 0.6 0.8 I 0 0.2 0.4 0.6 0.8 1

x Chrysene x Chrysene

$ 0.3
-
0.4
(b,) &*-;a%,
5 0.25
; 0.2
a
0.3 - b
R
-
'0
0.2 -
c
Eb 0.15
0.1
f k2 0.1 -
0.05 E
h
0 0,
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 I
x Chrysene x Chrysene

0.2 , I 025, I

'ok 0.1
c 0.1
x / x
0.05
h
.x
0.05
+
0 0.2 0.4 0.6 0.8 I 0 0.2 0.4 0.6 0.8 1
x Chrysene x Chrysene
Figure 20. Kinetics of chrysene hydrogenation. Yield (i) and selectivity (ii) vs conversion plots for dihydrochrysene (a),tetrahydrochrysene
(b), and hexahydrochrysene (4.(Curves represent the parameter estimation results of Figure 21.(m) Pure; (0) GAUS; (0) SR (A)TCFD;
(A) EQ; (-1 Est.)

to B-dodecahydro

OLW 1111.82

29.0 11 0.0362
to asym-octahydro
Figure 21. Proposed network for chrysene hydrogenation. ( K in L/(kg,t s). Numerator rate parameters underlined. F = 0.40912; estimated
deviation = 0.1013.)
middle ring was faster than hydrogenation at the parameters and equilibrium ratios larger than unity,
terminal ring for tetrahydrochrysene and of the same which is typical of the benzenic moieties. Hydrogena-
order of magnitude as that for chrysene. Based on these tion of the terminal ring in hexahydrochrysene was
values, chrysene was the most reactive of all molecules preferred to hydrogenation of the internal ring.
examined. The similar reactivities of the s- and as-octahydro-
The reactivity of dihydrochrysene was essentially chrysenes was also similar to or higher than the
limited to hydrogenation of its naphthalenic moiety to reactivity of other naphthalenic moieties. The A- and
hexahydrochrysene. Hydrogenation of the benzenic B-dodecahydrochryseneswere fairly reactive for single-
moiety in dihydrochrysene and further reactions of aromatic-ring hydrogenation, with A-dodecahydrochry-
hexahydrochrysene proceeded with very small rate sene the most reactive molecule in the benzenic class.
Ind. Eng. Chem. Res., Vol. 34,No. 1, 1995 113
Table 4. Aromatic Ring Number-Based L u m p regression to approximately 1700 experimental data
Adsorption Constants from Parameter Estimation to Eq points. Anthracene and tetralin data were not used, as
2 for the Emeriments in Table 2
these components were not present in the mixture
K (Umol) experiments. cis- and trans-decalins were considered
four ringdpyene 38.5 as one lump in the naphthalene network, as were all
four ringdchrysene 38.5 dimethylcyclohexanes and all xylenes in the o-xylene
three-ring lump 17.5 network. These approximations reduced the number of
two-ring lump 7.7 numerator rate and equilibrium parameters to 68.
one-ring lump 7.4
saturates lump 3.9 These were held constant, as the five adsorption pa-
rameters of the aromatic ring number-based lumps were
The general pattern appears to confirm the conclusion optimized t o the 1700 data points.
that reactivity increases with the addition of naphthenic No statistical difference was observed between LHHW
rings. models with denominator exponents n = 1 and n = 2
Figure 20a-c shows, as the curves, the results of rate (eq 2). Therefore, the simpler case of n = 1was adopted.
and equilibrium parameter estimation to the network The results are presented in Table 4 and their
of Figure 21. The parameter estimation results, rep- implications are illustrated in Figure 22a-b. Table 4
resented by the smooth curves of Figures 19 and 20, shows that the value of the adsorption parameters
represent the observed chrysene conversion kinetics increased with the number of aromatic rings, i.e., it was
well. lowest for the saturates lump and highest for the four-
ring compounds lump. A manifestation of this is
Adsorption Results demonstrated in Figure 22a, where the four-, three-, and
The compound-by-compound kinetic information pre- two-aromaticring number-based lump yields are plotted
sented thus far allowed the exploration of detailed with respect to time for all the initial compositions.
hydrogenation reaction pathways. The application of Although the absolute value of the yield for a particular
eq 3 resulted in the evaluation of 45 LHHW numerator time depends on the initial composition, it is clear that
rate parameters and 45 numerator equilibrium ratios, the rates are essentially independent of the initial
14 of which were constrained. This parameter set composition of the mixture for the four-aromatic ring
satisfactorily represented the yields of products with lump. The rates for the three-aromatic ring lump
respect to reactant conversion. The effect of the reacting decreased with increases in the heaviness of the com-
mixture composition on reactant conversion provides the position. This effect is even more pronounced for the
remaining piece of the quantitative rate law that will two-aromatic ring lump.
in turn allow predictions with respect to time or space The parameter estimation results are shown as the
velocity. smooth curves in Figure 1 for xylene, Figure 7 for
To this end, parameter estimation to eq 2 was naphthalene, Figure 10 for phenanthrene, Figure 16 for
performed using all available data simultaneously for pyrene, and Figure 19 for chrysene. In Figure 22b a
the time dependence of component yields. This utiliza- typical example of the results for the reactants of the
tion of all the experiments of Table 2 represented straight run composition experiment is presented (SR,

1.2 1.6 1.2


0 1 0 1.4 % 1
'j; 0.8
'j; 1.2

z 1 E 0.8
5
4
P
0.6 5 0.8
-1
0.6
g 0.4 0.6
M
t 0.4
3 0.2
0.4
4 0.2 a 0.2
-l

0 0 0
0 120 240 360 480 0 120 240 360 480 0 120 240 360 480
Time (min) Time (min) Time (min)

1
0.9
0.8
0.7
'j; 0.6
*
5 0.5
*
g 0.4
2 0.3
0.2
0.1
0
0 60 120 180 240 300 360 420 480
Time (min)
Figure 22. Adsorption parameter estimation results. (a) Aromatic ring number-based lump yields with respect to time for the four-
aromatic ring lump (i), three-aromatic ring lump (ii), and two-aromatic ring lump (iii).(Curves represent results with the adsorption
parameters of Table 4. (0)GAUS; (e) LCO; (0) SR (A)TCFD; (A) EQ; (-) Est.) (b) Reactant yields vs time for the straight run distribution
(SR).(Curves represent results with the adsorption parameters of Table 4. (W) o-Xylene; (0)
naphthalene; ( 0 )phenanthrene; (A) pyrene;
(A)chrysene; (--) Est.)
114 Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995

I I I
0.00947 10.0269 a I 0.394 @@ I
0.00356 b.00733 [r0.0992@@0 1 0.160 I 0.686 e 0.708 @@@
0.336 &
0.0743 @ 10.0937
I
0.0925 @ 0.403 @ I 0.393 @ 0.379 @
0.0288 @ 0.200@ 0.599
89
0 . 1 3 0 8 !0.0362@ 0.608

I 0.0253&@ 0.480 8 0.428 &@


o.Oo0 &Io.OOo I I
Benzenic Naphthalenic 'henanthrenic
Figure 23. Rate parameters for hydrogenation of polynuclear aromatics. (Shade indicates the ring being saturated. k in L/(kgCts).)

Table 3). The parameter set thus represents the yield The numerator rate parameters of Figure 23 implic-
vs time experimental data very well. itly contain a hydrogen pressure dependence, which, if
extracted, would further enhance the separation of the
three hydrogenation classes. That is, division of the rate
Discussion parameters by the hydrogen pressure raised to the
Qualitative Hydrogenation Reactivity Trends. hydrogen stoichiometric coefficient, a,would decrease
This experimental plan yielded reactivity information the rate constant most for the benzenic class (a= 3)
for a wide range of aromatic and hydroaromatic mol- and least for the phenanthrenic hydrogenation class (a
ecules under the same set of reaction conditions. Over- = 1).
all, 45 ring saturation reactions were probed. The The numerator rate parameters also encompass the
regressed numerator rate parameters suggest the fol- kinetically possible symmetries. For example, hydro-
lowing qualitative trends: (1) PNA hydrogenation pro- genation of phenanthrene at the terminal ring can occur
ceeded in a ring-by-ring manner; (2) hydrogenation a t two possible positions, while hydrogenation of tet-
reactivity increased with the number of aromatic rings; rahydrophenanthrene at the terminal ring can occur at
(3) for groups with the same number of fused aromatic only one position. Other examples of such symmetries
rings, hydrogenation reactivity increased with the pres- include hydrogenations of pyrene and chrysene a t the
ence of alkyl substituents and naphthenic rings; (4) for middle ring and hydrogenations of s-hexahydropyrene
moieties with one and two aromatic rings, hydrogena- and s-octahydrochrysene. All these factors need to be
tion of a ring located at the end of the molecule was taken into account for the quantitative interpretation
faster than hydrogenation of an internal ring. of numerator rate parameter information (Korre, 1994;
Korre et al., 1995).
These trends suggest that three broad categories of HydrogenatiodDehydrogenationEquilibria Al-
saturation, differentiated by numerator rate parameter though the selected conditions and most measurements
magnitude and driving force, are operative. These were intended to emphasize the hydrogenation rate, the
categories are illustrated in Figure 23. The first cat- longer-time mixture experiments allowed regression of
egory, single aromatic ring hydrogenation, was termed reliable dehydrogenation rate parameters for each step
benzenic hydrogenation (six hydrogen atoms added). in each network. This was revealed in the form of
Hydrogenation of one out of two fused aromatic rings equilibrium ratios, as shown in eqs 6 and 7. For
was the naphthazenic hydrogenation class, where four hydrogenation of an aromatic, (A), to a hydroaromatic
hydrogen atoms were added. Saturation of the terminal (HA),with n mol of H2, the equilibrium parameter, Kes,
of three- or four-hsed aromatic ring compounds has also can be obtained by dividing the equilibrium ratio, K,
been included in this group. The unique hydrogenation by the hydrogen pressure to the nth power:
of an aromatic ring fused between aromatic rings
defines the phenanthrenic hydrogenation category, where
two hydrogen atoms are added.
There are several possible causes for separation into
distinct hydrogenation classes based on the number of
fused aromatic rings. As the number of fused aromatic
rings increases, the resonance stabilization energy per (7)
aromatic ring decreases (Dewar, 1969) and the highest
electronic density increases (Neurock and Klein, 1993).
Both these factors can account for the observed increase The equilibrium ratio may be viewed as the equilibrium
in numerator rate parameter magnitude (Korre et al., constant for 1 atm of hydrogen pressure. It also
1995). possessed intuitive bounds. For example, an equilibri-
Ind. Eng. Chem. Res., Vol. 34,No. 1, 1995 115

I I I
1OOOOO I 151. a I 55.3 @@ I
78.9 I 242. 4.11 I 12.6 I 7.84 @@@ 0.934 @@@
64.7 & II 170. e 2.34 & II 6.16 I
I 8.33 0.582 &
10.5 8I 14.9 @ 0.645 @ I 6.67 &) I 2.74 @ 0.238 @

188. 6.48 @ 3.76 & 0.681 @ 91.2 &@ 0.708 @


6.16 5.12 @@ 14.2 8 11.8 &@ I 29.0 0.511 @
809. 818 6.72 I 3.31 &@ 2.29 &@ 0.353 &@
- &I- g?jQ@ I I
I Benzenic Naphthalenic Phenanthrenic
Figure 24. Equilibrium concentration ratios for hydrogenation of polynuclear aromatics. (Shade indicates the ring being saturated. K
= KePHzn-I

1000

100

6
e-
m
10

A;
1 I I
I I
I I
I I
I I
I I

0.1 I
I
I
I
I
I

0.1 1 10 100 1000


Frye et al.
Figure 25. Comparison between literature and experimentally determined equilibrium ratios. (Hydrogenations in naphthalene and
phenanthrene network a t 350 "C and 68.1 atm of Hz.)

um ratio K < 1implies more aromatic than hydroaro- only one additional aromatic ring, such as in the series
matic at equilibrium, and thus indicates a compara- naphthalene, tetrahydrophenanthrene, octahydrochry-
tively unfavorable reaction at the given conditions. In sene, etc. Note that 2 mol of H2 are added in these
the present work the equilibrium ratio was of order instances. These equilibrium ratios also dropped as
unity or higher. more naphthenic (and aromatic)rings were added to the
The best-fit equilibrium ratios K are listed in Figure unit sheet. Indeed, the value for s- and as-hexahydro-
24. Several qualitative trends are apparent. For pyrene hydrogenation a t an internal ring was less than
hydrogenation of an isolated aromatic ring, such as in unity. This implies that the hexahydropyrenes are more
benzenic hydrogenation, where three hydrogen mol- abundant than the decahydropyrenes a t equilibrium
ecules are added, the equilibrium ratios are usually and explains the equilibrium limitations observed in the
much larger than unity. This implies that the fully overall pyrene network.
saturated (perhydrogenated) molecule is much more Comparatively small equilibrium ratios K < 1were
abundant than the parent single-ring aromatic a t equi- also observed for the hydrogenation of the middle-of-
librium. The best-fit equilibrium ratio is very large for three or -four aromatic rings, such as in the addition of
xylene (K > 1000) and gradually decreases as more 1 mol of H2 to phenanthrene, pyrene, dihydropyrene,
naphthenic rings are added. Thus it appears that the chrysene, and tetrahydrochrysene.
more complex the hydroaromatic structure, the lower Figure 25 compares calculations of the equilibrium
the equilibrium ratio. ratios K based on experimental work by Frye and co-
Equilibrium ratios larger than unity were also found workers (Frye, 1962; Frye and Weitcamp, 1969) against
for hydrogenation of a terminal aromatic ring fused to those found in the present experimental work for
116 Ind. Eng. Chem. Res., Vol. 34, No. 1, 1995
reactions in the naphthalene and phenanthrene net- Three classes of hydrogenation were discerned, based
work at 350 "C and 68.1 atm of Ha. An excellent on the magnitude of saturation numerator rate par-
agreement persists over about 4 orders of magnitude, ameters: 1,benzenic (single isolated aromatic ring); 2,
lending credibility to the values reported here and naphthalenic (two isolated aromatic rings or terminal
summarized in Figure 24. of three- and four-aromatic ring systems); 3, phenan-
Adsorption. The adsorption constant trends based threnic (middle-of-three or four fused-aromatic rings).
on aromatic ring number are consistent with an as- Hydrogenation equilibrium ratios were much larger
sumption of acidhase interactions between the catalyst than unity for the benzenic hydrogenation class, gener-
and the adsorbed molecule. Indeed, the observed in- ally larger than unity for the naphthalenic hydrogena-
crease in adsorption constants with increasing aromatic tion class, and smaller than unity for the phenanthrenic
ring number may be attributed to the concurrent hydrogenation class. There was excellent agreement
increasing basicity of the PNAs. This has been revealed between the values reported here and those published
by both experimental measurements of gas-phase ba- by Frye and co-workers (Frye, 1962) for the same
sicity (LaVopa and Satterfield, 1988)) as well as com- conditions.
putational chemistry calculations of proton affinity Adsorption parameters were evaluated for aromatic
(Neurock and Klein, 1993). ring number-based lumps. They clearly increased with
The adsorption equilibrium constants for the aromatic increasing aromatic ring number, which is consistent
ring number-based lumps listed in Table 4 are es- with an acid-base interaction between catalyst and
sentially an average of the molecules present in mix- adsorbed molecule.
tures of different compositions that contained the same
number of aromatic rings. This renders them a very Acknowledgment
good estimate of the competitive inhibition in a mixture;
nonetheless, they must yet be considered as devoid of The authors acknowledge the financial support of
rigorous microscopic structural significance (i.e., adsorp- Mobil Research and Development Corp. (Paulsboro
tion energetics or other adsorption fundamentals). Research Laboratory) and the State of Delaware, as
It is thus reasonable to suspect that significant authorized by the State Budget Act of Fiscal Years
differences could be observed on a microscopic scale 1990-1992. Mr. Dennis Kalaygian's help with the
within a lump. They would depend on the number of laboratory experiments is greatly appreciated.
saturated rings, their position with respect to the
aromatic ring(s), and the nature of their adsorption on Literature Cited
a given catalytic site. A bulky hydroaromatic molecule, Aubert, C.; Durand, R.; Geneste, P.; Moreau, C. Factors Mecting
such as B-decahydropyrene, could adsorb flatly on the the Hydrogenation of Substituted Benzenes and Phenols over
catalytic surface, and thus reduce the number of sites a Sulfided NiO-MoO$y-A.l203 Catalyst. J. Catal. 1988,112,12-
available to other molecules. Even with the same 20.
energetics of its one aromatic ring adsorption lump, it Bhinde, M. V. Quinoline Hydrodenitrogenation Kinetics and
could thus appear in kinetics experiments to have a Reaction Inhibition. Ph.D. Thesis, University of Delaware, 1979.
higher adsorption constant than o-xylene, for example. Bhore, N. A.;Klein, M. T.; Bischoff, K. B. The Delplot Technique:
Such differences could be observed even between iso- A New Method for Reaction Pathway Analysis. Ind. Eng. Chem.
Res. 1990,29,313-316.
mers, such as A- and B-decahydropyrenes. Further Dewar, M. J. S. The Molecular Orbital Theory of Organic Chem-
elaboration of these ideas could reveal more information istry; McGraw-Hill New York, 1969.
and will be pursued in a following publication (Korre Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and
et al., 1995; Korre, 1994). Design; J. Wiley & Sons, Inc.: New York, 1990.
Frye, C. G. Equilibria in the Hydrogenation of Polycyclic Aromatics
J . Chem. Eng. Data 1962,7,592-595.
Conclusions Frye, C. G.; Weitcamp, A. W. Equilibrium Hydrogenations of
Multi-Ring Aromatics. J. Chem. Eng. Data 1969,14,372-376.
An extensive experimental plan provided a consistent Girgis, M. Reaction Networks, Kinetics and Inhibition in the
data base of quantitative reactivity information for 36 Hydroprocessing of Simulated Heavy Coal Liquids. Ph.D Thesis,
aromatic and hydroaromatic compounds containing up Delaware, 1988.
to four aromatic rings subject to 45 reactions at 350 "C Girgis, M. J.; Gates, B. C. Reactivities, Reaction Networks, and
and 68.1 atm of Hz. Kinetics in High-Pressure Catalytic Hydroprocessing. Ind. Eng.
Chem. Res. 1991,30,2021-2058.
o-Xylene hydrogenation proceeded with relatively Haynes, H. W. J.; Parcher, J. F.; Helmer, N. E. Hydrocracking
slow rates, inhibited by the presence of other aromatics. Polycyclic Hydrocarbons over a Dual-Functional Zeolite (Fau-
Tetralin hydrogenation to cis- and tram-decalin was jacite)-Based Catalyst. Ind. Eng. Chem. Process Des. Deu. 1983,
equally slow, while dehydrogenation to naphthalene 22,401-409.
occurred until equilibrium concentrations were reached. Huang, C.-S.; Wang, K.-C.; Haynes, H. W. J. Hydrogenation of
Phenanthrene hydrogenation proceeded through tet- Phenanthrene over a Commercial Cobalt Molybdenum Sulfide
rahydro- and octahydrophenanthrenes to complete satu- Catalyst Under Severe Reaction Conditions. In Liquid Fuels
from Coal; Academic Press: New York, 1977;pp 63-78.
ration, while experiments with dihydrophenanthrene Johnston, K P. Hydrogenation-Dehydrogenation of Pyrenes Cata-
established it as a "dead end" in the network, its only lyzed by Sulfided Cobalt-Molybdate at Coal Liquefaction Condi-
reaction being dehydrogenation t o phenanthrene. An- tions. Fuel 1984,63,463-468.
thracene pathways were similar to those for phen- Korre, S.C. Quantitative Structure/Reactivity Correlations as a
anthrene. Pyrene hydrogenation pathways were equi- Reaction Engineering Tool Applications to Hydrocracking of
librium limited, probably due to the high dehydro- Polynuclear Aromatics. Ph.D. Thesis, University of Delaware,
genation reactivity of the intermediate tetra- and hexahy- 1994.
dropyrenes; perhydrogenated products were detected, Korre, S. C.; Neurock, M.; Klein, M. T.; Quam, R. J. Polynuclear
Aromatic H y d " Hydrogenation 2.Quantitative Structure/
though. Finally, chrysene hydrogenation was the fast- Reactivity Correlations. Chem. Eng. Sci. 1995,in press.
est and the least affected by equilibrium or inhibition Landau, R. N. Chemical Modeling of the Hydroprocessing of Heavy
considerations. Oil Feedstocks. Ph.D. Thesis, University of Delaware, 1991.
Ind. Eng. Chem. Res., Vol. 34,No. 1, 1995 117
Lapinas, A. T. Catalytic Hydrocracking of Fused-Ring Aromatic Salim, S. S.; Bell, A. T. Effect of Lewis Acid Catalysts on the
Compounds: Chemical Reaction Pathways, Kinetics and Mecha- Hydrogenation and Cracking of Two-Ring Aromatic and Hy-
nisms. Ph.D. Thesis, University of Delaware, 1989. droaromatic Structures Related to Coal. Fuel 1982,61, 745-
Lapinas, A. T.; Klein, M. T.; Gates, B. C.; Macris, A.; Lyons, J. E. 753.
Catalytic Hydrogenation and Hydrocracking of Fluoranthene: Salim, S. S.; Bell, A. T. Effects of Lewis Acid Catalysts on the
Reaction Pathways and Kinetics. Znd. Eng. Chem. Res. 1987, Hydrogenation and Cracking of Three-Ring Aromatic and
26,1026-1033. Hydroaromatic Structures Related to Coal. Fuel 1984,63,469-
LaVopa, V.; Satterfield, C. Poisoning of Thiophene Hydrodesulfu- 475.
rization by Nitrogen Compounds. J . Catal. 1988,110,375-387. Shabtai, J.; Veluswami, L.; Oblad, A. G. Steric Effects in Phenan-
Lemberton, J.-L.; Guisnet, M. Phenanthrene Hydroconversion as threne and Pyrene Hydrogenation Catalyzed by Sulfided NiW/
a Potential Test Reaction for the Hydrogenating and Cracking A l 2 0 3 . Am. Chem. SOC.Diu. Fuel Chem. Prepr. 1978,23,107-
Properties of Coal Hydroliquefaction Catalysts. Appl. Catal. 112.
1984,13,181-192. Shaw, R.; Golden, D. M.; Benson, S. W. Thermochemistry of Some
McLafYerty, F. W. Interpretation of Mass Spectra; University Six-Membered Cyclic and Polycyclic Compounds Related to
Science Books: Mill Valley, CA, 1980. Coal. J. Phys. Chem. 1977,81,1716-1729.
Moreau, C.; Geneste, P. Factors Afecting the Reactivity of Organic Stein, S. E.;Golden, D. M.; Benson, S. W. Predictive Scheme for
Model Compounds in Hydrotreating Reactions. In Theoretical Thermochemical Properties of Polycyclic Aromatic Hydrocar-
Aspects of Heterogeneous Catalysis; Van Nostrand Reinhold: bons J . Phys. Chem. 1977,81.
New York, 1990;pp 256-310. Stephens, H. P.; Chapman, R. N. The Kinetics of Catalytic
Moreau, C.; Joffre, J.; Saenz, C.; Geneste, P. Hydroprocessing of Hydrogenation of Pyrene Implications for Direct Coal Liquefac-
Substituted Benzenes over a Sulfided CoO-Mo03/y-Al203Cata- tion Processing. Am. Chem. SOC.Diu. Fuel Chem. Prepr. 1983,
lyst. J . Catal. 1990,122,448-451. 28,161-168.
Neurock, M.; Klein, M. T. Linear Free Energy Relationships in Stephens, H. P.; Kottenstette, R. J. The kinetics of Catalytic
Kinetic Analyses: Applications of Quantum Chemistry. Poly- Hydrogenation of Polynuclear Aromatic Components in Coal
cyclic Arom. Compd. 1993,3,231-246. Liquefaction Solvents. Am. Chem. SOC.Diu. Fuel Chem. Prepr.
Press, W.H.; Flannery, B. P.; Teukolsky, B. P.; Vetterling, W. T. 1986,30,345-353.
Numerical Recipes; Cambridge University Press: Cambridge, van der Eijk, H.; den Otter, G. J.; Blauwhoff, P. M. M.; Maxwell,
U.K., 1986. I. E. The Application ofAdvanced Process Models in Oil Refining
Qader, S. A. Hydrocracking of Polynuclear Aromatic Hydrocarbons R&D. Chem. Eng. Sci. 1990,45(8), 2117-2124.
over Silica-Alumina Based Dual Functional Catalysts J. Znst. Wu, W.; Haynes, H. W. J. Hydrocracking Condensed Ring Aromat-
Pet. 1973,59, 178-187. ics Over Non-Acidic Catalysts. In Hydrocracking and Hy-
Qader, S. A.; Hill, G. R. Development of Catalysts for the drotreating; ACS Symposium Series 20,Philadelphia, PA, 1975;
Hydrocracking of Polynuclear Aromatic Hydrocarbons. Am. pp 466-477.
Chem. SOC.Diu. Fuel Chem. Prepr. 1972,16,93-106. Zeuthen, P.; Stolze, P.; Pedersen, U. B. Kinetics for Simultaneous
Qader, S. A.; McOmber, D. B. Conversion of Complex Aromatic HDS, HDN and Hydrogenation Model Reactions on a Co-Mol
Structures to Akylbenzenes, In Hydrocracking and Hydrotreat- A1203 Catalyst. Bull. SOC.Chim. Belg. 1987,96,985-995.
ing; ACS Symposium Series 20; Philadelphia, PA, 1975; pp
479-488. Received for review March 17,1994
Qader, S. A.; McOmber, D. B.; Wiser, W. H. Evaluation of Revised manuscript received August 10,1994
Mordenite Catalysts for Phenanthrene Hydrocracking. Am. Accepted August 26,1994*
Chem. SOC.Diu. Fuel Chem. Prepr. 1973,18,127-137.
Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Properties of Gases @ Abstract published in Advance A C S Abstracts, November
and Liquids; McGraw-Hill: 1987. 1, 1994.

You might also like